Apprehension and Argument: Ancient Theories of Starting Points for Knowledge (Studies in the History of Philosophy of Mind)

APPREHENSION AND ARGUMENT

STUDIES IN THE HISTORY OF PHILOSOPHY OF MIND Volume 3

Editors Henrik Lagerlund, Uppsala University, Sweden Mikko Yrjönsuuri, Academy of Finland and University of Jyväskylä, Finland Board of Consulting Editors Lilli Alanen, Uppsala University, Sweden Joël Biard, University of Tours, France Michael Della Rocca, Yale University, U.S.A. Eyjólfur Emilsson, University of Oslo, Norway André Gombay, University of Toronto, Canada Patricia Kitcher, Columbia University, U.S.A. Simo Knuuttila, University of Helsinki, Finland Béatrice M. Longuenesse, New York University, U.S.A. Calvin Normore, University of California, Los Angeles, U.S.A.

Aims and Scope The aim of the series is to foster historical research into the nature of thinking and the workings of the mind. The volumes address topics of intellectual history that would nowadays fall into different disciplines like philosophy of mind, philosophical psychology, artificial intelligence, cognitive science, etc. The monographs and collections of articles in the series are historically reliable as well as congenial to the contemporary reader. They provide original insights into central contemporary problems by looking at them in historical contexts, addressing issues like consciousness, representation and intentionality, mind and body, the self and the emotions. In this way, the books open up new perspectives for research on these topics.

APPREHENSION AND ARGUMENT Ancient Theories of Starting Points for Knowledge

by MIIRA TUOMINEN University of Helsinki, Finland

A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN 10: 1-4020-5042-9 (HB) ISBN 13: 978-1-4020-5042-8 (HB) ISBN 10: 1-4020-5043-7 (e-book) ISBN 13: 978-1-4020-5043-5 (e-book)

Published by Springer, P.O. Box 17, 3300 AA Dordrecht, The Netherlands. www.springer.com Printed on acid-free paper

Cover art: Albrecht Dürer, Unterweisung der Messung, University of Helsinki Library, The National Library of Finland.

All Rights Reserved © 2007 Springer No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.

TABLE OF CONTENTS

Acknowledgements

ix

Abbreviations and a Note on the Texts

xi

Introduction The Topic, Scope, and Aim of this Book The Structure of the Book A Brief Survey of the Existing Literature

1 2 6 12

PART I: PLATONIC-ARISTOTELIAN TRADITION 1. Theories of Argumentation 1.1 Plato Arguments as Socratic Discussions The Method of Hypothesis Collection and Division Philosophical Cosmology

17 17 18 22 32 35

1.2 Aristotle 1.2.1 Aristotle’s Inheritance from the Academy Dialectical Syllogisms Induction Conceptual Analysis 1.2.2 Science Being Better Known Premises of Scientific Proofs Proofs and Definitions Do the Sciences Have Something in Common? Remarks on Aristotle’s Scientific Practice Knowledge of the Premises

37 38 38 59 65 68 69 72 86 89 96 102

1.3 Later Developments 1.3.1 Some Developments in Platonism Galen Alcinous Plotinus

112 113 113 118 122

v

vi

TABLE OF CONTENTS 1.3.2 Greek Commentaries on Aristotle Alexander of Aphrodisias Themistius Philoponus Simplicius

2. Intellectual Apprehension 2.1 The Connection between the Two Contexts 2.2 Perception 2.2.1 Receptive Theories Causation through Medium 2.2.2 Projective Theories 2.2.3 Co-affection: Plotinus 2.2.4 Perceptual Realism and the Reliability of Perceptions Plato: Realism without Reliability? Aristotle’s Realism: Perceptibility as a Modalised Notion 2.3 From Perception to Intellection 2.3.1 Intelligible Forms Plato Aristotle 2.3.2 Later Developments Galen, Alcinous, Plotinus Alexander, Themistius, Philoponus

126 127 138 142 149 155 155 162 163 164 168 170 171 172 173 175 176 176 181 194 194 199

PART II: ALTERNATIVE APPROACHES 3. Hellenistic Philosophy 3.1 Is there a Starting Point for Knowledge? The Notion of a Criterion of Truth Perceptions and Cognitive Impressions Preconceptions The Problem of Vagueness 3.2 Is There a Transition from the Evident to the Non-Evident? 3.2.1 Epicurus Witnessing and Counter-Witnessing The Method of Elimination and the Method of Similarity 3.2.2 Stoics and Sextus Indemonstrable Argument Forms

219 222 222 228 238 251 254 255 255 260 265 265

TABLE OF CONTENTS

3.3 3.4

Proofs Arguments Involving a Non-Necessary Conditional Rejection of Proof What is Left for the Sceptic? Pyrrhonian Scepticism and Non-Dogmatic Beliefs What Does a Doctor Know? – Medical Empiricism as an Alternative Approach to Scientific Knowledge The Sorites Argument in Medicine Empiricist Expertise

vii 266 269 271 272 273

276 276 281

Conclusion

289

Bibliography

295

Index of Names

313

Index Locorum

317

Index of Topics

325

ACKNOWLEDGEMENTS

This book started living many years ago when, as an undergraduate student, I wrote a few small essays on Aristotle’s Posterior Analytics. In its present form, it has emerged from a doctoral dissertation at the University of Helsinki, which was supervised by Prof. Simo Knuuttila, to whom I wish to express my deepest gratitude. The thesis was preliminarily examined by Docent Marja-Liisa Kakkuri-Knuuttila and Prof. Juha Sihvola; both of them have also helped me later. Prof. Frans de Haas offered highly constructive criticism at the last stages of my thesis, and also acted as my opponent. I had the opportunity to carry out my work at the Department of Philosophy at the University of Helsinki, and I was supported financially by the Finnish Cultural Foundation, the Oskar Öflund foundation, and the University of Helsinki. I would like to thank once again all those who helped me during the process of completing my thesis. As for those who have helped me turn the thesis into a book, I would like to thank, first of all, Prof. Martha Nussbaum for inviting me to work as a visiting scholar at the University of Chicago, and for helping me during that period. I am also deeply grateful to Prof. Elizabeth Asmis and Prof. Rachel Barney for comments, criticism and encouragement. Of my colleagues, I wish first to thank PhD Håvard Løkke for generously commenting on previous versions of the manuscript. I am also grateful to Hallvard Fossheim, Prof. Monte Johnson, Prof. Taneli Kukkonen, PhD Pauliina Remes, PhD Mikko Yrjönsuuri and my fellow Galenites at the University of Chicago, for comments, discussions and support. My warm thanks to Prof. Mark Shackleton and Ranya Paasonen for polishing the English of my text, and to José Filipe da Silva, Vili Lähteenmäki and Mika Perälä for helping me proofread the manuscript. Between completing the dissertation and completing the book, I have been working at the History of Mind Unit at the University of Helsinki, financed by the Academy of Finland. I am grateful to all the leaders ix

x

ACKNOWLEDGEMENTS

and members of this project, as well as to the Department of Philosophy. The Academy of Finland also supported me financially during my stay in Chicago. Finally, I wish to thank all those who have simply been there for me. Helsinki 30.1.2006 Miira Tuominen

ABBREVIATIONS AND A NOTE ON THE TEXTS

The following list contains the abbreviations used in this book. An edition is included in this list if it forms the basis of the reference system I have used. For the works of Plato, Aristotle and Plotinus, the reader is instructed to consult editions in the Oxford Classical Texts series (OCT). If a specific edition is used and discussed, it will be referred to in the text or in the footnotes. 1. Collections of Fragments and Texts DK
H. Diels and W. Kranz (eds.), Die Fragmente der Vorsokratiker, 3 vols., 6th rev. ed. (Berlin: Weidmann 1951–1952). SVF
I. von Arnim (ed.), Stoicorum Veterum Fragmenta, 4 vols., (Leipzig: Teubner 1905 and 1924). LS
A. A. Long and D. Sedley (eds.), The Hellenistic Philosophers, 2 vols., (Cambridge: Cambridge University Press 1987). 2. Edition Series CAG
Commentaria in Aristotelem Graeca CLCAG
Corpus latinum commentariorum in Aristotelem Graecorum 3. Authors and Works Aët.
Aëtius, a reconstruction of his treatise (Placita) is found in H. Diels (ed.), Doxographi Graeci (Berlin: de Gruyter 1879). Alexander of Aphrodisias De Anima
Alexandrii Aphrodisiensis praeter commentaria scripta minora: De Anima liber cum Mantissa, ed. I. Bruns (Berlin: Reimer 1887). (CAG Suppl. II 1)

xi

xii

ABBREVIATIONS AND A NOTE ON THE TEXTS

De Mixt.
Alexandri Aphrodisiensis praeter commentaria scripta minora: quaestiones, de fato, de mixtione, ed. I. Bruns (Berlin: Reimer 1892). (CAG Suppl. II 2) in An. Pr.
Alexandri in Aristotelis Analyticorum Priorum librum I commentarium, ed. M. Wallies (Berlin: Reimer 1883). (CAG II 1) in Metaph.
Alexandri Aphrodisiensis in Aristotelis Metaphysica commentaria, ed. M. Hayduck (Berlin: Reimer 1891). (CAG I) in Top.
Alexandri Aphrodisiensis in Aristotelis Topicorum libros octos commentaria, ed. M. Wallies (Berlin: Reimer 1891). (CAG II 2) Arist.
Aristotle An. Post.
Analytica Posteriora An. Pr.
Analytica Priora Cat.
Categoriae DA
De Anima De Int.
De Interpretatione De Juvent.
De Juventute De Mem.
De Memoria De Som.
De Somno EN
Ethica Nicomachea GA = De Generatione Animalium HA
Historia Animalium Met.
Metaphysica Part. An.
De Partibus Animalium Phys.
Physica Rhet.
Rhetorica Soph. El.
Sophistici Elenchi Top.
Topica Cic.
Cicero Acad.
Academica Div.
De Divinatione Fin.
De Finibus Nat. deor.
De Natura Deorum Diog. Laert.
Diogenes Laertius, Lives of Eminent Philosophers Galen Plac. Hipp. et Plat.
De Placitis Hippocrates et Platonis Inst. log.
Institutio Logica Meth. med.
Methodo Medendi On Sects
/eo§ aßo2qewl rn‹p eåqacn,2lnip (On the Sects for Beginners), Claudii Galenii Pergamenii scripta minora vol. 3, ed. G. Helmreich (Leipzig: Teubner 1893).

ABBREVIATIONS AND A NOTE ON THE TEXTS

xiii

Lucr.
Lucretius Rer. nat.
De Rerum Natura Philoponus (?) in An. Post.
Ioannis Philoponi in Aristotelis Analytica Posteriora Commentaria cum anonymo in librum II, ed. M. Wallies (Berlin: Reimer 1909). (CAG XIII 3) in An. Pr.
Ioannis Philoponi in Aristotelis Analytica Priora commentaria, ed. M. Wallies (Berlin: Reimer 1905). (CAG XIII 2) in Phys.
Ioannis Philoponi in Aristotelis Physicorum libros tres priores commentaria, ed. H. Vitelli (Berlin: Reimer 1887). (CAG XVI) in De An.
Commentaire sur le De Anima d’Aristote, lat. traduction de Guillaume de Moerbeke, ed. G. Verbeke (Louvain: Publications universitaires de Louvain 1966). (CLCAG) Plato Ap.
Apology Prot.
Protagoras Rep.
Republic Tim.
Timaeus Plutarch Adv. Col.
Adversus Colotem Comm. not.
De communibus notitiis adversus Stoicos Sext. Emp.
Sextus Empiricus

Math.
Adversus Mathematicos Pyr.
PsooÍleini ∫/nrs/Íqeip (Outlines of Pyrrhonism) Simplicius in Phys.
Simplicii in Aristotelis Physicorum libros quattuor posteriores commentaria, ed. H. Diels (Berlin: Reimer 1895). (CAG XVI) Themistius in An. Post.
Themistii Analyticorum Posteriorum paraphrasis, ed. M. Wallies (Berlin: Reimer 1900). (CAG V 1) in De Anima
Themistii in libros Aristotelis De Anima paraphrasis, ed. R. Heinze (Berlin: Reimer 1890). (CAG V 3) in Phys.
Themistii in Aristotelis Physica paraphrasis, ed. H. Schenkl (Berlin: Reimer 1900). (CAG V 2)

INTRODUCTION

Every human effort aiming at improving, deepening or clarifying our conceptions of the world – at best to provide us with knowledge – involves some kind of starting point. One has to start from somewhere, end somewhere, and in the course of the inquiry, take something for granted. This means that an inquiry involves a distinction between two classes of statements: those for which the truth is questioned and those which are taken as accepted without further proof. The latter class includes the general principles of valid inference as well as the specific principles concerning the subject matter. In addition to these, we need some criteria which indicate that the inquiry is sufficient. All of these can be called starting points for knowledge in a broad sense. To talk about starting points or principles of any kind entails one thinking about a structure in which some components are prior to others. In the characterisation of starting points for knowledge just presented, the relevant priority is determined by whether or not a statement is accepted immediately in the context in which it occurs. The existence of starting points for knowledge is often established through a regress argument. Many philosophers, who discuss knowledge in different frameworks and whose general views on human knowledge differ a great deal, share the common conviction that knowledge claims cannot form an infinite structure. Basically this means that if we give reasons for the statement we claim to know, the chain of reasons must not be infinite. It is important to note that the mere fact that we consider the theme of starting points for knowledge does not necessarily commit us to any particular epistemology. It might easily come to mind that when we talk about the starting points for knowledge, we assume at the same time an epistemologically foundationalist framework. When the starting points are interpreted in a broad sense, as characterised above, this does not follow. The theme of 1

2

INTRODUCTION

starting points for knowledge is thus to be taken as being independent of any particular epistemological theory. THE TOPIC, SCOPE, AND AIM OF THIS BOOK The topic of this book is how ancient Greek and Roman philosophers1 treated the question of starting points for knowledge. In the Greek context, a term that was often used for such a starting point is archê (8.u3), which can be translated as ‘starting point’ or ‘principle’. 8.u3 is one of the central philosophical terms of Greek philosophy. It is also one of those terms that have several philosophically relevant meanings. Its most literal meaning is beginning or origin and it has political connotations of leading and ruling. The basic metaphor in connection with knowledge would be a leading or guiding principle from which other things follow. As such, the connotations of the word in the Greek context differ from contemporary metaphors in epistemology. In the contemporary context starting points for knowledge are often compared to the foundations of buildings. Such an idea of an underlying structure is not central in the connotations of the Greek 8.u3. I have often used the more flexible and more literal translation ‘starting point’ for 8.u3, because ‘principle’ typically refers to general truths or logical rules and these are propositional. In antiquity, however, we do find examples of starting points for knowledge which are not propositional in a straightforward sense. These include basic notions corresponding to natural kinds or to metaphysical structuring factors of reality. In the Neo-Platonic tradition we also find a form of immediate intellectual apprehension, which involves understanding a complex whole instantaneously. Such apprehension is not propositional either. There are also methodological reasons for formulating the topic loosely. As is well known to any scholar and student working on the history of philosophy, philosophical questions have not necessarily been formulated in quite the same terms and within the same conceptual framework in different periods. However, some crucial themes, such as basic questions concerning existence, the nature of good, and the nature and structure of human knowledge reappear in different periods even though the framework in which they are studied changes. It is inevitable that any reading of historical texts is influenced by the general intellectual climate and the more precise philosophical theories of 1

From this point onwards by the phrase ‘ancient philosophers’ I shall refer to Greek and Roman philosophers. The Asian philosophical tradition is outside of the scope of this study.

INTRODUCTION

3

the reader’s own time. The influence, however, comes in degrees. One rather obvious danger of anachronism is involved if the texts are expected to answer questions formulated completely in the framework of the reader’s time. This is very rarely done nowadays, and as such it is not a sufficient methodological principle in research into the history of philosophy to avoid doing so. Much more nuanced insight into how to combine systematic philosophical analysis with the aim at historical accuracy is needed. It seems to me that the insight into how we can do this is best described as a form of sensitivity. This is why I shall not attempt to nail down all the relevant considerations as a set of explicit methodological rules. In any case, I do follow the general strategy of formulating the topic of the book on a general level as a philosophical theme of starting points for knowledge. My task is to ask how philosophers in antiquity formulated questions related to this theme and how they answered them. In this book, the focus is on Aristotle’s theory. Both Plato as his predecessor and the later followers are treated in relation to his theory. One reason for doing so is that Aristotle provides a comprehensive and detailed systematic framework for the discussion of starting points for knowledge. However, this is not the only reason. It is a historical fact that the influence of Aristotle’s theory was incomparable. Also the commentators with Platonistic orientation infiltrated many basic assumptions from Aristotle. The scope of my study extends from Plato in the fourth century BC to Philoponus in the sixth century CE. To choose a scope this large serves the purpose of providing a synoptic view of certain key assumptions generally shared or rejected during that period. I believe that the more synoptic approach, which I have chosen, is valuable for scholars and students of ancient philosophy because in this way many central questions will be discussed in connection with each other. In addition, the book will provide material for scholars working on other periods of the history of Western philosophy. In fact, it seems to me that such diachronic co-operation between scholars is one of the most important things research into the history of philosophy can provide. Further, I hope that people studying contemporary philosophy will find fresh ways of thinking about, for instance, questions of self-evidence and direct knowledge. Finally, in modern discussions the expression ‘intuitive knowledge’ is used vaguely to refer to various kinds of immediate cognition. I shall show that the ancient theories can be useful in specifying the meaning of this common term. I have excluded the Pre-Socratic discussions of nature’s basic principles (8.ua4) and the later Neo-Platonist commentary tradition (Porphyry, Proclus, and Iamblichus). This is not because these authors would not be relevant for the current theme. However, there are philosophical and philological reasons

4

INTRODUCTION

for these exclusions. First, questions concerning the interpretation of the Pre-Socratic tradition are even more difficult than those concerning Plato, Aristotle and their followers. This is partly due to the fact that our evidence of the Pre-Socratics is so scarce. In addition, the literary form of the evidence we do have causes additional complications, and it would have required too much space and time to discuss these difficulties. Of the later NeoPlatonic tradition I have only included Plotinus, its founder, to see what direction Platonism was taking in late antiquity. I shall not enter the complex discussion concerning the metaphysics of the intellect prominent in the late Platonistic authors. I shall concentrate on those philosophers who think that starting points for knowledge do exist. In connection with the Hellenistic philosophers, however, I have included a short treatment of the basic forms of sceptical challenge of that time. It is an important development in that period that the sceptical challenge starts to guide the philosophical discussion of knowledge in a new way. The topics covered in this book will pertain to epistemological questions. However, they also involve questions we would classify as belonging to the philosophy of science and to a philosophically oriented psychological study of our cognitive capacities. Further, it is typical in antiquity in the framework of Plato, Aristotle and their followers that considerations of knowledge bring with them metaphysical assumptions. The nature of the discussion I am dealing with requires that borderlines between current specific fields of philosophy must be crossed. Much has been written about ancient philosophical discussion of knowledge. Almost every topic this study covers has been an object of lively scholarly attention for a long time. In addition, the Aristotelian framework for systematic inquiry into nature was utilised as the paradigm for science in some circles up to Descartes’ time, and Plato’s Theaetetus with its attempted definition of knowledge is still taken to provide the background for the discussion of a definition of knowledge. Thus, I am by no means making discoveries in an unknown territory. My main contribution to the existing literature on the topic lies in the explicit recognition that there was a powerful tradition in antiquity in which starting points for knowledge were discussed from two largely independent points of view: the point of view of argumentation and the point of view of psychology. In the context of argumentation, starting points are claims or statements that are used as premises. Within the psychological framework, the idea is not to analyse what kind of premises are used to enhance the credibility of a conclusion of an argument. Rather, the discussion concentrates on the question of how we come to have true cognition about external things. It

INTRODUCTION

5

is assumed that we have correct elements in our cognitive structure and such correct elements also function as starting points for knowledge. Both discussions of starting points involve subcategories. Starting points, understood as principles of argumentation, fall into three categories. Firstly, there are those starting points which we take for granted when we begin our inquiry into the nature and explanation of things. Secondly, in the course of inquiry, we may find general truths concerning reality, which function as starting points of explanations, or express the true nature of things. Those truths form the second class of starting points and they are not initially known. My discussion of starting points as premises concentrates mainly on these two classes of principles. In addition, all argumentation and inquiry presupposes some general logical principles that guide and regulate inferences. This class of principles was also recognised in antiquity, but it was assumed that general logical principles very rarely appear as explicit premises in arguments. It was presumed that such principles exist and that they are not proved. However, we do not have very much evidence about the question of how we come to possess them. From the psychological point of view, the starting points are divided into two: perceptual ones and intellectual ones. A basic assumption in the PlatonicAristotelian tradition is that we come to have valid general cognition about external things. Such cognition is acquired through a natural cognitive process in which perceptions are involved. This entails that a transition from perceived material into generalisations was not generally understood as an inferential process. As I said, the discussion of starting points as premises was largely independent of the discussion of elements of cognition. However, we do have some evidence of how the two were taken to be connected with each other. Basically, the connection comes from the idea that our knowledge claims are true, in the relevant sense of being true about things in the world, only if their elements have an appropriate link with the external things, and if the elements are correctly connected. The elements are assumed to have such a link if there is immediate intellectual apprehension of the elements. The apprehension is explained by psychological theories of human cognitive capacities. Behind such theories, there is the further metaphysical assumption that reality consists of immutable and discrete elements that have the intrinsic property of being intelligible. The ancient discussion, however, was not completely uniform. It is well known that there were also other influential threads in the ancient discussions of human knowledge. After my discussion of the Platonic-Aristotelian tradition, I shall turn to a brief treatment of the main Hellenistic schools, Epicureans, Stoics, and the Sceptics. In the Hellenistic discussion the question

6

INTRODUCTION

of whether we can attain the truth at all, becomes prominent. I shall show that it was originally a sceptical intuition to take this question in the sense of whether we are always capable of discerning true appearances from false ones. After the Hellenistic schools, I shall provide a discussion of the main sects of ancient medicine, namely the rationalists and the empiricists. I shall show that whereas the assumption that we can have immediate cognition of general truths is prominent in Plato and Aristotle, in the medical debate this assumption was challenged. THE STRUCTURE OF THE BOOK The book is divided into two parts. My main emphasis will be on the first part, on what can be called the ‘Platonic-Aristotelian tradition’. The philosophers I have counted in this tradition share some general assumptions which affect the discussion of starting points for knowledge. I shall now briefly present these assumptions and then proceed to describe the contents of the book. First of all, in the Platonic-Aristotelian framework it is thought that starting points for knowledge exist. This assumption is connected with a form of metaphysical realism. Reality is taken to have an intelligible structure with discrete elements called forms. The forms have necessary connections with each other and the structure involves relations of priority. Therefore, reality itself has an intrinsic order. Knowledge in the proper sense is taken to be knowledge about the structure of reality. Another important assumption in the Platonic-Aristotelian tradition is that there is an order in which human beings come to know things. The things we first come to know are secondary in the order of things. Nonetheless, through a natural cognitive process, we come to know some basic facts. To know these facts provides us with sufficient initial knowledge, which enables us to pose further questions about reality. Why are there eclipses of the moon? What kind of aquatic animals are dolphins? What is love? Is the human soul immortal? To pose these questions presupposes that we already know something. We know, at least in some sense of the word, that the moon has eclipses, that dolphins exist and that they are a kind of water animal, that love exists, and that the human species has a soul of a specific sort. It is also thought that all such questions finally lead to an answer which terminates the questions. For instance, when we have found out that the moon is eclipsed because the earth blocks the sun’s light and we understand that this is the way light is shadowed in general, we do not ask why there is an eclipse of the moon anymore. To take another example, when we have found all the properties necessary for dolphins to be dolphins, and when we understand that dolphins share many of those properties with other animals but only

INTRODUCTION

7

dolphins have all of them, we do not ask anymore what kind of animals dolphins are. We can start to ask how the properties identified as characteristically dolphin properties explain dolphins’ life and habits. It is also assumed that in this task there comes a point when we have found out the explanatory factors for the general peculiarities of dolphins’ lives. These examples enable us now to illustrate the general types of starting points for knowledge distinguished above. First we distinguished between the main types: premises of arguments and starting points as intellectual apprehension of elements of the structure of reality. Second, among the premises, we noticed the difference between premises as starting points of inquiry and premises as general explanatory truths found in the course of inquiry. The former are those starting from which we inquire into the nature of things; the latter ones are those towards which our inquiry proceeds. If we use dolphins as an example, we have intellectual apprehension on a general level that there are dolphins. This apprehension also involves grasping that dolphins are distinct from all the other kinds of things. In this way we have an accurate general element we call ‘dolphin’ in our cognitive structure. We have such elements, due to a natural cognitive process, which could also be called ‘concept formation’. However, the cognitive elements are not acquired in isolation. Rather, when we come to grasp that there are dolphins, we also grasp that they are animals, perhaps also that they are water animals. Such claims, then, function as the starting point of our inquiry. We know that there are dolphins and that they are water animals; we proceed to ask, ‘what are dolphins?’ In the end, we can come to find a comprehensive set of properties that are such that, taken together, they belong to dolphins alone. Such set provides a complete answer to our question about what dolphins are; they function as starting points in an explication of dolphin nature. Another basic type of questions, and this becomes explicit in Aristotle, is the question of explanation. To use moon and its eclipse as an example, we come to know through a natural cognitive process, first, that there is moon and, second, that it occasionally looses its light even though there are no clouds shadowing it. This fact raises in us the question: ‘why does the moon lose its light?’ Our initial knowledge concerning the fact that the moon loses its light enables us to initiate inquiry. In the course of our inquiry, we then find out that the loss of light from the moon is what we call ‘eclipse’. We might have to borrow some basic facts from astronomy which then enable us to understand that the moon has eclipses because the earth blocks sun’s light. The fact that the moon loses its light is a starting point we come to grasp through basic intellectual apprehension; we start our inquiry from that fact, and, eventually, find the reason: moon has eclipses because the earth blocks sun’s light. The claim that the loss of light is caused by the interposition of

8

INTRODUCTION

earth, functions as the starting point of the real explanation. An argument involving the real explanation is a proof in the proper sense; it starts from those truths that are found in inquiry. In the Platonic-Aristotelian tradition explanation is taken as a metaphysically based notion. The basic truths about reality explain the initially known facts, because the facts these truths express are real reasons or causes for the facts we initially know. The order of explanation corresponds to the intrinsic order of priority in reality. As noted above, in addition to these two classes of truths providing us with the starting point and an endpoint of inquiry, it is assumed that there are logical principles regulating all reasoning. The most important of such principles is that of non-contradiction. It is the basis of the argumentation techniques used in inquiry as well as of all the other logical principles. It is assumed that at least the principle of non-contradiction inheres in human rationality. In the Platonic-Aristotelian tradition perception is treated as a starting point for knowledge as well, but its role as a starting point is mainly instrumental. It is assumed that intellectual apprehension is possible only on the condition that we first become familiar with the perceptible aspects of reality through perception. It is granted that in a sense we can be said to have perceptual knowledge – the Aristotelians are more generous in this respect than the Platonists – but it is knowledge in a weaker sense of the word than knowledge of the intelligible structure. However, the explanation of how we attain the elements in our intellect requires an account of how we perceive the world. In the book, I shall first discuss the starting points as premises of argumentation, both in the sense of being starting points for inquiry and as general explanatory truths found in inquiry. I shall begin with Plato. In connection with Plato, my main aim is to show that many of the key assumptions articulated above, are found in his dialogues. At first, I shall discuss the general technique of argumentation known by the name ‘refutation’. The technique is built on the idea that we always have at least some true beliefs, and if we also have false beliefs the two subsets of our beliefs conflict. The conflicts can be brought to light with the refutational argument strategy. After refutation, I shall turn to consider Plato’s so-called method of hypothesis. In Plato’s discussion of the method of hypothesis, we find the assumption that reality itself has an intrinsic order; the order is taken to constitute explanatory relations finding of which is the aim of inquiry. The method of hypothesis involves the idea that we can ‘ascend’ from previously known truths to the explanatory ones. It is also assumed that eventually we are capable of finding an ultimate explanatory principle of the whole reality, which cannot be ascended. This is the form of good. Similar assumptions

INTRODUCTION

9

concerning explanatory directions are found in Plato’s discussion of collection and division. My discussion of Plato will provide the background for treating Aristotle. Aristotle formulates explicitly the idea that in argumentation, we need to distinguish between three classes of starting points. One is where our inquiry starts from; the other is the class of basic explanatory truths. Aristotle thinks that our inquiry can take the form of a dialectical argument or we can proceed empirically through intellectual generalisation. Which route is to be taken is determined partly by the nature of the topic, partly by the state of the existing research on the topic at issue. Aristotle also recognises that in accordance with the distinction between the two kinds of starting points, we must distinguish between two types of proofs. Another establishes more basic explanatory truths for us through the initially known facts; the other explains the initially known facts by starting from the basic explanatory truths. Only the latter provides proof in a strict Aristotelian sense. The unprovable premises of the proofs are unprovable only in the strict Aristotelian sense. Aristotle allows that they can be argued for even from true premises. Third class consists of logical principles that rarely appear as premises but form the basis of all argumentation. Then I shall move on to trace some later developments in the PlatonicAristotelian tradition. I have chosen as examples Galen, Alcinous and Plotinus from the ‘Platonist’ side and some of Aristotle’s late ancient commentators from the ‘Aristotelian’ side. I shall show that the basic assumption according to which we need to distinguish between starting points for knowledge in the sense of starting points for inquiry and basic explanatory truths expressing more primary facts, is preserved in the later Platonic-Aristotelian tradition in antiquity. After discussing different types of premises in the Platonic-Aristotelian tradition, I shall discuss starting points as elements of our intellectual structure. At the very beginning, I shall treat briefly the connection between the two discussions. My main claim is that even though the ancient discussion of premises is rather independent of the discussion concerning the elements, there are some common assumptions concerning the connection between them. Most importantly, it is assumed that in addition to discussing what kinds of premises can be accepted in different argumentation contexts, we also need an explanation of how we become familiar with the things in the world on a general level. The intuition behind this idea is that if we discussed premises and truth on a propositional level alone, we would not have explained the connection between the elements of reality and the elements of our statements. To explain intellectual apprehension of things on a general level, it is assumed, we also need a theory of perception. The ancient theories of perception discussed in this connection are realist in the sense that they assume

10

INTRODUCTION

that we are aware of the perceptible qualities of external things in an accurate manner. Almost all the theories take perception to involve physical contact between the object and the percipient; they vary with respect to the accounts of how this contact takes place. The theories also differ with respect to the question of whether or how perception is taken to contribute to intellectual apprehension. Plato assumes that the basic intelligible elements structuring the world need to be pre-existent in our minds and that they are not derived from experience. This is explained in the middle dialogues by recollection; in the later dialogues the theory of recollection is replaced by the theory according to which the basic constitutive elements of our soul are the same as the basic constitutive elements of reality, namely the very great kinds. Aristotle denies the assumption of the pre-existence of the elements of the intellectual structure in our minds. According to Aristotle, we come to grasp parts of the intelligible structure of reality, because the intelligible forms of the perceptible things come to be actualised in our intellect when a sufficient number of observations have been preserved in memory. Most of the later theories in the Platonic-Aristotelian tradition are versions of Plato’sandAristotle’stheories.IntheNeo-Platonists,however,wefindtheassumption that we must distinguish between ordinary conceptual thinking and instantaneous apprehension of the intelligible structuring factors of reality. The former is derived from perception; the latter is a separate intellectual occurrence, which is not part of the ordinary reasoning process but a kind of intellectual vision showing us a complex theoretical whole at one glance. After Aristotle, the framework involving assumptions of intelligible forms and an intrinsic order of reality was abandoned and the Platonic-Aristotelian tradition was interrupted. The assumption according to which knowledge is primarily about intelligible objects was also rejected. In the Hellenistic schools knowledge was taken, if granted possible at all, to be about the same objects as perception. Accordingly, perception was no longer seen as instrumental in attaining intelligible objects, but the question of the reliability of our perceptions came into the focus of the discussion. The rejection of the assumption that there are intelligible forms which make up a structured whole, brought along with it the rejection of the twoway model of starting points for knowledge. It continued to be thought by the Stoics and the Epicureans that there are evident facts through which we can establish and come to know other non-evident truths. However, it was no longer assumed that only an argument expressing those newly-found and explanatory truths would amount to a proof proper. By contrast, the arguments establishing non-evident truths of the Hellenistic period can be characterised as inferences towards an explanation, not from the explanation as the Aristotelian proofs.

INTRODUCTION

11

In the Hellenistic period, the notion of truth also gained new importance. In the Platonic-Aristotelian discussion the very possibility of knowledge is not in focus. By contrast, in the Hellenistic period it became a pressing question whether we can attain truth and have knowledge at all. This question was formulated as the question of whether or not a criterion of truth exists. In my discussion concerning the criterion of truth, I shall show that all the main schools, Epicureans, Stoics and Academic sceptics understood the notion of a criterion in different ways. For the Epicureans a criterion of truth is a kind of measuring stick providing us with a standard against which to judge the truth-value of our beliefs. The Stoics took the criterion of truth to be a special kind of truth-entailing appearance or notion. For instance, when I have that kind of appearance that there is a table in front of me, it is guaranteed to be the case that there is a table in front of me. The Academic sceptics, for their part, require that the criterion of truth has to provide us with a means to distinguish between true and false appearances in all possible situations. The Stoics claimed, rather, that the criterial role of appearances does not depend on discernibility in all possible situations. It was the sceptics who connected the notion of a criterion of truth with discernibility. The general treatment of the criterion will lead to the question whether the criteria of truth can be said to function as starting points for knowledge. I shall show that the Epicureans and the Stoics thought they can. Both these schools taught that the criteria of truth are evident and that through them we can establish previously unknown non-evident truths. The Epicureans conceived such transition from evident to non-evident as confirmation through evidence and disconfirmation through counter-evidence. The curious feature in the Epicurean way of understanding these notions is that the absence of counter-evidence is taken to provide us with reasons for accepting beliefs. This entails that multiple explanation is possible. I shall argue that the Epicureans in fact allowed that natural phenomena have multiple explanations; the same phenomenon can and actually does appear for different reasons. The Stoics claimed that we can prove a conclusion through premises by an argument in modus ponens form if three strict conditions are fulfilled. First, the premises of proofs need to be evident. Second, the conclusion must be non-evident at first but revealed by the premises. Third, the conditional premise involving the transition from evident to non-evident has to be a necessary one, i.e. that it does not allow falsity of the consequent when the antecedent is true. Such proofs, the Stoics thought, are rare but possible. The Academic sceptics thought that the Stoic candidate for a criterion of truth, a cognitive impression, cannot function as a criterion of truth, because it does not provide us with an absolutely unerring tool to discern between

12

INTRODUCTION

truth and falsity. However, some of them, at least Carneades, acknowledged that such clear and distinct appearances can be used as criteria for plausibility. The Pyrrhonian sceptics, most notably Sextus Empiricus, thought that a sceptic should not adopt too dogmatic an attitude towards appearances. From the Pyrrhonist perspective the Academic sceptics do so, because they claim that a criterion of truth does not exist. A Pyrrhonist, by contrast, will accept appearances in everyday life and have ordinary beliefs, but will suspend judgment as to whether or not any appearances are criteria of truth. One important context, where central questions concerning knowledge and inquiry were discussed in antiquity, is medicine. I shall not deal with the early Hippocratic origins of the discipline. Rather, I shall concentrate on a long-standing debate between two medical schools, the rationalists and the empiricists. We saw that in the Platonic-Aristotelian tradition it was assumed that it is possible for us to have true, immediate and general knowledge of things. This assumption is connected with assumptions concerning reality and the human cognitive capacities. Also the Hellenistic discussion involves the idea that because certain basic general notions called ‘preconceptions’ are criteria of truth, we do have accurate and general basic knowledge of what kinds of things there are. By contrast, in the ancient debate about the methodology of medicine, we do find discussions questioning the very possibility of arriving at accurate generalisation on the basis of observation. The rationalist doctors argue that because the empiricists think that medical knowledge is based solely on observation – and not on rational insight into the nature of things – accurate generalisation, and hence also medical knowledge, is impossible. However, I shall argue that the rationalist criticism against the ancient empiricists was in fact a little misplaced. A BRIEF SURVEY OF THE EXISTING LITERATURE As already noted, there is a great deal of literature on the topics discussed in my book. General studies of ancient epistemology are typically collections of articles by several authors. Such collections include, e.g. Companions to Ancient Thought I: Epistemology, edited by Stephen Everson (1990) and Language and Logos edited by Malcolm Schofield and Martha C. Nussbaum (1982). James Allen’s Inference from Signs: Ancient Debates about the Nature of Evidence (2001) is a chronologically wide-ranging monograph on ancient discussions on inferences based on evidence. A general collection of articles on Plato relevant for the present theme is found in Plato 1: Metaphysics and Epistemology edited by Gail Fine (1999). Plato’s method of hypothesis as well as the method of collection and division has also been subject to wide scholarly attention. A recent discussion of

INTRODUCTION

13

collection and division is found in Melissa Lane’s Method and Politics in Plato’s Statesman (1998). The theory of recollection is also one of the much discussed themes in connection with Plato. A provocative discussion is included in Dominic Scott’s Recollection and Experience: Plato’s Theory and Its Successors (1995). Theaetetus has of course greatly influenced philosophical theories of knowledge throughout the history of Western philosophy. This dialogue will not be in focus in the present study, however, because I will concentrate neither on discussing the definition of knowledge nor Plato’s attitude towards perceptual relativism. This notwithstanding, Myles Burnyeat’s monograph The Theaetetus of Plato (1990) should be mentioned. M. M. Lee’s recent book Epistemology after Protagoras (2005) includes a comprehensive discussion of Plato’s arguments against Protagoras’ perceptual relativism. Of Aristotle’s works the Posterior Analytics is highly relevant for the present study. Collections of articles on that treatise include, e.g., Articles on Aristotle I: Science edited by Jonathan Barnes, Malcolm Schofield and Richard Sorabji (1975) and Enrico Berti’s (ed.) Aristotle on Science (1981). Richard McKirahan’s Principles and Proofs (1992) is a fairly recent monograph on the Posterior Analytics. David Charles’ Aristotle on Meaning and Essence (2000) concentrates on Aristotle’s theory of definitions. Allan Gotthelf and James Lennox are well known for their work on Aristotle’s biology, such as Philosophical Issues in Aristotle’s Biology (1987). Lennox has also recently published a collection of his own articles, Aristotle’s Philosophy of Biology (2001). To G. E. L. Owen we owe many seminal contributions on ancient philosophy collected in Logic, Science, and Dialectic. In many studies on Aristotle’s Posterior Analytics, the relevance of the dialectical method presented in the Topics is recognised. A recent example is John Cleary’s Aristotle and Mathematics: Aporetic Method in Cosmology and Metaphysics (1995), which includes discussions on many topics in Aristotle relevant to this book. Studies on the Topics are mostly in article form by, e.g., Robert Bolton and Robin Smith. Marja-Liisa Kakkuri-Knuuttila’s dissertation (1993) deals with the Topics and she has kindly let me read her article on the dialectical argumentation technique which has come out during my writing process. Hellenistic epistemology has received increasing attention since the 1980s. Early collections include Doubt and Dogmatism (1980) edited by Schofield, Burnyeat, and Barnes, and Science and Speculation (1982) by the same editors, together with Jacques Brunschwig. Michael Frede’s and Gisela Striker’s main articles on Hellenistic philosophy are found in the collections Essays in Ancient Philosophy (1987) and Essays on Hellenistic Epistemology and Ethics (1996), respectively. For a collection of Hellenistic texts, translations, and comments, Anthony A. Long and David Sedley’s book Hellenistic Philosophers (1987), is indispensable.

14

INTRODUCTION

A general introduction to Neo-Platonism can be found in The Cambridge Companion to Plotinus (1996). Lloyd Gerson’s book Plotinus (1994) also helps to form a picture of Neo-Platonic thought. The easiest way to familiarise oneself with the ancient commentary tradition is Richard Sorabji’s recent and comprehensive sourcebook The Philosophy of the Commentators 200–600 AD in three volumes (2004). I shall mention two studies pertaining to the ancient discussion of human psychology. In Leen Spruit’s work Species Intelligibilis I (1994) the theme of the actualisation of intelligible form in the human soul is followed from antiquity through to the Middle Ages. Of more specific studies Eyjólfur Emilsson’s book Plotinus on Sense Perception: A Philosophical Study (1988), is useful for the theory of perception in Neo-Platonism. Bibliographies on topics related to my work are found in, e.g., Stephen Everson (ed.) Companions to Ancient Thought I: Epistemology (1989), in Long and Sedley’s Hellenistic Philosophers (1987), and in Lloyd P. Gerson (ed.) The Cambridge Companion to Plotinus (1996). Even though there is a great deal of literature on the topics I discuss here, there are few studies in which different ancient theories are compared, Spruit’s work being one exception. My study is the first monograph where the theme of starting points for knowledge is traced in antiquity over a period of almost a millennium.

PART I PLATONIC-ARISTOTELIAN TRADITION

CHAPTER ONE THEORIES OF ARGUMENTATION

1.1 PLATO It is notoriously difficult to write about Plato. The dialogue form, the eloquence of the text, and the fact that many of the dialogues end aporetically make it hard to assess whether Plato himself is behind the views presented. Further complications arise, for instance, on the basis of chronological questions and occasional discrepancies between different works, as well as due to the question which views possibly originate from the historical Socrates. In sum, it is not clear to what extent Plato is building systematic philosophical theories and, if he does, whether he himself advocates them or not. In my discussion on Plato, I shall concentrate on those themes that are central later in Aristotle, later Platonism and the commentary tradition. The most important points of view that became central in the later development of the Platonic-Aristotelian tradition are the following. Firstly, already in Plato we find the general distinction between two discussions about starting points for knowledge, one from the point of view of a theory of argumentation, another psychological. The psychological theories will be discussed in 2.3.1. The connection between the two in Plato will be at issue in 2.1. Basically, the psychological theories explain why and how it is possible to speak and argue truthfully about things in the world. Secondly, we also find the assumption that when talking about the starting points in the context of argumentation, we need to distinguish between the starting points from which we start our inquiry and those towards which our inquiry is directed. This assumption is connected with the metaphysical view according to which the intelligible structure of reality has an intrinsic order in which some things are prior to others. In inquiry we need to start from things that are familiar to us, but we must aim at finding the real orders of priority between things. 17

18

CHAPTER ONE

Plato divides the techniques of argumentation into two main groups. The first one is the refutational (or elenctic) technique ((kecunp), which is directed at an opponent and aims at improving his conceptions. The second group of argumentation techniques aims at finding the truth; in different dialogues Plato has different descriptions of these techniques. The metaphysical assumptions are important in the latter group of techniques but do not figure in the former.1 Arguments as Socratic Discussions I shall now turn to discuss the refutational conception of argumentation. Plato presents no systematic theory of argumentation, but he has a general conception of how it should proceed. This conception is exemplified by his early dialogues which are written more or less according to the conception of argumentation called refutation,2 adopted by the sophists and by Socrates.3 Refutational argumentation is conceived as a communicative process involving two participants, one who poses questions, and the other who answers (the interlocutor or opponent). The refutations start from a thesis accepted by the interlocutor. The questioner must ask questions that can be answered by a simple ‘yes’ or ‘no’; he aims at showing that the interlocutor in the discussion endorses statements that are incompatible with the thesis he4 at first accepted. The statements that the opponent accepts are used as premises of the argument. It is assumed that because one cannot hold conceptions that are shown to be mutually incompatible, the interlocutor should abandon the initial thesis. A Socratic refutation5 typically leads to an 8/n.4a, a situation in which the thesis at the beginning accepted by the opponent has been seen to be discordant with his other beliefs. The 8/n.4a shows that the set of statements that the interlocutor has accepted is contradictory, and hence he must admit that he does not know what he claimed to know. 1

Cf. Vlastos (1985). Of Plato’s early dialogues the Apology is an exception in the sense that it is not a refutational one, and the Crito, too, is a positive argument for the thesis that Socrates should not try to escape from prison. 3 It will not affect my discussion here to what extent the Socrates in Plato’s dialogues actually refers to the historical Socrates. 4 I have used the male pronoun here for the reason that all of Socrates’ interlocutors are male. Whether or not the interlocutor could also be female is another matter. 5 This title (
‘Socratic refutation’) for the technique is used, e.g., by Vlastos (1983) and Bolton (1993). Aristotle, by contrast, calls refutational argumentative strategy sophistical. For the historical background of Aristotle’s theory of sophistic refutations, see Dorion (1995); for a recent study of Aristotle’s Sophistical refutations, see Schreiber (2003). 2

THEORIES OF ARGUMENTATION

19

Now, it might seem that the refutational argumentative technique is sceptically motivated.6 The sceptics of the Hellenistic Academy in fact seem to think it is, and traces of scepticism are found in Middle Platonism as well.7 There is still scholarly debate over the question of what can be shown by a refutational argumentative strategy.8 However, it is important to note from the start that a refutational argument against a thesis is always an argument for the contradictory opposite of the very same thesis. In addition, at least in some dialogues elenctic arguments aim at showing that because of the incompatibilities within different statements that the interlocutor accepts, he should change his mind as to the question discussed.9 Quite often the Socratic arguments seem to contain the assumption that the premises the opponent accepts in the discussion are more acceptable10 than the opponent’s initial thesis. Hence, the opponent, rather than sticking to the thesis, should accept the premises and abandon the thesis. Therefore, the Socratic refutations differ from sceptical arguments which aim at showing that one has as strong grounds for accepting a conception as one has for accepting its negation and, consequently, that suspension of judgement should follow.11 The elenctic arguments of Plato’s early dialogues are based on the assumption that contradictions are a sign of the interlocutor having accepted both true and false statements. It is based on an important observation that a set of truths is never inconsistent. However, it is also important that it is not seen as a fatal threat for the refutational conception of argumentation that it is in principle possible that there is a consistent (and possibly even coherent) belief set which only contains false beliefs. It is assumed that in the course of the discussion 6

To claim that the Socratic arguments are sceptical would lead one to claim that when Socrates occasionally presents a view as one he has accepted (as, e.g., in Crito 49c) he does this without dogmatic commitment of any kind, and only reports how things seem to be to him. For the discussion whether Socrates’ claims of his own ignorance should be taken literally, see Vlastos (1985), Bolton (1993). 7 See, e.g., Tarrant (1985). 8 The dispute is sometimes formulated as whether or not the Socratic refutation has a constructive aspect. The question has recently been studied by Gonzalez (1998); for the controversy, see his notes 10–12 on page 21. 9 An instance of this is, e.g., the first book of The Republic. 10 For the acceptability of the premises of the Socratic refutations as analogous to what Aristotle calls reputable conceptions ((ldnmnl), see Bolton (1993). 11 For the arguments of equipollence (åqnqh2leia), see Bett (2000a, 208–209). For a short discussion of the differences between Socratic and sceptical arguments, see Kanayama (2000, 50–51) and Sedley (1983b, 9–10); cf. Vlastos (1983) (reprinted in Fine (ed.) 1999).

20

CHAPTER ONE

led by a skilful questioner, the interlocutor always accepts at least some true beliefs.12 In the discussion the false thesis of the opponent leads to contradiction, because he also has some true conceptions on the same subject. In addition to the elenctic arguments Socrates also uses inductions (see, e.g., Ap. 27b and Prot. 319b–c, 332c).13 For the Socratic use of induction it is characteristic that he uses it to obtain a general premise to be used in the argument. However, Socrates’ inductions are such that the instances to be listed are already general types such as builders or shipwrights, not singular instances of a more general kind (a particular builder) (e.g., Prot. 319b–c). The number of instances of types listed in Socratic inductive arguments is always fairly small: two to five, rarely more. The examples of induction in the dialogues involve the assumption that attaining truth on an abstract and general level – for instance whether builders are experts in building or shipwrights in the construction of ships – is not considered as problematic as such. Rather, it is thought that this level of abstraction is obtained rather automatically by human reason. Inductive argumentation need not start from particular instances (e.g., whether this man is a builder and whether he is an expert in building and so forth), but uses general types as its starting point. In the background of this practice we can find the assumption that human reason, if it functions normally, can grasp these kinds of generalisations directly. Arguments which are relevant for our quest for knowledge start from the abstract and general level, i.e. from species or types. Therefore, the modern problem of induction, namely, how one can justify universal generalisations in the first place by enumerating particular instances and their properties,14 is not at stake in the Platonic examples of inductive argumentation. The general truths of the relevant kind, such as whether builders are experts in building, are not taken to be grasped directly by reason because of being analytical. It is not assumed that we can take it as immediately accepted that builders are experts of building merely because of what the relevant words mean. Rather, the truth of the relevant propositions is dependent on whether builders really are experts of building or not.

12

The assumption has been pointed out by Vlastos (1983) (in Fine (ed.) 1999, 59). For a discussion of Socrates’ inductive arguments, see R. Robinson (1941, 35–50). Aristotle in fact says (Met. XIII 4, 1078b28–30) that Socrates is to be praised for introducing two things concerning the principles of knowledge (/e.§ 8.u¢l %/iqr3,gp), namely inductive arguments and definitions. 14 For a distinction between ‘intuitive induction’ and complete enumeration of instances in Plato’s views on argumentation, see Robinson (1941, 37–40). 13

THEORIES OF ARGUMENTATION

21

In the early dialogues Plato makes a contrast between Socratic elenctic arguments and the sophistic or eristic ones. However, sometimes it seems that in spite of the criticism towards sophistic arguments in the dialogues, Plato makes Socrates use fallacies of the sophistic or eristic kind in order to produce an 8/n.4a. It has been a matter of dispute among scholars whether the Socrates of the dialogues does so.15 The answer to this question depends mainly on how the arguments in the dialogues are reconstructed. Many of Socrates’ arguments claimed to be sophistic fallacies are alleged fallacies based on ambiguity. However, it is often the case that if we articulate some tacit assumptions Socrates is making, we can absolve him from the accusation that his arguments are fallacies. For instance, in the Gorgias there is an argument containing the terms e√rsu4a and e√/.am4a,16 which might be taken to be fallacious because it substitutes a misleadingly similar name e√rsu4a for the original term e√/.am4a. However, the same argument can be analysed as a valid one, if e√/.am4a is assumed to follow necessarily from e√rsu4a. Articulating this assumption as a premise would make the argument valid. One possibility also is that the Socrates of the dialogues is using fallacies in order to illustrate certain philosophical points.17 Therefore, it is possible that sometimes the use of fallacies is intentional and entails that the character of Socrates is fully conscious of the fallacious nature of the argument, but tries to provoke the interlocutor into thinking for himself. The refutational argument form is based on the principle of non-contradiction. It is assumed that because it is impossible to hold two contradictory beliefs at the same time, one of them must be abandoned. As suggested above, this is typically the initial thesis accepted by the interlocutor. This is because, in order to produce a good argument, the questioner aims at making the opponent accept claims that are more plausible than the opponent’s thesis. Therefore, rather than denying a great many plausible claims the interlocutor accepts in the discourse, he should accept that it is the initial thesis which causes the contradiction, not these plausible claims. It is noteworthy that Plato assumes that even the eristic fallacies (see Euthydemus 293b–c), are based on the principle of non-contradiction (see also Republic 437aff.).18 He does not explicitly discuss the possibility of denying the principle. Consequently, he is not led into the awkward position Aristotle finds himself in Metaphysics IV. There Aristotle is on the one hand producing

15 16 17 18

Sprague (1962), e.g., treats this question explicitly. This argument is discussed, e.g., by Gonzalez (1998). E.g., Sprague (1962) argues for this; for the discussion, see also Klosko (1983). This has been pointed out by Sprague (1962).

22

CHAPTER ONE

arguments in favour of the principle of non-contradiction even though he on the other hand asserts that all meaningful speech and argumentation is based on that principle. The Method of Hypothesis In the introduction we made a distinction between two kinds of principles used in argumentation. First type of principles consists of explicit premises of arguments. The second one includes logical principles that regulate argumentative activity but rarely appear as explicit premises. The explicit premises have content, and they divide into two subtypes. The first of these can be called starting points for us. Such principles function as starting points for inquiry and they are initially known to us. The other subtype of principles with actual content is a class of objective starting points. They are found in inquiry and function as the starting point of explanations expressing relations of priority according to the nature of things. Up to this point we have been discussing the starting points of refutation directed towards an opponent and the principle of non-contradiction among the logical starting points. We have not yet said anything about objective starting points. I shall now turn to those techniques Plato introduces that are relevant for the distinction between starting points for us and objective starting points. First, I shall discuss the method of hypothesis which is connected to the philosophical dialectic of the Republic. After that I shall deal with the method of collection and division. I shall not concentrate on analysing the connection between these two methods. The so-called method19 of hypothesis is discussed in several places (see Meno 86e4–87b2, Phaedo 99c5–d1; 99e5–100a7 and 101d5–e1 and the Republic 510b–511d; cf. also Phaedrus 245c–e). In the Meno Plato calls it ‘investigating from a hypothesis’ (%m ∫/nh2qewp qjn/e‹qhai, 86a4); in the Phaedo the peculiar name ‘second sailing’ (de6re.np /knflp, 99c9–d1) is used. The most Plato says about the so-called method of hypothesis is found in the Phaedo. The relevant passages are even characterised as ‘the obscurest parts of the dialogue (and perhaps even of Plato).’20 No common opinion is gained among scholars over the exact nature of the method.21 In fact it seems a little 19

We still lack a systematic monograph on the question of method in antiquity which would also trace the uses of the word ,2hndnp. Gentzler (1998) is a collection of articles which illuminate some aspects of the ancient discussions on method, but none of the articles is synoptic and comprehensive. 20 Rowe (1993, 50). 21 For discussions, see, e.g., Kanayama (2000), Gonzalez (1998, especially 188–208), Rowe (1993), Tait (1986), Vlastos (1988), Bluck (1957) and Robinson (1941).

THEORIES OF ARGUMENTATION

23

exaggerated to call the procedure introduced in the Phaedo a method in the proper sense. I shall, however, use this somewhat established description of the procedure. There are two passages in the Phaedo which describe the hypothetical method. The first of them is as follows: [T]aking as my hypothesis (∫/nh2,elnp) in each case the account (k5cnp) I judge to be the strongest (%..w,el2qrarnp), I would consider as true the ones that are in accordance (qs,twle‹l) with it and as false the ones that are not (diatwl4a) (100a3–7; transl. modified from Grube in Cooper (ed.) 1997).

Typically the discussion of the method of hypothesis has centred on the notions of accordance (qs,twl4a) and discordance (diatwl4a). Less attention is paid to the strength of the hypothesis. This, I believe, has led to some problems. In my discussion here, I shall also bring in the notion of strength and show how it can be used to solve some of those problems. Basically, it is vital to notice that the method of hypothesis is not introduced for the purpose of discriminating the truth of any chance statements whatsoever. Rather, the method is meant for cases where one is building a theory or a belief set describing a certain subject matter, and one is only interested in the truth of competing theories or explanatory models on the same question. Therefore, some kind of criterion of relevance has to be assumed in the background of the method. The existing suggestions do not, I think, take this point seriously enough. Let me start with discussing the scholarly literature on the notions of accordance and discordance. Basically, four different interpretations can be distinguished. It is most convenient to take the first two readings first. According to the first reading, accordance and discordance refer to logical consistency and inconsistency respectively.22 According to the second, qs,twl4a should be understood as logical entailment and diatwl4a as non-entailment. Both readings assume that the two notions qs,twl4a and diatwl4a are exhaustive and mutually exclusive. Every claim is either in accordance or in discordance with a given claim they are compared to, and two claims can never be both in accordance and in discordance with each other. Now, according to the first reading the method of hypothesis would recommend us to take every claim consistent with our hypothesis to be true. Such statements might very well include statements that we otherwise have strong reason to believe false.23 Therefore, the first reading seems to be too strong. It is clear that we cannot take all the claims which are merely consistent 22

E.g., Kanayama (2000, 62–68). Gentzler also makes this point (1991, 266); cf. Rowe (1996, 233). Robinson (1941) takes qs,twl4a to mean logical consistency; this leads him to condemn the whole method as worthless. 23

24

CHAPTER ONE

with our hypothesis to be true. For instance, if my hypothesis is that all beautiful things are beautiful by virtue of them taking part in beauty itself (some such version of the theory of forms is an example of the hypothesis in the Phaedo), it can be taken to be consistent with this hypothesis that there are small green men living under my bed. But it does not seem plausible that I could take it to be true that there are small green men living under my bed just because it happens to be consistent with my hypothesis of the explanatory power of the Platonic forms. However, even though consistency does not seem to be a sufficient criterion to judge a claim true, inconsistency seems sufficient for falsity. If p is judged true, and if q added to an otherwise consistent belief set including p makes the belief set inconsistent, q can be taken to be false. In the example, if my hypothesis is that all beautiful things are beautiful because they take part in beauty itself, the claim that they are beautiful not because they take part in beauty but because they are coloured in a certain way could be taken as false, because it is not consistent with my hypothesis. Therefore, consistency can be taken to be a necessary, but not sufficient condition for qs,twl4a. How about logical entailment, then? If we take qs,twl4a to mean logical entailment, the method suggests that if a statement is entailed by the hypothesis, we should take that statement to be true. There is no prima facie reason to believe that the method of hypothesis should not be understood in this way. However, again we can ask whether this is a sufficient condition for qs,twl4a. The condition implies that, for instance, we should take infinitely long disjunctions, where the first disjunct is identical with our hypothesis and the other disjuncts are any chance claims, to be true because they are entailed by the hypothesis. Plato might well agree that these disjunctions are true – even though he might be sceptical about the truth of infinitely long disjunctions – but it is another question whether they are in any sense relevant. We have now seen that both the first and the second reading involve problems. According to the third reading, the two notions do not have a univocal meaning throughout the relevant passages, but switch their meaning.24 The relevant passage is very short and, I believe, the third reading should be accepted only if all other possibilities turn out to be highly implausible or if it helps to solve some otherwise difficult problems. However, neither of these conditions is fulfilled with respect to the Phaedo and the method of hypothesis, and this reading should be rejected. The fourth suggestion concerning the reading of qs,twl4a and diatwl4a is based on the important point made, e.g., by Gentzler, that accordance and

24

Robinson (1941, 129–131) and Bostock (1986, 169).

THEORIES OF ARGUMENTATION

25

discordance need not be taken as contradictory opposites, but they can be contraries. Therefore, the two notions need not be taken in the sense that they should exhaust all possibilities. Rather, it is suggested that some claims can be such that they are neither in accordance nor in discordance with each other. Gentzler suggests that accordance should be taken to mean ‘coherence’ and discordance ‘inconsistency’.25 I agree with Gentzler that coherence – albeit a problematic notion –provides us with a reasonable initial clue for reading the notion of accordance.26 However, I would like to take a closer look at her articulation of this idea. Gentzler has two formulations, one for coherence and another for qs,twl4a, but she does not explain the relation between these two formulations. The condition for coherence is that: P coheres with Q if and only if P is consistent with Q and stands in either a suitable inductive or deductive inferential relation to Q.

The condition for qs,twl4a is: P qs,twle‹ with Q if and only if Q gives us some reason to believe that P is true.27

Gentzler leaves it open what the relation between the conditions for coherence and qs,twl4a is supposed to be. It is perhaps a little confusing that in the articulation of the two conditions the symbols P and Q take the opposite roles. P is said to stand in a suitable inductive or deductive relation to Q, whereas the condition for accordance is formulated the other way round; the two are in accordance (qs,twle‹l) if Q gives us some reason to believe that P. Later in the article (p. 269), however, it becomes clear that the phrase for P ‘to stand in either a suitable inductive or deductive inferential relation to Q’ is taken to be a formulation for the condition ‘to give some reason to believe that’. Thus, Gentzler’s suggestion is that P is coherent and therefore accords with Q if and only if P is in a suitable inductive or deductive relation to Q, and in such a case P gives good reason to believe that Q. Gentzler herself acknowledges that even this formulation of the condition for accordance faces the problem that the method seems to give us reason to reject propositions on insufficient grounds.28 Her example is the proposition that the charge of a proton is spread over a distance of 1015 metres. This, she says, is compatible with her beliefs, but it does not stand in an inductive or deductive inferential relation to her other beliefs and, hence, should be rejected by the suggested criterion. 25 26 27 28

Gentzler (1991, 270). Cf. Gonzalez (1998, 197), who talks about situating the hypothesis within a context. Gentzler (1991, 268–269). Ibid.

26

CHAPTER ONE

Further, according to the condition thus explicated, we are entitled to hold that P is true if our hypothesis Q is in a suitable inductive or deductive relation to P. If a sheer deductive relation is enough for a ‘suitable inductive or deductive relation’, then every statement that follows logically from our hypothesis must be considered true. This, however, brings us back to the second suggestion and the question of whether Plato’s method recommends us to consider, for instance, infinitely long disjunctions in which our hypothesis is the first disjunct to be true. As mentioned, I think that Plato might be sceptical of infinitely long disjunctions, but otherwise there seems to be no reason for him to deny that similar very long but finite disjunctions could be true. However, I do not think that the method of hypothesis is quite intended to cover such cases. Rather, limitations have to be posited with respect to the content of the statements to be investigated by the method of hypothesis. Later in her article, when discussing the second phase of the method of hypothesis, Gentzler acknowledges that Plato might not have assumed ‘a clearly defined decision procedure for restricting them [i.e. the results of the hypothesis] to a manageable number; rather, he took it for granted that his audience would have an intuitive, if rough, idea of the difference between the logical consequences that were worth considering and those that were not.’29 As I noted, a restriction of this kind seems necessary already in the first phase of the method for the reasons just discussed. Let us now go back to the text of the Phaedo once more and see if we can find yet another specification for qs,twl4a which could avoid the problems that the previously discussed suggestions face. The hypothesis which Plato is talking about in the relevant passage (Phaedo 100a3–7, quoted above) is described as the one ‘judged to be strongest’ (%..w,el2qrarnp, 100a3). I suggest that the most natural reading for the strength of the hypothesis and for its accordance with other propositions is its explanatory power. This in fact helps us understand the condition of qs,twl4a. A proposition P should be judged to be true on the basis of a hypothesis Q if and only if Q has explanatory power over P. If this suggestion is correct, the Greek %..w,el2qrarnp could be rendered ‘strongest’ (‘most powerful’ could also do)30 and it should be taken to refer to the hypothesis’ superior explanatory power. (I would like to remind the reader that I have agreed above with reading diatwl4a as ‘inconsistency’.)

29 30

Gentzler (1991, 271). ‘Healthiest’ has also been suggested; see Gregory (2000, 90).

THEORIES OF ARGUMENTATION

27

The connection between the method and search for explanations is quite clear in the context. The word ‘reason’ or ‘cause’ (aår4a)31 is used to specify what the method helps us find, and the description is preceded by Socrates’ famous account of how he was disappointed in his youth with the explanations given by natural philosophers.32 In addition, the theory of forms – presented as an example of the kind of hypothesis the method can investigate (100b1–9 and c3–7) – is taken to be explanatory of claims concerning perceptible things.33 Taking part in forms such as that of equality explains some properties of perceptible things (e.g., that this stick here is equal to that stick over there). Explanatory power also gives us a criterion for choosing the statements the truth of which one is to evaluate by the method. We do not need to take into consideration, e.g., infinite disjunctions including the hypothesis, because a good explanation should be specific concerning the question of which claim one is explaining. If we take the decisive criterion to be explanatory power, the first phase of the method provides us with a way of constructing a comprehensive and coherent set of statements involving explanatory relations. A set of statements of this kind can be called a theory. It includes the hypothesis which is the strongest in the sense that it gives the best explanation for the explananda, 31

The term aår4a is a very difficult one to translate; see Vlastos (1969) and Rowe (1993, 49). 32 Cf. Rowe (1993, 49), who suggests that what Socrates offers us in the relevant passages of the Phaedo is at its best ‘an answer to the question “why is x F?” or “why does x become F?”.’ 33 There is no general agreement among scholars about whether the presentation of the theory of forms does exemplify the method. A clear minority votes for the denial of this suggestion; cf. however, Tait (1986), who denies that the lines are an example of the use of the hypothetical method. For the adherents of the view that the lines provide us with an example of the method, see, e.g., Gallop (1975), Rowe (1993) and Kanayama (2000). Another disputed question is whether what Plato says in these lines amounts to a theory of forms or is it only a statement of the existence of forms. For the reading that the hypothesis amounts to a theory, see Gallop (1975) and Rowe (1993); they take the theory to consist of the statement of the existence of forms and of an aår4a proposition, i.e., one saying that other things which participate in forms come to have their names from them. For the reading that the hypothesis only consists of the statement of existence, see van Eck (1996); for the discussion, see Kanayama (2000, 53–62), who suggests that the hypothesis is only the aår4a proposition. To me it seems quite natural to take the hypothesis to refer to the existence of forms and to assume that the latter statement follows from the hypothesis, or that it is a further specification of what it means that there are forms.

28

CHAPTER ONE

plus the explananda itself. Because of the condition of diatwl4a, the set does not contain any statements that are inconsistent with the hypothesis. The theory can be evaluated on the basis of whether the hypothesis explains well the explananda (i.e., whether it is the strongest hypothesis), and whether the explananda themselves are credible.34 If the hypothesis makes us believe propositions we otherwise have strong reasons to disbelieve, this fact would undermine the credibility of the hypothesis as well. On the basis of what I have said above about the method of hypothesis, it might seem that Plato can be inferred to entertain a coherence theory of truth.35 However, it does not follow by necessity that Plato should be characterised as a coherence theorist of truth. Rather, I take it that some kind of coherence only functions as an indication of the truth value of the belief set. On the one hand, if someone has accepted a contradictory belief set, it must contain false beliefs. On the other hand, large and coherent belief sets are not possible if the beliefs are false. This does not entail that the truth of the beliefs should consist of coherence. I shall now summarise the most important aspects of the discussion above. In the first description of the method of hypothesis, Plato introduces an argumentative structure within which a statement, namely the hypothesis, is postulated as a starting point, i.e., a principle prior to the other statements in the structure. The other statements in the structure are judged to be true if they can be explained by the hypothesis; they are judged false if inconsistent with the hypothesis. Even though the first phase of the method is especially designed for evaluating the truth of the claims that are explained by the hypothesis, the truth of the hypothesis itself can be evaluated in the same

34

Some scholars have also suggested that the use of the method of hypothesis should be restricted to investigating the so-called theory of forms; see Gallop (1975) and Rowe (1993). However, such restriction seems unnecessary. In the Phaedo Plato says that the statements judged to be strongest on the basis of accordance with the hypothesis can be concerned with reasons or with all other things (/e.§ aår4ap ja§ /e.§ r‡l !kkwl 9/1lrwl). I take it to mean that even though the example of the method of hypothesis is concerned with the theory of forms, which can be taken as explanatory of perceptible things, this does not entail that the method could only be used in the case of building and evaluating the theory of forms. Further arguments against the restriction are found in Kanayama (2000, 58). In general Plato might sometimes be taken to assume that in fact all questions do derive their explanation from forms (see Phaedo 96ff.), but this assumption does not arise from the method. Such passages, however, are based on Plato’s metaphysical views, not on the character of the method of hypothesis. 35 See, e.g., Gregory (2000, 88).

THEORIES OF ARGUMENTATION

29

procedure on the basis of how plausible are the claims it can explain and on the basis of how well it explains them. The second passage concerning the method of hypothesis in the Phaedo concentrates on the question of how the hypothesis itself can be supported: If someone then attacked your hypothesis itself, you would let him be and not answer until you had examined whether the consequences that follow from it (r¡ %/’ %je‹la ….,gh2lra) agree or disagree (qs,twle‹ z diatwle‹) among themselves. And when you must give an account (k5cnl did5lai) of the hypothesis itself you will proceed in the same way: hypothetising another hypothesis, the one that seems to you best of those above (!lwhel) until you come to something sufficient (ri ßjal5l). (Phaedo, 101d3–8; transl. modified from Grube in Cooper (ed.) 1997.)

One way to support the hypothesis is to check whether what follows from the hypothesis includes mutually incompatible statements.36 If it does, then the hypothesis should be abandoned. If no such problems follow, the hypothesis can be left to stand. However, this is not the best confirmation the hypothesis can receive. Further confirmation for the hypothesis can be sought ‘from above’ (!lwhel). The hypotheses that are above possibly mean ones which have to be true if the hypothesis itself is to be true. Plato’s expression ‘from above’ (!lwhel) suggests that he assumes that there are ‘up and down’ directions in the context of explanation. What such directions mean, however, is not explained. In the Phaedo Plato talks about proceeding upwards until something sufficient is reached. It is not quite clear what we can infer on the basis of the meaning of the phrase ‘something sufficient’ (ri ßjal5l, 101e1) here. The expression as such suggests that the starting point we find is sufficient for the purposes of that particular inquiry. It leaves open the possibility that the adequate starting point can be traced further up. In the Republic we can find a somewhat similar passage concerned with philosophical dialectic. The connection between the dialectic of the Republic and the method of hypothesis is not made explicit, but the discussions do

36

On how compatibility or incompatibility between what follows from the hypothesis can be taken to confirm or disconfirm the hypothesis itself, see Kanayama (2000, 76–80). Gentzler (1991, 271–272) points out that in order to find out whether the hypothesis has inconsistent results we need not examine the consistency of all the results taken individually. Rather, if we wish to examine the truth of a result P it is sufficient that we examine whether it is consistent with the conjunction of the other results (Q&R&S) of the hypothesis.

30

CHAPTER ONE

resemble each other. Scholars often take the two to be concerned with the same thing.37 In the Republic the principle above is called ‘a non-hypothetical principle’ (8ls/5hernp 8.u3, 511b). Therefore, by using the method of hypothesis we can push our explanations further and further upwards, but unless we reach the ‘non-hypothetical principles’, there are always additional principles from which our previous hypotheses could be derived. Only when we reach the principle that is non-hypothetical have we reached the point where we cannot go on upwards to more and more primary truths which could explain our hypothesis. As I pointed out above, the first description of the method of hypothesis provides us with a structure of argumentation where the hypothesis functions as an unquestionable starting point on the basis of which other statements are judged true or false. Such a structure has directions in the sense that it involves an asymmetric and transitive relation. The hypothesis stands in an explanatory relation to the explananda but they do not stand in the same relation to the hypothesis. Further, explanatory power is transitive: if a is explained by the hypothesis and b is explained by a, then b is also explained by the hypothesis. The second description of the method of hypothesis together with the passage from the Republic adds to this picture an objective end of the process of postulating ascending hypotheses. It is assumed that the hypothesis – in addition to being confirmed on the basis of what follows from it – can be seen to require the truth of some truths that are above it. However, this cannot go on forever, but is only continued until one arrives at a nonhypothetical principle. The direction of the structure manifests itself by the explanatory power the truths have over the hypothesis, or over other statements in the structure. The explanatory power is not only an epistemological notion that is dependent on the human perspective. Rather, explanatory power is based on relations between things in reality. In an ideal argument this direction is preserved in the following sense: the reasons are expressed in the premises of an argument, and what they are reasons for figure in the conclusion (see Phaedo 101e1–6). 37

See, e.g., Rowe (1993, 52–53), Gallop (1975), Dillon (1993, 75) and Bluck (1957); cf., however, Tait (1986), who thinks that the phase of arriving at a non-hypothetical principle does not belong to the method of hypothesis, and Robinson (1941), who thinks that being sufficient means the same as being sufficient to convince the interlocutor in question; also Gonzalez (1998, 198–199) takes ‘being sufficient’ here to mean a hypothesis that is acceptable to the interlocutor. This, however, seems to be an insufficient criterion for the kind of sufficiency Plato is after.

THEORIES OF ARGUMENTATION

31

What, then, would a non-hypothetical principle be like? I think we go in the wrong direction if we assume it to be a giant cosmologic-philosophical principle from which we can deduce all the relevant truths of the world. A more likely candidate is found from the Republic, namely the form of the good. In the Phaedo, Socrates expresses his disappointment with the natural philosophers’ explanations which refer to the material constituents of things; he is also disappointed with the use Anaxagoras made of the intellect (lnflp). In the example (98b–99a), Socrates tries to explain the fact that he is sitting in prison by reference to his sinews and bones, and exclaims that if it were only for them, he would have left long ago for Megara – or somewhere else. He says that a reference to the parts of his body cannot explain his sitting in the prison. Rather, the explanation lies in the following facts: firstly, the Athenians had found it better to condemn rather than free him, and, secondly, he himself thinks it is better that he sits there and does not attempt to escape. These facts constitute a proper explanation because they refer to something good. In the Republic the form of the good functions as an ordering principle of forms themselves, and it is indicated that knowing the good is necessary for knowing explanatory relations (e.g., book VII 516c, where the sun is compared to the good). Through being an ordering principle of forms its existence accounts for the fact that the perceptible world is also well ordered.38 I have examined in some detail the so-called method of hypothesis in the Phaedo and similar passages in the Republic. These passages have made clear the assumption that when looking for principles in argumentation, mere logical criteria are not sufficient. Rather, we have to consider the content of the statements as well, and see whether they are connected in important ways. We also have to pay attention to whether principles we have found explain claims that we otherwise have strong reason to believe to be true. In addition, it is possible to find an objective starting point, i.e., such a principle that in reality has priority over all other things. In the context of the Republic the good is presented as such. I mentioned above that the Platonic method of hypothesis seems so unsystematic that it might be an exaggeration to call it a method. Does Plato have any other suggestions for a systematic procedure of inquiry?

38

See Fine (1989) (in Fine (ed.) 1999, 228–229) and Gregory (2000); cf. Annas (1982, 104), who claims that ultimate basic truths in the Republic concern the nature of goodness. It has, however, been suggested (Ross 1951, 243) that in fact unity would be the highest principle of ideas, and that goodness follows from it.

32

CHAPTER ONE

Collection and Division In Plato’s dialogues, the question of knowledge is often discussed in terms of the question of knowing what things are. In some of the dialogues, namely the Phaedrus, the Sophist and the Politicus, a general procedure is introduced to discover answers to that question. This procedure consists of two aspects, collection and division (dia4.eqip/qslacwc3), and it has been characterised as articulated generalisation.39 The outcome of a successful collection and division is a full definition which expresses the true nature of a thing. The central passages are the Phaedrus 265a–266b, the Sophist 219a–237a, and the Politicus 258b–267c. The method of collection and division is important to our topic, since it involves the assumption that the very nature of things is intrinsically ordered. Such an order is revealed in the process of collection and division which is a two-way method. First, when collecting, we ascend towards what is more general, and then come down by dividing the general into its specific elements. The general idea behind collecting and dividing is simple. When we are investigating into a phenomenon we first have to place it under a more general phenomenon under which it belongs. This part of the procedure is called ‘collecting’. Division consists in distinguishing the type under consideration form other subtypes of the more general phenomenon. I shall use the discussion of the nature of love in the Phaedrus as an example here. The collecting part is short: love is madness. This is an assumption shared by the two speeches of Socrates, the one he gives with his head covered (237b–238c, 238e–241d) and the reverse speech (/akil–d4a) given to reconcile the errors of the first speech (244aff.). Socrates makes clear that the first speech is erroneous in the following respect. In placing love under the genus of madness, it places ‘all mental derangements into one common kind’ and takes them to be ‘by nature one single kind within us’ (266a). Socrates notices that the flaws of the first speech – which make love an inappropriate desire of a sick man – are due to the fact that, contrary to what the first speech assumed, there are two kinds of madness. In addition to the madness conceived as a mental illness of human beings, there is another kind of madness, one which is due to divine inspiration. Distinguishing these two types of madness from each other belongs to the phase called ‘division’. The divine type of madness is further divided into four subtypes: inspiration of the prophet by Apollo, that of a mystic by Dionysus, that of a poet by the Muses and, finally, that of love by Aphrodite (265b–c). The madness inspired by Aphrodite, i.e. love, is said to be the best of all these. 39

The characterisation is from Ross (1951, 81), who takes generalisation to be the innovation of the method.

THEORIES OF ARGUMENTATION

33

Socrates also points out that in order to produce an accurate collection combined with division, one has to distinguish the classes and subclasses according to natural similarities and differences between things. He compares this with the work of a skilful butcher who knows how to cut according to natural joints (265e). When collection and division is performed in this way, the collection of subspecies cannot yield contradictions. Someone might infer from the above description that collection and division is just a method of producing analytical truths based on conceptual considerations. However, this is not the case. As Socrates’ remark on cutting according to natural joints makes clear, it is assumed that the collections and divisions are not analytic in this sense. Metaphysically speaking, the concepts are imitations of the forms,40 and the members of the subtypes of natural kinds are such that they take part in the same form. Accordingly, the distinctions are taken to apply within the realm of intelligible forms, and they do have consequences with respect to the perceptible world as well. But how do we know that the divisions are made at the right places? As I pointed out, one indication is that problems caused by erroneous collections and distinctions vanish. Although the fact that contradictions do not follow can be seen as a kind of proof for the correctness of the divisions, Plato seems to assume that direct intellectual apprehension plays a considerable role in how the right divisions are recognised. He seems to assume that human reason has the capability of recognising correct divisions, probably on the basis of understanding how they solve problems. In the late dialogues the significance of the very great kinds,41 namely being, sameness, difference, movement, and rest, becomes prominent. Of these, sameness and difference in particular can also be seen to play a key role in collection and division. Basically, the procedure of collecting requires that relevant similarities between several kinds of things are taken into account, whereas division involves recognising significant differences among the kinds of things one is investigating. In the Statesman the method of collection and division is complemented by the method of example.42 It is assumed that the relevant features of a general skill, such as statesmanship, can be illuminated by the characteristics of

40

For this assumption in Platonism, see Gerson (1999). When Plato uses the superlative expression (,2ciqra … r‡l cel‡l), e.g., in the Sophist (254d), he need not mean that these general categories are the only equally extensive and general categories. He might allow others that are as general as they are. 42 See, e.g., Lane (1998). I am grateful to Pauliina Remes for discussions on this point. 41

34

CHAPTER ONE

another general skill (such as weaving) – the method is also meant to work outside the genre of skills – and the other skill is taken to function as an example of the relevant aspects of the original object of inquiry. Now the question arises how the example is to be chosen and what it is supposed to show us. These questions are too large to be studied here in detail. In any case, the general notions of sameness and difference are again central. By revealing the relevant features of the example we can compare it to our original object of inquiry, and see more clearly where this overlaps and differs from the example. The general notions of sameness and difference are significant, because they point to a connection between the account of collection and division – which can be characterised as methodological – and Plato’s conception of human psychology. Sameness and difference are such general notions which are presupposed in all practices of argumentation and they are necessary even for the humblest forms of concept formation. Plato’s line of thought in several dialogues (e.g., the Sophist and the Timaeus) is that these general notions cannot be acquired from experience but are presupposed by any meaningful general experience: it is impossible to group together observations without these general notions. Like the method of hypothesis, the method of collection and division is so loose that one might wish it to be more rigorous. In fact, it is possible that the limit (/2.ap) and the unlimited (8/ei.4a) of the Philebus are introduced in order to meet this wish. If this is the case, the purpose of introducing them seems to be to emphasise that in making collections and divisions, one should be operating with the subtypes of a given genus and that the divisions are in this sense limited.43 Aristotle will later on make this point explicitly. Also in the Phaedrus, however, Plato assumes that there is a limit to distinctions. The divisions are to be continued until one arrives at something which can no longer be divided (Phaedrus 277b7–9). The most likely meaning of this criterion is that when one has reached that which cannot be further divided, one reaches such subtypes of more general types that cannot be further divided into natural kinds. Further division would force us to distinguish particular cases under these types. Examples of such indivisible types are the four types of divine madness in the Phaedrus. Like the method of hypothesis discussed above, the method of collection and division also entails an intrinsic direction in the procedure.44 Collection always aims at what is more general: it consists in locating the inquired

43

See D. Frede (1993, xxv); Thesleff (1999) takes these principles to be Pythagorean. Thesleff (1999, 68) has also paid attention to the fact that there are two directions in the Platonic methods. 44

THEORIES OF ARGUMENTATION

35

phenomenon under a more general type. Division, on the other hand, always heads towards what is more particular. It involves distinguishing the subtypes of the more general type collected in the first phase. Philosophical Cosmology Now we have seen that in Plato we do find conceptions as to how inquiry in general should be conducted. Both the method of hypothesis and that of collection and division can be seen to belong to the scope of the philosophy of science in a broad sense.45 We must keep in mind, of course, that science in those days was not quite what it is now. In any case, the method of hypothesis can be characterised as a method of constructing and evaluating theories, collection and division as one of making accurate classifications and finding correct definitions. There is yet one dialogue, which is particularly interesting for the history of the philosophy of science and hence needs to be discussed here; it is the Timaeus. It is an exaggeration to say that the Timaeus deals exclusively with questions pertaining to the scope of the philosophy of science. However, the dialogue’s philosophical cosmology can be seen to provide us with a general account of the intelligible structure of the world, which can also be used to account for the facts of perceptible reality. In this sense we can say that it touches upon themes which later become prominent in the philosophy of science. In addition, the dialogue places the human mind and its basic structures into this very same framework. In this way it provides an explanation of some basic cognitive functions by the same explanatory principles as reality in general is explained, particularly the notions of sameness and difference. Rather than classifying these explanations as naturalised psychology, they should be understood within the framework of metaphysical realism. The fact that the basic ingredients of reality are the same as those of the human soul explains 45

The term ‘science’ is precarious in connection with ancient philosophy. Some scholars have suggested that the term should not be used in translating %/iqr3,g; for this view with respect to Aristotle; see, e.g., Burnyeat (1981). However, some scholars have pointed out (e.g. G. E. R. Lloyd 1979, 1–9) that even though what might be called Greek science was in some sense different from our contemporary science, the term can be used if one keeps in mind that not all the connotations of the contemporary use of the term apply in antiquity. For a recent study underlining the connection between Plato’s discussions and the themes within philosophy of science, see Gregory (2000). For my taste Gregory’s view is at some points exaggerated. He, for instance, claims that the theory of recollection is supposed to solve the problem of under-determination of scientific theories by perceptual data. However, it is important to note that to some extent Plato’s dialogues discuss questions belonging to the early history of what is nowadays classified as philosophy of science.

36

CHAPTER ONE

our capacity to understand the world. I shall come back to this aspect of Plato’s Timaeus below. Plato claims in the Timaeus that his purpose is only to give a probable account (eåjÅp ,flhnp) for the perceptible world and its origin.46 However, even though classified as only probable, the account can be seen as a suggestion of what philosophical cosmology might look like. We can illuminate Plato’s approach in the Timaeus by comparing it to the Socrates of the Phaedo and his disappointment with natural philosophy and, in particular, its manner of providing material principles (water, fire, and so on) as basic explanatory factors of reality. Rather than choosing the path of pre-Socratic natural philosophy, Plato thinks that the cosmos is well ordered and structured, and this fact calls for explanation. The story of the demiurge creating the cosmos out of sameness and difference can be seen as Plato’s way of mythically explaining why the cosmos is well structured rather than an unintelligible chaos. Another question is to what extent the ordering activity of reason pertains to the perceptible realm which is in its nature such that it resists any kind of organisation made by reason. One possibility is that the rational order is not fully realised in the perceptible material world and our perceptual environment remains to be governed by chance. Vlastos, for instance, has suggested that rational and generally valid explanations do not strictly speaking apply to the realm governed by chance.47 In the Platonic framework, the intelligible order is not taken to be completely manifested on the perceptible level of reality. However, it is not completely absent from that level either, because to some extent perceptual phenomena can be explained by resorting to the intelligible order. Admittedly, rational explanations do not cover all particular events in the world history. Nonetheless, they do express general regularities and the general principles according to which the cosmos is ordered. *

*

*

At this point, we shall pause for a brief summary of the discussion above. Firstly, we have discussed the technique of argumentation called refutation, which is directed at an opponent. We have seen that arguments of this kind are typically non-sceptical discussions based on the observation that sets of true statements are consistent. In such refutations a critical 46

For the probable account, see, e.g., Berti (1997). Vlastos (1975, 28–30); cf. Cornford (1935, 173–175); for comments on these two works, see Lennox (2001a, 291). 47

THEORIES OF ARGUMENTATION

37

evaluator shows that his interlocutor cannot hold the thesis he at first accepted; this happens when the questioner succeeds in showing that the interlocutor’s initial thesis is incompatible with other statements he accepts in the discussion. Secondly, we have pointed out that Plato’s discussion of the method of hypothesis involves the idea that inquiry aims at finding explanatory structures which are intrinsically directed, and that the direction is determined objectively. The further upwards we go in the structure, the greater the explanatory power. Explanatory power is taken to be based on relations of objective priority among the things involved. The explanans is ontologically prior to the explanandum because the existence of the former is necessary for the existence of the latter. Also the conceptual analysis called collection and division presupposes two directions, towards what is more general and towards what is more particular. Eventually, the explanatory chains constructed and evaluated by the method of hypothesis can be attached to an unhypothetical principle that cannot be reduced to anything prior. In Plato the most likely candidate for such a principle is the form of the good in Republic VII. The good is a superior explanatory principle for all things in the sense that it accounts for the fact that the cosmos we live in is ordered. In addition, the fact that there is an intelligible structure governed by the good behind our perceptual environment, explains why teleological explanations apply. The assumption according to which there are explanatory principles which are objectively first and cannot be reduced to anything primary in the order of nature, becomes prominent in Aristotle and is preserved in the later Platonic-Aristotelian tradition. I shall next turn to discuss the later history of this assumption. 1.2 ARISTOTLE This chapter deals with the question of what kind of basic principles Aristotle thought we have to take as immediately acceptable in various forms of argumentation we engage in. There are basically three different kinds of such principles. Firstly, all argumentation presupposes some regulative (or logical) principles rarely articulated as explicit premises. These set some basic requirements on the form of the argument. Secondly, as we have seen above in the section concerning Plato, there is a powerful tradition in antiquity of understanding argumentation within a certain kind of social framework, where two discussants aim at evaluating the credibility of a given view. In such arguments there is the requirement that the premises must be accepted by the interlocutor or an opponent. With respect to

38

CHAPTER ONE

knowledge in the strict sense, commonly accepted views can at best function as starting points. From those views we can begin more systematic inquiry into the nature of things. In the inquiry, we proceed towards accurate explanations or scientific definitions for the phenomena and the kinds of things we encounter in perception. The third class of principles consists of the premises of scientific proofs, and knowing them requires knowing the definitions or explanations. All these kinds of principles will be discussed below. I shall begin with Aristotle’s analysis of arguments as encounters between two participants; here he clearly builds on his predecessors in the Academy. 1.2.1 Aristotle’s Inheritance from the Academy There are three main forms of argumentation Aristotle inherits from the Academy, namely dialectic,48 induction and conceptual analysis. Aristotle’s general exposition of dialectic is found in the Topics and it can be seen to originate from the refutational technique of argumentation, discussed above. Induction is treated both in the dialectical and in the scientific context. Aristotle in the Posterior Analytics also recommends a form of conceptual analysis as a useful tool for finding definitions. His version of such an analysis is very similar to Plato’s collection and division. Dialectical Syllogisms In order to understand Aristotle’s view of argumentation, we need to look at his analysis of dialectical arguments in some detail. In addition, Aristotle says explicitly (Top. I 2, 101a36–b4) that dialectical argumentation technique is vital to scientific principles, since only dialectic can have access to the premises of scientific proofs: a science does not strictly speaking prove its own premises. Therefore, an analysis of dialectical argumentation is a necessary prerequisite for assessing whether dialectical arguments may establish the premises of scientific proofs. It is characteristic of Aristotle to claim that in the case of all arguments, the context determines the conditions that the arguments should fulfil. From Aristotle’s point of view the differences between arguments are mainly differences in the criteria set for the premises. The main types of arguments are,

48

In Plato diak2ceqhai as a verb applies to argumentation quite generally. However, the combination diakejrij¢ r2ulg appears in a rather technical sense in the Republic to refer to a technique which aims at truth. For Aristotle ‘dialectic’ (diakejrij¢ r2ulg) refers to the communicative arguments involving two participants.

THEORIES OF ARGUMENTATION

39

according to Aristotle, scientific proofs (8/5deimip), dialectical arguments49 and contentious arguments50 (see, Top. I 1, 100a25–101a16; cf. Soph. El. 2, 165a35– b12).51 Rhetorical arguments are normally seen as a subtype of dialectical arguments (e.g., Rhet. 1356a31; cf. An. Post. I 1, 71a9–11). A dialectical argument always involves communication between the two participants where one is posing questions and the other answers them. The one who asks questions tries, by carefully chosen questions, to establish a conclusion, which follows from the statements that the interlocutor accepts in the course of the discussion. Often Aristotle distinguishes dialectical arguments from scientific proofs by the criterion that the premises of a dialectical argument are commonly accepted ((ldnmnl), whereas the premises of scientific proofs are true and fulfil additional very strict criteria (e.g., Top. I 1, 100a27–30). ‘Commonly accepted’ is defined (100b22–24) as that which is accepted either by everyone, by the majority, or by the wise – or by all, the majority, or the most reputable of the wise. However, because the degree of plausibility of the premises depends on the conclusion sometimes discredited premises must also be used in dialectical argumentation. If the conclusion is not plausible, it cannot be validly derived from reputable premises. In such cases we must use premises that are equally credible or more credible than the discredited conclusion, not as such reputable opinions (see Top. VIII 5). 49

Along with dialectical arguments Aristotle mentions in the Sophistical Refutations the peirastic ones. This indicates that Aristotle sees critical evaluation as one of the tasks dialectic can fulfil. For a general introduction to Aristotelian dialectic, see Smith (1993, xi–xxxiv). For more detailed studies, see Bolton (1994) and Kakkuri-Knuuttila (1993); the latter also links the dialectical modes of argumentation to scientific inquiry. Rhetorical arguments are considered as a subclass of dialectical arguments.They use plausible premises from which the conclusion follows. Often some of the premises are not explicitly stated. Cf., however, Burnyeat (1994), who claims that the inferential step in rhetorical arguments is not necessary and, hence, rhetorical arguments are not dialectical. 50 For a general study of Aristotle’s treatise on sophistical refutations, see Schreiber (2003). For similarities between the sophistical refutations Aristotle lists in the Sophistical Refutations and the ones Plato presents in the Euthydemus, see Dorion (1995, 91–104). 51 In the Sophistical Refutations Aristotle mentions didactic arguments as their own subclass. Didactic arguments use premises which are peculiar to the science being taught and they are, in Aristotle’s terms, ‘better known to nature’. This means that they might be quite unfamiliar to the pupil, who, however, is supposed to accept them and not question them as in dialectic. When hearing the proof, the pupil is probably expected to understand how the scientific premises explain the conclusions and, hence, to understand that they must be such premises. Cf. Aristotle’s discussion of premises he calls by the name a©rg,a in Posterior Analytics I 10.

40

CHAPTER ONE

Contentious (%.iqrij5p) arguments, according to Aristotle, differ from proper dialectical ones in the following sense. Whereas a dialectical syllogism draws a conclusion validly from reputable conceptions, the eristic one only appears to do so. This can happen in two ways. Either the premises only appear to be reputable but are not, or the argument only seems to be valid but is not. Aristotle specifies, however, that in the strict sense only the latter case qualifies as a contentious argument, whereas the former can be classified as a proper dialectical argument. (Top. I 1, 100b23– 101a4.)52 Therefore, strictly speaking, Aristotle considers all dialectical syllogisms valid, whereas the contentious arguments only appear to be such (cf. Top. VIII 11, 161a33–b5). Often the eristic arguments are such because they are competitive, and not only competitive, but a kind of dirty-fighting in argument, aiming at victory at any cost (Soph. El. 11, 171b22–34). If the arguments aim at victory for the sake of victory itself, they are called eristic (%.iqrij5p), if for the sake of reputation which enables one to make money out of apparent wisdom, sophistic (qntiqrij5p); as arguments eristic and sophistic ones do not differ. As is often noted, we are to some extent left on our imagination when it comes to the question of what the dialectical encounters were like, because Aristotle does not explain in detail how the discussions are supposed to work.53 However, this point should not be exaggerated. In a sense the Topics does provide an analysis of the nature of dialectical encounters by providing practical advice concerning how to deal with such encounters in an optimal way. Even though Aristotle does not provide us with an example of a complete dialectical syllogism, the treatise is full of examples of what kind of premises the questioner should ask the opponent to accept and in general of what kind of considerations are relevant for the encounter on both sides. In addition, it can hardly be doubted that the arguments, whose technique Aristotle discusses in the Topics, resemble the kind of philosophical encounters Plato describes in the early dialogues.54 Keeping this in mind, it even seems a little odd that Aristotle in such a straightforward manner says at the

52

See, however, Schreiber (2003), who points out that often in the Soph. El. Aristotle deals with cases that cannot be understood as logical fallacies. Rather, they are based on misconceptions concerning how things are. 53 This point has been vividly made by Smith (1997) in his introduction to the translation of books I and VIII of the Topics. 54 Bolton (1993, 121) says that no diligent student of these Aristotelian treatises can doubt this.

THEORIES OF ARGUMENTATION

41

end of the Sophistical Refutations (Soph. El. 34) that he has no predecessor in the task of systematising argumentation. However, as pointed out above, Plato does not analyse the rules and background assumptions of the arguments in a systematic manner. Therefore, Aristotle might well be right in his dictum that he is the first to provide such an analysis of disputations. In any case, in addition to the Topics, Plato’s early dialogues help us in getting a picture of what is going on in a dialectical encounter. Basically, a dialectical syllogism is one whose conclusion follows by necessity from the statements the interlocutor has accepted in the course of the discussion. It is a refutation if it concludes the contradictory opposite of the interlocutor’s initial thesis. Otherwise the definitions are the same for a dialectical syllogism and a refutation, as Aristotle explicitly says in the Sophistical Refutations (Soph. El. 1, 165a2–3, 6, 168a34–37, 10, 171a1–11). The necessity by which the conclusion follows is typically necessity in the following sense: asserting the negation of the conclusion would involve a literal contradiction with the premises. However, this is not the only possibility. Aristotle notes that sometimes the questioner succeeds in showing that the thesis the opponent has accepted entails a false or a highly implausible statement, not an utter contradiction (see Top. VIII 4, 159a19–21).55 In one of these ways the argument shows that one should rather accept the other beliefs accepted in the discussion as premises and revise the thesis that caused the false or implausible conclusion. When the opponent’s thesis entails a falsity or some otherwise implausible thesis, it is refuted only if the questioner succeeds in showing that the falsity or the implausibility follows from the thesis, not from the other premises accepted in the discussion (VIII 10, 160b24–33). It is important to note from the start that Aristotle makes it clear (Soph. El. 1, 165a2–3, 6, 168a34–37, 10, 171a1–11) that all refutations are at the same time positive arguments for the contradictory opposite of the answerer’s thesis. A dialectical argument can also take the form of a reductio ad impossibile when an indirect argumentative strategy is chosen.56 In an indirect argument a thesis is shown to hold by assuming its negation and concluding something impossible, false or very unconvincing on the basis of the negation of the thesis and, again, with the statements the answerer accepts in the discussion. If one prepares to act as an answerer defending a view, one should try to make sure that a questioner could not refute it on the basis of destructive strategies.

55

Cf. Smith (1993, 130). For different argumentative strategies from the impossible in Aristotle, see Smith (1993, 119–122). 56

42

CHAPTER ONE

If, on the other hand one wishes to argue for a view as a questioner, indirect arguments can be used.57 All dialectical arguments have an epistemic aim: the premises must be initially more credible than the conclusion and, hence, the idea is to enhance the credibility of the conclusion. Even in arguments that proceed from premises with a low credibility rate, the premises must at least appear credible to the answerer. Aristotle mentions two rather different criteria for evaluating the quality of the arguments. The first criterion is based on logic. From a logical point of view, the interlocutor might perform his task well and remain consistent, even though the premises he accepts are not plausible. However, in such a case argument cannot, according to Aristotle, be really good. To be a genuinely good dialectical argument, another criterion is needed. A good dialectical argument is not only valid, but the premises must be reputable in the sense defined in the Topics I 1: accepted either by the majority or the wise – or by all, the majority or the most noted of the wise. I shall next turn to discuss Aristotle’s definition of a dialectical syllogism and then provide a picture of dialectical argumentation with some examples. After that the central notion of a dialectical topos (r5/np) and the validity of dialectical arguments will be discussed. At the end of this section, we shall return to the question of what conditions a good dialectical argument – as opposed to any dialectical argument – should fulfil. The Definition of a Syllogism. Aristotle defines a syllogism at the beginning of the Topics. Virtually the same definition also appears in the Prior Analytics (An. Pr. I 1, 24b18–22) and the Sophistical Refutations (1, 165a1–4). It is a general definition of valid inference with some peculiarities. 57

Aristotelian dialectic strategies also resemble the Platonic method of hypothesis in the sense that they are used in investigating what follows from a given thesis in a structure of arguments. Aristotle uses the term ‘hypothesis’ (∫/5heqip) in two different senses to mean statements of existence of a scientific object (An. Post. I 2, 72a18–20) or statements which in fact could be proved from unprovable premises but which are in an argument taken as accepted without proof (e.g. An. Post. I 9, 76b35–39, 77a4). The former passage is the only one in Aristotle where the term has this strict meaning; for the uses of the term, see Ross (1949, 510–511; 540), Heath (1949, 53–57). The last mentioned meaning, which is the more common, resembles the Platonic use of the term. For analogies of the Platonic method of hypothesis in Aristotle, see also Top. II 4, 111b16–18: ‘look and see in regard to the thing in question, what is such that if it is the case the thing in question is the case, or what is necessarily the case if the thing in question is the case’. For the Aristotelian use of the term ∫/5heqip, see also De Caelo I 12, where Aristotle calls his indirect proofs ones from a hypothesis (%m ∫/nh2qewp).

THEORIES OF ARGUMENTATION

43

Syllogism is an inference (k5cnp) in which certain things having been laid down something other than those that are laid down necessarily follows on the basis of them (di¡ r‡l jei,2lwl) (I 1, 100a25–27; my translation).

It has often been pointed out that this definition does not restrict inferences to syllogisms in the three figures, i.e. the two-premise three-term inferences discussed in the Prior Analytics. Because of this some scholars have started to translate qskknciq,5p as ‘deduction’.58 However, others have pointed out that this translation is not quite appropriate because Aristotle’s definition of syllogism excludes some deductions, for instance ones whose premise or premises are identical with the conclusion and those that have redundant premises.59 I have here left qskknciq,5p untranslated as ‘syllogism’; the dialectical syllogism is not syllogism in the more narrow sense of the syllogistic figures of the Analytics. Nonetheless, we should also note that the very same definition is presented in the Prior Analytics. There has been a lot of discussion concerning Aristotle’s definitions of a syllogism. All the three instances, namely that in the Topics, that in the Sophistical Refutations and that in the Prior Analytics, are unequivocal in saying that the syllogism is an inference or argument (k5cnp) in which a conclusion follows from the premises by necessity (%m 8l1cjgp) and in such a way that the conclusion is something else ()re.5l ri) than the premises. We have already given short characterisations of the conditions of necessity and being an argument, and we shall return to these conditions when discussing the examples of dialectical syllogisms. I shall now turn to the third condition according to which the conclusion has to be something else than the premises. In the dialectical context the condition that the premises have to be something else than the conclusion is not restricted to the requirement that the conclusion should not appear in the premises. Rather, Aristotle lists five cases (Top. VIII 13, 162b34–163a13) of which only the first one has (i) a conclusion identical with one of the premises. In fact, an argument that has one or more premises identical with the conclusion would be extremely bad in a dialectical context: the questioner has to derive the conclusion from the claims the opponent accepts and the opponent tries to resist the conclusion. It is highly unlikely that the opponent would not notice that the intended conclusion is asked as a premise. The only possibility Aristotle mentions of how the premises and the conclusion could be identical is when the premise expresses the same claim as the conclusion but in synonymous terms. Aristotle would count that as a case of asking the conclusion as a premise. Hence, such an argument would not satisfy the general definition of a syllogism. 58 59

E.g. Smith (1994) and Corcoran (1974a). E.g. Kakkuri-Knuuttila (2005); cf. Bolton (1994).

44

CHAPTER ONE

The other cases Aristotle mentions as falling under the heading ‘begging the question’ (aåre‹qhai r• %l 8.u·), are the following: (ii) asking the interlocutor to accept a more universal version of the conclusion (e.g., if the intended conclusion is that there is one science concerning the contraries, and the questioner asks the interlocutor to accept that there is one science concerning opposites), (iii) asking the opponent to accept a particular case of a universal conclusion, (iv) if the conclusion is a conjunction, to ask both of the conjuncts separately, (v) if the conclusion is convertible, ask the conversion (e.g., if the conclusion is that a diagonal is incommensurable with the side, asking the interlocutor to accept that the side is incommensurable with the diagonal begs the question in Aristotle’s sense). The third clause (iii) entails that a dialectical syllogism cannot be in the form of an induction. This is well in line with Aristotle’s distinction between syllogisms and inductions elsewhere in the treatise (I 12, 105a11–12). The clause does not, however, rule out the possibility of getting some of the premises by induction from the interlocutor. It is just the final conclusion which cannot be drawn inductively. Aristotle also distinguishes (VIII 1, 155b20–21) between premises that are used to argue for the final conclusion of a dialectical argument and premises that are used to attain those premises from the opponent. He calls the former premises necessary (VIII 1, 155b20–21), even though they are not necessarily true; they are necessary in order to derive the conclusion. He says a little later (VIII 1, 155b36–38) that induction is always helpful in making the opponent concede the premises needed for the final conclusion, but it is not good in deriving the final conclusion because it cannot force the opponent to accept the conclusion (cf. Top. I 12, 105a17–19). When the argument is not deductively valid, it is unlikely that the opponent should concede its conclusion. In the remaining three cases (ii, iv and v) the problem is that the intended conclusion follows too obviously from the premises and an answerer would not accept them. However, it is possible that Aristotle’s remark is not merely pragmatic. He might also think that these are cases where the logical consequences of our beliefs are obvious to anyone. If someone believes something universally, the particular belief also necessarily follows;60 if someone believes two separate conjuncts (if a believes that p and a believes that q), she also believes the conjunction (a believes p&q) and similarly in the case of convertible terms. Inferences involving such premises might be admissible in the course of the debate in order to get a premise from the answerer, but the conclusion proper should not be this close to the premises.

60

This, however, is denied in An. Post. I 1.

THEORIES OF ARGUMENTATION

45

To sum up, the condition that the conclusion be different from the premises is not purely a logical one. It has got to do with the question to what extent the logical relations of our beliefs are obvious to us. We can characterise the five cases Aristotle lists as those where Aristotle takes the relation between the two claims to be so obvious that it would even be an exaggeration to distinguish two distinct beliefs there. There remains one condition in Aristotle’s definition of a dialectical syllogism, which is not put in exactly similar terms in the three treatises where the definition occurs. This is the condition that the conclusion has to follow because of the premises. In the Topics (100a25–27) he says that in a dialectical syllogism: something else than the premises follows necessarily because of the premises ()re.5l ri r‡l jei,2lwl %m 8l1cjgp qs,ba4lei di¡ r‡l jei,2lwl).61

In the Prior Analytics (24b18–22) he puts the definition as follows. A syllogism is an argument (k5cnp) in which certain things having been laid down something else than what is laid down follows necessarily because they are such (%l –\ reh2lrwl ril‡l )re.5l ri r‡l jei,2lwl %m 8l1cjgp qs,ba4lei r+ raflra e∆lai) (my transl.).

He explains that the clause ‘because they are such’ means that ‘the [conclusion] follows because of them’ (r• d£ di¡ raflra qs,ba4leil), which, in turn, is explained as meaning that no additional assumptions are needed in order for the necessity to come about, i.e. in order for the inference to be valid.62 Some scholars claim that the explanation of the Prior Analytics, according to which the premises have to be sufficient for the conclusion to follow necessarily, is not quite what Aristotle means in the Topics. One suggestion made by Bolton is that, in the Topics, the condition means that the premises should give an explanation or reason why the conclusion is as it is.63 I agree with Bolton that in a sense the fourth condition, according to which the conclusion must follow because of the premises, can be taken in the sense that the premises are reasons for the conclusion. However, we need to be careful in articulating this condition. It cannot be taken in the sense that the premises should express the ultimate real explanation of the conclusion, because such a condition is peculiar only to the premises of scientific proofs proper (e.g., An. 61

The same phrase di¡ r‡l jei,2lwl also appears in the version of the definition in the Soph. El. (165a3). 62 According to Smith (1994), the condition according to which the conclusion follows from the premises because of them is the same as the criterion of validity. This identification is problematic (see, e.g., Bolton 1994, 116–117). 63 Bolton (1994, 117–119).

46

CHAPTER ONE

Post. I 2, 71b9–16),64 and Aristotle makes it clear that dialectical arguments do not fulfil those conditions. Therefore, if the premises of dialectical arguments are understood as reasons for the conclusion, this means that they must be taken as reasons for the interlocutor to accept the conclusion. They do not necessarily explain why the conclusion holds. From a contemporary perspective this might sound odd, because explanation is nowadays typically taken to be an epistemological notion. If understood epistemologically, explanation involves precisely that something is made known or made comprehensible to us human beings. However, we can make a distinction between metaphysical and epistemological notions of explanation.65 If explanation is understood as a metaphysical notion, it is assumed that there is a factual basis in the nature and configuration of things for all explanations. Therefore, to say that the premises of proofs explain the conclusion in the metaphysical way, involves saying that the things that the premises and the conclusion refer to are connected with each other in reality. It seems that it is typical for Aristotle to understand explanation metaphysically. In the scientific proofs the premises must express the explanation of the conclusion, and this presupposes an essential connection between them. In addition, Aristotle’s so-called four types of causes are also better described as four types of explanation than as four causes,66 provided that we keep in mind that they are explanations in the metaphysical sense. Now, the relevant preposition ‘because of’ (di1) in the definition of a dialectical syllogism cannot be understood in the metaphysical sense because this would make dialectical arguments collapse with the scientific proofs. Rather, it should be taken as follows. In a dialectical argument, the interlocutor is shown that some of his beliefs make him accept the conclusion, which he previously rejected. He has to accept it because he accepted the other claims in the course of the inquiry and it follows necessarily from them. To sum up, a dialectical syllogism is an argument, where a conclusion other than the premises follows necessarily from the premises. Necessity is to be understood in one of the following two ways. Either the conclusion follows necessarily in the sense that its denial is contradictory to the accepted premises, or necessity is understood in a weaker sense: one of the premises leads to an obvious falsity or a highly implausible claim and should, therefore, be revised. The conclusion follows because of the premises in the sense

64

The conditions for scientific proofs will be discussed in 1.2.2. See, e.g., Moravcsik (1991). Moravcsik also points out that Aristotle is, in his theory of proofs, primarily interested in explanation understood in the metaphysical way. 66 Cf. Hankinson (1998a). 65

THEORIES OF ARGUMENTATION

47

that, on the one hand, no additional assumptions are needed for it to follow and, on the other hand, because the premises give reasons for the interlocutor to accept the conclusion. This reflects the general epistemic aim of dialectical arguments. Aristotle also lays out rather strict conditions on what it means that the conclusion is something else than the premises. It is not only the condition that the conclusion cannot be identical with one of the premises. Most importantly, an inductive inference cannot constitute the final step from the premises to the conclusion. In addition, the conclusion should not follow too obviously. Examples of a Dialectical Syllogism and of a Refutation. We shall now turn to discuss in more concrete terms what the dialectical arguments are like. The beginning moves of the questioner and the interlocutor are clear (see, e.g., I 4, 101b28–33). First, they agree on the problem (/.5bkg,a) they are going to discuss. For instance, what is the definition of man? These problems must be formulated as questions, which can be answered by ‘yes’ or ‘no’ (is ‘pedestrian biped animal’ the definition of a human being?). When they are thus formulated, they become suggestions or propositions (/.nr1qeip) that are put forward by the questioner. ‘Putting forward’ is understood in concrete terms: the questioner poses them as questions to the interlocutor. Suitable problems are the following three types of claims: (i) claims concerning which people do not usually have an opinion, (ii) claims about which most people have an opinion, but this opinion is contrary to that of the wise, and (iii) claims about which these classes (the majority and the wise) disagree among themselves. In general, dialectical problems are questions the answer to which is not immediately evident. However, they cannot be views which no-one would accept either. The former would cause no proper discussion at all. The latter would be silly to try to argue for because no one would accept premises from which it follows. (I 11, 104b1–17, I 10, 104a5–7 and I 14, 105b19–29.) However, a paradoxical view is worth examining if it is put forward by an eminent philosopher (e.g. Heraclitus, I 11, 104b18–24). An example of a dialectical syllogism would be one establishing the conclusion that, e.g., one should attack the Spartans (considered as enemies). The questioner needs67 premises of the following sort: If one should do good to friends, one should do harm to enemies. One should do good to friends.

67

As noted, Aristotle sometimes calls them necessary (VIII 1, 155b20–21). He does not mean that they are necessary in the sense of being necessarily true; they are necessary for the questioner to derive the conclusion validly.

48

CHAPTER ONE

From these the questioner could then draw a mediating conclusion that one should harm enemies. He should probably ask the interlocutor to accept some irrelevant premises in between asking those two which are necessary for the conclusion to follow. Particularly, he should be careful in asking the opponent to accept the following ‘protecting friends is doing good to them, whereas attacking an enemy is doing bad to them’; this links the two more general premises to the intended conclusion. As noted, if the questioner asks the two first-mentioned premises straight away, it is unlikely that any opponent accepts them as such. This is where the questioner needs the so-called rules of concealment (VIII 1). One way of hiding the strategy is not to ask the premises which apply to the case in point, but to keep as far as possible from it. Following this recommendation the questioner could, for instance, ask only a more general commonly accepted principle on which the first premise is based. Such a principle is somewhat cryptically expressed by Alexander of Aphrodisias (e.g. in Top. 126, 16–17) thus68: If a contrary belongs to a contrary, then the contrary belongs to the contrary.69

This means something like the following.70 There are two pairs of things so that each member of the pair is contrary to a member of the other pair (such as beneficial and good, harmful and bad). Now, if the contrary of the first member belongs to the contrary of the second member (harmful belongs to bad, i.e., bad is harmful), also the other pair behaves similarly (beneficial belongs to good, i.e., good is beneficial). Because this principle is classified as a dialectical topos, it is possible that the questioner could assume that the opponent accepts it without asking. Aristotle’s manual on dialectic is based on the idea that some such generally acceptable premises – even though they are not always true – can be taken for granted. However, another way would be to argue for it inductively, i.e. by asking some instances of the general principle (at the same time being careful not to ask those general principles on which the intended conclusion

68

For topoi concerning contraries in Aristotle, see, e.g., Top. II 7, and II 8, 113b27–34. 69 Alexander at this point concentrates on complaining about Theophrastus’ definition according to which as general principles the topoi are not determined with respect to their subjects, but can be applied to various cases. Alexander gives examples about the premises for which the general principle about contraries is useful. He mentions the following examples: ‘if bad is harmful, good is beneficial’, ‘if black colour compounds the sense of vision, white divides it’ and ‘if pain is bad, pleasure is good’ (in Top. 126, 17–30). 70 Cf. also Smith (1994, 145–146).

THEORIES OF ARGUMENTATION

49

hangs) and then ask the principle itself. In the dialectical context the opponent is supposed to accept a generalisation if he cannot provide a counterexample (Top. VIII 2, 157b31–32; cf. 157a34–38). Aristotle probably assumes that the questioner can rely on the opponent not having any. Note that even though the conclusion hinges on the general principle just mentioned, deriving it inductively does not beg the question, because the conclusion (we should attack the Spartans) is not an instance of the principle. In addition to the general principle concerning contraries, the only more substantial premise the questioner needs is that one should do good to friends. If the opponent accepts that, the questioner will have sufficient grounds for deriving the conclusion by articulating what follows from the general principle concerning pairs of contraries and the premise that one should do good to friends. If a contrary belongs to a contrary, then the contrary belongs to the contrary. (Accepted general premise, possibly derived by induction) If one should do good to friends, one should do bad to enemies. (Corollary from the principle) One should do good to friends. (Accepted premise) One should do bad to enemies. (Mediating conclusion) Attacking is bad. (Accepted premise) Spartans are enemies. (Accepted premise) Therefore, one should attack the Spartans. (Final conclusion)

This example is rather schematic. It is intended to illustrate the previous, fairly abstract discussion. An example of a refutation is easier to formulate. If, for instance, the participants choose to discuss the goodness of pleasures, the answerer might choose as the thesis the claim that pleasure is good.71 The general outline of the questioner’s argumentative strategy could in this case consist of the recognition that the thesis would amount to the claim that good is the genus of pleasure (e.g. Top. IV 1, 120b15–20). The general premise underlying the argument strategy is the following: The genus of a species is predicated of all the members of that species (Top. IV 1, 120b20; my transl.).

71

Aristotle, however, does not recommend the answerer to choose the thesis that pleasure is good, because – he says – it leads people to hate the answerer and think he has a bad character, not that he is holding the thesis just for the sake of argument (Top. VIII 9, 160b17–23). Another such example is the claim that it is better to commit an injustice than to suffer it. It is remarkable that both of these examples appear in Plato’s dialogues in the mouth of Socrates’ most hard-boiled interlocutors, Callicles and Polus in the Gorgias and Thrasymachus in the first book of the Republic.

50

CHAPTER ONE

The questioner would then need the answerer to accept the following premises: Excessive drinking is a pleasure. Excessive drinking is not good.72

Now, on the one hand good has been stated to be the genus of pleasure. On the other hand, the interlocutor has conceded that there is at least one pleasure, namely excessive drinking, which is both a pleasure and not good. This contradicts the principle that a genus belongs to all the sub-cases of its species. Therefore, good is not the genus of pleasure.73 In this case the questioner should again be extremely careful not to ask the premises straight away. The opponent would almost certainly see what kind of strategy is going to be used and reject claims he otherwise accepts – just because he would not like to lose the debate. The concealment rules together with induction would again be needed. The rules of concealment have caused much scholarly dispute because they have been thought to form obstacles in constructing genuinely good arguments instead of competing for victory.74 However, even though the questioner tries to conceal his argument strategy, this need not mean that this would affect the form of the argument itself. The rules of concealment provide the questioner with a means to construct an argument whose premises an opponent would accept. The strategies of concealment are concerned mainly with the social aspects of dialectic. I shall not, therefore, discuss them in more detail here. 72

Both Smith (1997) and Kakkuri-Knuuttila (2005) point out that this inference already involves the notion of a logical form. Kakkuri-Knuuttila uses this inference to illustrate how the theory of the syllogistic figures in the Analytics might have evolved from the dialectical syllogisms. Excessive drinking is not good. Excessive drinking is a pleasure. Therefore, not every pleasure is good. 73 An ancient example would be the following. First the questioner should ask the following fairly uncontroversial premise. ‘All good things are naturally desirable.’ From this the questioner could then infer that if all pleasures are good, they are naturally desirable. The questioner would now need the answerer to accept, for instance, that scratching an itch is a pleasure and to contrast it with naturally desirable things. Even though scratching an itch is a pleasure, it is not naturally desirable, because no-one wants to have an itch just in order to scratch it. The example appears in Philoponus’ commentary on the Prior Analytics (276, 20–29). Philoponus says explicitly that the premise that all pleasures are natural is generally accepted (jar¡ d5mal) but not true. Therefore, it is exactly the kind of premise used in dialectical argumentation. 74 For the discussion, see, e.g., Kakkuri-Knuuttila (2005) and Bolton (1994, 104–105).

THEORIES OF ARGUMENTATION

51

Validity and the Notion of Topos. As mentioned above, Aristotle says explicitly in all the three definitions of a syllogism that the conclusion follows necessarily from the premises. The passage from the Topics is clear that this applies to dialectical arguments as well. However, it has also been observed that the necessity by which the conclusion follows in a dialectical argument does not always mean that a denial of the conclusion would entail a literal contradiction with the premises. Sometimes there is only an obvious falsity or a highly implausible claim. This allows for refutational arguments, where no literal contradiction is found. Such a result is, according to Aristotle, sufficient for the questioner to show that the opponent should revise his belief set. Some scholars have suggested that there is also a further way in which the standards of validity are loosened in a dialectical context. According to this suggestion, Aristotle proposes non-deductive inference patterns such as ‘the generations of good things are good’ and ‘the generations of bad things are bad’, or ‘if a predicate belongs to a subject, then it also belongs to a subject to which it is more likely to belong’.75 However, I do not agree with this suggestion, basically because I think that the examples given of the supposed non-deductive inference patterns are commonly accepted dialectical premises, not inference patterns. Therefore, rather than introducing non-deductive inference forms, Aristotle allows that a dialectical argument can use general premises which are not universally true. I shall now turn to discuss this point more closely. Aristotle’s basic means for systematising argumentation are the so-called dialectical topics, topoi.76 As is well known, Aristotle does not define the central notion of the topos; hence it is no wonder that the notion has been a matter of dispute since antiquity.77 Alexander of Aphrodisias suggested that a topos should be understood as a heuristic advice for inventing arguments. Theophrastus, on the other hand, claimed that the topoi are generally accepted premises to be used by the questioner.78 A more recent suggestion is

75

E.g. Green-Pedersen (1984, 27) and Kakkuri-Knuuttila (2005). Other translations of this central notion include ‘commonplace’ and ‘location’. The literal meaning of the Greek r5/np, of course, is ‘place’ or ‘location’. 77 For the discussion, see Smith (1997, xxiv–xxviii), Brunschwig (1967), Stump (1978, 159–178) and De Pater (1968). 78 Alexander quotes Thephrastus’ definition according to which ‘a topos is a kind of principle (8.u3 rip) or element (qrniue‹nl) from which we get (ka,b1ln,el) the premises concerning each case (r¡p /e.§ )jaqrnl 8.u1p); [it is] determinate in outline (r· /e.ic.at· ,£l ®.iq,2lnp) but indefinite as to particulars (rn‹p d£ jah’ )jaqra 85.iqrnp)’ (in Top. 126, 13–15 and 5, 21–26). 76

52

CHAPTER ONE

that the topoi include an argument form which is used in the dialectical encounter.79 In what follows, I shall show how Aristotle’s notion of topos involves various things that are necessary for the questioner (Top. VIII 1, 155b4–10): it provides the questioner with a general premise, a piece of advice, and an argumentative strategy based on these aspects.80 In addition, I shall argue that the argumentative strategies are deductive. Consider some examples Aristotle gives of the topoi: If the species is a relative, [one must] examine whether the genus is also a relative: for if the species is a relative, the genus is also, for example double and multiple; for each is a relative (IV 4, 124b15–18, transl. Stump (1978, 167)). One commonplace [r5/np] is to look whether your opponent has assigned as an accident something which belongs in some other way (II 2, 109a34–36; Loeb translation by Forster 1960). Here is another commonplace (r5/np); when one predicate is applied to two subjects, then, if it does not belong to the one to which there is the greater likelihood of its belonging, it does not belong either to the one which it is less likely to belong; and if it belongs to that to which it is less likely to belong it belongs also to that to which it is more likely to belong. (II 10, 115a7–9; transl. Forster 1960.) For example the pleasant stands in the same relation to pleasure as the beneficial to the good; for in each case the one is productive of the other. If, therefore, pleasure is what is good, then the pleasant will be what is beneficial; for it is clear it would be productive of good, since pleasure is good. (IV 4, 124a16–20; transl. Forster.)

The first example includes all the relevant aspects of the notion of topos in a concise form. On the one hand, there is the premise ‘if the species is a relative, the genus is also [a relative]’. On the other hand, there is also a piece of advice ‘if the species is a relative, examine whether the genus is also a relative’. These can be used to construct an argumentative strategy. If, for instance, the opponent claims in a more particular form an instance of the case that a species is a relative but its genus is not, this involves a contradiction of the general principle. Such general principles as ‘if the genus is relative a species will be relative as well’, Aristotle assumes, are typically accepted. Often a dialectical argument is even characterised as one which uses generally accepted conceptions ((ldnma) as premises (e.g. Top. I 1). As the definition of a generally accepted 79

Smith (1997, xxvi) and Stump (1978). Also Kakkuri-Knuuttila (2005), Green-Pedersen (1984), Primavesi (1996, 82) and Smith (1997, xxiv–xxviii) agree that the notion of topos includes both the advice and the general premise; cf. also Stump (1978) and de Pater (1968). 80

THEORIES OF ARGUMENTATION

53

opinion shows, it is not only a majority opinion.81 Aristotle also recommends the questioner to point out that some conception is generally accepted (e.g., held by a famous thinker such as Empedocles, Top. I 14, 105b16–18; cf. ibid. 10–13). This usually makes the opponent accept the premise more easily. The second example involves a piece of advice for the questioner. There is, however, no reason to conclude from this that Aristotle’s notion of a topos does not include the dialectical premise. In addition to the first example, the third topos quoted above, for instance, provides us with fairly straightforward examples of general premises. First there is the generally accepted principle ‘if some predicate does not belong to a subject it is likely to belong to, it does not belong to another subject to which it is less likely to belong either’. A similar generally accepted principle is ‘if a predicate belongs to a subject to which it is less likely to belong, it also belongs to a subject to which it is more likely to belong’. These in the later Latin terminology are called a fortiori principles. The suggestion that dialectical arguments are not necessarily valid is based on the idea that a dialectical argument would be using the topoi as such as patterns of inference. Let me illustrate the point with the fourth example of a topos. For example the pleasant stands in the same relation to pleasure as the beneficial to the good; for in each case the one is productive of the other. If, therefore, pleasure is what is good, then the pleasant will be what is beneficial; for it is clear it would be productive of good, since pleasure is good. (IV 4, 124a16–20; transl. Forster.)

To claim that the argument strategy is non-deductive entails that the questioner must only ask premises of the following kind ‘Is pleasure good?’, and then go on to conclude that what is ‘pleasant will be beneficial’. However, a more likely reading is that the questioner uses a general premise such as ‘if pleasure is good, what is pleasant will be beneficial’. If the opponent then also accepts that ‘pleasure is good’, it can be inferred validly that what is pleasant is beneficial. Note, however, that the full conditional making the inference valid can sometimes be left unarticulated, and the topos can function as a tacit premise. Another possibility of non-deductive argument forms would be an argument based on the premise of the following form: Usually when one predicate is applied to two subjects, then, if it does not belong to the one to which there is the greater likelihood of its belonging, it does not belong either to the one to which it is less likely to belong. 81

As noted above, ‘commonly accepted’ is defined (100b22–24) as that which is accepted either by everyone, by the majority, or by the wise – or by all, the majority, or the most reputable of the wise.

54

CHAPTER ONE

If the questioner would then conclude, for instance, that wisdom cannot belong to a student if it does not belong to the teacher, this could be taken to involve a non-deductive inference. This example would be non-deductive because the general premise is not in a universal form. However, I do not think that Aristotle takes generalisations to function in dialectic in this way. Rather, it seems that the generalisations are taken universally even though this would be acceptable but not strictly speaking true. In fact, as we noted, Aristotle makes it clear (Top. VIII 2) that, in a dialectical context, inductive generalisations must be considered true if the opponent cannot produce a counter-example. Therefore, rather than being of the qualified general form ‘usually when one predicate is applied to two subjects, then, if it does not belong to the one to which there is the greater likelihood of its belonging, it does not belong either to the one to which it is less likely to belong’the premise would be genuinely general: ‘when one predicate is applied to two subjects, then, if it does not belong to the one to which there is the greater likelihood of its belonging, it does not belong either to the one to which it is less likely to belong’. If it is the case that the general principle does not hold in all particular cases, this will not affect the form of the argument, but the truth-value of its premises. As Eleanore Stump has pointed out, there are some three or four hundred topoi in the Topics.82 If the questioner has to remember all of them in order to be a good dialectician, it is an enormous task. However, there are some organising concepts that make it easier to memorise and use the topoi.83 One important set of general concepts according to which premises can be classified consists of dialectical tools (≈.cala). The most general of these are the so-called predicables, namely genus, unique property, definition, differentia and accident. The arguments are further classified according to general notions such as similarity, opposites and more or less equal. The predicables are taken to be part of Aristotle’s inheritance from the Academy and they play a prominent role in medieval theories of argumentation.84

82

Stump (1978, 177). It is often noted that Aristotle’s topoi resemble the topoi of the ancient mnemonic techniques. Aristotle makes this point himself (Top. VIII 14, 163b28–32). For the connection between Aristotelian dialectical topoi and the places of ancient mnemonic techniques, see Sorabji (1972); cf. Smith (1994, n. 21 p. 147). 84 See Smith (1997, xi). In medieval terminology the notions are, following Porphyry’s Isagoge, called quinque voces. The list is not exactly the same as Aristotle’s. In Porphyry the notion of a species is substituted by the notion of definition; see Brunschwig (1967) in the introduction to the French translation of the Topics. 83

THEORIES OF ARGUMENTATION

55

A Good Dialectical Argument. We have until now discussed what kind of conditions a dialectical argument must satisfy. Let us now turn to the conditions for a good dialectical argument. Aristotle notes (Top. VIII 11, 161a19–21) that none of the two participants alone can guarantee that the argument is good. It mostly remains the opponent’s task to see to it that the premises are as convincing as possible.85 Aristotle assumes that the best arguments are produced when the two participants share a common goal. Aristotle points out that the following two failures are possible on the opponent’s part: choosing the wrong thesis and failing to defend it (VIII 4, 159a23–25). We must not criticise an opponent when he accepts statements with a low credibility rate if he remains consistent. In such a case, in fact, the opponent wins. The opponent makes no mistake because the low credibility of the premises is due to the implausibility of the chosen thesis. The argument is a bad one (Top. VIII 11, 161b21), not because of the opponent’s bad argumentation skills but because the thesis is discredited. Therefore, partly the quality of a dialectical argument is dependent on the credibility of the premises. Aristotle gives the following rules for the relations in the degree of credibility of the dialectical premises (Top. VIII 5, 159a38–b24).86 These rules hold in arguments aiming at evaluating positions or making inquiry (Top. VIII 5, 159a33). Such cases do not involve competition (Top. VIII 11, 161a38–39). The conclusions of all arguments can, according to Aristotle, be placed in one of the three classes: the conclusion is plausible ((ldnmnl), discredited (!dnmnl), or neither, i.e. neutral in terms of credibility.87 If the conclusion is plausible, the answerer should only accept premises that are either more plausible than the conclusion or plausible simpliciter. If the conclusion is discredited, he should concede only to premises which are either plausible or less discredited than the conclusion. In the case of a neutral conclusion, everything which seems to be the case (r¡ tail5,ela) can be accepted, as well as all statements that are more acceptable than the conclusion. Similar rules can be applied when a particular philosophical position is evaluated (Top. VIII 5, 159b24–35). In such cases the degree of plausibility must be qualified accordingly. For instance, if one is evaluating Heraclitus’ position according to which good and bad are the same thing the opponent

85

See Kakkuri-Knuuttila (2005). For discussions, see, e.g., Kakkuri-Knuuttila (2005) and Bolton (1994). 87 If the argument is in the form of a refutation, the opponent’s initial thesis is contrary in the degree of credibility compared to the conclusion. If the conclusion is plausible, the thesis is discredited; if the conclusion is discredited, the thesis is plausible. The thesis and the conclusion are neutral at the same time. 86

56

CHAPTER ONE

must answer according to what is plausible, implausible, or neutral from the point of view of that position. As Kakkuri-Knuuttila has noted, the rules Aristotle gives for the degrees of plausibility in a good dialectical argument include the following condition which is not often highlighted.88 This is the condition that the premises of a dialectical argument should be initially more acceptable than the conclusion. Therefore, dialectical arguments always carry an intrinsic epistemic aim. The purpose is to find grounds for someone to believe a claim that is not identical with the premises. The epistemic condition is formulated as applying to good dialectical arguments (Top.VIII 5, 159b7–9) but, as also Bolton points out, there are indications that Aristotle intends it to hold for dialectical arguments in general. The premises of the dialectical argument must, according to Aristotle, contain at least a minimal degree of credibility. If one has to argue from false premises, they must be at least made to appear credible. Therefore, the epistemic nature of dialectical argumentation is built into the very concept; such conditions are not only specific to good dialectical arguments. In particular, the dialectical technique is designed to evaluate the credibility of fairly large belief-sets. Aristotle himself notes in chapter 5 of book II of the Topics (112a17–24) that when we accept a statement we do not just accept that statement: in fact we are committed to a number of other things that necessarily follow from that statement (%m 8l1cjgp 8j5knsha). This is how our beliefs lead us to other beliefs, and the problems in those beliefs that follow reflect problems in the initial beliefs. By critical evaluation the dialectical technique helps to transform one’s belief-set into a more plausible one by showing which ones have false, implausible or even contradictory consequences. Kakkuri-Knuuttila has proposed a more detailed analysis of how credibility should be evaluated.89 Her basic idea is that even though dialectic is to some extent a critical activity, the critical mode can easily be changed into a positive one by aiming one’s attention at the claims that remain intact in the dialectical encounters. Therefore, in addition to showing some beliefs to be such that they can only be members of relatively implausible belief-sets, dialectical arguments also show that some other belief-sets are not equally problematic. According to Kakkuri-Knuuttila’s suggestion, the dialectical technique can provide a high credibility rate for views that survive careful dialectical scrutiny; in such scrutiny they have been

88 89

Kakkuri-Knuuttila (forthcoming); cf. Bolton (1994). Kakkuri-Knuuttila (1993).

THEORIES OF ARGUMENTATION

57

competing with alternative views on the same issue with both pro and con arguments. Even though we can see that dialectical argumentation is capable of enhancing the credibility of large belief-sets and even of establishing, for instance, that one of two competing views is much more credible than the other, Aristotle never says that dialectic could establish truth.90 He always distinguishes dialectical arguments from proofs proper by the condition that whereas the latter have true premises, the premises of the former are plausible but not true. However, this does not mean that the dialectical technique should not have methodological value. In fact, one of Aristotle’s methodological key principles is that when one has examined closely the reputable conceptions concerning a certain issue, pointed out the problems that follow, and proposed qualifications that make the reputable conceptions more resistant towards dialectical criticism, one has shown enough. This is the core of the principle which is called by the name ‘saving the appearances’ (qÍfeqhai r¡ tail5,ela, EN VII 1, 1145b1–7). One important difference between the conditions of truth and that of being reputable needs to be observed here. The requirement of truth simply points out that the premises of scientific proofs have to be true; it says nothing about how we come to know them or how we come to distinguish between true and false conceptions. This can be contrasted with being reputable or being ‘endoxic’, which refers to groups of people who hold the conceptions Aristotle is talking about. As such, they can be true as well as false. Therefore, the distinction between true and reputable conceptions is not a proper dichotomy. However, sometimes – particularly in connection with dialectic – the idea of the distinction is to distinguish between true and reputable-but-not-true conceptions. This makes the methodological role of the reputable conceptions sound even more difficult because, basically, it means that we start our inquiry from falsities. When presenting the doctrine according to which appearances should be saved, Aristotle makes a remarkable methodological move, which connects the endoxic conceptions to his methodology. According to the basic formulation, it is the appearances (tail5,ela) which have to be saved which basically means that we should not let the force of our arguments lead us too far from how things appear to us.91 However, in the very same context (EN VII 1, 90

This is one of the reasons why I find Irwin’s suggestion (1988) that there should be a special truth-entailing sub-class of dialectic – the so-called ‘strong dialectic’ – artificial. More detailed criticism of Irwin’s suggestion is found in Kakkuri-Knuuttila (1993). 91 Cf. Nussbaum (1982, 277).

58

CHAPTER ONE

1145b1–7) Aristotle substitutes ‘reputable conceptions’ ((ldnma) for ‘the appearances’. This indicates that the reputable conceptions have something to do with how the world appears to us, i.e. perceptions. The problem with this is that the reputable conceptions include abstract philosophical views, the opinions of the wise. In a seminal article on the principle of saving the appearances, G. E. L. Owen argued (1961) that because Aristotle uses the notion of a tail5,elnl in the way that it allows both perceptions and endoxic conceptions, he is committed to a serious ambiguity with respect to the term. There is in Aristotle, on the one hand, an empirical or ‘Baconian’ meaning of ‘appearances’ connected to the use of the term in the Prior Analytics (I 30, 46a17–22), where it is used interchangeably with ‘experience’ (%,/ei.4a). On the other hand, there are shared appearances concerning human life such as our views on moral weakness (8j.aq4a), which Aristotle also calls ‘reputable opinions’. Owen concludes that only the former, not the latter, can be proper starting points for scientific inquiry. The latter can only be used as starting points for dialectic. However, as Martha Nussbaum has argued, the charge of ambiguity is based on the erroneous assumption that Aristotle uses the notion of a tail5,elnl to refer to hard empirical data unspoiled by interpretation.92 Nussbaum points out that rather than understanding tail5,elnl in this restricted sense and seeing it as in serious ambiguity with the (ldnmnl, we should note that the two are not that far apart. Even in the Physics, for instance, Aristotle’s discussion of time and place is based on shared conceptions. Even though Aristotle is clear that reputable conceptions suitable for dialectical evaluation cannot function as premises of scientific proofs as such, he never denies that they can function as a starting point for inquiry along with more strictly empirical appearances, such as observations concerning animal parts and behaviour. Aristotle’s general attitude towards reputable conceptions, namely the conceptions of the majority or the wise, or of the majority or the best of the wise, is that even though they are often distinguished quite sharply from true conceptions, they usually do contain the core of truth or are partly true. By the dialectical technique, we can show which parts or which conceptions are those that lead to problems and contradictions and which are the ones that do not. Understanding where the problems lie enables us to improve the conception we analysed. In addition, when we analysed a conception dialectically, many propositions were accepted and remained intact. They can be used to guide us to formulate a position that is better able to resist criticism. Rather than criticising Aristotle for not making his science empirical or experimental enough, I think we should note that he never proposes a strictly 92

Cf. Nussbaum (1982, 277).

THEORIES OF ARGUMENTATION

59

empiricist methodology. This is one of the important consequences of Nussbaum’s argument against Owen. Aristotle does not distinguish dialectic and science on the basis of the latter being empirical and the former being based on the predecessors’ theories. Both experience and predecessors’ theories are used as starting points for inquiry. This point will become clearer in the discussion below. Induction Let us now turn to discuss induction in Aristotle. As already mentioned, Aristotle distinguishes between dialectical syllogisms and dialectical inductions, and he excludes the possibility that a dialectical conclusion could be derived inductively. However, we also noted that Aristotle allows for inductive inferences within the dialectical context, and he emphasises their role in making the opponent concede some general principles that are needed in the final syllogism. In this section, I shall discuss the various forms of inductive inference that Aristotle talks about. Aristotle’s description of dialectical induction (%/acwc3) is very short, taking less than ten lines (Top. I 12, 105a12–19) in the Bekker numbering. At first we get the characterisation that induction is progress from particulars to universals (y 8/• r‡l jah’ )jaqrnl %/§ r¡ jah5kns (tndnp). Then Aristotle gives the following example. [I]f the skilled (%/iqr1,elnp) pilot is the best pilot and the skilled charioteer the best charioteer, then, in general, the skilled man is best in any particular sphere (I 12, 105a14–17; transl. Forster).

Aristotle’s example of dialectical induction strongly resembles the kind of inference that Socrates uses, e.g., in the Protagoras.93 In fact, Aristotle himself says that inductive inference was one of the things Socrates introduced into philosophy (Met. XIII 4, 1078b28–30). In the Protagoras Socrates questions the claim of virtue being teachable. At one point he needs the interlocutor to concede that whenever the Athenians want advice, they consult experts. For this conclusion he provides the following premises. [W]hen … the city has to take some action on a building project, we send for builders to advise us; if it has to do with the construction of ships, we send for shipwrights; and so forth for everything that is considered learnable and teachable (Prot. 319b–c; transl. Lombardo and Bell from Cooper (ed.) 1997).

The point of the argument is that in all the crafts that are supposed to be learnable and teachable there is a class of practitioners of that craft who are considered experts in that area. The argument is inductive in the way that its 93

Cf. Robinson (1941).

60

CHAPTER ONE

conclusion is general and it is argued for through particular instances. The idea is to provide the interlocutor with some instances that in his soul give rise to a generalisation: he realises that a similar conclusion applies to all the relevant cases. Aristotle’s dialectical inductions can be characterised similarly. In his example the questioner mentions two instances of the generalisation and then asks the general premises. If the generalisation is commonly accepted and plausible, the questioner can expect that the opponent cannot come up with a counter-example. And, as we have noted, when the opponent cannot provide a counter-example a dialectical inductive generalisation must be accepted. The inductive arguments in Plato and Aristotle have an important common feature: they do not start from particular individual cases as we nowadays take inductive inferences to do. Both in the Aristotelian dialectical induction and in Plato’s (or Socrates’) induction, the premises are already general. Aristotle, for instance, does not say that we take a particular person, e.g., Alcibiades, to be a skilled charioteer because he is knowledgeable and so on. Neither does Socrates say that when the Athenians want, for instance, to build a temple they go to Pheidias and so on. In both cases the dialectical inference starts from general classes such as charioteers and builders. In Plato and Aristotle the inductive inferences proceed from general types to still more general types; they do not start from individual cases.94 When discussing dialectical induction, Aristotle assumes that it is in principle possible to list all the relevant subtypes needed to establish the generalisation conclusively. He notes, however, that this is a very difficult task to perform (Top. VIII 2, 157a23–26) and is typically not done in dialectic.95 The assumption that it is possible to enumerate all the particular cases is based on the idea that the particulars in Aristotelian dialectical induction are all general subtypes, the number of which is finite. In fact, in the case of scientific induction Aristotle seems to assume that induction may be perfect. However, as we have observed, the acceptability of a dialectical induction is not dependent on whether all the subtypes are enumerated.

94

In the Posterior Analytics II 13 Aristotle does mention genuinely individual cases. Alcibiades, Achilleus and Ajax as well as Lysander and Socrates are listed as magnanimous human beings. However, the point of that example is not to make an inductive inference in the contemporary sense. Rather, he chooses the individuals to exemplify some general types of magnanimity and points to them only with respect to the common features these types of magnanimous people have. 95 Cf. Top. II 4, 111b5–10, where Aristotle instructs one to examine all the species of a genus in order to make general claims about the genus.

THEORIES OF ARGUMENTATION

61

Aristotle also treats induction in the Prior Analytics. The discussion is found towards the end of book II in chapter 23. He says that induction, that is, syllogism based on induction is an inference which ‘consists of proving the major term of the middle term by means of the minor’ (68b15–17). Such a claim sounds odd: a syllogism establishes the major term of the minor by means of a middle term. However, as we shall see, Aristotle names the terms here as they appear in the scientific proof, not on the basis of their positions in the syllogism from induction. To get a clearer picture of what is going on, let us consider Aristotle’s formal exposition and the example he gives. Aristotle describes the situation as follows. On the one hand, he says that we should consider a syllogism establishing that A belongs to all C through the middle term B. Now, the induction would be a syllogism showing that A belongs to all B through the middle term C. Both of these inferences are found in the following table. Syllogism

Induction

AaB BaC – AaC

AaC CaB – AaB

The terms Aristotle then recommends us to substitute for the inferences are the following: let A be long-lived, B bileless, and C all those which are longlived (jah’ )jaqrnl ,aj.5binl), such as human being, horse and mule. At some point we come to realise that our terms B and C are convertible: C belongs to all B and B belongs to all C. Then, given that C is convertible with B and that B is not broader than A,96 necessarily also A belongs to all B. In natural language we have the following two inferences which are equally valid. Syllogism

Induction

Being long-lived belongs to all bileless animal species.

Being long-lived belongs to all the following species: horse, mule and human being.

Being bileless belongs to all the following species: horse, mule and human being.

Being a horse, a mule, or a human being belongs to all the bileless species.

Therefore, being long-lived belongs to all the following species: horse, mule and human being.

Therefore, all the bileless species are long-lived.

96

This additional condition rules out the possibility of a counter-example that there are Bs outside of A.

62

CHAPTER ONE

As pointed out, given that Aristotle assumes that for such inductive inferences the minor and the middle terms need to be convertible, we will always in the case of induction have a kind of sister syllogism where the middle and minor terms have switched places. By converting the second premise and modifying the major premise accordingly, two syllogisms can be formed according to the tables. Aristotle’s manner of speaking reveals that the sister syllogism is an apodeictic syllogism (that is a scientific proof in the proper sense) and that he has the order of terms of the scientific proof in mind when he says that an induction proves the major term of the middle term by means of the minor. Towards the end of the chapter (68b30–37) he, for instance, distinguishes between induction and the syllogism in this case on the condition that the syllogism is ‘better known in nature’, whereas the induction is ‘better known to us’. This is his common way of distinguishing scientific proofs from other inferences. Therefore, in the context of the Analytics, induction is a very special type of inference where there are two convertible terms, in the above example B and C. The difference between the apodeictic syllogism and an induction is supposed to be that the premises of an induction are better known to us, whereas those of a proof in the proper sense are better known to nature. In Aristotle’s example, the premises better known to us are those stating, first, that horse, mule and human being are all long-lived animal species and, second, that all the animals belonging to these species are bileless. This makes up an inductive inference, whereas the proof in the proper sense goes as follows. All long-lived animal species are bileless; horse, mule and human being are all bileless animal species; therefore, horse, mule and human being are long-lived. Consider now the difference between these two inferences. In the induction the conclusion can be characterised as a result of scientific inquiry (all bileless animals are long-lived), and this statement is used as a premise in the scientific proof. In the apodeictic syllogism, on the other hand, the conclusion is a fact we are supposed to know initially before we inquire into the more detailed properties of long-lived species. The apodeictic syllogism concludes a list of all long-lived animal species, namely horse, mule and human being. This is typical of Aristotle’s definition of proofs in the proper sense. Their premises need to be less well known to us at the beginning and they need to explain the conclusion. Aristotle indicates at the end of his description of scientific induction that a syllogism from induction must be what we call perfect induction: all the relevant subtypes have to be taken into consideration. Hence the inference is deductively valid. Aristotle says that the term C, which in the induction appears as the middle term and which stands for the minor term in the scientific

THEORIES OF ARGUMENTATION

63

proof, needs to consist of all the subtypes (of B), because, he says, induction proceeds from all the cases (y c¡. %/acwc¢ di¡ /1lrwl, 68b28–29).97 But how are we supposed to know that these are all the cases, i.e. that B and C are convertible? Further, how do we know that there are no Bs outside of A? Aristotle’s discussion of induction in the Analytics does not pose these questions at all. Rather, the discussion presupposes that we know (or at leat that it is possible for us to know) all the relevant things. On that condition we can produce a syllogism from induction. If we consider the example we must know that human being, horse, and mule are long-lived animal species and that they are bileless; in addition, we must know that human being, horse and animal are all the bileless animals there are and that there are no bileless animals that would not be long-lived. Therefore, the conclusion does not provide us with new knowledge; everything is included in the preconditions. In addition, given that the relevant terms B and C are convertible and that we can produce the two syllogisms spelled out above, we must decide which one is the real proof and which one is inductive. The induction itself does not provide us with a criterion to make the decision. We could use as a criterion the general idea that the conclusions of the proofs are better known to us than the premises. The two conclusions are the following. ‘Being long-lived belongs to all the following species: horse, mule and human being’, and ‘all bileless animals are long-lived’. The former is closer to perception than the latter and, therefore, this indicates that we should consider that one as real proof. In addition, Aristotle perhaps assumes that if we know the basic principles of the science concerned with the generation and corruption of animals we understand that bile is the organ that produces heat in the animal. Because internal heat needs to be cooled down by the respiratory system this system is quickly exhausted if there is too much heat. Therefore, only animals which naturally do not have that much bile in them, can maintain life functions longer. However, it is important to note that the syllogisms themselves are not meant to settle these questions. Much is left to our intellectual insight: we have to understand how the premises explain the conclusions in order to recognise them as premises of scientific proofs.98 97

However, at the beginning of the relevant passage (68b15–29) the condition of being perfect is not mentioned. Rather, the relevant species, human being, horse and mule, are simply listed as examples (introduced by the expression ‘such as’) of longlived animals. However, the end of the passage makes quite clear that in order to actually produce the syllogism we must have a complete list of the subtypes. 98 Cf. also Hintikka (1980). He points out (on p. 433) that the references to complete induction are misleading in the sense that they give one the idea that the completeness of the lists of subtypes should be a guarantee of the truth of the induction. He says that the completeness is rather a means to find the relevant convertible term.

64

CHAPTER ONE

We have now seen that what Aristotle means by induction in the scientific context is an inference which has the same terms as the proof proper but in a different order. Therefore, he allows that at least the major premises of the scientific proofs can be validly deduced from other scientific statements in a way Aristotle characterises as induction. Given that this is the case, the premises of this apodeictic syllogism are not completely unprovable but can be validly argued for even from true premises. They still remain unprovable in the strict and technical sense Aristotle gives to this expression. We will discuss this kind of unprovability in 1.2.2 below. Aristotle also uses the term ‘induction’ (%/acwc3) once in the difficult last chapter (II 19) of the Posterior Analytics (100b4). At the end of his account of how we acquire starting points for sciences (100b3–5) he claims as a conclusion that it is clear that also ‘the first premises [of scientific proofs] must be known through induction’ (y,‹l r¡ /.‡ra %/acwc· clw.4feil 8lacja‹nl), because ‘perception installs the universals in the aforesaid way in us’ (y a©qhgqip n÷rw r• jah5kns %,/nie‹). As I shall argue in more detail below, Posterior Analytics II 19 points to a natural cognitive process, through which we come to have correct universal contents in our intellectual soul. This is the reference of the just-quoted phrase ‘perception installs the universals in the aforesaid way in us’. Aristotle uses this as a reason for saying that the principles of sciences must be known through induction. As will be argued below, it is not likely that this should apply to the principles in the sense of starting points for inquiry. Rather, the very cognitive process which explains why and how we come to have universal contents in our soul at all, explains how we come to know such basic facts that enable us to initiate inquiry. Therefore, the principles that are supposed to be known through induction are the real explanatory premises of the scientific proofs in the proper sense. As we have just seen, the major premise does appear as a conclusion of an inductive syllogism. It is possible that a student who did not know it before may come to know it through inductive syllogism. However, as we also saw, at least the teacher must know it beforehand in order to rule out possible counterexamples. The teacher’s knowledge of the connections and the order between the relevant general terms is based on an insight into how the phenomena are explained. This, I have suggested, is the basic cognitive mechanism through which we can find explanatory syllogisms and form inductions. The first transition from perception to universals does not come about through questions and explicit inquiry; rather, children learn basic universals quite naturally, just as they learn to walk.99 However, after we have reached

99

This example appears in Alexander of Aphrodisias; see below 2.3.2.

THEORIES OF ARGUMENTATION

65

the level of the basic universals, we start to inquire into their definitions and explanations. These can be found through a process where relevant connections between general terms are grasped as essential and explanatory. When they have been grasped, they can be put forward in the form of a scientific syllogism from induction. Aristotle seems to assume that our intellect is naturally capable of finding out essential and explanatory connections. We have now taken a look at Aristotelian inductive inferences. The most remarkable conclusion we can draw from this is that none of the types of inference listed here is inductive in the sense we would normally speak about induction, namely as a generalisation proceeding from particular individual instances. All the premises are already general. For instance, in dialectical induction the premises involve types of expertise (captains and charioteers), in scientific induction natural species and their general properties (mules, human beings and horses and their being long-lived). In addition, we have seen that in the scientific context inductive syllogisms presuppose intellectual insight into connections between the relevant terms involved. Aristotle’s account does not explain how this comes about; he seems to assume that our intellect is in principle capable of recognising explanatory connections when we have found the correct terms in the inquiry. Induction, as Aristotle conceives it, is not a method of inquiry. Conceptual Analysis In the Posterior Analytics Aristotle introduces a way of finding the predicates that essentially belong to a subject and make up its definition. He indicates that Plato’s collection and division is a forerunner of this procedure (see, e.g., An. Post. II 5, 91b13–15). I shall call Aristotle’s recommendations for finding the predicates belonging to the definition ‘conceptual analysis’. It is to be noted from the very start that the notion of a concept is precarious when ancient evidence is concerned, particularly with Plato and Aristotle. I shall discuss the problems more closely in 2.1. At this point it is sufficient to note that Aristotle’s procedure is not analytic in our sense. Rather, the aim of the analysis is to find predicates that are necessary for the subject in the following sense: if the subject were to lose some of those predicates, it would not be the same thing anymore. Therefore, even though we can, lacking a better alternative, talk about conceptual analysis here, we need to bear in mind that the analysis is supposed to be concerned with things themselves, e.g., what magnanimity is. Now, if we want to find out a definition, Aristotle recommends that we proceed as follows. First, we must limit our inquiries to the genus at hand (e.g. II 13, 96a24–26). For instance, if we look for the definition of the number three, we must restrict the analysis to predicates that belong only to

66

CHAPTER ONE

numbers. Inside the genus one must look for predicates which themselves belong to larger groups than the subject of inquiry. For instance, if we search for the definition of the number three, oddness is such a predicate. It belongs to the number three but it also belongs to other odd numbers. In addition, Aristotle mentions two further predicates which he takes to be part of the definition of the number three. He calls both of them ‘being first’ (/.‡rnp) and explains them as ‘not being measured by numbers’ (®p ,¢ ,er.e‹qhai 8.ih,+) and as ‘not being composed of numbers’ (®p ,¢ qscje‹qhai %m 8.ih,‡l). The latter means that three is a prime number, i.e. it is not a product of two numbers. The former is not equally clear but it seems to mean that three is not a sum of two numbers either, because in Aristotle’s time one was not taken to be a number in the same sense as other numbers but as a kind of unit for measuring that is presupposed by all numbers.100 Another example Aristotle discusses is that of magnanimity (,ecaknvsu4a, An. Post. II 13, 97b16–27). The example bears a strong resemblance to Plato’s analysis of love by collection and division in the Phaedrus, discussed above in 1.1.101 Both in the Phaedrus and in Aristotle’s example some phenomena are first located under a more general type. In the Phaedrus love is taken to be a kind of madness, in Aristotle’s example Alcibiades, Achilleus and Ajax as well as Lysander and Socrates are recognised as magnanimous men.102 In the Phaedrus categorising love under madness entails that love has some bad effects on the lover even though love is a god – and in fact one of the best things gods give to human beings. This causes problems that are solved when we notice that there are two subtypes of madness, namely human mental disorder and godly inspiration, and that love belongs to the latter not to the former. In Aristotle, on the other hand, Alcibiades, Achilleus and Ajax are considered magnanimous because they do not tolerate dishonour. Lysander and Socrates, in turn, are indifferent to fortune. Therefore, it seems that the two cannot belong to the same category. However, Aristotle mentions two ways out. Either magnanimity in them all is explained by some still more general property, or there are two kinds of magnanimity (as there are two kinds of love in the Phaedrus). Aristotle leaves the matter open, however. Aristotle does not provide clearer instructions about how we are supposed to find the predicates to be listed in the definition. One possibility is that we can find them out by, e.g., the kind of thought experiments he recommends

100

It was called the ‘principle of numbers’ (8.u¢ 8.ih,nfl); cf. Ross (1949, 656) and Heath (1949 (reprinted 1998), 83–84). 101 Also Kakkuri-Knuuttila and Knuuttila (1990, 297) mention the similarity. 102 Cf. above footnote 94.

THEORIES OF ARGUMENTATION

67

elsewhere (I 5, 74a37–b4) in the Posterior Analytics. Strip the property out you proposed as essential. If the thing can be thought to be the same thing without that property, then the property is not essential. For instance, being made of bronze can be stripped out from a triangle, but not being a plane figure or having the sum of the internal angles equal to two right angles. Similarly, the number three, for instance, can be thought of without being the number of, e.g., someone’s siblings, but it cannot be thought of without being odd. It has been pointed out that Aristotle’s remarks on searching for explanatory terms in definitions can be used to complement his analysis of induction as a means of finding the premises of scientific syllogisms.103 Concerning definitions Aristotle does not use the term ‘induction’ (%/acwc3), but his procedure resembles the description of the inductive syllogism. In both cases he aims at finding coextensive general terms so that one explains why the other belongs to the subject term in question. There are plenty of examples of this in biology as well.104 Both the Platonic and the Aristotelian analyses aim to uncover coextensive general terms and are based on the assumption that all the subtypes and all the relevant common features should be listed. In addition, Plato and Aristotle assume that the fact that contradictions follow is an indication that the analysis is not yet complete. Aristotle, however, criticises the Platonic method. The main point of his criticism is that the Platonic division is haphazard because it contains no structural background assumptions that could guide the procedures of collecting and dividing. The method can lead to correct results but this can in no way be guaranteed. Aristotle is ready to use the method of collection and division in finding out definitions, but he insists on certain improvements (An. Post. II 5; 13). Most importantly, he requires that the divisions be restricted within one genus at a time and that the differences within the genus in question be taken in the right order. He assumes that the differences do have a definite order. However, Aristotle’s version of conceptual analysis does not provide any explicit method of checking, for instance, whether we have taken the characteristic properties in the correct order. He seems to assume that consistency and explanatory power serve as our guides in searching for the correct definitions. All the strategies of argumentation that Aristotle has inherited from the Academy are based on the principle of non-contradiction. The conclusion

103

See Hintikka (1980) as well as Kakkuri-Knuuttila and Knuuttila (1990). See, e.g., the discussion concerning the number of stomachs in sanguineous viviparous animals (Part. An. III 14, 674a23–b17); for a discussion, see Lennox (2001b). 104

68

CHAPTER ONE

follows from the premises accepted by the interlocutor precisely because otherwise he would have to accept an explicitly contradictory set of statements. This is how the opponent can be forced to the intended conclusion. By contrast with Plato, whose sophistic adversaries also endorsed the principle of noncontradiction,Aristotle is concerned (in the Met. IV)105 with adversaries who do not accept even the principle of non-contradiction. Aristotle admits that it is in fact impossible to argue with such people, although he still tries to show that they are in fact implicitly committed to the principle. As noted above in 1.1., Aristotle’s position is a little awkward here. He argues against those who deny the principle of non-contradiction but his arguments at the same time presuppose this principle. Perhaps he thinks that the principle of non-contradiction is such a fundamental principle of all reasoning that its denial has to be somehow addressed. Except for induction in the scientific context, it is characteristic of all the strategies of argumentation discussed in this subchapter that they are not based on any special conception of argumentative form. In this respect they deviate from the syllogistic figures of the Prior Analytics.106 In the Prior Analytics Aristotle even goes as far as to claim that all valid arguments can be presented in the figures. As is well known, this claim is highly problematic.107 Problems related to this claim were already at stake in antiquity when the Stoics asserted that their logic was stronger than syllogistic logic in this respect.108 In any case, the theory of syllogistic figures in the Prior Analytics can be seen as a theory of valid arguments and, hence, it resembles a genuine logical theory. However, like the analysis of arguments by topoi, syllogistic analysis is based on similarities between actual arguments, not on pre-existing concepts of logical form and validity. 1.2.2 Science We have been dealing with the techniques of argumentation Aristotle inherits from the Academy and shall now turn to discuss Aristotle’s conception of scientific proofs and their premises. This topic has been much discussed in the 105

For Aristotle’s arguments, see Kirwan (1971) and Dancy (1975). It seems likely that Aristotle invented the syllogistic figures later when the Topics had already been completed. However, Smith (1994), for example, has analysed the differences between the Prior Analytics and the Topics without a reference to chronological considerations. I agree with Smith that chronological claims are difficult to confirm and should not be given too much weight. 107 For a discussion, see, e.g., Striker (1998). 108 See, e.g., M. Frede (1987b). 106

THEORIES OF ARGUMENTATION

69

literature109 and is considered highly problematic. However, I believe that a general picture of the most distinctive features of Aristotle’s theory can be drawn. I shall attempt to do this in the following, and at first I shall consider Aristotle’s distinction between two senses of the phrase ‘better known’. Before entering the discussion, however, a comment needs to be made on the term ‘science’. As is well known, there is no separate Greek word for knowledge and science, for both are translations of the Greek %/iqr3,g. In addition, our ‘science’ can be considered a misleading translation of %/iqr3,g for several reasons. For instance, science as today’s institutional and methodologically specialised activity did not exist in Aristotle’s time and many of the questions central in philosophy of science today had not been raised in the form we know them now. However, it is very difficult to find an alternative term translating the %/iqr3,g in the Posterior Analytics.110 If, for instance, we took %/iqr3,g to mean knowledge, this would be, perhaps, even more misleading than to talk about science in connection with Aristotle. The Posterior Analytics is not mainly an epistemological work as we understand epistemology today: it is not seeking to establish conditions that all knowledge claims must satisfy, as opposed to mere beliefs. Rather, Posterior Analytics to some extent deals with questions like how natural phenomena are explained and how objects of systematic inquiry are defined, and these questions are much more related to the scope of the philosophy of science than to epistemology. Therefore, I think that it is better to use them term ‘science’ after all, but we must keep in mind that Aristotle’s science was quite different from the science of today. Being Better Known In the previous subchapter we discussed Aristotle’s general conception of dialectical argumentation. We noted in that connection that dialectical arguments carry an intrinsic epistemic aim. The premises should be initially more acceptable or better known than the conclusion so that accepting them makes the conclusion more credible.An argument, even if it were valid, does not serve its purpose unless this condition is fulfilled. Typically, Aristotle assumes that 109

See, e.g., Berti (ed.) (1981), Barnes et al. (eds.) (1975) vol. 1, Gotthelf and Lennox (eds.) (1987), Irwin (1988), and Lennox (2001b), to mention just a few. 110 I think that Burnyeat’s suggestion (1981) that %/iqr3,g should in the An. Post. be taken to mean ‘understanding’ does not help us out either. If we say that Aristotle is talking about understanding, what are we supposed to understand? If we say that we should understand natural phenomena, how should we understand them? We should understand them by explaining and defining them. But should this not be what scientists are concerned with?

70

CHAPTER ONE

the conclusion of an argument is known beforehand. The conclusion is not known in the very strict and technical sense of knowledge (%/iqr3,g) Aristotle employs in the Posterior Analytics because such knowledge depends on knowing the proof.111 We must, then, have some other kind of preliminary ‘knowledge’ of it. Aristotle sometimes uses the Greek cl‡qip to refer to modes of knowledge in a loose sense. Always when Aristotle says that something is ‘better known’ than something else, we must bear in mind that he in several places underlines that there are two ways to understand this phrase. According to Aristotle, we must distinguish between what is better known to us and what is better known by nature (clw.i,Íre.nl y,‹l, or being prior to us, /o•p y,¡p /.5re.nl, and being better known [in nature], clw.i,Íre.nl [r· t6qei], or being prior in nature, /.5re.nl r· t6qei; see An. Post. I 2, 71b33–72a5, An. Pr. II 23, 68b35–37, Top. VI 4, 141b3ff., Phys. I 1, Met. VII 3, 1029b3–12, EN I 4, 1095a30–b5). Typically, what is better known to us is quite the opposite of what is better known in nature (see, e.g., An. Post. I 2). When the premises of dialectical arguments are called ‘better known’, it means that they have to be better known from the interlocutor’s perspective. The purpose of such an argument is to make the conclusion appear credible to the interlocutor on the basis of the premises accepted in the course of the discussion.

111

Aristotle has technical terms for knowledge in the Posterior Analytics. %/iqr3,g without qualifications (9/k‡p) is defined in An. Post. I 2 (71b9–12) as involving knowing (ciclÍqjeil) (i) that the reason why something is the case or something exists (di’ {l r• /.Øc,1 %qril) and (ii) knowing (ciclÍqjeil) that this really is the reason for it being the case or for the thing to exist, and (iii) knowing (ciclÍqjeil) that the thing could not be otherwise. The term %/iqr3,g in the Posterior Analytics denotes mainly the knowledge of the conclusions of scientific proofs in the way that also the premises are known in another sense. Sometimes (e.g. 71b20, 73a22 and 74b5) %/iqr3,g is qualified with the adjective 8/ndeijrij3 to make clear that it is knowledge of a proved conclusion. We also find (in e.g. 88b36) %/iqr3,g 8la/5deijrnp, knowledge of the unprovable premises. %/iqr3,g, however, can in Aristotle also refer to knowing scientific facts by means of a valid non-apodeictic syllogism (%/4qraqhai r• çri, An. Post. I 13, 78a22). lnflp is the usual term for knowledge of unprovable principles (An. Post. II 19), but the term is also used in a broad sense to refer to thinking in general (e.g. in DA III 4). cl‡qip (from the same root as the verb ciclÍqjeil in the definition of %/iqr3,g in the unqualified sense just quoted) refers to knowledge in a rather loose sense, mainly to types of knowledge which do not fulfil the conditions for %/iqr3,g. Often Aristotle uses the term cl‡qip for perceptual knowledge of contingent facts (see, e.g., An. Post. II 19, 99b38–39). For a discussion of Aristotle’s terminology concerning knowledge, cf., e.g., Burnyeat (1981).

THEORIES OF ARGUMENTATION

71

Aristotle’s distinction can be characterised as one between the order of things and the order of knowledge. Things are a certain way and have some permanent and necessary connections with other things quite independently of our conceptions of them. The fact that there are such permanent and necessary connections between things on a general level implies that reality has an intrinsic arrangement or order; that is the order of things in nature. In addition, there is a general order in which human beings normally learn things. Typically, we first learn some facts by making perceptions. After grasping that some facts are such that they are permanent and necessary or permanently recurring, we come to ask their explanations. We also learn in virtue of our intellectual capacity that there are general kinds of things which have some permanent properties. Learning this leads us to ask what their essential permanent properties are. These questions lead us to ascend the level of perceived facts and to ask about the more general permanent structure of nature. Aristotle is rather optimistic in the sense that he assumes that our inquiries guided by the two kinds of question just mentioned often lead to rather good results. We can come to find the explanations for regularly recurring natural phenomena and to understand the permanent necessary properties of things as properties which make those things the kinds of things they are. By ascending from the level of perception in the way just described we learn explanations and facts about the natures of things which were not known to us before. We learn them through knowing the facts we initially got to know. In this way we move from grasping what is better known to us to apprehending some aspects of what is better known by nature. Aristotle also calls what is better known by nature ‘prior in nature’ (/.5re.nl r· t6qei) and characterises such priority as follows. Something is prior to another thing in nature if the latter cannot exist without the former existing. For instance, a genus is prior to its species in this sense. If there are, say, animal species such as human beings, there are animals as well. It is imaginable that things of a certain genus, such as animals, could exist without creatures in one of the species (e.g., bees) existing.112 In Aristotle’s view explanatory factors are also prior to the phenomena to be explained by these factors.113 Truths that are prior in nature must, in Aristotle’s view, be expressed in the premises of the proofs, whereas the conclusion must be better known to us. We shall return to Aristotle’s distinction between the two ways of being better known below in connection with the conditions for the premises of scientific proofs. At this point it is sufficient to note that Aristotle’s distinction

112 113

The genus, however, could not exist if none of the things in its species existed. For different senses of the notion of priority in Aristotle, see Cleary (1988).

72

CHAPTER ONE

is based on an important feature of his metaphysical realism. We have noted already that Aristotle is a realist: according to him things are as they are quite independently of our conceptions about them. In addition, his realism involves the assumption expressed also in this distinction that things have a certain intrinsic order. A thing, say a, is prior in the relevant sense to another thing, say b, if a explains b, or if a is necessary for the existence of b in the sense that a is a part of what it is for b to be the kind of thing it is.114 In addition, Aristotle also assumes that the intrinsic order of reality is in principle knowable to us. As the distinction between the orders of human knowledge and the order of things makes clear, however, the order of things is by no means obvious to us. Rather, through things familiar to us – perceptible facts and basic generalisations – we come to learn some more basic truths about reality which, in turn, explain the facts we at first observed. The distinction between what is better known to us and what is better known in nature implies an intrinsic directedness within any structure of argumentation in Aristotle. We can either establish something less well known to us starting from what is well known to us, or we can establish something that is well known to us on the basis of premises prior in the order of nature. The former helps us establish the premises of the latter and only the latter counts as a proof in Aristotle’s strict and technical sense. In the Nicomachean Ethics, in one of the passages where Aristotle discusses his distinction, he compares the situation with a Greek stadium where the runners start from the judges, run towards a turning point, turn around and come back to the judges again. One of these ways corresponds to our ascent from the level of perception towards more general and explanatory truths; the other corresponds to the proof starting from such truths. In that context Aristotle also notes that Plato had recognised this important difference between argumentation towards some basic principles and argumentation from them. We have seen above that possible sources for the distinction in Plato are the method of hypothesis and conceptual analysis by collection and division. Premises of Scientific Proofs Up to this point I have mentioned but not yet discussed in detail the conditions for the premises of Aristotelian scientific proofs. In what follows, I shall first discuss these conditions. Then I shall discuss what the proofs are like by means of examples. I have also referred to a curious feature of Aristotle’s theory, namely the fact that he, on the one hand, claims that the premises are unprovable and, on the other 114

For Aristotle’s notion of explanation, cf. Moravcsik (1991). I shall discuss this notion more closely below in connection with the conditions set on the premises of scientific proofs.

THEORIES OF ARGUMENTATION

73

hand, allows various ways in which they are validly argued for even from true premises. The following discussion enables us to see more clearly what these prima facie rather problematic aspects of his theory amount to. I shall use Aristotle’s own examples in the discussion. This reflects the fact that I think, contrary to some scholars,115 that Aristotle’s examples in the Posterior Analytics are examples of the sort of premises he assumes we should aim at finding in scientific inquiry. I do not say that all of the examples would fulfil all the conditions for the premises of proofs. Rather, I think that when talking about the premises of proofs, Aristotle means something like those premises used in the examples. Another question is to what extent Aristotle himself proceeds according to his own methodological instructions in his scientific and philosophical works. Towards the end of this subchapter I shall also touch briefly on that question. Yet another question is to what extent Aristotle in fact succeeds in finding premises that satisfy his strict conditions. This question falls outside the scope of the present treatise. Proofs as Syllogisms. Aristotle’s scientific proofs are syllogisms in the sense of the syllogistic figures of the Prior Analytics. In science we most often need only the first figure universal affirmative syllogism also known in the medieval terminology as Barbara (cf. An. Post. I 14). It is useful to note from the very start that when Aristotle talks about proofs, he gives examples of two kinds of syllogisms. The first one can be called explanatory (or even causal, if we keep in mind that what Aristotle calls causes are to some extent different from our conception of causality), whereas the second one expresses parts of the genus-species structure with specific differences.116 Consider two examples of both types. Explanatory proofs: A [celestial body] which is shadowed by the earth is eclipsed. The moon is shadowed by the earth.

115

E.g., Barnes (1969, 124). Lennox (1987 and 2001b), for instance, has argued that in fact the syllogisms expressing the genus species structure do carry explanatory force for Aristotle. In biology in particular Aristotle often explains the animal organs and their behaviour by noting that they belong to a more general class of animals which share a common nature as those kinds of animals. I agree with Lennox. I do not want to deny that the syllogisms expressing genus-species structure can also in some sense be explanatory. 116

74

CHAPTER ONE

The moon is eclipsed. (An. Post. II 8, 93a30–31.)117 Those [celestial bodies] that are close to the earth do not twinkle. Planets are close to the earth. Planets do not twinkle. (An. Post. I 13, 78a39–b2.)118 Proofs expressing genus-species structure, ‘explicatory proofs’: All two-footed creatures are animals. Human beings are two-footed creatures. Human beings are animals. (See An. Post. II 4, 91a28–32 and II 14, 98a9–11.)119 All [things] having the sum of their internal angles equal to two right angles are [plane] figures. All triangles have the sum of their internal angles equal to two right angles. All triangles are figures. (Cf. I 4, 73b31.) These two types of syllogisms correspond to the scientific questions Aristotle mentions in the first chapter of book II of the Posterior Analytics. There Aristotle says that we can inquire into basically four things: the fact (r• çri), the reason (r• di5ri), if something is (eå (qri) and what it is (r4 %qril). The questions are clearly divided into two pairs: (a) the fact and the reason and (b) if something is and what it is. In both pairs we have to have answered the first question in the affirmative before initiating inquiry. We must have found out that something is the case (the fact, r• çri) or that something exists (eå (qri). In the examples this means that we have found out that the eclipse of the moon takes place or that planets do not twinkle, that human beings are animals or that triangles are plane figures. After that we can proceed to the 117

The terms of this syllogism are: A eclipse, B being shadowed and C the moon. The order of the terms is literally the following: eclipse belongs to being shadowed by earth; being shadowed by earth belongs to moon; eclipse belongs to the moon. However, it is more understandable for a reader not used to Aristotle’s manner of speaking to put it in the way I have done in the text. 118 Here the terms are: A non-twinkling, B being close to the earth, and C planets. A literal reading of the syllogism is as follows. Non-twinkling belongs to those [celestial bodies] that are close to the earth; being close to the earth belongs to the planets. Non-twinkling belongs to planets. In the example Aristotle treats the negative predicate non-twinkling in a rather straightforward manner as a major term in the universal affirmative first figure syllogism. This is a little suspicious, of course, but reflects Aristotle’s confidence in the universal affirmative first figure syllogisms in science. 119 This certainly is not a full definition of a human being according to Aristotle. However, it serves the purpose of illustrating how the definitions are related to the syllogistic structure.

THEORIES OF ARGUMENTATION

75

questions concerning the reason and the essence. We shall ask, for instance: ‘why is the moon eclipsed?’, ‘why don’t the planets twinkle?’ or ‘what kind of animals are human beings?’ and ‘what kind of plane figures are triangles?’ Basically, these are the kinds of questions Aristotle assumes we try to answer in scientific inquiry. It is possible that Aristotle’s conditions for the premises of scientific proofs are so strict because he wants to characterise premises that provide a complete answer to scientific questions. By ‘complete answer’ I mean here that when the explanatory and essential features have been articulated we cannot go on asking the same questions in the scientific sense any longer. I shall now turn to discuss the conditions. According to Aristotle, scientific proofs in the proper sense are a very limited subclass of affirmative universal first-figure syllogisms.120 The unprovable premises are qualified by strict criteria, which are even said to ‘take one’s breath away’.121 First of all, the premises of Aristotelian proofs in the strict sense are true (An. Post. I 2, 71b21). Truth, however, is not enough to distinguish unprovable premises from other scientific statements, but it is a necessary requirement for them. When presenting his conception of proofs, Aristotle does not concentrate on the question of how we can know whether scientific statements are true or not. Neither does he discuss the question of what the truth of a proposition consists. He simply says that the premises should be true because falsities cannot be known (An. Post. I 2, 71b26–27). Aristotle’s requirement of truth is related to his expectations concerning knowledge (see, e.g., An. Post. I 4, 73a21–23). He in fact expects that in addition to the simple condition of truth, the objects of knowledge have to be permanently true and cannot be otherwise. This aspect, in turn, is related to the typical ancient Greek way of analysing statements and their truth-values.122 For Aristotle, if we say that Socrates is white, this statement becomes false if Socrates gets tanned on the beach. A statement which can turn false in this way is not, according to Aristotle’s strict conditions, sufficient to qualify as an object of knowledge in the strict sense of %/iqr3,g. Such statements can only be known in a more loose sense. In fact the requirement that known

120

It has been pointed out that Aristotle also uses the term ‘proof’ (8/5deimip) in a loose sense so that it can also refer to, e.g. rhetorical arguments, see G. E. R. Lloyd (1990). Here, however, we are concerned with the notion of proof in the strict sense as scientific syllogism (qskknciq,•p %/iqrg,nlij5p) as Aristotle characterises it in An. Post. I 2 (71b17–18). 121 Lloyd (1990, 376). 122 A similar analysis also appears in Stoicism.

76

CHAPTER ONE

statements must be true already carries the expectation of necessary truth. We shall shortly discuss the requirement of necessity below. Other requirements for unprovable premises are the following. Unprovable premises are first (/.‡rnp) and immediate (!,eqnp); they are also better known than (clw.i,Íre.np), prior to (/.5re.np), and they express causes of (a©rinl) the conclusion (An. Post. I 2, 71b21–22). In addition, the premises are necessary (8lacja‹np, I 4, 73a24). When explaining the requirement of necessity, Aristotle explains it with the expressions ‘in every case’ (jar¡ /alr5p), ‘in itself’ (jah’ a∫r5), ‘as itself’ (¯ a√r5, 73b29) and ‘universally’ (jah5kns) (I 4, 73a26–27) or, rather, ‘firstly universally’ (/.Írns jah5kns, I 5, 74a14). Priority and Explanatory Power. First I shall pick up those conditions that are related to the distinction between the order of nature and the order of human knowledge we discussed above.123 These are the ones concerning priority (/.5re.np, /.‡rnp) and being better known (clw.i,Íre.np). The premises of scientific proofs are better known in the order of nature. In this connection Aristotle identifies the two conditions of being prior and being better known. He notes that these two expressions can be used in two ways. What is more familiar to us is closer to perception, whereas in the order of nature better known things are at a distance from the immediately perceived. Often in fact the examples of proofs are such that the conclusion can be perceived. For instance, we perceive that the planets do not twinkle, that the moon is eclipsed, or that there is noise in the clouds that we call thunder (for the last mentioned example, cf. II 8, 93b10–13). Perception together with a very basic intellectual capacity is sufficient to grasp that human beings are animals, triangles are plane figures and so on. Let us return to the example of planets and the fact that they do not twinkle; this is the one Aristotle himself uses to illustrate the requirement of priority later on in the treatise (I 13, 78a32–b2). Consider two syllogisms having the same terms but in a different order. 123

There has also been discussion whether the premises of proofs become evident to human beings after they have learnt specific sciences. Burnyeat suggests (1981, 130) that these principles become ‘second nature’ to people that through intellectual habituation acquire theoretical virtue. Burnyeat’s suggestion has been further developed by Smith (1993). For comments on Smith, see Hintikka (1993). I think that %/iqr3,g also refers to the state of the knowledgeable person; such a person is familiar with the general principles of sciences and can ask scientific questions. It is not farfetched to say that what is better known by nature becomes what is better known for such a person.

THEORIES OF ARGUMENTATION A non-twinkling B being near C planets

77

A being near B non-twinkling C planets

From these two configurations of terms we can form two syllogisms. Non-twinkling belongs to [celestial bodies] that are near. Being near belongs to planets. Non-twinkling belongs to planets.

Being near belongs to [celestial bodies] that do not twinkle. Non-twinkling belongs to planets. Being near belongs to planets.

Aristotle considers the premises of both of these syllogisms to be true and even to be necessarily true. However, only one of them is a proof in the strict sense. It is found in the left hand column. In the right hand column the syllogism has true premises and derives its conclusion validly but is not a proof in Aristotle’s sense because its premises are better known and prior in the wrong sense; they are such in relation to us, not in the order of nature. The real proof concludes that the planets do not twinkle because they are near, whereas the other syllogism concludes that the planets are near because they do not twinkle.124 Aristotle says that the syllogism in the right hand column is one that establishes a fact (qskknciq,•p rnfl çri), namely that the planets are near.125 The fact that planets are near (compared to other celestial bodies) is not familiar to us in everyday perception. Aristotle allows that such facts can be established by syllogisms of this kind. Therefore, to say that the premises of Aristotelian scientific proofs are unprovable is to say that they are unprovable in the technical sense. They can be argued for, even by a syllogism with true and necessary premises, but they are unprovable in the sense that such an argument never counts as a proof proper. In the case of the planets and their non-twinkling the reason for this is that the major and the middle term are in the wrong order; the premises are only better known to us, not prior in the order of nature. This is why they also fail to express the real reason for the conclusion.126 124

Cf. Charles (2000, 201) who discusses the passage briefly and emphasises the interdependence between a syllogism expressing a cause and a definition. It is to be noted that even though I distinguish between two types of syllogism here, I do not want to deny that definitions could be read from the explanatory syllogisms; cf. below. 125 Real proofs are called syllogisms of the reason (qskknciq,•p rnfl di5ri). The contrast is made in An. Post. I 2 (78a36–37 and 78b2). 126 Cf. I 2, 71b31, where Aristotle also connects the requirement of being prior in the order of nature with that of being explanatory.

78

CHAPTER ONE

In syllogisms like the one concerning the non-twinkling of planets the requirement of priority is connected with explanatory power. In such cases the link is fairly easy to understand. The nearness of the planets is a reason for their light being stable, not the other way around. Similarly, the earth’s interposition between the sun and the moon is the reason why an eclipse takes place. However, we also noted earlier that sometimes, when Aristotle talks about proofs (that is apodeictic syllogisms), he means syllogisms which express a subject’s place in the genus-species structure. In such a case the middle terms express specific differences and the major term the genus to which the subject belongs. In that kind of case it is at first sight not clear how and why the notion of priority should be connected with explanation. Why should a specific difference (such as having the sum of the internal angles equal to two right angles) explain that the species belongs to a genus (that triangles are plane figures)? The condition of priority does have a fairly straightforward interpretation in Aristotle that is closely connected to genus-species syllogisms. I noted above that Aristotle says that the genera are prior to their species in the way that it is conceivable that animals exist without one or some – though not all – of its species existing. However, an animal species could not exist without at the same time there being animals. The condition of priority can be taken in this sense in connection with syllogisms expressing the genus-species structure. We have also noted earlier in connection with inductive syllogisms that in the Aristotelian framework common features in general are considered explanatory. For instance, the fact that dolphins are mammals and hence viviparous, is supposed to explain why they have bones as opposed to a fishspine. Similarly, the bilelessness of certain animal species is assumed to explain their longevity.127 In such explanations, however, the conclusion is not that a species belongs to a genus. The conditions of being prior and being better known in the order of nature are comparative in the sense that they locate the premises of the proofs above the conclusions in the order of nature. However, Aristotle also adds an absolute condition according to which the premises are not only prior (/.5re.np) relative to the conclusion but first (/.‡rnp). He is clearly talking about the order of priority in nature again. Being first in the order of nature would then mean that we have found a premise above which there are no prior premises. Aristotle claims that all proofs end sooner or later; they cannot be continued infinitely, but, as

127

For a discussion of the relation between definitions and explanations, see Charles (2000).

THEORIES OF ARGUMENTATION

79

his comparative conditions make clear he allows for proofs which establish their conclusion on the basis of premises which themselves can be proved. However, at some point we arrive at first premises. To say that there are premises which are first in the order of nature seems difficult. How far, according to Aristotle, do we have to go in searching for such premises? If we think of the examples Aristotle gives, particularly the one concerning the non-twinkling of planets, it actually seems rather difficult to say what kinds of steps we should add in order for the syllogism to have premises that are not only prior to the conclusion but also first in the sense that there is no further explanation to be given for it. One aspect of this question is related to the strict limits Aristotle gives for the sciences. He allows that some premises that are unprovable in some sciences can appear as conclusions in others. This is particularly the case with sciences applying mathematics, such as optics and harmonics.128 It might be the case that Aristotle supposes that there are some optical results providing us with the additional information we need to claim that all celestial bodies close to the earth do not twinkle. However, it is also possible that Aristotle has something like the following in mind. If we look at distant lights at sea, they seem to twinkle, whereas those that are close seem to remain stable.129 The explanatory power of the planet example might be based on a generalisation of this observation: distance in general is considered explanatory of the twinkling of lights. Be that as it may, I think that it is vital to recognise that even though Aristotle’s examples are not perfect, they are at least something like the kind of premises he means when he talks about the premises of scientific proofs. In addition, I do not think that we should look for support for Aristotle’s example of non-twinkling from today’s physics or astronomy. Rather, I think we must refer to something which was accessible to Aristotle when putting forward that example. An example would be the observation that when we look at a distant lighthouse, for instance, its light seems to twinkle. When we come closer, we at some point notice that its light becomes stable. We have now discussed the conditions of being prior, first and better known in the order of nature, and seen that these conditions are closely related to the distinction between the order of things and the order of

128

Here we encounter the complex question of how the mathematical sciences, the objects of which are naturally secondary to the objects of sciences such as biology, can provide us with explanations of natural phenomena. 129 I am grateful to Håvard Løkke for discussion on this point.

80

CHAPTER ONE

knowledge. More specifically, the conditions are on the one hand related to the condition of being explanatory; the premises express explanatory factors of the facts figuring in the conclusions. Often the crucial factor appears in the middle term (this is the case, for instance, with the eclipse and being shadowed by the earth).130 In addition, Aristotle allows explanation in terms of categorising subtypes of natural things under more general kinds. From our perspective this procedure is not as clearly connected to explanation as explanation through syllogisms like those concerning the planets and the lunar eclipse. Necessity, Universality and Immediacy. Another group of requirements concerns necessity. Aristotle says explicitly that the condition of necessity is connected with assumptions concerning knowledge in the strict sense (%/iqr3,g 9/k‡p). What is known in the strict sense is the object of apodeictic knowledge (r• %/iqrgr•l r• jar¡ r¢l 8/ndeijrij¢l %/iqr3,gl), i.e. it is proved. What is proved cannot be otherwise and is, hence, necessary (I 4, 73a21–23). In explaining the notion of necessity Aristotle uses four auxiliary expressions ‘in every case’ (jar¡ /alr5p), ‘in itself’ (jah’ a∫r5), ‘as itself’ (¯ a√r5) and ‘universally’ (jah5kns) (I 4, 73a26–27; b29); the fourth one he also further specifies as ‘firstly universally’ (/oÍrns jah5kns, I 5, 74a14). When discussing the condition of necessity Aristotle at first clarifies what he means by saying ‘in every case’. His account makes clear that if something holds in every case this in fact amounts to omnitemporal truth. As noted above, it was typical for Aristotle, Plato and the Stoics to analyse statements as temporally indeterminate. This means that a standard statement for them was of the type ‘Socrates sits’ rather than ‘Socrates sits at time t0’. In this

130

However, this does not hold for all the examples we find in Aristotle. Philoponus notes this in his commentary, and it has recently been pointed out by Mariska Leunissen (forthcoming in de Haas (ed.)) that in the second book (II 11, 94b9–26), where Aristotle explains how the final cause is put into the syllogism, it is in fact quite difficult to take the final cause as the middle term. According to Leunissen, it should be taken as the major term. For instance, the example concerning the reason for walking after dinner should be taken as follows. A is to be healthy, B that the food will not float to the surface of the stomach, and C to walk after dinner. In this syllogism the conclusion reads ‘it is healthy to walk after dinner’. And why is that? It is that when one walks after dinner the food will not float to the surface of the stomach. This leads to enormous questions concerning the place of teleology in an Aristotelian science. Those questions cannot be dealt with here; for a discussion, see Johnson (2005).

THEORIES OF ARGUMENTATION

81

analysis, to say that the statement is true in every case amounts to saying that the state of affairs it expresses holds all the time.131 Next Aristotle goes on to discuss the condition of ‘in itself’ (jah’a∫r5).132 He distinguishes between four different ways of belonging in itself. It is often thought that only the first two are relevant to the scientific premises – and Aristotle himself seems to say so later on in the passage (An. Post. I 4, 73a34–b18).133 Those two ways in which predicates can belong to their subjects in itself are the following. (i) The first case is how a predicate belonging to the essence of a subject belongs to it. (ii) The second type is how specific differences belong to the genus which they differentiate into smaller species. Aristotle’s example of the first type (i) are lines which belong to a triangle; every triangle is essentially dependent on lines, because every triangle is a plane figure limited by lines and, hence, defined by reference to lines. Of (ii) there are more examples; they are all necessary attributes of the genus number, odd and even, prime and non-prime, and square and rectangular.134 These belong to the genus number in itself, because they are the differences of numbers, not of any other kinds of things. If we talk about rectangular and square figures, we are not in the Aristotelian context talking strictly about the 131

There has been discussion about whether the Aristotelian notion of necessity amounts to what is called a statistic interpretation of the modalities. According to this interpretation necessary truth means omnitemporal truth. Arguments for the statistic interpretation are found in, e.g., Hintikka, Remes and Knuuttila (1977). Knuuttila has recently withdrawn from this interpretation; he currently thinks that Aristotle allows for non-realised potencies but not for non-realised generic potencies (cf. Knuuttila 1993c). 132 Aristotle’s account of the predicates belonging in themselves has been much discussed in the literature since antiquity. For a discussion of Ancient and Renaissance commentators on belonging in itself, see Mignucci (1975, 55–85). For a more recent research, see van Rijen (1989, 132–156). The first type of the predicates is a quite straightforward case: it comprises the terms expressing genera and specific differences. The second type, by contrast, is more problematic, because Aristotle seems to include in it either disjunctive (all numbers are in themselves odd or even) or particular (some numbers are in themselves odd and some numbers are in themselves even) premises, neither of which he includes in the class of unprovable premises. Van Rijen claims that the second type of premises expressing in itself relations would consist of propria. This suggestion, however, runs counter to what Aristotle says about propria in the Topics (I 5, 102a18). 133 Cf., e.g., Ross (1949, 519–520). 134 The last mentioned refer to numbers here, not to figures. Those that are square are pretty clear; they are products of the same number with itself, e.g., 4 is square because 22
4. Rectangular numbers are such that they are products of two unequal numbers, e.g., 15 is rectangular (3  5
15).

82

CHAPTER ONE

same properties, namely about square and rectangular as belonging to numbers. The third case is (iii) an object that exists in itself, i.e. that it is not a predicate of another thing; fourthly, we have (iv) a reason that is reason for something in itself, i.e. not an accidental reason. Even though Aristotle emphasises the role of the first two in science, it is not impossible that the fourth case could also be of importance. We might take it as ruling out accidental causes (such as those discussed in the Physics II 5) and hence it could be included in the conditions for scientific premises. For example, it is not an accidental cause that the planets do not twinkle because they are near.135 If this is correct, the fourth way of belonging in itself refers to the explanatory apodeictic syllogism discussed above. Aristotle also adds a further expression according to which there are predicates belonging to their subjects when the subjects are conceived as themselves (¯ a√r5, ¯ in later Latin discussions is the qualification qua). At one point he says that the two qualifications, in itself and as itself, are the same (73b28–29). It is possible that he supposes that the two qualifications amount literally to the same thing. However, another, and to my mind a more plausible, possibility is that the qualification of belonging in itself (jah’ a∫r5) is larger in extension than as itself (¯ a√r5).136 I think it is plausible to make the following difference between the two qualifications. The qualification in itself (jah’ a∫r5) comprehends also such predicates that belong to the subject in question as qualified according to its generic nature (a human being as animal, for instance) or as one of the predicate’s subtypes (a triangle as isosceles). If this is correct, Aristotle would allow that, for instance, a human being in itself is a percipient self-moving

135

Even though Ross thinks that only the first two conditions are relevant, he says that in discussing the fourth way of belonging in itself Aristotle substitutes the phrase ‘in itself’ (jah’ a∫r5) for the normal ‘because of’ (di1). He notes that this fourth type reminds one of the first two because there is a necessary relation involved, but he distinguishes this condition from the first two because it involves a temporal sequence from cause to effect, whereas the first two are ‘timeless connexions’. I do not quite accept this argument because Aristotle in the Posterior Analytics clearly allows for cases where a temporal sequence is involved. For instance, a lunar eclipse takes place when the earth starts to shadow the moon, then the shadow becomes larger and finally the whole moon is shadowed. In addition, lunar eclipse is perpetually recurring and hence not timeless. Therefore, I do not think that, e.g., Ross provides us with very good reasons for excluding the fourth way of belonging in itself from the premises of proofs. 136 According to Philoponus (in An. Post. 71, 4–13), this view was taken over by Theophrastus; cf. Mignucci (1975, 81).

THEORIES OF ARGUMENTATION

83

thing and that an isosceles triangle has in itself the sum of its interior angles equal to two right angles. However, he would reject that a human being is as itself a percipient self-moving thing and that an isosceles triangle as itself has the sum of its interior angles equal to two right angles. A human being is such only as an animal and an isosceles as a triangle. However, we can insert one additional step in our syllogisms to make the connection. All isosceles figures are triangles, all triangles have the sum of its interior angles equal to two right angles; therefore all isosceles [figures] have the sum of its interior angles equal to two right angles. All human being are animals, all animals are percipient self-moving creatures; therefore, all human beings are percipient self-moving creatures. According to this suggestion the qualification ‘as itself’ restricts the scope of the qualification ‘in itself’ so that only the predicates that belong to their subjects in virtue of their own specific nature are included in the premises of scientific proofs.137 A similar point can be made in connection with the phrases ‘universally’ (jah5kns) and ‘universally first’ (/.Írns jah5kns). Aristotle makes clear that ‘universally’ is not merely the condition of being true in every case. Rather, he says that it comprises the three requirements of belonging in every case, in itself and as itself. The condition ‘as itself’, as we just saw, can be taken to restrict the predicates belonging in itself into those that belong to the subject only as such (not in virtue of a more general kind). Similarly, the condition of belonging universally first is, according to Aristotle, to be taken in the way that a predicate belonging universally applies, not to all predicates that belong to all the instances, but to those that form the largest class to which the predicate applies universally. For example, having the sum of the internal angles equal to two right angles belongs universally first to triangles because this is the most general class of which the property can be truly predicated. Even though the same property does belong to all isosceles triangles as well (and we might expect that it belongs to them universally), this is not the restricted meaning of the condition ‘universally’ Aristotle is talking about in this connection. Here Aristotle restricts the condition of belonging universally to cases that also satisfy the condition of belonging universally first. Aristotle does not explain in detail how the several requirements hang together. He does, however, make it clear that if the predicate belongs in itself (jah’ a∫r5) in one of the first two ways (either in the sense that the predicate belongs to the essence of the subject or in the sense that the subject divides the predicate into specific kinds) it belongs necessarily (73b16–19). Because belonging universally or universally first (jah5kns, /.Írns jah5kns), as he 137

For a discussion of reduplication and its connection with the requirement of immediacy, see Bäck (1996, 28–38).

84

CHAPTER ONE

says, involves belonging in itself and belonging as itself it is highly likely that belonging universally in the special meaning Aristotle employs here also entails necessity. Further, because the condition ‘as itself’ (¯ a√r5) is also defined through the condition ‘in itself’, it also entails necessity. The only condition of which it is not quite clear whether it entails necessity is that of belonging ‘in every case’. The question is whether there are necessary predicates belonging to their subjects in every case without belonging to them ‘in itself’. According to a traditional interpretation, Aristotle allowed such a class of necessities, i.e. the so-called peculiar properties (Greek ©dia, Latin propria) such as the capability of laughter among human beings.138 These peculiar properties or ‘inseparable accidents’ were in the Middle Ages oxymoronically called ‘accidental necessities’.139 The question how Aristotle interpreted modalities in general is much too large to be dealt with here. We can note that in science the most important predicates to consider are those which belong in a way defined through the condition ‘in itself’, i.e. such that are determined by the essence. Inseparable accidents are not of primary concern in Aristotle’s theory of science. In addition to the essential necessity involved in the ‘in itself’ relation between subjects and predicates, Aristotle also, particularly in the biological works, deals with another type of necessity called by scholars ‘hypothetical necessity’.140 We cannot, however, consider this topic in detail here. In general it can be said that the material causes in animals are hypothetically necessary in the following way. Animals must have some kind of material make-up in order to have the kind of organs they have and they need to have the kind of organs they have in order to be the kind of animals they are. In this sense the necessities involved in the material constitution of the animal are subordinated to the formal causes and to natural teleology. In the Posterior Analytics Aristotle does not explain in detail the metaphysical background for the properties that are determined by the essence. In the Metaphysics as well as De Anima he talks about intelligible (lngr5l) forms (e∆dnp). The forms are metaphysical structuring principles of all existing things 138

See An. Post. I 3, 73a7 and Top. I 5, 102a18–20; cf. Ross (1949, 517). For a discussion of these properties, see Van Rijen (1989, 132–156). Van Rijen defends a view according to which the inseparable accidents are not necessary, according to Aristotle, and hence this class of predicates constitutes a counterexample to the statistic interpretation of modalities. For a statistic interpretation, see Hintikka (1973) and Hintikka, Remes and Knuuttila (1977). At least, inseparable accidents are not necessary in the way that they would be necessarily contained in the substantial form, but they follow from it in a looser sense. 140 See, e.g., Cooper (1987a). 139

THEORIES OF ARGUMENTATION

85

that necessitate some properties, make some properties possible and exclude others.141 The terminology of forms and intelligibles is not present in the Posterior Analytics. However, in the last chapter (II 19), lines 100a7–8, the universal that is later also described as a starting point for science is characterised as being ‘one besides the many’ (*l /a.¡ r¡ /nkk1) and ‘the same in all [instances]’(%l %je4lnip r• a√r5).These expressions seem to contain in a nutshell the idea that in the many instances of the same species there is the same metaphysical principle in virtue of which things are the kind of things they are.142 Even though Aristotle lists the condition that the premises have to be necessary as one of the vital properties of scientific premises, he also later on in the treatise loosens his criteria quite a bit and notes that not all the premises need to be such. Some scientific generalisations are not necessary but only hold for the most part (®p %/§ r• /nk6, An. Post. I 30; II 12, 96a9–19). This, however, is to be distinguished from something purely contingent and they should not be analysed as statements of the form ‘some Bs are As’. Aristotle’s example of generalisation for the most part is that most men have beards but not all do. He allows such generalisations, valid only for the most part, to be included in scientific premises; he also indicates that they can appear in the first figure affirmative syllogism in the way that the syllogism remains scientific.143 He is not explicit whether in both the premises or only in one of them does the predicate belong only for the most part, but he is clear in saying that in the conclusion it will so belong. Aristotle assumes that in the sublunary world, events and the distribution of properties are not perfectly regular. However, the fact that he also allows for-the-most-part-generalisations to appear in science indicates that because in such cases the generalisation holds on a general level, it can be included in science even though it is not universally valid. One of the conditions for the premises of the scientific proofs which we have not yet discussed is the one according to which the premises are immediate (!,eqnp). This condition is closely connected to Aristotle’s view of the proofs as universal affirmative first figure syllogisms. As the term suggests, an immediate premise is such that no further middle term (r• ,2qnl) can be added between its terms. If the premises are first, not only prior in the order of nature, both premises express immediate connections. With respect to the syllogisms expressing a natural species within a genus this means that the nearest specific difference as well as the nearest genus have been found. In the case of syllogisms 141

For discussions of Aristotle’s conception of forms, see, e.g., Frede and Patzig (1988), Witt (1989) and Wedin (1991 and 2000). 142 I shall not discuss the complex questions related to the interpretation of the Metaphysics VII and its theory of forms here. 143 See Mignucci (1981).

86

CHAPTER ONE

which express the causes of natural phenomena (such as a lunar eclipse or thunder), the middle term expresses an immediate cause and there is no middle term expressing a necessary cause that could be added either between the lowest term and the middle term, or between the major and the middle term. Although connected with the assumption concerning sciences being syllogistic, the criterion of immediacy is not a purely logical one. By contrast, the alleged fact that no middle term can be added in such a way that a proof would result is a consequence of there being in reality no such mediating factor that would connect the things named by the middle term and the minor term and that would be a necessary cause.144 As mentioned earlier, I think that Aristotle’s examples can be used to illustrate the kind of syllogisms he means when he talks about scientific proofs. It seems, however, that in many syllogisms the premises are not immediate but seem to need further explanation. However, if the explanation falls within the scope of another science, we might be able to find examples of premises that are immediate in the limited sense of being immediate within a given science. This, however, remains a little unclear. Proofs and Definitions In addition to scientific proofs Aristotle also discusses scientific definitions in the Posterior Analytics.145 He says repeatedly (e.g. An. Post. II 3) that the definitions cannot be proved, and it would seem a rather natural assumption to make that Aristotle means that the unprovable premises of proofs are the definitions.146 However, there are some reasons to resist this prima facie plausible suggestion. In the second book of the Posterior Analytics chapters 8–10 Aristotle discusses definitions by means of the following examples.

144

Lunar eclipse

Thunder

A eclipse (
a certain loss of light) B shadowing by the earth C moon

A noise B quenching of fire C clouds

For a discussion of how the middle terms in Aristotelian syllogisms explain the conclusions, see Charles (2000, 204–265). Charles’s examples are mainly of the type of syllogism I have called ‘explanatory’. I do not intend to dispute that a definition of a natural phenomenon can be read from such premises. A more problematic case I have referred to above is whether the genus-species syllogisms that I have called ‘explicatory’ contain an explanation of the conclusion. 145 For Aristotelian definitions see, e.g., Charles (2000, 179ff.), Devereux and Pellegrin (1990), Gotthelf and Lennox (1987) and Ackrill (1981). 146 This is, for instance, the view taken by Kahn (1981).

THEORIES OF ARGUMENTATION

87

The syllogism expressing the reason why lunar eclipses take place is as follows: Those celestial bodies that are shadowed by the earth are eclipsed, i.e. suffer from loss of light. The moon is shadowed by the earth. The moon is eclipsed, i.e. there is a loss of light from the moon.

The syllogism for thunder would be: Quenching of fire causes noise. Fire is quenched in the clouds. There is noise in the clouds.

Both these syllogisms presuppose knowledge of the nominal definition of thunder and an eclipse. Aristotle requires that we have to know that some phenomenon takes place – in the Aristotelian framework that a predicate belongs to a subject – before we can inquire into its nature and cause. He points out that knowing the nominal definition of the phenomenon is a necessary condition for knowing that the phenomenon takes place. However, it is not a sufficient condition, because knowledge of the nominal definition needs to be accompanied with some kind of knowledge of the phenomenon or the object of inquiry as something existent. As Aristotle says, even though we know that a goat-stag (r.ac2katnp) is a combination of these two animals and in this sense we can know its nominal definition, we cannot know what a goat-stag is in the scientific sense, because it does not exist (II 7, 92b7). Therefore, Aristotle allows that it is possible that knowing a nominal definition is not sufficient for knowing the existence of something, or knowing that a phenomenon takes place. In many cases, however, existence can be perceived. In such cases, it is very easy to answer the first scientific questions ‘does it exist?’ or ‘does this attribute belong to that subject?’. Sometimes the existence or non-existence of the object of inquiry cannot be found out by mere sense perception. Aristotle also recognises this in connection with the scientific questions at the beginning of book II by choosing the existence of god as an example. In such cases, he assumes the existence of a scientific object must be ‘proved’ by using indirect argumentative strategy and pointing to the existence of some other facts that are better known in perception than the one intended to be proved (e.g. the arguments for the existence of the unmoved mover in the Physics VII and VIII 5–6). He also allows for the establishing of existence by a syllogism where the middle term is familiar in perception. This, I take it, is the meaning of the somewhat strange syllogism in chapter 8 of book II (93a37–b3). The major term A is an eclipse, C is the moon and B is ‘the incapacity of the full moon to make a

88

CHAPTER ONE

shadow even though there is no visible hindrance’. Aristotle probably refers to a case that someone doubts whether a lunar eclipse really takes place. In such a case it can be argued for as follows: we can point to the fact that even though there is a full moon it suffers loss of light. This, however, is not caused by any obstacle (for instance, it is not cloudy). Therefore, it has to be an eclipse: loss of light caused by a celestial body. These kinds of arguments are not scientific proofs in the strict sense because such phenomena ‘better known to us’ are not causes and, hence, the order of the arguments is the opposite of the one required for real proofs. In his discussion Aristotle takes the nominal definition of eclipse to be something like ‘loss of the light from the moon’ and for thunder ‘noise in the clouds’. When we know that sometimes the moon loses its light and not because of the usual reason that a cloud shadows it, we know that lunar eclipses take place. And when we know that there is noise in the clouds, we know that there is thunder. If we look at the examples, in both cases knowledge of the existence of the phenomenon is actually knowledge of the conclusion. If it is knowledge only of the conclusion, it does not yet entail knowledge of the whole explanation or the definition. Aristotle makes it clear that when we come to know the real explanation we also come to know the definition. What, then, do we need to know in order to know the definition? Aristotle’s answer is pretty clear. We need to know, for instance, that the loss of the light from the moon is caused by the earth shadowing the moon, or, we need to know that the noise in the clouds is caused by the quenching of fire (or other movement of the basic elements fire, air, water and earth causing noise) in the clouds.147 In both cases we know that the phenomenon is real or existent and we know its cause. If we now look back at the syllogisms, it becomes clear that the premises alone do not express the definition but all the parts of the syllogism are involved in the definition. The definition involves the conclusion ‘moon loses its light’, ‘there is noise in the clouds’. In an apodeictic syllogism, whether expressing a natural species, its genus and specific differences or natural phenomenon and its immediate cause, neither the premises nor the conclusion alone express the whole definition. In the former the specific differences are expressed by middle terms, in the latter type the middle term expresses the cause.148 In the examples above the conclusions read as follows. ‘The moon is eclipsed’ and ‘there is noise in the clouds’. In 147

Aristotle in fact (in the Meteorologica) seems to think that quenching of fire is not the actual cause of the noise in the clouds in thunder, but the example can in any case illustrate the relation between causes, proofs and definitions. 148 The middle term can also express a causal factor bringing about the intended end as in the example of II 11.

THEORIES OF ARGUMENTATION

89

both cases the middle terms, which do not figure in the conclusions, are important elements of the definition. When taken together, the terms express the definition.149 Aristotle might assume that the whole syllogism, even though not literally proving the definition as its conclusion, can be taken to confirm the definition. Therefore, as Aristotle says, the definition is not proved: the conclusion is only a part of it. However, the premises do not express the whole definition either but part of it needs to be taken from the conclusion. Do the Sciences Have Something in Common? We have up to this point discussed scientific premises with definite content. These premises are, according to Aristotle, specific for most sciences separately. However, he allows a class of general principles that are common to several sciences. In this section we shall discuss those principles. Common Axioms and the Departmentalisation of the Sciences. It is peculiar for his conception of proofs that Aristotle assumes the scientific context to be differentiated according to the objects of particular sciences. This differentiation is based on natural genera. According to Aristotle, reality consists of natural genera and each specific science has one natural genus as its object (An. Post. I 7, 75b1–11; I 10, 76b12–14). Usually, the premises of one science cannot be used in other sciences, because different sciences are concerned with different kinds of things. Basically, Aristotle claims that sciences are concerned with substances (e.g. An. Post. I 22, 83a24–35; cf. Met. III 2, 997a16–18) and that the substantial nature of each scientific object is partly determined by its generic nature.150 In order to be true to the order of nature, sciences must respect the generic differentiation of reality. One exception to Aristotle’s assumption of different sciences having different objects is the mathematical sciences. According to Aristotle, mathematical sciences, i.e. arithmetic and geometry, do not have a separate class of abstract entities as their objects as he supposes the Platonists assume. By contrast, according to Aristotle, the objects of mathematical sciences are natural things, but in arithmetic they are treated as countable and in geometry as

149

For a discussion of the unprovability of the Aristotelian definitions, see Byrne (1997, 147–163), who also suggests that the definition is included in the entire syllogism, not in any part of it. 150 This, however, is not to say that the sciences should make explicit reference to the notion of substance; only metaphysics does. E.g. when we prove something about an animal species we do not explicitly make it part of our proof that it is in the nature of that kind of a thing to have those features.

90

CHAPTER ONE

geometrical figures abstracted from any material content.151 Hence, according to Aristotle, there are no abstract substances, but mathematical sciences study natural things in an abstract way. In mathematics Aristotle also sticks to the general idea that different sciences do not use the same premises. He talks about common axioms (jnil•l 8m4w,a) that are used in all or many sciences. He gives two examples, one of which is a general logical principle, namely the law of the excluded middle (An. Post. I 11, 77a31–33).152 Another is mathematical: ‘if equals are taken from equals the results are equal’ (I 10, 76a37–b1). Aristotle says that in different sciences common axioms are used by analogy (jar’ 8laknc4al, An. Post. I 10, 76a38–39). The exact meaning of this remark is unclear. In connection with the mathematical sciences this principle was later called ‘the principle of purity’ and it had a strong history after Aristotle. I shall discuss below in 1.3 how some of the ancient Greek commentators on Aristotle reacted to this idea. Even though many commentators take it quite seriously, it is not impossible that Aristotle did not mean the remark to be a substantial one. He might only have meant that if one is to apply the principle concerning the subtraction of equals, for instance, he or she has to know what is counted. Numbers cannot be subtracted from spatial magnitudes. However, some of Aristotle’s background assumptions in fact point to the direction that he adheres to some version of the principle of purity. Firstly, it might be that common axioms cannot be exactly the same in different sciences because they do not express relations based on substantial sameness in natural objects. There is no substantial nature of, for instance, spatial quantity, which would include the generic necessary attributes of quantities. This is why all types of quantities are not synonymously quantities.153 Aristotle’s definition of synonymy does not require that only things sharing exactly the same substantial nature would be genuinely synonymous, i.e. that synonymy should only be restricted to the instances of the same species. According to Aristotle, there is synonymy between species within a genus, e.g., human beings and horses are animals synonymously (Cat. 1). Still, animal as a genus 151

For Aristotle’s philosophy of mathematics, see Annas (1976, 26–41). For an extensive study, see Cleary (1995) and Lear (1982) for a shorter article. 152 It is likely that to this class also belongs such principles as that of non-contradiction discussed in Metaphysics IV and that according to which, if two things are the same as a third, they are the same as each other (Soph. El. 6, 168b31–32). They are, however, not explicitly mentioned in the Posterior Analytics passage. 153 For Zabarella’s discussion of whether quantities should be taken to be quantities synonymically, see McKirahan (1992).

THEORIES OF ARGUMENTATION

91

has no substantial nature of its own. Horses and human beings have their own specific natures, which have a common generic element. But why are quantities not synonymously quantities in the same way? According to Aristotle’s philosophy of mathematics, mathematical objects do not form a separate class of substances of their own and, consequently, they do not have a substantial nature as mathematical objects. However, it is not likely that Aristotle would take quantities to be quantities in an accidentally homonymous way, i.e. in the sense that they should only have a common name. Rather, Aristotle seems to assume that different kinds of quantities are similar in important ways but there is no substantial sameness in them, because quantities lack substantial nature as quantities. This is why it can be said that what subtraction of spatial magnitudes is for spatial magnitudes, subtraction of numeric quantities is for numeric quantities. However, the two kinds of subtraction are not quite the same thing.154 We cannot subtract a number from a triangle. It is possible that Aristotle’s discussion of this ‘principle of purity’ is part of his critical responses to Plato.155 In the Meno the questions Socrates makes to the slave boy are concerned with numbers: ‘What is the number whose square is twice four?’. The answer is made clear by using a geometrical figure. But why should this be a problem? One reason might be that Aristotle considers the example in the Meno problematic because it involves an irrational number, namely the square root of two. In general the Greeks were not at that time able to deal with irrational numbers. Aristotle might have assumed that this is one reflection of the problems involved in crossing the borderlines between different kinds of quantities: from the geometrical picture we get numbers that do not belong to the class of comprehensible (i.e. rational) numbers. As already noted, however, it is not only general mathematical principles which Aristotle said are analogous in several sciences. This is the case with all the common axioms (jnil¡ 8miÍ,ara). Therefore, when we use, for instance, the principle of non-contradiction, we use it somehow differently if the objects of our science are different. But this sounds odd. However, I think it is typical of Aristotle to make this assumption. Basically, he claims that when we utter, for instance, the principle of non-contradiction, we need to know what we are talking about, i.e. what we are applying the principle to. He makes it clear that in fact such general logical principles such as that of noncontradiction are not usually used as premises in scientific proofs; they are so used only if we need to prove an instance of them (I 11, 77a10–21). However, 154 155

Cf. Aristotle’s ‘definition’ of analogy in the Top. I 17. I am grateful to Anna Somfai for pointing this out to me.

92

CHAPTER ONE

such proofs rarely occur, according to Aristotle. In any case, always when we use the principle we must have some particular application of it in mind, and this application involves taking some things into consideration. Then if I, for instance, think that some animal cannot be both human and non-human, I am already thinking of animals. If I then go on to apply the principle in another case, it possibly involves some things that are completely different from animals, and subtraction of geometrical figures from animals, for instance, is not a well-defined operation. Be that as it may, the important point is that Aristotle, even though using abstract symbols for the syllogistic terms, assumes that general mathematical and logical principles are always used with a particular application in mind. One question which Aristotle does not address explicitly, but which is related to the present topic, is the question of the status of syllogistic figures. As we have seen, Aristotle treats the structure of a science as syllogistic and assumes that proofs are – typically – first figure universal affirmative syllogisms. Logic, however, is not a science for Aristotle and the syllogistic argumentative form does not organise arguments in the same way as the form of natural substances determines the properties and potencies of these substances. Nonetheless, if arguments are considered they can be grasped as similar to each other although different in content. Aristotle expresses these similarities among arguments in the theory of syllogistic figures. The theory of the syllogistic figures is not a science in its own right. However, the figures are presupposed by any science as common forms of valid arguments of which the proofs are a subclass. Aristotle does not address the question explicitly, but it might be along his lines of thought to say that the syllogistic figures are also analogous in different sciences. As with the principle of non-contradiction, every application of a syllogistic figure in science has content of some sort and that content is always related to the genus that the science studies. In another science the content will be different. Therefore, it is not literally the same instance of the figure we have in the two sciences but on an abstract level it is similar. Even though Aristotle is particular on the point that the premises of different sciences are different, he allows that some sciences use premises which in another science appear as proven conclusions. This makes the science which applies the premises subordinated to the one which proves them (see An. Post. I 7, 75b13–16, I 9, 76a23–25). The subordinated sciences typically apply mathematical truths to explain perceptible phenomena. Aristotle mentions several examples of such cases, such as optics as subordinated to geometry, mechanics as subordinated to three-dimensional geometry, and harmonics as subordinated to arithmetic.

THEORIES OF ARGUMENTATION

93

Mathematical sciences are not in a very high position in the order of ontological priority, according to Aristotle. This is because he assumes that mathematical objects are not substances, and in this sense the objects of mathematical sciences are inferior to the objects of natural sciences and to metaphysics.156 Metaphysics, or possibly theology, is the highest of the sciences, because its objects are substances as substances and all the other modes of being depend on the being of substances. However, when Aristotle presents the idea of a hierarchy between sciences, he attributes great explanatory force to mathematical sciences: this is precisely why some other sciences are subordinated to them. This view causes some tension in Aristotle’s theory. Aristotle’s Model for the Sciences. Let us now consider some general aspects of Aristotle’s model for the sciences. As has been noted above, Aristotle assumes that all sciences are similar in the sense that they have a similar structure.157 More precisely, the proofs are similar in the sense that they are – for the most part – affirmative universal syllogisms in the first figure. These syllogisms express, basically, either how natural genera are divided into species by specific differences, how such differences explain some of the properties of the species, or the reasons for natural phenomena. From this it is sometimes concluded that Aristotle’s model for the sciences resembles axiomatised geometry.158 In a science following the model of axiomatised geometry many theorems are deduced from a restricted group of basic axioms, which are self- evident truths. However, the structure of the Aristotelian sciences deviates from such a model. There are many reasons for this and I shall only consider some of them.159 Firstly, in an axiomatised science, we start with a small number of selfevident assumptions and deduce from them – with the help of certain inference rules – a large group of theorems. Aristotle’s model can be contrasted with this because what we at first know when starting the inquiry, according to Aristotle, are the conclusions of the proofs, not the premises. This is precisely the idea in the contrast between what is better known to us and by nature. In the course of inquiry we need to find the premises, typically the syllogistic middle term or terms, knowledge of which makes us able to construct a 156

For this conception of Aristotle’s, see also Leszl (1980, 387). For the ‘unity of sciences’ in Aristotle, see Leszl (1980). 158 For such a conception see, e.g., Barnes (1969, 123) and Scholz (1975, 52) (originally published in 1930). 159 Other scholars, e.g. Kahn (1981, 388–389), have also pointed out that the model of axiomatised geometry is not appropriate to illustrate Aristotle’s conception of scientific proofs. 157

94

CHAPTER ONE

proof. Secondly, Aristotle makes it clear (An. Post. I 32, 88b3–6) that there are not many fewer premises than there are conclusions in a science. He assumes that all the proof chains terminate in premises that are not proved. However, he does not suppose that all proof chains lead to a small ultimate class of principles which are beyond doubt. Neither does he assume that the proof chains should be very long. Rather, he seems to conceive of sciences as collections of short proof chains concerned with the objects falling within the scope of a particular science. Further, some scholars have suggested that there is a kind of generic premise overarching each science concerning the generic nature of the things studied within that science.160 However, this does not entail that from this premise one could deduce all the ‘theorems’ of that science. Rather, Aristotle’s idea seems to be that the statements within a science are from the very start divided into many different topics, such as different parts of different kinds of animals, and that this division is in accordance with the nature of things and their multifarious nature. In general, Aristotle’s ‘theory of science’ is implicitly teleological in the following sense.161 When he discusses the general conditions set on the premises of scientific proofs at the beginning of the Posterior Analytics (I 2–6) Aristotle presents, as it were, a meta-theory concerning what a complete science should be like. Rather than claiming that a science is a linear structure of deductive chains Aristotle conceives it to be a fairly flexible set of proofs forming a net or a bundle of syllogistic chains around the object with which they are concerned – and there are many such objects within one science. He also clearly allows that the sciences contain additional remarks on, e.g. organs or parts of animals that do not seem to fit into a linear proof-structure with a single core, as they presuppose a more flexible analysis.162 In particular, the topics must have complex and flexible connections with each other. However, when presenting these remarks Aristotle seems to aim at producing short proof chains – either expressing parts of generic structure or the reason for natural phenomena – that could be taken as parts of a science conceived of as the kind of collection of proofs just explained.163 His aim is not to build 160

Hintikka (1972). For the expression ‘implicit teleology’, see Hintikka (1974). 162 An example is found, e.g., in Part. An. (III 14, 675a36–b1), where Aristotle notes that the fact that the gut is wider in the neighbourhood of the stomach and narrower towards the other end explains ‘why dogs have to strain so much in discharging their excrement’; see also Part. An. (III 11, 673b4–12). 163 This holds, for instance, of the remarks concerning sperm in the Generation of Animals I 17–23; see, e.g., I 17, 721a30ff. and I 18, 724a14ff. discussed in Bolton (1987, 151ff). Bolton points out that there is no exact equivalent in English for Aristotle’s term q/2.,a (n. 48, p. 151). See also Part. An. II 2, 648b4–10 and III 14, 161

THEORIES OF ARGUMENTATION

95

an architectonic system; rather he concentrates on one particular scientific object at a time and aims at collecting the genus and specific differences of a thing as well as terms explanatory of certain natural phenomena. Question of Circularity. We have indicated above that Aristotle allows that the unprovable premises of scientific proofs can be argued for in various ways, either dialectically from reputable conceptions or from true, even necessarily true premises as is the case in the example of the planets and nontwinkling.164 One might now ask how well Aristotle in fact succeeds in avoiding circularity, which he expressly claims to avoid in the Posterior Analytics I 3. Aristotle discusses the error of circularity, i.e. proving a conclusion on the basis of premises in the Prior Analytics II 16. He makes it clear that his arguments (called syllogisms of a fact) establishing the nearness of planets and proofs for the non-twinkling of planets (called syllogisms of the reason) employing the same terms but in a different order, are not to be counted as cases of begging the question (64b28–34).165 The argument that concludes that the planets are near on the basis of their non-twinkling has a premise which is naturally secondary to the conclusion because it is not explanatory of it in the metaphysical sense. The syllogism establishing that the planets are nontwinkling because they are near, by contrast, is a real proof because its premises are explanatory in that sense. Now, Aristotle claims, these two arguments do not lead to a circular conception of the premises of proofs, because – as he says – real premises are known through themselves (to them applies the phrase ‘that which is known through itself’, r• di’ a∫rnfl clwqr5l, 64b35; cf. Top. I 1 and Phys. I 1). However, given that Aristotle clearly allows that the minor premise of the proof (e.g. that the planets are near) can be established by a Barbara syllogism having true, and perhaps even necessarily true premises, we might wonder what Aristotle means by this. I shall return to this question at the end of this chapter. Before that I shall touch upon a very well-known problem related to Aristotle’s theory of science, namely its relation to his scientific practice. 674a9–b17 referred to by Gotthelf (1987). Gotthelf claims (1987, 186) that such passages can be fitted into a linear axiomatic structure, which seems a bit exaggerated. They seem rather to contain a fair number of unprovable premises, which cannot be characterised as self-evident as an axiomatic ideal would require. 164 I do not intend the phrase ‘necessarily true’here to mean that I would claim that the examples Aristotle presents have premises that actually are necessary. I simply point to the possibility that Aristotle might consider such premises necessary. However, the point about necessity is not crucial to my discussion here. 165 For Aristotle’s analysis of arguments that beg the question, see 1.2.1.

96

CHAPTER ONE

Remarks on Aristotle’s Scientific Practice In the contemporary scholarly literature it is rather widely recognised that, in many of his works, Aristotle proceeds according to the standards of dialectical argumentation. This seems to imply a serious conflict with the strict conditions Aristotle lays out for scientific proofs in the Posterior Analytics, namely that the best dialectic can do is to enhance the credibility of its conclusions and it can never guarantee truth, let alone necessary truth, priority or explanatory power. Therefore, it seems that Aristotle does not proceed according to his own methodological instructions. This problem has received a great deal of attention in the literature; so much so that G. E. R Lloyd has called it a ‘hoary old chestnut indeed’.166 When starting to discuss this old chestnut, namely the question of whether there is a contradiction or tension between Aristotle’s theory and practice of science, we need to be clear what we expect various scientific and philosophical works to be like and what we expect of the model of the Posterior Analytics. Firstly, if we expect the Posterior Analytics to provide a model for scientific and philosophical works in the sense that every sentence in those works should be a basic categorical sentence (AaE, AeE, AiE, AoE) and in one of the syllogistic relations to the other sentences that follow, then an appearance of contradiction is inevitable. Aristotle writes normal prose where the structure of the argument is sometimes rather difficult to follow. In addition, he clearly uses forms of argumentation classified as dialectical rather than scientific. Indirect argument forms, for instance, are frequent. On the other hand, if we renounce this expectation the appearance of a contradiction is not equally clear. This, taken together with what we said above about the structure of science, leaves us the possibility that the Posterior Analytics is intended as a general model for the kinds of predicational connections Aristotle is aiming at finding or arguing for in a somewhat dialectical manner in the treatises. A more detailed model for finding relevant kinds of syllogistic connections is found in the Prior Analytics. In chapters 27–30 of the first book Aristotle explains how premises can be found for all kinds of conclusions. He clearly indicates that the same scheme can also be used in science. The syllogistic scheme which Aristotle presents in the Prior Analytics I 27–30, can, he claims, be used to find the premises concerning any conclusion whatsoever. It is only in the Topics where he presents similar claims on the universal applicability of the technique of argumentation he introduces (I 1, 100a18–21). 166

See Lloyd (1990, 371). For the discussion, see also Barnes (1969 and 1981), Wians (1989 and 1990), Bolton (1987), Gotthelf (1987), Lennox (1987) and Leszl (1981).

THEORIES OF ARGUMENTATION

97

By contrast to the Topics, where the arguments are systematised according to the so-called predicables, the scheme of the Prior Analytics hangs completely on syllogistic analysis. Aristotle sometimes claims in the Prior Analytics that all valid arguments can in fact be either presented in the figures or reduced to them. Even though this claim is problematic, it is reflected in the opening words of chapter 27 of book I. Aristotle indicates that the scheme he there introduces can be used in any attempt to find premises for any given conclusion. In the framework of the syllogistic analysis the conclusion is in one of the following forms. Universal affirmative (AaE): A belongs to every E. Particular affirmative (AiE): A belongs to some E. Universal negative (AeE): A belongs to no E. Particular negative (AoE): A does not belong to some E.

To find premises for the conclusions, one should, according to Aristotle, do the following. Gather (a) the predicates which universally follow the relevant terms (A and E), (b) the predicates which are universally followed by the relevant terms, and (c) the predicates which do not belong to either A or E. These form the following groups.167 For A

For E

(a) B (XaA) (b) C (AaX) (c) D (XeA)

(a) F (YaE) (b) G (EaY) (c) H (YeE)

Basically, the idea is to find predicates which either universally belong to or are universally denied of other predicates. When we find the same term in both columns we can form a syllogism – provided that the predicate is found in the right places. To argue for a universal affirmative conclusion (AaE), we need to find a suitable middle term, i.e. one that is found in both C and F. Such a term is one to which A belongs universally and which itself belongs universally to Y (AaX, YaE) and, because Y
X, X belongs to all E; therefore, A belongs to all E. For instance, suppose that we have learned that all vines shed their leaves (AaE, where E stands for ‘the vine’ and A for the ‘leaf-shedding’) and we need premises establishing why this is so (An. Post. II 16, 98a38–b16). Suppose also that in our inquiries we have observed that all broad-leafed 167

A similar but not entirely identical scheme is also presented in Knuuttila’s Finnish translation with notes of the Prior Analytics (1994).

98

CHAPTER ONE

plants shed their leaves (AaX) and that the vine is a broad-leafed species (YaE and Y
X). Now we will be able to establish that vines shed its leaves on the basis of the following premises: All broad-leafed plants shed their leaves. All vines are broad-leafed plants. Therefore, all vines shed their leaves.

For a particular affirmative conclusion (AiE) we can use two kinds of premises. It is possible to establish that, for instance, some pleasures are good (AiE where A is good and E is pleasure).168 To follow the first path, we need to find a term so that (i) good belongs universally to that term (AaX), and that (ii) pleasure belongs universally to that term (EaY), which means that X
Y. Such terms in general can be found in the groups C and G from the table. Suppose that when gathering the relevant terms for pleasure and what is good we have noticed that all virtuous actions are good (AaX) and that all virtuous actions are pleasant (EaY, and X
Y). On the basis of these premises we can now conclude that some pleasures are good as follows. All actions in accordance with virtue are good (AaX). All actions in accordance with virtue are pleasant (EaX). Therefore, some pleasures are good (AiE).

The second path is to look for a common term from B and G so that it belongs to all good things (XaA) and pleasure belongs universally to it EaY (Y
X
 EaX). Thirdly, if we have a universal negative conclusion (AeE), we again have two paths to take. We can either look for premises of the form XeA and YaE (Y
X
 XaE), or we can seek those of the form XaA, YeE (Y
X
 XeE). If we take the first route we look for a term which does not belong to any A (XeA) but which belongs to all E (XaE); such a syllogism can be made if the same term is found in both D and F. The second way is to find a term which belongs to all A (XaA) but does not belong to any E. These are listed in groups B and H above. The following example can be adapted from the History of Animals (664a16–17). All oviparous water-animals have a fishspine (XaA). Dolphins do not have a fish-spine (XeE); they have a bone. Therefore, dolphins are not oviparous (AeE). A particular negative conclusion (AoE) can be shown as follows. The relevant term needs to be such that it does not belong to any A (XeA) and that E belongs to all of it (EaY (Y
X
 EaX) ). The example comes from

168

The example involving pleasure and what is good appears both in Alexander’s and in Philoponus’ commentaries on An. Pr. I 27 discussed in 1.3.2.

THEORIES OF ARGUMENTATION

99

Philoponus’ commentary. Let A be good and E be pleasure. Now we need a predicate of which goodness is universally denied, but to which pleasure belongs universally. Such predicates are in our scheme found in the groups D and G. Philoponus chooses ‘unnatural pleasure’. Now, no unnatural pleasures can be good, because being unnatural is completely outside the scope of being good (XeA). Because unnatural pleasures are pleasures (EaX), we can conclude that some pleasures are not good. For those who doubt whether there are unnatural pleasures, Philoponus points out (in An. Pr. 276, 20–28) that there is at least one. Scratching an itch is an unnatural pleasure. He probably does not mean that it is unnatural in the sense of being a perversion. Rather, it is unnatural in the sense of not being naturally desirable.169 Aristotle does not present many examples in the connection of the syllogistic scheme and the ones he mentions are neither particularly surprising nor helpful. Animal is said to be predicated universally of human beings (I 27, 43a25–33; cf. 43b21), it is implied that all justice is good (43b21), to say that ‘that approaching is Callias’or ‘that white is Socrates’are classified as accidental predications. These do not give us much information about how to use the scheme. In any case, Aristotle at one point says quite clearly that the scheme he introduces is perfectly general and can be used in any context where we need arguments for a conclusion. The procedure (…d5p) described is to be followed in the establishment of all conclusions, whether in philosophy (tiknqnt4a) or in any art (r2ulg) or field of study (,1hg,a); we must scrutinize of our two terms what belongs [universally] to them (r¡ ∫/1.unlra) and to what they belong [universally] (n˚p ∫/1.uei).170 One should have an abundance of these and we must proceed by way of three terms establishing conclusions in this way and refuting them in that way. If we want to proceed according to truth we need true premises, if we aim at dialectical syllogisms we argue from reputable opinions (jar¡ d5mal). (An. Pr. I 30, 46a3–10; my translation on the basis of Ross 1949.)

Here Aristotle does not use %/iqr3,g explicitly. However, some fifteen lines later he does so (46a17–22); there he starts to talk about how the premises of each particular science such as astronomy are found. Towards the end of the just quoted passage he also repeats the fairly standard distinction between arguments from true and those from merely reputable premises. It is reasonable to suppose that by talking about true premises Aristotle refers to science. In 169

Philoponus discusses this; see 1.3.2. Here Aristotle omits saying ‘those to which our terms do not belong at all’ (XeA and YeE) corresponding to the class of terms in (iii), i.e. in D and H. One likely explanation is that negative conclusions and premises are not that common in sciences about which he seems to be talking here – even though the word %/iqr3,g is not used. 170

100

CHAPTER ONE

addition, when he distinguishes different kinds of predicates belonging universally to their subjects he mentions those belonging in the essence of a thing (çqa %l r+ r4 eqri, I 27, 43b7) as one such group. Given that science is mostly interested in essential predications, this indicates that the scheme can be used in science too. At the end of his exposition of the scheme Aristotle says: This, then, is our general account of the selection of premises; we have discussed it more in detail in our work on dialectic (46a28–30; transl. from Ross 1949).

Therefore, at one point (46a17–22) he indicates that the scheme is used in the sciences to organise the material learned from experience so that explanatory syllogisms can be drawn more easily. On the other hand, he indicates that the selection and organisation of premises has been discussed in more detail in the Topics. As is clear, the premises the Topics deals with are dialectical; they are reputable but not true, whereas the premises in science have to be true. However, even though the kinds of premises we are looking for differ, Aristotle sees the procedure as in essence the same. Both in dialectic and in science we are looking for premises for a certain conclusion. In order to argue for a conclusion validly, we need to find universal connections between the terms. Whether the premises are true or not is a different question, and Aristotle’s scheme does not provide us with an answer to that question. In his work on Aristotle’s biology, James Lennox has pointed out that in the biological works Aristotle aims at finding precisely the same kind of connections between terms as the syllogistic scheme presented in the Prior Analytics I 27–30.171 Lennox also argues that in the biological works we find a great many examples that Aristotle considers common features to be explanatory. For instance, dolphins have a backbone (to be contrasted with having a fishspine) even though they live in water. When we learn that all animals that have a backbone are also viviparous we understand why the dolphins have a backbone, not a fish-spine: they have it because they are viviparous (HA 664a16–17; 489b2 and 566b3–26). Such explanatory features are found when we have identified the largest group to which the predicate belongs universally. A similar condition is found in the requirements for the premises of scientific proofs expressed as the requirement of belonging ‘firstly universally’.172 In the limits of the present study I cannot go into this topic in more detail. Lennox’s discussion on this is very helpful. 171

Lennox (2001b, 12 n. 11 and 14 n. 18; the article was originally published in 1987) referring to discussions with both Allan Gotthelf and Myles Burnyeat. Cf. Knuuttila’s (1994, 277–278) comments on the Finnish translation of the Prior Analytics. 172 Cf. also Balme (1987) and Pellegrin (1982).

THEORIES OF ARGUMENTATION

101

Aristotle introduces the syllogistic scheme as a universally applicable tool for organising terms so that conclusions of all forms can be argued for if we find the same predicates in suitable places. However, the scheme does not say anything about how we can find the relevant connections between the predicates. We can form the syllogism for a conclusion on the condition that we have the suitable predicates in our lists already; we also need to have them in the right places. Therefore, the scheme is not a proper method of inquiry; it is rather a guide for what kind of predications to look for and how to organise the material when it is found. There are cases where, according to Aristotle, we have a fairly straightforward way of finding the predicates by observation. He also at one point assimilates appearances (tail5,ela) with experience (%,/ei.4a); this is where he introduces his general syllogistic scheme for organising predicates in inquiry (see An. Pr. I 30, 46a17–22). However, in some cases the nature of things is itself such that truth concerning them is not perceptible in such a straightforward manner. In those cases Aristotle’s methodological instruction is, as with any case involving more strictly observational appearances, to lay out the appearances (EN VII 1, 1145b1–7). However, ‘appearances’, as we saw above, in Aristotle also mean the reputable conceptions ((ldnma) that are shared by all or the majority of people or by the wise – or by the majority or by the best of the wise (Top. I 1, 100b22–23). Therefore, it is Aristotle’s completely general methodological advice, that we should always start from laying out the appearances, whether more strictly observational or endoxic. Starting from them we should aim at finding universal predicational connections after the model of the Prior Analytics I 27–30. Basically, we should be able to uncover an intelligible structure organising and giving regularity to the appearances. This means, for instance, uncovering genus-species structures and the essential attributes of natural kinds and their parts, but also the kind of syllogisms I have called explanatory above (e.g. an eclipse is the loss of the light from the moon because of the interposition of the earth, thunder is noise in the clouds due to elementary movement caused by the quenching of fire) are expressed in Barbaras and can be formed by using the scheme of the Prior Analytics I 27–30. Further, it is not only in ethics and first philosophy where Aristotle starts from reputable conceptions. Aristotle also lays out his predecessors’ beliefs and theories on matters he is investigating in the biological works.173 In addition, 173

E.g. Parmenides and Empedocles on whether males are hotter than females (Part. An. II 2, 648a29–35) and Diogenes of Apollonia on bloodvessels (HA III 2, 511b31–512b12) and Polybius on the same topic (III 3, 512b13–513a7). I am grateful to John J. Cleary for reminding me of this. Nussbaum (1982) mentions the Physics.

102

CHAPTER ONE

as we noted earlier in the section on dialectic, in works like the Physics Aristotle starts with his predecessor’s conception, e.g., of space and time. Above we have recognised the idea emphasised by Aristotle that dialectic cannot guarantee the truth of its conclusions because the premises typically are not true but only endoxic. (This is not to say, of course, that it would not be possible for the conclusions of dialectical arguments to be true.) However, even though Aristotle is not providing us with a logically valid heuristic method for finding new knowledge, the devices he recommends – dialectic and the syllogistic scheme – do provide us with definite guidelines how to proceed in inquiry. In addition, dialectical criticism helps us evaluate the existing positions and see what is problematic about them. In this way we can at least improve our own conceptions and avoid the problems that the predecessors’ views encounter. It is also sometimes possible that if the dialectical criticism has been carried out carefully and extensively, we acquire such a comprehensive picture of the possible positions and essential problems in the field that we eventually succeed in hitting upon the truth ourselves. Yet, finding the truth in Aristotle is largely left to our intellectual capacity. Knowledge of the Premises We have seen that an assumption that arguments have an epistemic aim is built into the notion of a dialectical argument and is also present in scientific arguments. In the dialectical context the premises must be initially more acceptable than the conclusion from the point of view of the interlocutor to whom the argument is directed. In a scientific context the premises of proofs must be better known in a very special sense. They must be prior in the order of nature and hence explanatory of the conclusions. By explanatory in this context we mean a notion based on metaphysics. What is initially known to us in scientific inquiry is typically the conclusion of the proof because it is closer to perception and in general the task of scientific inquiry in the Aristotelian context is to find the premises. More specifically, the proofs are typically in the form of the first figure universal affirmative syllogism and the minor and the major term appear in the conclusion. Consequently, research mainly concentrates on finding middle terms in between those terms. There can very well be several such middle terms, but Aristotle trusts that the search will at some point terminate at a point where one needs no longer ask ‘Why is that the case?’ or ‘What are the specific differences of that object?’. This happens either in the way that these questions have received a final answer within the realm of a certain science, or in the sense that those questions do not allow a scientific answer at all, not even in a ‘superordinated’ science. Aristotle assumes that it is in principle possible to find such ultimate explanations or explications. As we have seen above, some of his requirements for the premises of proofs – particularly those of being immediate

THEORIES OF ARGUMENTATION

103

and being ‘first’ or absolutely primary, not only comparatively prior – apply to such ultimate cases alone. They are so strict because they apply to premises that ultimately answer the two basic scientific questions. Aristotle underlines that the premises of the proofs have to be unprovable. We have seen above that he means unprovability in a highly technical sense. The premises in fact can be argued for also from true and even from necessarily true premises, but such an argument is not a proof in the strict sense, because its premises are less well known in the order of nature, although better known to us. Therefore, Aristotle in the end allows many ways of establishing the premises. However, he is particular in pointing out – for instance in chapter I 3 of the Posterior Analytics – that the premises cannot be known in the same manner as the conclusion that is proved. The conclusion is known as proved (%/iqr3,g 8/ndeijrij3) and because the premises cannot be proved in the strict sense they cannot be known in this way either. If they were, an infinite regress would follow. But surely they have to be known in some way. Aristotle recognises this and calls such knowledge ‘unproved knowledge’(%/iqr3,g 8la/5deijrnp, An. Post. I 33, 88b35–37). In the last chapter of the Posterior Analytics Aristotle finally treats the question of how the unprovable starting points of proofs become known. Even in the context of the Posterior Analytics, its last chapter is considered especially problematic. The chapter has been a matter of almost incomparable scholarly controversy for a long time and many have found Aristotle’s answer obscure or even frustrating.174 In the following I shall discuss the chapter. My main aim is to attempt to clarify in outline how the chapter explains our knowledge of scientific starting points. More detailed scholarly questions will be primarily taken up in the footnotes. Posterior Analytics II 19 is likely to cause disappointment in a reader because the chapter is supposed to satisfy very high expectations: to explain how we can know the premises of scientific proofs in such a way that they are known through themselves and even better known than the conclusions. Then, when we arrive at the chapter, Aristotle provides a very short description of how we acquire universals from experience.175 After that description he provides an obscure metaphor (100a12–15) of soldiers making a turn and 174

Burnyeat (1981, 133) characterises it as ‘perfunctory in the extreme’. For discussions concerning the chapter, see, e.g., Kosman (1973), Lesher (1973), Hamlyn (1976), EngbergPedersen (1979), Kahn (1981), Modrak (1987) and Sorabji (1993). Cf. Ross (1949), Barnes (1975), Hintikka (1980), Couloubaritsis (1980), McKirahan (1992) and Frede, M. (1996a, 169 and 172–173). 175 Some scholars, e.g., Bolton (1991), however, consider the chapter not being psychological but methodological. Bolton argues that the final chapter of the Posterior Analytics should be read together with the first chapter of the Physics and, he argues, they together provide us with a description of an empirical scientific method.

104

CHAPTER ONE

returning to the original order; then he promises to explain the whole matter once again (100a14–b5), but the explanation is somewhat unclear. Aristotle indicates that the chapter should answer the question of how the premises of proofs are known. Such knowledge should by all reasonable expectations be propositional.176 The premises are clearly propositions. However, contrary to some suggestions, I think it is problematic to understand the chapter in a methodological way. What he talks about in the chapter are our cognitive abilities. He simply says that first we make perceptions which are stored in our memory, from these an experience is formed (a kind of generalisation concerning the instances we have observed) and through experience starting points for art and knowledge are consequently acquired in the mind (100a1–9). Later on he compares some cognitive processes to a group of soldiers making a turn and coming back, and talks about some undifferentiated (100a16) things first getting at rest in our soul and finally leading to some unanalysable entities (8,e.‚, 100b2) doing the same. If we expect this to be a methodological description of how the premises are found in scientific inquiry, it is unsatisfactory. I must now add that I do not intend to dispute that there is a connection between Aristotle’s description and how he assumes we come to know the premises. However, I do not think his description is methodological or that the connection is quite straightforward. I shall explain how I see the connection; before that I shall add one remark concerning methodology. According to Bolton’s suggestion this empirical method should be sharply distinguished from the dialectical argumentation technique which is introduced in the Topics and which has a prominent role in the Aristotelian methodology. I agree with Bolton that experience does have a prominent role in the Aristotelian methodology and that Aristotle thinks that many things can become known through perception. However, I do not quite agree with Bolton that there is such a sharp and mutually exclusive distinction between the dialectical and empirical aspects of what can be called Aristotle’s methodology. In addition, I do not agree with Bolton on the point that the final chapter of the Posterior Analytics should be in essence methodological either. Typically, when Aristotle discusses methodological matters he is much more explicit about what we should actually do: he gives instructions concerning what kind of premises we should look for, how to organise those that we find, and so on. Such pieces of advice are markedly absent from the chapter. Therefore, I think it is a much more plausible reading that in the II 19 Aristotle is concerned with describing our cognitive ability to acquire correct universals from perception. I shall argue further for this point in the text. I shall also discuss the connection between the methodological and the psychological accounts of how we come to know the principles. 176 The question of propositionality is discussed, e.g. in Kahn (1981) and in Sorabji (1993).

THEORIES OF ARGUMENTATION

105

If we take Posterior Analytics II 19 to be a straightforward methodological description of how we come to know the premises of proofs, it becomes unclear what the techniques of argumentation are used for. Even though, as we have seen above, Aristotle does not assume that the dialectical technique could in a non-contestable way secure the premises, he considers the technique an important means of examining and evaluating suggested premises and also of arguing for them. He also allows arguments that establish a statement that in the proof functions as a premise through other scientific truths (e.g. the one establishing that the planets are near on the basis of that they do not twinkle). If it just happens in our soul (as Aristotle says on 100a13–14) that we come to know the starting point of knowledge (%/iqr3,g) and art (r2ulg), why do we not know all the premises just like that without any special extra effort of arguing and investigating? Let us now go back to the beginning of the chapter where Aristotle formulates the questions he is going to study in the chapter. He starts with a short summary of the previous discussion and then goes on to state the purpose of the final chapter: it is to answer the question of how the starting points (8.u3, 99b17)177 of knowledge become known and which cognitive disposition ()mip) knows them. Whereas it is difficult to see what Aristotle’s exact answer to the first part of this question is, the answer to the second one is quite clear: the cognitive disposition which knows the starting points of knowledge is our intellectual capacity (lnflp). After these introductory remarks Aristotle discusses some problems (99b20–32) and viewpoints related to the question of how the starting points become known. He first opposes the theory of recollection. According to Aristotle, it is not plausible that any knowledge of the starting points should be innate within us; this would entail that we should have knowledge that is more accurate than proved knowledge but that would remain unnoticed (99b26–27). However, the assumption that we acquire such knowledge is also problematic because, according to Aristotle, it requires that we must acquire it from some pre-existing knowledge.178 This assumption entails the possibility of an infinite regress of types of more accurate knowledge or cognitive dispositions (cf. 100a11).

177

I shall use the more flexible translation ‘starting point’ instead of ‘principle’ here for the Greek 8.u3, because I think the word ‘principle’ too readily makes us assume that we are talking about propositional premises and general logical principles. It is to my mind too early to make this commitment in the translation at the very beginning of the chapter. 178 Cf. I 1, 71a2–3.

106

CHAPTER ONE

Aristotle proposes to solve this problem by claiming that the knowledge of the starting points is acquired from pre-existing knowledge but this is not more accurate than the knowledge of the starting points. Rather, our knowledge of the starting points comes about through perception, our most basic cognitive capacity, that is found in all animals and is ready to function right from birth onwards. According to Aristotle, perceptions are stored in the memory and from the memory a general grasp of all the observed cases called experience (%,/ei.4a) is formed.179 In addition to experience – which also exists in other non-human animals – human beings also have the rational capacity (k5cnp, 100a2) that is developed from experience. Reason is capable of making more universal generalisations in contrast with experience. Experience is concerned with observed cases alone, whereas rational generalisations are about all the instances of, e.g. a species (cf. 100a7–8). Experience, however, is sufficient for practising arts and producing things, but genuinely universal generalisation is needed to initiate scientific inquiry.180 At the end of the description, Aristotle claims that a universal acquired from perception is a starting point (8.u3) of art and knowledge or science (%/iqr3,g). After this Aristotle concludes that because the starting point of our cognitive development lies in perception there is no infinite regress of more accurate and cognitively higher dispositions. Rather, the disposition which knows the starting point comes to be from perception – the cognitive capacity all animals have. Now follows the metaphor: Aristotle compares what he has said to a group of soldiers who retreat from a battle. Their retreat takes place so that when the first soldier stops the others follow him until the original

179

I shall discuss Aristotle’s description more closely in 2.3.1. There has been wide scholarly discussion on Aristotle’s phrase ‘from experience or from a whole universal (jah5kns) which has come to rest in the soul, which is one besides the many and the same in all of them becomes the starting point for art and knowledge’. The crucial point is how we take the ‘or’ (Greek }). There are three possibilities: (a) a genuine disjunction (‘or’), (b) progression (‘or rather’) and (c) epexegetic (‘that is’). McKirahan, for instance, (1992, 243) rules the first reading out on the grounds that it would commit Aristotle to a Platonic view on the acquisition of the universals. Many (e.g. Tricot 1947, Ross 1949 and Barnes 1975) take the third route and see the ‘or’ as epexegetic. According to McKirahan that interpretation contradicts the Metaphysics I 1, and he chooses the second alternative. Therefore, he understands the phrase as follows: ‘universal at rest in the soul’ is an intermediate stage between experience and scientific knowledge. I agree with Mc Kirahan here. However, I would

180

THEORIES OF ARGUMENTATION

107

arrangement has been restored. According to Aristotle our soul is such that it can undergo such a process (100a13–14).181 Aristotle admits himself (100a14–15) that this is not a very clear exposition of what happens and he promises to explain the same thing he tried to illustrate by the metaphor again. His second description is as follows. When the first undifferentiated (8di1tn.nl) [object] has come to rest, there will be a first universal in the soul, for even though the particulars (jah’ )jaqrnl) are perceived, perception is about the general (y d£ a©qhgqip rnfl jah5kns %qr4l), for instance of a human being not of a human being Callias;182 we remain in these until those which are universal (r¡ jah5kns) and without parts (r¡ 8,e.‚) come to a stop, for instance an animal of this sort, and animal, and similarly in that case. It is now clear that the first [starting points] become familiar (clw.4feil) to us in this way, because perception also implants the universal into us similarly. (100a14–b5; my translation.)

At the beginning of this rather difficult passage, Aristotle expresses the idea that we acquire some universals from perception. They are as yet unspecific or undifferentiated. ‘Undifferentiated’ (8di1tn.nl) – literally to be translated as ‘that without a difference’ – has by some scholars183 been taken to refer to the so-called infimae species, that is to such species that cannot be further divided like to point out that I do not think his argument for ruling out (a) is conclusive because of the following reason. Aristotle is free to make the distinction between experience and the ‘whole universal’ without a commitment to a Platonic account for the acquisition of the universals. This is if experience concerns merely the cases we have perceived but is not about those yet to come. The ‘whole universal’, by contrast – and as Aristotle himself explains in the text – is about all the instances, not only about those we have encountered. Therefore, I think we cannot rule out (a) that easily. In fact, I think it gives us a plausible reading here. According to this reading, Aristotle is pointing out that experience about past cases through experience is sufficient for arts and production but in science we need genuinely universal insight which is achieved from experience but is not identical with it. For this distinction, see further 2.3.1. 181 The Greek for this is y d£ vsu¢ ∫/1.uei rnia6rg nœqa n˙a d6laqhai /1queil rnflrn. 182 I think this parenthetical remark is perhaps the most difficult aspect of this passage. Possibly Aristotle means something like the following. Even though we perceive only particular instances of a species, our perception in some sense is also about the species on a general level. If he is at all in line with what he says elsewhere (e.g. De anima) he cannot mean that the species as universals are objects of perceptions. However, even though we do not perceive the universals, it might be the case that perception provides us with the first familiarity with the universals. In the framework of Aristotle’s theory of human cognition, a universal might somehow carried to the soul in perception even though the perceptual faculty is unable to grasp it. 183 E.g. Ross (1949).

108

CHAPTER ONE

into species, e.g. human beings and horses. However, I think this is too technical and commits Aristotle to the position that we at first learn universals as species. This does not seem to be right because we often encounter things whose species is not clear to us. If I – who am not an ornithologist – for instance, see an exotic bird, I only recognise it as a bird, not as definite kind of bird.184 Aristotle recognises that there are such cases, and I do not think that he would say that we always learn the species first. Rather, he points out that we at first learn some universals from which we can proceed to more specifically articulated and more general ones, even up to the notions of the categories.185 Rather than to species, I take ‘undifferentiated’to point to the idea that at first our universal notions are unspecific and unarticulated.186 In spite of the difficulties, one thing becomes clear. Aristotle refers to species and genera. That is, he explains his previous rather unspecific description by pointing out that what we learn from perception through memory and experience are the universals as natural species and genera. If we now look back to the more detailed descriptions of Aristotelian scientific syllogisms and their premises we notice that in the explicatory syllogisms, expressing the structure of the species and genera with specific differences, predication of the genus of a species appears in the conclusion. The major term, the genus, appears both in the conclusion and in the first premise. Therefore, what we need to find in addition to that to complete the proof is the middle terms that express the specific differences (a specific difference is in Greek called diatn.1; cf. 8di1tn.nl). It is possible that a similar account applies to how we come to know the conclusions of the explanatory syllogisms even though Aristotle does not have an explicit example of that sort in the Posterior Analytics II 19. It is highly likely that Aristotle would say that also statements of the sort ‘planets do not twinkle’, ‘the moon is eclipsed’, or ‘there is a noise in the clouds called thunder’ are learnt through perception, memory and experience. It does not seem to require much intellectual effort to come to know these statements as such. 184

Here I agree with Bolton (1991). The suggestion that those ‘without parts’ (r¡ 8,e.‚) are the categories is found, e.g., in Ross (1949, 678). The expression r¡ 8,e.‚ ‘without parts’ is unusual. It is possible that the reference is to categories, but this reading cannot be confirmed with certainty. 186 Cf. the first chapter of the Physics, where Aristotle points out that children at first call all females ‘mother’ and all males ‘father’. Therefore, they are not able to make the difference between what a mother is and what a female is. However, this can still be understood as the child’s first ascent to the general level. 185

THEORIES OF ARGUMENTATION

109

Therefore, what Aristotle does in most of the Posterior Analytics II 19 is that he explains our knowledge of starting points for knowledge and science through a natural cognitive process of generalisation from perception through memory and experience. After that he lists our cognitive dispositions and distinguishes knowledge in the strict sense (%/iqr3,g) and intellect (lnflp) from other cognitive dispositions on the basis that they are always true. The first explicit reference to the intellect appears in the last section of the chapter (100b8). There it is classified among rational dispositions (/e.§ r¢l di1lnial )meip, 100b5–6), and among them it is supposed to be the one that grasps starting points for knowledge (100b12). This, however, is not explicitly linked with the preceding discussion concerning generalisation from perception. The only literal clue we have is the claim (on lines 100a6–9) that the starting point for knowledge is the universal we have in virtue of generalisation through perceptual experience. In addition, we have seen that Aristotle indicates later in the same paragraph that we come to know species and genera through that process. We also noted that in the scientific context this involves knowledge of the conclusions of explicatory syllogisms. We are now in a position to make the connection between knowing the premises and knowing the conclusions. The reader should keep in mind that in the Aristotelian context the conclusions function as the starting point for inquiry because they are better known to us. Inquiry is mostly concerned with the things ‘in between’ (cf. An. Pr. I 27, 43a42–43); i.e. those things we need to connect the minor and the major term of the syllogism in an explanatory (the nearness of the planets explaining why they do not twinkle, or the earth blocking the sunlight from the moon when the moon is eclipsed) or explicatory way (rationality or some such difference separating human beings from other animals, or having the sum of the internal angles equal to two right angles as the specific difference of triangles). I suggest that Aristotle assumes in the chapter that also recognition of the explanatory or explicatory factors is a function of the intellect. However, it does not happen to us automatically that we come to know complex theoretical structures. Rather, our intellect functions when we recognise the explanatory or explicatory force of the items expressed by the middle terms. For instance, we first grasp through perception that the planets do not twinkle or that the moon is eclipsed in the way that these phenomena are eternal and necessary or ever-recurring. Then by investigating the proposed solutions to how these facts are to be explained and by making observations ourselves, we can arrive at apprehending them as well. For instance, by perceiving that our hand can block the light of a lamp and a shadow falls on the table, we can come to apprehend that loss of light in general, and hence also an eclipse, must be explained in a similar way. Or, when at sea, we observe that distant lights

110

CHAPTER ONE

seem to twinkle; when we come closer they seem stable. From this we can come to understand – given that we know that planets are closer to us than other celestial bodies – that the non-twinkling of planets is explained similarly. The idea that the intellect is able to recognise explanatory relations appears in the first book of the Posterior Analytics (chapter 34). There Aristotle talks about those who have a quick intellect (the name of the property in Greek is 8cuiln4a). He explains that having a quick intellect involves the ability to recognise explanations quickly. Therefore, intellect as such is connected with recognising explanatory force. Let us now return to the question of propositionality. It seems clear that knowledge of the premises of scientific proofs is not quite the same as knowing the premises as simple propositions. To understand that planets do not twinkle because they are near – and that the short distance explains non-twinkling in general – is different from merely knowing the proposition ‘they are near’. Aristotle also points out earlier in the treatise (I 2, 71b33) that it is not sufficient merely to know the reason; we also have to know that it is the reason. It is difficult to say whether this kind of intellectual apprehension is propositional or not. On the one hand, our intellect is the disposition to grasp species and genera as species and genera. On the other hand, the same disposition is at work when we grasp explanatory connections and probably also specific differences. This leads me now to a comparison between the intellectual disposition and our capacity to move our limbs.187 All human beings who are not seriously disabled learn to walk quite naturally. Most of us are also in principle capable of performing advanced acrobatic movements. However, we are not capable of performing such movements just as naturally as we learn to walk, for we need a lot of practice. Similarly in the case of the intellect: we acquire, as naturally as we learn to walk, some genuine universals in the mind in virtue of our intellectual capacity. As in the case of moving around, some of us are naturally gifted (those who Aristotle thinks have a quick intellect) but no-one grasps all the explanatory relations or specific differences immediately at first glance. We need a lot of experience, systematic inquiry and argumentation in order to be able to understand these things. It is for such practice that the techniques of argumentation are useful. They enable us to analyse the matter, see where the problems lie and, perhaps, learn the truth for ourselves. *

*

*

We should now pause for a brief summary of the discussion concerning Aristotle. We have first discussed the dialectical technique of argumentation 187

A similar comparison also appears in Alexander of Aphrodisias; see 2.3.2 below.

THEORIES OF ARGUMENTATION

111

and recognised, with some other scholars, that it is a vital assumption behind that technique that the arguments have an epistemic aim to increase the credibility of the conclusions. The arguments manage to do that if their premises are initially more acceptable than the conclusion and the conclusion follows from the premises necessarily. Aristotle also excludes redundant premises and points out that the premises must contain sufficient reasons for accepting the conclusion. Aristotle systematises dialectical argumentation by the so-called topoi. They include general premises, advice concerning how to construct an argument, and general notions according to which premises and strategies can be classified. Many of the general premises are such that they are generally acceptable but perhaps not literally true. In the discussion of starting points in science I have emphasised the distinction between starting points for inquiry and such principles towards which our inquiry is directed, and that figure in the scientific proofs as premises. Aristotle sets very strict conditions on those premises; they are supposed to express the ultimate answers to scientific questions about explanations and definitions. Aristotle also recognises that we have to assume some general principles he calls ‘common axioms’; some of them concern the form of arguments and some are general mathematical principles. The axioms very rarely appear as explicit premises. Aristotle makes the peculiar point that general axioms are not the same if applied to different kinds of objects. The doctrine was particularly influential in mathematics. I have suggested that even though Aristotle is clear that every chain of proofs ends somewhere, namely to the ultimate principles, this does not entail that Aristotle’s model for the sciences should be an axiomatic-deductive ideal. In the axiomatic model, there is a small group of self-evident axioms from which all theorems can be deduced through valid inferential steps. In the Aristotelian science, the premises are not initially evident; they are found in the course of inquiry. In addition, as we saw, Aristotle makes it clear that in a science there are almost as many premises as there are conclusions. Therefore, I have suggested that an Aristotelian science should be rather understood as a complex network of fairly short syllogistic chains organised around the objects of the sciences. Finally, we have discussed the question of how the unprovable premises of sciences become known. I have suggested that Aristotle in the last chapter of the Posterior Analytics provides us with an account of how we acquire the starting points for sciences from everyday experience. The starting points, however, are not yet the premises; we start from what is better known to us and aim at finding the premises. The starting points for science include rather humble generalisations from experience, and they are formed in our soul quite naturally. From them we start to inquire into the explanations and

112

CHAPTER ONE

definitions. When they are found, we can formulate the proofs by positing the explanatory factors or specific differences as the middle terms of apodeictic syllogisms. I have also suggested that the same intellectual capacity that provided us with the initial generalisations to start from, functions when we recognise the explanations and the specific differences in the course of inquiry. Recognising them is based on understanding their explanatory force or how they serve to distinguish a particular natural species from others. Even though in my discussion of the Posterior Analytics II 19, I have emphasised universals acquired from experience through a natural cognitive process, I have recognised that sometimes often Aristotle assumes that we start our inquiry from appearances in the sense of reputable conceptions, most importantly, the predecessors’ theories. Perhaps it depends on the subject matter and the existing research on a topic which starting points are preferred. 1.3 LATER DEVELOPMENTS In the previous two chapters we have discussed Plato and Aristotle. We have seen that both Plato and Aristotle start with a communicative conception of argumentation pointing to an epistemic aim. Arguments are conceived as encounters between two discussants, one asking questions and the other answering. As Aristotle makes clear in the Topics, the aim of such arguments is to enhance the credibility of the conclusion by using premises that are initially more credible than the conclusion. Both Plato and Aristotle recognise a class of principles we could call ‘logical’ involving, most importantly, the principle of non-contradiction. As Aristotle points out explicitly, these principles appear very rarely as explicit premises of arguments. In addition to general principles regulating argumentation, both Plato and Aristotle share the view according to which we must recognise two basic kinds of premises which have content. One is that from which we start our inquiry. On the basis of such premises we then try to establish more general principles of reality; those principles explain the initially known facts. These two processes are intrinsically directed; the first aims at establishing objective explanatory principles, whereas the second starts from such principles. In this chapter we shall discuss later developments in the Platonic-Aristotelian tradition, paying particular attention to the distinction between the two directions: showing some basic principles of reality and proving initially known facts on the basis of them. The most important new developments are the following. Firstly, there is Galen’s attempt to understand medical science – which he takes to be the superior form of human knowledge – as being built on self-evident principles only. Eventually, Galen ends up with accepting something analogous to the Aristotelian distinction. He realises that a science needs principles

THEORIES OF ARGUMENTATION

113

concerning general regularities in nature, such as the causes and nature of things, and these are not evident to human beings in any ordinary sense of this word. Secondly, when Platonism starts to turn into Neo-Platonism, the distinction between intellectual activity and ordinary rational activity, pregnant with metaphysical background assumptions, comes into focus. To some extent this turn is already visible in Alcinous’middle Platonism and it receives its full formulation in Plotinus. The basic point of the distinction is that argumentation in general can only take us as far as the conclusions go. It can never force us into intellectual apprehension. Intellectual apprehension involves grasping complex wholes at a glance, as it were. Such apprehension amounts to understanding how the parts of a theoretical whole fit together. Such understanding can afterwards be articulated in reasoning as chains of propositions, but such chains cannot express the whole content and complexity of the intellectual insight. I shall first trace some later developments in the Platonistic side of the tradition. After that I shall do the same with respect to the Aristotelian side, and discuss some of Aristotle’s commentators in late antiquity. 1.3.1 Some Developments in Platonism Galen There was an active interchange between doctors and philosophers in antiquity in general. Doctors often relied on philosophers, but they also took part in philosophical discussion, particularly concerning the nature and methodology of medicine – so much so that the discussion at times was rather independent of the philosophical developments.188 Galen is by far the best-documented ancient doctor. Of the enormous number of works he wrote many have come down to us. He considered himself to be a Platonist, and in some cases this also affected his experimental results.189 Galen’s fame is so great that it might have shadowed the earlier doctors.190 In general, Galen is of the opinion that the best kind of knowledge we can have about the world is scientific knowledge and, more specifically, medical scientific knowledge conceived in Galenic terms. Galen’s medicine includes aspects that would nowadays be classified as biology.191 188

Frede (1987d). Galen was ready to use evidence he got by dissection to show that the seat of the intellect is in the head, because this was in harmony with Platonism. However, he did not want to use the same evidence to locate the centre of desire in the head, but claimed that it lies in the liver. See Tieleman (2002). 190 See Tuplin and Rihll (2002). 191 See, G. E. R. Lloyd (1996, 256). 189

114

CHAPTER ONE

According to Galen, scientific premises are true and appropriate relating to the essences studied. In addition, they are clear either to perception or to reason (Plac. Hipp. et Plat. 2.3.108, 26 ff. 3.1.168, 15–16). Further, the premises are also primary (/.‡rnp), unprovable (8la/5deijrnp), and convincing by themselves (%m ^asr‡l /iqra4) (Methodo Medendi 1.4.6).192 What is apparent to the senses or to reason does not, according to Galen, require proof (see Meth. med. 1.4.12; cf. also Inst. log. 1, p. 3 in Kalbfleisch’s edition of 1896).193 Galen is willing to accept as premises of scientific proofs two types of statements, axioms and definitions. Galen’s axioms include such principles as that of non-contradiction, and that of an excluded middle as well as the principle concerning the subtraction of equals (if equals are subtracted from equals the remainders are equal) (Meth. med. 1.4.10).194 The last-mentioned principle is also Aristotle’s example of common axioms in the Posterior Analytics. These principles can be characterised as being clear in the sense of being self-evident or being beyond reasonable doubt.195 However, Galen also allows other axioms the self-evidence of which is not equally clear. These include a number of axioms concerning causes such as ‘nothing occurs without a cause’, ‘everything comes to be from something which exists’, ‘nothing is annihilated into the absolutely non-existent’ (Meth. med. 1.4.10),196 ‘nature does nothing in vain’,197 ‘what is changed takes on a 192

For a study of such Aristotelian concepts in Galen, see Tieleman (1996); for the last three requirements, see also Hankinson (1991a, 117–130), and for Galen’s epistemology in general, see Frede (1987e). 193 The edition is published in Teubner’s series, Leipzig. The text is damaged on the first page of Galen’s Institutio logica. Nevertheless, the idea of these two classes of principles can be understood from the text without Kalbfleisch’s suggestions concerning lacunas. 194 Galen also mentions the following Euclidean principles: ‘two quantities equal to a given quantity are equal to each other’ and ‘equals added to equals yield equals’ (Meth. med. 1.4.10). 195 It is true that the sorites argument, which some philosophers today understand in the way that it shows that the principle of the excluded middle is not universally valid, was known to Galen. He is our main source of evidence for the debate between medical empiricists and rationalists, and the sorites argument appears in the debate. Barnes (1982) has suggested that the medical empiricists were predecessors of manyvalued logic; this is not a necessary consequence of the discussion. Be this as it may, Galen does not seem to lose his confidence in the principle of the excluded middle. For the debate and the sorites argument, see below chapter 3. 196 For a discussion of these principles, see Hankinson (1998b, 376). 197 This principle is mentioned by G. E. R. Lloyd (1996, 266), who points out that Galen admits that this principle might not be true without exception.

THEORIES OF ARGUMENTATION

115

form similar to that which causes the change’, and ‘it is impossible that, when two bodies come together, they should not both act and be acted upon’ (Plac. Hipp. et Plat. 5. 566–567). In addition, he mentions a generally accepted medical axiom according to which opposites cure opposites.198 Galen must have been aware that there was a philosophical debate about the axioms concerning causes during and before his time. The Epicurean atomistic theory, for instance, includes the assumption that atoms sometimes swerve from their usual routes without a cause, and also the teleological principle was questioned by the ancient atomists. Therefore, these general principles listed by Galen cannot be taken as being beyond doubt in the way the first mentioned logical-mathematical axioms can. Furthermore, there are two types of scientific premises, which cannot be self-evident in any ordinary sense. Firstly, Galenic definitions form the second class of premises of scientific proofs. Galen is an essentialist of a kind and assumes that essences cannot be perceived. He seems to suppose that they are not directly known by reason either, because inquiry and conceptual analysis are necessary requisites for knowing them. Secondly, as has been pointed out by G. E. R. Lloyd,199 Galen allows that premises are found out by dissection, and that these kinds of premises should not be considered in any way as self-evident. Now Galen faces the following problem. On the one hand, he requires that scientific proofs should start from self-evident principles no one in his or her right mind would question. On the other hand, he is aware of the fact that medical science cannot be built on logical axioms alone. Two suggested solutions have been put forward to this problem in scholarly literature. According to the first one made by Jonathan Barnes, Galen claims that in the course of inquiry scientific premises other than logical axioms become evident to a scientist. Therefore, they can be called self-evident in the limited sense that they are evident to a specialist who has learned the relevant field of study thoroughly.200 In some cases Galen allows that one can argue in favour of the scientific premises,201 and such arguments can probably make the scientific premises evident to those who are not experts. The second suggestion put forward by Jim Hankinson is that Galen eventually endorses the Aristotelian distinction between two types of principles. Some are principles

198

For a discussion of the last-mentioned principle, see G. E. R. Lloyd (1996, 267–269). 199 G. E. R. Lloyd (1996, 271). 200 See Barnes (1991). 201 Ibid. p. 70 n. 65.

116

CHAPTER ONE

more familiar to us,202 others are prior in the order of nature. We can establish and make known the principles prior in the order of nature through principles more familiar to us, but this is not strictly speaking a proof.203 A proof in the proper sense starts from premises prior in the order of nature and its conclusion is more familiar to us. Ultimately, these two solutions are not very far from each other. We can assume that in the course of inquiry the cognitive structure of the scientist becomes altered. What before starting to do research seemed remote and unknown, even suspicious, becomes familiar and evident in the process of inquiry when its explanatory force with respect to perceived phenomena becomes evident. This can be described as a transformation of the cognitive structure into one which is – with respect to the studied field – in accordance with the order of things. Medical science, according to Galen, need not be a comprehensive system of proofs organised into continuous deductive chains. By contrast, it contains, as it were, separate proofs with not so many deductive inference steps. It might be thought that in this respect Galen’s conception differs from the Aristotelian one. However, as I have argued above, the Aristotelian sciences, too, consist of short proof chains rather than of continuous deductive chains proceeding from a small number of unprovable premises. Therefore, these kinds of structural considerations do not make the Aristotelian and Galenic conceptions of scientific proofs different from each other. We saw above that one group of scientific axioms, according to Galen, consists of definitions. In searching for definitions, Galen says, one should discover correct specific differences.204 In listing different diseases one should use the method of division from first specific differences until such species are reached that cannot be further divided (Meth. med. 1.3.10).205 Galen points to the method of division which is used to articulate the specific 202

For these kinds of starting points, see Meth. med. 1.5.3 and 1.4.3. Galen also calls them ‘common conceptions’ (jnila§ (llniai); cf. Meth. med. 9.178; 199. 203 See also Hankinson (1991a, 131). 204 Galen’s formulation of how one should proceed seems a little problematic. He says that in order to find the correct specific differences, one must first present a correct definition, i.e. a formula (k5cnp) of substance (Meth. med. 1.3.8). But this, again, requires that one knows the correct specific differences, and the process is in danger of becoming circular. Problems of circularity also emerge in Aristotle; for a discussion of these problems in the commentary tradition, see below 1.3.2. 205 Galen also says that even Plato and Aristotle did not develop the method of division to its full maturity (Meth. med. 1.3.11). Possibly he thinks that he himself has succeeded in accomplishing this task.

THEORIES OF ARGUMENTATION

117

differences. Therefore, the definitions Galen talks about are not self-evident either. Rather, they are found as a result of inquiry using methods of conceptual analysis resembling those we find in Plato and Aristotle. Galen classifies this method as a method of discovery. As such it is to be distinguished from criteria of truth. The criteria, on the basis of which the truth of the discoveries is to be decided, are for all human beings experience and reason. In Galen we find in a nutshell the idea that scientific discovery aims at building theories which can be tested by experience. In building theories ‘logical’ techniques such as the method of division are used, whereas in testing them we use the two basic cognitive capacities of perception and reasoning. Sometimes Galen also presents views on the metaphysics of knowledge, although he in some cases wants to withdraw from speculations concerning the nature of a scientific object.206 He endorses a version of the PlatonicAristotelian conception of an intelligible structure of reality, although he does not seem to talk about forms.207 He says that there is a factor in, for instance, dogs and human beings in virtue of which they are dogs or human beings (Meth. med. 2.7, 3–4).208 He says that ‘insofar as [a man] is a man he has the very same feature as any man you care to mention’ (Meth. med. 2.7, 6). By contrast to Aristotle in particular, Galen conceives the thing in virtue of which all the instances of a certain species are the things they are as a property. According to Galen, this property is not visible, but he says that it would be childish to imagine that for this reason there would not be such a thing [i.e. property] as man in itself or disease in itself (Meth. med. 2.7, 36). On the basis of the treatise on therapeutic method (Meth. med.) we can draw the following picture of Galen’s views concerning human knowledge and its metaphysical background. Galen assumes that there are natures or essences of things which cannot be perceived and are not directly knowable to human reason either. They are expressed in definitions that constitute an important subgroup of unprovable premises. The definitions are such premises of scientific proofs which are a result of scientific inquiry rather than its starting point. This resembles Aristotle’s conception according to which the premises of scientific proofs are found in the inquiry and when we have found them, we can construct a proof in the proper sense from them. 206

Galen is, for instance, reluctant to present a doctrine on the nature of the soul, see Hankinson (1991b). 207 It has been suggested (Hankinson 1991a, 209) that Galen’s metaphysics of knowledge is not a developed theory but is based on his quite robustly realistic intuitions on language and his reference to classes. 208 Galen in fact claims that both Aristotle and Theophrastus take forms to be general (Meth. med. 2.7, 16).

118

CHAPTER ONE

(However, as we have seen above, for Aristotle the conclusion of a proof is an important part of the definition.) In addition, Galen’s use of the notion of self-evidence points to an interesting tension in his position. On the one hand he wants, in a manner resembling the Hellenistic discussion (see below chapter 3), that the premises of proofs are self-evident and, hence, unquestionable. On the other hand, he acknowledges that not much can be derived from self-evident premises by logically valid inferences. This makes him enlarge the class of self-evident principles. However, this manoeuvre leaves open the question of to whom the principles are supposed to be self-evident. Further, this question reveals the unwelcome feature of the notion of self-evidence that it is not objective. It is always related to a subject to whom something appears or does not appear self-evident. If the subject of self-evidence is understood as the numeric majority of humankind, it is clear that the scientific premises cannot – and even should not – be self-evident in this sense. If, on the other hand, it refers to scientists or philosophers, it is well known that their opinions diverge. It is possible that Galen means something like ‘being evident to an expert’ which, on the other hand, comes close to the ideal of objective self-evidence. If something is evident to someone who has learnt a science thoroughly, this is as close as we can get to what it is to be evident as such or prior in the order of nature. The learned scientist can be taken to have the best perspective on things and their nature. In his accounts of the premises of scientific proofs Galen infiltrates two different ancient tendencies, the first of which can be called Platonic-Aristotelian and the second Hellenistic. The Platonic-Aristotelian tendency is to see reality as such as intelligible and to take the nature of things to be in principle apprehensible by human beings. This is reflected in the idea also accepted by Galen that definitions express the nature of things and that they can be found out by methods of conceptual analysis. The second Hellenistic tendency is to aim at fortifying the conception of knowledge against sceptical attacks by accepting only evident premises and valid inferences from them. Alcinous209 The middle Platonist Alcinous is perhaps an earlier figure than Galen, who lived from 129 to c. 210 AD. However, it is far from certain when Alcinous

209

Scholars are nowadays in agreement that the author of the Handbook of Platonism (Didaskalikos) is to be called Alcinous; see Sedley (1996, 300–301 n. 1), Dillon (1993) and Whittaker (1990). This means that the conjecture – made about a hundred years ago by Freudenthal – that the author should be identified with a Platonist called Albinus, has been abandoned.

THEORIES OF ARGUMENTATION

119

lived. Many scholars place him in the second century AD,210 but some take it to be possible that he lived in the first century.211 In any case, Alcinous serves to exemplify some developments which were to become dominant in Neo-Platonism. Alcinous is a good example of a general late ancient tendency to include in one’s own doctrine elements from Plato and Aristotle and also from the Stoics. Alcinous combines elements from Platonic collection and division and the Aristotelian syllogistic without excluding one or the other. It is possible that Alcinous considers the Platonic method particularly helpful in the process of inquiry, whereas syllogistic is suitable for presenting the results in an appropriate manner (see Didaskalikos 5–6). This idea is in fact not alien to Aristotle either, but Aristotle insists on certain improvements on the Platonic division.212 It is understandable that the Aristotelian syllogistic can be fitted fairly easily into a Platonist framework. A Platonist is willing to express relations between forms and the Aristotelian syllogistic fits this purpose quite well. In the syllogistic figures relations between – usually – general terms are expressed and the validity of syllogisms is based on the transitivity of class inclusion. Alcinous’ presentation of cosmology (Didaskalikos 12–16) builds on the Timaeus and he expresses Plato’s assumption that the whole cosmos is, as it were, a living organism arranged according to how the forms are interconnected in the creator god’s thought. Alcinous seems to assume that in order to have real knowledge we should get to know the necessary arrangement of ideas and in this enterprise the method of collection and division should be used. When the interconnections have been grasped, they can be expressed in Aristotelian syllogisms.213 In what Alcinous says about dialectic, he uses the distinction between principles as starting points for inquiry and as premises of proofs expressing the relations of priority in nature. Alcinous claims that dialectic proceeds either ‘from above’ by definition and division or from below by means of analysis (Didaskalikos 5.1). He says that definition, division and analysis differ from inductions and syllogisms. Inductions and syllogisms are used in examining attributes that belong to the substances. Induction is said to be a way of 210

See Dillon (1993) and Donini (1994), 5057–5058. Sedley (1996, 300–301 n. 1). 212 For division in Aristotle, see An. Post. II 5; 13 discussed above pp. 65–68. 213 In addition to the Aristotelian syllogistic, Alcinous’ view on argumentative form also has Stoic ingredients. In what he says about hypothetical inference, he uses the Stoic unprovables, particularly the first one (modus ponens) (see, e.g. chapters 6.3 and 6.6 of the Didaskalikos). There are instances of such inferences in Alcinous’ own argumentation as well (e.g. 10.7). 211

120

CHAPTER ONE

inquiry from the point of view of individuals, syllogism from the point of view of universals. We can get a clearer picture of the situation on the basis of Alcinous’ example where the immortality of the human soul is proved on the basis of the necessary characteristics of the human soul. Here Alcinous in fact distinguishes between three different ‘methods’of analysis. The first one does not seem to be a method or involve any argument structure. Rather, it is a process where we come to grasp more and more general intelligible forms; starting from perceptible beauty we proceed to beauty in souls and, finally, enter the great sea of beauty. The language is highly reminiscent of the Symposium of Plato. The second type of analysis is described in some detail. The passage reads as follows. [O]ne must postulate what is being sought (∫/nr4heqhai de‹ r• fgrn6,elnl), and then consider what other propositions are prior (/.5re.a) to it, and demonstrate (8/ndeijl6eil) these by ascending from secondary propositions to more primary ones, until we come to know that which is absolutely primary (/.‡rnl) and admitted (by all) (…,nkncn6,elnl), and beginning from this we will arrive at what is being sought by a procedure of synthesis (jarekesq5,eha qslherij+ r.5/–). For example, if I am enquiring whether the soul is immortal, I first postulate this very thing and then enquire whether it is ever-moving. Having demonstrated this (rnflrn 8/nde4map), I enquire if what is ever-moving is self-moving; and in turn, having demonstrated this, I investigate whether what is self-moving is a first principle of motion; and then, whether a first principle is ungenerated, which is taken as universally agreed (®p …,nkncn6,elnl), the ungenerated being also imperishable.And starting from this proposition, which has the quality of self-evidence (%la.cnflp ≈lrnp), I produce by synthesis (qslh3qw) such a proof (8/5deimip) as follows: a first principle is something ungenerated and imperishable, the first principle of motion is the self-moved; but the self-moved is the soul; therefore the soul is imperishable and ungenerated and immortal. (Didaskalikos 5.5, 157, 22–36; translated by Dillon 1993, with minor modifications.)

Basically, the idea of the procedure is the following. In order for us to prove that the soul is immortal we must first show that the soul is the first principle of motion. If this is shown, it will be necessary that the soul is also immortal. Alcinous assumes that it is universally agreed that the first principle of motion is necessarily ungenerated, imperishable and immortal. To show that the soul is the first principle of motion Alcinous presents two inferences that are syllogistic. The terms involved are: A immortal, B first principle of motion, C self-moving, D ever-moving and E soul. In the first syllogism the terms are self-moving (C), ever-moving (D) and soul (E); it shows that the soul is self-moving through the middle term that it is evermoving as follows. All ever-moving things are self-moving. (CaD) Every soul is ever-moving. (DaE) Therefore every soul is self-moving. (CaE)

THEORIES OF ARGUMENTATION

121

The second syllogism uses the conclusion of the first one as its second premise to show that the soul is the first principle of motion. The major premise is that self-moving is the first principle of motion. The syllogism is as follows: Every self-moving thing is the first principle of motion. (BaC) Every soul is self-moving. (CaE) Therefore every soul is the first principle of motion. (BaE)

This procedure of establishing the premises needed in the proof in the proper sense is called analysis. Now the proof, according to Alcinous, can be constructed by synthesis from the universally agreed and unprovable premise that the first principle of motion is immortal, and from the premise shown by the above analysis according to which the soul is the first principle of motion. The conclusion will be that the soul is immortal, and the syllogism goes as follows. Every first principle of motion is immortal. (AaB) Every soul is the first principle of motion. (BaE) Therefore, every soul is immortal. (AaE)

In the first syllogism the crucial assumption is that souls are ever-moving. In his description of the procedure Alcinous says that he has demonstrated this (8/nde‹map rnflrn). However, he does not explain how this is demonstrated. The syllogism (or ‘analysis’, as Alcinous calls it), presupposes that we manage to establish that all souls are ever-moving quite independently of this particular piece of analysis. The syllogisms involved in the analysis do not prove it to us. Hence, we might easily object that even though we have conceded that the soul is a first principle of motion, it is such only in those things whose soul it is. The argument has not shown us that there are no perishable ensouled beings. Such an idea is involved in the claim that all souls are ever-moving. But, as we saw, that particular claim was not demonstrated in the procedure. Despite the problem related to how we are supposed to know the premises, we can see that Alcinous’ description of the analytic-synthetic argumentative procedure resembles Plato’s method of hypothesis discussed above in 1.1. Especially at the beginning of the quotation above, Alcinous’ description is very much like Plato’s second phase of the hypothetical method in which an argument for the hypothesis itself is sought. As in Plato’s presentation, we have here the assumption that there is a point ‘above’ which one cannot ascend. Such a point is called by Alcinous ‘absolutely primary’ and ‘admitted by all’. Plato’s expressions are ‘something adequate’ (ri ßjal5l) in the Phaedo and ‘a non-hypothetical principle’ (8ls/5hernp 8.u3) in the Republic. However, Alcinous briefly discusses the method of hypothesis just after the passage we quoted (Didaskaliskos 5.6) and this discussion includes both

122

CHAPTER ONE

phases of the Platonic method. Therefore, Alcinous distinguishes his analyticalsynthetic method from the Platonic method of hypothesis. Alcinous presents a short summary of Plato’s method of hypothesis and calls it ‘analysis through hypothesis’. Presumably this indicates that the procedure is the third type of analysis Alcinous recognises. His description is a paraphrase of what Plato says in the Phaedo.214 Nonetheless, the important common feature in Alcinous’ procedure and the discussions of Plato and Aristotle is the following. It is assumed that there is an objective order between things so that some are primary to others in nature. When we ascend towards these naturally prior things we find something that is supposed to explain and, also, to prove (in Aristotle and Alcinous) the claim we started with. In Plato the original hypothesis, as in the Phaedo, can already be quite high in the explanatory structure of reality; the example is a version of the theory of forms. In Alcinous’ case the alleged facts that the soul is a first principle of motion and that a first principle of motion is also immortal explains the soul’s immortality. This immortality follows from being the first principle of motion. As we noted, the additional assumption is needed that the soul is ever-moving, and it is not at all clear how it was ‘proved’. As has been already indicated, Alcinous claims that the premises of proofs should be ‘accepted by all’. It is not quite clear what he means by this phrase. The unprovable premise, which Alcinous takes as being accepted by all and being self-evident, is that ‘every first principle of motion is immortal’. It is unlikely that such a principle could be understood as self-evident. Perhaps, if it referred to the divine first principle of the universe we could grant that within a certain framework such a principle must be accepted. However, here the first principle of motion cannot be understood in this sense because the human soul is also assumed to be the first principle of motion in this sense. Plotinus When we turn to Plotinus we encounter Neo-Platonism in full bloom, but I shall not be able to explore the details here. Rather, I will inquire whether we encounter the kind of distinction we have traced above in various thinkers according to which we must distinguish between a natural order of priority and priority in human knowledge. I shall lay out the bare bones, so to speak, and mention some crucial assumptions that are characteristic of Plotinus’ 214

Alcinous says: first we must examine what follows from our hypothesis, second when we are supposed to give an account of the hypothesis itself, we must postulate other, more primary, hypotheses until we reach a non-hypothetical principle (8ls/5hernp 8.u3) (Didaskalikos 5.6).

THEORIES OF ARGUMENTATION

123

Neo-Platonism. First of all, according to Plotinus, reality is a hierarchical structure which consists of three different layers: the One, the Intellect and the Soul.215 The higher layer is always assumed to be in some sense the cause for the lower ones.216 The One is the most primary, and all being in a sense flows from it. On the level of the One, there are no distinctions whatsoever. It is above the intelligible forms. However, human beings are somehow capable of taking part in it in a highly specialised ‘experience’ which involves losing sense of a distinction between a subject and an object. This experience is also the highest goal for human beings. Such an experience – if we can use that word here – is not essentially connected to the theme of this book, and I shall leave it aside here. The Intellect is the layer below the One. It involves basically two aspects: the intelligible forms or objects that form a structured whole and, on the other hand, an eternal perfect thinker, or knower, grasping those objects in eternal intellectual apprehension. Human beings are also capable of occasionally taking part in this eternal intellectual activity. This is experienced as a kind of intellectual vision which can be compared to the following situation familiar to most scientists and philosophers. We have for a long time tried to solve a particular scientific or philosophical problem. All reasoning, calculations and attempts of a proof have been in vain. The problem remains unsolved. Then, suddenly, we grasp how the elements of the problem fit together and explanatory relations between them are conceived.217 It is important for Plotinus that this immediate grasping is not part of the reasoning process but separate from it. In metaphysical terms it is described as our taking part in the Intellect. As an experience it resembles seeing something complicated, say, a room with furniture and people sitting on chairs in front of tables. Just as seeing the room captures the spatial relationships between things in the room, intellectual ‘vision’ apprehends relationships between intelligible forms.218 This kind of apprehension is the supreme cognitive achievement concerning the intelligible structure imperfectly realised – or copied as Plotinus says – in the perceptible cosmos. According to Plotinus, the cosmos itself is ensouled and the third level of the ontological hierarchy, the Soul, reflects the life-like regular course of natural events in the world. In human beings’ cognitive capacities, discursive reasoning 215

The relevant terms are spelled with capital initials here in order to distinguish them from the more ordinary uses of the same words. 216 For this assumption, see Emilsson (1996). 217 The comparison is from Emilsson (2003). 218 This analogy is also from Emilsson (2003).

124

CHAPTER ONE

(di1lnia), together with perception, belong to the level of Soul. Discursive reasoning is reasoning by concepts; it involves a combination of concepts and transitions from one concept to another and from one proposition to another. Discursive reasoning is, according to Plotinus, always inferior to intellectual ‘vision’ of many forms at a single glance. Concepts, in the Plotinian framework, are imperfect images of the intelligible forms in the human mind.219 All activities involving argumentation belong to the level of discursive reasoning and, therefore, fall short of the highest cognitive achievement, taking part in the Intellect. However, reasoning activities, even though separate from intellectual ‘vision’ do have a role in our attempts to achieve such vision. Sorting out the elements of a problem, trying to find their definitions and inferring on the basis of those definitions – to mention a few examples of discursive reasoning – do function as stepping stones towards an instantaneous grasp of how the problem can be solved or a theory be made to work. But the instantaneous grasp can never be produced by the discursive activities. Plotinus is not particularly interested in logic and he does not seem to present anything that could be called a scientific methodology. However, he does have a tractate on dialectic (Enneads 1.3), albeit a short one. We have to bear in mind that in Plotinus’ case ‘dialectic’ does not mean the technique of argumentation we encountered in Aristotle above. Plotinus’dialectic is closer to Plato’s dialectic in the Republic that involves reasoning concerned with intelligible forms and their interconnections. It turns out that the Ennead concerning dialectic points to a distinction resembling the distinction we have encountered above between the order of human knowledge and the order of things.220 Let us now take a look at the text. The relevant Ennead begins as follows: ‘What art is there, what method or practice, which will take us up there where we must go (g: ,¡p n˚ de‹ /n.esh‚lai 8l1cei, 1.3.1, 1–2)?’ By the method or practice leading us ‘up there’ Plotinus means dialectic. He says: It [the disposition] stops wandering about the world of sense and settles down in the world of intellect, and there it occupies itself, casting off falsehood and feeding the soul in what Plato calls ‘the plain of truth’[see, Phaedrus 248b6], using his method of division to distinguish the Forms (e∆dnp), and to determine the essential nature of each thing (u.w,2lg d£ ja§ eåp r• r4 %qri), and to find the primary kinds (u.w,2lg d£ ja§ %/§ r¡ /.‡ra cel3 ), and weaving together by the intellect all that issues from these primary kinds (ja§ r¡ %j rn6rwl lng.‡p /k2jnsqa), till it has traversed the whole intelligible world (/Øl r• lngr5l); then it resolves again the 219

For this assumption, see Gerson (1999). It has, in fact, been suggested (Leroux 1974) that there is something like an ascending and descending dialectic in Plotinus, but this is denied by A. C. Lloyd (1990, 11). Perhaps Lloyd means that even though something like ascending and descending can be found in Plotinus, he is not talking about a technique of argumentation.

220

THEORIES OF ARGUMENTATION

125

structure of that world into its parts (ja§ 8l1/akil 8lak6nsqa), and comes back to its starting-point (eåp Ω -l %/’8.u¢l (kh¶);221 and then, keeping quiet (for it is quiet in so far as it is present There) it busies itself no more, but contemplates, having arrived at unity. It leaves what is called logical activity (r¢l kecn,2lgl kncij¢l /.ac,are4al), about propositions and syllogisms (/e.§ /.nr1qewp ja§ qskknciq,‡l), to another art, as it might leave knowing how to write. (1.3.4, 9–20; transl. by Armstrong from the Loeb edition.)

Here Plotinus is all the time speaking of a disposition ()mip) which he connects with the adjective ‘dialectical’ (diakejrij3). This disposition is assumed to be able to move in the intelligible realm. First it is said that the disposition distinguishes the Forms there, discovers the essential natures of things and the primary genera (/.‡ra c2lg, which is Plotinus’ term for Plato’s very great kinds, ,2ciqra c2lg). This first phase seems to be like an inventory of the elements of the part of the intelligible realm one is studying. Then the dialectical disposition is said to ‘weave together’ (/k2jw) what consists of these primary kinds. The inquiry is probably concerned with the question of how the elements are interconnected and what kinds of permanent combinations they form. This is followed by a procedure called ‘analysis’, the analysis being assumed to lead us back to the elements from the combinations. In the dialectical procedure the part of the intelligible realm under inquiry is supposed to receive its full analysis into its ingredients and their necessary connections with each other. Finally, the dialectical disposition is ready to contemplate the intelligible objects without weaving things together and analysing any longer. It is puzzling that in the passage quoted above Plotinus clearly assumes that we have a disposition which moves in the intelligible realm. If this dialectical disposition is assumed to belong to the abilities of discursive reasoning, it should not, according to Plotinian assumptions, be able to enter the intelligible at all. One might suggest, however, that Plotinus means to say that we can, by using our reasoning capacities, get in touch with the intelligible in the following qualified sense. In reasoning we use our concepts that are pictures of the real intelligibles. They imitate and resemble the order of the forms in the metaphysical intellect. In virtue of this resemblance, analysis of our concepts gives us some sort of a map of the intelligible. However, in the treatise on dialectic Plotinus talks in the way that dialectical analysis takes place in the intelligible realm.222 221

Cf. Arist. An. Post. II 19, 100a13. It has been a matter of dispute among scholars to what extent we can make it easier to reach the upper intellectual level by any argumentative enterprise. Gerson (1999), for instance, seems to think that getting involved in intellectual acts is not a result of the use of argumentative techniques. The view that I have expressed is also found in Pauliina Remes’s forthcoming monograph on Plotinus’ philosophy of self. 222

126

CHAPTER ONE

Later in his treatise on dialectic Plotinus asks ‘From where does the science [of dialectic] derive its principles?’ and answers: Intellect gives clear principles to any soul which can receive them: and then it combines and interweaves and distinguishes their consequences till it arrives at perfect intelligence (1.3.5, 1–5; Armstrong’s translation).

At first sight this passage gives the impression that in the last analysis our reasoning activities do not have a role in the apprehension of the intelligible realm, but some intelligence outside our mind just hands the intelligible order of reality over to us in a way analogous to divine intervention. However, the passage does not necessarily mean that our dialectical activities would not have a role in the acquisition of intelligible objects. Rather, it is likely that Plotinus assumes that dealing with concepts prepares our soul so that we can ascend to the level of intellectual apprehension proper. The quoted sentence should rather be taken to mean that even though our reasoning activity can make it more likely that we attain an intellectual apprehension into the intelligible, such an apprehension cannot be produced by these activities in a necessary way. Given the general Neo-Platonic framework, it is understandable that Plotinus pays little attention to questions like: What should the structure of a science be like? Or: Are there principles common to all sciences? Plotinus seems to suppose that we should be concerned with metaphysical explanations, and not so much involved in empirical scientific activity.223 For Plotinus, things receive their final explanation when they are seen as originating from the One,224 and when we understand our own place as creatures capable of intellectual cognition within the metaphysical framework.225 In fact, as we have seen, Plotinus’ treatise on dialectic contains very little concerning the details of any reasoning process or an argument structure. It is an important development in Plotinus’ Neo-Platonism that the techniques of argumentation do not receive very much attention.The psychological side starts to dominate, and much more attention is paid to analysing human thinking. 1.3.2 Greek Commentaries on Aristotle I shall now turn to discuss how certain basic themes were treated in the late ancient commentary tradition on Aristotle. I shall not attempt to present an 223

For Plotinus’ use of the term ‘scientific knowledge’ or ‘science’ (%/iqr3,g), see Enneads 5.9.7, 5–6. 224 See Hankinson (1998a, 409–425; 446). 225 For the Plotinian conception of self-knowledge, see Emilsson’s forthcoming study on Plotinus on thought and Remes (forthcoming).

THEORIES OF ARGUMENTATION

127

overall picture of the commentators’ thought, but I am particularly interested in how some of them treated Aristotle’s distinction between what is better known to us and what is better known by nature. I will also pay attention to the discussions concerning the possibility of principles shared by several sciences. Many of the passages to be discussed below have not been translated into English before, and I shall provide a translation of the key passages. Alexander of Aphrodisias The commentator first in chronological order is Alexander of Aphrodisias, who lived in the late second century AD, and died in 211.226 Of Alexander’s commentaries the following are relevant to my purposes here: the commentary on books I-V of the Metaphysics (the commentary on books VI-XIV is generally taken to be by Michael of Ephesus, who lived in the 12th century),227 the commentary on the first book of the Prior Analytics and that on the Topics. Of the lost commentary on the Posterior Analytics we only have quotations by other commentators.228 The most important passages where Aristotle mentions the distinction between what is better known to us and what is better known by nature are the following: An. Post. I 2, 71b33–72a5, An. Pr. II 23, 68b35–37, Top. VI 4, 141b3ff., Phys. I 1, Met. VII 3, 1029b3–12, EN I 4, 1095a30-b5. Of these we only have an extant commentary by Alexander on the Topics. Alexander refers to the beginning of the Physics in his commentary on the Metaphysics (12, 6–14 ad I 2, 982a25) and this short remark is of interest for the present topic too. Alexander follows Aristotle quite closely in distinguishing between proofs in the strict sense and valid syllogisms from true premises. In his commentary on the Topics he says: Not every syllogism from true premises is a proof in the strict sense (js.4wp 8/5deimip), but [only] if the [premises] through which such a syllogism shows [its conclusion] (r¡ diX ‘l … rninflrnp de4jlsrai qskknciq,5p), in addition to being true, are also primary (/.‡ra). In this way they are also reasons (a©ria); a syllogism through reasons (diX aår4wl) is a proof. The primary [things] (r¡ /.‡ra) are reasons for what comes after them (r‡l ,er¡ raflra). The premises (r¡ ka,bal5,ela) have to be not only prior by nature to that which is shown by them, but also primary by being immediate (/.‡ra r+ !,eqa e∆lai) and not needing to be shown (ja§ ,¢ de4mewp de‹qhai). Being as such they will be known on the basis of themselves (%m ^asr‡l) by nature (r· t6qei); the principles (8.ua4 ) of proofs are the ones that are known on the basis of themselves and the immediate [principles] are known on the 226 227 228

For Aristotle’s ancient Greek commentators in general, see Sorabji (1990a). See, e.g., the introduction by Sorabji (1990a) and Sharples in Dooley (1989, 3). Collected by Moraux (1979).

128

CHAPTER ONE basis of themselves, because they are unprovable. (in Top. 16, 1–8 ad I 1, 100a27–29; my translation.)

Here Alexander expresses the assumption that there is an order in which things are known for nature and that there are immediate premises which do not need to be shown (de4jls,i, de‹mip). This is a little strange, because on the basis of what he says it is clear that they can be shown, even from true premises; they just cannot be proved (8/nde4jls,i, 8/5deimip) in the strict Aristotelian sense. However, as we shall soon see, Alexander does allow that the premises of proofs are shown ‘in a secondary way’. What is ‘known by nature’ is not yet compared to the order in which human beings come to know things, but about ten lines later Alexander brings in this comparison too. At first Alexander mentions again the distinction between a proof in the strict sense (jso4wp 8/5deimip) and a certain kind of syllogism, which has true but secondary premises (qskknciq,5p rip… di’ 8kgh‡l ,£l ∫qr2.wl). According to Alexander the premises of the latter type of syllogism are true, accepted and reputable ((ldnmnl). Then he says: The former is a proof in the strict sense (js.4wp ,£l %je‹lnp 8/5deimip), the latter in a secondary way (desr2.wp); it is, as it were, a proof for us, because the premises assumed are better known to us (®p c¡. /.•p y,Øp, %/e§ y,‹l clw.i,Íre.a r¡ ka,bal5,ela) (in Top. 16, 29–30 ad I 1, 100a27–29; my translation).

The distinction between two kinds of priority is also mentioned in Alexander’s commentary on the fourth chapter of book VI of Topics. There he points out that a definition should be a clarification of what each thing is (r• r4 ;l e∆lai, 435, 17 ad VI 4, 141a26) and that the genus and the specific differences are prior by nature and better known simpliciter than the species which is the definiens. The reason for this is that a species can be destroyed without the genus being destroyed, but not the other way round (436, 9–11 ad VI 4, 141b15; cf. Arist. Cat. 12, 14a30–35). Such a criterion of what is prior by nature is applicable to cases where one is searching for definitions in the genus-plus-specific-difference form. It is not equally clear whether a similar condition is satisfied in cases where the inquiry aims at finding an explanation for a phenomenon like the longevity of some animal species (An. Pr. II 23). There the question is whether bilelessness is the cause of longevity or the other way round. In such a case we might well think that the destruction of one destroys the other as well, but I am not sure whether this helps us find out which one is the cause of the other. Alexander does mention induction in connection with unprovable premises. He says: Of all existing things the form is that according to which the thing is; this premise is arrived at by induction (a÷rg c¡. y /.5raqip di’ %/acwc‚p ,£l qsl4qrarai),

THEORIES OF ARGUMENTATION

129

and it is unprovable and primary supposing that all that actually is consists of matter and form (jei,2lns rnfl /Øl r• %le.ce4ˇ æl %m ÷kgp ja§ e©dnsp e∆lai), for it follows from this supposition that all [things] consisting of matter and form are what they are according to their form ()/erai c¡. rn6rns jei,2lns [r+] /alr§ r+ %m ÷kgp ja§ e©dnsp ≈lri r• jar¡ r• e∆dnp e∆lai rnflrn ç %qri) (in Top. 17, 1–6 ad I 1, 100a27–29; my translation).

Here Alexander might be taken to express the view that induction is the way towards unprovable premises in general. However, he does not specify in what sense he talks about induction. There are two relevant possibilities. The first one is that the statement that the form in all cases determines what the thing in question is, is discovered by induction through examining all the cases in which we know what the thing is. The second alternative is that the statement concerning what it is for a thing to be the thing it is, is in all cases established by induction. Only the latter reading would make Alexander’s comment of general methodological interest. The former would be a metaphysical claim. However, it seems that we have no means of excluding the first reading and, therefore, the passage should not be given too much weight, particularly because here Alexander does not connect induction with the distinction between what is better known to us and what is better known by nature. In addition to the passages listed above, the first chapters of the Metaphysics are relevant to Aristotle’s distinction between two directions. On lines 982a25–b10 in I 2 Aristotle uses the distinction to argue for the hierarchical order of the special sciences according to the degree of universality or abstractness of the premises. When commenting on these lines Alexander points to a contrast between what is known by nature and what it is easy for human beings to learn. But things that are in the highest degree universal are also the most difficult of the things that can be known to man because they are completely removed from the senses, … for of the things that are, the first and most simple are farthest from the senses. Aristotle adds ‘to men’ [982a4], to point out a contrast with nature; for by nature first and simple229 things are more knowable than sensible objects. (in Metaph. 11, 8–13 ad I 2, 982a21; transl. Dooley 1989.)

There is a variant reading of this quotation of the penultimate line 11, 12. It is translated here as ‘for by nature first and simple things are more knowable than sensible objects, but in manuscripts LF it is: ‘for nature knows itself far better than it knows sensible objects’. The relevant difference between the 229

The concept of simplicity is also connected to that of primacy in later commentaries; see especially Simplicius’ commentary on the Physics (discussed below pp. 149–151).

130

CHAPTER ONE

two traditions is that the latter contains a clear reference to personified nature knowing itself, whereas the former does not necessarily contain that idea. Dooley, the translator of the quoted passage, points out230 that also his version just quoted might also contain an intended personification of nature. He also refers to Alexander’s commentary on the Prior Analytics and its line (in An. Pr. 3, 20), which could be taken as further evidence of the idea of personified nature in Alexander. If Alexander did entertain the idea of a personified nature knowing herself his references to the distinction between the order of human knowledge and the order of nature should be taken in the following way: we human beings come to know reality in a certain way but nature knows herself in quite the opposite way. However, in his commentary on the Metaphysics (103, 5–104,18 ad I 9, 991a23), when rejecting the view according to which there is a demiurge who uses Platonic ideas as models for the creation of the universe,Alexander specifies that he means by ‘nature’ a formative principle or factor which is in the things themselves. In that connection he does not personify nature. Therefore, it is not necessary to attribute the personification of nature to Alexander. A page after the passage just quoted, Alexander refers to Aristotle’s formulation of the distinction between the two directions made in the Physics. The first chapter of the Physics is often taken to put the distinction in a way which is contradictory to Aristotle’s other formulations.231 The commented line from the Metaphysics is the following: ‘The most exact of the sciences are those that deal most with first principles’ (982a25). Alexander says: We should not suppose that this statement goes counter to what is said in the introduction to the Physics. There, in saying from what point we ought to begin, that is from things knowable by us, Aristotle says: ‘For it is the whole that is more knowable by sense perception, and the universal is a kind of whole’ (184a24). For what is first is not the same as what is universal: at any rate, the first cause imparting motion (a©rinl jilgrij5l) to which he here refers is indeed prior to everything else, but is not universal in the way in which the genera are. But neither did he say, in the text from the Physics, that what is called universal as genus is first in relation to sense perception, but rather (that this latter is) what is more common and an attribute (qs,bebgj5p) of a number of things, as he made clear by his examples. (in Metaph. 12, 6–14 ad I 2, 982a25; transl. Dooley 1989.)

Here Alexander points to the possibility of the formulation of the distinction between two directions in epistemic context in the Physics being contradictory to the others.232 However, he rejects the idea that it should differ from them in a 230

Dooley (1989, 29 n. 53). However, see, e.g., Bolton (1991) discussed in 1.2.2 above, who argues for a reading where this contradiction does not follow. 232 The different interpretations of this formulation of the distinction in the commentary tradition are discussed in de Haas (2002). 231

THEORIES OF ARGUMENTATION

131

serious way and thereby should lead to modifications concerning what Aristotle says elsewhere. Alexander seems to assume that in the passage from the Physics Aristotle uses the term ‘universal’ in a non-technical sense, probably meaning something that is commonly known. Alexander’s expression ‘universal … as first in relation to perception’ might mean something similar to what Aristotle means by the expression /.‡rnl jah5kns, ‘first universal’ in An. Post. II 19, 100a16. This refers to the first familiarity with universals in the temporal sense; the first universal is first in the sense that we learn it earlier. This temporal priority should not be confused with the order of natural priority. In the Metaphysics, by contrast, Aristotle is, according to Alexander, dealing with priority in the natural sense. The quotations above provide us with some evidence that Alexander embraces the Aristotelian distinction between the order of things and the order of human knowledge. Alexander is committed to the idea that there is a certain order in which human beings learn to know reality (in Metaph. 12, 6–14). He also claims that there are two kinds of valid arguments from true premises of which only the kinds of arguments proceeding according to the objective order of priority are to be called proofs in the strict sense (in Top. 16, 1 ad I 1, 100a27–29). In addition, Alexander keeps to the Aristotelian idea of form as an inner principle of things by virtue of which combinations of matter and form are the things they are (in Top. 17, 1–6).233 It is not entirely clear whetherAlexander endorses theAristotelian idea I have called the departmentalisation of sciences above. Departmentalisation in this sense means that the so-called common axioms are not strictly speaking the same in different sciences. Aristotle’s example is that the Euclidian principle ‘if equals are subtracted from equals, the remainders are equal’ is not strictly speaking the same in geometry and arithmetic, because in the former it is concerned with geometrical entities, whereas in the latter it applies to numbers. In the collection of quotations of Alexander’s lost commentary on the Posterior Analytics, no mention concerning the relevant lines 76a37–40 has survived. Neither is the idea explicitly stated in the other fragments of the commentary. Nonetheless, it is clear thatAlexander takes overAristotle’s idea that ‘being’ has several meanings and finds it relevant for the question of scientific premises. When discussing axioms used in several sciences Alexander clearly expresses his commitment to the Aristotelian idea that although neither the proper premises of proofs nor the common axioms can be proved in the proper sense, they can be validly and convincingly argued for on the basis of true premises; they can be proved to us. In the following Alexander uses indirect 233

However, Alexander’s conception of form deviates to some extent from Aristotle’s; see, e.g., Tweedale (1984). I shall not be able to discuss this topic here, however.

132

CHAPTER ONE

argument, setting out to prove a common axiom that ‘things similar to the same thing are also similar to one another’. He says: He Aristotle says that this the principle of non-contradiction234 is the principle of all the axioms – not that it demonstrates them (for the axioms do not even need demonstration, for if [they were demonstrated] they would no longer be axioms or principles) – but because on many occasions we use it to confirm the axioms and [our] conviction (/4qrip) [about them]. For example, to confirm that things similar to the same thing are also similar to one another, we assume (ka,b1ln,el) that, if this is not the case, they will not be similar to one another; but if they are not similar to one another, they could not be similar to the same one thing either, but would rather diverge from it, as they diverge from one another; but it was posited that they were similar [to the same thing]; so the same things would be similar to and not similar to the same thing at the same time. To the extent that we have brought the argument to an evident impossibility, we believe that we have helped to confirm the axiom that says that things which are similar to the same thing are also similar to one another. (in Metaph. 271, 12–21; transl. Madigan 1993.)

Here Alexander treats the general axiom ‘things similar to the same thing are also similar to one another’as an implication (if p then q) ‘if two things are similar to the same thing, they are similar to each other’in which the antecedent (p) states that two things are similar to the same thing and the consequent that they are similar to each other (q). He sets out to prove that this implication is in fact one that would nowadays be called logical consequence. This he wishes to show as follows. First the axiom is assumed together with the additional assumption that the two things are not similar to one another (not-q). From this it is concluded, on the basis of the rule later to be called modus tollens, that they are not similar to the same thing (not-p). But it was assumed that p and therefore the assumption that the antecedent of the implication (p) should be true and the consequent false (not-q) leads to an open contradiction (p and not-p). Alexander’s ‘proof’ of the axiom seems in fact somewhat circular. One might ask why it is more obvious that from not-q it follows not-p – i.e. from the assumption that the two things are not similar to one another it follows that they are not similar to the same thing – than that from p it necessarily follows q – i.e. that if two things are similar to the same thing, they are similar to one another. However, it is clear that Alexander does not intend this as a proper proof of the axiom; such would indeed be impossible, but it is just presented to ‘confirm the axioms and [our] conviction (/4qrip) [about them]’. Above in connection with Aristotle we discussed the question – called ‘the hoary old chestnut’ by Geoffrey Lloyd – whether there is a contradiction or tension between Aristotle’s theory and the practice of science. We discussed chapters I 27–30 of the Prior Analytics and saw that there Aristotle presents a 234

These two additions inside  parentheses are mine.

THEORIES OF ARGUMENTATION

133

syllogistic scheme which, according to him, can be used to find premises for any given conclusion. The important upshot of this discussion is that Aristotle there seems to see no essential methodological difference between the procedure of searching for dialectical premises and looking for scientific premises. Here we shall see that Alexander conceives of the issue rather similarly. In addition, in the commentary of chapter I 2 of the Topics, Alexander presents a view on how dialectical arguments are related to scientific proofs. There are two commentaries on chapters I 27–30 of the Prior Analytics, one by Alexander, another by Philoponus. They are both usually considered authentic.235 Ammonius has also written a commentary on the Prior Analytics, but for some reason he stops right before chapter 27 at the end of chapter 26. There is not indication why this is so. Both Alexander and Philoponus use the following example to illustrate Aristotle’s syllogistic scheme. They have gathered a number of predicates which belong or are universally denied of good and pleasure. Alexander uses the scheme to produce all the possible conclusions involving good as the predicate and pleasure as the subject: all pleasures are good, some pleasures are good, some pleasures are not good, and no pleasures are good. Alexander’s lists the following predicates according to Aristotle’s model: A, good (8cah5l)

E, pleasure (ydnl3)

B (XaA): useful (_teki,5l) preferable (r• aß.er5l), to be striven after (diwjr5l), appropriate (nåje‹nl), profitable (ksqirek2p), agreeable (qs,t2.nl), desirable (¬.ejr5l)

Z (YaE): soft movement (y ke4a j4lgqip), the unhindered actuality of a natural disposition (%l2.ceia r‚p jar¡ r¢l t6qil )mewp 8le,/5diqrnp), untroubled (r• 8lel5ukgrnl), non-violent (r• 8l5.cgrnl), pleasing (r• 8.eqr5l), effortless (r• !/nlnl), painless (r• !ks/nl), without fear (r• !tnbnl), natural (r• jar¡ t6qil), preferable (r• aß.er5l)

G (AaX): happiness (e√dai,nl4a), being perfect (r2keinl), virtues (8.era4 ), action in accordance with virtue (jar’ 8.er¢l %l2.ceia), bodily goods (qw,arij¡ 8cah1), the external [goods] (r¡ %jr5p), the natural [goods] (r¡ jar¡ t6qil)

H (EaY): health (∫ce4a), being lucky (e√rsu4a), having a good progeny (e√rejl4a), action in accordance with virtue (%l2.ceia jar’ 8.er3l), wealth (e√/n.4a)

D (XeA): to be avoided (tesjr5l), hurtful (bkabe.5l), bad (jaj5l), useless (8ksqirek2p), non-profitable (8q6,tn.nl), shameful (aåqu.5l), imperfect (8rek2p)

Q (YeE): sickness (l5qnp), trouble (/5lnp), distress (k6/g), fear (t5bnp), difficulty/poverty (8/n.4a)

(Alexander of Aphrodisias, in An. Pr. 301, 17–32 ad 43b39; my transl.) 235

See, e.g. Sorabji (1990a, 27–28).

134

CHAPTER ONE

In the column for good and that for pleasure only the following three predicates are literally the same: natural (r¡ jar¡ t6qil), desirable (r• aß.er5l) and activity according to virtue (jar’ 8.er¢l %l2.ceia). They would be sufficient to show either that all pleasures are good, or that some pleasure is.236 However, Alexander does not stop with these literally identical terms.237 To establish that every pleasure is good, Alexander does not use the predicate ‘natural’, but goes on to identify some terms from one column with some others from the other column. For instance, he identifies perfect (r2keinl) with unhindered activity of a natural disposition (%l2.ceia r‚p jar¡ r¢l t6qil )mewp 8le,/5diqrnp). ‘Taking these two to be one,’ Alexander says, ‘we make a middle term’ (302, 6). This move is based on a kind of definition of pleasure Aristotle gives in the Nicomachean Ethics book VII chapter 12 (1153a14) according to which pleasure is unhindered activity of a natural disposition. Alexander, however, does not identify unhindered activity of a natural disposition only with pleasure – which seems to be the idea of Aristotle’s ‘definition’ – but also with being perfect.238 Alexander’s argument showing the negative universal conclusion (no pleasure is good) trades on the notion of movement or change (j4lgqip) and its imperfect (8rek2p) character. This discussion also builds on Nicomachean Ethics VII, where Aristotle speaks about pleasure being a process (c2leqip) and a process not being the same in kind as a goal (r2knp). Alexander’s syllogism is as follows. Every pleasure is a soft movement; every movement is

236

For the conclusion that (a) all pleasures are good, ‘natural’ would be a suitable middle term, whereas the latter two (being desirable and activity in accordance with virtue) would show only that some pleasures are good (b and c). (a) Being good belongs to all natural things; being natural belongs to every pleasure. Therefore, being good belongs to every pleasure. (b) Every good thing is desirable; every pleasure is desirable. Therefore, some pleasures are good. (c) All activities according to virtue are good; all activities according to virtue are pleasant. Therefore, some pleasures are good. 237 In fact, he uses literally identical terms only in establishing the particular affirmative conclusion ‘some pleasures are good’ (all activities according to virtue are good, all activities according to virtue are pleasant; therefore, some pleasures are good, Alex. in An. Pr., 302, 18–25) and in a remark on the universal affirmative conclusion (302, 7–13) implying that it can be established by using the ‘natural’ as a middle term. Therefore, he does not employ the two terms ‘natural’ and ‘desirable’ that are identical in the two columns, which is rather surprising. Philoponus does discuss these; cf. below pp. 147–148. 238 For this cf. Aristotle’s definition for being perfect in Met. (V 16, 1021b12– 1022a3).

THEORIES OF ARGUMENTATION

135

imperfect; no good is imperfect. Therefore, no pleasure, since it is a movement or a process, is good. A similar argument appears in Aristotle’s Ethics; it derives from Plato’s Philebus239 and is central in Aristotle’s discussion with a slightly different emphasis. The argument for a negative particular conclusion that some pleasures are not good (304, 19–30) is a little unclear. Alexander selects unprofitable (8ksqirek2p) from D and effortless (r• !/nlnl) from H! This is what he says even though effortless is actually in Z. In any case, he goes on to identify unprofitable with effortless (on 304, 26 we find: 8ksqirek2p, rnflr’ (qril !/nlnl) and argues as follows. No unprofitable, in other words effortless, thing is good. Every effortless thing is pleasant. Therefore, there are some pleasures which are not good. This is a kind of ‘no pain, no gain’ argument. Perhaps there was a proverb of this sort in Greek and Alexander used it here as a basis of his identification of effortless and unprofitable. When presenting the two syllogisms, Alexander does not comment on the quality of their premises at all. He is also silent about the moves he makes to identify terms that are not literally the same. This identification makes the whole scheme much more complicated than it seemed at first. If we are not only mechanically looking for the same terms from the two columns the use of the method requires even more pre-existent knowledge than it seemed at first glance. It is not sufficient that we have gathered all the predicates; we also have to have a grasp of what kinds of identities can be postulated between various terms and on what grounds. Later on Alexander makes quite clear that the scheme is used in science. He uses the Greek %/iqr3,g explicitly. This is a procedure (…d5p) and method (,2hndnp)240 in all the sciences and arts which prove something appropriate by means of syllogisms (%l /1qaip %/iqr3,aip ja§ r2ulaip ra‹p di¡ qskknciq,‡l 8/ndeijlsn6qaip ri r‡l nåje4wl) … the procedure and method is necessary for a philosopher (tik5qntnp), a doctor (åar.5p), an orator (’3rw.), a cultured person (,nsqij5p) and everyone alike who is establishing something through syllogism (qskknciq,5p) (Alexander in An. Pr. 330, 32–331, 1; my translation).

In addition, at the end of his comments on the relevant chapters (331, 22–24), he connects the scheme both with making apodeictic syllogisms in science and with forming dialectical arguments. He says that formation of apodeictic syllogisms is discussed in the Posterior Analytics, whereas making dialectical 239

Philebus 53c–54a expresses the idea that pleasures are processes and as such are different from the goals of the processes. 240 Aristotle uses …d5p in the corresponding passage, not ,2hndnp. The term ,2hndnp, however, appears in Topics I 2, 101b4, where Aristotle explains that dialectic enables us to investigate into the premises of sciences; cf. also Top. I 3, 101b5.

136

CHAPTER ONE

syllogisms is treated in the Topics. Therefore, Alexander does not see any essential difference between the processes of trying to find premises for a given conclusion in these two contexts, even though the requirements for the premises differ (331, 17–24).241 In his comments on the first chapter of the Topics Alexander quite straightforwardly repeats and elaborates Aristotelian ideas about the difference between proofs in the proper sense and dialectical arguments on the one hand and between proofs and non-apodeictic arguments in the scientific context on the other. He also makes a few remarks on the premises of proofs; they show that he is committed to the Aristotelian view of the criteria for the premises of proofs as well. He even repeats some of Aristotle’s examples.242 Alexander’s commentary on the Topics I 2 is of great importance for the question of the relationship between Aristotle’s dialectic and science. Let us now turn to discuss Alexander’s comments on the crucial methodological lines of that chapter. First, Alexander comments on the line 101a26 (27, 7–29, 16), where dialectic243 is said to be useful for training, rhetorical encounters and for the philosophical sciences. Alexander identifies these with logic, ethics, physics, and that which comes after physics, i.e. metaphysics.244 Alexander’s main 241

Alexander distinguishes between (1) syllogisms from true premises, which can be (a) apodeictic with true, appropriate (nåje‹a) and primary (/.‡ra) premises and (b) deictic proceeding from the genus, difference, from a proprium, from definition or from reason (%m aår4ns) and (2) syllogisms from reputable or endoxic premises, i.e. dialectical syllogisms. It is not quite clear to me why the requirement of proceeding from reasons is listed as that for a deictic syllogism. Elsewhere, for instance, in the commentary on the Topics – the passage is quoted above – Alexander presents the idea that it is the apodeictic syllogism that proceeds from reasons. Perhaps he here means a reason for us to accept a conclusion. 242 For Alexander’s comments, see also Moraux (1979, 13–16). 243 In the prooemium (1, 8–19) Alexander points out that the term ‘dialectic’ is used differently by different authors. He says that, on the one hand, the Stoics use is for ‘speaking well’ which has two subspecies: speaking truthfully and speaking appropriately. Plato, on the other hand, uses it for a philosophical method of collection and division. In Alexander’s discussion of Aristotle’s definition of dialectic the main question seems to be how it differs from rhetoric. The difference is taken to be that rhetoric only deals with ethical or political questions, whereas dialectic is completely general. 244 Strangely enough, Alexander does not mention biology; it would involve a rather obvious application of the scheme. For some reason the commentators in general were not that interested in Aristotle’s biology. We only have commentaries by Michael of Ephesus on the Parts of Animals and the Movement of Animals, but they are not particularly helpful. I am grateful to Devin Henry and Monte Johnson for a discussion on this point.

THEORIES OF ARGUMENTATION

137

point is that dialectic is good for philosophy, because it trains us by making us examine the opposing positions. According to Alexander, it teaches us to find out easily on which side the truth lies. In addition, when making us familiar with the convincing (but not true) conceptions and teaching us the nature of the convincing, we learn how not to be misled by convincing (but not true) positions and taking them erroneously to be true. At the end of his comments he says that dialectic teaches us to apprehend (qsln.Øl) how apories are solved. Throughout he emphasises the gymnastic aspect of dialectic. When commenting on the line 101a36 where Aristotle says that dialectic is good for examining the first principles of science, Alexander worries for a long time how this function is related to the previously outlined three applications (training, rhetoric, philosophy). He is finally satisfied with the answer that dialectic as used to examine the first principles of science is subordinated to its use in philosophical science. Alexander repeats Aristotle’s point that the first principles (i.e. the unprovable premises) cannot be proved because nothing is prior to them in the order of nature, and the premises of proofs must be prior to their conclusions in this sense. He also notes that mostly our conviction of the principles comes about through induction (%/acwc3). Finally, Alexander proves to be very helpful. He explains how dialectic can be used in philosophy.245 He uses examples from geometry. According to Alexander, a geometrician posits it as a premise (r4herai) that a surface is that which only has length and breadth; a geometrician also posits that a line is length without breadth and that a point (qg,e‹nl)246 is that which has no parts (n∑ ,2.np n√d2l). However, Alexander says that some people oppose this by saying that no quantity (,2cehnp) can be two-dimensional,247 and a point has even less, only one. Therefore, according to them, a point cannot exist at all; for there is nothing which does not become bigger through addition and smaller through diminution. This argument is said to originate from Zeno the Eleatic. (30, 19–26.) After this Alexander begins a lengthy exposition (30, 26–31, 21) concerning how the existence of a point can be established dialectically. The main 245

His first example is Aristotle’s argument in the Physics that shows that no body can be infinite (!/ei.nl) on the basis of, according to Alexander, an endoxic premise that every body is limited or defined by a plain (30, 12–17) to which Aristotle, according to Alexander, added that ‘no infinite thing can be limited’ (n√d£l d£ ®.iq,2lnl !/ei.nl). Alexander continues that this makes clear that it is a dialecticians job to judge or argue (k2ceil) about the principles. 246 Alexander also uses the more familiar qric,3 interchangeably with qg,e‹nl. 247 I take the Greek di1qrg,a and its derivatives here to refer to the dimensions; this becomes clearer when I proceed.

138

CHAPTER ONE

point of the argument is that a limit is never the same as that whose limit it is. ‘Limit’ is here used in a rather abstract sense as referring to an entity appearing in the definition of another object; for instance, a line is a limit of a triangle, because a line is needed in the definition of a triangle. The limit cannot be the same as that whose limit it is, because a limit has one dimension less than that whose limit it is. The sequence Alexander has in mind is the following: body → surface → line → point. It is assumed that such defining by limits cannot go on forever but some notions are basic. A point is here considered such a basic notion. Because, Alexander’s argument goes, lines exist and points are limits of lines, points must exist. It is not completely clear to me why Alexander takes this argument to be dialectical. Earlier when commenting on the line 101a26 (27, 7–29, 16) he has underlined the difference between dialectical and scientific arguments. The difference, according to Alexander, is that the premises of a dialectical argument are convincing but not true. Therefore, the fact that he identifies Aristotle’s premise here as a reputable conception ((ldnmnl), seems to indicate that he wants to say that it is not true. However, he does not say why we should take it to be convincing but false. The relevant aspect of Alexander’s discussion of the Topics is that when discussing the argument for the existence of a point he gives an example how dialectic is used to establish the premises of a science. His discussion shows that Alexander takes it for granted that dialectic can be used to establish the premises to someone who does not believe that they are true. In sum, Alexander takes dialectic, on the one hand, to train us to distinguish between truth and falsity and, on the other hand, to provide us with means to establish the truth of the premises to someone who does not yet grasp that they are true. Therefore, it seems that for Alexander dialectic is not a method of inquiry; we are just trained to ‘see’ the solutions or distinguish truth from falsity. Themistius We shall next discuss Themistius (AD c. 317–c. 388), who wrote paraphrases of Aristotle’s works, probably for pedagogic purposes. Among his paraphrases is one on the Posterior Analytics. He also has a paraphrase of the Physics, and we shall also take a look at it below. In his paraphrase of the Posterior Analytics I 2 Themistius discusses at some length Aristotle’s distinction between two types of priority, i.e. priority by nature and priority for human beings. He expresses the distinction as follows: If the premises are primary (/.‡rai) and prior (/.5re.ai) in this way, it is clear that the priority here means (r• /.5re.nl -l (uniel) not to us (r• /.•p y,Øp), but

THEORIES OF ARGUMENTATION

139

to nature (r• /.•p r¢l t6qil) … for to us the perceptibles (r¡ aåqhgr1) and the particulars (jah’ )jaqra) are prior, whereas by nature it is the universals (r¡ jah5kns) and such [objects] that are further from perception (/5..w r‚p aåqh3qewp), [that are prior] (in An. Post. 6, 14–16 ad I 2, 71b32–72a8; my translation).

Here Themistius does not contribute personally or make explanatory additions; his presentation is a rather straightforward paraphrase. Priority to us and by nature is distinguished on the basis of whether the object we are talking about is perceptible or not. If it is, then it is prior to us; if it is not, it is taken to be universal and classified as being prior to nature. This does not tell us much about what it means to be universal and prior in the order of nature. In his paraphrase of the first chapter of the Physics Themistius makes explicit the connection between priority for us and the order in which we come to know things. The priority is of two kinds (r• /.5re.nl nœl diu‡p): for us and for nature. For us prior is that which is better known to us and easier for us to grasp (Ω ’∏nl -l jarak1bni,el), like the names and syllables of letters; for nature prior [the objects] are by essence simpler (r¡ jar¡ r¢l n√q4al 9/kn6qre.a), like the letters of names, and these [directions] are opposite (8l1/akil d£ y /n.e4a). We analyse [starting] from the composites [and proceeding] towards the simpler (8/• r‡l qslh2rwl %/§ r¡ 9/kn6qre.a) [objects] that are by nature prior; nature in turn produces (,gualØrai) the composites from the simpler ones. (in Phys. 1, 14–20 ad I 1; my translation.)

Here Themistius refers to the idea that what is prior for us is more familiar to us and also somehow composite, whereas what is prior by nature is simpler. In addition, he says, not only that what is simpler is universal or not graspable in perception, but that nature produces (,gualØrai) the composite objects from simpler things. There seem to be three relevant possibilities of understanding this expression. The first is that Themistius refers to the Aristotelian idea that all things in the sublunary world are composites of form and matter and they are in a sense generated (and hence ‘produced’) from these metaphysical principles. The second possibility refers to another Aristotelian idea according to which all the definitions are composed of the genus term and the specific differences. If we take Themistius to refer to this assumption, then his explanation of the two types of priority would be close to what we find in Alexander’s commentary on the Topics, where he points out that a definition should be a clarification of what each thing is (r• r4 ;l e∆lai, 435, 17 ad VI 4, 141a26) and that the genus and the specific differences are prior by nature. If the genus is destroyed, the species will be destroyed as well, but not the other way round. The third possibility would be a Platonist reading containing a reference to a demiurge that produces things from general metaphysical

140

CHAPTER ONE

principles such as the great kinds. The context does not allow us to decide which meaning is at stake. Themistius also repeats the Aristotelian view according to which there are two types of proofs distinguished on the basis of whether the premises are prior by nature or prior for us. He says: When we prove something about the composites (/e.§ r‡l qscjei,2lwl) we will prove on the basis of what is prior by nature (%j r‡l t6qei /.nr2.wl a√r• 8/nde4mn,el), whereas when we prove something about the starting points (/e.§ r‡l 8.u‡l), [we prove] on the basis of what is prior for us (%j r‡l /.•p y,Øp /.nr2.wl). The former is a proof in the proper sense (y js.i‡p 8/5deimip); the latter, although not [a proof] in the proper sense, is [a proof] to the degree that is sufficient for us (y,‹l ßjal‡p). (in Phys. 1, 20–2, 3 ad I 1; my translation.)

I have paid attention to the fact that Aristotle’s formulation of the distinction between priority for us and by nature at the beginning of the Physics appears to be contradictory to his other formulations. Themistius neither in his paraphrase of the Physics nor of the Posterior Analytics makes any direct comments on the difference between Aristotle’s formulations. However, it is likely that he assumes that there is no contradiction involved (see, in Phys. 2, 5–20). What Themistius says on the distinction in the first chapter of the Physics can be used to clarify how he takes the distinction there. He assumes that at first our concepts are somewhat blurred but they become clearer when we grow up. Themistius’ idea is that in the relevant passage Aristotle is talking about the difference between a still undistinguished general concept and a thought out definition. The former is better known to us, because that is the one we come to know first; the latter is better known for nature (see, in Phys. 2, 20–25).248 Therefore, from Themistius’ perspective the first chapter of the Physics does not contradict the other formulations of that Aristotelian distinction because it refers to a process where our notions are developed from vague pre-theoretical notions into worked out scientific definitions. Themistius uses the distinction between proofs for us and proofs for nature in his paraphrase of the Posterior Analytics I 3 to argue for the view that proofs, if conceived in the Aristotelian manner, are not circular: [A] proof takes as [its] premises always [those that are] prior in another way (™ ce 8/5deimip 8e§ r•l )re.nl r.5/nl r‡l /.nr2.wl /a.aka,b1lei); if [an argument proceeds] from [premises] prior for us, it could also be called a proof (8/5deimip), 248

However, when presenting an example, he talks about material parts, not parts of the definition, of a man as prior by nature (ibid. 2, 7–11). It is a problem also for Aristotle to what extent the parts of the composite particular are also parts of the form (see Met. VII 10–11).

THEORIES OF ARGUMENTATION

141

but not [a proof] simpliciter (9/k‡p), but only for us (/.•p y,Øp) (in An. Post. 9, 21–23 ad I 3, 72b15–73a20; my translation).

This passage expresses in a concise form the typical Aristotelian assumption that the unprovable and naturally prior premises of scientific proofs can in a sense be proved: they can appear as conclusions of valid arguments from true premises. However, according to Aristotle – and this conception Themistius reflects here – these arguments are not proofs in the strict sense because they proceed in an order opposite to the order of things. Demonstrations establishing the premises of scientific proofs start from what is better known to us, but secondary in the order of natural priority. Only arguments proceeding from premises ‘better known for nature’, i.e. prior in the order of things, are classified as proofs in the full Aristotelian sense. Themistius takes quite seriously the Aristotelian view according to which general axioms used in different sciences are not strictly speaking the same.249 This can be seen on the basis of Themistius’ account of Bryson’s alleged quadration of the circle. According to Themistius (in An. Post. 19, 12–17), Bryson claimed to quadrate the circle, i.e. to construct a square in size identical to a circle, as follows. A circle is always greater than a polygonal figure (/nk6cwlnp) drawn inside it and it is always smaller than a polygonal figure drawn around it. If we draw a polygonal figure in between the one drawn inside the circle and the one drawn around it, we get a polygonal figure that is equal to the circle; this conclusion is based on the general axiom ‘if some things are greater and smaller than the same things, they are equal to each other’.250 At first sight Bryson’s quadration seems, at least on the basis of Themistius’ account, rather strange. Why should the fact that there are polygonals smaller and bigger than a circle make some of them equal to the circle? A possible way of reconstructing the attempt is as follows. Themistius’ account of Bryson’s enterprise seems to tacitly assume that the polygonal figure drawn inside the circle is that much smaller than the circle than the polygonal figure drawn outside of the circle is bigger than it. Therefore the polygonal figure drawn midway between these two polygonal figures is shown to be equal in size to the circle. The general axiom behind this would 249

When commenting on the line (in An. Post. 18, 31–19, 2 ad I 7, 75b1–20) where Aristotle expresses the idea of departmentalisation, Themistius simply repeats Aristotle; cf. in An. Post. 24, 26–28 ad I 11. 250 Philoponus in his commentary on the Posterior Analytics also refers to Bryson’s enterprise. The two versions differ in the respect that Themistius talks about polygonals, Philoponus about rectilinear figures; see below in this chapter the section concerning Philoponus.

142

CHAPTER ONE

then mean on the one hand that if there are two things (a and b) that are equally smaller than a given thing (c a
cb), they are equal to each other (a
b) and, on the other hand that if two things are equally greater to another given thing (a d
bd), the two things are equal (a
b).251 Rather than pointing to the falsity of Bryson’s premises, Themistius’ criticism is aimed at the common character of the axiom Bryson uses. Themistius says that such axioms, like the one used by Bryson, do not apply only to spatial magnitudes but also to numbers and times and many other kinds of things. (in An. Post. 19, 7–10.) Therefore, what is wrong in Bryson’s enterprise to quadrate a circle is not a false assumption that he makes (according to which the polygonal figure drawn inside the circle is that much smaller than the circle than the polygonal figure drawn outside of the circle is bigger than it). The problem is that the axiom behind this assumption (ca
cb
 a
b and ad
bd
 a
b) is too general. However, even though the example contains much confusion, it provides us with evidence that Themistius takes quite seriously the Aristotelian assumption which I have called departmentalisation of the sciences. Philoponus We shall now move on to Philoponus who lived in Alexandria c. 490–570. Two commentaries bearing Philoponus’ name are relevant here. One is a commentary on the Physics, the other is a commentary on the Posterior Analytics I. Both these commentaries are considered authentic.252 Philoponus takes Aristotle’s formulation of the distinction between the two senses of priority at the beginning of the Physics to be of particular interest.253 He devotes ten CAG254 pages to the relevant line 184a16. At the end of his comments Philoponus mentions that the formulation might appear to be contrary to that presented in the Analytics. However, according to Philoponus there is no contradiction between the two formulations. He says: [i]how is it that [Aristotle] here says that the universals (r¡ jah5kns) are better known in perception and for this reason we should begin from the universals and

251

Proclus claims that the axiom is false, but he sees the axiom in a slightly different way. Philoponus follows Proclus, see below. 252 See Sorabji (1990a, 28). The commentary of book II of the Posterior Analytics attributed to Philoponus will be discussed below in 2.3.2. That is not usually considered authentic. 253 For late ancient interpretations of the first chapter of the Physics, cf. also de Haas (2002). 254 Commentaria in Aristotelem Graeca; see the list of abbreviations for more information about the volumes.

THEORIES OF ARGUMENTATION

143

proceed towards the particulars (r¡ jah’ )jaqra), but in the Apodeictic he says that the particulars are nearer to perception, the universals further from perception; and [ii] how is it that in the Apodeictic he says that the universals are primary by nature (/.‡r1 tgqil e∆lai r· t6qei r¡ jah5kns), but here he says that the particulars are primary by nature (in Phys. 17, 15–20 ad I 1, 184a16; my translation).

Philoponus explains the situation as follows: It is true that the particulars are better known in perception (because perception does not grasp the universals at all (çkwp c¡. rnfl jah5kns y a©qhgqip n√j 8lrika,b1lerai)); but because all nature proceeds from imperfect to the perfect (/Øqa t6qip %j rnfl 8reknflp %/§ r• r2keinl /.5eiqi), perception is confronted (%/ib1kkei) with the particulars in a mixed (qscjeus,2lwp) and undistinguished manner (8dia.h.Írwp) unable to distinguish the peculiar characteristics from the others straight away (diaj.4leil e√hºp rnºp ådijnºp rnfl jah’ )jaqra ua.ajr‚.ap 8/• r‡l kni/‡l n√j %miqu6nsqa). For this reason he named this kind of knowledge based on perception [knowledge] of a universal (çhel di’ a√r• rnflrn rnfl jah5kns r¢l rnia6rgl r‚p aåqh3qewp cl‡qil /.nqgc5.esqel), whereas in the Apodeictic he takes the universals in a strict sense. Therefore [the formulations] are compatible, not contrary to each other. (in Phys. 17, 27–18, 4; my translation.)

According to Philoponus, the problem is two-fold: the first problematic question is related to the order of what is better known to us and what is better known by nature, and the second one to what is prior by nature. The first problem is whether Aristotle characterises the order in which we come to know things in the Physics in a way different from the formulation of this order in the Posterior Analytics. The second problem, in turn, pertains to whether Aristotle defines priority in existence differently in the two places. Philoponus observes that in the Physics Aristotle seems to be using the expression ‘universals’(r¡ jah5kns) in a somewhat loose sense.255 Philoponus says that in the loose sense the meaning of the term is as follows: ‘that which is a part (r• ,e.ij5l) and poured together (qscjeus,2lnp) because of fitting to several [things] (di¡ r• /ke4nqil %ta.,5feil)’ (in Phys. 17, 25–27).256 Philoponus says r• ,e.ij5l is a ‘particular as indefinite’ (®p 85.iqrnl jah’ )jaqra) (in Phys. 13, 3) such as an animal as yet unspecified (ri f+nl) or an indefinite man (rip !lh.w/np) (ibid. 11, 7–8). His solution to the first problem, i.e. that the order of what is better known to us seems on the basis of the characterisation in the first chapter of the Physics to be contrary to what is said in the Analytics, is the following. According to Philoponus, Aristotle in the Physics is by ‘universal’ (jah5kns) referring to our first acquaintance with the universal, when we start to see 255

Cf. Alexander and Themistius above. The expression r• ,e.ij5l also appears in Simplicius (in Phys. 17, 17) in the phrase r• ,e.ij•l ©dinl. 256

144

CHAPTER ONE

instances of animal species as animals of a kind. In inquiry we start from this kind of unarticulated familiarity with universals and aim at specific definitions of natural species. The definitions of particular species are, then, characterised as being better known by nature than our first grasp of universals. Philoponus claims that in the Analytics Aristotle points to the fact that the particular instances of natural species and genera are the ones we encounter first in perception.257 In the Physics Aristotle is contrasting an initial grasp of the universal (this is a kind of animal) to a full scientific definition of the same species. To solve the second part of the problem, i.e. that related to the priority in existence, Philoponus says (in Phys. 18, 5–19, 2) that one has to distinguish between two things. On the one hand there is the whole of properties, which, e.g., a particular animal necessarily is, and this is different from the cause or reason why it forms such a whole (see in Phys. 18, 16–17). He says that in the Physics, when Aristotle uses the term ‘universal’, he refers to the whole of properties which we at first know in an indeterminate way, and this is why it is prior for us. This is to be contrasted with the Analytics where ‘universal’ refers to the reason or cause why the instances of a certain species share a common nature (r¢l aår4al … r‡l jah5kns t6qil, 19, 1–2).Therefore, the two works should not be considered as contradictory: the relevant term is used in two different ways. According to Philoponus, an alternative solution to the problem has also been suggested. It is not impossible that he is referring to the solution found in Simplicius (see below in this chapter). Philoponus rejects the other suggested solution (see in Phys. 11, 24–13, 4). In that solution, according to Philoponus, it is assumed that we come to know the genus at first in perception. Philoponus claims that this means that we come to know a collection of the particular species that fall under the genus, because genera are no more than such collections, i.e. they do not exist in their own right. But, Philoponus says, if we perceive a man coming we do not take him to be such a collection (ibid. 12, 5–10). It is likely that Philoponus reads the alternative suggestion in a misleading way because he wants to show the superiority of his own 257

A factor that increases the confusion related to Aristotle’s formulations of the distinction between two types of priority within the epistemic context is that the expression ‘particular’, r• jahX )jaqrnl, is also ambiguous inAristotle. Barnes, in his commentary on the Posterior Analytics (1975), talks about systematic ambiguity. The term can mean particulars in the sense of singular instances as contrasted to universals, but it can also mean particular species as contrasted to, e.g., genera. The English word ‘particular’is ambiguous in a similar way. It can be used of a particular instance (i.e. a singular substance) or of a particular species. Philoponus, however, does not pay attention to the ambiguity of ‘particular’.

THEORIES OF ARGUMENTATION

145

solution. It seems that the alternative solution expresses the idea that in perception our first acquaintance with the universal is of the genus. That claim cannot be destroyed as easily. Sometimes Philoponus seems to understand Aristotle’s phrase ‘better known by nature’ quite literally in the sense that nature actually knows or makes things. He says: ‘each nature knows at first that of which it is also the cause’ (^j1qrg t6qip rnflrn cilÍqjei /.‡rnl n∑/e. ja4 %qril aår4a, in Phys. 18, 22–23) and ‘that which nature makes, it also knows to the greatest extent, and it makes the particulars (in Phys. 17, 23–24). One might ask whether in general he assumes that the distinction, according to Philoponus, presupposes a personified nature knowing itself. On the basis of the latter quotation it would seem that personified nature would have something like a producer’s knowledge of the various forms which things can take in the course of nature. We cannot deal with this issue more deeply, however. In the commentary on book I of the Posterior Analytics Philoponus clearly endorses the Aristotelian distinction between what is better known by nature and what is better known to human beings together with this distinction’s relevance to the notion of proof. He says that the syllogisms that are proofs (8/ndeijrij•p qskknciq,5p) have to start from what is prior to nature (%j r‡l r· t6qei /.Írwl) and that strictly speaking (js.4wp) prior, i.e. prior to nature (r· t6qei), are the universals. Philoponus also repeats Aristotle in saying that the two orders of priority are opposite to each other. He says explicitly that what is prior for human beings is what we at first come to know (r¡ y,‹l /.Írwp cilwqj5,ela) by using the perceptive faculties. (See in An. Post. 29, 1–14 ad I 2, 71b33.) Let us now turn to the question of whether Philoponus endorses the idea of departmentalisation of the sciences and claims that common axioms are not literally the same in all sciences. Philoponus in fact says that the common axioms are homonymous in different sciences (see in An. Post. 122, 26–123, 14). However, it seems improbable that he would mean that they are homonymous in the sense that in different sciences they only have a common name, but nothing else in common. Philoponus uses the same example as Themistius, namely Bryson and his attempted quadration of a circle. He presents it in a slightly different way, however. According to Philoponus, Bryson tried to show that there is a rectilinear figure equal to a circle, whereas Themistius talked about polygonals. Like Themistius, Philoponus also presents the attempt in connection with the criterion that the principles of Aristotelian proofs have to be appropriate for the subjects discussed (in An. Post. 111, 20–31). According to the reading, which Philoponus attributes toAlexander, the quadration is carried out as follows. One should draw two rectilinear figures, one

146

CHAPTER ONE

around the circle and one inside it, the former being as small as possible, the latter as large as possible.Then a rectilinear figure is drawn half way between these two figures. It is claimed that the last figure drawn is equal in size to the circle. Again the quadration works only if it is assumed – falsely – that the figure drawn outside the circle differs as much in size from the circle than the one drawn inside it. On the basis of this assumption and the axiom that if two things are equally smaller than the same thing or equally bigger than the same thing they are equal to each other (c a
cb
 a
b), the conclusion is assumed to follow. Philoponus, however, goes on to say that Proclus has observed that the common axiom used by Bryson is false (ibid. 112, 8–14). Proclus’ counterexample, according to Philoponus, is as follows. Both 9 and 10 are greater than 8 but smaller than 12. However, 9 and 10 are not equal to each other. Therefore, from the common axiom that if two quantities are greater and smaller than the same quantities, it does not follow that they are equal to each other. It can be observed that Proclus understands the axiom in the way that it should apply to quantities that are to some extent – but not necessarily equally – smaller or larger than the same thing. If understood in this way the axiom is clearly false. However, Philoponus’ presentation of Bryson’s attempt points quite clearly to the way that Bryson understood the axiom in the sense that if two things are equally larger (or smaller) than the same thing, they are equal to each other. Philoponus is critical of Proclus’ counterexample (in An. Post. 112, 25–36). Strangely enough, he does not present the qualification just mentioned, which would make the axiom true. Instead the point of Philoponus’ complaint is that Proclus’counterexample is mistaken because it uses numbers, whereas Bryson’s attempt concerns spatial magnitudes (ibid. 113, 1–4). Philoponus constructs a counterexample of his own in terms of geometrical magnitudes in order to show the falsity of the axiom, which means that his own interpretation of the axiom is the same as Proclus’ (see ibid. 113, 4–114, 17.) Unfortunately, Philoponus does not distinguish between the false assumption that the rectilinear figure drawn inside the circle differs as much in size from the circle as a similar figure drawn outside it and the true axiom (ca
cb
 a
b). In any case, his criticism towards Proclus shows commitment to the Aristotelian assumption of the equivocity of the common axioms. According to Philoponus, even counterexamples have to be formulated by using quantities of the same kind. We can now leave the question of departmentalisation behind and take a brief look at whether Philoponus, in his comments concerning Aristotle’s general ‘method’ of finding syllogisms for all purposes in the Prior Analytics I 27–30, recognises any tension between Aristotle’s theory and practice of science. Philoponus presents a highly similar list of predicates for pleasure and good as Alexander (in An. Pr. p. 274). There are, however, some changes and terminological differences.

THEORIES OF ARGUMENTATION

147

A, good (8cah5l)

E, pleasure (ydnl3 )

B (XaA): useful (_teki,5l), preferable (aß.er5l), to be striven after (diwjr5l), appropriate (nåje‹nl), profitable (ksqirek2p), agreeable (qs,t2.nl), to be pursued (%ter5l)

Z (YaE): movement (j4lgqip), activity according to nature (%l2.ceia jar¡ t6qil), unhindered life (8le,/5diqrnp fw3 ), untroubled (r• 8lel5ukgrnl), preferable (r• aß.er5l), naturally desirable (r• jar¡ t6qil ¬.ejr5l)

G (AaX): happiness (e√dai,nl4a), being perfect (r2keinl), virtuous life (jar’ 8.er¢l b4np), natural (jar¡ t6qil), good disposition (e√em4a), that for the sake of which (n∑ )leja)

H (EaY): health (∫ce4a), having a good reputation (e√dnm4a), having a good progeny (e√rejl4a), virtuous life (jar’ 8.er¢l b4np), wealth (e√/n.4a), painlessness (8/nl4a), freedom from sorrow (8te.e/nl4a)

D (XeA): to be avoided (tesjr5l), hurtful (bkabe.5l), bad (jaj5l), inappropriate (8kk5r.inl), non-profitable (8ksqirek2p), damaging (fg,i‡dgp), imperfect (8rek2p)

Q (YeE): sickness (l5qnp), trouble (/5lnp), fear (t5bnp), difficulty/poverty (8/n.4a), unnatural movement (/a.¡ t6qil j4lgqip)

(Philoponus in An. Pr. 274 ad 43b1; my transl.)

Philoponus puts these predicates into a star-shaped map illustrating how the syllogisms are constructed. One key assumption is left tacit in the picture: in order for us to make syllogisms involving predicates from the groups distinguished here, the predicates we pick from the two columns have to be the same. In Philoponus’ example this condition is fulfilled in only one case: ‘virtuous life’ is found in both G and H, and these would entitle us to conclude a particular affirmative conclusion (some pleasures are good). ‘Natural’is found in both G and Z, but not in exactly the same form. Philoponus, however, considers them to be the same. We can contrast Philoponus with Alexander, who did not say anything about the quality or truth of the premises, whereas Philoponus makes a comment on this issue. He says (276, 20–29) that it is true that movement (j4lgqip) and being desirable (¬.ejr5l) follow pleasure, i.e. that every pleasure is a movement and that every pleasure is desirable. However, according to Philoponus, it is only plausible (jar¡ d5mal) but not true that every pleasure would be natural, because, for instance, scratching an itch is not natural. It is highly likely that Philoponus means that scratching an itch even though being pleasant is not naturally desirable: people do not want to have an itch just for the pleasure of scratching it.258 Anyway, Philoponus points out (276, 20–28) 258

Philoponus is probably thinking of Callicles in Plato’s Gorgias, who takes hedonism up to the point where a pleasure derived from scratching an itch can in itself be desirable and make life happy (cf. Gorgias 494c–d).

148

CHAPTER ONE

that if we showed by a syllogism with the middle term ‘natural’ that every pleasure is good, this syllogism would not have true premises. When commenting on line 46a3 of the Prior Analytics I 30, where Aristotle refers to the applications of the syllogistic scheme, Philoponus – in contrast with Alexander – does not use the word %/iqr3,g explicitly. He says that logic (kncij3) is a proper tool for philosophy. However, Philoponus also says that a similar procedure is used in ‘vulgar’ (b1lasqnp) fields of study, too. In the vulgar context, he says, inferences are made on the basis of common notions (jnila§ (llniai), but people making them are ignorant of the method (,2hndnp). (305, 12–21.) Philoponus probably means that all arts (he uses r2ulg on 305, 16–18) presuppose some kind of inference capacity, which in fact functions similarly to the syllogistic scheme of the Prior Analytics. However, many people working in vulgar fields do not realise that they are using inference schemes of some sort and, therefore, they are in a way ignorant of the method but are implicitly using something similar. It is not completely clear what Philoponus means by vulgar fields of study. On the basis of how he explains the difference it seems that all such arts or even more theoretical studies that do not make their inference patterns explicit count as vulgar. However, even though the reference to science is not explicit in this context, Philoponus’ rather keen interest in the quality of the premises as well as his sharp distinction between what is only endoxic and what is true, indicates that Philoponus assumes that the syllogistic scheme can be used both for scientific and for dialectical purposes. Throughout the commentary on chapters 27–30 of the Prior Analytics I, Philoponus uses distinctions which point to the scientific context. For instance, in 276, 10–16 he distinguishes between predicates belonging to essence (r¡ n√qiÍdg) and those somehow supervening on or following from it (r¡ %/nsqiÍdg). I think it is reasonable to suppose that essential predications are needed in scientific proofs, whereas other kinds of predicates are useful for argumentation with less strict conditions. Philoponus also repeatedly uses Greek words for proving (8/nde4jls,i and its derivatives; see, e.g., 276, 31–32; 273, 10). In addition to the distinction between truth and reputability, we also have some other references to the idea that a similar scheme can be employed in dialectic. When introducing the general outline of the scheme, Philoponus repeats Aristotle in saying that the predicates which belong to the middle ground between the most general and the most particular predicates are the most important objects of inquiry. There his example is whether soul is immortal or not (273, 10). That question is just of the sort which, according to Aristotle, is suitable for a dialectical problem (Topics, I 11, 104b1–17, I 10, 104a5–7 and I 14, 105b19–29). It is a point on which the majority and the wise disagree with each other and among themselves.

THEORIES OF ARGUMENTATION

149

Simplicius I shall now make a few remarks on Simplicius, since his discussion is directly related to that of Philoponus.259 We do not have his commentary of the Posterior Analytics but we do have one on the Physics. Like Philoponus, Simplicius also comments quite extensively (6 CAG pages) on line 184a16 (in Physics I 1) concerning the distinction between two directions in epistemic context. When commenting on that line, Simplicius is exemplifying the late Platonist tendency to present the Aristotelian and Platonic views as compatible with each other. He repeats the Aristotelian distinction between two kinds of epistemic priority and illustrates it with Platonic examples. Let us consider the following passage. The better known [premises] are taken either [i] as principles (8.ua4) and reasons (a©ria) of that, which is shown, as happens in proofs in the strict sense (js.4wp) … as when [one proves] the cosmos to be good (jak5p) on the basis of the demiurge’s being good (8cah5p) … or [ii] as following necessarily from that, which is proved (®p 8jnknshnfllra %m 8l1cjgp rn‹p 8/ndeijls,2lnip), and concluding them [i.e. the better known premises] on the basis of that (di¡ rnflrn qsleiq1cnlra a√r1). The better known [premises] are assumed (/a.aka,b1lerai) this latter way, as when [one shows] god to be good (8cah5p) on the basis of the cosmos being good (jak5p) and ordered (rerac,2lnp), because this is more at our disposal in perception (/.nuei.nr2.ns rn6rns jar’ a©qhgqil y,‹l ≈lrnp); … the latter [ii] is more like a syllogism from sign (rej,g.iÍdgp ,Økknl), not a proof (8/ndeijrij5p).And [the premises] assumed for this kind of conviction are not principles (8.ua4) of that which is proved, for they follow ()/nlrai) them rather than guide (/.ngcnfllrai) them,260 but [the premises of] the former kind of proof [i] are principles, because they are better known and clearer (/.ntal2qre.a) and on the basis of them comes the conviction of that which is proved (8/’ a√r‡l y /4qrip c4lerai rnfl 8/ndeijls,2lns). (in Phys. 15, 15–29 ad I 1, 184a16; my translation.)

This quotation suggests that at least in some sense Simplicius endorses the idea that we can refer to a kind of objective priority within an epistemic context: reasons can be described better known than the facts that they are reasons for, even though they are by no means self-evident. Rather, syllogisms based on facts that follow from those reasons can be used to establish the reasons. Simplicius connects this idea with the definition of proofs in the proper sense. In this example he says that the fact that the cosmos is ordered is more 259 The details of Simplicius’ life and career have been the object of research in recent years (see, e.g., Hadot 1990 and Sorabji’s introduction 2004). In any case, he studied in Alexandria at some point between 475 and 526 but apparently wrote most of his commentaries after Justinianus had closed the schools in 529. 260 The idea appearing in Simplicius’ comments that the real principles are in some sense necessary following from that which is proved are still in an objective way prior to the conclusions, is also found in Alcinous (Didaskalikos 5.5) discussed above.

150

CHAPTER ONE

familiar to us in perception and that on the basis of this the goodness of the demiurge can be shown in a syllogism, which is not a proof but a syllogism from a sign (rej,g.iÍdgp).261 A real proof, according to Simplicius, proceeds in the opposite direction: it proves the cosmos to be good, taking as its premise that the demiurge is good. Because the goodness of the demiurge is, according to Simplicius the real reason for the cosmos being good and, therefore, only the latter counts as a proof. A similar remark concerning the explanatory order also appears in Aristotle (Met. XII 10, 1075a13–15). He talks about the good – which can be understood as divine262 – and its relation to the order in the world. His analogy indicates that there is order in the world because its organising principle, divine good, is good. It is not that the good as a divine principle is good because of the world-order but the other way round. We can also compare Simplicius’ example to that of Aristotle concerning the non-twinkling of the planets and the shape of the moon (An. Post. I 13, 78a31–b11). All these cases involve syllogisms where the terms are the same but in the opposite order, and only one of them is recognised as a real proof. The explanation for the one being a proof and the other not is the same in Simplicius as in Aristotle. Only such arguments that proceed according to relations of priority in nature are real proofs. Their premises can be established to us human beings from facts initially known to us, but these arguments do not count as proofs proper because they proceed in the wrong direction. What follows a little later gives us further evidence that Simplicius takes the basic idea of the distinction to be that what human beings at first come to know and what is easy to attain in perception is to be contrasted with what is really primary (see in Phys. 16,7–20). Simplicius takes Aristotle’s formulation of the distinction between two directions in the Physics as one which presents Aristotle’s considered view on the matter. He says that the principles that one is looking for in science should be conceived of as elements. In addition, the principles are simple (9/kØ), 261

Philoponus, too, uses the adjective rej,g.i7dgp for non-apodeictic syllogisms, which use premises better known to us in order to show real principles; he also calls such non-apodeictic syllogisms didactic (didaqjakij5p; in Phys. 9, 12–17). Simplicius’ and Philoponus’ conception of syllogisms from signs is discussed in Morrison (1997), who argues that these commentators conceive the syllogisms from signs in a way which is different from Aristotle. Morrison claims that the class of syllogisms from signs in Philoponus and Simplicius also includes other types of arguments concerning unprovable principles than those classified by Aristotle as syllogisms from signs. For Aristotle’s theory of inference from signs, see Allen (2001). 262 Cf. Menn (1992).

THEORIES OF ARGUMENTATION

151

whereas the things better known to us are compounds and compositions (qscjeus,2lnp, q6lhernp) of the principles (see, e.g., in Phys. 16, 7–14). It is not entirely clear what Simplicius means by the principles, and his discussion is a little problematic. At one point he talks about the four material elements (16, 14), but more often he discusses the elements of a definition (16, 20–24; 16, 31–17, 25). He also mentions Aristotelian physical principles, form, lack of form and the underlying subject (20, 17), but it is not clear whether he takes them to be principles in the intended sense. At one point he claims that the elementary principles should not be taken as unmoved (in Phys. 20, 9–11) and this seems to entail that the principles Simplicius has in mind are the material elements. Otherwise, however, Simplicius is a Platonist and it would be quite strange for a Platonist to consider principles of sciences as material elements. I shall, however, leave this problem for later research to solve. *

*

*

In this chapter I have treated discussions of principles of argumentation in the later Platonic-Aristotelian tradition in antiquity. I have concentrated on the question whether the later tradition endorsed the distinction formulated by Aristotle between what is better known to us and what is better known by nature, and whether this distinction was taken to entail a distinction between proofs in the proper sense and valid demonstrations from true premises which are not proofs, because they proceed in the wrong direction. We have traced the distinction and its connection to forms in Alcinous and some of the late ancient commentators on Aristotle. Alcinous does not make it explicit but in the commentaries the assumption is widespread that the order is dependent on explanatory power understood in metaphysical terms. In Galen we find the interesting move towards an attempt to make all the principles self-evident either to reason or to the senses. However, Galen recognises that in medical science, which he takes to be the highest form of human knowledge, we also need definitions as premises, and he admits that the scientific definitions are not self-evident in any normal sense of the word. Therefore, he ends up with accepting something analogous to the Aristotelian premises of proofs: we must accept a class of principles that are found as a result of inquiry but that function as a starting point for scientific explanations. They are not self-evident initially but can become rather evident to a learned scientist. In Plotinus the role of discussions concerning general requirements for argument forms and premises becomes much less prominent than before. The psychological distinction between ordinary conceptual thought and intellectual

152

CHAPTER ONE

non-mediated apprehension starts to dominate the discussion. However, in his treatise on dialectic, as we have seen, Plotinus recognises that ordinary thinking involving analysis of the conceptual order also promotes understanding of the structure of the intelligible. Behind this there is the assumption that our concepts are images of the intelligible objects and hence are similar to them. That is why operating with concepts and being involved in discursive thought can prepare us for intellectual apprehension: however, such apprehension is never the outcome or a conclusion of ordinary reasoning. One remarkable feature of the commentaries of Themistius and Philoponus is that they emphasiseAristotle’s idea that the general mathematical axioms are not strictly speaking the same in different mathematical sciences. They both address the example of Bryson’s attempt to quadrate a circle; both find from it a common axiom concerning magnitudes. Themistius does not even mention that the axiom is used falsely by Bryson. Rather, his criticism concentrates on the axiom being too general. Philoponus pays attention to Proclus’ criticism showing that the axiom seems to be false. Philoponus, then, criticises Proclus on the grounds that Proclus’ criticism uses numbers even though the original example was geometrical. Therefore, Philoponus assumes that even counterexamples have to be formulated by using the same kinds of quantities. I have also discussed some commentators’ reactions to Aristotle’s formulation of the distinction between what is better known to us and what is better known by nature that at first sight seems contradictory to his other formulations. All of them seem to assume, and some of them argue, that the formulation is not in fact contradictory. A common way of solving the problem is to point out that Aristotle is not using ‘universal’ in the same sense in the passages. In the Physics, it is argued, ‘universal’ refers to a vague general notion we initially have, and it is to be contrasted with a worked out definition. Finally, I have argued that we find evidence from Alexander and Philoponus showing that they did not recognise any major tension between Aristotle’s theory and his practice of science. Rather, they thought that both dialectic and science employ the same general scheme – which is presented by Aristotle in the Prior Analytics I 27–30 – of constructing valid arguments by listing general terms according to their connections with each other. If the premises express connections between those terms that are merely plausible but not true the argument is dialectical. If the connections are essential and the premises express reasons for the conclusion the argument is a proper proof. Alexander also discusses the question of how dialectic is used in science. He recognises four such functions: by analysing arguments pro and con we can, on the one hand, learn to distinguish the true position from false ones; on the other hand, such an exercise enables us not to be deceived by seemingly plausible but not true conceptions. Third, evaluating arguments on both sides

THEORIES OF ARGUMENTATION

153

enables us to see what the solutions of the problematic aspects of the competing views are. Fourth, we can argue dialectically for our premises against someone who is not yet convinced by them by using the dialectical techniques. He never says that dialectic could or even should establish the premises of proofs in a straightforward sense.

CHAPTER TWO INTELLECTUAL APPREHENSION

2.1 THE CONNECTION BETWEEN THE TWO CONTEXTS We have discussed above what kind of premises the philosophers working within a Platonic-Aristotelian framework claim we can take as immediately accepted in various situations involving argumentation. It has already been indicated that, in the Platonic-Aristotelian tradition, starting points for knowledge are also discussed from another perspective. In a rather homogenous way the philosophers belonging to this tradition, give an explanation or an account of how we acquire the basic contents of our intellect. Such an explanation is psychologically oriented; it is formulated with reference to the cognitive capacities of the human soul. It also relies on some metaphysical background assumptions. Most importantly, it is assumed that reality as such is intelligible and that human reason is capable of grasping it, at least to some extent. Further, the intelligible structure of reality is taken to consist of discrete elements, often called forms, which due to their necessary interconnections, form an organised whole. We do not find much explicit discussion concerning the question how the theories of principles of argumentation are related to the discussion of how the elements of the intelligible structure become known to human beings. The basic intuition behind the theories, however, seems to be that our claims and arguments are true and are concerned with the structure of reality only if we are capable of referring to external things in an appropriate way. Such reference, then, is supposed to require that at least small fragments of the intelligible structure are realised in our minds. How this happens is explained in theories referring to our psychic capacities and their interaction with reality. The process in which we gain an initial familiarity with the structure of reality can, with some accuracy, be called the process of concept formation. 155

156

CHAPTER TWO

However, the notion of a concept is precarious here for the following reasons. Firstly, concepts are often nowadays seen as abstract entities, as ingredients of propositions. However, Plato and Aristotle did not seem to posit any such abstract entities as propositions, and it is not quite clear whether they can be said to have a theory of concepts. They rather assume that because of the uniform effects the objects and their intelligibility have in our soul we are able to refer to things appropriately and we are capable of thinking of them. Such appropriate contact with the general intelligible aspects of things is necessary for any true discourse concerning them. The main epistemological upshot of the psychological theories is that there is a natural cognitive process taking place in all human beings through which we gain initial ‘knowledge’ of what kinds of things there are in the world. Such initial knowledge can be used to initiate inquiry into the nature of things in more detail. Secondly, and this is explicit in Aristotle, concepts – with the precautions made above – are not in essence dependent on language. Rather, the uniform effects the world has in human souls are primary to any conventions concerning language. In the Platonic-Aristotelian tradition we find the metaphor where thinking is compared to inner discussion or argumentation (diak2ceqhai). However, all such thinking presupposes that we have at least some grasp of the intelligible elements, and such apprehension is compared to seeing. The assumption of the intelligibility of reality also carries with it the assumption that there are special objects of knowledge, namely the intelligible forms that are unchangeable and eternal. Plato and Aristotle interpret the ontological status of the intelligible objects differently, and also their explanations of how we come to grasp them deviate from each other. However, both of them assume that knowledge must be about the general and unchangeable structure of reality and this requires that we are immediately familiar with at least some parts of that structure. I shall now briefly introduce the evidence of the assumptions concerning the connection between the theories of starting points in the framework of argumentation, and those of how we acquire the intelligible object in the first place. I shall begin with Plato. Three passages in particular must be considered here. The first one is found in the Meno, where belief is said to turn into knowledge on the condition that it is ‘tied down’ by an account of the reason why (aår4ap knciq,5p), and that is said to be recollection (98a).1 One fairly natural way to take this would be that all our knowledge claims are derived from some privileged group of innate truths which are latent but can be recollected. This would mean that the basic truths are known in the way that they do not presuppose an account of the reason why. The Meno, in fact, seems to 1

Cf. Phaedo 76b–c.

INTELLECTUAL APPREHENSION

157

provide evidence for such a reading. When the process of recollection is explained, the objects of recollection seem to be a class of basic arithmetical truths. However, it remains unclear whether all our knowledge claims should be derived from such truths only. I shall return to this passage below in 2.3.1. In the Phaedo, recollection has other kinds of objects. What we recollect are not basic mathematical truths but the forms, such as equality, the greater and the smaller, even the beautiful, the good and the just (75c–d). It is argued that if we did not have a grasp of pure equality as such, we could not learn it from experience either because all perceptible things are both equal and unequal with some other things. The Phaedo, then, points to the assumption that we could not form any general concepts or classify things appropriately if we did not have a pre-existing cognition or knowledge of the forms in our soul. Rather, when we see instances that are in some respects equal with each other, beautiful or good, we recollect or recognise equality itself, beauty itself, or goodness itself, and we also recognise that those instances are imperfect instantiations of these forms. In this way recollection of the intelligible form is a precondition of any apprehension or discourse concerning the general structure of reality. Another way of explaining how we come to apprehend the intelligible structure of reality is found in the Republic. In the metaphor of the cave it is suggested that we have a capacity of intellectual vision of the good, which can be attained if we manage not to concentrate too much on the perceptible world and its changeable properties. If we succeed in ascending to grasp the good, we then also understand perceptible reality in a new light. Such apprehension brings along with it understanding that everything around us manifests the good, because goodness accounts for all order we encounter. If the reality was not ordered in accordance with the good, it would be an incomprehensible chaos where there would be no regularity and we could not understand it at all. In addition, as pointed out above, the good as a basic structuring principle of reality also accounts for the fact that the best explanations we can have are teleological. Plato indicates that explanations referring merely to the material constitution of things are always deficient; they do not make things understandable to us, whereas teleological explanations make us see why certain events take place. The intellectual vision of the Republic can be contrasted with recollection in the Phaedo. Whereas the theory in the Phaedo seems to be designed to explain the possibility of truthful discourse in general, the Republic refers to the ultimate explanatory principle of reality. Both these passages, however, share the assumption that discourse and inferential thinking have their limits. They are always based on immediate cognition – or recognition – of some basic elements of the world’s intelligible structure. At the very top, so to

158

CHAPTER TWO

speak, we need immediate intellectual vision to guide our understanding of the world’s order. When we move to the late dialogues, the forms and recollection are not prominently present any longer. However, the assumption remains according to which our reason is at birth not ‘empty’, but some pre-existent contents are, as it were, programmed into it. The very great kinds (,2ciqra c2lg), most importantly the notions of sameness and difference, are posited as general categories that are already there in our reason, and this is used to explain the basic functions of reason along the following lines. We are able to classify things into natural classes and so on because the very same general notions that structure the functioning of our reason structure reality as well. In addition, in the Sophist (253d) the basic notions constituting the greatest kinds are taken to explain that we can get correct results by the method of collection and division. In that method, when things are similar in relevant respects they are collected under the same general type; relevant differences, on the other hand, determine the divisions among such general types. In Plato we find the basic intuition according to which some preexistent content in our reason, be they the basic truths referred to in the Meno, the cognition of forms figuring in the Phaedo, or the very general and abstract categories of the late dialogues, explain the alleged fact that we can grasp the truth concerning reality around us. The explanation of how we acquire truths, cognition of forms or the general categories is put in a mythical form both in the Phaedo and in the Timaeus. The pre-existent cognition of forms is explained by reference to our soul’s pre-natal existence; after the soul’s birth into the body we can then recognise the imitations of these forms in our environment. In the Timaeus, the mythical account of the presence of the basic categories or the very great kinds in our soul involves cosmological elements. The demiurge is said to make our soul of the very same ingredients as those the world is made of. Because the make-up of our soul is in principle the same as that of the world, we can have true cognition about it. As has been already indicated, Aristotle does provide us with some fairly detailed evidence of how the discussion concerning principles of argumentation and starting points for knowledge from a psychological point of view, are connected to each other. In the Posterior Analytics he discusses for a long time what knowledge in the strict sense requires. His general line of argument is built on epistemological considerations (in I 3) according to which all knowledge does not require proof. This position, Aristotle notes, avoids both an infinite regress and circularity. It is only in the last chapter of the Posterior Analytics where he finally turns to explain why and how.

INTELLECTUAL APPREHENSION

159

In the Posterior Analytics II 19, we then find the view that there is a natural cognitive process in virtue of which we acquire basic contents in our intellect. This process is described in psychological terms: first we make perceptions, these perceptions are stored in our soul, they make up a single experience and, finally, we come to have a universal fixed in our soul. The examples of human beings and animals are given a little later. A similar process is said to take place in other cases as well. In fact Aristotle has not much more to say about the psychological side of the process. This passage is notoriously difficult to understand as indeed is the whole chapter; I shall return to it below.2 What he seems to mean is that, in addition to experience, which is a kind of tacit generalisation concerning the perceived instances, which appears also in some non-human animals, we human beings come to grasp, e.g., species as species. When we have seen sufficiently many horses, we come to grasp that the horse is an animal species distinct from all the other animal species. We do not typically at this point know very clearly what the features are that, on the one hand, are shared by horses and some other animals and, on the other hand, distinguish horses from all other animals. However, at this point we in some sense understand that we are dealing with a unique animal species and we can identify the instances of that species. Aristotle mentions in various places that our intellect is unerring. This should not be taken in the sense that we can immediately and infallibly identify all the essential properties that are characteristic of a given species. Rather, Aristotle seems to mean that when we have attained the level of abstraction where we can recognise a species as distinct from all other species, we are unerring in having attained the level of universals. In the Metaphysics (IX 10, 1052a1–2; cf. DA III 6, 430b26–30) Aristotle says that not grasping the primary, non-composite objects of intellect is not error, but ignorance. Ignorance is described as a lack of contact (1051b25). There is no such thing as wrong apprehension of the primary intelligible objects: they are either apprehended or not (lne‹l z ,3, 1051b32). When, for instance, we start to inquire into water-animals we at some point recognise various species as different species even though we do not necessarily know their names (‘this is a kind of water-animal; this is another kind of water-animal’). Before that, it was not the case that we had a false grasp of the situation; we simply had not yet attained that level of abstraction. However, when we start to combine abstract notions with each other, error becomes possible (cf. DA III 6). Aristotle does not deny that we can err in claiming, e.g., what kind of a water-animal some animal is (‘dolphins are

2

Cf. also above the last section in 2.2 ‘Knowledge of the Premises’.

160

CHAPTER TWO

fish’). When we continue our inquiry, we learn what the distinguishing features are and why the species has them (‘this is a dolphin, which is a mammal, not a fish’). It is characteristic of Aristotle to assume that there is a psychological explanation based on some metaphysical background assumptions concerning how we gain an initial familiarity with the kinds of things there are in the world. By virtue of our intellectual capacity, we are able to refer to things correctly on a general level and recognise the instances of different natural kinds. This enables us to start inquiry. Because of our natural cognitive capacities of perception and intellection, we can typically answer the two questions at the beginning of the inquiry, namely ‘Does it exist?’ and ‘Does this attribute belong to that subject?’, in the way that we can start inquiring into what it is and why the attribute belongs to the subject. Aristotle’s example in Posterior Analytics II 19 is concerned with basic general notions; his examples are ‘human being’ and ‘animal’. The example thus indicates that one of the basic functions of the intellect is to apprehend the elements of the intelligible structure of the world, species, genera and the essential properties. However, Aristotle also assumes that the intelligible structure of reality has an intrinsic order and its elements are connected with each other in various ways. Typically, it seems that the intellectual capacity at first grasps parts of the genus species structure (e.g., ‘human beings are animals’, ‘dolphins are water-animals’, and ‘white is a colour’). Now, grasping in this way the basic elements of the intelligible structure in their natural order already brings with it apprehension of some basic truths. In the core of how Aristotle relates his discussion on the psychology of the intellect to his discussion of scientific proofs there is the assumption that we are only able to entertain propositional truths about the world if we are in an appropriate relation with the elements of its intelligible structure. Aristotle is not interested in investigating at length about whether we are capable of attaining general truths about reality in the first place; he assumes that it is rather clear that we do. He also assumes that he needs to present a psychological explanation of how we come to have basic contents in our intellect to explain our apprehension of true statements, not enter lengthy debates about the justification of this claim. However, Aristotle also seems to suppose that our intellect not only provides us with the starting point of inquiry, at least in some cases. It is also involved in finding the proper principles of reality, namely those truths that in the scientific proofs appear as premises. I have suggested above that Aristotle claims that it is the very same capacity which on the one hand, accounts for our basic ability to refer to things accurately on a

INTELLECTUAL APPREHENSION

161

general level and, on the other hand, enables us to recognise explanatory premises of proofs in inquiry when we understand how they explain the phenomena. Before turning to a brief discussion of the theories of perception, let me comment on some later developments in the Platonistic tradition. Galen, who considered himself a Platonist, shares some of the basic assumptions discussed above. In his treatise on logic (known by the Latin title Institutio logica), he at one point stops to discuss the question of how the basic notions are acquired. He presents an explanation very much like the one we find in Aristotle according to which we attain contents in our reason from experience. Therefore, he seems to assume that the premises of our inferences can capture truth in the relevant sense only if the basic notions are acquired in an appropriate way. However, Galen also claims that our reason contains some basic logical principles as innate. Therefore, he is clear in saying that even though the basic contents of our reason are acquired from experience, the reason must contain some pre-existing principles guiding its activity. There is no need to assume that Galen would refer to the theory of recollection, but he presents the claim that reason must have some basic functioning principles that are not acquired from experience. In Neo-Platonism we encounter another comprehensive view of the relation between theories of premises of argumentation and the psychology of the intellect. It is typical of Neo-Platonism that very much attention is paid to the metaphysics of thought, especially to the distinction between ordinary human thought, which is discursive, and intellectual apprehension proper, where thought does not involve transitions between concepts or propositions but is instantaneous apprehension of complex wholes. It is assumed that the metaphysical intellect and its eternal thinking constitute a necessary prerequisite for all human thinking. Were the Intellect not grasping the forms eternally, there would be no world to understand because the world around us somehow flows from the Intellect as an external aspect or product of its thinking activity. In addition, if there were no Intellect, there would be no ordinary human thought either because human concepts are metaphysically secondary to the forms in the level of the intellect. However, on the level of individuals – and on the condition that the metaphysical intellect exists – ordinary thinking can be initiated without an intellectual insight into the world’s eternal aspects. Human concept formation takes place as abstraction from experience, and concepts thus formed can be combined into complex propositional chains in ordinary thinking. However, as a description of reality such chains of statements are always deficient if one has not transcended the level of human thought and acquired an intellectual vision into the nature of things.

162

CHAPTER TWO 2.2 PERCEPTION

Even though their explanation of the relation between perception and intellectual apprehension differs, there is a common assumption among philosophers in the Platonic-Aristotelian tradition according to which our cognitive development starts from perception. Recollection cannot take place, the general categories of sameness and difference cannot be used to classify things, the intelligible structure of the world cannot be actualised in the soul, or we cannot obtain an intellectual level of thought without first making perceptions. Perceptions are, in that tradition, starting points in a very literal sense. In perception we receive basic information about our environment. From that we can start our development into knowers of more general and permanent aspects of reality. In Aristotle we also find the assumption that having perceptions is necessary for the actualisation of intelligible forms in the human intellect for the following reason. Intelligible form, which is actualised in things as their ordering principle, must somehow come into our soul. Even though both the intellect and the form are immaterial, the form cannot enter our minds directly. Only by virtue of some kind of physical interaction between the thing and our mind can we eventually come to grasp the form. The relevant interaction is perception. Perception, so to speak, brings intelligible forms into our soul and constitutes our first cognitive contact with the things. However, even though the form comes to us in this way in any ordinary perception, we are not ready to apprehend the intelligible aspects of the thing before we have had a sufficient amount of experience. It is a widely shared assumption among ancient non-sceptical philosophers that in sense perception there is a causal interaction between the object and the percipient. Another basic assumption is that causal interaction can only take place if the effective and the effected are in physical contact with each other. According to this latter assumption, there is no causation from a distance. However, at least in some places Plotinus seems to make an exception to the latter assumption. I shall return to this point below. The question of how distant objects can affect our sense organs is particularly pressing in the case of seeing; we see the object at a distance, but we do not see anything flying in the air in between us and the object. I shall refer to this question as the question of perceptual transmission in general and as that of visual transmission in particular.3 There are basically two different ways of explaining perceptual transmission in antiquity. According to one explanation, the object somehow comes to the sense organs of the percipient; according 3

For the ancient theories of visual transmission, cf. Emilsson (1988).

INTELLECTUAL APPREHENSION

163

to the other, the sensitive soul, as it were, extends to the object directly or indirectly. In addition, there is the possibility that both of these happen and perception either takes place somewhere between the object and the percipient or the object’s effect is brought back to the percipient. The question why physical contact with the object causes mental changes (e.g. why an effect in the eyes makes us see) is not in focus in the ancient discussion. Rather, it is assumed that a physical effect can be transmitted to the soul, the central sense faculty or the soul’s commanding centre, and this brings about a mental change (from not-seeing to seeing, for instance). An influential way of analysing the mental change is the Aristotelian one according to which the mental change is not a change of quality but a transition from not exercising a capacity into exercising it. This kind of change is called perfection; in the case of perception it is an actualisation of our potency to perceive; it is caused by the object affecting our sense organs. 2.2.1 Receptive Theories The theory according to which seeing occurs in virtue of effluences coming to our eyes from the visible objects, is found in Empedocles (Theophrastus in de Sensibus § 7
DK A 86; cf. Plato Meno 76c), who might have borrowed the theory from Alcmaeon of Croton.4 The early versions of the effluence theory were rough. General reliability of sense perception was explained by the assumption that an object literally sends parts of its body to our sense organs. The sense organs were taken to be specialised in the way that different organs allow only one particular sort of effluence. The organs contain pores, which are of different shapes and sizes. We, for instance, cannot hear colours, because the pores in our ears are of a different shape than those in our eyes specialised in receiving colour effluences.5 Of the earlier atomists, Leucippus and Democritus, it is difficult to say whether they are of the opinion that the causal traffic in the case of seeing goes from the object to the percipient or from the percipient to the object, or both. On one reading they are saying that the object affects the air next to it and that this effect is propagated to the part of the air adjacent to the affected 4

See, e.g. Barnes (1996, 478). However, some scholars such as Lindberg (1976) suggest that Empedocles did not take mere reception of effluences to be sufficient for perception, but some kind of activity was assumed on the part of the percipient. 5 A similar kind of roughness is still encountered in Epicurus’ theory, which will be discussed below. For a discussion on Epicurus’ theory of seeing, see Annas (1992, 157–163); cf. Everson (1990a) and Asmis (1984, 104–126).

164

CHAPTER TWO

air, and so on, until it reaches our eyes. According to this reading, we see the air next to our eyes as it is affected by the object.6 However, another possibility is that Democritus claims that the eye has an effect of its own in seeing. According to this interpretation, Democritus would claim that ‘a cone-shaped radiation of atoms’ proceeds from the eye to the object and that this radiation meets the films of atoms (e©dwka) coming from the object. When they meet each other, an image is formed in the air between the object and the percipient and this image is somehow brought back to the percipient.7 The latter reading would make the theory projective. I shall not deal with the question of which reading is more plausible. Causation through Medium Aristotle’s conception of seeing is connected with his general analysis of change. According to Aristotle, every perception is a special kind of change, which involves two different aspects. One is a mental change; another consists of an alteration in the sense organ. (See DA II 5, 416b34–35 and 417a30–b1). Aristotle is clear that the alteration is caused by the perceptible object (DA II 5, 417b16–26; cf. I 4, 408b16–17). Aristotle says that the object is ‘able to produce the activity’. However, this is not all there is for our capacity to perceive. In the general framework of Aristotle’s theory of change, interaction presupposes that the object that is capable of producing activity is capable of producing it only in things capable of performing that activity. For instance, even though in some sense we can say that colours also affect stones, they cannot affect them in the relevant way because stones do not have a potency to receive the effect.8 By contrast, in human beings who have the capacity to be affected by colours in the relevant way, the activity of seeing comes about if certain conditions are met. There is a single activity in which both these potencies are realised. In the case of seeing, this means that the visibility of the object and the percipient’s capacity to see are both actualised in the same act of seeing. This act takes place in the perceiver, i.e. that which is acted upon. (DA III 2, 425b26–426a11; cf. Phys. III 3, especially 202b6–8.)9 From the point of view of the sense faculty the act of seeing is a fulfilment of the subject’s essential capacity to see. 6

For this reading, see Furley (1989, 131–132) and Emilsson (1988, 39). This reading is also suggested by Furley (1987), see pages 132–133. 8 In the medieval Aristotelian tradition the assumption was formulated in the way that the object has an active potency to be perceived, whereas the percipient’s capacity is a passive potency. For the kinds and types of potencies in Aristotle, see Knuuttila (1993c). 9 See also Lear (1988, 102). 7

INTELLECTUAL APPREHENSION

165

Aristotle does not accept the atomist analysis according to which some effluences or material particles should arrive at our organs from the objects. He criticises the atomists for making seeing perfectly analogous to touch. Rather, he claims that the visible object must change the illuminated air or water adjacent to it and the air or water transmits the effect to our eyes. According to Aristotle, visual transmission takes place when the colour of the object which is at the limit of the transparent medium on the surface of the body (De Sensu 3, 439a30; cf. DA II 7, 419a17–20) changes (jile‹l) the transparent medium, which in turn changes the eye. The medium is such that light makes it transparent; only when it is made transparent by light is it capable of functioning as the medium for vision (DA II 7, 418b9–13). Aristotle is not quite clear on the details of how the effect through the medium is supposed to take place. However, he points out that both of these changes, namely the one between the limit of the body and the medium and that between the medium and the eye are changes of quality (8kkn4wqip); no movement of place is involved (De Sensu 6, 446b28–447a12). The change of quality in the medium is invisible; we do not see colours in the medium in between the object and the eye. A more controversial issue is the change taking place in the inner core of the eye (j5.g). Aristotle says that ‘a sense is a power of receiving into itself the sensible forms of things without matter’ (see, e.g. DA II 12, 424a18–19; cf. 424b2; III 2, 425b22–23, III 4, 429a13–18, III 12, 434a29). It has been suggested that the reception of a perceptible form entails that the sense organ literally takes on the quality perceived. This literalist reading, suggested by Richard Sorabji, takes Aristotle to be claiming that seeing red requires that something inside the eye turns red.10 The literalist view has been criticised, and some scholars, such as Myles Burnyeat, have claimed that according to Aristotle ‘the organ’s becoming like the object is … a noticing or becoming

10

See Sorabji (1974); the literalist thesis is also embraced by Everson (1997). Slakey (1993, 85) claims that in the De Anima Aristotle seems to be advocating the literalist view, but this view is absent from his other writings on senses (De Sensu 436b7, De Mem. 450a27–29, De Som. 454a7–11). Slakey (1993, 85) takes the De Anima to contain Aristotle’s mature view on the subject. Here he is following Nuyens (1948, 57), taking the Parva Naturalia to be earlier than the De Anima. Usually the order is taken the other way round (see, e.g., Kahn 1966, Block 1961 and 1988). It is, however, possible to find evidence for the literalist reading also in the Parva Naturalia (see, e.g., De Sensu 7, 448a2–5) and a sharp distinction between the views on these two works seems to me to be overstated.

166

CHAPTER TWO

aware of hardness and warmth’.11 According to Burnyeat’s reading, this is all there is to perception in Aristotle. No physical change is involved. There is considerable evidence that Aristotle takes physical effects to be necessary for perception to take place (see, e.g., DA II 12, 424a28–32; III 4, 429a29–b5, De Sensu 436a7–8; 436b7, Phys. VII 2, 244b11).12 There is also some evidence, which seems to support the literalist reading in the De Anima.13 This aspect of Aristotle’s theory has been the object of enormous scholarly interest and discussing it in detail goes beyond the scope of the present book. However, the following can be noted here. Firstly, Aristotle’s idea seems to be that there are two changes involved in sense perception; one is the change in the organs, the other the mental change in the perceptual faculty. Sorabji’s analysis concentrates on the former, Burnyeat’s on the latter change. However, it seems that none of these aspects alone is sufficient for perception to take place. Rather it seems that the change in the sense organs is a necessary but not sufficient condition for perception. The change in the sense organs brings about another change in the sense faculty, which can be characterised as perfection, i.e. a transition from not exercising a capacity to exercising it.14 However, that perfection could not take place without the change in the organs. Secondly, even though it is rather clear that some kind of physical change in the sense organs is necessary for there to be perception, it is not necessary to take it literally in the sense that the organ needs to become similar to the object in quality. Another possibility is to understand the change in the sense organ as follows.15 The form that is received does not inhere in the sense organ quite in the same way as it does in the perceived object, i.e. in the manner of making it coloured. Rather, it might inhere in the organ in a similar way to how it inheres

11

Burnyeat (1992, 21); cf. Burnyeat (1993). He also lists Philoponus, Aquinas and Brentano as former adherents of his interpretation (1992, 18). Burnyeat’s view has been criticised by Everson (1997) and Everson’s arguments against Burnyeat by Sisko 1998. 12 Aristotle explains the effect in terms of ratios (k5cnp). The sense organ itself is a neutral ratio (a mean) between two extremes such as sharp and flat, loud and soft, smooth and rough in sounds. The sensible qualities are also analysed as ratios. In virtue of being a mean the sense can register the perceptual qualities, i.e. ratios, within the range of these extremes. For this doctrine, see further, DA II 11, 422b29–30; 424a4–6; cf. Ward (1988). 13 Evidence for the literalist view is found in, e.g., 417a20, 418a3–6, 422a7, 423b30–424a3; cf. 435a22–24. 14 Cf. Lear (1988). 15 This reading is suggested by Thomas Aquinas; see Summa Theologiae Ia, q. 78 a. 3.

INTELLECTUAL APPREHENSION

167

in the medium. This would mean that it is possible that the quality also causes a change in our sensory system and that this effect is transmitted to the central sense faculty. Such transmission need not change the colour of its transmitter.16 The perceptual centre or the common sense faculty (jnil¢ a©qhgqip)17 seems to be the proper location of perceptions in the Aristotelian analysis.18 In sanguineous animals the centre is in the heart (De Juvent. 3, 469a11–15).19 I shall not discuss the theories of perception in the late ancient commentaries on Aristotle. Sorabji has argued that the commentators tend to dematerialise Aristotle’s account in the sense that they do not take an effect in the sense organs as a necessary condition for perception anymore.20 I do not wish to contest Sorabji’s general argument. However, I would like to point out that, for example, Pseudo-Simplicius explicitly points out that physical change in the sense organs is necessary for sense perception.21

16

For a suggestion that there is an analogy between the Aristotelian theory of perception and information coding, see Bynum (1993, 108); cf. Ackrill (1981, 66–67). There is, however, a danger in this comparison. Given that the word ‘information’ originates in Aristotle’s theory – it contains ‘form’ – and that Aristotle thinks that perception is reception of the perceptible form, the comparison easily becomes circular. 17 Aristotle also has other terms for the centre of perception, e.g., ‘common capacity’ (jnil¢ d6la,ip, in De Somno), ‘nonspecific sense faculty’ (aåqhgrij•l /1lrwl, in De Sensu) and ‘primary sense’ (/.‡rnl aåqhgrij5l, in De Memoria), see Modrak (1987, 67); cf. Block 1988. Hamlyn (1993, 128), however, takes them to refer to different faculties and gives different functions for them. This multiplication of faculties is not necessary. 18 See, e.g., De Sensu 7, 449a2–10: ‘But if we do not perceive the objects of a single sense in an indivisible moment, neither shall we so perceive the objects of different senses; for it was more possible to perceive the former together than objects different in genus. But if we perceive sweet and white with different parts of the soul, these parts either form or do not form a unity. But they must form a unity, since the percipient faculty is one thing; what one thing, then, does it perceive? These objects form no unity. Therefore there must be a single faculty of the soul by which it perceives everything; and yet it perceives different kinds of object by means of different organs.’ (Transl. Ross.) Cf. III 1, 425a14–b3 19 Aristotle explains the transition of the quality into the centre as follows: ‘[T]he air modifies the pupil in this or that way and the pupil transmits the modification to some third thing (and similarly in hearing), while the ultimate point of arrival is one, a single mean, with different manners of being’ (III 7, 431a17–19). 20 Sorabji (1991); for the evidence, see Sorabji (2004). 21 The relevant passage reads as follows: ‘Perceptual reception (y … rnfl aåqhgrijnfl ∫/ndnu3) happens with a movement (r.n/3), change (,erabnk3) and bodily subjection (qw,arij¢ /e‹qip) (for the sense organ has to be affected by something if the

168

CHAPTER TWO

2.2.2 Projective Theories Aristotle’s version of the receptive theory of perception was influential in later antiquity as well as in the Middle Ages. It was a rival to Plato’s theory introduced earlier, though Plato’s theory was also prominent in later antiquity. By classifying Plato’s account as a projective theory, I mean that it explains seeing by assuming that something goes out of the eyes to the objects. Probably the motivation of the projective theories was to explain that we see only those objects we look at. From the projective theorist’s point of view the atomist theory – as well as the Aristotelian theory of an affected medium – leaves this fact unexplained. If the illuminated air around us is affected by the colours all over, why do we not see the colours behind us? However, the Aristotelians probably thought that the effect has to enter our eyes at the right angle. First we shall take a quick look at Plato’s account of visual transmission. Consider the following passage from the Timaeus. Now, the pure fire inside us … they made to flow through the eyes: so they made the eyes – the eyes as a whole but its middle part in particular – close-textured, smooth and dense, to enable them to keep out all the other, coarser stuff, and let that kind of fire pass through pure by itself. Now whenever daylight surrounds the visual stream, like makes contact with like and coalesces with it to make us a single homogenous body aligned with the direction of the eyes. This happens wherever the internal fire strikes and presses against an external object it has connected with. And because this body of fire has become uniform throughout and thus uniformly affected, it transmits the motions of whatever it comes in contact with as well as of whatever comes in contact with it, to and through the whole body until they reach the soul. This brings about the sensation we call ‘seeing’. (Plato Tim. 45b–c; transl. Zeyl from Cooper (ed.) 1997.)

Plato makes it clear here that in seeing, the internal light of the eye comes out and is mixed with the external light, meets the objects out there and brings the effect it received from the object back to the soul where seeing takes place. The passage provided material for later ancient versions of the Platonic theory. Before going to the later developments on the basis of Plato’s account of seeing in the Timaeus, I would like to point to the Theaetetus, where Plato also discusses the question of how perception takes place. There perception is explained as an interaction of two powers, one active (r• /ninfll) and the other

sense is to be activated (/ahe‹l c1. ri de‹ r• aåqhgr3.inl ∫/• r‡l aåqhgr‡l, eå %le.ce‹l ,2kkni y a©qhgqip)), whereas intellect is unmoved (!r.e/rnp) and the grasping (,er1kgvip) happens according to its proper activity (jar’ nåje4al %l2.ceial y ,er1kgvip)’ (CAG XI, 228, 17–20; my translation).

INTELLECTUAL APPREHENSION

169

passive (… /1quwl) (159c).22 The object is active, we percipients passive. This is worth noting, because on the basis of the account of the visual transmission we might be inclined to suppose that the percipient is the active counterpart in perception, because he or she sends internal fire out from the eyes.23 However, it is not completely clear how the account in the Timaeus is related to that in the Theaetetus. One of the later Platonist theories of vision is Galen’s theory.24 Galen interprets the light coming out of the eyes as a visual ray which affects the air between the percipient and the object (Plac. Hipp. et Plat. 7.5, 5–10).25 Galen claims that in virtue of the visual ray’s coming out of the eye, our nerves, as it were, extend to the seen object. He says that when the external light, the visual ray and the air intermingle, the sensitive air is in the same relation to the organ of sense as the nerve which leads from the organ to the brain is to the brain (Plac. Hipp. et Plat. 7.5, 32–33).26 Galen’s account seems to entail that when the visual ray exits from our eyes, the air around us becomes sensitive just as our nerves are. This would make all perception analogous to the internal perception inside a living organism. However, external perception would still only be analogous to but not the same as internal perception. Therefore, we could take Galen’s account in the sense that the relevant property of the nerves he uses in the comparison between them and the air affected by a visual ray is the capacity to transmit qualities and effects. This would fit quite well with the following comparison of Galen’s. According to Galen, placing an obstacle between our eyes and the object is analogous to a nerve being severed. That is, when a nerve is severed it cannot transmit anything; similarly, when an obstacle is put between an object and our eyes, we do not see because the quality of the object cannot be transmitted to the eyes. Therefore, we need not conclude that Galen completely assimilates seeing with internal perception. 22

I am grateful to Lesley Brown for reminding me about this passage. In the Theaetetus perceptible qualities are analysed as twin-products of the object and the percipient. The overall aim of the Theaetetus is to discuss perceptions from an epistemological point of view and their unreliability is underlined. By contrast, the Timaeus is closer in spirit to natural philosophy. There is no particular need to deny the possibility that the analysis of active and passive counterparts presented in the Theaetetus applies to the Timaeus as well. 24 The Stoic theory of visual transmission was also a version of Plato’s theory. For general accounts of the Stoic theory of seeing, see Annas (1992, 71–72); cf. Hahm (1977). 25 For the questions which Galen found urgent in the theory of perception, see Plac. Hipp. et Plat. 7.7, 9–10. 26 See further, Emilsson (1988, 59). 23

170

CHAPTER TWO

2.2.3 Co-affection: Plotinus Plotinus’ theory of vision is difficult to classify. On the one hand, he endorses Plato’s conception of internal light. On the other hand, as Emilsson has argued,27 Plotinus’ theory differs from Galen’s in not involving a visual ray. Further, as anticipated above at the beginning of this chapter, Plotinus sometimes seems to make an exception to all other theories in the sense that he does not take physical contact to be necessary between the object and the percipient. He at times seems to claim that the object can affect our eyes in the way that no physical intermediary is needed (Enneads, 4.4.23 and 4.5). This is why it deserves a section of its own. Visual transmission without physical contact is explained by the so-called co-affection (qs,/1heia) based on similarity between different parts of the cosmic organism. By virtue of this kind of co-affection psychic states can be transformed from one place to another without any kind of physical contact.28 In the case of seeing, the relevant similarity is that both the eye and the colour seen are of luminous nature (for the eyes, see 5.5.7, 28–29; for the colours, 2.4.5, 10–11; 5.3.8, 20; 5.5.7, 30). Plotinus does not explain in detail how co-affection is supposed to take place. However, he does not rest the theory of seeing completely on coaffection either. He also claims that forms are present in the air all the time, and presumably these forms play a role in establishing the connection between our soul and the object.29 These forms are invisible, but they somehow link the organ of vision and the visible object together in an invisible way. Plotinus says that the forms are not present in the air ‘as a bodily affection but in accordance with high psychic forces of a unitary sympathetic organism’ (4.5.3, 36–38).

27

Emilsson (1988, 43–44). Plotinus gives no clear account of the co-affection, but it is clear that his conception differs from the Stoic one. For Plotinus’ conception of co-affection, see Emilsson (1988, 48–62). Plotinus uses it to explain, e.g., the efficacy of magic and prayers and how the celestial bodies affect us (see, e.g. 4.4.26, 14–16; 4.4.32, 1–17; 4.4.40; 4.9.3, 1–6). This can be contrasted with the Stoic account of co-affection, which involves physical contact between the co-affected objects. Such physical coaffection is possible in the Stoic framework, because everywhere in the cosmos there is /lefl,a, a mixture of fire and air, which physically connects even distant parts of the cosmos. The Stoic theory of co-affection was taken to explain such natural phenomena as tides and the change of seasons, but also such ‘super-natural’ phenomena as divination, see Sext. Math. 9.79, Cic. Div. 2.34 and Nat. Deor. 2.19. 29 Emilsson (1988, 62). 28

INTELLECTUAL APPREHENSION

171

We might take Plotinus’ forms in the air as analoguous to air capable of transmitting qualities and effects that we found in Galen. However, unlike Galen, who assumes that the air becomes such only when a visual ray exits the eyes, Plotinus assumes that the presence of forms in the air is a permanent condition. This assumption seems problematic, though. If the presence of forms in the air is taken to be sufficient for there being co-affection between the eye and the object – on the condition that the object is of luminous nature (i.e. that there is either light coming from the object or that the environment is illuminated) – it is not clear any longer why we do not see everything all the time. 2.2.4 Perceptual Realism and the Reliability of Perceptions It is characteristic for the ancient theories of perception discussed above, to assume that perceptions are mental changes brought about by physical interaction between the perceptible object and our sense organs. From the metaphysical point of view the theories can be classified as forms of direct realism. According to the ancient theories, in perception we are aware of the external world. This is to be distinguished from indirect realism, which denies that we are aware of external objects directly. According to indirect realism, we are directly aware only of some kinds of mental intermediaries, which are somehow related to the external objects.30 In addition, many theories discussed above accept that when we perceive we become aware of the perceptible aspects of reality in a reliable way. This means that when we perceive a certain thing to have a perceptible quality, we are mostly right if we take it to have that quality. However, this does not mean that perceptions must be completely unerring. Mistakes might occur, but by making perceptions we find out what kinds of perceptible qualities things have. However, it is well known that in Platonism there is a tendency to be dubious about perceptions. Let us now turn to discuss this attitude. 30

For representational theories, see, e.g., Smith and Jones (1986, 85–89); for a recent defence of direct realism, see Smith (2002). Sextus Empiricus argues that it is possible that we are not aware of external objects but only of some mental intermediaries, or that we are not aware of the objects at all, not even through the intermediaries (see Math. 7. 364–367; cf. 7. 354). The Cyrenaics might have advocated a form of radical perceptual scepticism according to which we are only aware of some inner affections, not of external objects. ‘They affirm that mental affections can be known (jarakg/r1), but not the objects from which they come’ (Diog. Laert. 2. 92; cf. Sext. Emp. Pyr. 1.215. and Plutarch Adv. Col. 1120c–d). However, it is equally possible that the Cyrenaic dictum was intended to support their hedonistic ethics, not as a separate claim about the objects of mental acts.

172

CHAPTER TWO

Plato: Realism without Reliability? I shall consider two alternatives how Plato’s alleged suspicion towards perception can be understood. The first alternative is the following. In Plato’s framework the senses can be taken to be inferior in the sense that even though we perceive the perceptible aspects of reality in an accurate manner, perceptual ‘knowledge’ always falls short of knowledge proper. This is due to the requirements concerning knowledge proper, not to the capacity of our senses to present to us the perceptible aspects of reality. According to this reading, perceptual ‘knowledge’ can never be knowledge in the full sense because the perceptible properties of reality change. If I know perceptibly that Socrates is white, this is not knowledge in the proper sense of the word, because Socrates can cease to be white in the summer in the sun. In this reading the reliability of my perception that Socrates is white is not in doubt; it is rather that the statements we know in the full sense, cannot be false. Another possibility is that Plato is casting doubt on our senses’ capacity to reach the perceptible aspects of reality. This would amount to the claim that our senses are not reliable witnesses concerning the perceptible qualities. It is true that in some places Plato seems to have a very low opinion of sense perception. In the Phaedo (65b), for instance, Plato points out that even the poets tell us that we do not see or hear accurately. He asks whether we find any truth in perceptions and answers in the negative. It would appear that Plato uses arguments involving both these points of view against perception in different places. Concerning the first alternative we have passages like that found in Republic VII (529b–530b). There, even though Plato is criticising the use of perceptions in astronomy, he does not question our capacity to see the star constellations’ visual form accurately at any given moment. In addition, in another passage of the Republic (VI, 507c) he makes quite explicit that audible things are perceived by hearing, and so on. He does distinguish between these aspects of reality and the intelligible realm but, at least in these passages, the point does not seem to be that we cannot trust our senses at all in the case of perceptible matters. In addition, also in the Theaetetus, Socrates seems to concede that perception in relation to perceptible qualities is infallible and that it would, with respect to those qualities, fulfil one crucial requirement for knowledge.31

31

Cf. Spruit (1994, 33–36). He argues that the Theaetetus contains the claim that although perception can in a sense be taken as infallible – infallibility being one of the characteristic marks of knowledge – perception cannot be knowledge proper, because it does not have the real [i.e. intelligible forms] as its object.

INTELLECTUAL APPREHENSION

173

Furthermore, Plato is clear that perceptual experience is needed to initiate the recollection process (e.g. Phaedo 72e–75e) and that it is a prerequisite for philosophical dialectic (Rep. VII 523a–524d). Therefore, it seems that at least as far as perception is relevant for our present topic, it is much more likely that Plato’s suspicion towards the senses is to be taken in the first way. This means that we do capture the perceptible aspects of reality as they are, but perception is not knowledge in the proper sense, because the perceptible world changes and is hence not a proper object of knowledge. Aristotle’s Realism: Perceptibility as a Modalised Notion It is typical for Aristotle to analyse perception as an actualisation of two distinct potencies. In the Aristotelian analysis perceptible qualities are potencies to be perceived, and these potencies are realised in the perceiver. In modern parlance the perceptible qualities as analysed by Aristotle could be called dispositional. Although the perceptibility of the object is actualised in the perceiver it is not produced by him. The object’s perceptibility consists in its capacity to affect the right kind of subject – one that is capable of receiving the perceptible form – in a determined way. Therefore, Aristotle’s theory of perception can be characterised as a type of direct realism in which perceptibility is analysed as a modalised notion. Because perceptible qualities are in this way dispositional, it would make no sense to talk about perceptible qualities if there were no subjects capable of perception. It is most likely that Aristotle would say that the objects would be there and that they would be exactly as they are even without perceivers (cf. his discussion concerning air being breathable in Top. V 9, 138b30–37).32 However, it would make no sense to say that the world is perceptible if in world history there were not a single actual perceiver. In that case, their generic potency to be perceived would not be a real possibility.33 Aristotle assumes that the perceptibility of the world is actualised in perceivers in an accurate way because sense organs do not add any contribution of their own to the effect caused by the perceptible object. In relation to their proper objects, senses are never mistaken (DA II 5, 418a12–16; III 3, 427b12– 13; cf. also III 6, 430b29) or ‘admit the least possible amount of falsehood’ 32

It is true that this passage is dialectical and some scholars might doubt its value as evidence. However, it seems to me that often these rather casual remarks reflect Aristotle’s attitude. 33 Saying that perceptibility is a generic potency leaves open the possibility that there is a single perceptible object that remains unperceived throughout its entire existence, provided that other instances of its kind are perceived in the eternal history of species; for generic potencies in Aristotle, see Knuuttila (1993c).

174

CHAPTER TWO

(DA III 3, 428b18–19).34 In the case of proper sensibles, when a sense is in good condition it is a neutral mean between a pair of extremes determining the range of qualities perceived by it (DA II 11, 422b29–30; 424a4–6). In this case the sense does not add anything of its own; it only registers the effect caused by the object. Aristotle, however, does not claim that perceptual error is impossible. He only says that perceptions are highly reliable and typically in the case of perceptual error we can find an explanation why the error occurred. An example of a case where perceptual error has a clear explanation is when sickness causes us, for instance, to find everything that we taste bitter because the tongue is ‘overflowing with bitter moisture’ (DA II 10, 422b9). In that case the malfunction caused by sickness can be explained physically. It is likely that, according to Aristotle, in that abnormal state the sense organ does have a perceptible ratio. The organ registers this ratio together with the ratio of the object, and the result is not the same as the ratio of the object. Perceptual error is, according to Aristotle, most likely to occur in cases where there is an incidental object of perception and a common sensible object connected with it. Such mistaken perceptions are explained in the De Anima by talraq4a.35 talr1q,ara stored in the memory are connected with actually perceived objects. We, for instance, mistakenly take the perceived white object to be Diares’ son, whereas it is not. However, Aristotle seems to assume that cases of error are exceptional and in general the senses are reliable. In addition, it is usually not difficult to tell whether the sense is in good shape and whether the conditions are otherwise normal or not (see, e.g., Met. IV 5, 1010b3–10).

34

The combinatory perceptual functions are more complicated and more liable to error. Error is most likely in the case of common sensibles (movement, rest, figure, magnitude, number, and unity) which accompany the incidental objects of perception (DA III 3, 428b18–26, cf. II 6, 418a16–17). This seems to entail that we are liable to mistakes as to, for instance, whether someone we see is moving or not or with respect to the number of people we see, and so on. Would it not be more plausible to assume that error takes place more often in recognising who the person is than with respect to his or her properties classified as common sensibles? However, a mistaken recognition is classified as a type of misrepresentation: a wrong t1lraq,a from memory is connected with a perception, which is as such accurate. 35 In fact Aristotle sometimes seems to be restricting the talk of perception to veridical cases (see, e.g., DA III 3, 428a11–15). The cases of misrepresentation are classified under the title talraq4a. However, he does not restrict talraq4a to cases of ‘non-paradigmatic sensory experiences’ as Schofield has suggested (1978, 101). For the roles of talraq4a, see D. Frede (1992) and Everson (1997, 169).

INTELLECTUAL APPREHENSION

175

It seems that in the Aristotelian framework perceptions as psychological processes are different from hallucinations, illusions, afterimages and such because they have different objects. Dorothea Frede has suggested that a passage (De Somno, 460b1–3) where Aristotle distinguishes between awareness of afterimages and awareness of perceptible qualities of objects is evidence for the view that awareness of afterimages is awareness of talr1q,ara,36 not of the qualities of the objects.37 It is a general tendency in Aristotle to make a distinction between perception proper and something just seeming to be the case but not really being so. This tendency is combined with a tendency to use ‘perception’ as a success word and to use talraq4a to denote cases when we are in error. When we perceive an object, we capture its perceptible aspects in an accurate manner. If we are mistaken in perception, e.g., a person approaching seems to be the son of Diares, but on closer inspection turns out to be the son of someone else, we at first fail to be aware of that person, we are only aware of our own t1lraq,a of a different person stored in our memory. In this chapter we have discussed some aspects of ancient theories of perception. The theories of perception are starting points for our cognitive development in a literal sense. In the Platonic-Aristotelian tradition there is also the tendency to see perceptions as instrumental to a further goal: our becoming aware of the intelligible structure of the world. The question of how visible objects affect us received a lot of attention in antiquity. This question was often understood as an explanatory one; a descriptive account was provided of how the interaction between the object and the subject takes place. Much less attention was paid to questions of how such interaction between the object and the subject produces mental changes, or whether our perceptual beliefs are justified at all. This is in particular typical of Aristotle. His attitude towards sceptical doubts about whether external objects really have those perceptible qualities we perceive them to have was rather dismissive. 2.3 FROM PERCEPTION TO INTELLECTION We have now taken a look at Plato’s and Aristotle’s theories of perception. In the Platonic-Aristotelian tradition it is a common assumption that, starting from perception, we can ascend to a level of intellectual cognition of the world where we come to know things, not only with respect to their perceptible appearance but 36

The Greek t1lraq,a (of which we have the plural in the text) is a difficult one to translate. Like many other scholars, I have chosen not to translate it. Here t1lraq,a refers to some kind of representation of the object. 37 See D. Frede (1992, 286).

176

CHAPTER TWO

with respect to their unchangeable nature. I shall now turn to discuss how the philosophers in this tradition explain the transition from perception to intellectual cognition. I shall first deal with Plato and Aristotle, who both consider the immutable structure of reality to consist of intelligible forms. The assumption of the existence of forms is often preserved also in the later PlatonicAristotelian tradition. However, the theories are developed in other respects. Most importantly, the Platonic tradition turns into Neo-Platonism and the commentators on Aristotle give a prominent role to the discussion of an active intelligence and its role in human cognitive development. 2.3.1 Intelligible Forms Plato We noted above at the beginning of chapter 2 that, according to Plato, our reason must have some kind of content or general structuring categories in it before we are born. We cannot learn such basic structures from the perceptible world around us because there the general categories are imperfectly realised. Plato has two ways of explaining what the basic contents are, how we come to know them, and how they are related to the order of reality. The first account is the theory of recollection, the second the theory of the greatest kinds. I shall not discuss these theories in detail here; my main objective is to provide their basic contours in order to make a contrast both with Aristotle and with the developments in later Platonism. Recollection. According to the theory of recollection, there are pre-existing basic contents in our reason, which are forgotten at the time of birth but can be remembered or recollected during our bodily life. However, it is not entirely clear what we are supposed to recollect. The theme of recollection has been widely discussed in the scholarly literature and basically four different readings can be distinguished. The first one can be called ‘traditional’. According to the traditional reading, recollection refers to the view that we have grasped the intelligible forms in their perfection before our birth into body. In birth we have forgotten what we learnt. When perceiving things in our perceptual environment, however, we can come to recognise them as imperfect copies of those forms, because – as copies – they resemble the forms themselves. The main evidence for the first account of recollection is found in the Phaedo. The example is that we can recognise that certain perceptible things are equal because we have before birth cognised pure equality.38 The first 38

See, e.g., Bostock (1986).

INTELLECTUAL APPREHENSION

177

account takes Plato to claim that all ordinary concept formation and any classification of perceptible things presuppose pre-existing cognition of the forms. The assumption behind this view is that, for instance, equality even as a humble human notion could never be learned from experience only, because all the things we perceive and which are in some respect equal, are at the same time not equal in other respects. Equality in general – which is the same in all cases of imperfect equality – cannot even be used as a classificatory notion if there is no pre-existing cognition of it. The first reading is closely tied to the traditionally understood Platonism exemplified in the dialogues of the middle period, though, as is well known, Plato distanced himself from that theory in the late dialogues. To mention just two examples, the relationship between the forms and particulars comes under critical scrutiny in the Parmenides. Another assumption that Plato seemed to abandon later was the assumption that relational predicates can have forms. Relative forms appear in the theory of recollection in the Phaedo. The second reading is based on the Meno and, more specifically, on Socrates’ discussion with Meno’s slave boy. According to this view, the point of the discussion is to show that even though the boy has never studied mathematics and even though Socrates does not teach some basic arithmetical truths to him, he can recognise them in the discussion.39 This is taken to imply that the fact that the boy accepts some basic mathematical truths is an indication that he has those truths in him, but he does not know it at first. Such knowledge can be explained by appealing to recollection. The boy has learned the truths before birth, but has forgotten them during his bodily existence. Third, it has been suggested by Dominic Scott40 that the theory of recollection is not designed to explain learning in general, but that it only refers to the higher understanding which philosophers can have of the forms. This suggestion is problematic with respect to the Meno because it is made quite clear in the dialogue that recollection applies to Socrates’ discussion with the slave boy. In that discussion the recollected objects seem to be simple mathematical truths, certainly not advanced philosophical theories.41 The fourth suggestion is that we should not take recollection to imply that the slave boy should have some pre-natal explicit truths in his soul. By contrast, Lesley Brown argues, the example points solely to the fact that the boy has an innate capacity to use his reason.42 Even when untrained – as is the case with the slave boy – this capacity can lead people to follow inferences and 39 40 41 42

Cf. Vlastos (1965). Scott (1995). For criticism of the third reading, see also Brown (1997). For this reading, see Brown (1991).

178

CHAPTER TWO

recognise simple logical-arithmetical truths in communication. I agree with Brown that to some extent the theory of recollection can be taken to explain our reasoning capacity. However, even if we accept that the discussion with the slave boy has its emphasis on the boy’s ability to reason, this does not rule out the possibility that there are also supposed to be pre-existing contents, truths or notions, in our reason. To sum up, different dialogues point to different aspects of the process of recollection and, at the same time, to different aspects of our rational capacity. The Phaedo brings forward our capacity to make accurate classifications emphasised by the first reading; the Meno involves the capacity to recognise mathematical truths, and perhaps some other simple necessary truths. The fourth suggestion points to our natural ability to follow reasoning and recognise contradictions. Very Great Kinds. In the late dialogues we no longer encounter the theory of recollection. A crucial role is given to the very great kinds (,2ciqra c2lg). In the Sophist (253d) Plato suggests that all dialectical activity, collecting and dividing, is based on the basic categories called ‘the greatest kinds’; they are a little later (254b–256a) identified with being, change, rest, sameness and difference.43 Those very great kinds are taken to be such general abstract categories that everything there is has in some respect a share in all of them. The aim of the activity of collecting and dividing aims at capturing the relevant sameness, difference, change, rest and being in the thing which is chosen as the object of inquiry. What, then, are the very great kinds? One possibility seems to be that they are categories used by the human mind to organise perceptual material so that natural kinds can be distinguished from each other and the representatives of general kinds can be grouped together. However, to say that the very great kinds are solely categories used by the human mind would be misleading. In the Timaeus being, sameness and difference are identified with the ingredients out of which the demiurge composes the cosmos and souls. Therefore, the very great kinds are not merely notions of the human mind; they are also categories according to which perceptible reality is organised. The fact that we are able to recognise these categories is explained in the context of the Timaeus’ cosmological-mythological story by saying that our soul is made out of the same ingredients as the world.

43

Cf. also the common kinds (r¡ jnil1) of the Theaetetus discussed in D. Frede (1996). Notice that Aristotle’s so-called common sensibles include most of what Plato classifies as very great kinds (being, number, movement, rest).

INTELLECTUAL APPREHENSION

179

It is important to note that Plato’s discussions concerning the very great kinds in different places all point to the assumption that to explain how we can get anything out of our perceptions, we already need to have some general notions in us; we can then organise and interpret the perceptual data according to these notions. Plato’s explanation of why this enables us to make true classifications is in the Timaeus put in the somewhat robust way of claiming that the soul has as its ingredients the same abstract categories as structure perceptible reality. Dorothea Frede calls this feature the ‘economy of Plato’s psychology’. No new explanatory principles need to be postulated; the same ones explain the order both in the world and in our thought. On the one hand, the natures of all things in the perceptible cosmos are a mixture of sameness and difference. All things are in some respects similar to some other things and in other respects different from others. However, because sameness and difference as such are not there in the perceptible realm – we only find similarities and differences between some things in some respects there – we cannot learn them from there either; they must be in our soul. Seeds for Later Developments. As has been mentioned above, in later Platonism, especially in the Platonism of Plotinus, the following assumption becomes prominent. All so-called discursive reasoning (di1lnia) is inferior to a mode of intellectual apprehension of a comprehensive whole of intelligible forms in their interrelations.44 According to Plotinus, the content of intellectual apprehension can only be communicated through discursive reasoning, by telling what one ‘saw’ and formulating in statements the relations between the intelligible objects one grasped. This involves articulating propositions and arranging them into chains. According to Plotinus, these propositions and their chains – even though they are the results of attempts to articulate the content of intellectual apprehension – can never incorporate the whole richness and exactitude of intellectual apprehension itself. We can compare this to attempts of putting into words a beautiful landscape we see containing people, food, wine, furniture, musical instruments and so on. All descriptions sound blunt understatements compared to actually seeing it. In Plato we can find anticipations of the kind of intellectual ‘vision’ we encounter in Plotinus. An example of such anticipation is the analogy of the cave and grasping the form of the good in Republic VII. The apprehension of the good does not solely consist of grasping the good itself, but it makes one grasp the whole reality in a different way, namely, as manifesting order, which is produced by the good. Understanding that the good explains nature’s 44

See, Emilsson (2003).

180

CHAPTER TWO

regularities and order, is compared to grasping how the sun causes the seasons and all other things as well (Rep. VII, 516b–c). When the person who turned around and exited from the cave goes back and tells the others what he or she saw outside, telling about the vision will not be quite the same as seeing it. This aspect of the analogy of the cave can be seen to anticipate Plotinus’ doctrine of the inferiority of discursive reasoning to intellectual apprehension. Another point where we can find material for the later Neo-Platonic developments is the Seventh Letter.45 I shall suspend judgment concerning the question whether the letter is authentic or not,46 but be that as it may, the letter reflects the direction Platonism was taking either immediately after Plato or late in his lifetime. Especially the paragraph 342a–343b emphasises that an understanding of the nature of a thing by reason (lnflp) is always superior to a verbal expression of it.47 The definition is merely ‘a combination of nouns and verbs, and there is nothing surely fixed about it’. The sharpest point of the criticism is directed at language. Because language is conventional, all expressions of a nature of a thing will always be dependent on contingencies. By contrast, a true intellectual apprehension captures the real nature of the thing without such conventional intermediaries. 45

I am grateful to Rachel Barney for a discussion on this point. For a short comment on the letter’s authenticity, see Cooper (1997). 47 The relevant passage is as follows: ‘For every real being, there are three things that are necessary of it is to be acquired: first, the name; second, the definition; third, the image; knowledge comes fourth, and in the fifth place we must put the object itself, the knowable and truly real being. To understand what this means, take a particular example, and think of all other objects as analogous to it. There is something called a circle, and its name is this very word we have just used. Second, there is its definition, composed of nouns and verbs. “The figure whose extremities are everywhere equally distant from its center” is the definition of precisely that to which the names “round,” “circumference,” and “circle” apply. Third is what we draw or rub out, what is turned or destroyed; but the circle itself to which they all refer remains unaffected, because it is different from them. In the fourth place are knowledge (%/iqr3,g), reason (lnflp), and right opinion … ; of these, reason is nearest the fifth in kinship and likeness, while the others are further away … On this account, no sensible man will venture to express his deepest thoughts in words, especially in a form which is unchangeable, as is true of written outlines … And we say that their names are by no means fixed; there is no reason why what we call “circles” might not be called “straight lines,” and the straight lines “circles,” and their natures will be none the less fixed despite this exchange of names. Indeed the same thing is true of the definition: since it is a combination of nouns and verbs, there is nothing surely fixed about it.’ (342a–343b.) (Translation by Glenn Morrow from Cooper (ed.) (1997), italics mine.) 46

INTELLECTUAL APPREHENSION

181

Aristotle Let us now move on to Aristotle and his theory of the transition from perception to intellection in the context of his general theory of human cognitive capacities. Posterior Analytics II 19. In the last chapter of the Posterior Analytics Aristotle presents, as we saw above, an account of how the starting points for proofs become known. It is important that normative discussions concerning the justification of generalisation or the nature of sufficient evidence, are markedly absent from that discussion. He has epistemologically oriented arguments, also in the Posterior Analytics, to show that the starting points for proofs need to be known independently of proof (I 3). However, his account of the acquisition of the starting points is better characterised as psychological than epistemological, and his approach is descriptive. Basically, he gives an account of how our intellectual disposition becomes activated through perception and memory. Intellect (lnflp), asAristotle indicates in the PosteriorAnalytics II 19, is a disposition ()mip). As such it is an acquired capacity and can be contrasted with perception, called an innate capacity of making distinctions (d6la,ip q6,tsrnp j.irij3, 99b35). To say that perception is innate means that all animals are ready to perceive straight after they are born; intellect is not innate in this sense. In the De Anima (II 5), our intellect as a potency is compared to a very young child’s capacity to learn geometry. Such potencies can be called innate in the sense that all human beings are capable of acquiring geometrical skills, but they need to be distinguished from potencies immediately ready for use; e.g., a potency to use knowledge which one has already acquired.48 Perception is a potency of the latter, intellect of the former sort. We have already been given a preliminary picture of Aristotle’s account of how the intellectual contents are acquired from experience. Let us now consider the following lines: For some animals there remains a trace (,nl3) of what has been perceived; some animals do not retain such a trace. For those incapable of retaining the trace, there is no knowledge (cl‡qip) outside perception; for those who can the trace remains in the soul. When many such [traces] have occurred, there will be yet another distinction so that in some animals reason (k5cnp) develops from such traces; in others this does not happen. So from perceptions comes memory, as has already been said, from memories of often having observed the same thing, experience (%,/ei.4a); for many memories make up a single experience. From experience, or from the whole universal (%j /alr5p … rnfl jah5kns) which has come to rest in 48

For the difference between the types of potencies, see, e.g., Sisko (1996, 142–148) and Lear (1988, 103).

182

CHAPTER TWO our soul – from the one besides the many (rnfl ^l•p /a.¡ r¡ /nkk1), which is the same in all the instances – comes the starting point (8.u3) of art (r2ulg) and science (%/iqr3,g). If it is concerned with coming to be (c2leqip), it is the starting point of art, if with being (r• ≈l), of science. Therefore, the dispositions ()meip) do not belong to us as determined; neither do they become from other cognitively higher (8/’ !kkwl )mewl … clwqrijwr2.wl) dispositions; they come from perception, as in a battle, when the soldiers retreat, after one has stopped the others do until the original order has been reached ()wp %/§ 8.u¢l ;khel). The soul is such that it is capable of undergoing this (y d£ vsu¢ ∫/1.uei rnia6rg nœqa n˙a d6laqhai /1queil rnflrn). (99b36–100a14; my translation.)

Before we turn to discuss the contents of these lines, it can be noted as an aside that Aristotle’s account at the beginning of the quotation on lines 100a3–4 is almost word for word the same as the one Plato rejects in the Phaedo (96b). Above in chapter 1, we have discussed the Posterior Analytics II 19 in the context of Aristotle’s theory of scientific proofs. I would now like to pay attention to the following three aspects in Aristotle’s account. Firstly, in this short passage Aristotle describes in a nutshell how we acquire our rational capacity (k5cnp on 100a2, quoted above). His account makes it clear that in the core of this capacity, there is an apprehension of a genuine universal. Therefore, for Aristotle, reason is not merely the capacity of making inferences. It is a capacity which is characterised by its contents.49 This is not to say, of course, that reason would not involve a capacity for making inferences. The examples he gives of such contents are notions of general kinds, ‘human being’ and ‘animal’. Later on, at the very end of the chapter, he calls such apprehension ‘intellect’ (lnflp) and argues that intellect must always be correct in its apprehension. Therefore, to have reason is not merely to have some universal contents in the mind but to have some correct contents. As has been noted above, this does not mean that having such correct contents would involve full knowledge of the scientific definition of the species and genera. Rather, it is apprehension that there is such a natural kind of thing, as distinct from all other kinds of things. In addition, the correctness of the notion also involves the assumption that it is not vacuous, but refers to an existing natural kind of thing. Secondly, acquiring universal and correct contents in one’s intellect is for Aristotle not in essence an inferential process. As mentioned, Aristotle does not address questions like ‘How do we infer the universal from perceptions?’, ‘Can perceptions justify a universal generalisation?’or ‘How is inductive inference – being not deductively valid – capable of justifying generalisations?’I would like 49

Cf. M. Frede (1994); his emphasis is on the Stoics but he makes a similar point concerning Aristotle.

INTELLECTUAL APPREHENSION

183

to emphasise the following two aspects of this second point. On the one hand, the acquisition of the universal is described as a natural psychological process, not as an inference. One the other hand, Aristotle’s approach is not normative; he is not concerned with the question of how generalisations can be justified. Thirdly, Aristotle’s explanation of a cognitive process in which we acquire starting points for science50 and arts from perception and memory involves a distinction between having experience and having a universal at rest in the soul. In fact he does not explain here what kind of difference that difference is from a psychological point of view. The most important parallel passage for our passage here is Metaphysics I 1, 981a5–17, where Aristotle distinguishes between experience on the one hand, and art and science on the other. Experience is there characterised as cognition of particulars. I can have experience that this cure has helped Callias and Socrates in this illness. This makes me capable of curing them both when they have that illness. However, I will lack knowledge of the general class of people who will be helped by the cure, be it, e.g., phlegmatic or bilious. Art and science are distinguished from experience because they involve that knowledge. However, this distinction, as also McKirahan for instance notes,51 is in danger of making experience collapse with memory. Experience in the passage quoted above is distinguished from memory on the basis of being a single thing consisting of many memories. Even though experience does not involve explicit cognition of the fact that, e.g., the cure helps both Callias and Socrates because they have something in common, e.g. being bilious, it probably contains tacit or unarticulated and unnoticed cognition of that fact, and this aspect distinguishes it from mere memory. People having experience, as Aristotle points out, are often better when it comes to actually curing people. Therefore, they seem to be able to make the relevant distinction (e.g. between bilious and phlegmatic patients) even though they have no explicit knowledge of what the distinction is.52 One noteworthy aspect remains. When talking about experience in the Posterior Analytics passage, Aristotle distinguishes it sharply from reason. Experience is for him basically a cognitive capacity of the animal soul and, therefore, it must not presuppose reason.53 The acquisition of reason is explained through experience. However, in the passage of the Metaphysics, 50

We have discussed in the previous chapter, pp. 102–110, how the account is related to Aristotle’s scientific proofs. 51 McKirahan (1992, 241–242). 52 My account of this passage is in essence the same as that found in McKirahan (ibid.); cf. also Ross (1953 I, 116). 53 For animal cognitive functions in Aristotle, see Sorabji (1993).

184

CHAPTER TWO

Aristotle is clearly talking about experience of adult human beings, who also have reason. Therefore, it is not so clear whether we can use the Metaphysics passage as our guide concerning the Posterior Analytics passage. I think that, ultimately, we can, if we keep the following in mind. Experience in non-human animals needs to consist wholly of unarticulated generalisation because they do not have a similar rational apparatus to include even the thought that this cure has helped Callias and Socrates. Animals do recognise, e.g., people from each other even though they do not have such thoughts. In the Aristotelian framework reason must originally be developed from such unarticulated or tacit generalisation in human beings. This, I think, is what Aristotle is talking about in the Posterior Analytics II 19. On the other hand, in the Metaphysics Aristotle is talking about essentially the same cognitive process and how rational creatures, namely adult human beings, can use their experience together with the rational capacity. It might now seem that the difference between having mere experience as some non-human animals do and having also the universals in the soul is a difference between not having and having language. However, there is quite clear evidence that Aristotle considers thinking independent of language. For instance, in De interpretatione (16a7–9), he distinguishes between things, affections in the soul (/ah3,ara r‚p vsu‚p) and language.54 The former two are primarily connected with each other; things cause changes in our souls and these changes are the same in all people. Naming things in language is an additional, conventional aspect of our having those affections in the soul. De Anima III. Let us now take a look at how Aristotle accounts for the acquisition of the intellectual disposition in the De Anima. Also there Aristotle points out that the actualisation of that disposition requires the actualisation of intelligible objects in the intellect. That is, reason without content is mere potentiality; to have reason in a full sense is for it to have contents. In Aristotle’s framework, it is one thing to acquire the contents in the intellect and another thing to use or apply them in thought. Acquisition of contents is a transition from a first potentiality to first actuality, whereas starting to think is one from first actuality to second actuality (DA II 5). In the De Anima, Aristotle compares intellectual apprehension to perception. He says: If intellectual apprehension (lne‹l) is like perceiving, it must be either a process in which the soul is acted upon by what is capable of being apprehended (lngr5l), 54

Cf. also the distinction between fallacies due to language and not due to language in the Sophistical Refutations. Schreiber, in his monograph on that treatise (2003), notes that Aristotle has the idea that thinking is independent of language. He condemns the idea as dubious (2003, 52).

INTELLECTUAL APPREHENSION

185

or a process different from but analogous to that. The intellectual part of the soul must therefore be, while impassible (8/ah3p), capable of receiving the form (e∆dnp) of an object; that is, it must be potentially identical in character with its object without being the object. Intellect (lnflp) must be related to what is intelligible (lngr5l), as sense (r• aåqhgrij5l) is to what is sensible (aåqhgr5l). (DA III 4, 429a13–17; cf. III 7, 431b6–8.)

Here Aristotle makes it clear that, analogously to our capacity to perceive, also our intellectual capacity is, before its first actualisation, passive or receptive. The acquisition of intellectual contents requires that the intelligible objects have an effect on our intellect. Aristotle’s argument in De Anima III 4 makes clear that he does not take intellect to be literally affected by the intelligible objects. However, something analogous takes place. ProbablyAristotle refers to a distinction he makes in II 5 (417b3–7).There he distinguishes being affected as being changed, where one quality is extinguished by another, and being affected as perfection, where such extinction does not take place but a natural capacity starts to perform its characteristic activity. Reception of intelligible objects would, then, be analysed as the latter kind of change. When our intellect comes to grasp the intelligible objects for the first time, it does not undergo change in the sense of losing some property – it did not have any actual properties before that – but its capacity to think or grasp is activated. Only when our intellect has received the objects from outside for the first time is it capable of initiating thinking on its own (DA III 4, 429a29–b8).55 Therefore, before we are capable of thinking, we need to acquire contents in our intellect. Aristotle explains this to be a process which is very much like being affected by the perceptible objects in perception. In a famous analogy, Aristotle compares the human intellect before it starts to receive the intelligible objects to an empty writing tablet ‘on which as yet nothing stands written’ (DA III 4, 430a1–2). He makes clear that our intellect is initially like a pure potency (DA III 4, 429a21–22; 24); it does not have a nature of its own. Aristotle provides arguments for this, and a highly influential one is the argument that if the intellect had a nature, this nature would prevent it from thinking. But because everything can be thought, intellect cannot have a nature. 55

Aristotle literally says that after having received some contents in the intellect, the intellect is capable of thinking of itself (d£ a∫r5l), which seems difficult. Accordingly, an alternative reading (di’ a∫rnfl) is suggested by Ross and Bywater in their edition (Oxford Classical Texts) to make sure that the phrase means ‘thinking on its own initiative’. In general, I think that textual emandation should be done only if it is absolutely necessary. This does not seem to be such a case. Rather, in the Aristotelian framework, when the intellect thinks, it is identical with the object. Therefore, it makes sense to say that when the intellect thinks it thinks of itself.

186

CHAPTER TWO

The background for this argument is Aristotle’s view that intellect thinks by becoming identical in form with the object. Basically, the argument can be taken in two ways.56 The first reading of the argument would be the following. Given that we grasp intelligible objects so that our intellect becomes identical in form with them, our intellect could not grasp itself because it could never become what it already is. Another way to take the argument would be to say that the intellect cannot have a nature of its own, because this nature would prevent it from becoming completely identical in form with something else. In this second reading, the basic idea of the argument is to say that the intellect can only become completely identical in form with X if it itself lacks form. This is because if the intellect had a form, say Y, then it could not become X entirely; it could at best become YX. Aristotle does not make it explicit, but it seems to me that the second reading of the argument is the intended one. In fact, the argument is taken that way both by Alexander (De Anima, 84, 15–17)57 and by the author of the treatise on the intellect in the supplement to Alexander’s De Anima, which is possibly unauthentic (106, 18–113, 24). In this supplement, often called Mantissa, the argument is specified as follows. If the intellect had a form, ‘its own nature … would prevent it from grasping external things (r‡l %jr•p 8lr4kgvip), because it would constitute an obstacle (%,/5dinl) to grasping those things’ (106, 28–29). The author might be aware that the argument can be understood in two ways. In any case, he goes on to give further grounds for taking the argument in the sense he does. He says (106, 31–107, 1) that sight, being the capacity to discern colours, has a colourless organ – for water is colourless – in which sight resides and through which the discernment of colours occurs, and similarly in the case of smell and touch (107, 1–4). He concludes (107, 7–9) that likewise the intellect cannot be any one of the objects it is supposed to grasp. The idea of the analogy with sense perception is apparently the following. If we were to perceive colour Z through sight and our sense organ had a colour U, we could not perceive Z but only ZU. Aristotle has an example (DA II 10, 422b8–10) where he talks about a sick man tasting wine as bitter because his tongue is overflowing with bitter moisture. In this example it would not work to say that if our tongue has a tastable quality U, it could not become U because it already is such. Rather, Aristotle’s example is that when

56

Distinguished, e.g., by Calvin Normore in a conference presentation at the University of Victoria, Canada, October 2004. 57 Alexander says that the intellect’s own form would prevent it from grasping what is different (jwk6ei r¢l rnfl 8kknr.4ns k‚vil 84, 16).

INTELLECTUAL APPREHENSION

187

a healthy person tastes wine, he will find it sweet, but a sick person tastes the same wine as bitter because of the bitter moisture on his tongue. If the argument were of the form that what already is something cannot become what it is, the example should go as follows. The sick person cannot taste bitter wine, because tasting bitterness involves becoming bitter and his tongue is already bitter. But this is not what Aristotle says in the example. The intellect, of course, is different, because it does not have an organ of its own. Still, it seems to me that the argument is intended in the sense that the nature of the intellect would prevent it from grasping other things. Aristotle does not connect this example with the argument of the intellect and the evidence from the Mantissa is not conclusive. In any case, the argument is not our main topic here, and we can move on. Independently of how we take this argument, it carries along the following important assumption: the intellect comes to think of objects by becoming identical with them (see, DA III 5, 430a14–15; 430a20 and III 6, 431a1–2).58 In a sense we can say that, for Aristotle, the objects of our intellection are external things. For instance, if I think of a stone, the object of my thought is the stone out there. However, when saying that the object is actualised in the intellect, Aristotle clarifies that he does not mean the object in the same sense as it exists out there. The stone as a composite entity is not in my intellect when I think about it; only its form is actualised there (DA III 8, 431b29–432a1). The form is the intelligible aspect (lngr5l) of the stone. Even though Aristotle claims that the intellect does not have a material nature of its own, he probably thinks that it is somehow located in, or near, the perceptual centre of the human soul in the chest. In addition, in spite of the intelligible form being immaterial as well (being the structuring principle of a thing), the form cannot act upon our intellect from a distance just by being actualised there in the thing outside. Rather, the thing needs to have material impact on our soul so that the intelligible form will be transmitted there as well. It is likely that this happens in perception. However, the lower cognitive capacities – perception and memory, found in animals as well – are unable to operate with the intelligible forms. It seems that in human beings who have the intellectual capacity, perceptions, memory and experience work in the way that they finally lead to actualisations of the forms in the intellect as well. We discussed above the question of what kind of mental change is it to change from having mere experience to having a universal in the soul. From

58

Cf. D. Frede (1992, 288).

188

CHAPTER TWO

the metaphysical point of view, Aristotle makes the difference clear. The intelligible form is not actualised in the animal soul in the relevant manner, whereas in human beings there is a change from mere experience without reason to intellectual apprehension; such change takes place when the form is received in the intellect. The discussion concerning the De Anima and the passive or receptive nature of the intellect can now be connected with the last chapter of the Posterior Analytics. Even though the analysis of the intellect as a pure potency becoming identical with the intelligible aspect of things is not present there, the following line points to a connection: ‘the soul is so constituted that it is capable of undergoing this’ (y d£ vsu¢ ∫/1.uei rnia6rg nœqa n˙a d6laqhai /1queil rnflrn, An. Post. II 19, 100a13–14). Here Aristotle refers to the idea that the transition from having mere experience to having intelligible objects actualised in the soul – and thus having reason – is something that takes place in the human soul rather automatically because of the soul’s natural constitution. Our soul undergoes the change from a non-rational entity to a rational one; it is something that happens to us. It is a remarkable assumption of Aristotle’s that our intellect is basically a receptive or passive capacity. We observed above that it is a basic intuition behind Aristotle’s theory that the reliability of cognitive capacities is explained by their passivity or receptiveness. This holds both for the perceptual and the intellectual faculty. The idea that our intellectual potency is receptive is supposed to explain why and how we can have an accurate conception of the units of the intelligible structure of the world. There is no intervening activity of the intellect; the same form that is the organising principle of an external thing is actualised in the intellect. It is clear, however, that Aristotle does not think that the passive aspect is the only aspect of our intellectual capacity; he admits that we can use these passively or receptively acquired basic contents in ordinary thought. But such thought involving combination and distinction always involves the possibility of error. I shall return to this point in the next section. What are the Basic Contents of Intellect? So far I have mainly concentrated on Aristotle’s account of how our intellectual capacity becomes activated. Aristotle does not provide a proper systematic discussion of what the basic contents are; the main source concerning the contents are his examples, and they are few. Those in Posterior Analytics II 19 are fairly simple: a species and a genus. The evidence from the De Anima is more complex. I shall discuss that evidence after first treating the Posterior Analytics. In contemporary scholarly literature Aristotle’s view of the objects of intellectual apprehension has caused some controversy. The debate is concerned

INTELLECTUAL APPREHENSION

189

with the question of whether such apprehension is propositional or not.59 The answer is, of course, dependent on what we take propositionality to be.60 If we think that mastery of any single concept is propositional in the derivative sense that its mastery enables us to utter some correct statements about the object, then we have to say that all thinking acts are propositional. However, this issue can be dealt with in more detail. It is useful to make the following distinction explicit from the very beginning. lne‹l, as is well known, is a Greek verb that in Aristotle can refer to ordinary thinking in general (as, e.g., in the beginning of his discussion of the thinking faculty in the DA III 4). However, it also has the very technical meaning of intellectual apprehension of indivisible intelligible objects. I agree with Lloyd and Sorabji that it would be strange to say that ordinary thinking could be non-propositional. In fact, I do not think we should attribute the doctrine that there is non-propositional ordinary thought to any philosopher discussed in this book. However, it is far less obvious that intellectual apprehension could not be non-propositional. We can now move to the more detailed points concerning the issue of nonpropositionality. As I have pointed out above, even though the beginning of the Posterior Analytics II 19 makes one expect that Aristotle is going to explain how the premises of scientific proofs become known, this is not in a straightforward sense what he does in the chapter. The account Aristotle gives there concentrates on what we would characterise as natural concept formation. When enough perceptions are stored in the memory and they have become a unified non-intellectual experience, this leads to the activation of our reason. The notions of, e.g., human being and animal are formed in the mind. Therefore, Aristotle is mainly speaking of how we acquire the elements which make up our rational thoughts. Consequently, he is not talking mainly about propositions. Aristotle’s account of how our intellectual capacity is activated cannot accurately be described as our coming to know some definite propositions. What happens when the form becomes actualised is not that some propositions are ‘downloaded’ into our intellect. Rather, having the form in the intellect equips us with some abilities we did not have before. Even though it is 59

The controversy mainly focuses at Plotinus. The problem has been put forward by Lloyd (1969–70) whose analysis, however, is strange. Non-propositional thought is considered a mistaken or interrupted form of propositional thinking. Sorabji (1982; 1983) has argued that non-propositional thought is a myth; ancient philosophers did not assume that non-propositional thought would be possible. 60 E.g. Sorabji (1993) also takes animal perception to be propositional, because we human beings can formulate in propositions what animals perceive.

190

CHAPTER TWO

difficult to pinpoint the exact difference in the abilities we have by virtue of mere experience and those we have by virtue of our reason, the basic point is that the actualisation of the form should not be taken to involve the fact that we merely come to know a group of propositions. If one expects an explanation of how the premises of proofs become known Aristotle’s examples come as a disappointment. The only kind of propositional principles we can extract from the Posterior Analytics is of the form ‘the human being is an animal’. Such claims in the Aristotelian science appear as conclusions of proofs expressing the definition of a human being. However, even though emphasising that Aristotle is not mainly talking about propositional starting points in the Posterior Analytics II 19, I do not want to say that he is not talking about propositional starting points at all. On the contrary, some kind of propositional structure is formed in our mind in that very process. By the example – and also in his ‘improved’ explanation (100a15–b5) following the passage quoted above – Aristotle indicates that when we acquire notions of a natural kind, they come to be in our mind so that they are related with other notions. Expressing these relations is propositional (human beings are animals) and, hence, the account concerns propositions in a derivative sense. This leads us now to the evidence from the De Anima. There Aristotle quite explicitly makes the distinction between non-propositional and propositional objects of the intellect. The former are also called ‘primary notions’. The distinction is made in connection with the remark that only propositional thought involves both truth and falsity. In a sense, nonpropositional intellectual apprehension can be said to be true, but falsity does not apply to it. Consider, for instance, De Anima III 6, where Aristotle says: ‘what is true or false involves a synthesis of objects of intellect’ (%l n˚p d£ ja§ r• vefldnp ja§ r• 8kgh2p, q6lheq4p rip }dg lng,1rwl). And further: ‘where the alternative of true and false applies, there we always find a sort of combining of objects of intellect (l5g,a) in a quasi-unity (πq/e. *l ≈lrwl)’ (DA III 6, 430a27–29). Such combining is contrasted with grasping primary objects of the intellect (/.‡rnl l5g,a) (DA III 8, 432a12). Only combined objects of intellect are propositional; as Aristotle notes, they have a truth value. We have already referred to the passage in the Metaphysics where Aristotle makes a similar distinction. There he talks explicitly about primary, non-composite objects of intellect (Met. IX 10, 1052a1–2; cf. DA III 6, 430b26–30). The alternative for grasping them is not error, but ignorance. There is no such thing as grasping them erroneously, but we can fail to grasp them. Such failure is described as lack of contact. Therefore, Aristotle allows primary objects of the intellect that are not propositional.

INTELLECTUAL APPREHENSION

191

However, it is difficult to formulate with accuracy what Aristotle takes the primary objects of the intellect to be because his examples are problematic. When discussing thinking of the indivisibles (y r‡l 8diai.2rwl l5gqip, 430a26), which he elsewhere calls ‘primary objects of intellect’ (/.‡ra ln3,ara, 432a12), ‘incomposites’ (8q6lhera, Met. IX 10, 1051b17) and ‘undifferentiated’ (8di1tn.a, An. Post. II 19, 100a16), Aristotle gives examples such as ‘incommensurate’ and ‘diagonal’. None of them is likely to be indivisible in the sense required by Aristotle. ‘Incommensurate’ is a complicated notion, and so is ‘diagonal’. More likely objects would be the infimae species, probably at least in the first two categories, i.e., substance and quality, possibly also in the category of quantity.61 Examples of these would be the human species in the category of substance, white or black in the category of quality, and one in the category of quantity. Of these examples, white is mentioned in DA III 6 (430b2) as a primary object of intellect. In addition, Aristotle says that black is grasped as privation (DA III 6, 430b23). This might entail that its opposite, white, can be taken to be a primary object. ‘Human being’ appears in Posterior Analytics II 19. One is not mentioned, and of my examples it is the most speculative. There are other examples of the category of quantity, such as the line and the point, but neither of them is said to be an indivisible object of the intellect. On the contrary, a line is indivisible only actually; potentially it is divisible into the elements of its definition, such as a point. And a point is grasped as a privation, probably of dimensions (see, DA III 6, 430b6–14). Evil is classified as being grasped through privation (DA III 6, 430b22–23). Saying this might entail that the opposite, the good, is an indivisible intelligible object. Also generic notions, such as ‘animal’ mentioned in An. Post. II 19 (100b3), are indivisible only actually;62 potentially they are divisible into species. Even the infimae species, which constitute the most likely candidate for indivisible intelligible objects are potentially divisible in the sense of being divisible into their instances. However, because they are only divisible into particulars, they can be taken to be the ‘smallest’ universal objects and hence indivisible. As we noted, some modern scholars have found it implausible that thinking of the indivisibles would involve a non-propositional object. For this reason, it has been suggested that thinking of the indivisibles is the propositional

61

Modrak (1990–91, 762) also suggests that the infimae species are the most likely objects. 62 In this case the translation ‘undivided’ would be a better one; cf. Berti (1978, 144 n. 22) and Hamlyn (1993, 142).

192

CHAPTER TWO

thought of a definition.63 If what I have suggested above is correct, then actualisations of single intelligible objects in the intellect would not involve thinking of a definition. Thought of a definition requires that the properties necessitated by the form are articulated in statements. The definition is the outcome of systematic inquiry, not the initial intellectual grasp of a species at the beginning of inquiry. The initial intellectual grasp of a species as a species would primarily imply only the ability to refer to a species as a kind in thought. Probably we often learn the generic notion at the same time and come to grasp their relation. However, the initial objects are undifferentiated (8di1tn.a, An. Post. II 19, 100a16): they do not include the specific differences. Sometimes also the genus might be problematic; Aristotle might have thought of examples like the starfish. Is it a plant or an animal? Although not involving explicit articulation of the specific differences, our initial intellectual grasp of an object has something resembling a predicational structure. According to Aristotle, when intellectual thinking occurs, there is always also a t1lraq,a present in the perceptive faculty (see, e.g., DA III 7, 431a14–17 and 431b2). Aristotle describes this metaphorically: the thinking faculty thinks by means of talr1q,ara ‘just as if it were seeing’ (πq/e. ….‡l, 431b7). We might think that even though our initial grasp of human beings does not require explicit knowledge of what properties are essential to human beings, at least some of those properties are somehow present to us, namely in the talr1q,ara. Like the notion of an indivisible intelligible object, the relation between definitions and simple intelligible objects is also difficult in Aristotle. Consider the following problem. Aristotle thinks that all the infimae species have essences, which are expressed in definitions. The elements of essence, genus and the specific differences, belong to the definiens by virtue of its own nature (in Aristotle’s terms jah’ a∫r5). This suggests that the constituents of a definition are contained in the definiens. However, when talking about the grasping of indivisibles, Aristotle assumes that the primary objects of intellect are not only actually but also potentially indivisible. This seems to entail that the elements of the definition cannot be contained in the definiens but have to consist of the subject’s relations to other indivisible universals. Therefore, it would seem that in fact what belongs to a subject per se (jah’ 63

Sorabji has presented this opinion (1982, 1983) according to which thinking of indivisibles is propositional, because its objects are definitions. Also Lloyd (1969–70), though accepting non-propositionality, finds the existence of the contemplation of single essences implausible and calls the non-propositional thought ‘an enigma of Greek philosophy’; cf. Berti (1978), who takes the objects to be single essences. See also Wedin (1988, 125–131).

INTELLECTUAL APPREHENSION

193

a∫r5) can be grasped only in the subject’s relation to other species within the same genus. As has already been said, there are important metaphysical assumptions behind Aristotle’s theory. (Similar assumptions can also be traced in Plato and in the later Platonic-Aristotelian tradition.) Particularly important is the assumption according to which there is a kind of correspondence, not only between true propositions and facts, but also between the simple objects of the human intellect and the things or kinds of things in the external world. In Aristotle’s theory this correspondence is articulated in the doctrine that it is the same intelligible form, which is actualised in the external things as a metaphysical factor determining the necessary properties of things of that kind on the one hand and in the human intellect on the other. Aristotle takes the intelligible forms as objects of basic intellectual acts to be universal. However, in recent scholarship on the seventh book of Metaphysics it has become increasingly evident that we have to take seriously the possibility that in the Metaphysics Aristotle abandons the assumption of universal forms and considers that the forms as substances are particular.64 According to a particularist reading, Aristotle completely abandons the idea of the reality of universals and takes our talk about species to be just a manner of speaking of which there are no ontological consequences. The question of the interpretation of Metaphysics VII is much too large to be dealt with here. It is useful to note, however, that some tension seems to remain in Aristotle’s discussion in the Metaphysics VII. From the point of view of existence, particular forms seem a much stronger candidate for being substances. From the point of view of knowledge and definition, by contrast, universal forms seem to be prior to particulars. Active Intellect, a Note. In the Aristotelian account of intellection there is a highly problematic question of the status of the so-called active intellect. The question is related to one rather unclear passage in Aristotle, namely the fifth chapter of book III of the De Anima. The term translated as ‘active intellect’ (lnflp /nigrij5p) does not appear as such in the relevant passage. The passive intellect (/ahgrij•p lnflp) is explicitly mentioned (430a24–25); the active counterpart is present as a pronoun (rnflrn, 430a24). The passage on the active intellect is very short and it does not in fact have many explicit connections with the rest of the De Anima. I shall not discuss the long-standing debate concerning Aristotle’s doctrine of the active intellect here. However, the theme is prominent in the later commentary tradition and we shall return to it in connection with the commentaries below. 64

Frede and Patzig 1988; for the discussion, see Wedin (2000).

194

CHAPTER TWO

2.3.2 Later Developments Next we shall move to discuss some later developments in the PlatonicAristotelian tradition. First we shall briefly consider later Platonism and then turn to some of the ancient Greek commentaries on Aristotle. Galen, Alcinous, Plotinus In his treatise on logic (Institutio logica),65 Galen discusses basic logical concepts and introduces the valid inference forms. However, he also occasionally comments on issues we would classify as epistemological or psychological. On the one hand, he makes a distinction between things that are evident and things known through demonstration, and comments briefly on how nondemonstrated evident things are known (Inst. log. 1.1). On the other hand, Galen also gives an account of how some general logical principles become known. Finally, he also explains how terms and concepts underlying them are acquired. Galen does not explicitly comment on his remarks concerning the epistemological and psychological issues and their relation to the discussion of logic. However, the fact that he makes these remarks indicates that he assumes that his treatise on logic is not only concerned with validity as an abstract notion. Valid forms of inference are interesting particularly because they can be used in proving things – particularly proving something about the natures of things (cf. Inst. log. 1.5). This kind of proving implies that the premises have to be known. Galen’s basic realistic assumption seems to be along the lines similar to what we have found in Plato and Aristotle. For the premises to be known in the sense that something can be established about the nature of things through them, they have to be about existing things. The premises are about things only if they consist of certain kinds of terms. These terms must correspond to a special kind of concept, namely those which are formed in the human mind naturally in interaction with existing things.

65

The text of this treatise was discovered as late as 1844. There are two editions; one from 1844 (by Mynas) and another, the Teubner edition, by Karl Kalbfleisch (1896). The manuscript has been damaged, and there has been discussion about the restorations. Mynas’ suggestions have been very much criticised (see, e.g., Kalbfleisch’s preface). Also the authenticity of the work has been questioned, but the current opinion among scholars seems to follow Kalbfleisch’s article from 1897, where it is argued that Galen wrote it. On the discussion, see, e.g., Kieffer (1964, 3–5).

INTELLECTUAL APPREHENSION

195

The two aspects, the knowledge of general logical principles and the account of concepts are involved in an aside Galen makes after first interrupting the discussion concerning basic forms of premises to comment on the copula. As a kind of footnote to the comments on the copula he says: [W]hen we have memories of perceptible things (aåqhgr•l /.Øc,a), whenever we call them up in connection with motions, as of Athenians (çral ,£l <jar¡> jil3qeip ra6rap /nigqÍ,eha jah1/e. eå r6uni ’Ahgla4wl), let this action be called ‘thought’ (¬ln,af2qhw rnflrn y,‹l l5gqip), but when they are at rest, ‘conception’ (çral d£ yqsu1fnsqai r6uwqil, (llniai); there are also other such conceptions, not derived from memory of perceptions, but existing naturally in all men ((,tsrni /Øqil ∫/1.unsqai), and the ancient philosophers call them, when they are expressed in language, ‘axiom’ (jaknflqi d£ a√r¡p nß /akain§ r‡l tiknq5twl, çral ^.,gle6wlrai di¡ twl‚p, 8m4w,a); often, however, the Greeks call conception, ‘thought’ (/nkk1jip ,2lrni ja§ r¢l (llnial l5gqil ¬ln,1fnsqil nß ]Ekkglep). (Inst. log. 3.2; transl. Kieffer (1964); for the text, see Kalbfleisch (1896).)

The text is slightly damaged here and it begins quite abruptly. Some important points can, nonetheless, be perceived. Firstly, in this little excursion Galen provides us with some evidence for how the theories of the principles of argumentation are connected to the discussions of philosophical psychology. Secondly, he makes a distinction between concepts formed through perception and memory and with some, presumably logical, principles the knowledge of which is not based on experience. Galen notes that the conceptions not formed through experience were called ‘axioms’ by philosophers of the past. It is highly likely that he uses the term in a similar way as, for instance, Aristotle does. In chapter 1 section 5 Galen gives an example: ‘Things equal to the same thing are also equal to one another’. This particular principle does not appear in Aristotle’s presentation of axioms in the Posterior Analytics I. However, Aristotle’s example ‘if equals are subtracted from equals the remainders are equal’ (I 10, 76a41–b1) is close enough to indicate that Galen means the same kind of principles: some very basic arithmetical or logical truths. Both examples also belong to Euclid’s common notions (jnila§ (llniai), roughly corresponding to what Aristotle calls ‘axioms’. The axioms, Galen says, belong naturally to all human beings. What, then, does Galen mean when he says that the axioms belong naturally, or are innate – as the crucial Greek word (,tsrnp can also be translated? He makes it clear that the axioms are not acquired from experience. However, it is not equally clear whether they are innate in the sense that they are necessary for the formation of concepts.

196

CHAPTER TWO

Galen does not give any very detailed account of how the concepts are formed in the soul through perception and memory.66 However, the little he says refers to the kind of explanation we have already encountered in Aristotle.67 His example of the perceptible things the concepts of which he is talking about is Athenians. Even though the example is somewhat nonstandard, it seems that there are some basic realist assumptions involved. There are perceptible things which belong to various natural classes and we have general notions corresponding to those classes. There is no special reason to say that Galen would not accept examples like ‘human being’ and ‘animal’ as concepts formed on the basis of experience. However, because Galen does not endorse the theory of intelligible forms, he does not assume that our natural concepts should correspond to some basic intelligible objects in reality. Nonetheless, he talks about the natures of things and refers to factors which make, e.g., human beings the kind of creatures they are (see, e.g., Meth. med. 2.7, 3–4). Therefore, it is probable that he would claim that at least some of our general concepts have a metaphysical basis. Alcinous – as indicated above – might have been Galen’s temporal predecessor, but he is philosophically closer to Neo-Platonism. In Alcinous’ treatise ‘on the criterion’, he discusses under a Hellenistic heading the question of how we can arrive at truth.68 In that connection he makes a distinction between intellectual acts proper (l5gqip) and ordinary thinking which Alcinous take to involve ‘natural conceptions’ (tsqija§ (llniai). A crucial passage for the distinction is the following. Intellection (l5gqip) is the activity of the intellect as it contemplates the primary objects of intellection. There seem to be two forms of this, the one prior to the soul’s coming to be in this body, when it is contemplating by itself the objects of intellection, the other after it has been installed in this body. Of these, the former, that which existed before the soul came to be in body, is called intellection (l5gqip) in the strict sense, while, once it has come to be in the body, what was 66

Galen’s use of the verb ‘to come at rest’ (yqsu1feil) brings to mind a similar use of the verb x.e,e‹l in Aristotle’s Posterior Analytics II 19 (100a6). Galen’s contrast between the stability of concepts ((llniai) and moving thoughts, however, is probably different from Aristotle’s distinction between experience and having universals in the soul. Galen probably means the same as Plutarch (de Sollertia Animalium 961c). Plutarch refers to the difference between stored concepts not currently appearing in our thoughts and concepts used in the thoughts we have at a given moment; cf. Kieffer (1964, 72). 67 Rather similar accounts for the preconceptions also appear in the Stoics and the Epicureans. 68 A discussion of the passage is found in Sedley (1996).

INTELLECTUAL APPREHENSION

197

then called intellection is now called ‘natural conception’ (tsqij¢ (llnia), being, as it were an intellection stored up in the soul. (Didaskalikos 4; 155, 21–27; transl. Dillon 1993.)

Alcinous denies the possibility of our having intellectual acts in the proper sense in the incarnated life. During our bodily existence, we can only have some kinds of memories of these thinking acts and these memories are called natural conceptions (tsqij¢ (llnia). The expression tsqij¢ (llnia also appears in Stoic sources. The Stoics’ natural concepts are also called ‘preconceptions’ which are genuinely acquired from experience. Several scholars have suggested that Alcinous’ natural concepts correspond to Stoic preconceptions.69 However, it is not quite clear how this suggestion could be fitted to Alcinous’ remark that the natural conceptions are some kinds of memories of the discarnated intellectual acts. As Dillon remarks70 we should, in describing Alcinous’ natural conceptions, emphasise the Platonic aspects of the theory. Alcinous does not explain in more detail how he thinks the natural conceptions are acquired, but he does say that in order to have such natural conceptions we must have had intellectual thinking acts before birth into our body. Also his reference to the Phaedrus at the end of the paragraph (Didaskalikos 4; 155, 30–34) points to a Platonic rather than a Stoic reading.71 Be this as it may, Alcinous distinguishes between pre-bodily intellectual thinking (l5gqip) and ordinary thinking, and claims that the former can never be attained in incarnated life. Alcinous’ view that intellectual acts in the strict sense have to be distinguished from the memories we have of them in our bodily life brings along with it a distinction between the objects of these acts. He claims that the objects of intellection proper (l5gqip) have to be different from the objects of our bodily thought. The former has as its object metaphysical forms which 69

The suggestion is made by Schrenk (1991) and accepted by Sorabji (1993, see especially 73) as well as Gerson (1999, 73). Gerson claims that Alcinous assumes that natural concepts are acquired from experience, because he uses the term ‘comprehension’ (/e.4kgvip) in connection with them and this is the term which Alexander of Aphrodisias uses of empirical generalisation (see, e.g. De Anima 83, 11–19 and in Metaph. 166, 8–13). 70 Dillon (1993, 67–68). 71 Sedley (1996), however, considers the picture to be more complicated. He refers to Alcinous’ distinction between epistemonic and doxastic reason (k5cnp) (154, 25–32) and suggests that the natural concepts belong only to the former. Therefore, it seems to follow that our ordinary belief and opinion formation should involve still some other concepts belonging only to the level of the doxastic reason. Cf. however (155, 3–7).

198

CHAPTER TWO

are separate from the perceptible world and are concepts or thoughts in the divine mind (chapters 9–10 of the Didaskalikos).72 When in the bodily life we remember these acts and grasp perceptible instances as imitating those forms, our acts do have forms as their object. However, they are not the separate forms which are only conceivable by a discarnate soul. They are directed at lower forms, forms in matter actualised in the perceptible instances.73 Plotinus deviates from Alcinous in thinking that proper intellectual acts are possible also during our incarnated life. Plotinus thinks that the intelligible objects are eternally in the metaphysical Intellect, which thinks of them all the time. On the level of the metaphysical Intellect the contact with forms is immediate to such a high degree that it is even said that differences between the act of thought and its object vanish.74 Even though the metaphysical Intellect is not part of our soul – it belongs to a higher level of being – we do have mental contact with it, and are occasionally capable of taking part in the eternal intellectual activity. In the following passage Plotinus discusses the question of how, for instance, beautiful things can provoke an apprehension of beauty in us: We must consider him [i.e. a musical type of person] as easily moved and excited by beauty, but not quite capable of being moved by absolute beauty; he is however quick to respond to its images when he comes upon them, and just as nervous people react readily to noises, so does he to articulate sounds and the beauty in them; and he always avoids what is inharmonious and not a unity in songs and verses and seeks eagerly after what is rhythmical and shapely. So in leading him on, these sounds and rhythms and forms perceived by the senses must be made the starting point. He must be led and taught to make abstraction of the material element in them (uw.4fnlra r¢l ÷kgl) and come to the principles from which their proportions and ordering forces derive and to the beauty, which is in these principles, and he must have the doctrines of philosophy implanted in him (k5cnsp rnºp tiknqnt4ap %lher2nl); by these he must be brought to firm confidence in what he possesses without knowing it. (Enneads, 1.3.1, 22–35; Armstrong’s translation.)

For the present purposes we can note the following. Our responsiveness to, for instance, beauty in perceptible things functions as a starting point from which we can proceed to a more abstract understanding of beauty. Therefore, 72

It has been suggested that this interpretation goes back to Plato, more specifically to Republic VI and Timaeus 29a; e and 39e; see Dillon (1993, 67). For the interpretation of Platonic forms as thoughts of the divine mind, see further Whittaker (1990), Rich (1954), Armstrong (1960) and Dillon (1977). 73 Gerson (1999, n. 12, p. 73) points out that Alcinous marks this distinction by calling the forms in matter forms (e∆dnp), whereas the separate forms are called ideas (åd2a). 74 See Emilsson (1996). Hence it has been characterised as ‘non-intentional cognition’, see O’Meara (2001).

INTELLECTUAL APPREHENSION

199

perceptions function as an initiator of our cognitive development towards an intellectual understanding of the form of beauty and its role in a more complex network of intelligible forms as organising principles of reality. Therefore, on the one hand, Plotinus claims that some preliminary grasp of beauty in perceptible beautiful things initiates a cognitive process, which leads to the mastery of abstract concepts, e.g., beauty. On the other hand, it is an important assumption in his Neo-Platonism that mastery of abstract concepts presupposes that we are related to the metaphysical Intellect in which the forms are apprehended in their complex interconnections eternally and from which the structure and existence of the perceptible world originates. Even though in the Plotinian framework all ordinary propositional thinking by means of concepts (di1lnia) is dependent upon and secondary to the existence and activity of the metaphysical Intellect, in human life ordinary thinking temporarily precedes the occasional moments of intellection proper. In addition, even though the possibility of normal conceptual and propositional thinking is explained by some kind of connection we have with the metaphysical intellect, the two are distinct. Intellectual acts are never a necessary outcome of propositional thinking. However, ordinary thinking and making conceptual distinctions can lead us towards grasping parts of the world’s intelligible structure as a whole at a single glance. Our concepts can promote intellectual apprehension because in a sense both the concepts and the perceptible objects themselves are imitations of, and hence resemble, the forms (1.2.3, 27–30). Alexander, Themistius, Philoponus It is a common feature of Aristotle’s ancient Greek commentators that they start to operate with a distinction between different intellects or different aspects of intellect. The main distinction is that between the so-called active and passive intellect. For these aspects such terms are used as … /nigrij•p lnflp, which can be translated as ‘active intellect’ or ‘intellect in actuality’, … lnflp … %le.ce4ˇ, which can be translated as ‘productive intellect’ and, further, … dsl1,ei lnflp (potential intellect) and … /ahgrij•p lnflp (passive intellect or passible intellect).75 The expression … /nigrij•p lnflp is never used by Aristotle as such, and the expression … /ahgrij•p lnflp appears only once at the end of chapter 5 of the third book of the De Anima (430a23–25).76 From Alexander 75

For the terminology, see, e.g., Blumenthal (1991). For interpretations of the passive intellect, see Blumenthal (1991). Some commentators identify the passive intellect with talraq4a (reported by Proclus, see, e.g., Philoponus in De Anima 6, 1–2). In the Latin commentary, possibly by Philoponus, however, the author denies that talraq4a could be identified with the passive intellect (see in De An. CLCAG Verbeke p. 13, lines 00–6). 76

200

CHAPTER TWO

of Aphrodisias onwards, however, the expression … /nigrij•p lnflp is in frequent use. Alexander. The relevant text material concerning Alexander in this section is mainly found in his own treatise on the soul (De Anima), which is based on Aristotelian theory but is not a commentary. Alexander also wrote a commentary on Aristotle’s De Anima but it is lost, and we only have quotations from it in other commentators.77 Alexander is well known for identifying the active intellect with God, i.e., with the first cause and the unmoved mover of the universe (De Anima 89, 9–19). Alexander is with Aristotle in claiming that when we human beings are born, we do not have an intellectual disposition ready to be used when we wish, but we only have a potentiality for receiving such a disposition (81, 13–15). Alexander calls this potential of ours ‘material intellect’ (∫kij•p lnflp), because, he says, it is characterised by a potentiality to become something and that which becomes something serves in a sense as matter to that which it becomes (81, 22–28). Therefore, it is not meant as the claim that the intellect is a material entity. Like Aristotle, Alexander also compares our potential intellect to an empty writing tablet on which nothing stands written (84, 24–25).78 However, he makes the analogy even more specific by pointing out that our potential intellect should not be taken as the tablet itself, because the tablet suffers material change and is affected when letters are written on it. Rather, our intellect is like the tablet’s ability to receive writing on it, its suitability for being written on. His idea is that, contrary to the tablet, this capacity is not affected in the strict sense by the letters written on the tablet, but is perfected or fulfilled (De Anima 84, 25–85, 5). Alexander also follows Aristotle in arguing that the potential intellect is without a nature, because having a nature would prevent it from grasping other things (84, 15–17). He is quite clear in taking the argument in the second way distinguished above – that is in the sense that the intellect’s nature would somehow prevent it from becoming completely identical with any other object in form. According to Alexander, the intellect’s own nature would prevent it from grasping something alien (8kk5r.inl) to it. Now, given that the material intellect is a receptive capacity, we can ask what it receives. Two different possibilities suggest themselves. First, we could take it as receptive of intelligible objects (lngr1). Second, it is possible that our potential intellect is not receptive of intelligible objects but of the 77 78

For these quotations, see Moraux (1942). Cf. Aristotle DA III 4, 429b31–430a2.

INTELLECTUAL APPREHENSION

201

intellectual disposition ()mip). Both of these alternatives have advocates among scholars. To assume that our material intellect receives intelligible objects leads to a problem which has been recognised for a long time.79 The problem arises because of the following assumptions. Alexander says – as Aristotle once does – that enmattered things are potentially intelligible (dsl1,ei lngr1, 87, 29). He says explicitly that the intellect – the context being now a discussion of the human intellect, not the divine active intellect80 – has to make perceptible things intelligible to itself by separating or abstracting (uw.4feil) them from their material realisation, and thus turning them into pure objects of thought (84, 20–21).81 But now it seems extremely problematic to claim that the human potential intellect would do this. How could a mere pure potency of reception perform the activity of separating the form from its material circumstances or conditions? To say that it does would be like saying that the tablet’s capacity to receive writing would somehow make letters appear on the tablet (or such that they can appear on the tablet), which seems absurd.82 If we say that our potential intellect is receptive of the intellectual disposition, not intelligible objects, leaves us with the question of how we received the intellectual disposition in the first place. It has occurred to various scholars that this problem is a possible application of the active intellect. The suggestion is often coupled with the claim that the treatise on intellect in the Mantissa – a supplement to Alexander’s De Anima – was also written by Alexander, and that the treatise solves this problem by referring to the active intellect.83 I shall not be able to discuss the Mantissa here. However, it has also been suggested that the perplexity concerning the De Anima can be solved without reference to the active intellect. Schroeder has argued against Moraux and Bazán that the Greek word d6la,ip, which is crucial in Alexander’s description of the human potential or material intellect, should not be taken to indicate passivity.84 Rather, it should 79

For the problem, see, e.g., Moraux (1942) and Bazán (1973). Moraux claims that because of this problem, Alexander’s theory is contradictory; see Moraux (1942, 132–142). This claim has been accepted by Bazán, but denied by Schroeder (1982). 80 Cf. Sharples (1987, 1207). 81 Cf. 87, 24–29, where Alexander says that the intellect makes it intelligible to itself by abstraction of what it is for the composite entities (r¡ q6lhera) to be the kinds of things they are. 82 Cf. Moraux (1942, 75). 83 See, e.g., Bazán (1973) 472; cf. Moraux (1967) despite his original rejection of this view. 84 Schroeder (1982).

202

CHAPTER TWO

be understood as a power to perform the act of separation of intelligible objects. It is undeniable that d6la,ip also means power. It is also clear that Alexander is careful in specifying that the human potential intellect is not strictly speaking affected by anything. However, I do not agree with Schroeder in the analysis of the latter claim for the following reason. According to Schroeder, Alexander’s main reason for saying that the material intellect is not affected is in order to argue that the acquisition of intelligible objects is not passive or receptive, but is active separation of the objects from their material circumstances. But Alexander’s explanation is not that the material intellect is not passive because it is active. Rather, Alexander says that only actual things can be affected in the strict sense of the word. The potential intellect is not an actual thing because it has no form before taking on the objects; being no actual thing, it cannot be affected (n√d£l /1quei, 85, 3) in this sense. Therefore, Alexander denies that the intellect is affected in the literal sense, because this kind of affection involves a change in an actual thing. However, he is not denying that the human potential intellect is receptive. Indeed, when arguing that the human potential intellect is receptive of all objects precisely because it is a pure potentiality, he says quite clearly that the material intellect is a capacity for receiving the forms (%/irgdei5rgp … /.•p r¢l r‡l eåd‡l ∫/ndnu3l, 84, 24–25). This strongly suggests that Alexander considers that the forms are received. The question still remains whether abstraction can be understood as reception. There are also references to the active intellect in the De Anima. Therefore, it is possible that Alexander solved the previously articulated problem by referring to the active intellect in that very treatise. In a sourcebook on the commentaries, Richard Sorabji in fact suggests that in the De Anima Alexander introduces the idea that the active intellect turns our potential intellect into dispositional intellect, i.e. such that we can think on our own initiative.85 His suggestion includes two components: (i) we are able to abstract concepts from perceptible objects when they are present to us, (ii) we are not capable of thinking on our own initiative when the objects are absent until the active intellect ‘gives the material intellect its proper disposition’.86 I agree with Sorabji that Alexander’s text strongly suggests that the human potential intellect is capable of abstracting concepts (ln3,ara) from perceptible objects (see pp. 84–85 of the De Anima). Alexander is also explicit in saying that the active intellect is a reason for the potential intellect’s becoming dispositional. However, I do not quite agree on the claim (ii) just mentioned that the active intellect should give us our proper disposition to think – as 85 86

Sorabji (2004, 104). Ibid.

INTELLECTUAL APPREHENSION

203

distinguished from the ordinary concept formation taking place in the potential intellect. Firstly, the crucial evidence for the claim that the active intellect gives us our disposition to think comes from the following passage: And since there is a material intellect of a sort (ja§ %/e4 %qril ∫kij5p rip lnflp), there must also be a productive [or active, /nigrij5p] intellect, which comes to be the reason for the disposition of the potential intellect (Ωp a©rinp r‚p )mewp r‚p rnfl ∫kijnfl lnfl c4lerai) (88, 23–24; my translation).

One reason why I think that this does not necessarily involve the claim that the active intellect gives us our disposition to think, is related to how we read the a©rinp. I cannot enter into discussing this in detail, but several scholars have pointed out that the unmoved mover as the a©rinp of human thinking in Aristotle is not to be understood as a cause in any straightforward sense.87 Therefore, the evidence does not necessitate the reading that the active intellect should directly bring about the coming to be of the intellectual disposition in us. Rather, the passage can be taken in a weaker sense to point out that the active intellect is a general explanatory principle for intellectual thinking. It is of course possible that Sorabji’s formulation is intended rather vaguely and it could be taken to be consistent with what I have said here. However, another, perhaps more important reason for why I am not completely happy with the suggestion that the coming to be of the human dispositional intellect is explained by a reference to the active intellect, is that there is evidence in Alexander’s De Anima for another explanation of how we acquire that disposition. Before returning to that evidence, let us, however, see how Alexander distinguishes between merely having concepts and intellectual thinking (lnflp). Initially this kind of disposition (y rni1de )mip) arises in the intellect through a transition from continuous activity related to the perceptible objects into obtaining a kind of theoretical vision concerning the universal (πq/e. ≈vil ril¡ 8/’ a√r‡l ka,b1lnlrnp rnfl jah5kns hew.grij3l), which is at first called a concept (l5g,a) or a notion ((llnia). When we have more such experiences and the notions become variegated and manifold so that it is possible to grasp the objects without the underlying perceptual conditions (®p d6laqhai ja§ uw.§p r‚p aåqhgrij‚p ∫/nb1h.ap /nie‹l rnflrn), it is intellectual thinking (lnflp). (De Anima 85, 20–25; my translation.)

Now the question is how we attain the disposition to think in the proper sense as opposed to merely having some concepts. The evidence that Alexander does answer this question without referring to the active intellect comes from 87

M. Frede (1996b), Caston (1999b). For more general accounts of causes or reasons in antiquity, see, e.g., Vlastos (1969), Frede (1987c) and Hankinson (1998a).

204

CHAPTER TWO

the beginning of the section of the treatise concerning the thinking faculties. There Alexander makes is quite clear that the proper disposition of the intellectual capacity comes about through habituation and instruction (di¡ … didaqjak4ap re ja§ %h‡l, 81, 25). This, he says, cannot be acquired by all human beings, but only by the most intelligent ones. Therefore, he does explain how the intellectual disposition is supposed to be acquired: through using and habituating our cognitive abilities as well as receiving teaching. In that passage (81, 22–82, 15) Alexander makes a further distinction between (a) common (jnil5p) intellect belonging to all as an elementary conceptual ability and (b) proper intellectual disposition including advanced theoretical understanding. This distinction is not followed systematically in the treatise. However, it matches quite well with what we found from the passage (85, 20–25) quoted a little while ago. Both passages pay attention to the fact that all human beings have a basic ability to master concepts, but not all of them grasp complex theoretical wholes concerning the nature of things and the explanations of natural phenomena. The latter involves theoretical or scientific knowledge (%/iqr3,g) and comes about in those who are gifted through instruction and habituation, that is, through education and research. We are now in a position to return to the problem articulated by Moraux in Alexander’s doctrine. The problem is how the human material intellect, which is characterised as pure receptivity, can manage to perform the abstraction of concepts from perception. I do not propose to solve this problem, but suggest that we look at it from another angle. When distinguishing between the ‘common intellect’ and the dispositional intellect at the beginning of his discussion of the various aspects of intellect, Alexander indicates that the common intellect, i.e. mastering some basic concepts, is acquired without any special effort by human beings. To explain the distinction, he compares intellectual capacities with our ability to learn to walk (82, 5–19). All human beings are able to have some concepts and thus a minimal share in intellectual activity because of our natural ability to have concepts. This is analogous to the fact that all of us – who are not seriously disabled – learn to walk without any special effort. By contrast, Alexander says, we do not come to have comprehensive theoretical knowledge (%/iqr3,g) so easily. Indeed, the acquisition of that kind of knowledge requires active work. From this perspective the receptivity of our material intellect can be taken to refer to the manner in which we learn the basic concepts. We come to do this naturally as children and, therefore, such basic concept formation simply happens in us. This is why it can be characterised as reception. Therefore, from the point of view of calculating actualities and potentialities and explaining change on the basis of basic Aristotelian notions, Alexander’s doctrine in the De Anima is problematic in the sense that the potential intellect

INTELLECTUAL APPREHENSION

205

is said to abstract concepts. However, there is a clear and plausible explanation of how we acquire the intellectual disposition, which is distinguished from merely having abstracted some concepts. Alexander claims that being capable of intellectual thinking requires that we master complex theoretical wholes and also know some basic principles of sciences. Learning them is not automatic reception like abstracting some basic concepts; it does not happen to us as naturally as we learn to walk. Rather, to learn theoretical principles and grasp them as explanatory requires education and research. Themistius. Of Themistius’ paraphrases on Aristotle’s works the one on the De Anima is the most ambitious and it almost amounts to a worked out commentary.88 His general account of the soul is fairly Aristotelian.89 Themistius endorses the distinction between animal and intellectual parts of the soul and – in a typical late ancient fashion – he makes a distinction within the intellectual soul between potential intellect and intellect in actuality.90 Themistius follows Aristotle in saying that potential intellect has no actual nature of its own: ‘the intellect is potentially all objects of thought (lngr5l), yet it is in actuality (%lrekeue4ˇ) nothing until it thinks’ (in De Anima 97, 19–20 ad III 4; Todd’s translation 1996; cf. 94, 18–20). The reason, according to Themistius, for potential intellect being without a nature of its own is the same as is found in Aristotle and Alexander. If intellect had an actual nature of its own, this nature would somehow restrict the scope of the objects it can think of. Because the intellect is in no such way restricted, ‘it is naturally disposed to grasp all [forms]’ (jaraka,b1leil #/alra /2tsjel, 94, 26–27), it cannot be of any definite nature. In his paraphrasis of the Posterior Analytics Themistius repeats Aristotle in saying that we achieve conceptual structure from experience. He says: This seems to belong to all animals: they [all] have a natural capacity to make distinctions (d6la,ip q6,tsrnp j.irij3), which is called perception ({l jaknfl,el a©qhgqil). For some animals, when they have perceptions (%ln6qgp d’ aåqh3qewp), there remains a trace (,nl3) of the percept (a©qhg,a) [in their soul] and they for

88

For a short introduction to the paraphrasis, see Todd (1996); cf. also Todd (1990). For Themistius’ definition of soul, see Blumenthal (1990b). 90 According to Themistius, the potential intellect (lnflp dsl1,ei) is not the same as the passive intellect (… /ahgrij•p lnflp). Themistius talks about a common intellect, which is perishable, passive, mixed with and inseparable from the body. The potential intellect, by contrast, is unaffected, unmixed with the body and separate; it is characterised as a forerunner of the intellect in actuality (see in De Anima 105, 26–30). I find this distinction somewhat difficult to follow. It is not quite clear to me what aspects of human thinking the distinction is supposed to capture. 89

206

CHAPTER TWO the most part preserve [the percept] from what they have perceived; for some animals no [such trace] remains … Such animals, for which the percepts do not remain in the soul, do not have knowledge outside perceptions (rn6rnip n√j (qri cl‡qip (mw r‡l aåqh3qewl). For others, it is possible to have the percept of the perceptible (rnfl aåqhgrnfl … r• a©qhg,a) in two ways: [i] as when Socrates has gone away, the imprint (r6/np) and form (,n.t3) [remains]; for these there is another kind of knowledge (cl‡qip), [i.e.] memory (,l3,g), but these are still less perfect, simply capable of [keeping] the imprint of many perceptions; [ii] for others it is possible to join and fit together (r¡ ç,nia qsl1/reil … ja§ %la.,5feil) the similars; and those for whom there is contemplation (hew.4a) and separating (di1j.iqip) of similars are already perfect (r2keia) and have reason (kncij1).91 On the basis of perception memory is formed, as we said, and on the basis of the memory of something which happened many times, experience arises; for many memories are in number one experience.92 That this is purifying hellebore he or she perceived and remembered, but on another occasion that this is hellebore and purifying, he or she again perceived and remembered, and [this happens again and again case after case]. Of many perceptions a single experience is gathered of hellebore being purifying. And when this experience concerning perception and memory of similars grows and increases ('p qslasmaln,2lgp re ja§ /.nqka,baln6qgp a©qhgqil …,n4al ja§ ,l3,gl), the universal [conception] becomes firm and remains in the soul (/3clsrai }dg r• jah5kns ja§ %,,2lei r· vsu·), [namely] that all hellebore is purifying. And the universal is the similar and the same in the particular and that which is in many ((qri r• jah5kns r• ç,ninl ja§ ra√r•l %l rn‹p jah’ )jaqrnl ja§ r• %l rn‹p /nkkn‹p), and this is the principle of art and knowledge (rnflrn r2ulgp 8.u¢ ja§ %/iqr3,gp); [it is the principle of art] if it is concerned with that which art produces (eå /e.§ r‡l ∫/’ a√r‚p ciln,2lwl), [and it is the principle] of knowledge if it is concerned with that which comes into existence by nature (eå /e.§ r‡l ∫/• t6qewp). (in An. Post. 63, 2–26 ad II 19; my translation.)

Themistius’ account is very similar to Aristotle’s, but there is one notable difference. Whereas Aristotle mentions as examples of the universals we acquire from experience species and genera (‘human being’ and ‘animal’), Themistius’ example is propositional and contains explicit quantification (all hellebore is purifying). The example enables Themistius to draw a clear borderline between having experience and having a universal in the soul. Experience involves a non-universal generalisation ‘hellebore is purifying’, and a universal generalisation ‘all hellebore is purifying’. The example is clear but it is slightly problematic with respect to the claim that non-human animals can also have experience. Perhaps Themistius is thinking of experience in the sense of a rational creature’s experience.

91

This latter group of rational animals contains only human beings. This sentence in Themistius’ Greek is almost a literal quotation from Aristotle An. Post. II 19, 100a3–6. 92

INTELLECTUAL APPREHENSION

207

However, the examples are also similar in an important respect: both could be seen as a conclusion of a scientific proof. For Themistius it would be an argument proving that hellebore is purifying through some explanatory middle term or terms. Similarly, if we connect the general terms Aristotle mentions into a statement, we get ‘human beings are animals’, or ‘all human beings are animals’. Such a claim could also appear as a conclusion in a scientific proof, where the specific differences function as middle terms. In his paraphrasis of the Posterior Analytics Themistius follows Aristotle also in not mentioning the distinction between potential intellect and intellect in actuality, or any other distinction of different aspects of the intellect or different intellects. In his paraphrasis of the De Anima, however, he uses these distinctions and presents the theory according to which the potential intellect is perfected by an intellect in actuality. He says: The potential intellect (… dsl1,ei lnflp) must be perfected by some other intellect that is already perfect, i.e. actual, not potential … And this intellect is separate, unaffected, and unmixed … this actual intellect (… lnflp … … %le.ce4ˇ) advances the potential intellect, and not only makes it an actual intellect, but also constitutes its objects of thought as actual objects. These are the enmattered forms, i.e. the universal concepts assembled from particular objects of perception. (in De Anima 98, 29–99, 4)93

How, then, does the intellect in actuality94 perfect the potential intellect? Themistius’ analogy is not surprising. The intellect in actuality is said to advance the potential intellect ‘as light when supervening on potential sight and potential colours produces both actual sight and actual colours’ (98, 36–99, 1).95 However, Themistius’ account differs from Alexander’s in the important respect that according to Themistius both the potential intellect and the intellect in actuality are in us (see, 102, 30ff.). Themistius criticises Alexander for placing the intellect in actuality or the productive intellect outside the human soul. This is, according to Themistius, problematic because Aristotle himself in De Anima III 5 talks about potentiality and actuality in human soul. 93

Here I have quoted Todd’s translation (1996) with the exception that Todd translates the word l5g,a as ‘a thought’. I prefer the translation ‘concept’ not because it would be completely unproblematic but because it refers to the possibility that the basic intellectual acts might not be propositional. 94 In similar contexts Themistius also calls the intellect perfecting the potential intellect productive intellect (… lnflp … /nigrij5p, e.g. 99, 11–12). 95 Themistius (in De Anima 95, 17–18) also compares intellectual acts to seeing; the same comparison appears in Aristotle (DA III 7, 431b7) and Alexander (in De Anima 85, 14–22).

208

CHAPTER TWO

In spite of the analogy between light and intellect in actuality, in Themistius’ account it is left quite unclear how exactly he assumes that the perfection of the potential intellect by the intellect in actuality happens. In his paraphrasis of the last chapter of the Posterior Analytics Themistius rejects the possibility of unnoticed principles in the soul and he also says that the potential intellect is in us prior in time (99, 30–31). Therefore, the analogy of illumination cannot refer to any kind of recollection of known but forgotten general notions or truths. It is not impossible that Themistius assumes that the intellect in actuality, although being in the human soul, does not have to be in the same individual as the potential intellect which is perfected by it. According to this assumption, later explicitly expressed by the Latin Philoponus (on the Intellect, 48, 32–33), teachers have, at least to some extent, perfected their intellects and they can probably also perfect the potential intellects in their pupils. Evidence for the possibility that Themistius also attributes a role to teacher’s intellect is found in the following. Similarly with bodies of knowledge, the teacher’s objects of thought are identical to those of the learner (… did1qjwl r+ ,alh1lnlri r¡ a√r¡ lne‹); for there would not even be any teaching and learning unless the concepts (l5g,a) possessed by teacher and learner were identical. And if, as is necessary, [that concept] is identical, then clearly the teacher also has an intellect identical to that of the learner, given that in the case of the intellect its essence is identical with its activity. (in De Anima, 104, 7–11)96

Themistius, however, also has another description of how the potential intellect is perfected by the intellect in actuality. He says that before the actual intellect has made the potential intellect into an intellect in actuality and the potentially intelligible objects into actually intelligible objects, the potential intellect cannot distinguish (diaj.‹lai) between different universal concepts (ln3,ara), cannot combine (n√d£ qslrih2lai) or divide (diai.e‹l) them and it cannot make transitions from one concept to another (%m ^r2.wl eåp )re.a ,eri2lai) (99, 5–6). In this state the potential intellect is compared to a storehouse of concepts (hgqas.•p lng,1rwl). Only after the productive intellect 96

Here I have also deviated from Todd’s translation of l5g,a as ‘thought’ and translated it as ‘concept’. In this passage, the idea clearly is that the intelligible objects, i.e. the forms, are the same for teachers and learners. Because our intellectual capacity is conceived of as a capacity to become identical with these objects, the intellects in the moment of thinking these objects have to be identical as well. However, it is not necessary that the teachers and the learners think exactly the same thoughts, but the thoughts have to consist of the same elements. For problems related to ‘concept’, see above p. 156.

INTELLECTUAL APPREHENSION

209

(… lnflp … /nigrij5p)97 has encountered (%/2khnlrnp) the potential intellect, is the potential intellect capable of making transitions, combinations and divisions between concepts (99, 9–10). This passage emphasises the passivity of the potential intellect. It is taken to be incapable of thinking on its own without an impetus from an actual intelligence. However, even though we took Themistius quite literally, the matter is left unclear. If we do have intellectual contents in our potential intellect, but we cannot think on our own initiative before an encounter of the actual intellect, this leaves open the question how we got the contents of our potential intellect in the first place. According to Themistius, the intellect in actuality is something that is shared by all human beings (104, 2–3). A possible meaning of this is that we all have at least some concepts which are the same in all of us (cf. Arist. De Int. 1, 16a6–7). Themistius points out that we could not understand each other unless our concepts (l5g,a) were the same (103, 36–104, 11). Here Themistius builds on basic Aristotelian assumptions. The intelligible objects, i.e., the forms, are the same for all and the forms are actualised in our intellect. When the form is actualised in our intellect, the intellect becomes identical with its object and is, before actualising the form, without a nature of its own. Therefore, the forms actualised in the intellect as elementary thought contents, are the same for all. Themistius thinks that our ability to grasp common notions (jnila§ (llniai) and first axioms (/.‡ra 8miÍ,ara) is also due to the alleged fact that we all somehow take part in one single intellect in actuality (103, 38–104, 3).98 He does not explain, however, what the common notions and first axioms are and what it means that they come from the intellect in actuality. It is possible that Themistius means both by ‘common notions’ and ‘first axioms’ some set of logical-arithmetical principles which all human beings would take as true when hearing them. To say that they also come from the intellect in actuality perhaps means that we cannot apply such principles before the intellect in actuality has made us able to think on our own initiative. 97

Here Themistius changes terms; up to this point he has been talking about an actual intellect (… lnflp … %le.ce4ˇ). It is difficult to say, whether the switch of terms have consequences as to the doctrine presented. In 99,11–12 he uses the term ‘productive intellect’ (… /nigrij•p lnflp) as an alternative to the ‘intellect in actuality’ (… lnflp … %le.ce4ˇ) whereas in 103, 37–38 he seems to mean different things by these two terms. It is not entirely clear what the difference is supposed to be. 98 Here Themistius assumes that intellect is composed of a potential intellect and an intellect in actuality, but he does not specify whether he thinks that we take part in a single intellect in actuality. This is probably his meaning.

210

CHAPTER TWO

Three Commentaries under Philoponus’ Name. After Themistius we shall consider four Greek commentaries traditionally attributed to Philoponus on Aristotle about how our intellect becomes activated. The authenticity of all of them has been questioned.99 One of these commentaries has come down to us only as a Latin translation made by William of Moerbeke in 1268 (probably for Thomas Aquinas). It is on Aristotle’s discussion of the intellect (DA III 4–8), and it is the one modern scholars would most readily but not unanimously accept as expressing the historical Philoponus’ views.100 I shall call the author of this commentary the Latin Philoponus without committing myself to the view that the commentary was actually written by the historical Philoponus. Of the two other commentaries appearing under Philoponus’ name one is on the Posterior Analytics (CAG XIII 3), and I shall call the author Greek Philoponus A.101 The other one is on the De Anima (CAG XV); I shall call the author of this commentary Greek Philoponus B.102 In the commentary on the Posterior Analytics, Greek Philoponus A expresses the Aristotelian view according to which a basic conceptual structure is acquired from experience. He says: For those in whom there is the trace of the perceived when they have perceived … in addition to there still being the trace in them [it is also possible for them] to have something which is one in the soul (qºl r+ e∆lai %l a√rn‹p r¢l ,nl¢l (ri ja§ (ueil %l r· vsu· )l ri), i.e. a capacity to join together the similar percepts, of which they had perception (}cnsl d6la,4l rila %/iqsl1/rnsqal r¡ ç,nia r‡l aåqhg,1rwl ‘l (qunl r¢l a©qhgqil). For instance, in the case of men (%/§ 8lh.Í/wl), I have once seen (e∆dnl) that Socrates has purified his bile by drinking hellebore, and I have also seen this in the case of Plato. These percepts were printed in the phantasia (raflra r¡ aåqh3,ara r· talraq4ˇ %lers/Íhgqal). Then occasionally coming across hellebore (,er¡ /a.ad.n,3l rilnp jai.nfl ådÅl ^kk2bn.nl) I have been able to join this together with cases in which I know on the basis of similarity that also this purifies the bile (%dsl3hgl %/iqsl1vai rnflrn n˚p e∆dnl di¡ r¢l …,ni5rgra ja§ %lreflhel cl‡lai çri ja§ n∑rnp jelwrij•p unk‚p %qril). (in An. Post. 435, 2–10; my translation.) 99

For the commentaries, see Sorabji (1990a). For the discussion, see Charlton (1991, 4), who identifies the author as Philopous. A Greek commentary on the De Anima going by the name of Philoponus printed in the CAG XV is taken to be authentic as to the commentaries on book I and II but not on book III. My concern is here on the commentaries on book III. 101 The views presented in the commentaries bearing Philoponus’ name are not compatible with each other and the authors do not explain differences in the views presented. It is likely that the commentary on book II of the Posterior Analytics is not by Philoponus. 102 The author is identified as Stephanus by, e.g., Hayduck in his edition, Hayduck’s arguments are criticised by Charlton (1991, 4–12). 100

INTELLECTUAL APPREHENSION

211

And further: [O]n the basis of many memories experience and knowledge has been formed in me (c2cnl2 ,ni %,/ei.4a ja§ cl‡qip) that the hellebore has the power of purifying the bile (®p … ^kk2bn.np d6la,il (uei jelwrij¢l unk‚p). When this knowledge has become still (x.e,3qaqa), fixed (%,/ace‹qa) and stable (^d.aiwhe‹qa) in my soul that the hellebore is like this and not otherwise, the universal has come together (qsl‚mel), namely ‘all hellebore is purifying’, and the universal is the principle of proofs (Ω jah5kns %qr§l 8.u¢ r‡l 8/nde4mewl). (in An. Post. 435, 23–27 ad II 19; my translation.)

These quotations are, again, highly similar to the account in Posterior Analytics II 19. However, Greek Philoponus A also uses the same example that we find in Themistius. Perhaps it was already a standard example before Themistius. In his commentary Greek Philoponus A mentions the possibility of latent knowledge in us, but he takes this to be impossible. He says: ‘If we had these [dispositions] in the soul, but did not know that we have them, this would be out of place’ (eå ,£l d¢ (un,el a√r¡p %l r· vsu· ja§ 8clnnfl,el çri ra6rap (un,el, !rn/nl) (in An. Post. 433, 15–16).103 The explanation for this is also adopted from Aristotle; it refers to levels of exactness (8j.4beia) of cognitive dispositions (see 433, 16–21). As a conclusion the author says that it is not possible that the principles (8.ua4) are innate (q6,tsrnp), but neither is it possible to learn them without having them, and therefore there has to be, I quote: a capacity of the soul (d6la,ip vsuij3), which provides us with a starting point for the knowledge of them [i.e. the principles of proofs]. This capacity is not more appreciated (ri,iÍre.np) and exact (8j.ib2qre.np) than the one concerning the principles (433, 31–34; my translation).

It is, of course, the capacity for sense perception (434, 3). Along with Greek Philoponus A, Greek Philoponus B also rejects the Platonising interpretation of Aristotle, which he attributes to Plutarch (in De Anima CAG XV, 520, 2–3).104 In this connection, Greek Philoponus B criticises Alexander, too, whom he understands as saying that the intellect in children is only potentially existent: Plato says that the children’s intellect is habitual (jah’ )mil) and that they have definitions (k5cni) of things, but [this is] not [accepted by]Aristotle. Plutarch thinks that Aristotle, too, is saying the same thing. But how would he not talk falsely when Aristotle refutes Plato? He [Aristotle] says children’s intellect is like a writing tablet not yet inscribed upon, because it is suitable for receiving the definitions (k5cnp) of things, which, however, are not yet received. This is why both Plutarch and Alexander are mistaken together and separately. Let us tell the truth. Let it be fixed (åqr2nl) that we believe that the intellect is always a form (8e§ r•l lnfll e∆dnp e∆lai). 103 104

Cf. Arist. An. Post. II 19, 99b26. For the rejection of the Platonising interpretation, see also 523, 23–26.

212

CHAPTER TWO For the intellect of children is the form, and is potentially capable of receiving forms (dsl1,ei … dejrij•p r‡l eåd‡l). By being a form it is not potentially [existent] (n√j (qri dsl1,ei) as it seems to Alexander; by being potentially capable of receiving forms it does not possess the definitions of things, as Plutarch thinks. Let us say what is meant by the intellect. First, the intellect means that which potentially knows things (dsl1,ei eådÅp r¡ /.1c,ara), like it is in the case of children. Secondly, the intellect means that which knows things dispositionally (jah’ )mil eådÅp r¡ /.1c,ara), like in the case of those who have perfected their knowledge, but are at rest and hence do not actually use their knowledge. Thirdly, it means the intellect, which does not come from outside (n√u … h6.ahel eåqiÍl), like they think,105 but the one which is a disposition in action (jah’ )mil #,a ja§ %l2.ceial), like in the case of the intellect of [persons] who have perfected [it] and know things (r‡l reke4wl lnflp r‡l eåd5rwl r¡ /.1c,ara) and who have the knowledge at their disposal and act according to it (/.nuei.ifn,2lwl r¢l e©dgqil ja§ %le.cn6lrwl jar’ a√r3l). (in De Anima 520, 1–20 in CAG XV; my translation.)

According to Greek Philoponus B, potential knowledge is transformed into actual knowledge by what is enmattered (r¡ (lska, !cerai c¡. ∫/’ a√r‡l 8/• r‚p dsl1,ei clÍqewp eåp r¢l %le.ce4ˇ cl‡qil, 526, 15–16). The author does not mention the substantive; we will add ‘objects’. He then states that the perfection of the potential intellect by enmattered objects seems to lead to a problem: the intellect is on the one hand affected by the enmattered objects (/1quei ∫/’ a√r‡l), which seems to entail that it is of the same material character (…,5sknp) as they are, i.e., enmattered and corruptible. However, on the other hand the intellect is said to be immaterial and not mixed with the body. (526, 16–20.) The problem is, according to the author, solved by the Aristotelian distinction between destructive affection (/1hnp … tha.rij5l) and perfective affection (/1hnp rekeiwrij5l).106 The intellect is affected in the perfective way by the perceptibles (r¡ a©qhgra, 526, 29–30) and the author says that in fact this perfective affection should not be called affection at all (526, 30–34).107 Therefore, according to Greek Philoponus B, the intellect is not affected in the way that one quality should be replaced by another (e.g. by turning from not-white to white), but in another ‘perfective’ way. This perfective affection – which is not genuine affection at all – is a transition from not performing a natural function to performing it. In the case of the intellect this would mean a transition from not grasping to grasping. 105

Here the plural ‘they’ probably refers to Alexander. The plural can be used for a single commentator, see Cherniss (1971, 76 n. 1) and Todd (1996, 26 n. 22). 106 Cf. Aristotle DA II 5, 417b3–7; discussed above in section 2.2.1. 107 The commentators usually discuss at considerable length the question to what extent the potential intellect can be said to be affected. For Alexander, see his De Anima (86, 6–14; cf. 84, 3–13), and Themistius, in De Anima (94, 5–16, 56, 1–12); cf. also (112, 25–33).

INTELLECTUAL APPREHENSION

213

Greek Philoponus B’s point against Alexander seems to be that he takes Alexander to claim that the intellect is only potentially existent in children. By contrast, he himself claims that the intellect exists (not only potentially but actually), but it only potentially has the forms of things in them. This in fact seems a little unfair towards Alexander. In sum, Greek Philoponus B claims that we come to grasp the intelligible forms of things, because the things affect our potential intellect in a perfective way – and this is not affection in a proper sense. Therefore, no references to divine or even teacher’s intellects are needed in order for us to be able to achieve a basic conceptual structure from perceptual experience. The objects’ intelligibility is sufficient to account for this. By contrast to the Greek Philoponuses, the Latin Philoponus abandons the Aristotelian view of how we come to grasp basic intelligible objects in favour of a Platonic alternative. The Latin Philoponus enters into a lengthy discussion trying to do away with the consequences of Aristotle’s comparison (DA III 4, 430a1–2) of the intellect with an empty writing tablet awaiting inscription on it. He claims that Aristotle in fact means to express a view similar to the Platonic theory of recollection, and tries to make Aristotle identify with the two types of potentiality he distinguishes from each other in the De Anima II 5 (417a21–b7). The Latin Philoponus says: [W]e ought to interpret what Aristotle says here carefully and thoughtfully with regard to his whole thought and to what he says everywhere about the intellect. If we have shown a thousand times over, quoting Aristotelian texts, that he wants the rational soul to be separate and immortal, it is plain that even if he here likens it to an uninscribed thing we write on he does not mean that it has forms in potentiality in the first sense (the sense in which semen is a man in potentiality). But a certain latitude must be recognised in both meanings of ‘potentiality’… . so intellect which is in actuality perfects intellect which is in potentiality and brings it to actuality not by putting into it forms that are not there, but by bringing into light forms which are non-evident and hidden because of the state of swoon which is the effect of birth. And it is this he calls ‘potentiality’ in the first sense. For there is a difference between the geometer who is in a cataleptic swoon though he still possesses [knowledge] dispositionally, or who is asleep, and one who is in none of these conditions but is not exercising the disposition; … The intellect which enters the world of becoming is like a person asleep or delirious. (The Latin translation of On the Intellect 38, 99–40, 43; 40, 35–43; transl. Charlton 1991.)

Therefore, the author thinks that the forms, i.e., the intelligible objects,108 are already there in the potential intellect when we are born, but we do not notice 108

The Latin Philoponus distinguishes between enmattered forms as configurations and forms as substances (65, 64–65). The former are not intelligible as such, but the intellect knows them by the aid of imagination. The ‘abstracted accounts of the enmattered forms intellect understands by itself without imagination’ (77, 83–87).

214

CHAPTER TWO

that we have them because of the disorder of the soul caused by birth into the body. The latent forms have to be illuminated by an intellect in actuality in order for us to notice that we possess them. According to the Latin Philoponus, the intellect in actuality is in the human soul, but not in the soul of the person, whose potential intellect is perfected.109 It is the intellect of the teacher (48, 32–33; see also 45, 53–59) who perfects the potential intellect of the pupil in such a way that it ‘makes the potential intellect receptive of all’ (51, 95–99). He might mean something analogous to the questioning Socrates addresses to the slave-boy in the Meno. He states that Aristotle’s remark that there is always intellectual thinking means that there always are and always have been human beings who have perfected their intellect (51, 11–52, 29) to a certain extent,110 and that they can pass on the capacity of abstract thinking in a way similar to natural reproduction (48, 27–49). When it is said that also the teacher’s intellect was once brought into actuality, one might ask whether there is a first actualisation of an intellect caused by a divine intellect. The author states that the divine creative intellect is not part of the causal chain in the process of perfection of a human intellect. It is a cause at a higher level, like ‘the Sun is said to generate men’ (51, 7–10). Therefore, he seems to conceive the creative intellect as something like a necessary condition of human intellectual thinking. The Latin Philoponus does not say whether he thinks there once was a first human being or how his or her intellect was perfected. The Latin Philoponus is explicit in saying that the senses do not have any positive role in the recollection of intelligible objects. They are only said to remove an impediment (116, 75–80). The impediment to our being aware of the intelligible objects consists of false conceptions due to phantasia (33, 80–85). Propositions are, according to the Latin Philoponus, thought by the intellect as well and it is the intellect which also proves statements such as that there are three figures in categorical syllogisms (98, 35–44).111 The premises of sciences are thought to be derived from the intellect in actuality. The Latin Philoponus says explicitly that the intellect ratiocinates, and that this is why our actual intellect is never perfect in the sense of having everything intelligible presented to it at the same time. The human intellect in actuality is 109

As such it is to be distinguished from a creative intellect (see, 5, 82–90; 5, 99). He denies that those who have attained the highest intellectual virtue possible for human beings would be thinking all the time (47, 96–47, 4). 111 In such thinking the intellect does not, according to the Latin Philoponus, use phantasmata. 110

INTELLECTUAL APPREHENSION

215

always in a way potential, because all the intelligibles that the intellect is not thinking of at the moment are only potentially present to it. (4, 60–65.) Even though the Latin Philoponus thinks that the intellect ratiocinates, i.e. moves from one proposition to another, he also says that the intellect grasps propositions as well as indivisible intelligible objects instantaneously (for indivisible intelligible objects, see, e.g., 73, 44–45; 75, 20–21; for propositions, see 73, 52–53; cf. 72, 38–73, 64). His example of a proposition which is grasped by an instantaneous act of the intellect is the one that the sum of the angles of a triangle is equal to two right angles (in 73, 52–53). Therefore, the Latin Philoponus does not endorse the Neo-Platonic assumption according to which conceptual and propositional ordinary thinking is performed by a capacity different from the intellectual capacity of understanding complex wholes at a single glance. *

*

*

We have now discussed some later developments in the PlatonicAristotelian tradition on the question of how we human beings acquire basic contents into our intellect. I shall now summarise the most important aspects of the discussion above. As representatives of the later Platonism, we have chosen Galen, Alcinous and Plotinus. Galen does not assume that there are intelligible forms, but provides a realistically oriented account of how some general notions are acquired from experience. In addition to general notions from perception, we have some basic innate logical-arithmetical principles in us that are not learnt from perception. Alcinous, for his part, does endorse the theory of intelligible forms, and he also endorses the theory of metempsychosis. According to Alcinous, proper intellectual acts are possible only in our pre-bodily life. During our bodily existence, we can at best have some kinds of memories of them but we are never able to ascend to pure intellectual cognition. Plotinus also distinguishes between intellectual acts proper and ordinary thinking, but considers proper intellectual apprehension possible in the incarnated life. We can abstract concepts from perception and think about things using these concepts but such thinking never amounts to or necessarily produces full intellectual comprehension. Such comprehension occasionally takes place in us when we, instantaneously, grasp a complex theoretical whole as a whole. In the commentary tradition on Aristotle, much attention is paid to questions concerning activity and passivity or actuality and potentiality with respect to the acquisition of intelligible objects. The process is understood in basically Aristotelian terms: it is assumed that there is a passive or receptive aspect in the human intellect, and it needs to be explained how it becomes

216

CHAPTER TWO

activated. Alexander claims that some concepts are abstracted to the human intellect rather automatically but learning complex theoretical wholes, e.g., sciences, requires education and research, and not all human beings grasp the basic principles of the sciences. Intellectual thinking in the proper sense, however, requires understanding complex wholes; thinking with just a small, and not internally organised, collection of concepts, does not count as proper intellectual apprehension. In Themistius, we encounter various versions of Aristotle’s theory. He ends up by claiming that an active intelligence, perhaps understood as a teacher, needs to activate the passive aspect of our intellect in order for us to grasp things on our own. At the end of the chapter I have discussed three commentaries traditionally attributed to Philoponus. Not all of them are authentic; two are in Greek (one on the Posterior Analytics, one on the De Anima) and one is preserved only as a Medieval Latin translation. The Greek commentaries follow Aristotle rather closely; the commentator on the Posterior Analytics mentions the possibility of latent truths or principles in the soul but rejects the interpretation. The ‘Latin Philoponus’, however, argues for a Platonising interpretation of Aristotle. He claims that we have truths in our soul but they are hidden from us because of a shock caused by carnal birth. Perception, he goes on to argue, does not have any positive role in the process in which we rediscover the truths from our soul; they merely remove our false illusions and imaginations. A teacher, who is conceived of as the owner of an active intellect, can affect our potential intellect so that we come to grasp the truths we already have.

PART II ALTERNATIVE APPROACHES

CHAPTER THREE HELLENISTIC PHILOSOPHY

It is appropriate to begin with a brief summary of the first part of the book. Above we have discussed what I call the Platonic-Aristotelian tradition. Philosophers forming part of this tradition share the basic metaphysical assumption according to which reality is intrinsically intelligible and consists of intelligible elements, namely the forms. They also think that the intelligibility of reality entails that we human beings are capable of immediate familiarity with its elements by virtue of our reason. In addition, reality is taken to be ordered according to natural relations of priority between things. Basically, such relations consist of the following. The existence of some things is necessary for other things to exist (e.g. the existence of animals is necessary for the existence of human beings). The natural relations of priority also constitute basic explanatory relations (the earth being in between the sun and the moon explains the occurrence of a lunar eclipse), and as we noted, explanation is used as a metaphysical and not solely an epistemological notion. In the Platonic-Aristotelian tradition, reality is taken to be intelligible; it is taken to have an intrinsic order and we are taken to be capable of immediate familiarity with its elements. This, however, does not mean that the explanatory structure of reality should be somehow evident to us. On the contrary, it is a common assumption among philosophers in the Platonic-Aristotelian tradition that the order in which we come to know things is typically quite the opposite to the order of nature. This means, basically, that we usually first come to know facts the explanatory principles of which then have to be found in systematic inquiry. Therefore, there are two basic kinds of principles: (i) those which we know at first and on the basis of which we can initiate the inquiry (in Aristotle an example is that the planets do not twinkle) and (ii) those which are found in the inquiry and which explain the facts we at first learnt (that planets are near and that celestial bodies that are near do not twinkle). 219

220

CHAPTER THREE

Notice that both of these kinds of principles are general. It is typical in the Aristotelian tradition to assume that we have an immediate familiarity with facts on a general level. This is explained by a natural cognitive process in which we come to have an initial grasp of the forms and the intelligible level of reality. According to the Aristotelians, we attain intelligible objects from perceptual experience, whereas the Platonists think that we could not learn universally applicable generalisations or reach any kind of universals merely on the basis of perceptual experience: we need to have some pre-existing cognitive structures in our minds which are later recollected or which help us to form universal generalisations. Within such a framework, perception is understood as a starting point for knowledge in a very literal sense: it is where our cognitive development starts. A theory of perception is needed to explain how we get information from the world in the first place. In the Platonic-Aristotelian tradition, it is thought that generalisation is a natural cognitive process which metaphysically speaking involves an immediate contact with the intelligible forms. Therefore, an assumption according to which there should be inferential justification for basic generalisations is quite absent from the Platonic-Aristotelian tradition. The general picture of how starting points of knowledge are discussed in Hellenistic philosophy is different. I shall mention five basic differences. Firstly, mainly due to the influence of the Academic sceptics, the question concerning the very possibility of knowledge comes into focus.1 In the Hellenistic context, discussion on this question concentrates on the question of whether there is a criterion of truth. The question of truth becomes relevant particularly with respect to perception. Do our perceptions really tell us the truth about the world? If we perceive something to be the case, can we take that to be the case? And further, can perceptions be used as decisive evidence for the truth-value of some other beliefs? This means that by contrast to the Platonic-Aristotelian tradition, where the discussion of perception is mainly subordinated to questions concerning intelligibility, in the Hellenistic framework the question of the reliability of our perceptions becomes pressing. Perception is no longer understood as 1

This is widely noted among scholars; see, e.g. Brunschwig (1999), Striker (1990, p. 150 in the 1996 reprint). Brunschwig, however, emphasises Plato’s and Aristotle’s familiarity with the sceptical challenge and points out that we should not separate them too strictly from the Hellenistic schools on this point. Note that I by no means intend to say that Plato and Aristotle were not familiar with the sceptical challenge or that they did not react to it. Rather, I want to point to the status of the sceptical challenge in the theories: the Platonic-Aristotelian approach is not guided by an attempt to answer the sceptics.

HELLENISTIC PHILOSOPHY

221

merely instrumental to a further goal of grasping parts of the intelligible structure of reality. Secondly, the very notion of knowledge involved in the Hellenistic discussion is to some extent different from the one the Platonic-Aristotelian tradition concentrates on. For Plato and Aristotle, knowledge in the strict sense (%/iqr3,g) is such that if we know something in the strict sense, it cannot be otherwise. They also formulate knowledge claims with no reference to time. We are said to know, e.g., that justice is good or that planets do not twinkle because they are near; we are not said to know that this is the case at time t0. Therefore, for Plato and Aristotle the objects of knowledge are unchangeable. The assumption that there should be unchangeable objects of knowledge is absent from the Hellenistic discussion. Even the knowledge (%/iqr3,g) the Stoic Sage has, is about the same objects that we perceive; the Sage only knows these things in a different way. Whereas the best kind of ‘knowledge’ non-Sages can have within the Stoic framework is cognition (jar1kgvip), the Sage has knowledge which is perfectly firm. We ordinary fools change our minds and waver in our conceptions, but the Sage’s cognition remains unchanged by any arguments, thought or reasoning.2 Thirdly, the metaphysical background assumptions are not the same. Basically, the difference consists of the fact that, in the Hellenistic framework, the theory of the intelligible forms is abandoned. The Stoics suppose that reality has an intrinsic rational order. However, they do not think that there are general intelligible forms that are the objects of knowledge. The Epicureans, for their part, abandon the whole idea of reality being intrinsically intelligible. They emphasise the role of perception and downgrade that of reason. Fourthly, along with the intensifying sceptical challenge and reactions to it, notions of clarity and self-evidence gain new importance. In the PlatonicAristotelian tradition, we find a distinction – explicit in Aristotle – between what is better known to us and what is better known to nature. What is better known to us could in some sense be characterised as being evident. However, the notion of evidence is not used to indicate immediate justification. Rather, it is taken to be a fact about human psychology that we first learn things that are close to perception. In the Hellenistic discussion, by contrast, the notion of evidence comes to have the connotation of epistemological safety: what is evident must be accepted as true. Fifthly, new assumptions emerge concerning the role of knowledge in human life. Within non-sceptical Hellenistic schools, the discussion of knowledge becomes subordinated to ethics in a new way. In particular, Epicurus emphasises that trying to find natural explanations to phenomena is 2

Cf. the Stoic definition of knowledge found, e.g., in Sextus (Math. 7.151–152).

222

CHAPTER THREE

necessary only to the extent that it prevents maleficent emotions, most importantly, the fear of death (Kuriai doxai 11). However, also for the Stoics the purpose of trying to achieve knowledge is ethical. Wisdom is constituted by the condition that one never assents to a false presentation, particularly in value questions. This does not necessarily require that one should be widely knowledgeable in natural philosophy. Three preliminary remarks need to be made before we enter the discussion. The first one is the most important: we are not as well equipped with evidence about the Stoics as we are with respect to Plato and Aristotle or the ancient Greek commentaries. Our evidence concerning early Stoicism is mostly second hand and often comes from sources critical or even hostile towards the school. Galen and Plutarch are indispensable and they sometimes even quote Chrysippus’ works. Also Cicero’s Academica is a very important source. Diogenes Laertius is rather late, probably from the time when ‘the last Stoic’, namely the emperor Marcus Aurelius, was already dead, which means from at least about half a millennium after the founder Zeno. Sextus Empiricus is not an impartial judge of the debate between sceptical and non-sceptical schools: he is one of the participants. However, Sextus discusses the key questions and is an invaluable source of evidence, though he needs to be treated with care. The early history of the sceptical movement is not well documented either.3 In Epicurus’ case, by contrast, three probably authentic letters have been preserved in Diogenes. My second preliminary remark concerns the aim of my discussion. I shall attempt to give a synoptic picture of the main aspects of the Hellenistic discussion concerning starting points for knowledge. This means that I shall not concentrate on distinguishing between the individual thinkers unless necessary. Thirdly, ancient scepticism is not my primary concern here. This is mainly a study of those philosophers who think there are starting points for knowledge, not of those who doubt it. However, it is necessary to bring some of the main sceptical arguments into the discussion in order to understand better the non-sceptical views. 3.1 IS THERE A STARTING POINT FOR KNOWLEDGE? The Notion of a Criterion of Truth In the Hellenistic debate the question of whether we can attain truth is formulated as a question of whether a criterion of truth (j.ir3.inl r‚p 8kghe4ap) 3

The most debated question concerning the early phases of Greek scepticism is whether the historical Pyrrho himself was a sceptic or not. Apart from the later legend-like stories about Pyrrho, we only have one crucial piece of text referring to Pyrrho’s own words and it is fourth-hand. For the evidence, see, e.g., Bett (1994).

HELLENISTIC PHILOSOPHY

223

exists. j.ir3.inl became a central technical term for all the schools and it was also utilised in the Pyrrhonian sceptical movement.4 The Greek j.ir3.inl is normally left untranslated and ‘criterion’ is used. At first we need to find out how the notion of criterion was understood. I suggest that there were three different ways to understand the notion of criterion, one for each of the main Hellenistic schools: Epicureans, Stoics and the Academic sceptics.5 Therefore, I take it that the Academic sceptics are in fact employing the very notion of a criterion of truth in a different way from the Stoic one. I shall argue for this in more detail below in the section on cognitive impressions. The seminal work on the notion of criterion has been done by Gisela Striker (1974). She points out that j.ir3.inl was probably a philosophical term of art from the very beginning. One philosophical meaning of the term was to refer to the human cognitive faculties, most importantly the senses and reason, as criteria (as, e.g., Epicurus does in Diog. Laert. 10.38 and 51). If, for instance, the senses are called criteria in this sense it means roughly that the senses are a means of discovering or judging the truth. However, this is not the main use of the term and it is not typical of the Stoics.6 The philosophers who claimed that a criterion of truth exists assumed that the criterion must itself be true. This implies that truth is also attributed to objects other than propositions. For instance, perceptions and certain kinds of impressions are classified as criteria and hence as true. The Stoics made the distinction between propositions and impressions and explained that an impression is not strictly speaking true, but only in the sense that its propositional content is true (see Sext. Emp. Math. 8.10; cf. 7.244). According to the Stoics, all the impressions of adult human beings are rational (kncij5p, Diog. Laert. 7.51). This means that they can be expressed in a propositional manner.7 This clarification does not occur in Epicurus. Epicurus’ use of the term ‘criterion’ has connotations with the term ‘standard’ or ‘ruler’ (jalÍl) which was actually the title of Epicurus’ lost treatise 4

Cf. also Striker (1974). Here I take it that the Academic sceptics can be identified as a philosophical school, but the Pyrrhonists do not form a school. However, nothing significant is implied by this. I simply discuss the Academic understanding of the Stoic notion of the criterion of truth here, because it to some extent illuminates the crucial points on both sides. 6 Cf., however, Boethus’ criteria in Diog. Laert. 7.54. 7 Cf. M. Frede (1983, p. 154 in the 1987 reprint). Sextus says (Math. 8.70) that according to the Stoics a rational impression is such that it can be expressed in language (k5c– /a.aqr‚qai). 5

224

CHAPTER THREE

dealing with criteria.8 For Epicurus it is crucial that the criterion of truth resembles a ruler or a straightedge in two respects. Firstly, in order for something to function as a criterion of truth, it must itself be true. Analogously, a straightedge must itself have a straight angle. Secondly, it must be able to function as a kind of tool for deciding the truth of other beliefs. The idea is that as we can judge whether an angle is straight or not by using a straightedge, we can, by comparing other beliefs with the criterion, distinguish whether the other beliefs are similar with respect to truth-value as the criterion (i.e. whether they are true). In addition, the criterion needs to be such that its truth need not be decided on the basis of some further criterion, because this would involve an infinite regress. Our sources mention three different criteria as Epicurean, namely perception (a©qhgqip), preconception (/.5kgvip) and feeling (/1hnp) (Diog. Laert. 10.31; cf. Cicero Acad. 2.46, 142). Of these the first two are listed as criteria of truth; feelings are criteria of choice and avoidance (Diog. Laert. 10.34; cf. Cic. Fin. 1.22–23).9 Given that perceptions are criteria, they must also be true. Epicurus is famous for holding the problematic doctrine that all perceptions are true.10 I shall discuss this doctrine more closely below. In addition to perceptions, the Epicureans took preconceptions to be criteria of truth. At this point it is sufficient to note that the truth of the preconceptions entails two things: (i) the attributes assigned to the things in the preconception really belong to those things; (ii) the things the preconception refers to exist, i.e. the preconceptions are not vacuous.

8

Diogenes Laertius probably had access to information about the treatise; he is taken to be summarising some of its content in 10.31; see Striker (1974, p. 30 in 1996 reprint) and Asmis (1984, 87). 9 It has, however, been suggested by some scholars (e.g. Bailey (1928, 249–250); cf. also Rist (1972, 31)) that feelings can in fact be seen as functioning similarly with perceptions as criteria of truth. Asmis discusses this more closely (1984, 96–99) and suggests that whereas perceptions are criteria of truth concerning external reality, feelings can be seen to function as criteria of truth concerning inner conditions. 10 There are some verbal differences in the evidence concerning this doctrine. (i) Sextus says once that every percept (aåqhgr5l) is true (Math. 8.9) and once (ii) that every appearance (talraq4a) is true (Math. 7.203–204). He also says that (iii) perception (a©qhgqip) tells the truth (di¡ /alr5p re 8kghe6eil) (Math. 8.9; cf. 8.185 ‘the perceptions never lie’ (,gd2/nre vesdn,2lgp r‚p aåqh3qewp)). In Aëtius we find the formulation that every perception (a©qhgqip) and every appearance (talraq4a) is true (see Usener 1887, no. 248). And in Plutarch (Adv. Col. 1109a–b) that every appearance occurring by means of perception (a©qhgqip) is true. The evidence is gathered by Striker (1977); it is presented and discussed also by Taylor (1980).

HELLENISTIC PHILOSOPHY

225

The Stoic use of the term ‘criterion’, as Gisela Striker points out, differed from that found in Epicurus. Whereas for Epicurus the metaphor of a measuring stick is crucial – and implies that a criterion can be used and is used to decide the truth value of some beliefs – the criterion in the Stoic sense is used to establish what is or is not the case. In addition, the Stoics also deviate from Epicurus who thinks that all perceptions are equally powerful or persuasive (åqnqh2leia). The Stoics analysed perceptions by means of the notion of impression (talraq4a).11 They thought that not all impressions can function as criteria, but only a certain subclass of them called ‘cognitive impressions’ (jarakg/rij¢ talraq4a).12 The cognitive impressions are said to be criteria of facts (j.ir3.ia r‚p ∫/1.mewp). Basically this means, for instance, that if I have a cognitive impression that Socrates is in front of my eyes, it is the case that Socrates is in front of my eyes. The doctrine that the cognitive impression is the (or a) criterion of truth is said to be maintained by Chrysippus and Antipater; it was also endorsed by the later Stoic school. Zeno – the founder of the school – was, however, reported to claim that it is not a cognitive impression which is the criterion, but rather cognition (jar1kgvip) (see Math. 7.152; Cic. Acad. 1.41–42).13 However, if we bear in mind that cognition involves assent to a cognitive impression (Sextus Math. 8.397 and 11.182; cf. Pyr. 3.242), the difference does not seem to be very big. To say that cognition, not the cognitive impression, functions as a criterion, might be meant to underline the following aspect of the criterion. Sometimes we fail to assent to a cognitive impression because we have other false beliefs which prevent us from doing this. In such a case, we have a cognitive impression but do not assent to it: we have the cognitive impression but we do not have a true belief. Therefore, in order for the impression to function fully as a criterion, it must be assented to. And assenting to the cognitive impression is the same as having cognition. 11

talraq4a is also sometimes translated as ‘appearance’; for a discussion of these two translations, see Barney (1992). I shall mainly use ‘impression’ and occasionally also ‘appearance’ to indicate that if we have an impression, something appears to be the case. The core meaning to which I refer is in both cases ‘something appears to be the case’. 12 This key term is translated in many different ways; also such translations as ‘apprehensive presentation’, or ‘apprehensive appearance’ occur. To underline the fact that this term is highly technical, I shall also occasionally use ‘cataleptic impression’ or ‘cataleptic appearance’. 13 Zeno is not explicitly mentioned in the passage from Sextus, but it can be inferred on chronological grounds that the reference is to him. For a discussion of this difference, see Striker (1990, p. 158 in the 1996 reprint) and M. Frede (1999, 317 and 319).

226

CHAPTER THREE

Diogenes Laertius also lists various other kinds of items in the passage concerning the Stoic view on the criteria (7.54). The most important of them are the preconceptions (/.nk3veip); preconceptions are characterised as ‘natural conceptions of universals’ ((llnia tsqij¢ r‡l jah5kns), and they were listed as criteria by Chrysippus.14 In addition, Diogenes reports Posidonius to have said that some older Stoics took right reason (¬.h•p k5cnp) to be a criterion.15 Right reason can be understood basically in two ways. Either it can mean a kind of principle of coherence. If our beliefs are inconsistent, they cannot all be true. Or, right reason can refer to our rational faculty; in this sense it will be a criterion in the loose and traditional sense. 14

Chrysippus is also said to call perception (a©qhgqip) a criterion as well. Diogenes suggests that the fact that Chrysippus on the one hand takes only cognitive impressions as criteria, but on the other hand names perceptions and preconceptions as well is a sign of him contradicting himself. However, preconceptions do not seem to be at all difficult to fit the same pattern as the cognitive impressions. The case of perception, in its turn, seems to be analogous to cognition (jar1kgvip). In order for us to be right about things concerning perceptual matters we need to assent to a sense impression. Aëtius in fact mentions (4.8,12) that perception (a©qhgqip) is assent (qscjar1heqip) to a sense impression (talraq4a aåqhgrij3) and is also cognition. In addition, ‘perceive’ is often in antiquity used as a success word and this entails that if I perceive an x or if I perceive that it is the case that p, then x is there or p is the case. Therefore, we can take the reference to perception to amount to saying that in order for us to have true perceptual beliefs assent needs to be involved. Another question is whether all cataleptic impressions are perceptual. Some scholars such as Striker (1974, p. 52 in the 1996 reprint; cf. also her appendix to that article) take it that they are. Plutarch, however, says (Comm. not. 1084f) that notions, i.e. preconceptions are some kind of appearances (talraq4a c1. rip y (llni1 %qri). This does not entail that the earlier Stoics should take preconceptions to be appearances. It is not completely clear whether the early Stoics thought that cataleptic impressions could extend outside the scope of perception. I have for clarity’s sake spoken of ‘cognitive sense impressions’ when I wish to discuss cataleptic impressions as perceptual impressions. 15 In addition, Boethus is said to hold that all of the following, intellect (lnflp), perception (a©qhgqip), desire (≈.emip) and scientific knowledge (%/iqr3,g), were criteria. Boethus’ list carries connotations of the general use of the term criterion as referring to mental faculties. However, desire (≈.emip) and scientific knowledge (%/iqr3,g) do not fall under such a description. Desire might correspond to the Epicurean primary feelings (pleasure and pain) as criteria of pursuit and avoidance. Scientific knowledge, in its turn, can be taken as a criterion of truth in a way analogous to the cognitive impression. If we have %/iqr3,g that something is the case about a certain subject, then that is the case; having scientific knowledge thus entails knowing the truth about the matter. However, Boethus’ list is somewhat idiosyncratic and I shall not concentrate on it here.

HELLENISTIC PHILOSOPHY

227

A preconception can be taken to be a criterion in a similar way as the cognitive sense impression. Both cognitive sense impressions and preconceptions seem to be criteria in the following sense: if we have a preconception, or a cognitive sense impression, that something is the case, then it is the case. For instance, if we have the preconception ‘if something is a man, that thing is a rational mortal animal’ (Sextus, Math. 11.8–11
SVF 2.224, see LS 30I), then it is true that human beings are mortal animals. Plutarch and Alexander of Aphrodisias also say that, according to Chrysippus, nature gave us common notions (jnila§ (llniai) as criteria of truth (see Plutarch Comm. not. 1060a
LS 40R; cf. Alex. De Mixt. 217, 2–32). They actually can to some extent be assimilated with the preconceptions. I shall discuss the preconceptions and common notions below in a section called Preconceptions pp. 238–250. The Stoic view of the criterion of truth was subject to sceptical criticism from the very beginning. The main target of the criticism was the cataleptic impression. Of the Academics Carneades argued against the cognitive impression along the following lines. Because for every cognitive impression a false one with identical intrinsic quality can be found, the impression does not enable us to decide conclusively that something is the case. Therefore, the Stoics’ alleged criterion does not function as a criterion in the way they think it does. The argument is built on the idea that in order for the Stoic criterion to function as a criterion of truth, it should provide us with a means to distinguish between true and false impressions in all circumstances. As Carneades understands it, no matter how clear and persuasive a cataleptic impression is, it is always possible that there is an equally clear and persuasive impression which is false. Carneades, however, concluded that even though the Stoic criterion is not acceptable as a criterion of truth, we are entitled to use some similar criteria to decide whether it is reasonable to assent to an impression or not. If there is a convincing (/ihal5p) impression, which is not contradicted by other equally convincing impressions, and we are not in exceptional perceptual circumstances, we are allowed to accept the impression even though it is possible that it is false. (Cf. Math. 7.176, Cic. Acad. 2.31; 99–101 and Math. 7.184.)16 Therefore, an Academic sceptic like Carneades can take something to be the case and have grounds for doing so. However, they wish to deny the Stoic notion of a criterion on the basis that, according to Carneades’ arguments, it cannot do what it was supposed to do, namely to secure that the cognitive

16

Cf. Striker (1974, p. 55 in Striker 1996).

228

CHAPTER THREE

impressions can be distinguished from non-cognitive impressions and guarantee the truth of the resulting belief with certainty. Perceptions and Cognitive Impressions I shall now turn to the criteria more closely and see what kind of conditions they were taken to satisfy. At first I shall discuss the Epicurean doctrine according to which all perceptions are true, and its meaning and status in Epicurus’ theory. After that I will turn to the debate between the Stoics and the Academic sceptics on the cognitive sense impression. The crucial question will be what, according to the Stoics, distinguishes cataleptic sense impressions from non-cataleptic ones and how this was taken to support their criterial role. This question will be discussed with reference to the Academic criticism. We noted above that Epicurus held the problematic doctrine that all perceptions are true. Epicurus argues for this doctrine as follows. There is nothing which could refute perceptions: perceptions made by the same sense organ (Epicurus uses the Greek adjective …,ncel3p)17 cannot refute another perception deriving from the same sense organ because they are equally powerful or convincing (this phenomenon is in Greek called åqnqh2leia); perceptions from different organs (Epicurus refers to this by the Greek 8ln,nc2leia) cannot refute one another, because their objects are different. The only alternative that is left is that reason refutes perceptions. However, reason is wholly dependent (}.rgrai) on perception.18 Therefore, Epicurus argues, all perceptions must be taken as equally true. (Diog. Laert. 10.32) The crucial point here is the equal strength (åqnqh2leia) of all perceptions of the same sense organ. Later on in the same passage Epicurus says that dreams and madmen’s fantasies are also true in the sense that they produce effects; they make those people react. This could be taken to entail that by saying that all perceptions are true, Epicurus just says that all of them are real. If taken in this way, the point of equal strength would be that dreams and hallucinations are equally capable of causing bodily movement. However, if this were all there is to Epicurus’ doctrine, his argument for the truth of all perceptions would be quite pointless. It would merely say that perceptions of all sense organs are equally real. But who would doubt that? Epicurus.

17

I take the reference to the origin of the perception to be to the sense organs, because of the explanation that different origin (8ln,nc2leia) means different objects. Here it seems that Epicurus refers to the fact that I cannot decide the truth of, for instance, my visual perceptions by smelling. 18 Here Epicurus probably means that because our rational capacity grows out from perceptions, its reliability derives from that of the perceptions.

HELLENISTIC PHILOSOPHY

229

Rather, it seems that the equal power of the perceptions means the following. The only condition on which we could make comparisons concerning the truth value of our perceptions is their persuasive power or self-evidence. Now, the argument goes, because all our perceptions are equally evident or equally persuasive, we cannot but infer that they are equally true. The most natural way to take equal truth, then, is to say that they have the same truth value. If all perceptions were false, we would have no means to reach the truth. Therefore, they must all be true. Epicurus might also think that it is clear that some of our perceptions are true and, therefore, because all of them are equally powerful candidates for being true, they must all be true. Epicurus’ doctrine entails the obvious difficulty that if we can use all perceptions as equally powerful criteria to decide the truth of other opinions, the whole point of the criterion of truth seems to be watered down. If also dreams and madmen’s fantasies are classified as perceptions, and hence as true, we get strange results when evaluating the truth-value of our other opinions on the basis of them as well as on the basis of our waking perceptions. It seems that Epicurus classifies dreams and madmen’s fantasies as well as mistaken perceptions as perceptions, because he thinks that in all these cases there is a physical object from which the mental state arises. However, he makes a sharp distinction between the perceptual aspect and the aspect of belief. Perceiving involves reception of something from the external world and there is no additional element of belief in it.19 Another possibility of reading the argument would be to take truth to mean the following: in all perceptions, fantasies and dreams, there is an object we are aware of. This suggestion has been put forward by several scholars.20 However, this reading would again render Epicurus’ argument vacuous. If Epicurus were saying that all perceptions and fantasies are equally powerful because they are all about something, it is not clear why he would go on to discuss the possibility of one sense being refuted by another sense or by reason. Further, as Stephen Everson has argued,21 the suggestion that reality makes the difference between opinions and perceptions does not work: perceptions and opinions are distinguished on the grounds that one, namely perception, is always true whereas opinions might be false as well (Math. 7.203). If we take ‘true’ to mean ‘real’, this renders false opinions as unreal, non-existent, or

19

Epicurus is quite clear that even though all perceptions are true, many perceptual beliefs are false (cf. Diog. Laert. 10.51). 20 E.g. Rist (1972, 20); cf. also Long (1971a, 116). 21 Everson (1990a, 166). Further arguments against the suggestion can be found in Striker (1977).

230

CHAPTER THREE

without an object. This clearly is not the intended meaning of the doctrine. Finally, to take the doctrine to mean that all sensations and feelings are awarenesses of something entails that when, for instance, we have a headache, we need to postulate a mental object – a headache – which we are aware of when having a headache. One possibility that would make the doctrine easy to understand is to say that Epicurus uses ‘to perceive’ as a success word. In this case truth would follow from the definition of perception. However, this is not the point of the argument. Epicurus says that also the dreams and fantasies of madmen are true. The argument is that dreams and hallucinations cause people to act. There is, however, one further possibility suggested by Sextus and Lucretius which is in agreement with what Plutarch says. According to Sextus: In the case of Orestes, when he seemed to see the Furies, his sensation, being moved by the images (e©dwka), was true, in that the images objectively existed (∫/2jeirn c¡. r¡ e©dwka); but his mind, in thinking that the Furies were solid bodies, held a false opinion (Math. 8.63; transl. LS 16 F).

Lucretius, for his part, points out that when, for instance, a square tower seems round from far off, this is due to the fact that when the films of atoms fly from it to our eyes ‘through a large expanse of air the corner is forced to become blunt by the air’s repeated buffetings’ (4.353–363; transl. LS 16 G). Therefore, both Sextus and Lucretius take it that the doctrine according to which all perceptions are true should be understood in the context of Epicurus’ atomic theory of perception. This means that when saying that all perceptions are true, Epicurus is saying two things. Firstly, always when we perceive, there is a real (meaning physical) external object, i.e. the film of atoms or some kind of atomic effluence, coming from the world and affecting our sense-organs. Secondly, that particular film of atoms determines the content of our perception. For instance, an object is red if its surface has a certain kind of atomic structure and when we receive a film with that structure, we see red (cf. Plutarch, Adv. Col. 1110c
fragm. 30 Usener). According to this suggestion the doctrine of the truth of all perceptions means that whatever we perceive, this perception is always produced by a film of atoms with a structure that determines the content of our perception. Sextus’ clarification of the doctrine also supports this suggestion. He says that it means that ‘every perception comes about from a real object and in accord with that object’ (Math. 7.205). Further, Plutarch says (Adv. Col. 1109d) that the Epicureans also explained perceptual conflict by appealing to the atomic theory. The theory presupposes that we receive only those kinds of films of atoms that are suited to the pores in our sense organs. According to Plutarch, the Epicureans say that cases when,

HELLENISTIC PHILOSOPHY

231

for instance, the same water feels cold to one person and warm to another should be explained as follows: the former person’s pores allow those congregates of atoms which produce cold-feeling movements in the body and the latter allows atoms that cause warm-feeling motions.22 He concludes that Epicurus should have joined the Cyrenaics in saying that we are not aware of external objects at all, but merely of our own appearances. The Cyrenaics, in Plutarch’s view, close themselves to a siege of their own feelings and appearances in such a way that they cannot even say that we see a man or a horse but we need to say that we are ‘manned’ and ‘horsed’. According to Plutarch, the Cyrenaic dictum according to which there is no cognition of things should be understood in this way (Adv. Col. 1120e): we only cognise our own affections. In the end, Plutarch says, this school will teach me to say: ‘I receive a manlike impression, but do not perceive, whether a man is there’ (1121c), which he takes to be absurd. In spite of Plutarch’s rhetorical elegance, I think that Epicurus is not going in the direction of the Cyrenaics at all. Rather, the atomic films come from the surfaces of the external bodies and are thus literally parts of the perceived body. It is quite clear that the image (e© dwknl) Epicurus is talking about cannot be an inner mental item.23 Therefore, rather than locking himself into the siege of some kind of internal feelings or representations, Epicurus refers to physical objects, the films of atoms, which, according to him, are always there when perception occurs. When they fly in the air, they are diminished in size,24 but normally preserve the mutual arrangement; it is the arrangement which determines the perceptual quality we become aware of. Sometimes the flow of atoms is, for one reason or another, disrupted. The disruptions might occur for several reasons. One common example is the one where we think we see a round-shaped tower, even though from close up the tower is seen to be square. Here the apparent perceptual conflict is explained by the alleged fact that the atoms are under pressure when they fly in the air a long distance, and, the pressure makes them round. There are indications that Epicurus would not count perceptions from a distance as 22

Cf. 1109f–1110b. This also entails that Epicurus’ theory of perception is not representational; for a contrary view, see Taylor (1980, 117). For representational theories, see, e.g., Smith and Jones (1986, 85–89). 24 This is the reading adopted by Annas (1992, 159–160). Cf., Avotins (1980), who claims that the films of atoms are not diminished, but only take in parts of them and construct a scale model of the object on the basis of these parts of films. Asmis (1984, 131, see especially n. 23) follows Avotins in this respect. 23

232

CHAPTER THREE

criteria of truth, but as cases where the truth-value has to be decided by resorting to evident perceptions.25 This leaves us with dreams and madmen’s fantasies. We have seen in both Diogenes and Sextus that Epicurus takes both to count as cases of true perceptions. It seems to me that the reference to the atomic theory helps us understand this as well. To say that dreams and hallucinations are true amounts to saying that, in having one of them, we are affected by atoms, which determine the content of our experience.26 I would like to point out that it might be important for Epicurus to explain dreams and hallucinations by atomic interaction because he wants to underline their natural or physical explanations. I have not seen this point spelled out elsewhere in the literature. Like celestial phenomena, such as thunder, dreams have often been taken to be a kind of divine sign about the fate of human beings. As is well known, Epicurus wanted to remove all such superstition, and the theory of the natural origin of all dreams and hallucinations could have served this purpose. To say that also dreams are true might make us think that they are true in the following sense: if I dream about a god sending a message to me, a god is sending a message to me. However, Epicurus would probably say that the dream should not be understood that way. Rather, we should take it as saying: god is not sending a message to me, there just happened to be some films of atoms wandering around in the air, which caused me to have this dream. The unfortunate conclusion of taking the doctrine of the truth of all perceptions in the way we have just outlined is that it seems to undermine the evidential value of perceptions,27 which can be taken to be their main epistemological role. If all ‘perceptions’, including hallucinations and dreams, are taken to be true and to report to us the kind of organised whole of atoms that is affecting us, a problem emerges. We cannot decide on the basis of perception alone whether there is a solid body in front of us, or just a film of atoms presenting, e.g., a Fury-like dark-haired woman, as in Orestes’ case. However, Epicurus seems to have a way out of this problem. He thinks that from solid bodies there is a constant and homogenous flow of films of atoms (Diog. Laert. 10.46–48
LS 15A) towards our sense organs. In cases where 25

See, e.g., Sextus (Math. 7.211; 215–216): seeing someone from a distance is not yet evident, but needs to be witnessed by a close-up perception; cf. Diog. Laert. 10.34. 26 In dreams this might be somewhat difficult to explain, because most people sleep with their eyes shut and this usually prevents the films of atoms from affecting our eyes. Possibly we blink a little and then some films manage to enter our eyes and affect us. 27 Cf. also Taylor (1980, 112).

HELLENISTIC PHILOSOPHY

233

the atomic films are distorted, there is no constant flow of similarly distorted films. Therefore, Epicurus is not saying that we should have an unquestioned confidence in whatever perceptions we have, but that we should instead be careful not to run to rash opinions about external things on the basis of one perception only; the atoms might not have preserved the same arrangement they had on the surface of the body. We should rather wait and see whether similar films keep coming to our eyes. In the end, Epicurus’ much-debated doctrine according to which all perceptions are true, has acquired the form that perceptions are reliable witnesses about external things. In every perception there is a physical object, namely the film of atoms which is affecting our sense-organs and which determines the content of our perception. It makes us aware of an object as it is presented in the film. In normal cases the film has preserved the order and configuration of the atoms and, hence, the thing is as we perceive it to be. In some cases distortion occurs. However, also in such cases a real physical object is affecting us; it does not come from a solid body (i.e. a source of a constant flow of similar atomic films) – nor is it a divine sign: it is a chance configuration of flying atoms. These cases, however, are exceptional. Stoics. How, then, did the Stoics characterise cognitive impressions? We find three different descriptions in the main sources. Let us now discuss the first two descriptions. According to Diogenes Laertius, Chrysippus characterised a cognitive impression as follows: (1) a cognitive impression arises from what is (8/• ∫/1.unlrnp) (Diog. Laert. 7.54
LS 40A). (2) a cognitive impression is stamped and impressed exactly in accordance with what is (jar’ a√r• r• ∫/1.unl %la/eqt.aciq,2lgl ja§ %la/n,e,ac,2lgl) (Diog. Laert. 7.46
LS 40C).

A description highly similar to description (2) is found in Sextus. He adds that such impression presents the object ‘with all their peculiarities in a craftsmanlike way’ (talraq4al … /1lra reulij‡p r¡ /e.§ a√rn‹p ådi7,ara 8la,e,ac,2lgl) (Math. 7.247–252
LS 40E). Cicero also says that cataleptic impressions according to the Stoics ‘have a peculiar power of revealing their objects’ (propriam quandam haberent declarationem earum rerum quae viderunt, Cic. Acad. 1.41; transl. LS 40B). There are two crucial points in these Stoic descriptions of cataleptic impressions. Firstly, such impressions come from an external and existent object and, secondly, they present it in its detailed properties.

234

CHAPTER THREE

What kinds of cases are those where we have impressions coming from existing objects and they are presented in all their peculiarities? Consider the following quote from Sextus: Of true impressions, some are cognitive, others not. Non-cognitive are ones people experience when they are in abnormal states (jar¡ /1hnp). For very large numbers of people who are deranged or melancholic take in an impression which is true but non-cognitive, and arises purely externally and fortuitously, so that they often do not respond to it positively and do not assent to it.28 (Math. 7.247; transl. LS 40E)

Therefore, it is not sufficient for a cognitive impression to be true. We might have true impressions in abnormal psychic states and in that case the causal history of the impression would not be normal.29 Abnormal psychic states, however, are not the only ones which might prevent a true impression being cataleptic and functioning as the criterion of truth. Sextus lists the following conditions which need to be fulfilled in order for a cataleptic impression to occur and to produce a true belief, i.e. cognition. For a cognitive impression to occur, e.g., one of sight, five factors in their [the Stoics’] view must concur: the sense-organ, the sense-object, the place, the manner and the mind; since if all of these but one are present (i.e. if the mind is in an abnormal (/a.¡ t6qil) state), the perception, they say, will not be secured. For this reason some said that the cognitive impression is not a criterion universally, but when it has no impediment. (Math. 7.424; transl. LS 40L)30

In addition to such abnormal states of mind as drunkenness and melancholy, false beliefs can also form a hindrance which prevents the cognitive impression from functioning as a criterion of truth. If we stick to our previously formed false beliefs, we will not have the true belief entailed by the cognitive impressions. An example is the case when Heracles brings Alcestis back from the dead; Admetus sees her but does not believe she is there because he thinks she is dead (Math. 7.253–260
LS 40K). Further, the environment also has to be such that the impression can present the object to us in a craftsmanlike way. We, for instance, cannot be too far from the object (Sextus mentions the place), or there needs to be enough light and the

28

After this Sextus goes on to deal with the clauses we have just discussed. This entails that cognitive impressions have what Michael Frede calls ‘normal causal history’; see Frede, M. (1983). 30 In fact the condition that there must be no impediment was, according to Sextus (Math. 7.253), added by later Stoics. This is one of the indications that Academic sceptical criticism affected the Stoic analysis of the criterion of truth. Cf. Annas (1990). 29

HELLENISTIC PHILOSOPHY

235

acoustic environment must not disturb us. In spite of Sextus’list, we do not have a detailed analysis of what, according to the Stoics, it means that the mind is in a natural (jar¡ t6qil) state. Notice, however, that neither of the clauses refers to how we can recognise cognitive impressions or whether we can always distinguish them from non-cataleptic ones. However, the Academic sceptics kept insisting on the point that it is vital for a criterion of truth to be such that it provides us with a means of distinguishing between true and false impressions and, among the true ones those that are cognitive and those that are not. According to Carneades, cataleptic impressions as the Stoic criterion of truth cannot function as a criterion because they are indiscernible from non-cataleptic ones (see, e.g., Sext. Math. 7.401–420). There are two relevant kinds of indiscernibility (Math. 7.407): (i) the impressions we have in dreams are indiscernible from those we have in waking state and (ii) highly similar things produce in us completely indiscernible impressions. The argument from dreams goes as follows: Just as in waking states a thirsty man gets pleasure from drinking and someone who flees from a wild beast or any other terror shouts and screams, so too in dreams people satisfy their thirst and think they are drinking from a spring, and it is just the same with the fear of those who have nightmares (Math. 7.402–410; transl. LS 40H).

The idea is to point to the fact that in dreams we have clear and persuasive impressions even though the objects we have impressions of are not there.31 Therefore, having a clear and persuasive impression is not sufficient for ruling out the possibility that the impression is not caused by a present object. The second case is that of highly similar objects: In the case of things which are similar in shape but different objectively (jar¡ r• ∫/nje4,elnl) it is impossible to distinguish the cognitive impression from that which is false and incognitive. E.g. if I give the Stoic first one and then another of two exactly similar eggs to discriminate, will the wise man, by focusing on them, be able to say infallibly that the one egg he is being shown is this one rather than that one? The same argument applies in the case of twins. For the virtuous man will get a false impression, albeit one from what is and imprinted and stamped exactly in accordance with what is, if the impression he gets from Castor is one of Polydeuces. (Math. 7.410; transl. LS 40H)

31

The Stoics – at least Chrysippus – would, however, distinguish between impressions (talraq4a) and figments (t1lraq,a), the latter of which is ‘a fanciful thought which occurs in dreams’ (Diog. Laert. 7.49–51
LS 39A; cf. Aët. 4.12, 1–5
LS 39B); therefore, it is possible that the dream impressions Sextus refers to should not even be classified as impressions in the Stoic framework. Cf., however, empty impressions discussed in Sext. Emp. Math. 8.85 and Cic. Acad. 2.51.

236

CHAPTER THREE

The second part of the argument refers to the familiar idea that it is possible to have equally clear and persuasive impressions which come from objects there in front of us but are misleading in another way: they are indiscernible with respect to their intrinsic quality from impressions coming from different objects. Therefore, there cannot be impressions which are ‘imprinted and stamped exactly in accordance with what is’. The reason for this is that clarity does not guarantee discernibility. The final conclusion of the sceptic argument is that indiscernibility prevents cataleptic impressions from functioning as criteria of truth. Carneades’ arguments insist on the requirement that a criterion of truth must bear an internal distinguishing mark such as self-evidence in order for it to function as a criterion. According to Carneades, the mere possibility of there being an equally evident but false impression, indiscernible in its inner quality from the cognitive impression, entails that the cognitive impression cannot be a criterion of truth. From the sceptic’s point of view, the notion of a criterion of truth becomes an epistemic notion involving aspects like selfevidence or certainty. For them to say that we possess a criterion is to say that we possess a means to tell whether we possess it or not, and to do this in an infallible way. I have now argued that internal discernibility is not, for the Stoics, the decisive property that distinguishes cognitive impressions from others. However, there is some evidence that could be taken to point to the idea that a cognitive impression is in principle internally distinguishable from non-cognitive ones. Diogenes Laertius (7.46 in LS 40C) describes a non-cognitive impression as being not clear and not distinct (,¢ r.al‚ ,gd£ (jrs/nl) and this seems to entail that the cataleptic ones are clear and distinct.32 In addition, Sextus reports that, due to the Academic criticism, the Stoics add a further clause in their description of the cataleptic impression, namely the following:33 (3) A cataleptic impression is ‘of such a kind that it could not arise from what is not’ (n˙a n√j -l c2lnirn 8/• ,¢ ∫/1.unlrnp) (Math. 7.402
LS 40H).

This passage claims that an impression which is caused by an external object could not arise from what is not, i.e. it could not have been produced by an 32

Cicero also says that cognitive impressions have ‘a peculiar power of revealing their objects’ (propriam quandam haberent declarationem earum rerum quae viderentur) (Ac. 1. 41–2). 33 Some scholars, however, such as Striker (1974, p. 51 in Striker 1996) hold that this additional clause goes back to Zeno. The question of the accurate dating of this clause is not my main point here. The relevant question is the meaning of the clause; I discuss this question in the text.

HELLENISTIC PHILOSOPHY

237

absent object. In addition, it states that cataleptic impressions are ‘of such a kind that could not arise from what is not’. This additional clause might be taken to refer to the inner quality of the impression. This would amount to saying that, from the Stoic point of view, dream impressions are in fact not of such a kind that they could not come from what is not. Neither would two different impressions from different but qualitatively highly similar objects be completely similar. If this is what the passage is claiming, it is based on the idea that non-identical objects do differ from each other in their attributes34 and, therefore, an impression which grasps them in a craftsmanlike manner, and in all their peculiarities, also preserves these tiny differences. Although it is not impossible to read some of the evidence in a way that points to the idea that cataleptic appearances bear an internal distinguishing mark, we must bear in mind that, as I have argued above, internal distinguishability is not the decisive property of a criterion of truth, understood in a Stoic manner. Rather, a cataleptic impression is such that, if it is accepted, it necessarily produces a true belief and, hence, makes us possess the truth regarding the fact which the impression presents to us. In addition, what makes cataleptic appearances even more powerful criteria of truth in the Stoic sense is that they – at least almost – force us to assent to them (Math. 7.403–408, Cic. Acad. 2.38; 88–90; cf. Math. 7.253–260).35 A consequence of this idea is that all human beings have at least some true beliefs. Therefore, according to the Stoics, people in general have fairly reliable conceptions of perceptual matters. Mainly in ethical issues their beliefs become perverted due to the emotional ways of thinking wide-spread in human societies. Emotional reactions observed in fellow citizens tend to give one the erroneous idea that external things are good or bad. It is true that the Stoics acknowledge the possibility that we might fail to accept a cataleptic impression if some false beliefs we stick to contradict the impression. This was the case, for instance, with Menelaus, who encountered Helen on the island of Pharos. He failed to assent to the cognitive impression of Helen because he could not believe she was there due to his false belief that he had left her on a ship. As we noted, the later Stoics apparently added that a cognitive impression functions as a criterion given that it has no obstacle. However, even though it is possible to fail to accept a cognitive impression, they are usually accepted. This is primarily, it seems, when we do not yet have so many beliefs, i.e. when we are children or very young.

34 35

For this doctrine in Stoicism, see Long and Sedley (1987 I, 173–174). See also Striker (1990, p. 159 n. 14 in Striker 1996); cf. Annas (1990).

238

CHAPTER THREE

Preconceptions In addition to perceptions or cognitive sense impressions, the Epicureans and the Stoics both posited preconceptions as being criteria of truth. The term ‘preconception’ (/.5kgvip) was probably introduced by Epicurus. According to Diogenes Laertius, the Epicureans define preconceptions in the following manner:

Epicurus.

Preconception (/.5kgvip) … is as it were an apprehension (jar1kgvip), or correct opinion (d5ma ¬.h¢), or conception ((llnia), or universal ‘stored notion’ (jahnkij¢ l5gqip %la/njei,2lg) (i.e. memory (,l3,g)) of that which has frequently become evident externally: e.g. ‘such and such a kind of thing is man’ (rninflr5l %qril !lh.w/np). For as soon as the word ‘man’ is uttered, immediately its delineation also comes to mind by means of preconception since senses give the lead. Thus what primarily underlies each name is something self-evident (%la.c2p). (Diog. Laert. 10.33; transl. LS 17E with slight modifications.)

Above we identified two crucial properties of Epicurean preconceptions as criteria of truth. Firstly, they are true: if we have the preconception that, e.g., a human being is such and such a kind of thing, then human beings are such things. (Here the phrase ‘such and such a kind of thing’ probably refers to a kind of memory image of observed human beings.) Another important aspect of the preconceptions in Epicurus is that they refer to existing kinds of things. The fact that we have a preconception of human beings entails that human beings exist. This is due to the fact that the preconception is formed through memories of observations and we cannot observe non-existing things.36 Therefore, preconceptions cannot be vacuous (cf. Diog. Laert. 10.37). Epicurus’ account of how the preconceptions are formed in the human mind is short. Memory is involved; the preconceptions are even identified with memory. Otherwise Epicurus just says that the preconceptions are formed through something becoming evident to us frequently. The preconceptions might in the Epicurean framework be described as such modifications of the soul that result from many similar films of atoms imprinting the soul material.37 What, then, if any is the epistemological role of preconceptions? Epicurus’ answer to this question can be found in the following paragraph, which is concise to the point of being cryptic. First, Herodotus, it is necessary to grasp what underlies the utterances (r¡ ∫/nrerac,2la rn‹p th5ccnip) so that we have them as a reference point against which to judge matters of opinion, inquiry and puzzlement, and not have everything

36

‘Observe’ is here taken as a success word. On the basis of what was said about perceptions and their truth, we must conclude that reliability of perceptions is sufficient for our having accurate preconceptions. 37 This has also been suggested by Long (1971a, 120–121).

HELLENISTIC PHILOSOPHY

239

undiscriminated for ourselves as we attempt infinite chains of proofs (eåp !/ei.nl 8/ndeijl6nsqil), or have words which are empty. For the primary concept (/.‡rnl %ll5g,a) corresponding to each word (jah’ )jaqrnl th5ccnl) must be seen (bk2/eqhai) and needs no additional proof (,gh£l 8/nde4mewp /.nqde‹qhai) if we are going to have a reference point for matters of inquiry, puzzlement and opinion. (Diog. Laert. 10.37–38; transl. LS 17C with small modifications.)38

As is immediately clear, this is a kind of regress argument. There are basically two ways of understanding the argument. The first possibility is (i) if all our knowledge is based on proof, an infinite regress follows. Because such regress would render all knowledge impossible, there have to be some basic premises we know in an immediate manner and on the basis of which we can prove other statements. The second possibility does not refer to our knowledge of some basic premises, but to our having some basic concepts. According to the second reading the argument is: (ii) if our understanding of all the words should be based on knowing an account or a definition, this would entail that all the words making up the account should also be thus understood, and hence an infinite regress would follow. Therefore, our understanding of some basic words must be based on our having some natural concepts which we ‘see’ (bk2/eqhai) and of which we do not need an account. I think it is quite clear that the point of the argument is the latter (i.e. ii), not the former (i). Epicurus is not saying that in order for us to have knowledge it must be derived from some basic premises. Nor does he say that some terms are such that their definitions must be immediately familiar to us. Rather, he is saying that in order for us to refer to things correctly and non-vacuously, we need to have some basic concepts that are such that when we have them we ‘see’ what they refer to and, hence, understand the meaning of the corresponding words. Even though Epicurus does not say it explicitly in this passage, it is quite clear that what he takes to underlie the utterances (r¡ ∫/nrerac,2la rn‹p th5ccnip) are the preconceptions. In Diogenes’ summary of the Epicurean criteriology, we find the phrase ‘the first [notion] underlying each word’ (/alr§ nœl ¬l5,ari r• /.‡rnp ∫/nrerac,2lnl), referring to self-evident preconceptions,39 and also elsewhere in Diogenes’ presentation of Epicurus we find similar references.40 38

For discussions of the argument, see Long (1971a, 119–122), Asmis (1984, 24–34) and Scott (1995). 39 The passage reads as follows: ‘For as soon as the word “man” is uttered, immediately its delineation (r6/np) also comes to mind by means of preconception (jar¡ /.5kgvil), since the senses give the lead (/.ngcns,2lwl r‡l aåqh3qewl). Thus what primarily underlies each name is something self-evident (/alr§ nœl ¬l5,ari r• /.‡rnp ∫/nrerac,2lnl %la.c2p %qri).’ (Diog. Laert. 10.33; transl. LS 17E.) 40 Cf. 10. 72 and Key doctrine 37. Later authors also confirm this (e.g., Sext. Math. 1.57). For the evidence, see Striker (1974, p. 38 in 1996).

240

CHAPTER THREE

One could, of course, resist my reading of the argument by saying that it is the definition that Epicurus thinks we ‘see’ and the phrase ‘such and such a kind of thing is man’ is meant to refer to the definition. However, rather than suggesting that we should know some basic definitions and derive our knowledge of other things from these definitions, Epicurus was known in antiquity for rejecting the idea that definitions could make us understand words better. Epicurus does not say this explicitly in the preserved writings but we do have evidence for this in other sources. The most important piece of evidence is found in Cicero’s On Ends (see esp. Fin. 1.22), where the Epicurean spokesman Torquatus – when asked by Cicero to define pleasure – points out that everyone knows what pleasure is and it cannot be better known through a definition. Also in the letter to Pythocles (Diog. Laert. 10.87) Epicurus himself insists that ‘nature should not be investigated through empty axioms and stipulations’ (n√ c¡. jar¡ 8miÍ,ara jel¡ ja§ ln,nheq4ap tsqinkncgr2nl; transl. from Asmis with small modifications). It is most natural to take the ‘empty stipulations’ to refer to definitions.41 Therefore, I take it that the regress argument shows that, according to Epicurus, we have immediate knowledge of things (which Epicurus compares to seeing) through having preconceptions that are formed from perceptual experience either by direct confrontation with the objects (/e.4/rwqip), through analogy (8laknc4a), similarity (…,ni5rgp), or composition (q6lheqip) (Diog. Laert. 10.32). Knowledge like this is not knowledge of the definition of the thing, but some kind of familiarity with it, possibly through a mental image which we can be taken to ‘see’. However, even though Epicurus to some extent explains thinking by images, such images cannot be completely pictorial, but have to be rather abstract and general. For instance, we have a preconception of god as a blessed and immortal being free from all concern, anger and favour (see, e.g., Diog. Laert. 10.123–124 and 10.76–77). It is not easy to understand what kind of picture could capture all these features. In fact, Epicurus’ regress argument comes quite close to what Aristotle says in the Posterior Analytics II 19. Even though the metaphysical background assumptions and some basic suppositions concerning the nature of knowledge are very different in Aristotle and Epicurus, both of them express the idea that proofs cannot go on for ever. And what is the starting point they both bring forward? It is the fact that we are immediately familiar with things on a general level through the process of natural concept formation. In Aristotle this includes the assumption that we have an intellectual capacity (lnflp) transcending the perceptual memory and enabling us to grasp parts 41

The evidence for Epicurus’ rejection of definition is discussed in Asmis (1984, 39–47).

HELLENISTIC PHILOSOPHY

241

of the world’s intelligibility in a truly universal manner, by having the forms actualised in our rational soul. In Epicurus no such assumptions occur. Nonetheless, he notices that there must be some basic elements in knowledge claims which are not known through an account. Both Aristotle and Epicurus think that we are able to refer to existing things in a general manner because through having some concepts we are immediately familiar with those things. In Epicurus’ regress argument we encounter the epistemological upshot of the theory of preconceptions as criteria. For Epicurus preconceptions are true in the sense that they reflect things in the world. If we have, for instance, a preconception of a horse, this means that horses exist and that we have correct criteria for identifying them when we see them. We do not need to be able to articulate the conditions for identification as definitions. In a passage (10.33) where Diogenes describes how the preconceptions are formed according to Epicurus (quoted above), the phrase ‘such and such a kind of thing is a man’ is used to refer to a preconception of a man. This indicates that having a preconception of a man entails that we can pick a presentation of a man from our memory. In fact, the theory of preconceptions in Epicurus – as in the Stoics for that matter – involves the assumption that by having concepts we have some basic knowledge about the world: we know, at least in some sense of the word, what kinds of things there are. Notice that even though Epicurus can with some accuracy be characterised as a radical empiricist – he thinks that all perceptions are true and that in addition to them, we only have the preconceptions to rely on, and moreover they are formed through repeated perceptions – his regress argument is not a typical empiricist one. We would expect an empiricist to argue along the lines that all knowledge must ultimately be based on perceptions. In a sense Epicurus does think along these lines: the preconceptions are criteria of truth because they are completely based on perceptions. However, his regress argument shows that his approach deviates from foundationalist empiricist epistemology. He does not argue that there are some basic statements which are immediately justified through perception. Rather, he argues that all discourse is based on our ability to refer to things on a general level. This ability is explained by referring to the process of natural concept formation. There is one preconception which received a lot of attention in antiquity and which is problematic in the Epicurean framework. This is the preconception of god. In Velleius’ Epicurean argument related by Cicero in his On the Nature of Gods, the existence of gods is established on the basis of a preconception. According to Velleius, gods can be said to exist because nature herself has imprinted a preconception of god in the minds of human beings. (Cic.

242

CHAPTER THREE

Nat. deor. 1.43, 62–63).42 The argument at one point refers to a consensus. However, the core of the argument is not to refer to actual universal consensus among human beings.43 Rather, the argument resorts to the natural origin of the preconception: the preconception of god is not a fiction, because it originates in nature. However, it is not entirely clear how the acquisition of a natural preconception of god can be explained in the Epicurean framework. Epicurus thinks that gods are maximally happy and indestructible living beings who are not concerned with human affairs.44 They are sometimes taken to live in a place between the universes (intramundia). However, the Epicurean universes are assumed to be causally closed, and it does not seem plausible to assume that the gods could from outside of our universe cause us to have preconceptions of them. However, the argument establishing the existence of gods on the basis of the preconception does not quite work unless the origin of the preconception is causal. Lucretius (Rer. nat. 5.1161–1225) provides the following explanation for the formation of the preconception of gods.45 According to Lucretius, already in the early days of the human race, people used to see both in dreams and with their waking minds creatures that move their limbs – on the basis of which we know that they are living and, hence, sensitive – and who seem maximally happy and eternal.46 It is not entirely clear why, even if we agreed that it is empirically true that such visions and dreams existed, their existence should justify the belief in the existence of gods. It is possible that the idea that gods are only encountered in dreams is supposed to indicate that they cannot affect us normally through perception. If this is the case, it seems that we must allow a non-standard form of causation which explains such dreams. However, even though the Epicurean explanation for our preconception of gods is not totally 42

‘For he [Epicurus] alone saw, first, that the gods existed, because nature herself had imprinted the conception of them in all men’s minds. For what human nation or race does not have, without instruction, some preconception of the gods? Epicurus’ word for this is /.5kgvip, that is what we may call a delineation of a thing, preconceived by the mind, without which understanding, inquiry and discussion are impossible. The power and value of this reasoning we have learnt from Epicurus’ heaven-sent book on the yardstick and criterion. Thus you see the foundation of this inquiry admirably laid. For since the belief has not been established by any convention, custom or law, and retains unanimous consent, it must necessarily be understood that there are gods, given that we have ingrained, or rather innate knowledge of them.’ (Cic. Nat. deor. 1.43; transl. LS 23E.) 43 See Knuuttila and Sihvola (2000). 44 For the Epicurean theology, see, e.g., LS 13F, G and H; 23. 45 Discussed, e.g., by Scott (1995, 191–194). 46 Cf. Diog. Laert. 10.123–124
LS 23B.

HELLENISTIC PHILOSOPHY

243

clear, it shows that Epicurus thought that we can appeal to the existence of a preconception in order to establish the existence of the corresponding thing or class of things. Let us now briefly return to the doctrine that all perceptions are true. According to Diogenes Laertius, [a]ll sensation (a©qhgqip), he says, is irrational and does not accommodate memory. For neither is it moved by itself, nor when moved by something else is it able to add or subtract anything. (Diog. Laert. 10.31; transl. LS 16B.)

Epicurus wants to make a sharp distinction between what can be called the phenomenal content of a perception and a perceptual belief. The former is completely determined by the film of atoms we receive; the latter involves an added element of conceptualisation or identification of the phenomenal content with a propositional content.47 He thinks that the former cannot go wrong, but error in the latter is possible and not unusual. Such error takes place if we, erroneously, combine the phenomenal content of our perception with a concept of some other thing. If, for instance, we perceive a horse, we might mistakenly connect it with the preconception of a cow. In this case we err in our perceptual judgment, not in identifying the thing we see with a non-existing thing, but with an existing thing which, however, is wrong.48 Therefore, it seems that even though the preconceptions are criteria of truth, this does not entail that we can apply them infallibly.

47

The later Epicureans are said (Diog. Laert. 10.31) to have posited a third criterion of truth in addition to perceptions and preconceptions. The Greek phrase y talraqrij¢ %/ibnk¢ r‚p dialn4ap expressing this criterion is very difficult to translate. The literal meaning of the phrase is ‘impression-like attention by the mind’, or ‘presentational application of the mind’ as Elizabeth Asmis (1984, 87) translates it. This criterion seems to involve the idea that as long as we just pay attention to what appears to us through the senses, we cannot go wrong. But, as soon as we formulate it as a belief there is always the possibility of error, because of the reasons we have just discussed. If we say ‘this is a cow’, we might go wrong, but if we only pay attention to the phenomenal content of the appearance, we cannot err in the sense that those contents are always determined by the films of atoms, which are external objects, not mental items and they usually retain the qualities (as atomic configuration) of the things they come from. 48 We might also err in identifying the thing with something non-existent. For instance, if we see a white horse in the moonlight and it seems to bear a horn, we might take it to be a unicorn. In this case, in Epicurus’ terms, we would not have connected the perception with a preconception, but to a concept which has been made up by us by composition.

244

CHAPTER THREE

Stoics. I shall now turn to discuss the Stoic theory of preconceptions. Along with Epicurus, the Stoics took preconceptions to be criteria of truth. The Stoics also agree with Epicurus in calling the preconceptions natural; they are characterised by Diogenes as ‘natural conceptions of universals’ ((llnia tsqij¢ r‡l jah5kns) (7.54).49 According to the following passage from Aëtius their naturalness is connected to the way they are formed. When a man is born, the Stoics say, he has the commanding-part of his soul like a sheet of paper ready for writing upon (πq/e. u1.rgl e≥e.cnl eåp 8/nc.at3l).50 On this he inscribes each one of his conceptions (eåp rnflrn ,4al ^j1qrgl r‡l %llni‡l %la/nc.1terai). The first method of inscription is through the senses (/.‡rnp d£ … r‚p 8/nc.at‚p r.5/np … di¡ r‡l aåqh3qewl). For by perceiving something, e.g. white, they have a memory of it when it has departed. And when many memories of a similar kind have occurred (çral d£ …,neide‹p /nkka§ ,l‚,ai c2lwlrai), we then say we have experience. For the plurality of similar impressions is experience (%,/ei.4a c1. %qri r• r‡l …,neid‡l talraqi‡l /k‚hnp). Some conceptions ((llniai) arise naturally (tsqij‡p c4lnlrai) in the aforesaid ways and undesignedly (8le/ireul3rwp), others through our own instruction and attention (di’ y,er2.ap didaqjak4ap ja§ %/i,eke4ap). The latter are called conceptions ((llniai) only, the former are called ‘preconceptions’ (/.nk3veip) as well. Reason (k5cnp) for which we are called rational, is said to be completed from our preconceptions during our first seven51 years. (Aët. 4.11.1–4
SVF 2.83; transl. LS 39E.)

What, then, are the natural conceptions, i.e., preconceptions taken to be according to the Stoics? In the quoted description only the concept of white is mentioned as an example and one might think that only perceptual concepts in a fairly narrow sense could be acquired naturally in the way described in the quotation. Diogenes Laertius says in a passage we shall soon quote (7.53) that we grasp something just and good naturally (tsqij‡p lne‹rai) and therefore also the concepts of justice and goodness belong to the class of natural concepts. In addition to these, the concepts of natural species seem to be natural in origin (Cic. Acad. 2.21
LS 39C) and also the concept of god is mentioned as being natural (Plut. Comm. not. 1075e
SVF 2.1126
LS 54K, Cic. Nat. deor. 2.12–15
LS 54C).52 However, in some 49

Cf. the Epicurean ‘conception’ ((llnia) or universal ‘stored notion’ (jahnkij¢ l5gqip %la/njei,2lg) in Diog. Laert. 10.33 quoted above. 50 Cf. Arist. DA III 4, 430a1–2. The whole description bears a resemblance to An. Post. II 19. 51 Cf., however, Diog. Laert. 7.55–56 (
LS 33H3), where it is said that it takes fourteen years for reason to grow into maturity. See also Galen Plac. 5.2.49, 5.3.1
SVF 2.841. 52 The identification of the concept of god as natural points to the idea that we could on the basis of having this natural concept conclude that gods exist. On the other hand, Diogenes Laertius says that the existence of god is known through proof (di’ 8/nde4mewp). We might be inclined to think that if we need a proof for the

HELLENISTIC PHILOSOPHY

245

sources the notion of good is said to be formed by analogy (see, e.g., Cicero Fin. 3.33
LS 60D), and it is not immediately clear how this fits in with Diogenes’ account, where analogy is distinguished as a way of forming concepts from natural concept formation. Diogenes Laertius provides a fairly comprehensive list of ways in which, according to the Stoics, we can form concepts. It is by confrontation (jar¡ /e.4/rwqil) that we come to think of (%ln3hg) senseobjects (r¡ aåqhgr¡). By similarity (jah’ …,ni5rgra), things based on thoughts of something related, like Socrates on the basis of a picture. By analogy (jar’ 8laknc4al), sometimes by magnification, as the case of Tityos and Cyclopes, something by diminution, as in the case of the Pigmy; also the idea of the centre of the earth arose by analogy on the basis of smaller spheres. By transposition (jar¡ ,er1heqil), things like eyes on the chest. By combination (jar¡ q6lheqil), hippocentaur. By opposition (jar’%lalr4wqil), death. Some things are also conceived by transition (jar¡ ,er1baqil), such as sayables and place. The concept of something just and good is acquired naturally (tsqij‡p). That of being without hands, for instance, by privation (jar¡ qr2.gqil). (Diog. Laert. 7.53; transl. LS 39D slightly modified.)

The paragraph is based on the idea that, given that it is obvious that we have concepts of things we have not encountered, we need an account of how such concepts can be formed. (For instance, some people might have actually seen the Pigmies, but those who have not cannot form the corresponding concept through confrontation, i.e. actually coming across a Pigmy.) However, in this passage all concepts that do not arise from perception are lumped together. This is a little difficult, because in many cases the typical examples Diogenes lists are concepts of imaginary things (like the Cyclopes and the hippocentaurs, for instance) and concepts of imaginary things cannot be preconceptions, i.e. natural conceptions of existing things.53 However, in some cases, for instance god, the good and the just, the Stoics do recognise that there are natural conceptions of objects which are not perceived in any normal sense. existence of gods, it is not plausible to claim at the same time that we have a natural conception of divinity on the basis of which it is clear that gods exist. However, this does not seem to be the Stoic attitude to the question. By contrast, as Schofield (1980) points out, the Hellenistic philosophical schools did not assume that the fact that they produce proofs or other more loosely defined arguments for the existence of gods did not undermine the claim that we do have a natural conception of divinity. 53 Sandbach (1971) argues that all these procedures concern the formation of natural concepts but this cannot be the case without qualification for the reason just stated. The natural concepts, i.e. preconceptions, are such that if we have a natural concept of something (such as a human being or god) there are human beings or a god. Diogenes’ examples, however, are mostly concepts of imaginary things.

246

CHAPTER THREE

By contrast to the philosophers in the Platonic-Aristotelian framework, the Stoics do not think that there are intelligible forms by confronting which we could come to have general concepts.54 However, they do think that cosmic rationality structures bodily particulars.55 They do not deny that the instances of the same species are similar to each other; they only deny that the similarity would be due to metaphysical forms. Therefore, according to the Stoics, the fact that there are similarities in particulars is sufficient for us to have general concepts. Our general concepts are not based on there being a form actualised in the human intellect. The Stoics seem to think that even though there are no general human beings or general animals nor forms of human beings, the direct confrontation in perception with individual human beings and animals is sufficient for us to form an accurate general concept of those things. It is possible that the Stoics think that this is due to the instances of the same species being somehow more similar to each other than to instances of other species. Perhaps the human soul is taken to be such that appearances of things of the highest degree of similarity are stored in the same place and, by being thus stored, cause a permanent modification of the soul /lefl,a. But what can we say about the concept of good? Surely there are no corresponding forms which could cause us to have the concept. But, in contrast with the concepts of natural species where there are no general forms either, in the case of the good, we even lack proper instances. As the Stoics think, only virtue is good, and normally there are no Sages i.e. virtuous persons around. How, then, can we have the notion of goodness at all? Note that god is not equally problematic in Stoicism because god is in some sense present in all nature as its organising principle. We do not directly perceive god, but god is manifested in our perceptual environment. According to Cicero, the notion of the good is acquired by analogy (collatio rationis, Fin. 3.33
LS 60D), and he explains it as follows. The mind ‘climbs up from those things which are in accordance with nature’ and then ‘arrives at the conception of the good’. It is well known that the Stoic identification of the human good as that which is in accordance with nature means virtuous rational life. Therefore, it is quite clear that Cicero cannot mean that we form the notion of the good by climbing up from things in accordance with nature in this sense. A more plausible suggestion is that Cicero refers to the process of appropriation (nåje4wqip) in which an animal or a human child pursues what is in 54

Cf. 2.3 above. For Stoic metaphysics, see Long and Sedley (1987 I, 163–164); cf. Long (1986, 152–163). For the functioning of cosmic rationality in the world through /lefl,a, see Cic. Nat. deor. 2.37
LS 54H; see also Hankinson (1998a, 239–241). 55

HELLENISTIC PHILOSOPHY

247

accordance with its nature and avoids what is contrary to that nature. From this process, then, the notion of the good is somehow generalised. In the passage quoted above, Aëtius wants to make a distinction between concepts that arise in us undesignedly and without special effort and those that only arise through our own instruction and attention. Only the former are identified as being natural. Therefore, the passage in Aëtius points in the direction that when elsewhere it is said that the notion of the good is natural this means that there is some kind of automatic, or undersigned, generalisation taking place in our minds which leads from the process of appropriation to a general and abstract notion of the good. However, the correct analysis of what this notion implies and what is the correct definition of goodness is a matter of philosophical instruction and attention. The dialectical techniques such as division of a genus can be used in searching for definitions, i.e. in the quest for a more articulated concept.56 Some scholars have suggested that the Stoic preconceptions should be understood as being innate.57 In fact Cicero does refer to them by the Latin innatus (see, e.g., On the nature of gods 2.12–15 in LS 54 C). However, as the Aëtius passage shows, the Stoics are pretty clear that we form our concepts on the basis of experience. Dominic Scott has argued that if we want to speak of innatism in Stoicism, it must be understood as dispositional innatism.58 Such innatism involves the assumption that we are naturally capable of forming certain notions from experience, but they are not pre-existent in our souls. We discussed above the question of propositionality of the preconceptions in relation to Epicurus. The Stoics make it clear that they think that the preconceptions are propositional. For instance, we have a preconception, i.e. a natural conception of god ‘not only as immortal and blessed but also as benevolent, caring and beneficient’ (Plut. Comm. not. 1075e
SVF 2.1126
LS 54K). As a formulation of a preconception the Stoics prefer the conditional ‘if something is a god, it is immortal, blessed, caring and benevolent’. Up to this point we have only been talking about the Stoic views on preconceptions or natural conceptions. In Plutarch and Alexander of Aphrodisias 56

For the Stoic views on dialectic and definitions, see Long and Sedley (1987 I, 183–195). 57 See, e.g., Christensen (1962, 58) and Green-Pedersen (1984, 61). Christensen (ibid.) suggests that common notions are patterns according to which the rational mind works and such universal notions as the part-whole concept, the concept of consistency and the good as the end of purposive behaviour (ibid. p. 43). According to Green-Pedersen (ibid.) common notions exemplify the correspondence between the world and the mind’s understanding of it. 58 Scott (1995).

248

CHAPTER THREE

we also find a reference to common notions (jnila§ (llniai).59 Common notions are even said to be criteria par excellence (,1kiqra) (Plutarch Comm. not. 1060a
LS 40R; cf. Alex. De Mixt. 217, 2–32). In most lists of the Stoic criteria of truth the common notions are not mentioned.60 What, then, are such common notions? According to Alexander (De Mixt. 217, 2–19), Chrysippus argues for the Stoic view that there are three types of mixture, juxtaposition (/a.1heqip), fusion (q6cusqip) and mixture proper (j.Øqip) on the basis of common notions. Mixture proper is not accepted by the Aristotelians and the argument centres on it. According to Alexander, Chrysippus points out that: We have one appearance (talraq4a) of bodies composed by juncture [juxtaposition type of mixture], another of bodies confused together and destroyed together [fusion], and a third of bodies blended and mutually coextended in their entirety in such a way that each of them preserves its own nature [and this is the mixture proper, j.Øqip].61

The idea of the argument is that we have a common notion of mixture proper (i.e. of bodies blended in their entirety and at the same time retaining their own natures) and therefore such a mixture exists. Alexander also says that, according to Chrysippus, we have appearances (talraq4ai) of all the types of mixture mentioned. It is not clear whether the use of the term talraq4a here goes back to Chrysippus or whether it is Alexander’s. Therefore, it is not clear whether the Stoics would have thought that we perceive such a mixture. It seems that in any case we do not perceive it as that kind of a mixture. Rather, the point of the Stoic argument is that we tend to think in ways that presuppose the existence of such a mixture. Therefore, it would be inconsistent for us to assume that such a mixture does not exist. The notion of a mixture in which the ingredients are completely coextended but still retain their own nature is a highly technical one. It is an important part of the Stoic doctrine because pneuma as the vehicle of cosmic rationality is blended in this way with all the bodies in the world. We might wonder why the Stoics claim that such a highly technical notion can have a natural origin and therefore reflect the nature of things accurately. In fact it seems that according to the Stoics all our appearances are potentially very rich in the following sense. In addition to producing in us a fairly 59

The expression jnil¢ (llnia originates from Euclid. He uses it to mean the axioms of the geometrical system. See Heath (1956, 62). 60 For these lists, see Long and Sedley (1987, 40A-T). 61 The translation is slightly modified from Obbink (1992, 204), who has adapted it from Schofield (1980, 296–297). The additional explanations of the types of mixture are mine.

HELLENISTIC PHILOSOPHY

249

unarticulated preconception, which we have without any instruction or analysis, having a preconception enables us to articulate it into a worked out definition. As was noted, the Stoics express such expert definitions in a conditional manner ‘if something is a human being, it is a rational, mortal animal’. Probably Alexander’s reference to the mixture proper is, according to the Stoics such a definition ‘if something is a mixture proper, then that thing is a completely coextended blend of its ingredients which preserve their nature’. To take an example, think now of a painted white surface seen both by an expert and a layman. Both of them have the preconception of white and are capable of recognising the colour as white. However, the expert knows what it means to be white in contrast with other colours, how one can produce a beautiful shade of white from available ingredients, and so on. The layman does not know these things. We can say that in the Stoic analysis they both have a preconception, and perhaps even that they have the same preconception. However, the expert is capable of uttering a number of true propositions about the subject and he can give a much more detailed definition of it.62 A layman has only a non-expert perception of white as he is only capable of identifying the thing as being white. The expert’s appearance is already different, because it involves much more of the conceptual structure in which white is located.63 Therefore, on the basis of Alexander’s testimony it in fact seems that the common notions function similarly to the preconceptions. Because we tend to think in ways that presuppose the existence of the mixture referred to above, this indicates that we have a preconception of such a mixture and, therefore, such a mixture exists. Plutarch complains that the Stoics, though claiming that the preconceptions are criteria of truth, argued against common notions (Comm. not. 1059f–1060e). Therefore, he seems to identify the preconceptions with the

62

For the distinction between ‘expert’ (talraq4a reulij3) and non-expert appearances, see Diogenes Laertius (7.51); they are discussed, e.g., in Annas (1992, 81–82). 63 In fact Plutarch calls notions appearances of a kind (talraq4a c1. rip y (llni1 %qri, Comm. not. 1084f). This points to the idea that when the conceptual structure is articulated in inquiry, this also makes a difference in our appearances; they become conceptualised in a more accurate and richer manner. Whether Plutarch’s dictum can be used to indicate that the earlier Stoics took preconceptions to be appearances, is to some extent an open question. However, the indefinite pronoun rip in Greek is very weak in these kinds of cases. It need not carry the meaning that we should in fact subordinate the thing we speak of under the genus. Cf., e.g., Aristotle calling talraq4a a kind of thought (l5gq4p rip) (DA 433a10); he does not mean that having a talraq4a should amount to thinking.

250

CHAPTER THREE

common notions. He probably points to another feature of Stoic preconceptions: they have to survive tests of consistency. Plutarch argues that this is not the case with the Stoic notion of the good. According to Plutarch, the problem lies in the fact that the Stoics on the one hand claim that the notion of the good is a natural one and nature herself equips us with repugnance against injurious things, pains, mutilations and such things. On the other hand, they render these things indifferent in their ethics. In addition, the Stoics say that living in accordance with nature is the goal of human life and the highest good. However, they also say that natural things, which Plutarch identifies here with external goods, are indifferent (1060e). As noted above the Stoics would not think that living in accordance with nature should involve living in accordance with the external ‘goods’, such as health and food, as Plutarch thinks it does. Therefore, the last argument does not seem to hit its mark. However, a more difficult question for the Stoics is constituted by the first argument. Given that the Stoics also think that the preconception of good is learned through a process where first some external things are perceived as suitable, it is not quite clear how we can arrive at the technical Stoic notion of the good through natural analogy. In addition, it is also somewhat problematic that according to the Stoics external ‘natural’ things on the basis of which the notion of good is learned are excluded from the final and correct notion of the good. Their basic idea is that when forming concepts and becoming rational, human beings are changed in the way that what was good for them as infants and from which they learned the concepts is not good for them anymore as rational creatures. The fact that both the Stoics and the Epicureans accepted the preconceptions as criteria of truth entails that both schools assumed that in addition to statements which we can know in direct perception, in the process of natural concept formation we learn some general truths about reality. In acquiring a general concept we become capable of referring to kinds of things in speech and learn how to identify its instances later on. In addition, acquiring the concept entails that we can utter at least some true propositions concerning the kinds of things to which the concept applies. The simplest case is the Epicurean ‘such and such a kind of a thing is man’, where the description probably refers to some kind of memory image of men. Having the concept also enables us to initiate inquiry concerning a more detailed analysis of the concept. Particularly the Stoics seem to assume that a preconception includes potentially very rich description of things. An example is the analysed notion of a mixture proper: ‘if something is a mixture proper, then the bodies it contains are blended and mutually coextended in their entirety in such a way that each of them preserves its own nature’.

HELLENISTIC PHILOSOPHY

251

The Problem of Vagueness We have now discussed the role of preconceptions as criteria of truth and we have seen they are highly important both for the Epicureans and the Stoics. I shall now briefly discuss one sceptical challenge towards an important preconception, namely the problem of vagueness in the case of divinity. There were of course many other sceptical arguments against the preconceptions, and an important target was the notion of god.64 One fairly easy way for the sceptics to attack the preconception of god was the one already referred to above: to point out that the two schools had conflicting preconceptions.65 I shall not, however, discuss the sceptical arguments against the arguments for the existence of god based on preconceptions in general. Carneades argued against Stoic theology in soritical manner as follows. If Zeus is a god, Poseidon also is a god: Brethren three were we, all children of Cronos and Rhea, Zeus and myself and Hades, the third with the Shades for his kingdom. All things were parted in three, and each hath his share of the glory.66 So that if Zeus is a god, Poseidon also, being his brother, will be a god.And if Poseidon is a god, Achelous, too, will be a god; and if Achelous, Neilos; and if Neilos, every river as well; and if every river, the streams also will be gods; and if the streams, the torrents; but the torrents67 are not gods; neither, then, is Zeus a god. But if there had been gods, Zeus would have been a god. Therefore, there are no gods. – Further, if the sun is a god, day will also be a god; for day is nothing else than sun above the earth. And if day is god, the month too will be god; for it is a composite made up of days. And if the month is god, the year too will be god; for the year is a composite made up of months. But this is not true, neither then is the original supposition. And besides, they say, it is absurd to declare that the day is god, but not the dawn and midday and the evening. (Sext. Emp. Math. 9.182–184; transl. Bury.)

The sorites argument like this is a chain of implications together with the first antecedent. Therefore, its validity derives from the validity of the modus ponens. The argument seems paradoxical, because its premises are chosen in a way that the first antecedent and the implications seem clearly true, but the conclusion is clearly false. The name of the argument derives from the Greek qw.5p which means heap.68 64

For a discussion of such arguments, see Knuuttila and Sihvola (2000). For references and a discussion, see Schofield (1980). 66 This is a quote from Homer Iliad XV, 187–189; Poseidon is speaking. 67 Here the text reads ‘the streams’ (’6ajep), but it is more coherent to read ‘torrents’ (ua.1d.ai), because the word ‘torrents’ appears in the last consequent; see Sedley (1977, n. 89); cf. Burnyeat (1982, 326). 68 According to Diogenes Laertius (2.108) the sorites argument was first introduced by Eubulides, who was a contemporary of Aristotle. Eubulides is reported to have been fond of all kinds of logical paradoxes. In addition to the sorites he also proposed the Liar, the Elusive Man, the Electra, the Veiled Man, the Horns, and the Bald Man 65

252

CHAPTER THREE

Carneades uses the argument to undermine the credibility of Stoic theology by showing that the notion of god is vague and, hence, even Zeus’ status as a god is problematic. However, the argument does not seem to be a particularly good one against Stoic theology: on the one hand the Stoics take Zeus to be the only god; on the other hand, they are ready to accept that there is divinity everywhere in nature. Therefore, they do not seem to be employing a vague notion of divinity. Nonetheless, the sorites argument is of a more general interest because it could have been used against the theory of preconceptions. The Stoics, particularly Chrysippus, were interested in the sorites argument, and it is likely that at least in some form vagueness was recognised as a general problem on both the sceptical and on the ‘dogmatist’ side. Unfortunately, both Chrysippus’ treatises, three books under the heading On Soritical Arguments against Words (Diog. Laert. 7.192) and two books On the Argument from Little-by-Little to Stesagoras (ibid. 197), are lost. We only have later reports on Chrysippus’ reaction.69 Chrysippus recommended answering arguments in soritical form as follows. As long as the premises are clearly true, they can be accepted. But one must stop doing this before they turn unclear. This means that one must resist accepting some clear cases, but those must not be denied either. Only when the implications have already for some time been clearly false, can they be denied. This recommendation applies to arguments between two discussants, the sceptic presenting the sorites and his Stoic interlocutor. In such a case, the intended conclusion can only be derived on the basis of premises accepted by the interlocutor. Now, Chrysippus’ strategy is to prevent the sceptic from arriving at the conclusion by not answering crucial questions. However, as Cicero reports, Carneades had a response to this: silence will not help; it is no better than snoring in a discussion. The sceptic can always wake the Stoic up and start questioning again from the point where he fell asleep, i.e. stopped answering. (Cic. Acad. 2.93–94.) However, this is not the only piece of evidence on the basis of which we can infer the Stoic reaction to the problem of vagueness. It is nowadays widely recognised that the Stoics introduced the interpretation of vagueness called ‘the epistemic interpretation’.70 The point of this interpretation is that (ibid.). For these paradoxes, see Kneale and Kneale (1962, 114) and Williamson (1994, n. 2 on p. 8). In fact the ancient sources do not unanimously confirm that all these paradoxes were introduced by Eubulides; Diodorus (Diog. Laert. 2.111) and Chrysippus (ibid. 7. 187) are also mentioned as their inventors; see Barnes (1982, 36 n. 28). For the name of the argument, see Barnes (1982, 32 n. 18) and Burnyeat (1982, 316, n. 3). 69 Discussed in Barnes (1982); cf. Burnyeat (1982, 333–334). 70 Barnes (1982), Burnyeat (1982), Williamson (1994).

HELLENISTIC PHILOSOPHY

253

vagueness is not a problem of logic, neither is it a problem in the coherence of our conceptual system.71 (Cf., however, Sext. Math. 7.416–421.) It is a result of our epistemic powers. The core of the epistemic interpretation is the claim that for every predicate there is a sharp cut-off point such that on one side the predicate applies, and on the other side it does not apply. Vagueness is due to our not knowing where the cut-off point lies. In these cases even the Sage has to suspend judgment. It is clear that sceptics like Carneades were not satisfied with the Stoic answer to the sorites argument. Carneades thinks that it is insufficient to claim that the cut-off point exists somewhere on the borderline between the clear scope of a concept and its contradictory. From Carneades’ point of view the silence Chrysippus recommends as a response to soritical questioning is a kind of dogmatic sleep from which the philosopher must, sooner or later, wake up and answer the sceptic’s questions. On the other hand, from the Stoic point of view, Carneades’ requirement seems misplaced. Even though we do not know everything for certain, this does not mean that reality is as vague as our knowledge of it. In the Platonic-Aristotelian context the intelligible structure of reality is taken to be constituted by discrete intelligible elements determined by the forms. Particularly Aristotle took it that the forms are actualised in the human intellect and by virtue of this we can grasp the basic elements of the intelligible structure such as natural kinds and their necessary properties. The sorites argument was not explicitly directed at this view, but it could have been so directed. In this case the sceptical point would be that because our concepts are vague, no intelligible structure consisting of discrete elements can be reflected in the natural concepts. Therefore, the natural conceptual structure cannot function as a starting point for knowledge in the sense it is supposed to do. Galen, who considered himself a Platonist and who in important respects belongs to the Platonic-Aristotelian tradition, writes about the argument as it was used by the rationalist doctors against the medical empiricists. He takes the argument to show the non-existence of things that allow borderline cases. Galen argues that if the rationalists want to use the sorites against the empiricists, they 71

These are two common reactions to the problem in the contemporary context. Manyvalued logic in general was first developed by Peirce and carried forward by, for instance Lukasiewicz. Neither of them, however, proposes a many-valued approach to vagueness. According to Williamson, the first attempt to handle vagueness in many-valued logic goes back to 1949, to the Swedish logician Sören Halldén. Barnes (1982, 64) suggests that the ancient empiricist doctors can be seen as forerunners of this view. I shall deal with the ancient medical empiricist below in 3.4. For the interpretation that vagueness points to semantic incoherence, particularly in observational predicates, see Wright (1975 and 1976). For a defence of the epistemic interpretation, see Williamson (1994).

254

CHAPTER THREE

have to accept that the same argument can also be used to establish that there are no heaps (On Medical Experience XVII, p. 115–116 Walzer), no mountains (chapter XVII, p. 118), ‘neither nation, nor row, nor flock’ (ibid.), and so on. However, as such he takes the argument to be so absurd that the rationalists, rather than accepting the argument, should not use it against the empiricists either. Therefore, Galen sees the sorites argument as a powerful one, and his response is to direct it back to those who presented it. 3.2 IS THERE A TRANSITION FROM THE EVIDENT TO THE NON-EVIDENT? We have up to this point discussed the criteria of truth posited by the Epicurean and the Stoic schools and some sceptical challenges towards them. We shall now move on to ask whether the criteria of truth were taken to function as starting points for knowledge concerning non-evident things. It became an important assumption in Hellenistic philosophy that we must distinguish between what is clear or evident to us (d‚knl) and what is unclear or non-evident (!dgknl). What is clear or apparent is – or can be – directly apprehended by us, whereas some parts of reality are hidden or in their nature such that they cannot be perceived by our sensory system. The Stoics recognise hidden facts or things which in some cases can be revealed to us through proof (such as the existence of intelligible but invisible pores in the skin); the Epicureans posit a class of imperceptible entities (the atoms) as basic explanatory elements in the world structure. Now the question is, whether there is a transition from what is evident to what is non-evident. The transition (,er1baqip) was taken, if possible, to happen through signs or demonstrations (e.g., Sextus Math. 7.24–26, 8.140 and 319; cf. 7.396 and Pyr. 2.96). Signs were typically understood in a conditional form ‘if this, then this’ (Math. 8.276), and conditionals play a major role in Stoic logic. Even though the Epicureans did not formulate explicit views on logical form, their arguments took the form of conditional inference. There is a recent comprehensive monograph on the ancient theories of signs by James Allen72, and I shall not concentrate on the notion of sign here. We can note at the beginning that the distinction between what is clear and what is non-evident introduces a direction related to the human epistemic powers. We saw above that the idea of intrinsic directions plays an important role in the Platonic-Aristotelian tradition. It constitutes the basic distinction between two classes of starting points: those from which our inquiry starts 72

Allen (2001).

HELLENISTIC PHILOSOPHY

255

and those at which it is directed. Starting from the latter, we can – if our inquiry is successful – construct the proofs in the strict sense that express the explanatory relations between things. By contrast, in the Hellenistic context only one of the directions is relevant; the discussion concentrates on the question what non-evident facts can be established on the basis of what is clear to us. This corresponds to the direction from what is better known to us towards what is better known to nature in the Platonic-Aristotelian context. The idea that we should then ‘turn back’ and establish the evident facts through the explanatory principles is not prominent in Hellenistic philosophy. 3.2.1 Epicurus Epicurus, as is well known, was a strong advocate of the senses. However, he accepted the idea that we can, at least in some cases, establish the truth of beliefs or theoretical views which are not directly based on perception. His pronounced statements concerning how to do this are connected to the function of the criteria of truth as criteria. In addition, we also find some general outlines for an Epicurean methodology in the arguments establishing his most important doctrines concerning the make-up of reality, namely the existence of atoms and the void.73 I shall first discuss how the Epicurean criteria of truth function as criteria, and then I shall turn to the general argument forms Epicurus accepts as a means to establish theoretical views. Epicurus rejected dialectic as superfluous (Diog. Laert. 10.31) and was reluctant to discuss logic on a general level. However, we do find evidence that he ended up accepting some general argument forms through which the existence of some non-evident things can be established. Witnessing and Counter-Witnessing As noted above, for Epicurus the criteria of truth are used to decide the truth value of our beliefs when it is not evident. From Sextus we find several classifications of non-evident objects or facts (see Math. 8.142–155; 316–319, Pyr. 2.97–99).74 For the Epicurean criteria the most important is the one found in Math. 8.317–319. According to this distinction, things can either be 73

In Asmis (1984) we find a more comprehensive discussion of Epicurean methodological doctrines. 74 Sextus’ presentation is vague with respect to the question of whether non-evident entities are things or facts. Sometimes he talks about things (the city of Athens), sometimes about facts (the stars are even in number or odd). Asmis notes (1984, 190) that the more detailed classification of Math. 8.142–155 fits Epicureanism rather badly because it is based on Stoicism. These problems are to some extent avoided if

256

CHAPTER THREE

non-evident naturally or they can be homonymously called non-evident with the genus (…,wl6,wp kec5,ela r+ c2lei !dgka).75 If we are dealing with a naturally non-evident object or fact, we must accept that it is unknowable to us. Such is, for instance, the number of stars: we cannot know whether it is odd or even.76 But if something is homonymously non-evident, it is not altogether unknowable; it is hidden but can be made known to us. In Sextus’ example the existence of atoms and the void is said to be non-evident in this homonymous way. In Diogenes (10.34) we have a further Epicurean way of classifying nonevident things or facts. He adds to the picture the class of what is expected to be evident (r• /.nq,2lnl, e.g. a tower as seen from afar).77 Things or facts which are expected to be evident are not hidden in their nature. Neither can they be made known through proof. By contrast, they will be evident when perceived (e.g. the tower when seen close up).78 To use the criteria as criteria to decide the truth of other beliefs on the basis of them, Epicurus introduces a pair of concepts, that of being we use the less detailed two-fold classification from Math. 8.316–319 as I have done where the example is also Epicurean. Allen has recently argued (in his Study II, pp. 87–146 in Allen 2001) that Sextus’ classification in 8.142–155 is not appropriate from the point of view of Stoicism either, but is based on the medical debate between the empiricists and the rationalists. 75 Sextus’ terminology, however, is not completely consistent. In another place (Math. 8.145–150; cf. Pyr. 2.97–99) where he classifies things into different categories of what is non-evident, the expression ‘naturally non-evident’ is used to refer to the second category, i.e. to facts that are hidden, but can be made known through proof. 76 According to Sextus, it would be an exaggeration to say that such things have an unknowable nature, because saying this would entail that we do know something about their nature – namely that it is unknowable. By contrast, we should just say that such things are unknowable, and say nothing about their nature. 77 For a discussion, see Asmis (1984, 176–177; 190–196). 78 In Sextus we find examples which are difficult to classify in the Epicurean framework, such as the reverse side of the moon. Some scholars (e.g. Striker 1974) have suggested that these kinds of non-evident entities should be classified in the Epicurean category of what is expected to be evident (r• /.nq,2lnl); according to this suggestion the reverse side of the moon, for instance, is in principle perceptible and hence does not allow knowledge through demonstration but only possible explanation. Asmis (1984, 193), however, has argued that these kinds of entities should rather be classified into what is genuinely non-evident. She argues that for Epicurus it would be inconceivable that we could ever perceive these things. I also find it unlikely that Epicurus would have thought that, e.g., the reverse side of the moon will some day be evident. It is a more difficult question, however, whether it is hidden or naturally non-evident.

HELLENISTIC PHILOSOPHY

257

witnessed (%/i,a.r6.gqip) (or ‘attested’) and that of being counter-witnessed (8lri,a.r6.gqip) (or ‘contested’) (Math. 7.211–216; cf. Diog. Laert. 10.51). Attestation is available only in cases which are not in their nature hidden, and, hence, can become evident. Contestation is used in cases which are expected to be both evident and hidden.79 It might seem that only perceptions are counted into what is evident and can attest or contest our beliefs. Sextus, who is our most detailed source for the doctrine of attestation and contestation, says that the beliefs (d5ma) are judged on the basis of whether they are attested or contested by what is evident, and he often uses ‘evident things’ (r¡ %la.c‚) as synonymous with ‘perceptible things’ (aåqhgr1).80 However, I take it that because preconceptions are also said to be evident, there is no reason to exclude them from the class of evident things on the basis of which the truth of the beliefs are decided through contestation or attestation. This will become clearer below. The outline of the theory – if I may use that word here – of being witnessed or counter-witnessed by perceptual evidence seems simple. If we have clear evidence for a view in perception, we can accept it as true. If, by contrast, our perceptions witness against the view, it must be rejected as false. The examples of something being witnessed in both Diogenes and Sextus involve perception at a distance. Diogenes explicitly classifies such a case in the category of what is expected to be evident. His example is ‘waiting and getting near the tower and learning how it appears from close up’. In Sextus, the example is a case when someone is approaching and it is not yet evident to me, whether it is Socrates or Plato. I expect it to become evident to me when I go closer. Suppose that it is Plato. In this case the belief ‘this is Plato’ will be attested by the perception from close up, whereas the belief ‘this is Socrates’ will be contested by evident perception. Some complications arise, however, from the fact that in addition to the cases of evidence and counter-evidence, Epicurus also treats negations of 79

Contestation and attestation are said to apply to opinions (d5ma) both in Diogenes and in Sextus. Being non-evident, by contrast, can be applied to things themselves (/.1c,ara, in Math. 8.316). This causes some confusion, because at times it seems that it is that which is non-evident (i.e. the things, not the beliefs or opinions) which the method of attestation and contestation applies to. It seems that even though contestation and attestation strictly speaking apply to opinions – their truth-value is judged on the basis of these two notions – there is the related notion of non-evidence, which is used loosely both of facts and even of things along the following lines. The belief ‘this is Plato’ can be judged to be true or false by attestation or contestation through perception if (i) the fact that Plato is in front of me is not evident or (ii) Plato himself is not yet evident to me because of the distance. Cf. Striker (1974, p. 43 in Striker 1996). 80 Cf. Striker (1974, p. 42 in Striker 1996).

258

CHAPTER THREE

both of these. In addition to saying that something can be attested or contested by the senses there is the possibility of not being attested (n√j %/i,a.r6.gqip) by the senses. The most natural way of reading this phrase would be to say that it means that we simply do not have evidence, and must suspend judgment. However, Sextus’ explanation identifies non-attestation with the case that we have perceptual evidence against our belief: ‘we recognise through self-evidence that it is not Plato’ (Math. 8.215; transl. LS 18A). Sextus does not discuss cases where we remain without evidence with respect to something expected to be evident. For hidden things we cannot have direct evidence. According to Epicurus, we must in these cases inquire whether there is contestation or indirect evidence to the contrary. If there is, we must judge the belief to be false. If there is not, it can be taken to be true. Sextus takes the condition according to which a belief is true if it is not counter-witnessed by our perceptions in the sense that there has to be indirect positive evidence for its truth (Math. 8.213). However, Elizabeth Asmis has argued that Sextus’ discussion of the Epicurean notion of counter-witnessing is guided by Stoic terminology and is probably misleading with respect to the Epicurean theory. She points out that Epicurus himself in the letter to Pythocles (Diog. Laert. 10.86; cf. 10.87; 93; 95 and 112) explains those categories by the terms agreement (qs,twl4a) and disagreement or conflict (diatwl4a, r¡ ,au5,ela). On the basis of Diogenes it becomes clear that a theory must be taken to be true if it is in agreement with the phenomena or perceptions (tail5,ela, aåqh3qeip) and false if it disagrees with them. Therefore, as Asmis concludes, the Epicurean condition of not being counterwitnessed must be understood in the way that a belief or theory must be taken to be true if there is no counter-evidence to it. Consequently, Sextus’ suggestion according to which there must be indirect positive evidence for a theory, should be rejected. In any case, and this is the important point, it is quite clear that Epicurus accepts that, contrary to what one might expect, when a belief or a theory is not counter- witnessed by our perceptions, we are not supposed to suspend judgment. Rather, the belief or theory must be taken to be true. His methodological principle has the astonishing implication that some phenomena allow for several explanations (Diog. Laert. 10.80; 86),81 given that they are not contradictory to our perceptions or preconceptions and do not have implications which are contradictory to perceptions or preconceptions. They might, however, be incompatible with each other. Epicurus himself uses

81

Cf. Asmis (1984, 179).

HELLENISTIC PHILOSOPHY

259

this methodological principle several times. This happens, for instance, when he explains perception by the atomic theory in the letter to Herodotus (Diog. Laert. 10.47). An explanation for this peculiar doctrine has been suggested by James Allen. He argues that Epicurus means to say that if an explanation is compatible with what we perceive and it does not have implications contradicting perceptions either, then it actually is the explanation for the phenomena at least in some of the occurrences or in some worlds. The crucial reason for finding explanations for natural phenomena in the first place is, for Epicurus, to do away with superstitious fear related to phenomena such as thunder, whose explanation is unknown to ordinary people.82 I think Allen provides a plausible way to understand Epicurus’ methodological ideas and there is good evidence for it in the letter to Pythocles (especially Diog. Laert. 10.80 and 86). Particularly, Epicurus takes this approach to apply to celestial phenomena. Consider, for instance, the following passage: In the first place, remember that, like everything else, knowledge of celestial phenomena, whether taken with other things or in isolation, has no other end in view than peace of mind (8ra.am4al) and firm conviction (/4qril b2bainl). We do not seek to wrest by force what is impossible, nor to understand all matters equally well, nor make our treatment always as clear as when we discuss human life or explain the principles of physics in general – for instance, that the whole of being consists of bodies and intangible nature (8lat¢p t6qip), or that the ultimate elements of things are indivisible, or any other proposition which admits only one explanation of the phenomena to be possible. But this is not the case with celestial phenomena: these at any rate admit of manifold explanations (aår4a) for their occurrence and manifold accounts, in accordance with perceptions (ra‹p aåqh3qeqi q6,twlnl), of their nature (n√q4a). (Diog. Laert. 10.86; Loeb translation with slight modifications.)

We could also think in the opposite way and take the same methodological principle to function against the intended goal. If we find several explanations for a phenomenon some of which are incompatible with each other, this could undermine rather than fortify our belief in the existence of a natural explanation. However, Epicurus probably thinks that in a universe where atoms move randomly in a void it is more likely that similar occurrences happen for different reasons than it is that the same occurrence would always have the same explanation. Therefore, for Epicurus an explanation, in order to be a real one, does not need to be universally applicable. 82

Allen (1998, 308–311).

260

CHAPTER THREE

In the end, however, Epicurus makes it clear that he thinks that there are some natural explanations which involve reference to what is not evident, but which do not allow alternative explanations. The example is the existence of atoms and the void. But how does he think that these claims can be established? The Method of Elimination and the Method of Similarity We have discussed how attestation and contestation establish the truth or falsity of expected beliefs and how non-contestation confirms the truth of explanations. We have not yet discussed, however, contestation of facts that are non-evident in a homonymous way. Such contestation is of crucial importance in Epicurus. On the basis of the letter to Pythocles it seems that such contestation can establish a theory in the way that it no longer allows for alternatives. Some of our evidence for contestation as elimination comes from a later Epicurean, Philodemus, whose treatise has been preserved as a papyrus.83 Philodemus also lists another method, namely that of similarity along with elimination. These two procedures can be described as Epicurean versions of indirect proof and induction, and I shall turn to discuss them now. Eliminative contestation of homonymously non-evident objects will be discussed separately from the previous cases of attestation and contestation, because such contestation entails a way of establishing theoretical views in a manner that alternative explanations are excluded. Such contestation involves an argumentative step from what is evident to non-evident, whereas the previous types were rather straightforward cases of having or not having evidence. Sextus describes contestation as follows: It is the ‘elimination (8laqjes3) of that which is evident by the positing of a non-evident thing’(Math. 8.214; transl. LS 18A). His example is an argument concluding that the Stoic view according to which the void does not exist, is erroneous. For this purpose something nonevident is supposed, namely that the void does not exist. From this it then follows that motion does not exist either. It is concluded that, because motion clearly exists, the void must exist as well. Also the evidence from Philodemus’ treatise confirms that this is how elimination (8laqjes3) works.84 The basic idea of the argument is, as I said above, that the existent of motion necessitates the existence of void. In Sextus (Math. 8.214
LS 18A) it is formulated as a modus tollens. If there is no void, there is no movement either. But there is movement. Therefore the void must exist as well. 83

For an edition with a translation and a commentary, see De Lacy and De Lacy (1978). 84 For a discussion, see Asmis (1984, 199).

HELLENISTIC PHILOSOPHY

261

In the letter to Herodotus (Diog. Laert. 10.39–40)85, the argument is rather similar, but it is not formulated explicitly as a modus tollens.86 Epicurus himself clearly indicates that the inference (knciq,5p) is one which extends itself into what is non-evident (!dgknl). Therefore, the inferential step itself is not self-evident and the question becomes what, in the Epicurean methodology, justifies the inference from the existence of movement to the existence of void. It is highly likely that the Epicureans assume that the notion of void is based on perceptions of movement in air or water. However, the ancient atomists did not typically perceive movement in a void. Perhaps the only example of perceptions of void they might have had, is the medical treatment of blood-letting. In this treatment, which was common in antiquity, a vein is cut to make a wound. A cupping glass is then applied to the wound. Burning a candle in the glass causes a void and blood is sucked out of the vein. This treatment was not unquestionable evidence for the existence of void, however. The plenum theorists, who claimed that no void exists, used precisely these kinds of perceptions to show that nature cannot maintain a void. The problem in the Epicurean inference from the existence of movement to the existence of void is not the transition from the existence of bodies to the existence of empty space of some sort. I do not think there would be much point in doubting that inference. Rather, because the Epicureans were arguing against the plenum theorists, they had to be able to justify the inference from the existence of movement to the existence of void that is completely devoid of matter: no thin air-like ether or anything of that sort is allowed. And this is a conclusion that is much more difficult to argue for.87 The example makes clear that even though Epicurus in general may with some accuracy be characterised as an empiricist, he accepts that there are evident truths which are not literally based on observation. Even though we can observe that there is movement and the movement we observe happens in 85

For the text, see Asmis (1984, 238 n. 1). Sextus’ formulation might have resulted from Stoicising tendencies. However, to (re)construct the inference as a modus tollens is a way of making it logically valid. This is not to say that the Epicureans themselves conceptualised it as such. 87 In the history of physics it is Einstein’s general theory of relativity that marks the rejection of the notion of ether. If we need such a complex theory to establish that a void rather than some sort of ether exists, we have come a long way from the Epicurean methodology. There are many interesting issues related to the question of the justification of the Epicurean proof of the existence of void. I shall treat them in more detail in a forthcoming essay on conceptual analysis in Epicurus. 86

262

CHAPTER THREE

space, we do no perceive that if there is no void, there is no movement either. This inference seems to require some abstract conceptual knowledge which is used in deriving further knowledge. It is also assumed that such knowledge is not analytic but synthetic: it is not only about internal necessary relations within our conceptual structure but about necessary relations between things in the world. However, as I have pointed out, the general justification of this inference is not entirely clear in the Epicurean framework. As already pointed out, Philodemus argues (de Signis, cols. 11.32–12.31 in De Lacy
LS 18F) that, in addition to the method of elimination induced by the notion of contestation, there is another Epicurean way of establishing conclusions, namely inference by similarity, that is assumed to be equally valid. However, Philodemus does not provide a technical elaboration as to how this method should be used. The notion of similarity and inferences based on it also appear in connection with the doctrine of atomism and that of minimal parts (Diog. Laert. 10.58–59
LS 9A and ibid. 72
LS 7B). The relevant similarities can hold between several observed facts or between observed and unobserved ones.88 Philodemus’ example of an inference based on similarity is the following: if Plato is a man, Socrates will be a man too. He wishes to distinguish these kinds of inferences from those based on elimination. In an inference by elimination, there is an assumption (there is no void), which leads to the elimination of an observed phenomenon (there is no movement). The two methods differ at least in the sense that one is indirect (it assumes the negation of the intended conclusion), the other direct. However, I believe that there is more to be said on the difference between the two. In the method of elimination a non-evident conclusion is established on the grounds that if the relevant non-evident thing (e.g. the void) did not exist, a familiar and evident ‘thing’ (e.g. movement) would not exist either. In this way it is inferred that the existence of the non-evident thing is a necessary condition for the existence of the evident thing. Because we know that the evident thing exists, we can infer that the non-evident thing exists as well. By contrast, in the method of similarity the inference is – as the name of the method indicates – based on a similarity between things. Plato and Socrates are similar to such a high degree that it is inconceivable that one should be a human being and the other not. The method of similarity involves the assumption that if two perceived things are similar to a high degree, they must be of the same natural kind. In the method of elimination, by contrast, the inference is based on a necessary connection between two

88

See Asmis (1984, 177).

HELLENISTIC PHILOSOPHY

263

notions (based on a necessary connection between the things), not on any kind of similarity between notions or things. Another example of the method of similarity is found in the arguments for the theory of minimal parts according to which a finite body cannot have an infinite number of infinitely small bits in it (Diog. Laert. 10.56–59
LS 9A). In connection with this doctrine the inference based on the method of similarity is one concluding what properties the unobserved minimal parts have on the basis of what properties minimal observed parts of things have. Long and Sedley point out that the method of similarity is not used to prove the existence of minimal parts (this is done before the references to analogy, in LS 9A 1–6), but to explain how the minimal parts function in larger bodies.89 By contrast to the inference based on observed similarities, this inference is based on a similarity which we do not observe. This makes the justification of the inference highly problematic in the Epicurean context. The only Epicurean methodological principle I can think of here is the one according to which explanations concerning non-evident things are to be accepted as true if they do not conflict with our perceptions. But this is also problematic. We can admit that it is compatible with what we observe in general that imperceptibly small particles behave similarly compared to perceptible particles. However, it is equally compatible with the perceptual evidence that imperceptibly small particles are radically different from perceptible particles. This is what physics tells us nowadays. In this case Epicurus’ atomism appears dogmatic: alternative explanations are allowed elsewhere but not with respect to the atomic theory. Philodemus says several times in his treatise that in inferences based on similarity it is inconceivable (8dial5grnl) that the antecedent be true and the consequent false.90 This could be taken in the sense that some of the preconceptions appearing in the premises and the conclusions are incompatible with each other. However, the problem is that the conclusion refers to imperceptible entities, and it can be asked how, given that preconceptions are formed through perceptual experience, we can have preconceptions of them. The Epicurean arguments based on similarity can be characterised as inductive. The first type involves generalisation on the basis of observed similarity; the second one proceeds from the observed to the unobserved on the basis of assumed similarity. It is important that we start to find such arguments proceeding from individual instances to generalisation here. As we saw above, in the Platonic-Aristotelian framework inductive inferences proceed 89

Long and Sedley (1987 I, p. 42; cf. pp. 96–97). ‘Inconceivable’ (8dial5grnl) appears in cols. 14.17, 15.37 and 21.29; cf. Asmis (1984, 201). 90

264

CHAPTER THREE

from premises that are already general. Such generalisations are based on the natural process of concept formation or on a pre-existent universal in our minds. In that framework there is no need to infer on the basis of Plato being a man that Socrates is one too; this is known immediately on the basis of perceptions of Socrates and on the recognition of him as a man. We have now discussed how Epicurus thought we can establish or decide the truth-value of claims concerning non-evident things. First the notions of attestation and contestation were dealt with. The notion of attestation only applies to temporarily non-evident facts or things which are expected to be evident. Its meaning is simply that our belief about an object seen from afar can be confirmed or disconfirmed by our perception when we get near to it. Non-attestation in this case is understood as direct contrary evidence of the type ‘I saw that it was not Plato’. In the case of non-evident things, which are the proper subject of the chapter, we cannot have direct evidence either way and we must apply the notion of contestation. We have evidence that Epicurus thought that contestation is a conclusive method for establishing a theoretical view through indirect argument. The most important (and possibly the only) example is to show that the void exists by assuming that it does not and inferring that on this assumption motion does not exist either. Because it is obvious that motion exists, Epicurus concludes that the void exists as well. I have indicated above some problems concerning this inference. A peculiar case is non-contestation. Epicurus thought that if we do not have contrary evidence to a theoretical view in perception, that view can be taken to be true. This entails that natural phenomena allow for several even mutually contradictory explanations. This doctrine might be connected to Epicurus’ idea that superstition should be removed by finding natural explanations for phenomena which are most likely to cause it, particularly celestial or meteorological phenomena. In addition to the method of elimination based on the notion of contestation, Epicurus – as Philodemus testifies – also took the method of similarity to be conclusive. This is problematic, because the method of similarity consists of inductive inferences based on similarity between things. The invisible minimal parts are concluded to behave similarly compared to the visible minimal parts in this way: they are assumed to be so similar that it would be inconceivable that they should not behave similarly. Epicurus’ method of similarity is the first clear case in antiquity where we encounter the possibility of questioning the justification of inductive inferences from individual instances. In the PlatonicAristotelian tradition such a problem does not properly arise because of the assumption that metaphysical forms structure reality and we have immediate cognition of them.

HELLENISTIC PHILOSOPHY

265

3.2.2 Stoics and Sextus Indemonstrable Argument Forms By contrast to the Epicureans, the Stoics were very much interested in logic. They posited five unprovable argument forms (8la/5deijrni k5cni). In modern propositional logic they can be presented as follows: (a) (b) (c) (d) (e)

If p, then q; p; therefore q. If p, then q; not-q; therefore not-p. Not (both p and q); p; therefore not-q. Either p or q; p; therefore not-q. Either p or q; not-q; therefore p. (Diog. Laert. 7.76–81
LS 36A, Sext. Emp. Pyr. 2.156; Math. 8.223.)

According to the Stoics, all actually valid arguments are either in one of these five basic forms or reducible to one of them by means of a set of rules called h2,ara.91 According to the Stoic logic, therefore, the argumentative form of every valid argument is known.92 Of these argument forms the first one is the most important for the Stoics. The fact that these argument forms are called unprovable perhaps indicates that the Stoics were rather strongly committed to these forms. The undemonstrable argument forms can be seen as being constitutive of reason and they also regulate human reasoning. This, however, does not mean that the Stoics take logic to be psychological. By contrast, they seem to think that logical necessities are built in to the rational guiding principle of the cosmos.93 Further, the assumption that reality is basically knowable to us, also through valid inference, is probably based on this assumption. We noted that the Stoics do not assume that human reason should have some innate principles in any other sense than that of a capacity. There is no reason to suppose that the valid argument forms should be present in the human reason as innate structures either. How, then, are they learned? When Sextus presents how we are supposed to understand signs, he notes that we do have the notion of a conditional (8jnknsh4a, Math. 8.276).This is probably related to the fact that the Stoic preconceptions were analysed as conditionals (‘if something is a human being, it is a mortal rational animal’). Michael Frede has argued that according to the Stoics human reason is not essentially a capacity of inference but is constituted by a kind of knowledge of the world through preconceptions. However, his argument continues, the natural psychological 91

See Frede, M. (1974, 105–107, 118–121 and 124–127). On this aspect of Stoic logic, see Frede ibid., (196–198). For a recent presentation of Stoic logic, see Bobzien (2003). 93 Cf. Long (1986). 92

266

CHAPTER THREE

process of how we acquire the preconceptions at the same time teaches the notions of a conditional or consequence (8jnknsh4a) and that of incompatibility (,1ug). These notions form the basis of four of the five indemonstrables, that is those which are of the form ‘if p then q’ or ‘either p or q’.94 Also the remaining indemonstrable (c) (not (bothp and q); p; therefore not-q) is equivalent to the conditional ‘if p then not-q’, and therefore the Stoics might think that such an argument form is also learnt in the natural process of concept formation. However, for some reason they did not formulate it as a conditional but as a conjunction. In any case, it seems highly likely that according to the Stoics basic argument forms are learnt in the natural process of concept formation. Therefore, reason, in addition to being a collection (!h.niq,a) of preconceptions as Chrysippus himself is reported to have said (Galen Plac. Hipp. et Plat. 5.3.1
SVF 2.841), is also a capacity to make inferences according to the basic notions of consequence and incompatibility. From the fact that two statements are incompatible and one of them is true, it can be concluded that the other one is false. It is inconceivable that they could both be true. Proofs Like the Epicureans, the Stoics also allowed that in principle facts not known to us through perception can be established through proof. According to the Stoic definition, a proof reveals a non-evident conclusion on the basis of evident premises.95 Now we have got a general picture of what – according to the Stoics – is counted as being evident: namely cognitive impressions, preconceptions (and common notions) and indemonstrable argument forms. Now we need to find out more precisely how non-evident conclusions are established. Above I referred to Sextus’classifications into evident and non-evident things or facts. Sextus’ classification is probably connected to the medical debate between the rationalists and the empiricists, but it can also be used to illustrate Stoic proofs.96 According to the following classification found in Sextus (Math. 8.145–150), there are three different classes of what is non-evident. Something can be non-evident absolutely (jah1/am), temporarily (/.•p jai.5l) or naturally (t6qei). If something is non-evident absolutely it is completely unknown to us and cannot be revealed by proof (like the number of stars). If, on the other hand, something is non-evident only temporarily (like the city of Athens to me right now), it can be known by observation under different circumstances (me going to Athens later on). Only if something is non-evident naturally, can it be proved. Sextus’example of a naturally non-evident thing is the pores of the skin. 94 95 96

M. Frede (1994, esp. p. 55). For discussion of the definition, see Allen (2001, 171–173) and Barnes (1980). Cf. Allen (2001, 87–146).

HELLENISTIC PHILOSOPHY

267

The classification of non-evident things brings forward the important assumption according to which there are knowable non-perceptible things. In the characterisation of being naturally non-evident there is the built-in requirement of being existent or being real and, hence, being in principle knowable. Non-evident intelligible facts have to be proved by a valid argument from evident premises in one of the five indemonstrable forms, or by an argument reducible to them. The most important form of argument for proofs seems to be the first indemonstrable equivalent to the modus ponens. The classification of non-evident things is connected to the Hellenistic discussion of signs. If something is naturally non-evident, observed phenomena can function as what was then called ‘indicative signs’ (%ldeijrij¡ qg,e‹a) for them (the flow of sweat is a sign of the invisible pores). If something is an indicative sign for a naturally non-evident thing, the latter can be proved on the basis of the former. Indicative signs must be distinguished from commemorative signs (∫/n,lgqrij¡ qg,e‹a): we can be reminded of temporarily non-evident things by a commemorative sign – smoke can make me recall a fire (Math. 8.153) – but this is not a proof. The distinction between these two types of signs is, again, related to the medical discussions and is probably not Stoic as such. However, as James Allen has recently argued, the Stoics probably accepted a very similar distinction even though they did not use these medical terms.97 An example of a proof which the Stoics would accept is the following. If sweat flows through the skin, there are intelligible but invisible pores in it. Sweat flows through the skin. Therefore, there are intelligible but invisible pores in the skin. (Sextus Math. 8.306, cf. Pyr. 2.98).98

The second premise can reasonably be taken to express a cataleptic sense impression. We observe that sweat flows through the skin. But how is the first, conditional premise known? The most likely explanation is that the first premise is based on our preconceptions; we recognise skin as something solid and we also have a natural conception of solidity formed through our observations of all kinds of solid things. The preconception can, following the Stoic practice, be formulated as follows: if something is a solid body, another solid body cannot pass through it. Therefore, the above proof can be further analysed as follows. 97

Allen (2001, 159–160). As I pointed out in the introduction to this book, literally the conclusion reads ‘there are intelligible pores …’. Barnes (1980) translates lngr5l as ‘invisible’ in order to make clear that intelligible means intelligible but not visible. I use both in order to make the matter clear and at the same time to preserve the original reference to intelligibility. 98

268

CHAPTER THREE If skin is a solid body, liquid cannot pass through it. Sweat flows through the skin. Therefore, skin is not a solid body, but must have intelligible but invisible pores in it.

As we noted above, the Stoics assume that the acquisition of natural concepts entails acceptance of conditional propositions like the first premise in our latest example. Therefore, the conditional premises of the proofs analysed in the Stoic manner are ones we have in virtue of the process of natural concept formation. If someone has the concept of solidity, she also accepts that it entails nonpenetrability by another solid body. It is highly likely that the Stoic notion of a proof entails that the conditional involved has to be a necessary one.99 The Stoics distinguished between two kinds of conditionals, one equivalent to what is today called ‘material implication’ which is false only when the antecedent is true and the consequent false);100 the other is equivalent to logical consequence. The first type was called Philonian; the second one had a technical name of its own, qsl1.rgqip. In arguments satisfying the conditions of proofs the conditional has to be of the latter type; a material implication would not be enough. The Stoic proofs involving the necessary connection between the premises (qsl1.rgqip) can be taken to establish an explanatory factor for something we observe. In the example the non-evident (intelligible) pores explain the flow of sweat through the skin; in another example found in Sextus’ presentation of the medical debate, the soul explains the capacity of self-initiated movement. Therefore, a Stoic proof can be taken as a valid argument the conclusion of which is an explanation for the phenomenon mentioned in the premises.101 However, it seems that even though the Stoics took it to be possible that we can sometimes grasp necessary conditional premises and they thought that it is in principle possible to prove the existence of a non-evident fact, they seemed to think that this is quite rare. First of all, actual human beings are not ideal Sages and their knowledge of non-observable aspects of reality is very limited. In addition, they might have thought that even the Sage knows very little about such necessary conditionals which could function as the premises involving the transition from the evident to the non-evident.102 The Stoic attitude involves the idea that the most significant things to know are ethical, and 99

Cf. Allen (2001, 160–178). A Stoic description of it is found in Sextus: ‘that which does not begin with a truth and end with a falsehood’ (e.g., Pyr. 2.110). 101 Barnes (1980) has characterised it as an inference to the best explanation; cf. also Kapp (1931). 102 This is suggested by Allen (2001, 188). 100

HELLENISTIC PHILOSOPHY

269

the key property of a Sage is that he never assents to a false impression particularly in evaluative matters. Even though the Stoics were not too optimistic towards our knowledge of the non-observed explanatory aspects of reality, we can formulate the difference between Stoic and Aristotelian proofs by their relation to explanation. Stoic proofs contain an explanation in the conclusion and the explanandum in the premises and, hence, run in a direction exactly the opposite to Aristotelian proofs. Whereas Aristotle is clear that the explanations can be established by valid arguments from true premises, these arguments are not proofs in the strict sense precisely because they have the explanation in the conclusion and the explanandum in the premises. The proofs in the strict Aristotelian sense express the explanation in the premises and the explanandum in the conclusion. Arguments Involving a Non-Necessary Conditional The Stoics also discussed arguments involving a Philonian conditional and took them to be able to establish their conclusions in a plausible manner even though they were not proper proofs. An argument based on divination, for example, would be classified into this class. Next we shall discuss these kinds of arguments. An example of an argument involving a Philonian conditional based on divination is the following: If a god has told you that you will not die at sea, you will not die at sea. Zeus has told you that you will not die at sea. Therefore, you will not die at sea.

Other examples of arguments involving a Philonian conditional premise are found in Chrysippus’ arguments aiming to show that the commanding faculty (r• yceln,ij5l) of the soul is located in the chest. Chrysippus points to poetry, our use of language (see Galen de Plac. Hipp. et Plat. 3.5,28
SVF 2.896) and, finally, bodily movements simultaneous to the utterance of the word %cÍ and related expressions (Galen ibid. 2.2,10–11 and 3.5.8
SVF 2.895; 892).103 Premises related to our use of language are, for example, that it is commonly said that anger ‘rises’ and statements are ‘not swallowed’, and these are taken to reflect the alleged fact that the commanding faculty lies in the chest. Arguments using gestures as premises include the one in which Greek women in general point to their chest when denying a statement and say: ‘this does not go down here’. Perhaps the most famous of the arguments is based on the movement of the lips when the word %cÍ is uttered. The point of this argument is that the alleged fact that our upper lip points in a certain direction – to the chest – when we say %cÍ, reflects the location of the commanding faculty. 103

For the arguments, see Tieleman (1996, 196ff).

270

CHAPTER THREE

Galen vehemently criticises Chrysippus’ arguments.104 Galen is unhappy with the fact that Chrysippus resorts to plausible arguments (contrasted with proofs) in an important case like this. By doing so he seems to give up the hope of finding a proof for the location of the commanding faculty. Rather than proofs, Chrysippus can be seen to present cumulative arguments in which the premises taken from various sources feature as additional points of view taken to give further support to the conclusion. In such cumulative arguments premises pointing to how language is used and what kind of gestures we make probably function in the following way. In the Stoics’ audience someone might be convinced of the Stoic conclusion on the basis of these premises. If these points manage to convince someone, the argument is successful. It is possible that such premises are not even true (is our lip pointing to the chest when we say ‘%cÍ’?), but people in the audience might have many false conceptions; therefore it would not be possible to transform their whole belief systems into the philosophical world view. Still, it might be possible to convince them of some Stoic conceptions. Rather than expecting from Chrysippus arguments satisfying the criteria of proofs in the Stoic sense, the purpose of his arguments seems to be to convince an audience. However, it seems that particularly in the argument based on our lips pointing towards the chest when we utter the word %cÍ, the premises are extremely unconvincing if not clearly false. The argument can be reconstructed as: If we make a gesture towards a part of our body when we utter %cÍ, this gesture indicates the location of the commanding faculty. Our upper lip points to the chest when we utter %cÍ. Therefore, the commanding faculty is in the chest.

Even saying that both of these premises are false and that the argument is a bad one would not constitute a complete violation of the principle of charity. Even a bad argument can convince someone and it would be in line with the Stoic spirit to say that if it does, then it is not a thoroughly bad argument; it has served a purpose. It is characteristic of the Stoics that they think that in all cases an argument can be produced for a statement in dispute.105 For instance, the Stoics think 104

For references to, and a discussion of, Galen’s criticism, see Tieleman op. cit., (38ff and 288–290); cf. Allen (2001, 164–165). 105 According to some sceptical arguments this is a problematic feature in dogmatic philosophy. The sceptics think that given that the Stoics assume that there is a universally shared preconception of god, they should not present further arguments for the existence of gods. The sceptics think that the argument here makes the claim of the preconception suspicious. For the discussion, see Schofield (1980).

HELLENISTIC PHILOSOPHY

271

that all human beings have the preconception of god and the existence of gods is clear on the basis of our having this preconception, but they also have many other arguments for the existence of gods as well (see Cic. Nat. deor. 2). Also in the case of perceptual statements the Stoics think that although what is perceived can be taken as immediately acceptable, perceptual statements can be argued for.106 Rejection of Proof In the previous chapter we saw that both the Epicurean and the Stoic conception of proofs involve evident premises and a transition from them to a nonevident conclusion. In the case of the Stoics, the conclusion is non-evident in the natural sense in Sextus’ three-fold classification (found, e.g., in Math. 8.145). The sceptics, in addition to questioning the possibility of our having cognition of evident things, also challenge the notion of a proof. Sextus has various arguments against proofs in general. He proceeds in the usual ad hominem manner, namely by assuming the adversary’s position concerning proofs and showing that the position at hand leads to serious difficulties, if not to utter contradictions. Of the more specific arguments against proofs some are designed to attack the Stoic conception of proofs. To some extent the attack is also applicable to the Epicureans, but the Stoics are clearly the main target. According to Sextus’ arguments, proofs conceived as valid arguments in modus ponens form revealing a non-evident conclusion are either useless or they do not succeed in proving what they were supposed to prove. In order to satisfy the Stoic definition of proofs, the arguments must have evident premises which are connected in the necessary way and a non-evident conclusion which is revealed by the premises. The transition from the one to the other is made by means of a conditional premise (e.g. if sweat flows through the skin, there are intelligible but invisible pores in it). Sextus’basic criticism is twofold. Firstly, if the conditional premise is evident, the proof is useless (e.g. Pyr. 2.116). Secondly, Sextus’argument against the non-evident conditional is based on disagreements concerning the form of the argument. He points out that the Stoics do not even agree among themselves about the nature of the conditional, because some of them take it to be a necessary connection (qsl1.rgqip), others a material implication (i.e. the Philonian conditional) (ibid.). The second argument is not quite in line with the first, which trades on the content of the implication, not its form. We could add that if the conditional 106

Cf. the Stoic characterisation of dialectic as an expertise that ‘specialises in distinguishing true from false impressions’ (LS 31B7; G, 37H, 40I2), see also Long and Sedley (1987 I, 190).

272

CHAPTER THREE

premise is not evident, the proof does not satisfy the definition because one of its premises is not evident. Sextus himself does not mention this. Anyway, he concludes that proofs conceived in the Stoic manner cannot do what they are supposed to do. As for the first point, Sextus seems to think that if a conditional were evident, the knowledge of its antecedent would automatically entail knowledge of the consequent. Given that the proof is typically in the form of a modus ponens, the antecedent must be true, and also evident, according to the Stoic definition. If the conditional were evident, its consequent would also be evident and there would be no point in presenting the proof. Or, even if there were proof, it would not satisfy the Stoic definition according to which a proof must reveal its conclusion. If the conditional is already evident and the antecedent is evident as well, the consequent cannot be revealed. Therefore, the only way in which the proof can manage to establish a nonevident conclusion is the case in which the conditional is not evident. However, as is clear on the basis of the definition, an argument involving a non-evident premise does not satisfy the Stoic definition of proof. Sextus’ arguments are successful, I think, in showing that neither the Stoics nor the Epicureans provide an exact explanation of all the types of premises needed in proofs. The most problematic group of premises has been seen to be the one containing what I have called transition from what is evident to what is non-evident. Typically this transition is expressed as a conditional, and in the Stoic context it has to be understood as a necessary one. I have pointed out that both the Stoics and the Epicureans have an open possibility of claiming that such conditionals involve knowledge based on necessary connections between natural concepts, and even though the premises were not strictly speaking evident, they can be recognised as such by making the connections between natural concepts clearer. From the sceptic’s point of view the arguments against proofs show that Stoic proofs do not fulfil the definition they employ. Therefore, there is no reason for the sceptic to accept those proofs. The Stoics might have answered that even though it is very rare that we in fact know the necessary conditionals involved in proofs, this does not mean that the whole notion of proof should be abandoned. Again, as in the case of the sorites argument, they could say that if there are some deficiencies in our knowledge about the world, we should not draw too far-reaching conclusions on the basis of that fact. 3.3 WHAT IS LEFT FOR THE SCEPTIC? We have now seen that the Academic sceptics have argued against what they take to be the Stoic notion of a criterion of truth and Sextus as a Pyrrhonian

HELLENISTIC PHILOSOPHY

273

sceptic does not accept proofs. Sextus typically argues against every ‘dogmatic’ position in an ad hominem way. This procedure, according to Sextus, has led him in all the cases he has studied so far to a state in which he needs to suspend judgment: he has equally powerful grounds for accepting two opposing statements. This suspension of judgment has accidentally brought him freedom from disturbances – the result the dogmatic schools were seeking. In a famous simile, this is a compared to how a painter, after trying for a long time to paint foam in a horse’s mouth, finally threw his sponge on the canvas and, by accident, a perfect depiction of foam resulted. I shall not enter the question whether Sextus’ suspension of judgment results just as easily – in fact I believe that it does not. Rather, I shall discuss another question, namely what Sextus’ Pyrrhonian suspension of judgment amounts to. Pyrrhonian Scepticism and Non-Dogmatic Beliefs A sceptic, as the name indicates, makes inquiry (qj2vip). He examines different possibilities and tries to find out what the truth is. A typical choice is to inquire into things which are important in ‘dogmatic’ positions. The Stoics take certain kinds of appearances to be criteria of truth, and Sextus also presents Epicurus’ doctrine that all perceptions are true in the way that all appearances are true (e.g., Math. 7.203–204). However, when a sceptic like Sextus inquires into appearances in general, he notices that there are equally strong but contradictory appearances; therefore, no decision can be made as to which one is right and which one wrong. In this case, the sceptic suspends judgment (%/2ueil). He does not claim one to be right, the other to be wrong. Therefore, the Pyrrhonian sceptic concludes, appearances cannot function as criteria of truth. Sextus’ inquiry concerning preconceptions leads to a similar situation as that with appearances. The sceptic notices that the preconceptions proposed by the Stoics and the Epicureans, typically of god, are contradictory. The Stoics claim god to be providential, whereas the Epicureans say that the preconception of god includes the alleged fact that gods do not care about human matters. What can a true inquirer do in this kind of situation but suspend judgment concerning the preconception of god? Also in cases where the dogmatists argue for their views, the Sextan sceptic finds arguments which to him appear to be equally strong, but which have a conclusion contradictory to the dogmatist’s argument. Again, Sextus has to suspend judgment. It might seem that suspension of judgment resulting from sceptical arguments would force Sextus to claim suspension of belief altogether.107 107

This is the view defended, e.g., by Barnes (1983) and Burnyeat (1983).

274

CHAPTER THREE

However, this does not seem likely. Within the limits of the present study I cannot provide a detailed argument for this. Rather, I shall point to some aspects of Sextus’ scepticism which I see as supporting the view that suspension of judgment does not entail suspension of all ordinary beliefs.108 A common argument against scepticism in antiquity was the one according to which a sceptical attitude makes human action impossible, because any action requires that some commitments are made to how things are. Now, if we took suspension of judgment as amounting to suspension of any ordinary beliefs whatsoever, this argument would seem a rather powerful one against Sextus. However, it seems that Sextus did have a good answer to the argument. According to Sextus, appearances are criteria for the sceptic after all. However, they are criteria in a very different sense from the one suggested by the dogmatists. For dogmatists take appearances to function as criteria of truth. By contrast, for the sceptic they only function as criteria of action. There seems to be no reason to suppose that this should mean that the sceptic’s psyche functions in a different way, namely in a way that appearances as criteria of action are not his beliefs. It is more natural to take appearances as criteria of action in the way that they are beliefs and that the beliefs guide the sceptic’s action. The difference between the sceptic’s and the dogmatist’s attitude towards appearances, then, is that the dogmatist takes some of them to be true, whereas the sceptic merely acts on the basis of beliefs concerning the way the world appears to him. Brennan makes this point powerfully in criticising Barnes’ formulation of the view that suspension of judgment according to Sextus amounts to suspension of ordinary beliefs.109 Barnes claims that a sceptic cannot have ordinary beliefs, because this would entail that the sceptic has a criterion of truth, which Sextus denies. However, it is a very different thing to have ordinary beliefs and to have a criterion of truth. Moreover, it is compatible with the sceptic’s position to have a criterion of truth in the sense of having, or even accepting appearances, including the kind of appearances the dogmatists take to count as a criterion of truth. What the sceptic denies is that he thinks that he possesses a criterion of truth. Therefore, the sceptic is free to have all kinds of ordinary beliefs in various everyday situations and still deny the claim that there is a criterion of truth.

108

For the claim that suspension of judgment according to Sextus is compatible with having ordinary beliefs, see Frede, M. (1979); cf. Brennan (1994) and Striker (1981; in reprint 1996 p. 142). 109 For Barnes’s position, see Barnes (1983); for Brennan’s criticism, see Brennan (2000), a slightly revised version of Brennan (1994).

HELLENISTIC PHILOSOPHY

275

It is not likely that, from the sceptic’s point of view, the dogmatists would be wrong in having beliefs. Rather, the sceptic does not accept that some appearances are claimed to be criteria of truth. On the other hand, the sceptic denies that ordinary beliefs could be taken to be premises of proofs. It seems that the sceptics think that for perceptual beliefs to function as premises of proofs, we should assume that they are permanently true, or at least their status as true is somehow permanent. By contrast, the sceptic notices that conflicting appearances are ubiquitous. A tower appears round from a distance, square from close up, a stick appears bent in water and straight in the air (Pyr. 1.118–119). To take perceptual appearances to provide us with starting points for proofs would from the sceptical point of view be a blunt dismissal of these varieties in appearances in general. In fact, as is clear on the basis of the initial description of typical Pyrrhonian arguments, it seems that for the Pyrrhonian sceptic conflicting appearances seem equally powerful. Suspension of judgment is taken to follow because the sceptic cannot help having conflicting but equally powerful appearances. From this perspective the dogmatists make a mistake in choosing only one member of a conflicting pair of equally powerful appearances. However, as I indicated at the beginning of this section, Sextus is somewhat tendentious in collecting appearances and making them appear equally powerful. By contrast to Sextus, the Academics did not think that all appearances are equally convincing. Rather, they thought that one appearance can be clearer and more distinct than another, and if this is the case, the Academic sceptic can accept the clearer and more distinct appearance as the one which it is more rational to follow. However, to say that some appearances would be clear and distinct to such a high degree that they must be true (i.e. such that if they are accepted a true belief results), would be an exaggeration. This would from the Academic point of view be a piece of Stoic doctrine.110 One possible reaction to the fact that some appearances are clearer and more distinct than others would be to say that their propositional content is more likely to be true than that of the less clear appearances. However, it does not seem that Academic sceptics, particularly Carneades, took the clarity and distinctness of an impression to be a criterion of probable truth. Rather, it seems that Carneades concentrates on an ad hominem argument showing to the Stoics that a clear and distinct impression is not a sufficient criterion of truth, because we can find cases where two impressions are equally clear and distinct, but one is true and the other false. Therefore, an Academic sceptic could – in arguing within the framework of Stoic epistemology – accept that when an appearance is clear and distinct and more so than another appearance, the sceptic accepts 110

For references and a discussion, see Striker (1981).

276

CHAPTER THREE

the clearer one. However, this is not sufficient as a general criterion of truth or even of probability. Conflicting appearances can be equally strong and sometimes clarity and distinctness are not indications of truth.111 Cicero, a Roman follower of the Academic sceptics, made the move of claiming that even though we should abandon the idea that we can attain truth and certainty, our conceptions of the world can be truth-like (veri simile). It seems in fact that Cicero’s scepticism is a form of probabilism.112 Cicero’s remarks on truth-likeness as well as the whole discussion concerning the degrees of clarity of appearances can be seen as parts of the early history of the notion of probability.

3.4 WHAT DOES A DOCTOR KNOW? – MEDICAL EMPIRICISM AS AN ALTERNATIVE APPROACH TO SCIENTIFIC KNOWLEDGE We have seen that Hellenistic non-sceptical schools were rather modest in their claims concerning the extent to which we can have knowledge that is either of things which are not directly perceived, or knowledge that does not result from an analysis of the preconceptions. For the Stoics it was also typical to claim that it is possible to have knowledge even in the strict sense of %/iqr3,g. Such knowledge, however, only exists in a Sage and, as is well known, Sages are extremely rare in the history of the human race. Epicurus, for his part, thought that inquiry into nature is needed only to the extent that it diminishes our fear of death and other superstitious misunderstandings. Therefore, even though to some extent Epicurus provides us with an approach to natural science which is an alternative to the Aristotelian model, his main focus is never on scientific activity for its own sake. I shall now briefly discuss a methodological debate between the rationalists and the empiricists in antiquity. The main aim of my discussion is to provide a picture of an alternative ancient approach to methodology, a fairly radical form of empiricism. The Sorites Argument in Medicine Most of our evidence concerning the ancient medical debate between the empiricists and the rationalists comes from Galen’s treatises On the Sects for 111

Cf. Striker (1981, in reprint 1996, p. 136 n. 2), who refers to Burnyeat’s unpublished article ‘Carneades was no probabilist’. 112 See Glucker (1995); cf. Niiniluoto (2000).

HELLENISTIC PHILOSOPHY

277

Beginners, An Outline of Empiricism and On Medical Experience.113 The debate went on for centuries and there were several competing schools.114 For the present purposes, the rationalists and the empiricists are the most important ones, the empiricists being the main object of attention. Basically, the empiricists and the rationalists disagreed on the nature of the method of medical science. A standard presentation of the disagreement is the following: Some say that experience alone suffices for the [medical] art, whereas others think that reason, too, has an important contribution to make. Those who rely on experience are accordingly called empiricists (%,/ei.ijn4 ). Similarly, those who rely on reason are called rationalists (kncijn4 ). And these are the two primary sects in medicine. (Galen On Sects I pp. 1, 14–2, 1 from Helmreich’s edition; transl. from Walzer and Frede 1985.)

In the debate we encounter the sorites argument again and it takes us to the very core of the methodological disagreement between the two schools. On the basis of the discussion on vagueness above we might expect that it is the empiricists who use the sorites in the debate. Carneades had used the sorites against Stoic theology, and Sextus Empiricus is one of the empiricist doctors. It would suit his project to attack the rationalists on the grounds that because our general concepts are vague, we should not expect them to give us an accurate picture of reality. However, this is not the case. In the medical context it is the rationalists who use the sorites. A full presentation of the argument goes as follows. For they [i.e. the empiricist doctors] say that a thing seen but once cannot be accepted nor regarded as true, neither what was seen a few times only. They believe something can only be accepted and considered true, if it has been seen very many times, and in the same manner every time. I [a rationalist doctor] would ask them, therefore, if that which has been observed ten times is included in that which has been seen very many times, and their answer to this is ‘No’. Then I would say to 113

All the translations of these three texts are printed in Walzer and Frede (1985). The Greek texts of the On Medical Experience (Med. Exp.) and the Outline of Empiricism (Outline) have not survived. We do, however, have fairly early translations of both. The Arabic translation of On Medical Experience is made on the basis of a translation into Syriac; it has been edited and translated by Walzer (1947). The Outline of Empiricism has survived as a Latin translation made by Nicolaus from Reggio in the 14th century. For the text, see Deichgräber (1930, 7). Deichgräber has also made a translation back into Greek on the basis of Nicolaus’ Latin. Both the Latin and the Arabic translations are, on the basis of a comparison of the extant fragments, considered reliable. The page numbers are to Walzer’s and Deichgräber’s editions and translations (found in Walzer and Frede 1985), and in the case of On the Sects for Beginners (On Sects) to Helmreich’s edition. 114 See, e.g., Frede (Walzer and Frede 1985, xx–xxxi).

278

CHAPTER THREE them: And what has been seen eleven times? – and they say ‘No’. Then I would ask them further … And so I never cease asking and adding another number to each until I reach a high number. Nothing remains for him thus questioned except either to deny at a given time that the number has reached the limit when one can say it constitutes very many times, or, should he admit that it has, to make himself a laughing-stock for men, since he would thus require them to allow him a number reached solely by a usage fixed by himself, and a decision made by him alone … For if something that was seen forty-nine times and yet in all these times was not accepted nor considered to be true, now by the addition of this one single time comes to be considered acceptable and true, it is obvious that only by being seen a single time has it become acceptable and true. The inevitable conclusion is that seeing a thing once – although at the outset this was not accepted and considered true – has on this occasion such force that when added to something which was not acceptable and not considered true as to make it acceptable, and vice versa (Med. Exp. VII, pp. 96–97 in Walzer; transl. from Walzer and Frede 1985).

Galen himself does not provide us with any detailed analysis of the argument.115 After presenting the argument, he leaves it somewhat abruptly. He picks the argument up again towards the end of the treatise in chapter XV, and it is in that connection that we find a formulation of the sorites involving the predicate ‘heap’ (chapter XVII). There Galen himself uses the sorites against the rationalists as follows. The core of the rationalist methodology, according to Galen, is to make inferences (k5cni) from the visible to the invisible. One inference which they use – not in medicine itself but in the methodological debate – is the sorites. Now, Galen argues, if the rationalists want to use the sorites to show that medical knowledge consisting of experience is non-existent, they have to accept that the same argument can also be used to establish that there are no heaps (chapter XVII, p. 115–116 Walzer), no mountains (chapter XVII, p. 118), ‘neither nation, nor row, nor flock’ (ibid.) and so on. This is the basic idea of Galen’s response to the sorites presented against the empiricists. He repeats the same point very many times choosing many different predicates, but the argument remains the same: if the rationalists want to destroy the empiricist conception of experience by the sorites, they have to accept that the same argument destroys many obvious things as well. Further, he seems to assume that if such abstract arguments can be used to show obviously existing things such as heaps, mountains and crowds to be non-existent, the rationalists should admit that they are wrong in relying more on the power of arguments than that of observations. But let us now return to the argument as it was used against the empiricists. Before the exposition of the sorites, Galen presents two other rationalist attacks against the empiricists along similar lines. The very first one starts 115

He does not do so when discussing the argument elsewhere either; see, e.g., de Locis Affectis Book I p. 25ff. Kühn.

HELLENISTIC PHILOSOPHY

279

from the empiricist view according to which medical knowledge consists of technical observations and the technical observations consist of very many observations. The argument is that given that the empiricists do not accept a single observation to be technical, they should not do this in the case of very many observations either. The reason for this is that observing very many times is equivalent to many single observations. (Med. Exp. VII, p. 94 Walzer.)116 The idea is to show that ‘very many’ or ‘many’ cannot be well defined and distinguished from ‘one’ or ‘single’, because many always consist of the single instances.117 Perhaps the idea is to point out that, according to the rationalists, there is or should be a qualitative difference between a single ordinary observation and technical observation, but, according to the empiricists, this difference is merely quantitative. The second critical point is formulated as a simple question: ‘Can you tell us, Empiricists, how many times very many times is?’ (Med. Exp. VII, p. 95 Walzer.) The rationalist argument pays attention to predicates like ‘very many’ or ‘many’ and the argument looks like a sorites. These facts indicate that we are dealing with the problem nowadays known as that of vagueness.118 However, it seems to me that the debate does not provide us with any particularly insightful views concerning that problem. Let us take a closer look at the argument. The argument starts from the empiricist assumption according to which a single observation cannot constitute medical expert experience on the basis of which the doctor could cure his patients. We could express this, e.g., as follows: (i) f (1) is non-E, where f (n) refers to the predicate ‘has been observed (n) times’; E means ‘constitutes medical expert experience’.

Later on the rationalists get the following additional premises from the empiricists. (ii) f (10) is non-E. (iii) f (11) is non-E.

116

Galen mentions that he has actually made this first formulation himself (Med. Exp. VII, p. 95 Walzer). 117 Cf. also Sextus (Math. 7.415–421). 118 The sorites argument in the medial debate has, accordingly, been analysed as such by Barnes (1982). According to Barnes, the rationalists’ sorites shows that the empiricist notion of technical experience as the basis of their methodology is vague. Further, he suggests that the empiricists proposed a solution to the problem of vagueness by denying the logical principle of bivalence.

280

CHAPTER THREE

However, only at the following stage do we encounter the clue to the argument. It is the following transition, again taken to be accepted by the empiricists. (iv) If f (49) is non-E, f (50) is non-E.

Here the fact that the rationalists have chosen actual numbers, namely 49 and 50, is unimportant. The crucial point is that the rationalist argument can be generalised. They could use any numbers for 49 and 50. Therefore, the argument has the following general form: (i) f (1) is non-E. (v) If f (n) is non-E, then f (n
1) is non-E.

To say that arguments of this form are soritical involves the idea that at some point n has clearly grown large enough so that at such a point, call it i, f (i) is E. The problem is that we cannot quite tell when this happens. Undeniably, one aspect of the argument is that it is supposed to function on empiricist premises. Empiricists accept (i) and the rationalists also attribute (v) to them. It is not, however, completely clear whether the empiricists need to accept (v). They could claim that even though they deny that one observation alone suffices for medical expertise, this is not equivalent to accepting (v). I shall return to this point below. From the rationalists’ point of view, however, the problem is not vagueness or the absence of an exact borderline between non-technical and technical observation. From the rationalist perspective, the problem is that no amount of observation is sufficient for medical knowledge. For the rationalist, there is nothing paradoxical in the combination of (i) and (v), because rationalists do not accept that the number of observations ever grows large enough to warrant the name of ‘medicine’. From a rationalist perspective, the argument could be analysed similarly as above. (i) f (1) is non-E. (v) If f (n) is non-E, then f (n1) is non-E.

However, if we do not assume that f (i) is E for some i, we do not get the ‘paradox’ of the sorites. Rather, what we get is an argument which is highly similar to mathematical induction. Something (in this case ‘not being expert experience’) is first agreed to hold in the case of a small number (‘having observed something once does not constitute expert experience’), then a step is taken: if that predicate holds for n, it holds for n1. The crucial point in mathematical induction is this kind of inductive step: it is shown that if the same property belongs to an arbitrary number n, it also belongs to n1. This is exactly what the rationalists are doing. They show through such an ‘inductive step’ that no number is high enough to transform observation into medical expertise.

HELLENISTIC PHILOSOPHY

281

It is tempting to see some of the rationalists’ own assumptions lurking in the background. According to the rationalists, medical knowledge is in essence general, and knowledge concerning the general aspects of reality is gained by human reason. If they have this assumption in mind, the argument trades on the following point. If the empiricists think that by a multitude of observations alone we could justify generalisations, they are wrong. Only rational insight, according to the rationalists, can justify generalisation from observation. The rationalist methodology involves precisely this kind of assumption. Their methodology is based on the idea that human reason has a capacity to understand the nature and causes of illnesses and necessary connections between causes and effects, illnesses and their cures. In fact, the symptoms are taken to work as indications for the rationalist on the basis of which the cure is made evident (On Sects III, p. 5, 5–18). If it is assumed that our reason is able to apprehend general natures and general connections in the way the rationalists do, the above problem does not follow: our apprehension of the cause and powers of the illness is taken to justify the inference of the cure. Empiricist Expertise Even though logically elegant, the rationalist argument does not quite hit its mark. On the one hand, the empiricists are not interested in providing a universally applicable borderline between non-technical and technical experience. On the other hand, as pointed out above, they do not need to accept (v), namely the conditional premise: if n observations are not sufficient for expertise, n1 are not either. In addition, if we take the argument in the light of the rationalist assumptions and take up the issue of generalisation and rational insight, we must note that the empiricists do not make universal generalisations.119 Consider the following quotations. [E]xperience is the observation and the memory of those things which one has seen to happen often and in a similar way, or one can just say that it is the memory of these things. For observation is already implicit in memory, since we cannot remember those things which have been seen to happen often and in a similar way, unless we in some way make their observation. (Outline IV, pp. 51, 2–9; transl. from Walzer and Frede 1985.) (Cf. Outline III, p. 48, 2–4 and 14ff.) By experience, we mean the knowledge of those things which have become apparent so often that they already can be formulated as theorems, i.e., when it is known whether they always have turned out this way, or only for the most part, or half of the time, or rarely (Outline II, p. 45, 21–26; transl. from Walzer and Frede 1985; cf. Outline VI, p. 58, 10–25).

119

For the empiricist methodology, see also Frede, M. (1988).

282

CHAPTER THREE

The crucial point in these passages is that experience defined as a kind of memory and the ‘theorems’, which arise on the basis of memory, concern only cases one has observed.120 Therefore, the only generalisations the empiricist doctor makes are of the following form: ‘This has happened always’, ‘This has happened for the most part’, ‘This has happened half of the time’, or ‘This has happened rarely’. They do not involve generalisation concerning the future. Even though in the strict sense experience according to empiricist doctors applies to cases in the past, it is clear that they also think that experience of past cases guides our action and beliefs about future cases. Given that this is not due to generalisations made on the basis of observations, what is the empiricist’s attitude towards a case in the future? Their reply is, expectation: For, since he has often seen in cases like this that evacuation is beneficial, he expects (%k/4feil) that it will also prove useful when he uses it now (On Sects IV p. 7, 6–8; transl. from Walzer and Frede).

When the empiricist doctor equipped with the experience that evacuation, i.e. letting blood out from the body, has helped patients suffering from a certain kind of symptoms in the past, meets a new patient suffering from the same kind of symptoms, he or she expects that evacuation will be beneficial in the following case as well. In fact, the epistemic attitude the doctor has towards the effect of the treatment in the subsequent case is even weaker than the translation ‘expectation’ indicates. Behind the relevant verb there is the Greek noun %k/4p, which means ‘hope’. The doctor hopes the cure will help. If the doctor’s expectation or wish is sometimes not fulfilled by the subsequent observation, this does not falsify anything. It could only falsify a claim concerning the outcome of that case or a universal generalisation, but the empiricist makes neither one of these claims. If a contrary outcome results the empiricist doctor lists it as one observed case. Hence we will also say that a theorem is the knowledge of something which has been seen often but a knowledge which involves at the same time a distinct knowledge of results to the contrary. This will be a distinction between what

120

The medical empiricists’ conception of experience bears a resemblance to some other ancient views on experience. The closest relatives in this sense are the Epicureans (see De Lacy and De Lacy 1978, 165ff.) and the Stoics (Aët. 4.12). However, also Aristotle, who did not of course wish to minimise the role of reason, had a rather similar view of experience as being developed from perception and memory (see Met. I 1 and An. Post. II 19); cf. further Plato Philebus 32a–36a.

HELLENISTIC PHILOSOPHY

283

happens always (as something whose contrary never makes its appearance),121 what happens for the most part (as something whose contrary does appear, but rarely), what happens either way (as something whose contrary appears equally often), and finally what happens rarely (because its contrary does appear, not just sometimes, but for the most part). (Outline II, p. 46, 2–13; cf. Outline VI, p. 58, 10–25; transl. from Walzer and Frede 1985.)

The degree of expectation is determined by the frequency with which the similar result has been observed to happen. The more often the observation has been made, the stronger is the doctor’s expectation that the same cure (e.g. evacuation) will work in the following case. Note that the strength of expectation does not carry the connotation of greater justification. The point is not that the doctor should have greater confidence in predictions that are based on a higher frequency than in those that have been observed only rarely. Rather, the point is what happens in the doctor’s mind: the more frequent the similar observations, the more intense the expectation. The frequency, however, is not the only thing which affects the degree of expectation. If it were, medical experience conceived in the empiricist manner would in fact resemble a heap of observations. As a heap it would be characterised by two things: the quantity of its ingredients and their juxtaposition.122 However, experience is more than that, according to the empiricists. It has an inner structure, which is the result of a special kind of experience, namely imitative (,i,grij5p) experience. An experience is imitative if something, which has proved to be beneficial or harmful … is tried out again for the same disease. It is this kind of experience, which has contributed most to their art. (On Sects, II p. 3, 4–9; transl. from Walzer and Frede.)

An experience is imitative in the case when a doctor has seen that many patients suffering from the same symptoms have been cured by the same remedy, and then he or she tries the same remedy out in a subsequent similar case. If the imitation proves successful, i.e. if the patient is helped by the cure the doctor tried out, this corresponds to having seen the same thing happening in the same way very many times. 121

This might be taken to entail that the empiricists are ready to make universal generalisations after all. However, ‘happening always’ need not refer to all cases including the future ones, but can be taken in the sense of what has been observed to have happened always and whose contrary has neither been observed nor reported. 122 For these criteria for heaps, see, e.g., Med. Exp. XVII, p. 115 in Walzer. It is true that heaps also have an inner structure. The grains, for instance, which go to make up the heap, must support each other so that other grains can be placed on top of them. For structural models for heaps, see Hart (1992). However, the structure of experience is more complex than that of heaps.

284

CHAPTER THREE [O]nce one has put what one expected to the test, it already is trustworthy, if it has been confirmed by this, no less than if it had been observed many times to happen in the same way (On Sects II, p. 4, 8–10; transl. from Walzer and Frede).

And why is imitative experience so important for the empiricists that Galen says that it has contributed most to their art? The most likely explanation is that imitative experience is the way our minds naturally build up a cognitive structure. This holds in the case of medicine as well as in ordinary cases of children learning things about the world. For example, if a child sees that putting an egg in a hot frying pan makes it solidify, the next time she will probably expect that if she puts an egg in a frying pan it will solidify in a similar way. It is again an empirical fact that human minds work in the way that perceiving similarities produces expectations concerning future cases. If we go with the empiricists, we do not assume that this has any consequences concerning, for instance, the uniformity of the world. The empiricists have yet another ingredient in their expertise, which can be employed analogously to observations as a source of ideas how to cure patients, namely predecessors’ testimonies (ßqrn.4a) (On Sects II p. 3, 18–19). Such testimony is particularly useful in cases which the doctor has not observed. In these cases the doctor can see whether predecessors have encountered a similar case. If the observation confirms the testimony, this again intensifies the doctor’s expectation. Even though the empiricists seem to escape the rationalists’ sorites argument concentrating on the problem of distinguishing between technical and non-technical observation, the rationalists could repeat the soritical criticism and ask in the following manner: ‘How many times does one need to observe a thing happening similarly in order to start expecting it to happen similarly in the following case?’ or ‘How many times does something need to be observed to take place similarly in order for the empiricist to think that it has always turned out similarly?’. As we saw above, the empiricists think that there is not and there need not be a universally applicable answer to the question of how many observations need to be recorded in our memory. They seem to think that it is also an empirical question to determine how many observations are needed each time for each doctor to start thinking ‘this has always turned out similarly’ or expecting a similar result in the following case. There is no definite number which answers all such questions.123 123

Cf. the shoemaker example Galen gives: asking for a universal borderline between non-technical and technical observation is like trying to have one last which can be used to make shoes for everyone (Med. Exp. VII, p. 95 Walzer). Barnes’s (1982, 59) example is: there is no duration which is common to all vacations, though all particular vacations do have a definite duration. Barnes thinks that it would be

HELLENISTIC PHILOSOPHY

285

The disagreement between the rationalists and the empiricists around the sorites argument can also be located in a difference in their approach to crucial questions. The rationalists seem to be after some kind of justification for the empiricist notion of medical expertise. The contrast with the empiricists comes from the fact that the empiricists were not interested in the question of justification. Rather, their approach can be described as descriptive: they want to know how medical expertise is formed in the human soul. The rationalists, however, would not be satisfied with this move either. They think that even from a descriptive point of view the empiricist answer is insufficient. From the rationalist point of view, even taken in the sense just explained, experience presupposes reason. [E]ven when one grants them that one can see a thing happen in the same way very many times, they still are in no position to see, or to remember, or to write down, such myriads of differences as one finds in patients. Which library would have place for so large a history, which soul could store the memory of so many things? (Med. Exp. VII, p. 94; transl. from Walzer and Frede; cf. VIII, pp. 97–98, VI, pp. 91–92, and III, p. 88).

The idea of this argument is that the material we receive in perception is so rich, manifold and unorganised that if we did not assume that reason is organising and interpreting this material in our soul, we could not understand or even remember anything that we observe. In this case, the rationalists assume, no patterns should be formed and relevant similarities could not be recognised. The main critical point is that the perceptual capacity and memory alone could not make the difference between relevant and irrelevant similarities, and, hence, lead to any significant distinction between similar and dissimilar cases and possible cures in them. The empiricists admit that it is not trivial to decide which similarities are relevant. However, they claim that in the long run the difference can be sorted out. In the beginning, as is reasonable, he [i.e. the empiricist] observed what is beneficial and what is harmful, not only among things which it is useful to observe but also among those which it is useless to observe. In the course of the long time down to the present, though, with a multitude of observers having observed a vast array of things, many things have been found to have been observed in vain … For one has observed that the colour of clothes in many diseases is of no use, whereas in a few it is useful. For somebody who suffers from ophthalmia is helped by the colours blue, green, and black, whereas a light and gleaming colour

insufficient to answer the rationalists according to these lines. The empiricists, however, might well respond that it is insufficient only on the assumption that one is talking about the criteria of justification of medical expertise, but as I have suggested the empiricists are not looking for this kind of justification.

286

CHAPTER THREE is most adverse, and other colours are somewhere in the middle between both. (Outline VI, p. 54, 13–29; transl. Walzer and Frede.)

In this case predecessors’ testimony is used to bring in a kind of evolutionary perspective. From the reports of past doctors younger generations can learn what has been observed in the past and in this way the number of observations cumulates in history. Predecessors’ testimony is also very useful in situations which are as yet unknown to the doctor. If a doctor has not encountered a similar case in the past, he or she has in principle no guide other than to see how older generations have dealt with similar cases. We can now conclude that the empiricist conception of medicine deviates radically from the kind of idea of a science we have encountered in Aristotle. For Aristotle, as we saw, finding out necessary relations between the nature of things and permanent explanations for ever-recurring natural phenomena is vital. However, the empiricists could not accept the Stoic or Epicurean proofs either, because they do not accept reasoning that makes a transition from the evident to the non-evident. It can be suggested that from the point of view of a medical empiricist medicine is not a science but a craft (r2ulg). If this is the case, we should not expect that the doctor has an insight into the nature of the diseases and their causes; it is sufficient that they can successfully cure their patients. The idea that knowledge should be useful for practical purposes is markedly absent from Aristotle. Theorising, namely grasping the general outlines of scientific explanations, is for Aristotle an activity which is valuable for its own sake – and also for the sake of human happiness – but not for the sake of its technical applicability. The medical empiricists’ attitude is quite the opposite. We should not aim at understanding explanations or the nature of things; we should not speculate about them at all. We should only aim at curing patients in the most efficient way. We are now in a position to summarise the above discussion on Hellenistic philosophy and ancient medicine. One significant development is that when the question of whether we can attain truth becomes the focus of the discussion, the notions of evidence and clarity gain new importance. We have seen that it was originally a sceptical move to insist upon the point that we can be said to attain truth only if we have means to distinguish between true and false appearances in all possible situations. However, from the very beginning, also the non-sceptical schools attributed an evident character to the criteria of truth. In addition to the discussion of whether what is evident can be taken as a criterion of truth, there was some discussion concerning the question of whether we can use the evident criteria as starting points for new knowledge. The non-sceptical schools taught that we can, provided that we can make a

HELLENISTIC PHILOSOPHY

287

transition from what is evident to what is non-evident through safe steps. For the Stoics, such a transition takes place if we have evident premises from which we proceed to what is non-evident through necessary implication. They probably thought that we can in principle know such necessary implications because they are built in to the preconceptions we learn through a natural cognitive process. The Epicureans saw transitions from what is evident to what is non-evident in a framework of confirming and falsifying beliefs or theories on the basis of observations. Most notably, Epicurus allowed multiple explanation if it was compatible with perceptual evidence. We have seen, however, that in some cases Epicurus also thought that we can eliminate alternative suggestions and establish a single view. This can be done by his methods of elimination and similarity; they were also used to establish the basic tenets of the atomic theory. Elimination is a form of indirect proof that involves a transition from what is evident to what is non-evident. The justification of this transition is not completely clear. An even more problematic case is the inference from observed things to what is unobserved on the basis of similarity. Epicurus does not explain what entitles us to assume that what is unobserved is similar to the observed. A Pyrrhonian sceptic will suspend judgment concerning the question of whether we can know about non-evident things. The denial of this possibility would have been taken to count as dogmatism, and, therefore, it had to be left open. The medical empiricists, for their part, did not leave open the question of whether we can know about hidden things. They thought that rationalist theories involving hidden causes and natures were to be dismissed. Theorising is not necessary for medicine at all. Rather, medicine is to be conceived as a craft consisting of technical experience based completely on observation. The doctor preserves in his memory the cases he has observed and forms a kind of frequency table of them: treatments that have been observed to work in all cases, those that have been observed to help for the most part, those that have been observed to help in half of the cases, and, finally, those that rarely help. He expects that the same treatment will work in similar future cases as well. However, he will not make a universal generalisation claiming that this kind of cure works in such and such cases, let alone, for such and such a reason.

CONCLUSION

In this study I have discussed ancient philosophical theories concerning starting points for knowledge. The main emphasis has been on what I have called the Platonic-Aristotelian tradition. This tradition is characterised by some common assumptions. Most importantly, the philosophers classified in this tradition believe that starting points for knowledge exist. Plato and Aristotle both argue that the existence of knowledge is possible only if there are starting points for it. In the first part of this book, I have concentrated on the question of what philosophers working in this tradition take these starting points to be. The main finding of my discussion is that in the Platonic-Aristotelian tradition the starting points are taken to fall into two main types: starting points as premises of argumentation and starting points as general notions. We have seen that the ancient discussion of starting points for knowledge is largely based on the assumption that an account of such starting points is not sufficient if it only deals with premises. Premises can be suitable starting points for knowledge only if they are concerned with external existing things in an appropriate way. It is assumed that we can present true general claims about external existing things when the elements of the claims correspond in an accurate manner to the kinds of things in the world. This is assumed to require that basic contents of our cognitive structure are attained correctly: it is assumed that there is a natural cognitive process, which at least to some extent takes place in every human soul, and which leads us to have a basic grip of the world’s structure. In the background of these assumptions, we find a form of metaphysical realism, where it is assumed that reality itself has an intrinsic and immutable structure quite independently of what we human beings think about it. It is also assumed that the structure has an intrinsic order and that it is in principle knowable by human beings. However, it is also a common assumption that 289

290

CONCLUSION

the structure cannot be transformed into our minds directly as it is, but the order in which we come to know things is typically quite opposite to the intrinsic order of the structure of reality. We have seen that because the order of things is taken to be opposite to the order in which we come to know things, we need to make a distinction between two types of starting points within theories of argumentation. Aristotle formulates this distinction explicitly, but as we have seen, a similar assumption is found in Plato. According to Plato, there is an intrinsic order within the structure of reality and when we inquire into the nature of things, we often start from things which are secondary in that order. From those we can ascend to more primary truths about reality and at some point this ascent terminates in a principle which is prior to everything else. Such principle is assumed to explain the order and intelligibility of reality. In Aristotle the distinction between the order of our knowledge and the order of things determines two basic classes of premises as follows. Firstly, there are premises used as a starting point of inquiry in scientific non-apodeictic arguments or in dialectical arguments. Secondly, we can come to know more basic and more general truths about the world, which explain the facts we have at first been acquainted with. The principles in the second class come to be known through those belonging to the first class. Aristotle also assumes that in every field of study we can arrive at such basic truths that they do not allow further explanation in the same manner. Aristotle claims that basically such truths are peculiar to each of the sciences and there are almost as many basic explanatory principles as there are conclusions within a science. In addition to the premises with explicit content, it is assumed that in all argumentation and reasoning we need some general logical principles. Aristotle is not clear how the logical principles are acquired, but he makes it clear that all inferences are regulated by some basic logical principles. Plato and many other Platonists indicate that we cannot even learn anything from experience if the basic logical principles are not already there in our reason. One of the upshots of my discussion is that in the Platonic-Aristotelian tradition generalisation as such is not seen as inferential. By contrast, it is thought that human beings come to know some universal truths on the basis of a natural cognitive process. Through that process we come to have accurate contents in our intellect; having such contents, enables us to identify the general kinds to which the particular instances belong and to refer to those general kinds accurately in speech. It also enables us to present some true and universal generalisations about things (such as ‘human beings are animals’, or ‘virtue is good’). The accuracy of generalisation in that framework is guaranteed by a kind of correspondence between the elements of the structure of reality and the elements of the contents of our reason. As Aristotle puts it,

CONCLUSION

291

the same forms that are actualised there in the objects, are also actualised in the intellect. Even though it is assumed that there is a natural cognitive process providing us with some correct contents in the mind, it is granted that human conceptions are often false. Aristotle explains this by saying that whereas the apprehension of elements is always correct – if it has happened – connecting and separating such elements with each other may be false. Plato in particular tended to claim that common beliefs are usually misleading. However, he also assumed that we always accept at least some true conceptions and, therefore, it is in principle possible to show that false beliefs cause contradictions and should be revised. Aristotle thinks that through the cognitive process which activates our intellect we come to grasp some facts and relations within the intelligible structure in a reliable manner, and this allows us to start inquiry into explanations and definitions. Typically, the facts we at first learn appear in scientific proofs as conclusions. He also assumes that in many disciplines reputable opinions can be used as starting points for inquiry. If not literally true, they are often partly right. I have also touched upon the question of how the Platonic-Aristotelian tradition was continued in late antiquity. I have shown that the commentators accepted the basic distinction between the discussion of starting points in argumentation and in philosophical psychology. They also endorsed the idea that we must distinguish between premises as starting points for inquiry and basic explanatory truths found in inquiry. An important figure in the later development of the Platonic-Aristotelian tradition is Galen, who in many ways is very close to Aristotle. He to some extent departs from the Aristotelian two-way model of premises and attempts to claim that the explanatory premises of science should be self-evident. However, Galen ends up with recognising that all the explanatory premises cannot be evident in any normal sense; they can only become evident to the scientist in the inquiry. The second part of the book has concentrated on another powerful ancient tradition. The Hellenistic schools dominated the philosophical scene for about half a millennium. After Plato and Aristotle, there was a turn in the philosophical discussion concerning knowledge. First, the sceptical challenge concerning the very possibility of knowledge started to guide the discussion in a way unprecedented in Plato and Aristotle. This brought about a new emphasis on the notion of truth. Second, the general metaphysical framework according to which reality has an intrinsic intelligible structure consisting of discrete elements called forms was no longer assumed as a background for the discussion of knowledge. Hence, also the assumption according to which knowledge in the proper sense requires special immutable objects was abandoned. Rather, it was thought that knowledge must be about the same

292

CONCLUSION

things as our beliefs and ordinary conceptions. If we have knowledge, we cognise the same objects in a different, more secure way. Hellenistic philosophical discussion concerning knowledge concentrated on the question of whether we can attain truth at all. In the terminology of that time, this was formulated as the question whether there is a criterion of truth. I have distinguished above three ways of understanding the notion of the criterion of truth. For the Epicureans, the criterion is like a measuring stick against which the truth-value of other beliefs can be decided. According to Epicurus, we can decide the truth-value of our beliefs by comparing them to our perceptions and to a class of general notions called preconceptions. The natural origin of the preconceptions is taken to guarantee their accuracy. For the Stoics the most important criteria of truth are a special kind of assented perceptual appearance and a natural general conception. They are criteria in the sense that when we have such an assented appearance amounting to a belief or such a general conception, our belief or conception is guaranteed to be true. The Academic sceptics, in their attack against the ‘dogmatic’ conceptions of the criterion, take the criterion of truth to mean that it must provide us with a means to distinguish between true and false appearances in all possible situations. Because, they argue, the Stoic criterion cannot do this, it cannot function as a criterion of truth. From my argument concerning the Hellenistic discussion of the criterion of truth, we can draw the following conclusion. It was originally a sceptical move to focus attention on the question of whether true appearances are always discernible from false ones. This brought with it the assumption that we should try and find intrinsic qualities of appearances which would enable us to decide their truth value. I have argued that the Stoic notion of a criterion of truth does not depend on internal discernibility. Many philosophers provide a regress argument for the existence of starting points for knowledge. Epicurus provides one as well, and the argument is of particular interest. According to Epicurus, we can only discuss things on a general level if our understanding of all the general terms is not based on a definition. According to Epicurus, some of our general notions have to be basic; they are natural in origin and come from repeated sense perceptions. I paid attention to the fact that even though Epicurus can in some sense be characterised as a radical empiricist, his regress argument does not show that all knowledge must be inferred from sense contents. Rather, he argues that all our inferences and all human communication presuppose general notions that have their origin in experience and the human mind is such that it acquires them naturally. At the end of the second part of the book, I have addressed an approach to science alternative to the Platonic-Aristotelian framework. One of the rival

CONCLUSION

293

schools in ancient medicine, the empiricists, took medical science to be possible on the basis of experience alone. I have shown that their adversaries, rationalist doctors, argued that the empiricist conception of science is problematic because it does not involve rational insight into the nature of things. Rather, they argued, no number of observations is sufficient to generate medical knowledge because single observations are unable to justify generalisations. I have shown that this argument does not quite hit its mark, empiricist methodology. The rationalists assume that the empiricists take observations to justify generalisations but a medical science, according to the empiricists, does not consist of universal generalisations. Nonetheless, it is noteworthy that the problem of whether we can justify knowledge claims through observation alone did appear at that point. As we saw above, because of the assumption that the elements of the contents of our intellect match the elements of the intelligible structure of the world, this problem does not arise in the Platonic-Aristotelian framework.

BIBLIOGRAPHY

Ackrill, J. L. (1981), Aristotle the Philosopher, Oxford: Oxford University Press. Algra, K., van der Horst, P. W. and Runia, D. (eds.) (1996), Polyhistor: Studies in the History and Historiography of Philosophy (Philosophia Antiqua 72), Leiden: Brill. Algra, K., Barnes, J., Mansfeld, J. and Schofield, M. (eds.) (1999), Cambridge History of Hellenistic Philosophy, Cambridge: Cambridge University Press. Allen, J. (1998), ‘Epicurean Inferences: The Evidence of Philodemus’s De signis’, in Gentzler (1998) (ed.), 307–350. Allen, J. (2001), Inference from Signs: Ancient Debates about the Nature of Evidence, Oxford: Oxford University Press. Annas, J. (transl. with introduction and notes) (1976), Aristotle’s Metaphysics Books M and N (Clarendon Aristotle Series), Oxford: Clarendon Press. Annas, J. (1982), ‘Knowledge and Language: the Theaetetus and the Cratylus’, in Schofield and Nussbaum (eds.) (1982), 95–114. Annas, J. (1990), ‘Stoic Epistemology’, in Everson (ed.) (1990b), 184–203. Annas, J. (1992), Hellenistic Philosophy of Mind, Berkeley: University of California Press. Armstrong, A. H. (1960), ‘The Background of the Doctrine that the Intelligibles Are Not Outside the Intellect’, in Les Sources de Plotin (Entretiens sur l’antiquité classique 5), Vandoeuvres and Geneva: Fondation Hardt, 393–413. Ashbaugh, A. F. (1988), Plato’s Theory of Explanation (SUNY series in Philosophy), Albany, NY: State University of New York Press. Asmis, E. (1984), Epicurus’ Scientific Methodology, Ithaca, NY: Cornell University Press. Avotins, I. (1980), ‘Alexander of Aphrodisias on Vision in the Atomists’, Classical Quarterly 30, 429–454. Bäck, A. (1996), On Reduplication. Logical Theories of Qualification (Studien und Texte zur Geistes Geschichte des Mittelalters 49), Leiden: Brill. Bailey, C. (1928), The Greek Atomists and Epicurus, Oxford: Clarendon Press. Baldassari, M. (1986), Alessandro di Afrodisia. Dal commento agli ‘Analitici primi’, dal commento ai ‘Topici’ (La logica Stoica. Testimonianze e frammenti. Testi originali con introduzione e traduzione commentata 5 a), Como: Libreria Noseda. Balme, D. M. (1987), ‘Aristotle’s Use of Division and Differentiae’, in Gotthelf and Lennox (eds.) (1987), 69–89.

295

296

BIBLIOGRAPHY

Baltzly, D. (1997), ‘Plato, Aristotle, and the logos ek tôn pros ti’, Oxford Studies in Ancient Philosophy 15, 177–206. Barford, R. (1976), ‘A Proof in the Peri Ideôn Revisited’, Phronesis 21, 198–218. Barnes, J. (1969), ‘Aristotle’s Theory of Demonstration’, Phronesis 14, 123–152, reprinted as a revised version in Barnes et al. (eds.) (1975), 65–87. Barnes, J. (1975), Aristotle’s Posterior Analytics (Clarendon Aristotle Series), Oxford: Clarendon Press (2nd ed. 1994). Barnes, J. (1976), ‘Aristotle, Menaechmus, and Circular Proof’, Classical Quarterly, 278–292. Barnes, J. (1980), ‘Proof Destroyed’, in Schofield et al. (eds.) (1980), 161–181. Barnes, J. (1981), ‘Proof and the Syllogism’, in Berti (ed.) (1981), 17–59. Barnes, J. (1982), ‘Medicine, Experience, and Logic’, in Barnes et al. (eds.) (1982), 24–68. Barnes, J. (1983), ‘The Beliefs of a Pyrrhonist’, Elenchos 4, 5–43. Barnes, J. (1990), The Toils of Scepticism, Cambridge: Cambridge University Press. Barnes, J. (1991), ‘Galen on Logic and Therapy’, in Kudlien and Durling (eds.) (1991), 50–102. Barnes, J., Schofield, M. and Sorabji, R. (eds.) (1975–1979), Articles on Aristotle I–IV, London: Duckworth. Barnes, J., Brunschwig, J., Burnyeat, M. and Schofield, M. (eds.) (1982), Science and Speculation. Studies in Hellenistic Theory and Practice, Cambridge: Cambridge University Press. Barnes, J., Bobzien, S., Flannery, K. and Ierodiakonou, K. (1991), Alexander of Aphrodisias. On Aristotle Prior Analytics 1.1–7 (Ancient Commentators on Aristotle), London: Duckworth. Barney, R. (1992), ‘Appearances and Impressions’, Phronesis 37, 283–313. Bazán, B. C. (1973), ‘L’authenticité du “De Intellectu” attribué à Alexandre d’Aphrodise’, Révue Philosophique de Louvain 71, 468–487. Berti, E. (1978), ‘The Intellection of “Indivisibles” According to Aristotle, De Anima III 6’, in Lloyd and Owen (eds.) (1978), 141–163. Berti, E. (ed.) (1981), Aristotle on Science. The Posterior Analytics (Proceedings of the Eighth Symposium Aristotelicum), Padova: Antenore. Berti, E. (1997), ‘L’oggetto dell’ eikôs muthos nel Timeo di Platone’, in Calvo and Brisson (eds.) (1997), 119–131. Bett, R. (1994), ‘Aristocles on Timon on Pyrrho’, Oxford Studies in Ancient Philosophy 12, 123–161. Bett, R. (2000a), Pyrrho, His Antecedents and His Legacy, Oxford: Oxford University Press. Bett, R. (2000b), ‘What Does Pyrrhonism Have to Do With Pyrrho?’ in Sihvola (ed.) (2000), 11–33. Block, I. L. (1961), ‘The Order of Aristotle’s Psychological Writings’, American Journal of Philology 82, 50–77. Block, I. L. (1988), ‘Aristotle on the Common Sense: A Reply to Kahn and Others’, Ancient Philosophy 8/2, 235–249. Bluck, R. S. (1956), ‘Logos and Forms: A Reply to Professor Cross’, Mind 65, 522–529. Bluck, R. S. (1957), ‘Hupothesis in the Phaedo and Platonic Dialectic’, Phronesis 2, 21–31. Bluck, R. S. (1963), ‘ “Knowledge by Acquaintance” in Plato’s Theaetetus’, Mind 72, 259–263. Blumenthal, H. J. (1987), ‘Alexander of Aphrodisias in the Later Greek Commentaries on Aristotle’s De Anima’, in Wiesner (ed.) (1987), 90–106. Blumenthal, H. J. (1990a), ‘Neoplatonic Elements in the de Anima Commentaries’, in Sorabji (ed.) (1990b), 305–324.

BIBLIOGRAPHY

297

Blumenthal, H. J. (1990b), ‘Themistius: the Last Peripatetic Commentator on Aristotle?’, in Sorabji (ed.) (1990b), 113–123. Blumenthal, H. J. (1991), ‘Nous pathêtikos in Later Greek Philosophy’, in Blumenthal and Robinson (eds.) (1991), 191–205. Blumenthal, H. J. and Robinson, H. (eds.) (1991), Aristotle and the Later Tradition, (Oxford Studies in Ancient Philosophy suppl. vol.) Oxford: Clarendon Press. Bobzien, S. (1998), Determinism and Freedom in Stoic Philosophy, Oxford: Oxford University Press. Bobzien, S. (2003), ‘Stoic Logic’, in Inwood (ed.) 85–123. Bolton, R. (1976), ‘Essentialism and Semantic Theory in Aristotle: Posterior Analytics, II 7–10’, Philosophical Review 85, 514–544. Bolton, R. (1987), ‘Definition and Scientific Method in Aristotle’s Posterior Analytics and Generation of Animals’, in Gotthelf and Lennox (eds.) (1987), 120–166. Bolton, R. (1990), ‘The Epistemological Basis of Aristotelian Dialectic’, in Devereux and Pellegrin (eds.) (1990), 185–236. Bolton, R. (1991), ‘Aristotle’s Method in Natural Science. Physics I’, in Judson (ed.) (1991), 1–29. Bolton, R. (1993), ‘Aristotle’s Account of the Socratic Elenchus’, Oxford Studies in Ancient Philosophy 11, 121–152. Bolton, R. (1994), ‘The Problem of Dialectical Reasoning (Rskknciq,5p) in Aristotle’, Ancient Philosophy 14, 99–132. Bostock, D. (1986), Plato’s Phaedo, Oxford: Clarendon Press. Boudouris, K. J. (ed.) (1994), Hellenistic Philosophy (Studies in Greek Philosophy 12), Athens: International Center for Greek Philosophy and Culture. Brennan, T. (1994), ‘Criterion and Appearance in Sextus Empiricus’, Bulletin of the Institute of Classical Studies 39, 151–169. (Reprinted in Sihvola (ed.) (2000), 63–92.) Broackes, J. (1999), ‘Aristotle, Objectivity and Perception’, Oxford Studies in Ancient Philosophy 17, 57–113. Broadie, S. (1993), ‘Aristotle’s Perceptual Realism’, Southern Journal of Philosophy 31 (suppl.), 137–159. Broadie, S. (1998), ‘Interpreting Aristotle’s Directions’, in Gentzler (ed.) (1998), 291–306. Brown, L. (1991), ‘Connaissance et réminiscence dans le “Ménon” ’, Revue philosophique 4/1991, 603–619. Brown, L. (1997), Book Review of Scott’s Recollection and Experience, The Philosophical Review 106/2, 270–272. Brunschwig, J. (1967), Topiques I. Livres I–IV (Collection Budé), Paris: Les Belles lettres. Brunschwig, J. (1986), ‘The Cradle Argument in Epicureanism and Stoicism’, in Schofield and Striker (eds.) (1986), 113–144. Brunschwig, J. (1999), ‘Introduction: The Beginnings of Hellenistic Philosophy’, in Algra et al. (eds.), 229–259. Brunschwig, J. and Nussbaum, M. C. (eds.) (1993), Passions and Perceptions. Studies in Hellenistic Philosophy of Mind (Proceedings of the Fifth Symposium Hellenisticum), Cambridge: Cambridge University Press. Burnyeat, M. (1980), ‘Can the Sceptic Live his Scepticism?’ in Schofield et al. (eds.) (1980), 20–53. Burnyeat, M. (1981), ‘Aristotle on Understanding Knowledge’, in Berti (ed.) (1981), 97–139. Burnyeat, M. (1982), ‘Gods and Heaps’, in Schofield and Nussbaum (eds.) (1982), 315–338. Burnyeat, M. (ed.) (1983), The Skeptical Tradition, Berkeley: University of California Press.

298

BIBLIOGRAPHY

Burnyeat, M. (1984), ‘The Sceptic in His Place and Time’, in Rorty et al. (eds.) (1984), 225–254. Burnyeat, M. (1992), ‘Is an Aristotelian Philosophy of Mind Still Credible?’, in Nussbaum and Oksenberg Rorty (eds.) (1992), 15–26. Burnyeat, M. (1993), ‘Aristote voit du rouge et entend un “do”: Combien se passe-t-il de choses? Remarques sur De Anima, II 7–8’, Revue philosophique de la France et de l’étranger 118, 263–280. Burnyeat, M. (1994), ‘Enthymeme. Aristotle on the Logic of Persuasion’, in Furley and Nehamas (eds.) (1994), 3–55. Bynum, T. W. (1993), ‘A New Look at Aristotle’s Theory of Perception’, in Durrant (ed.) (1993), 90–109. Byrne, P. H. (1997), Analysis and Science in Aristotle (SUNY Series in Ancient Greek Philosophy), Albany, NY: State University of New York Press. Calvo, T. and Brisson, L. (eds.) (1997), Interpreting the Timaeus-Critias (Proceedings of the IV Symposium Platonicum), Sankt Augustin: Academia Verlag. Cashdollar, S. (1973), ‘Aristotle’s Account of Incidental Perception’, Phronesis 18, 156–175. Caston, V. (1997), ‘Epiphenomenalisms, Ancient and Modern’, Philosophical Review 106/3, 309–363. Caston, V. (1998), ‘Aristotle and the Problem of Intentionality’, Philosophy and Phenomenological Research 58, 249–298. Caston, V. (1999a), ‘Something and Nothing: The Stoics on Concepts and Universals’, Oxford Studies in Ancient Philosophy 17, 145–213. Caston, V. (1999b), ‘Aristotle’s Two Intellects: A Modest Proposal’, Phronesis 44/3, 199–227. Catan, J. R. (ed.) (1981), Aristotle. The Collected Papers of Joseph Owens, Albany, NY: State University of New York Press. Charles, D. (1994), ‘Aristotle on Names and Their Signification’, in Everson (ed.) (1994), 37–73. Charles, D. (2000), Aristotle on Meaning and Essence (Oxford Aristotle Studies), Oxford: Oxford University Press. Charlton, W. (transl. with an introduction and notes with the assistance of F. Bossier) (1991), Philoponus On Aristotle on the Intellect (Ancient Commentators on Aristotle), London: Duckworth. Cherniss, H. (1944), Aristotle’s Criticism of Plato and the Academy, Baltimore: The Johns Hopkins Press. Cherniss, H. (1971), ‘Review of Saffrey’, in Saffrey (ed.) (1971), 71–79 (originally published in Gnomon 31 (1959), 36–51). Christensen, J. (1962), An Essay on the Unity of Stoic Philosophy, Copenhagen: Munksgaard. Cleary, J. J. (1988), Aristotle on the Many Senses of Priority (The Journal of the History of Philosophy Monograph Series), Carbondale, IL: Southern Illinois University Press. Cleary, J. J. (1995), Aristotle and Mathematics. Aporetic Method in Cosmology and Metaphysics (Philosophia Antiqua 67), Leiden: Brill. Cleary, J. J. (ed.) (1997), The Perennial Tradition of Neo-Platonism, Leuven: Leuven University Press. Cleary, J. J. (ed.) (1999), Traditions of Platonism. Essays in Honour of John Dillon, Aldershot: Ashgate. Cohen, S. M. (1971), ‘The Logic of the Third Man’, Philosophical Review 80, 448–475. Cooper, J. M. (1975), Reason and Human Good in Aristotle, Cambridge, MA: Harvard University Press.

BIBLIOGRAPHY

299

Cooper, J. M. (1987a), ‘Hypothetical Necessity and Natural Teleology’, in Gotthelf and Lennox (eds.) (1987), 243–274. Cooper, J. M. (1987b), ‘Contemplation and Happiness. A Reconsideration’, Synthese 72, 187–216. Cooper, J. M. (ed.) (1997), Plato. Complete Works, Indianapolis, IN: Hackett. Corcoran, J. (1974a), ‘Aristotle’s Natural Deduction Deduction System’, in Corcoran (ed.) (1974b), 85–132. Corcoran, J. (ed.) (1974b), Ancient Logic and Its Modern Interpretations (Synthese Historical Library 9), Dordrecht: Reidel. Corcoran, J. (1983), ‘The Stoicism of the New Academy’, in Burnyeat (ed.) (1983), 31–63 (translated by Jennifer Barnes and M. Burnyeat from the original version, ‘Le Stoïcisme de la Nouvelle Académie’, Revue histoire de la philosophie 3, 241–276). Cornford, F. M. (transl. with a commentary) (1935), Plato’s Cosmology. The Timaeus of Plato, London: Routledge. Couloubaritsis, L. (1980), ‘Y-a-t-il une intuition des principes chez Aristote?’ Revue internationale de philosophie 34 (133–134), 440–489. Crane, T. (1988a), ‘The Waterfall Illusion’, Analysis 48, 142–147. Crane, T. (1988b), ‘Concepts in Perception’, Analysis 48, 150–153. Crick, F. and Koch, C. (1990), ‘Towards a Neurobiological Theory of Consciousness’, Seminars in the Neuro-Sciences 2, 263–275. Crombie, I. M. (1962–1963), An Examination of Plato’s Doctrines (International Library of Philosophy and Scientific Method), London: Routledge. Dancy, R. M. (1975), Sense and Contradiction. A Study in Aristotle (Synthese Historical Library 14), Dordrecht: Reidel. Deichgräber, K. (1930), Die griechische Empirikerschule, Berlin: Weidmann. De Lacy, P. H. and De Lacy, E. A. (1978), Philodemus. On Methods of Inference, Napoli: Bibliopolis. De Pater, W. A. (1968), ‘La fonction du lieu et de l’instrument dans les Topiques’, in Owen (ed.) (1968), 164–188. Devereux, D. (1978), ‘Nature and Teaching in Plato’s Meno’, Phronesis 23, 118–126. Devereux, D. and Pellegrin, P. (eds.) 1990, Biologie, logique et Métaphysique chez Aristote, Paris: Centre national de la recherche scientifique. Di Liscia, D., Kessler, E. and Methuen, C. (eds.) (1997), Method and Order in Renaissance Philosophy of Nature. The Aristotle Commentary Tradition, Aldershot: Ashgate. Dillon, J. (1977), The Middle Platonists, 80 BC to AD 220, Ithaca, NY: Cornell University Press. Dillon, J. (1989), ‘Tampering with Timaeus: Ideological Emendations in Plato with Special Reference to the Timaeus’, American Journal of Philology 110, 50–72. Dillon, J. (transl. with an introduction and commentary) (1993), Alcinous. The Handbook of Platonism, Oxford: Oxford University Press. Donini, P. -L. (1994), ‘Testi e commenti, manuali e insegnamento: la forma sistematica e i metodi della filosofia in età postellenistica’, Aufstieg und Niedergang der römischen Welt Bd 36.7, Berlin: de Gruyter, 5027–5100. Dooley, W. E. (transl.) (1989), Alexander of Aphrodisias. On Aristotle’s Metaphysics 1, Ithaca, NY: Cornell University Press. Dooley, W. E. and Madigan, A. (transl.) (1992), Alexander of Aphrodisias. On Aristotle Metaphysics 2 & 3 (Ancient Commentators on Aristotle), London: Duckworth. Dorion, L. -A. (1995), Aristote: Les réfutations sophistique (Histoire des doctrines de l’ antiquité classique) Paris: Vrin.

300

BIBLIOGRAPHY

Durrant, M. (ed.) (1993), Aristotle’s De Anima in Focus, London: Routledge. Düring, I. and Owen, G. E. L. (eds.) (1960), Aristotle and Plato in the Mid-Fourth Century (Symposium Aristotelicum 1957), Göteborg: Almqvist. Eck, J. van (1996), ‘Resailing Socrates’ deuteros plous: A Criticism of Rowe’s “Explanation in Phaedo 99c6–102a8” ’, Oxford Studies in Ancient Philosophy 14, 211–226. Ellis, J. (1990), ‘The Trouble with Fragrance’, Phronesis 35/3, 290–302. Emilsson, E. K. (1988), Plotinus on Sense-Perception: A Philosophical Study, Cambridge: Cambridge University Press. Emilsson, E. K. (1996), ‘Cognition and Its Object’, in Gerson (ed.) (1996), 217–249. Emilsson, E. K. (2003), ‘Discursive and Non-Discursive Thought’, in Larsen et al. (eds.), 47–65. Emilsson, E. K. (forthcoming 2007), Plotinus on Intellect, Oxford: Oxford University Press. Engberg-Pedersen, T. (1979), ‘More on Aristotelian Epagoge’, Phronesis 24, 301–319. Engberg-Pedersen, T. (1990), The Stoic Theory of Oikeiosis. Moral Development and Social Interaction in Early Stoic Philosophy (Studies in Hellenistic Sivilization), Aarhus: Aarhus University Press. Engberg-Pedersen, T. and Sihvola, J. (eds.) (1998), The Emotions in Hellenistic Philosophy, Dordrecht: Kluwer. Evans, G. and McDowell, J. (eds.) (1976), Truth and Meaning: Essays in Semantics, Oxford: Oxford University Press. Evans, J. D. G. (1977), Aristotle’s Concept of Dialectic, Cambridge: Cambridge University Press. Everson, S. (1990a), ‘Epicurus on the Truth of the Senses’, in Everson (ed.) (1990), 161–183. Everson, S. (ed.) (1990b), Companions to Ancient Thought. Epistemology I, Cambridge: Cambridge University Press. Everson, S. (ed.) (1991), Companions to Ancient Thought. Psychology II, Cambridge: Cambridge University Press. Everson, S. (ed.) (1994), Companions to Ancient Thought. Language III, Cambridge: Cambridge University Press. Everson, S. (1997), Aristotle on Perception, Oxford: Oxford University Press. Ferejohn, M. (1991), The Origins of Aristotelian Science, New Haven, CT: Yale University Press. Fine, G. (1979), ‘False Belief in the Theaetetus’, Phronesis 24, 70–80. Fine, G. (1990), ‘Knowledge and Belief in Republic 5–7’, in Everson (ed.) (1990b), 85–115. Fine, G. (1993), On Ideas. Aristotle’s Criticism of Plato’s Theory of Forms, Oxford: Clarendon Press. Fine, G. (ed.) (1999), Plato 1. Metaphysics and Epistemology (Oxford Readings in Philosophy), Oxford: Clarendon Press. Frede, D. (1978), ‘The Final Proof of the Immortality of the Soul in Plato’s Phaedo 102a–107a’, Phronesis 23, 27–41. Frede, D. (1992), ‘The Cognitive Role of Phantasia in Aristotle’, in Nussbaum and Oksenberg Rorty (eds.) (1992), 279–295. Frede, D. (1993), Philebus, Indianapolis, IN: Hackett. Frede, D. (1996), ‘The Philosophical Economy of Plato’s Psychology: Rationality and Common Concepts in the Timaeus’, in Frede and Striker (eds.) (1996), 29–58. Frede, M. (1974), Die Stoische Logik (Abhandlungen der Akademie der Wissenschaften in Göttingen. 3; Phil.-hist. Kl. 88), Göttingen. Frede, M. (1979), ‘Des Skeptikers Meinungen’, Neue Hefte für Philosophie 15/16, 102–129 (an English translation ‘The Skeptic’s Beliefs’ in Frede (1987e), 179–200). Frede, M. (1983), ‘Stoics and Skeptics on Clear and Distinct Impressions’ in Burnyeat (ed.) (1983), reprinted in Frede (1987e), 151–178.

BIBLIOGRAPHY

301

Frede, M. (1987a), ‘On Galen’s Epistemology’, in Frede (1987e), 279–298 (originally published in Nutton (ed.) (1981)). Frede, M. (1987b), ‘Stoic versus Aristotelian Syllogistic’, in Frede (1987e), 99–124 (originally published in Archiv für Geschichte der Philosophie 56). Frede, M. (1987c), ‘The Original Notion of Cause’, in Frede (1987e), 125–150. Frede, M. (1987d), ‘Philosophy and Medicine in Antiquity’, in Frede (1987e), 225–242. Frede, M. (1987e), Essays in Ancient Philosophy, Minneapolis, MN: University of Minnesota Press. Frede, M. (1988), ‘The Empiricist Attitude towards Reason and Theory’, Apeiron suppl. vol. 21/2, 79–97. Frede, M. (1994), ‘The Stoic Conception of Reason’, in Boudouris (ed.) (1994), 50–63. Frede, M. (1996a), ‘Aristotle’s Rationalism’, in Frede and Striker (eds.) (1996), 157–173. Frede, M. (1996b), ‘La théorie aristotélicienne de l’intellect agent’, in Romeyer-Dherbey (dir.), 377–390. Frede, M. (1999), ‘Stoic Epistemology’, in Algra et al. (eds.), 295–322. Frede, M. and Patzig, G. (1988), Aristoteles ‘Metaphysik Z’. Text, Übersetzung, Kommentar, München: Beck. Frede, M. and Striker, G. (eds.) (1996), Rationality in Greek Thought, Oxford: Oxford University Press. Furley, D. (1967), Two Studies in the Greek Atomists, Princeton: Princeton University Press. Furley, D. (1987), The Greek Cosmologists I. The Formation of the Atomic Theory and its Earliest Critics, Cambridge: Cambridge University Press. Furley, D. (1989), Cosmic Problems. Essays on Greek and Roman Philosophy of Nature, Cambridge: Cambridge University Press. Furley, D. (1993), ‘Democritus and Epicurus on Sensible Qualities’, in Brunschwig and Nussbaum (eds.) (1993), 72–94. Furley, D. and Nehamas, A. (eds.) (1994), Aristotle’s Rhetoric. Philosophical Essays, Princeton: Princeton University Press. Gallop, D. (1975), Plato: Phaedo, Oxford: Clarendon Press. Gentzler, J. (1991), ‘Sumphônein in Plato’s Phaedo’, Phronesis 36, 265–276. Gentzler, J. (ed.) (1998), Method in Ancient Philosophy, Oxford: Clarendon Press. Gerson, L. P. (1990), God and Greek Philosophy. Studies in the Early History of Natural Theology, London: Routledge. Gerson, L. P. (ed.) (1996), The Cambridge Companion to Plotinus, Cambridge: Cambridge University Press. Gerson, L. P. (1999), ‘The Concept in Platonism’, in Cleary (ed.) (1999), 65–80. Gill, C. (1998), ‘Did Galen Understand Platonic and Stoic Thinking on Emotions?’, in Engberg-Pedersen and Sihvola (eds.), 113–148. Glucker, J. (1995), ‘Probabile, Veri Simile and Related Terms’, in Powell (ed.) (1995), 115–143. Goldin, O. (1989), ‘Problems with Graham’s Two-Systems Hypothesis’, Oxford Studies in Ancient Philosophy 7, 203–213. Goldin, O. (1996), Explaining an Eclipse, Ann Arbor: University of Michigan Press. Goldman, A. I. (1986), Epistemology and Cognition, Cambridge, MA: Harvard University Press. Gonzalez, F. J. (1998), Dialectic and Dialogue. Plato’s Practice of Philosophical Inquiry, Evanston, IL: Northwestern University Press. Gotthelf, A. (1987), ‘First Principles in Aristotle’s Parts of Animals’, in Gotthelf and Lennox (eds.) (1987), 167–198. Gotthelf, A. and Lennox, J. (eds.) (1987), Philosophical Issues in Aristotle’s Biology, Cambridge: Cambridge University Press.

302

BIBLIOGRAPHY

Graeser, A. (ed.) (1987), Mathematics and Metaphysics in Aristotle/Mathematik und Metaphysik bei Aristoteles (Akten des X Symposium Aristotelicum 1984), Bern: Haupt. Graf, D. and Williamson, T. (eds.) (2002), Vagueness (The International Research Library of Philosophy 27), Aldershot: Ashgate. Graham, D. (1987), Aristotle’s Two Systems, Oxford: Clarendon Press. Graham, D. (1989), ‘Two Systems in Aristotle’, Oxford Studies in Ancient Philosophy 7, 215–231. Green-Pedersen, N. J. (1984), The Tradition of the Topics in the Middle Ages. The Commentaries on Aristotle’s and Boethius’ Topics, München: Philosophia. Gregory, A. (2000), Plato’s Philosophy of Science, London: Duckworth. Haack, S. (1993), Evidence and Inquiry. Towards Reconstruction in Epistemology, Oxford: Blackwell. Haaparanta, L., Kusch, M. and Niiniluoto, I. (eds.) (1990), Language, Knowledge, and Intentionality: Perspectives on the Philosophy of Jaakko Hintikka (Acta Philosophica Fennica 49), Helsinki: The Philosophical Society of Finland. Haas, Frans de (2002), ‘Modifications of the Method of Inquiry in Aristotle’s Physics I 1: An Essay on the Dynamics of the Commentary Tradition’, in Leijenhorst et al. (eds.), 31–56. Hadot, I. (1990), ‘The Life and Work of Simplicius in Greek and Arabic Sources’, in Sorabji (ed.) (1990b), 275–303. Hagen, C. (1994), Simplicius. On Aristotle Physics 7 (Ancient Commentators on Aristotle), London: Duckworth. Hahm, D. E. (1977), The Origins of Stoic Cosmology, Columbus, OH: Ohio State University Press. Hamlyn, D. W. (1955), ‘The Communion of Forms and the Development of Plato’s Logic’, Philosophical Quarterly 5, 289–302. Hamlyn, D. W. (1961), Sensation and Perception. A History of the Philosophy of Perception (International Library of Philosophy and Scientific Method), London: Routledge. Hamlyn, D. W. (1976), ‘Aristotelian Epagoge’, Phronesis 21, 167–184. Hamlyn, D. W. (1993), Aristotle De Anima. Books II and III with Passages from Book I (Clarendon Aristotle Series), Oxford: Clarendon Press. Hankinson, R. J. (1987), ‘Evidence, Externality and Antecedence: Inquiries into Later Greek Concepts’, Phronesis 32/1, 80–100. Hankinson, R. J. (transl. with introduction and commentary) (1991a), Galen on the Therapeutic Method Books I and II, Oxford: Clarendon Press. Hankinson, R. J. (1991b), ‘Greek Medical Models of Mind’, in Everson (ed.) (1991), 194–217. Hankinson, R. J. (1998a), Cause and Explanation in Ancient Greek Thought, Oxford: Oxford University Press. Hankinson, R. J. (ed. with an introduction, translation and commentary) (1998b), Galen: On Antecedent Causes (Cambridge Classical Texts and Commentaries 35), Cambridge: Cambridge University Press. Hardie, W. F. R. (1980), Aristotle’s Ethical Theory, Oxford: Clarendon Press. Harlfinger, D. (1975), see Leszl (1975). Hart, W. D. (1992), ‘Hat-Tricks and Heaps’, Philosophical Studies 33, 1–24 (reprinted in Graf and Williamson (eds.) (2002), 253–276). Hayduck, M. (1891), Alexandri Aphrodisiensis in Aristotelis Metaphysica commentaria (Commentaria in Aristotelem Graeca I), Berlin: Reimer. Heath, T. (1949), Mathematics in Aristotle (reprinted 1998 in the Classic Studies in the History of Ideas), Bristol: Thoemmes.

BIBLIOGRAPHY

303

Heath, T. (transl. with introduction and commentary) (1956), The Thirteen Books of Euclid’s Elements, New York: Dover. Heinaman, R. (1990), ‘Aristotle and the Mind-Body Problem’, Phronesis 35, 83–102. Hintikka, J. (1972), ‘On the Ingredients of an Aristotelian Science’, Nous 6, 55–69. Hintikka, J. (1973), Time and Necessity. Studies in Aristotle’s Theory of Modality, Oxford: Oxford University Press. Hintikka, J. (1974), Knowledge and the Known. Historical Perspectives in Epistemology (Synthese Historical Library 11), Dordrecht: Reidel. Hintikka, J. (1980), ‘Aristotelian Induction’, Revue internationale de philosophie 34 (133–134), 422–439. Hintikka, J. (1993), ‘Commentary on Smith’, Proceedings of the Boston Area Colloquium in Ancient Philosophy 9, 186–195. Hintikka, J. and Remes, U. (1974), The Method of Analysis. Its Geometrical Origin and Its General Significance (Synthese Library 75), Dordrecht: Reidel. Hintikka, J., Remes, U. and Knuuttila, S. (1977), Aristotle on Modality and Determinism (Acta Philosophica Fennica 29/1), Helsinki: Philosophical Society of Finland. Inwood, B. (ed.) (2003), The Cambridge Companion to the Stoics, Cambridge: Cambridge University Press. Irwin, T. H. (1977), ‘Plato’s Heracliteanism’, Philosophical Quarterly 27, 1–13. Irwin, T. H. (1987), ‘Ways to First Principles: Aristotle’s Methods of Discovery’, Philosophical Topics 15/2, 109–134. Irwin, T. H. (1988), Aristotle’s First Principles, Oxford: Clarendon Press. Isnardi-Parente, M. (1987), ‘TA META TAS IDEAS: Figures idéales ou premières figures?’ in Graeser (ed.) (1987), 261–280. Jaeger, W. (1948), Aristotle. Fundamentals of the History of His Development (the German original Aristoteles, Grundlegung einer Geschichte seiner Entwicklung 1923), Oxford: Clarendon Press. Johnson, M. R. (2005), Aristotle on Teleology, Oxford: Oxford University Press. Judson, L. (ed.) (1991), Aristotle’s Physics. A Collection of Essays, Oxford: Clarendon Press. Kahn, C. H. (1966), ‘Sensation and Consciousness in Aristotle’s Psychology’, Archiv für Geschichte der Philosophie 48, 43–81. Kahn, C. H. (1973), The Verb ‘Be’ in Ancient Greek (Foundations of Language Supplementary Series 16), Dordrecht: Reidel. Kahn, C. H. (1981), ‘The Role of Nous in the Cognition of the First Principles in the Posterior Analytics II 19’, in Berti (ed.) (1981), 385–414. Kahn, C. H. (1992), ‘Aristotle on Thinking’, in Nussbaum and Rorty (ed.), 359–379. Kakkuri-Knuuttila, M. -L. (1993), Dialectic and Inquiry in Aristotle, Helsinki: Helsinki School of Economics. Kakkuri-Knuuttila, M. -L. (2005), ‘The Relevance of Dialectical Skills to Philosophical Inquiry in Aristotle’, Rhizai: A Journal for Ancient Philosophy and Science II 1, 31–74. Kakkuri-Knuuttila, M. -L. and Knuuttila, S. (1990), ‘Induction and Conceptual Analysis in Aristotle’, in Haaparanta et al. (eds.) (1990), 294–303. Kal, V. (1988), On Intuition and Discursive Reasoning in Aristotle (Philosophia Antiqua 46), Leiden: Brill. Kanayama, Y. (2000), ‘The Methodology of the Second Voyage and the Proof of the Soul’s Indestructibility in Plato’s Phaedo’, Oxford Studies in Ancient Philosophy 18, 41–100. Kapp, E. (1931), ‘Syllogistik’, in Real-Encyclopädie der klassischen Altertumswissenschaft IV A, 1046–1067. (A shortened version ‘Syllogistic’, in Barnes et al. (eds.) (1975), 35–49.)

304

BIBLIOGRAPHY

Kapp, E. (1942), Greek Foundations of Traditional Logic (Columbia Studies in Philosophy 5), New York: Columbia University Press. Kieffer, J. S. (1964), Galen’s Institutio Logica. English Translation, Introduction, and Commentary, Baltimore: The Johns Hopkins Press. Kirwan, C. A. (1971), Aristotle’s Metaphysics, Books IV, V, VI (Clarendon Aristotle Series), Oxford: Clarendon Press. Klosko, G. (1983), ‘Criteria of Fallacy and Sophistry for Use in the Analysis of Platonic Dialogues’, Classical Quarterly 33/1, 363–374. Kneale, W. and Kneale, M. (1962), The Development of Logic, Oxford: Oxford University Press. Knuuttila, S. (ed.) (1981), Reforging the Great Chain of Being. Studies in the History of Modal Theories (Synthese Historical Library 20), Dordrecht: Reidel. Knuuttila, S. (1993a), ‘Remarks on Induction in Aristotle’s Dialectic and Rhetoric’, Revue internationale de philosophie 47/184, 78–88. Knuuttila, S. (1993b), Review on van Rijen’s Aspects of Aristotle’s Modal Theory (1989), Nous 27, 553–555. Knuuttila, S. (1993c), Modalities in Medieval Philosophy, London: Routledge. Knuuttila, S. (transl. with introduction and notes) (1994), Aristoteles. Ensimmäinen analytiikka, Helsinki: Gaudeamus. Knuuttila, S. and Sihvola, J. (2000), ‘Ancient Scepticism and Philosophy of Religion’, in Sihvola (ed.) (2000), 125–144. Kosman, L. A. (1973), ‘Understanding, Explanation and Insight in the Posterior Analytics’, in Lee et al. (eds.) (1973), 374–392. Kosman, L. A. (1992), ‘What does the Maker Mind Make?’ in Nussbaum and Rorty (eds.) (1992), 343–358. Kraut, R. (ed.) (1992), The Cambridge Companion to Plato, Cambridge: Cambridge University Press. Kudlien, F. and Durling, R. J. (eds.) (1991), Galen’s Method of Healing (Studies in Ancient Medicine 1), Leiden: Brill. Lammenranta, M. (1996), ‘Reliabilism and Circularity’, Philosophy and Phenomenological Research 56, 111–124. Lane, M. (1998), Method and Politics in the Statesman (Cambridge Classical Studies), Cambridge: Cambridge University Press. Larsen, T. M., Fossheim, H. and Sageng, J. R. (eds.) (2003), Non-Conceptual Aspects of Experience, Oslo: Unipubforlag. Lear, J. (1982), ‘Aristotle’s Philosophy of Mathematics’, Philosophical Review 91, 161–192. Lear, J. (1988), Aristotle: The Desire to Understand, Cambridge: Cambridge University Press. Le Blond, J. M. (1939), Logique et Méthode chez Aristote. Etude sur la recherche des principes dans la physique aristotélicienne, Paris: Vrin. Lee, E. N., Mourelatos, A. and Rorty, R. (eds.) (1973), Exegesis and Argument. Studies in Greek Philosophy Presented to Gregory Vlastos (Phronesis Suppl. vol. 1), Assen: Van Gorcum. Lee, M. M. (2005), Epistemology after Protagoras: Responses to Relativism in Plato, Aristotle, and Democritus, Oxford: Clarendon Press. Leijenhorst, C. H., Lüthy, C. H. and Thijssen, J. M. M. H. (eds.), The Dynamics of the Aristotelian Natural Philosophy from Antiquity to the Seventeenth Century (Medieval and Early Modern Science), Leiden: Brill. Lennox, J. G. (1987), ‘Divide and Explain. The Posterior Analytics in Practice’, in Gotthelf and Lennox (eds.) (1987), 90–119 reprinted in Lennox (2001b).

BIBLIOGRAPHY

305

Lennox, J. G. (2001a), ‘Plato’s Unnatural Teleology’, in Lennox (2001c) (originally appeared in D. O’Meara (ed.), Platonic Investigations, Washington: Catholic University Press of America (1985), 195–218. Lennox, J. G. (2001b), Aristotle’s Biology: Studies in the Origins of Life Science (Studies in Biology and Philosophy), Cambridge: Cambridge University Press. Leroux, G. (1974), ‘Logique et dialectique chez Plotin’, Phoenix 28, 180–192. Lesher, J. H. (1969), ‘Gnôsis and Epistêmê in Socrates’ Dream in the Theaetetus’, Journal of Hellenic Studies 89, 72–78. Lesher, J. H. (1973), ‘The Meaning of Nous in the Posterior Analytics’, Phronesis 18, 44–68. Leszl, W. (1975), Il ‘De ideis’di Aristotele e la teoria Platonica delle idee (Accademia Toscana di scienze e lettere ‘La Colombaria’ 40), Firenze: Olschki. Leszl, W. (1980), ‘Unity and Diversity of the Sciences: The Methodology of the Mathematical and of the Physical Sciences and the Role of Nominal Definition’, Revue internationale de philosophie 34/133–134, 384–421. Lindberg, D. C. (1976), Theories of Vision from al-Kindi to Kepler (University of Chicago History of Science and Medicine), Chicago: University of Chicago Press. Lloyd, A. C. (1969–1970), ‘Non-Discursive Thought – An Enigma of Greek Philosophy’, Proceedings of the Aristotelian Society 70, 261–274. Lloyd, A. C. (1981), ‘Necessity and Essence in the Posterior Analytics’, in Berti (ed.) (1981), 157–171. Lloyd, A. C. (1990), The Anatomy of Neoplatonism, Oxford: Clarendon Press. Lloyd, G. E. R. (1979), Magic, Reason and Experience: Studies in the Origins and Development of Greek Science, Cambridge: Cambridge University Press. Lloyd, G. E. R. (1990), ‘The Theories and Practices of Demonstration in Aristotle’, The Proceedings of the Boston Area Colloquium in Ancient Philosophy VI, 371–401. Lloyd, G. E. R. (1996), ‘Demonstration in Galen’, in Frede and Striker (eds.) (1996), 255–277. Lloyd, G. E. R. and Owen, G. E. L. (eds.) (1978), Aristotle on Mind and the Senses (Proceedings of the Seventh Symposium Aristotelicum), Cambridge: Cambridge University Press. Long, A. A. (1971a), ‘Aisthêsis, Prolêpsis and Linguistic Theory in Epicurus’, Bulletin of Institute of Classical Studies 18, 114–133. Long, A. A. (ed.) (1971b), Problems in Stoicism, London: Athlone. Long, A. A. (1986), Hellenistic Philosophy. Stoics, Epicureans, Sceptics, Berkeley: University of California Press. Long, A. A. and Sedley, D. (1987), The Hellenistic Philosophers. Translations of the Principal Sources with Philosophical Commentary, Cambridge: Cambridge University Press. Lyons, J. (1963), Structural Semantics: An Analysis of Part of the Vocabulary of Plato, Oxford: Blackwell. Madigan, A. (1993), Alexander of Aphrodisias. On Aristotle Metaphysics 4 (Ancient Commentators on Aristotle), London: Duckworth. Mahoney, E. P. (1973), ‘Themistius and the Agent Intellect in James of Viterbo and Other Thirteenth-Century Philosophers: Saint Thomas Aquinas, Siger of Brabant and Henry Bate’, Augustiniana 23, 422–467. Mahoney, E. P. (1982), ‘Neoplatonism, the Greek Commentators and Renaissance Aristotelianism’, in O’Meara (ed.) (1982), 169–177. Mansion, S. (ed.) (1961), Aristote et les problèmes de methode, Louvain: Publications universitaires. Mates, B. (1961), Stoic Logic, Berkeley: University of California Press.

306

BIBLIOGRAPHY

McCabe, M. M. (1994), Plato’s Individuals, Princeton: Princeton University Press. McDowell, J. (1973), Plato. Theaetetus (Clarendon Plato Series), Oxford: Oxford University Press. McGinn, C. (1999a), ‘The Appearance of Colour’, in McGinn (1999b), 314–325. McGinn, C. (1999b), Knowledge and Reality, Oxford: Oxford University Press. McKirahan, R. D. Jr. (1992), Principles and Proofs. Aristotle’s Theory of Demonstrative Science, Princeton: Princeton University Press. Mellor, D. H. (1988), ‘Crane’s Waterfall Illusion’, Analysis 48, 147–150. Menn, S. (1992), ‘Aristotle and Plato on God as Nous and as the Good’, Review of Metaphysics 45, 543–573. Menn, S. (1999), ‘The Stoic Theory of Categories’, Oxford Studies in Ancient Philosophy 17, 215–247. Mignucci, M. (1969), Aristotele. Gli Analitici primi. Traduzione, introduzione e commento (Filosofi antichi 3), Napoli: Loffredo. Mignucci, M. (1975), L’Argomentazione Dimostrativa in Aristotele. Commento agli Analitici secondi, Padova: Antenore. Mignucci, M. (1981), ‘Hôs epi to polu et nécessaire dans la conception aristotélicienne de la science’, in Berti (ed.) (1981), 173–203. Mikkeli, H. (1992), An Aristotelian Response to Renaissance Humanism. Jacopo Zabarella on the Nature of Arts and Sciences (Studia Historica 41), Helsinki: The Historical Society of Finland. Modrak, D. K. W. (1987), Aristotle. The Power of Perception, Chicago, IL: The University of Chicago Press. Modrak, D. K. W. (1990–1991), ‘The Nous-Body Problem in Aristotle’, Review of Metaphysics 44, 755–774. Moraux, P. (1942), Alexandre d’Aphrodise. Exégète de la Noétique d’Aristote (Bibliothèque de la faculté de philosophie et lettres de l’université de Liège 99), Paris: Faculté de philosophie et lettres. Moraux, P. (1968), ‘La joute dialectique d’après le huitième livre des Topiques’, in Owen (ed.) (1968), 277–311. Moraux, P. (1979), Le Commentaire d’Alexandre d’Aphrodise aux ‘Secondes Analytiques’ d’Aristote (Peripatoi 13), Berlin: de Gruyter. Moravcsik, J. (1991), ‘What makes Reality Intelligible? Reflections on Aristotle’s Theory of Aitia’, in Judson (ed.) (1991), 31–47. Morgan, M. L. (1984), ‘Sense Perception and Recollection in the Phaedo’, Phronesis 29, 237–251. Morrison, D. (1997), ‘Philoponus and Simplicius on Tekmeriodic Proof’, in Di Liscia et al. (eds.) (1997), 1–22. Morrow, G. (transl. with introduction and notes) (1970), Proclus Diadochus. A Commentary on the First Book of Euclid’s Elements, Princeton: Princeton University Press. Mourelatos, A. P. D. (1970), The Route of Parmenides. A Study of Word, Image, and Argument in the Fragments, New Haven, CT: Yale University Press. Mueller, I. (1969), ‘Stoic and Peripatetic Logic’, Archiv für Geschichte der Philosophie 51, 173–187. Mueller, I. (1981), Philosophy of Mathematics and Deductive Structure in Euclid’s Elements, Cambridge, MA: MIT Press. Mueller, I. (1987), ‘Aristotle’s Approach to the Problem of Principles in Metaphysics M and N’, in Graeser (ed.) (1987), 241–259. Mueller, I. (1990), ‘Aristotle’s Doctrine of Abstractionism in the Commentators’, in Sorabji (ed.) (1990b), 463–480.

BIBLIOGRAPHY

307

Niiniluoto, I. (2000), ‘Scepticism, Fallibilism, and Verisimilitude’, in Sihvola (ed.) (2000), 145–169. Nuchelmans, G. (1973), Theories of the Proposition. Ancient and Medieval Conceptions of the Bearers of Truth and Falsity (North-Holland Linguistic Series), Amsterdam: NorthHolland Publishing Company. Nussbaum, M. C. (1978), Aristotle’s De Motu Animalium, Princeton: Princeton University Press. Nussbaum, M. C. (1982), ‘Saving Aristotle’s Appearances’, in Schofield and Nussbaum (eds.) (1982), 267–293. Nussbaum, M. C. (1986), The Fragility of Goodness. Luck and Ethics in Greek Tragedy and Philosophy, Cambridge: Cambridge University Press. Nussbaum, M. C. (1994), The Therapy of Desire. Theory and Practice in Hellenistic Ethics, Princeton: Princeton University Press. Nussbaum, M. C. and Oksenberg Rorty, A. (eds.) (1992), Essays on Aristotle’s De Anima, Oxford: Oxford University Press. Nutton, V. (ed.) (1981), Galen. Problems and Prospects, London: Wellcome Institute. Nuyens, F. (1948), L’Evolution de la psychologie d’Aristote, Louvain: Editions de l’institut supérior de philosophie. Obbink, D. (1992), ‘ “What All Men Believe – Must Be True”: Common Conceptions and Consensio Omnium in Aristotle and Hellenistic Philosophy’, Oxford Studies in Ancient Philosophy 10, 193–231. O’Meara, D. J. (1982) (ed.), Neoplatonism and Christian Thought (Studies in Neoplatonism), Albany, NY: State University of New York Press. O’Meara, D. J. (2001), ‘Intentional Objects in Later NeoPlatonism’, in Perler (ed.), 115–125. Owen, G. E. L. (1957), ‘A Proof in the Peri Ideôn’, Journal of Hellenic Studies 77, 103–111 (in Logic, Science and Dialectic, 165–179). Owen, G. E. L. (1960), ‘Logic and Metaphysics in Some Earlier Works of Aristotle’, Düring and Owen (eds.) (1960), 163–290 (in Logic, Science and Dialectic, 180–200). Owen, G. E. L. (1961), ‘Tithenai ta phainomena’, in Mansion (ed.) (1961), 83–103 (in Logic, Science and Dialectic, 239–251). Owen, G. E. L. (1965), ‘Inherence’, Phronesis 10, 97–105 (in Logic, Science, and Dialectic, 252–258). Owen, G. E. L. (1966), ‘The Platonism of Aristotle’, Proceedings of the British Academy 51, 125–50 (in Logic, Science and Dialectic, 200–220). Owen, G. E. L. (1968), ‘Dialectic and Eristic in the Treatment of the Forms’, in Owen 1968, 103–125 (in Logic, Science and Dialectic, 221–238). Owen, G. E. L. (ed.) (1968), Aristotle on Dialectic. The Topics (Proceedings of the Third Symposium Aristotelicum), Oxford: Clarendon Press. Owen, G. E. L. (1970), ‘Plato on Not-Being’, in Vlastos (ed.) (1970), 223–267 (in Logic, Science and Dialectic, 104–137). Owen, G. E. L. (1986), in M. C. Nussbaum (ed.) Logic, Science and Dialectic. Collected Papers in Greek Philosophy, London: Duckworth. Owens, J. (1964), ‘The Aristotelian Conception of the Sciences’, International Philosophical Quarterly 4, 200–216. Patzig, G. (1968), Aristotle’s Theory of the Syllogism (Synthese Library), Dordrecht: Reidel (German original Die Aristotelische Syllogistik 1963). Patzig, G. (1975), ‘Theology and Ontology in Aristotle’s Metaphysics’, in Barnes et al. (eds.) (1975), 33–49. Peck, A. L. (1962), ‘Plato’s Sophist: the sumplokê tôn eidôn’, Phronesis 7, 46–66.

308

BIBLIOGRAPHY

Pellegrin, P. (1982), Classification des animaux chez Aristote: statu de la biologie et unité d’aristotélisme (Collection d’études anciennes), Paris: Les Belles lettres. Perler, D. (ed.) (2001), Ancient Theories of Intentionality (Studien und Texte zur Geistesgeschichte des Mittelalters), Leiden: Brill. Philip. J. A. (1969), ‘The Megista Gene of the Sophistes’, Phoenix 23, 89–103. Pinborg, J. (1975), ‘Historiography of Linguistics: Classical Antiquity: Greece’, Current Trends of Linguistics 13, 69–126. Powell, J. G. F. (ed.) (1995), Cicero the Philosopher: Twelve Papers, Oxford: Clarendon Press. Primavesi, O. (1996), Die Aristotelische Topik: Ein Interpretationsmodell und seine Erprobung am Beispiel von Topik B (Zetemata: Monographien zur klassischen Altertumswissenshaft Heft 94), München: Beck. Prior, W. J. (1985), Unity and Development in Plato’s Metaphysics, London: Croom Helm. Remes, P. (forthcoming 2007), Plotinus on Self: The Philosophy of the ‘We’ Cambridge: Cambridge University Press. Rijen, J. van (1989), Aspects of Aristotle’s Logic of Modalities (Synthese Historical Library 35), Dordrecht: Kluwer. Rich, A. N. M. (1954), ‘The Platonic Ideas as Thoughts of God’, Mnemosyne 4/7, 123–133. Rist, J. M. (1972), Epicurus: An Introduction, Cambridge: Cambridge University Press. Robinson, H. (1991), ‘Form and the Immateriality of the Intellect from Aristotle to Aquinas’, in Blumenthal and Robinson (eds.) (1991), 207–226. Robinson, R. (1941), Plato’s Earlier Dialectic, Ithaca, NY: Cornell University Press. (2. revised edition 1953.) Rohr, M. (1981), ‘Empty Forms in Plato’, in Knuuttila (ed.) (1981), 19–56. Romeyer-Dherbey, G. (dir.) and Viano, C. (ed.) (1996), Corps et âme: Sur le De anima d’Aristote (Bibliothèque d’histoire de la philosophie), Paris: Vrin. Rorty, R., Schneewind, J. B. and Skinner, O. (eds.) (1984), Philosophy in History. Essays on the Historiography of Philosophy, Cambridge: Cambridge University Press. Ross, W. D. (1923), Aristotle, London: Methuen. Ross, W. D. (1949), Aristotle’s Prior and Posterior Analytics. A Revised Text with Introduction and Commentary, Oxford: Clarendon Press. Ross, W. D. (1951), Plato’s Theory of Ideas, Oxford: Clarendon Press. Ross, W. D. (1953), Aristotle’s Metaphysics. A Revised Text with Introduction and Commentary, Oxford: Oxford University Press (1. ed. 1924). Rowe, C. (1979), ‘The Proof from Relatives in the Peri Ideon: Further Reconsiderations’, Phronesis 24, 270–281. Rowe, C. (1993), ‘Explanation in Phaedo 99c6–102a8’, Oxford Studies in Ancient Philosophy 11, 49–69. Rowe, C. (1996), ‘A Reply to van Eck’, Oxford Studies in Ancient Philosophy 14, 227–240. Runciman, W. G. (1962), Plato’s Later Epistemology (Cambridge Classical Studies), Cambridge: Cambridge University Press. Ryle, G. (1939), ‘Plato’s Parmenides’, Mind 48, 129–151 and 302–325. Ryle, G. (1968), ‘Dialectic in the Academy’, in Owen (ed.) (1968), 69–79. Saffrey, H. D. (ed.) (1971), Le Peri Philosohias d’Aristote at la théorie platonicienne des idées nombres (Philosophia Antiqua 7), Leiden: Brill. Sandbach, F. H. (1971), ‘Ennoia and Prolêpsis in the Stoic Theory of Knowledge’, in Long (ed.) (1971b), 22–37. Sandbach, F. H. (1985), Aristotle and the Stoics (Proceedings of the Cambridge Philological Society, suppl. 10). Schofield, M. (1978), ‘Aristotle on the Imagination’, in Lloyd and Owen (eds.), 99–140.

BIBLIOGRAPHY

309

Schofield, M. (1980), ‘Preconception, Argument and God’, in Schofield et al. (eds.) (1980), 283–308. Schofield, M. and Nussbaum, M. C. (eds.) (1982), Language and Logos in Ancient Greek Philosophy, Cambridge: Cambridge University Press. Schofield, M. and Striker, G. (eds.) (1986), The Norms of Nature. Studies in Hellenistic Ethics, Cambridge: Cambridge University Press. Schofield, M., Burnyeat, M. and Barnes, J. (eds.) (1980), Doubt and Dogmatism. Studies in Hellenistic Epistemology, Oxford: Clarendon Press. Scholz, H. (1975), ‘The Ancient Axiomatic Theory’, in Barnes et al. (eds.) (1975), 50–64 (originally published in German 1930). Schreiber, S. G. (2003), Aristotle on False Reasoning: Language and the World in the Sophistical Refutations, Albany, NY: State University of New York Press. Schrenk, L. 1991, ‘Faculties of Judgment in the Didaskalikos’, Mnemosyne 44, 347–363. Schroeder, F. M. (1981), ‘The Analogy of the Active Intellect to Light in the De Anima of Alexander of Aphrodisias’, Hermes 109, 215–225. Schroeder, F. M. (1982), ‘The Potential or Material Intellect and the Authorship of the De Intellectu: A Reply to B. C. Bazán’, Symbolae Osloenses 57, 115–125. Schroeder, F. M. (1984), ‘Light and the Active Intellect in Alexander and Plotinus’, Hermes 112, 239–248. Scott, D. (1988), ‘Innatism and the Stoa’, Proceedings of the Cambridge Philological Society 34, 123–153. Scott, D. (1995), Recollection and Experience: Plato’s Theory and Its Successors, Cambridge: Cambridge University Press. Sedley, D. (1977), ‘Diodorus Cronus and Hellenistic Philosophy’, Proceedings of the Cambridge Philological Society 203 (New Series 23), 74–118. Sedley, D. (1982), ‘On Signs’, in Barnes et al. (eds.) (1982), 239–272. Sedley, D. (1983a), ‘Epicurus’ Refutation of Determinism’, Suzêtêsis. Studi sull’ epicureismo greco e latino offerti a Marcello Gigante, 11–51. Sedley, D. (1983b), ‘The Motivation of Greek Scepticism’, in Burnyeat (ed.) (1983), 9–29. Sedley, D. (1996), ‘Alcinous’ Epistemology’, in Algra et al. (eds.) (1996), 300–312. Sellars, W. (1963), Science, Perception and Reality, London: Routledge. Sellars, W. (1967), Philosophical Perspectives. History of Philosophy (American Lecture Series), Springfield, IL: Thomas. Sharples, R. (1987), ‘Alexander of Aphrodisias: Scholasticism and Innovation’, Aufstieg und Niedergang der römischen Welt Bd 36.2 (Berlin: de Gruyter), 1176–1243. Shields, C. (1988), ‘Soul and Body in Aristotle’, Oxford Studies in Ancient Philosohy 6, 103–137. Shields, C. (1995), ‘The Proofs of the Peri ideôn: A Review of Gail Fine, On Ideas’, Oxford Studies in Ancient Philosophy 13, 269–281. Sihvola, J. (ed.) (2000), Ancient Scepticism and the Sceptical Tradition (Acta Philosophica Fennica 66), Helsinki: Philosophical Society of Finland. Sisko, J. E. (1996), ‘Material Alteration and Cognitive Activity in Aristotle’s De anima’, Phronesis 41, 138–157. Sisko, J. E. (1998), ‘Alteration and Quasi-alteration. A Critical Notice of Stephen Everson’s Aristotle on Perception’, Oxford Studies in Ancient Philosophy 16, 331–352. Slakey, T. J. (1993), ‘Aristotle on Sense-Perception’, in Durrant (ed.) (1993), 75–89. Smith, A. D. (2002), The Problem of Perception, Cambridge, MA: Harvard University Press. Smith, P. and Jones, O. R. (1986), The Philosophy of Mind. An Introduction, Cambridge: Cambridge University Press.

310

BIBLIOGRAPHY

Smith, R. (1989), Aristotle. Prior Analytics, Indianapolis, IN: Hackett. Smith, R. (1993), ‘What Use is Aristotle’s Organon?’ Proceedings of the Boston Area Colloquium in Ancient Philosophy 9, 261–285. Smith, R. (1994), ‘Dialectic and the Syllogism’, Ancient Philosophy 14, 133–151. Smith, R. (1997), Aristotle. Topics, Books I and VIII with Excerpts from Related Texts (Clarendon Aristotle Series), Oxford: Clarendon Press. Solmsen, F. (1968), ‘Dialectic without the Forms’, in Owen (ed.) (1968), 49–68. Sorabji, R. (1972), Aristotle on Memory, London: Duckworth. Sorabji, R. (1974), ‘Body and Soul in Aristotle’, Philosophy 49, 63–89. Sorabji, R. (1980), Necessity, Cause, and Blame. Perspectives on Aristotle’s Theory, London: Duckworth. Sorabji, R. (1982), ‘Myths about Non-propositional Thought’, in Schofield and Nussbaum (eds.) (1982), 295–314. Sorabji, R. (1983), Time, Creation, and the Continuum, London: Duckworth. Sorabji, R. (1990a), ‘The Ancient Commentators on Aristotle’, in Sorabji (1990b) (ed.), 1–30. Sorabji, R. (ed.) (1990b), Aristotle Transformed. The Ancient Commentators and Their Influence (The Ancient Commentators on Aristotle), London: Duckworth. Sorabji, R. (1991), ‘From Aristotle to Brentano: the Development of the Concept of Intentionality’, in Blumenthal and Robinson (eds.) (1991), 227–259. Sorabji, R. (1993), Animal Minds and Human Morals. The Origins of the Western Debate, Ithaca, NY: Cornell University Press. Sorabji, R. (2004), The Philosophy of the Commentators 200–600 AD: A Sourcebook, London: Duckworth, 3 vols. Sprague, R. K. (1962), Plato’s Use of Phallacy. A Study of the Euthydemus and Some Other Dialogues, London: Routledge. Spruit, L. (1994), Species Intelligibilis. From Perception to Knowledge I. Classical Roots and Medieval Discussions (Brill’s Studies in Intellectual History 48), Leiden: Brill. Striker, G. (1974), ‘Kritêrion tês alêtheias’, Nachrichten der Akademie der Wissehschaften in Göttingen, Phil.-hist. Kl. 2, 47–110. Striker, G. (1977), ‘Epicurus on the Truth of Sense Impressions’, Archiv für Geschichte der Philosophie 59, 125–142. Striker, G. (1980), ‘Sceptical Strategies’, in Schofield et al. (eds.) (1980), 54–83. Striker, G. (1981), ‘Über den Unterscheid zwischen den Pyrrhoneern und den Akademikern’, Phronesis 26, 153–171 (an English translation ‘On the Difference between the Pyrrhonists and the Academics’, in Striker (1996), 135–149). Striker, G. (1990), ‘The Problem of the Criterion’, in Everson (ed.) (1990b), 143–161. Striker, G. (1996), Essays on Hellenistic Epistemology and Ethics, Cambridge: Cambridge University Press. Striker, G. (1998), ‘Aristotle and the Uses of Logic’, in Gentzler (ed.) (1998), 209–226. Stump, E. (transl. with notes and essays on the text) (1978), Boethius’ De topicis differentis, Ithaca, NY: Cornell University Press. Tait, W. W. (1986), ‘Plato’s Second Best Method’, Review of Metaphysics 39, 455–482. Tarán, L. (1981), Speusippus of Athens (Philosophia Antiqua 39), Leiden: Brill. Tarrant, H. (1985), Scepticism or Platonism? The Philosophy of the Fourth Academy (Cambridge Classical Studies), Cambridge: Cambridge University Press. Taylor, C. C. W. (1980), ‘All Perceptions Are True’, in Schofield et al. (eds.), 105–124. Taylor, C. C. W. (1990), ‘Aristotle’s Epistemology’, in Everson (ed.) (1990b), 116–142. Thesleff, H. (1999), Studies in Plato’s Two-Level Model (Commentationes Humanarum Litterarum 113), Helsinki: Societas Scientiarum Fennica.

BIBLIOGRAPHY

311

Thillet, P. (1981), ‘Matérialisme et théorie de l’âme et de l’intellect chez Alexandre d’Aphrodise’, Revue philosophique de la France et de l’Etranger 171, 5–24. Tieleman, T. (1996), Galen and Chrysippus on the Soul. Argument and Refutation in the De Placitiis Books II–III (Philosophia Antiqua 68), Leiden: Brill. Tieleman, T. (2002), ‘Galen on the Seat of the Intellect: Anatomical Experiment and Philosophical Tradition’, in Tuplin and Rihll (eds.) (2002), 256–273. Todd, R. (1973), ‘The Stoic Common Notions’, Symbolae Osloenses 48, 47–75. Todd, R. (transl. with F. M. Schroeder) (1990), Two Greek Aristotelian Commentators on the Intellect (Medieval Sources in Translation 33), Toronto: Pontifical Institute of Mediaeval Studies. Todd, R. (transl. with introduction and notes) (1996), Themistius: On Aristotle On the Soul, (Ancient Commentators on Aristotle), London: Duckworth. Tricot, J. (transl.) (1947), Organon IV: Les ‘Secondes Analytiques’, Paris. Tuplin, C. J. and Rihll, T. E. (eds.) (2002), Science and Mathematics in Ancient Greek Culture, Oxford: Oxford University Press. Tweedale, M. (1984), ‘Alexander of Aphrodisias’ Views on Universals’, Phronesis 29/3, 279–303. Upton, T. V. (1981), ‘A Note on the Aristotelian Epagoge’, Phronesis 26. Vlastos, G. (1965), ‘Anamnesis in the Meno’, Dialogue 4/2, 143–167. Vlastos, G. (1969), ‘Reasons and Causes in the Phaedo’, Philosophical Review 78, 291–325. Vlastos, G. (ed.) (1970), Plato: Critical Essays I, Metaphysics and Epistemology, Garden City, NY: Doubleday. Vlastos, G. (1975) Plato’s Universe, Seattle, WA: University of Washington Press. Vlastos, G. (1983), ‘The Socratic Elenchus’, Oxford Studies in Ancient Philosophy 1, 27–58 (reprinted in Fine 1999, the page numbers are to this version). Vlastos, G. (1985), ‘Socrates’ Disavowal of Knowledge’, Philosophical Quarterly 35, 1–31. Vlastos, G. (1988), ‘Elenchus and Mathematics: A Turning-Point in Plato’s Philosophical Development’, American Journal of Philology 109, 362–396. Von Fritz, K. (1945–1946), ‘NOUS, NOEIN and their Derivatives in Presocratic Philosophy (excluding Anaxagoras) I–II’, Classical Philology 41 and 42, 223–242; 12–43. Walzer, R. (1947), Galen on Medical Experience, Oxford, reprinted in Walzer and Frede (1985), 49–106. Walzer, R. and Frede, M. (transl. and introduction) (1985), Three Treatises on the Nature of Science, Indianapolis, IN: Hackett. Ward, J. K. (1988), ‘Perception and logos in De Anima ii 12’, Ancient Philosophy VIII/2, 217–233. Wardy, R. (1990), The Chain of Change. A Study of Aristotle’s Physics VII (Cambridge Classical Studies), Cambridge: Cambridge University Press. Wedin, M. V. (1988), Mind and Imagination in Aristotle, New Haven, CT: Yale University Press. Wedin, M. V. (1991), ‘PARTisanship in Metaphysics Z’, Ancient Philosophy XI, 361–385. Wedin, M. V. (1993), ‘Tracking Aristotle’s Noûs’, in Durrant (ed.) (1993), 128–161. Wedin, M. V. (2000), Aristotle’s Theory of the Substance: The Categories and the Metaphysics (Oxford Aristotle Studies), Oxford: Oxford University Press. White, N. P. (1971), ‘Aristotle on Sameness and Oneness’, Philosophical Review 80, 177–197. White, N. P. (1976), Plato on Knowledge and Reality, Indianapolis, IN: Hackett. Whittaker, J. (1990), Alcinous. Enseignement des doctrines de Platon (Collection Budé), Paris: Les Belles lettres. Wians, W. (1990), ‘Commentary on Lloyd’, The Proceedings of the Boston Area Colloquium in Ancient Philosophy VI, 402–412.

312

BIBLIOGRAPHY

Wians, W. (1989), ‘Aristotle, Demonstration and Teaching’, Ancient Philosophy 9, 245–253. Wiesner, J. (ed.) (1987), Aristoteles: Werk und Wirkung. Paul Moraux gewidment vol. 2, Berlin: de Gruyter. Wiggins, D. (1986), ‘Teleology and the Good in Plato’s Phaedo’, Oxford Studies in Ancient Philosophy 4, 1–18. Wildberg, C. (1989), ‘Two Systems in Aristotle?’ Oxford Studies in Ancient Philosophy 7, 193–202. Williamson, T. (1994), Vagueness (The Problems of Philosophy, Their Past and Present), London: Routledge. Witt, C. (1989), Substance and Essence in Aristotle: An Interpretation of Metaphysics VII-IX, Ithaca, NY: Cornell University Press. Wright, C. (1975), ‘On the Coherence of Vague Predicates’, Synthese 30, 325–365 (reprinted in Graf and Williamson (eds.) (2002), 109–149). Wright, C. (1976), ‘Language-Mastery and the Sorites Paradox’, in Evans and McDowell (eds.), 223–247. Zeyl, D. J. (transl. with an introduction and notes) (2000), Plato’s Timaeus, Indianapolis, IN: Hackett.

INDEX OF NAMES

Achelous 251 Achilleus 60 n94, 66 Ackrill, J.L. 86 n145, 167 n16 Admetus 234 Aëtius 224 n10, 226 n14, 244, 247 Ajax 60 n94 Albinus 118 n209 Alcestis 234 Alcibiades 60, 66 Alcinous 9, 113, 118–122, 149, 151, 196–198, 215 Alcmaeon of Croton 163 Alexander of Aphrodisias 48, 51, 64 n99, 98 n168, 110 n187, 127–139, 143 n255, 152, 186, 197 n69, 199–207, 211–213, 215, 227, 247–249 Allen, J. 12, 150 n261, 254, 256 n74, 259, 266–268, 270 n104 Ammonius 133 Annas, J. 31 n38, 90 n151, 163 n5, 169 n124, 231 n24, 234 n30, 237 n35, 249 n62 Antipater 225 Aphrodite 32 Apollo 32 Aquinas; see Thomas Aquinas Aristotle 3, 4, 6, 7, 9, 10, 13, 17, 18 n5, 19 n10, 20 n13, 21, 34, 35, 37–119, 122, 124, 126, 127, 129–152, 156, 158–162, 164–168, 173–176, 178, 181–196, 199–201, 203, 205–207, 211–216, 219–221, 240, 241, 249, 251, 253, 269, 282, 286, 290, 291 Armstrong, A.H. 125, 126, 198 Asmis, E. 163 n5, 224, 231 n24, 239 n38, 240, 243 n17, 255, 256, 258, 260 n84, 261 n85, 262 n88, 263 n90 Avotins, I. 231 n24

Bäck, A. 83 n137 Bailey, C. 224 n9 Balme, D.M. 100 n172 Barnes, J. 13, 69 n109, 73 n115, 93 n158, 96 n166, 103 n174, 106 n180, 118 n195, 115, 144 n257, 163 n4, 252, 253, 266–268, 273 n107, 274, 279 n118, 184 n123 Barney, R. 180 n45, 225 n11 Bazán, B.C. 201 Berti, E. 13, 36 n46, 69 n109, 191 n62, 192 n63 Bett, R. 119 n11, 222 n3 Block, I.L. 165 n10, 167 n17 Bluck, R.S. 22 n21, 30 n37 Blumenthal, H.J. 199, 205 n89 Bobzien, S. 265 n92 Boethus 223 n6, 226 n15 Bolton, R. 13, 18 n5, 19, 39 n49, 40 n54, 43 n59, 45, 50 n74, 55 n86, 56, 94 n163, 96 n166, 103–104 n175, 108 n184, 130 n231 Bostock, D. 24 n24, 176 n38 Brennan, T. 274 Brentano, F. 166 n11 Brown, L. 169 n22, 177, 178 Brunschwig, J. 13, 51 n77, 54 n84, 220 n1 Bryson 141, 142, 145, 146, 152 Burnyeat, M. 13, 35 n45, 39 n49, 69 n110, 70 n111, 76 n123, 100 n171, 103 n174, 165, 166, 251 n67, 252, 273 n107, 276 n111 Bynum, T.W. 167 n16 Byrne, P.H. 89 n149 Callicles 49 n71, 147 n258 Carneades 12, 227, 235, 236, 251–253, 275–277 Caston, V. 203 n87 Castor 235 Charles, D. 13, 77n124, 78 n127, 86

313

314

INDEX OF NAMES

Charlton, W. 210, 213 Cherniss, H. 212 n105 Christensen, J. 247 n57 Chrysippus 222, 225–227, 233, 235 n31, 248, 252, 253, 266, 269, 270 Cicero 222, 224, 233, 236 n32, 240, 241, 245–247, 252, 276 Cleary, J.J. 13, 71 n113, 90 n151, 101 n173 Cooper, J.M. 23, 29, 59, 84 n140, 168, 180 Corcoran, J. 43 n48 Cornford, F.M. 36 n47 Couloubaritsis, L. 103 n174 Cronos 251 Dancy, R.M. 68 n105 Deichgräber, K. 277 n113 Democritus 163, 164 De Lacy, E.A. 260 n83, 282 n120 De Lacy, P.H. 260 n83, 262, 282 n120 de Pater, W.A. 51 n77, 52 n80 Descartes, R. 4 Devereux, D. 86 n145 Diogenes Laertius 171 n30, 222–224, 226, 228, 229 n19, 232, 233, 235, 236, 238–245, 249 n62, 251 n68, 252, 256–259, 261–263, 265 Diogenes of Apollonia 101 Dillon, J. 30 n37, 118 n209, 119 n120, 120, 197, 198 n72 Diodoros Chronus 252 n68 Dionysus 32 Donini, P.-L. 119 n210 Dooley, W.E. 127 n227, 129, 130 Dorion, L.-A. 18 n5, 39 n50 Ebbesen, S. 210 n101 Eck, J. van 27 n33 Electra 251 n68 Emilsson, E.K. 14, 123, 126 n225, 162 n3, 164 n6, 169 n26, 170, 179 n44, 198 n74 Empedocles 53, 101 n173, 163 Engberg-Pedersen, T. 103 n174 Epicurus 163 n5, 221–226, 228–233, 238–244, 255–264, 273, 276, 287, 292 Eubulides 251 n68 Euclid 114 n194, 131, 195, 248 n59 Everson, S. 12, 14, 163 n5, 165 n10, 166 n11, 174 n35 Fine, G. 12, 19 n11, 20 n12, 31, n38 Forster, E.S. 52, 53, 59 Frede, D. 34 n43, 174 n35, 175, 178 n43, 179, 187 n58

Frede, M. 13, 68 n108, 85 n141, 103 n174, 113 n188, 114 n192, 182 n49, 193 n64, 203 n87, 223 n7, 225 n13, 234 n29, 265, 266, 274 n108, 277, 278, 281–286 Furley, D. 164 Galen 9, 112–118, 151, 161, 169–171, 194–196, 215, 222, 244 n51, 253, 254, 269, 270, 277–279, 284, 291 Gallop, D. 27 n33, 28 n34, 30 n37 Gentzler, J. 22–26, 29 Gerson, L.P. 14, 33 n40, 124 n219, 125 n222, 197 n69, 198 n73 Glucker, J. 276 n112 Gonzalez, F.J. 19 n8, 21 n16, 22 n21, 25 n26, 30 n37 Gotthelf, A. 13, 69 n109, 86 n145, 95 n163, 96 n166, 100 n171 ‘Greek Philoponus A’ 210, 211 ‘Greek Philoponus B’ 210–213 Green-Pedersen, N.J. 51 n75, 52 n80, 247 n57 Haas, F. de 80 n130, 130 n232, 142 n253 Hades 251 Hahm, D.E. 169 n24 Halldén, S. 253 n71 Hamlyn, D.W. 103 n174, 167 n17, 191 n62 Hankinson, R.J. 46 n66, 114, 115, 116 n203, 117, 126 n224, 203 n87, 246 n55 Hart, D.W. 283 n122 Hayduck, M. 210 n102 Heath, T. 42 n57, 66 n100, 248 n59 Helen 237 Helmreich, G. 277 Henry, D. 136 n244 Heracles 234 Heraclitus 47, 55 Herodotus 238, 259, 261 Hintikka, J. 63 n98, 67 n103, 76 n123, 81 n131, 84 n139, 84, 103 n174 Homer 261 n66 Iamblichus 3 Irwin, T.H. 57 n90, 69 n109 Johnson, M.R. 80 n130, 136 n244 Jones, O.R. 171 n30, 231 n23 Kahn, C.H. 86 n146, 96 n159, 103 n174, 104 n176, 165 n10 Kakkuri-Knuuttila, M.-L. 13, 39 n49, 43 n59, 50, 51 n75, 52 n80, 55–57, 66 n101, 67 n103 Kalbfleisch, C. 114, 194 n65, 195

INDEX OF NAMES Kanayama, Y. 19 n11, 22 n21, 23 n22, 27 n33, 28 n34, 29 n36 Kapp, E. 268 n101 Kieffer, J.S. 194 n65, 195, 196 n66 Kirwan, C.A. 68 n105 Klosko, G. 21 n17 Kneale, M. 252 n68 Kneale, W. 252 n68 Knuuttila, S. 66 n101, 67 n103, 81 n131, 84 n139, 97 n167, 100 n171, 164 n8, 173 n33, 242 n43, 251 n64 Kosman, L.A. 103 n174 Kühn, K.G. 271 n115 Lane, M. 13, 33 n42 ‘Latin Philoponus’ 208, 210, 213–216 Lear, J. 90 n151, 166 n14, 181 n48 Lee, M.M. 13 Lennox, J. 13, 36 n47, 67 n104, 69 n109, 73 n116, 86 n145, 96 n166, 100 Leroux, G. 124 n220 Lesher, J.H. 103 n174 Leszl, W. 93, 96 n166 Leucippus 163 Leunissen, M. 80 n130 Lindberg, D.C. 163 n4 Lloyd, A.C. 124 n220, 189, 192 n63 Lloyd, G.E.R. 14, 35 n45, 75, 96, 113 n191, 113–115, 132 Løkke, H. 79 n129 Long, A.A. 13, 14, 229 n20, 230, 234, 235, 238, 239, 243–245, 246 n55, 247 n56, 248 n60, 263, 265, 271 n106 Lucretius 230, 242 Lysander 60 n94, 66 Marcus Aurelius 222 McKirahan, R.D.Jr. 13, 90 n153, 103 n174, 106 n180, 183 Menelaus 237 Menn, S. 150 n262 Meno 177 Michael of Ephesus 127, 136 n244 Mignucci, M. 81 n132, 82 n136, 85 n143 Modrak, D.K.W. 103 n174, 167 n17, 191 n61 Moraux, P. 127 n228, 136 n242, 200 n77, 201, 204 Moravcsik, J. 46 n65, 72 n114 Morrison, D. 150 n261 Morrow, G. 180 n47 Neilos 251 Nicolaus of Reggio 277 n113

315

Niiniluoto, I. 276 n112 Normore, C.G. 186 n56 Nussbaum, M.C. 12, 57 n91, 58, 59, 101 n173 Nuyens, F. 165 n10 Obbink, D. 248 n61 O’Meara, D.J. 198 n74 Orestes 230, 232 Owen, G.E.L. 13, 58, 59 Parmenides 101 n173 Patzig, G. 85 n141, 193 n64 Pellegrin, P. 86 n145, 100 n172 Pheidias 60 Philodemus 260, 262–265 Philoponus, John 3, 50 n73, 80 n130, 82 n136, 98 n168, 99, 133, 134, 141–150, 152, 166 n11, 199, 208, 210–216 Peirce, C.S. 253 n71 Plato 2–4, 6, 8–10, 12, 13, 17–41, 49 n71, 60, 65–68, 72, 80, 91, 112, 116 n205, 117, 119–122, 124, 125, 135, 136 n243, 147 n258, 156–158, 163, 168–173, 175–180, 182, 193, 198 n72, 210, 211, 220–222, 282, 289–291 Plotinus 4, 9, 14, 113, 122–126, 151, 152, 162, 171, 179, 180, 189 n59, 198, 199, 215 Plutarch 171 n30, 196 n66, 211, 212, 222, 224 n10, 226 n14, 227, 230, 231, 247–250 Polus 49 n71 Polybius 101 n173 Polydeuces 253 Porphyry 3, 54 n84 Poseidon 253 Posidonius 226 Primavesi, O. 52 n80 Proclus 3, 142 n251, 146, 152, 199 n76 Pyrrho of Elis 222 n3 Pythocles 240, 258–260 Remes, P. 33 n42, 125 n222, 126 n225 Remes, U. 81 n131, 84 n139 Rhea 251 Rich, A.N.M. 198 n72 Rihll, T.E. 113 n190 Rijen, J. van 81 n132, 84 n139 Rist, J.M. 224 n9 Robinson, R. 20, 22 n21, 23 n23, 28 n34, 30 n37 Ross, W.D. 31 n38, 32 n39, 42 n57, 66 n100, 81 n133, 82 n135, 84 n138, 99, 100, 103 n174, 106 n180, 107 n183, 108 n185, 183 n52, 185 n55 Rowe, C. 22, 23 n23, 27, 28 n34, 30 n37

316

INDEX OF NAMES

Sandbach, F.H. 245 n53 Schofield, M. 12, 174 n35, 245 n52, 248 n61, 251 n65, 270 n105 Scholz, H. 90 n158 Schreiber, S.G. 18 n5, 39 n50, 40 n52, 184 n54 Schroeder, F.M. 20, 202 Scott, D. 13, 177, 239 n38, 242 n45, 247 Schrenk, L. 197 n69 Sedley, D. 13, 14, 19 n11, 118 n209, 119 n211, 196 n68, 197 n71, 230, 234, 235, 237 n34, 238, 239, 243, 244, 245, 246 n55, 247 n56, 248 n60, 259 n67, 263, 271 n106 Sextus Empiricus 12, 170 n28, 171 n30, 221 n2, 222–225, 227, 230, 232–236, 239 n40, 251, 253–258, 260, 261 n86, 265–268, 271–275, 277, 279 n117 Sharples, R. 127 n227, 201 n80 Sihvola, J. 242 n43, 251 n64 Sisko, J.E. 168 n11, 181 n48 Slakey, T.J. 165 n10 Smith, A.D. 171 n30 Smith, P. 171 n30, 231 n23 Smith, R. 13, 39 n49, 40 n53, 41, 43 n58, 45 n62, 48 n70, 50 n72, 51 n77, 52, 54, 68 n106, 76 n123 Socrates 17–21, 27, 31–33, 36, 49 n71, 59, 60, 66, 91, 177, 214 Somfai, A. 91 n155 Sorabji, R. 13, 14, 54 n83, 103 n174, 104 n176, 127, 133 n235, 142 n252, 149 n259, 165–167, 183 n53, 189, 192 n13, 197 n69, 202, 203, 210 n99 Sprague, R.K. 21 Spruit, L. 14, 172 n31 Stephanus 210 n102 Striker, G. 13, 68 n107, 220 n1, 223–225, 226 n14, 227 n16, 229 n21, 236 n33, 237 n35, 239 n40, 256 n78, 257, 274 n108, 275 n110, 276 n111 Stump, E. 51 n77, 52, 54

Tait, W.W. 22 n21, 27 n33, 30 n37 Tarrant, H. 19 n7 Taylor, C.C.W. 224 n10, 231 n23, 237 n27 Themistius 138–143, 145, 152, 205–211, 212 n107, 216 Theophrastus 48 n69, 51, 82 n136, 117 n208 Thesleff, H. 34 n43 Thrasymachus 49 n71 Tieleman, T. 113 n89, 114 n92, 269 n103, 270 n104 Todd, R. 205, 207 n93, 208 n96, 212 n105 Torquatus 240 Tricot, J. 106 n180 Tuplin, C.J. 113 n190 Tweedale, M. 131 n233 Usener, H.

224 n10, 230

Velleius 241 Verbeke, G. 199 n76 Vlastos, G. 18, 19, 20 n12, 22 n21, 27 n31, 36, 177 n39, 203 n87 Walzer, R. 254, 277–279, 281–286 Ward, J.K. 166 n12 Wedin, M.V. 85 n141, 192 n63, 193 n64 Whittaker, J. 118 n209, 198 n72 Wians, W. 96 n166 William of Moerbeke 210 Williamson, T. 252, 253 n71 Witt, C. 85 n141 Wright, C. 253 n71 Zabarella, I. 90 n153 Zeno of Citium 222, 225, 236 Zeno of Elea 137 Zeus 251, 252, 269 Zeyl, D.J. 168

INDEX LOCORUM

Alcinous Didaskalikos 4.3. 154, 25–32 197 n71 4.5. 155, 3–7 197 n71 4.6. 155, 21–27 197 4.6. 155, 30–34 197 5.1 119 5.6 122 5–6 119 6.3 119 n213 6.6 119 n213 9–10 198 10.7 119 n213 12–16 119

Alexander of Aphrodisias De Anima 81, 13–15 200 81, 25 204 81, 22–28 200 81, 22–82, 15 204 82, 5–19 204 83, 11–19 197 n69 84–85 202 84, 3–13 212 n107 84, 15–17 186, 200 84, 20–21 201 84, 24–25 200, 202 84, 25–85, 5 200 85, 3 202 85, 14–22 207 n95 85, 20–25 203, 204 86, 6–14 212 n107 87, 24–29 201, 201 n81 88, 23–24 203 89, 9–19 200

106, 18–113, 24 186 106, 28–29 186 106, 31–107, 1 186 107, 1–4 186 107, 7–9 186 De Mixtione 217, 2–32 248 In Metaphysics 11, 8–13 129 11,12 129 12, 6–14 130 103, 5–104, 18 130 166, 8–13 197 n69 271, 12–21 132 In Prior Analytics 3, 20 130 301, 17–32 133 302, 6 134 302, 7–13 134 302,18–25 134 304, 19–30 135 330, 32–331, 1 135 331, 17–24 135, 136 In Topics 1, 8–19 136 n243 5, 21–26 51 n78 16, 1–8 128, 131 16, 29–30 128 17, 1–6 129, 131 27, 7–29, 16 136, 138 30, 12–17 137 n245 30, 19–26 137 30, 26–31, 21 137

317

318 126, 13–15 51 n78 126, 16–17 48 126, 17–30 48 n69 435, 17 128, 139 436, 9–11 128

Aëtius Placita 4.8, 12 226 n14 4.11, 1–4 244 4.12 282 n120, 235 n31

Aristotle Analytica Posteriora I 1 39, 44 n60 71a2–3 105 n178 71a9–11 39 I 2–6 70–77, 94 71b9–16 46, 70 n111 71b17–18 75 n120 71b20 70 n111 71b21 75 71b26–27 75 71b31 77 n126 71b32–72a8 70, 127, 139, 145 72a18–20 42 n57 I 3 84, 95, 103, 158 72b15–73a20 141 73a7 84 n138 73a21–23 70 n111, 75 73a24 76 73a26–27 76 73a34-b18 81 73b16–19 83 73b28–29 82, 76 73b31 74 74a14 76 74a37–b4 67 74b5 70 n111 75b1–20 89, 92, 141 n249 76a23–25 92 76a37–b1 90, 131, 195 76b12–14 89 76b35–39 42 n57 77a4 42 n57 77a10–21 91 77a31–33 90 78a22 70 n111 78a31–b11 76, 77 n125, 150 83a24–35 89 88b3–6 94 88b35–37 103, 70 n111

INDEX LOCORUM 91a28–32 74 91b13–15 65 92b7 87 93a37–b3 87 93a30–31 74 94b9–26 80 n130 96a9–19 85 96a24–26 65 97b16–27 66 98a9–11 74 II 19 64, 85, 102–110, 112, 159, 160, 181, 182, 184, 188–192, 206, 211, 240, 244 n50, 282 n120 99b17 105 99b20–32 105, 211 99b35 181 99b36–100a14 182, 70 n111 100a1–9 85, 104, 106, 184, 206 n92, 109, 196 n66 100a11 105 100a12–15 103, 105, 107, 125 n221, 188 100a14–b5 64, 104, 107, 131, 191, 192 100b5–12 109 Analytica Priora 24b18–22 42, 45 I 27–30 96–102, 146–152 43a25–33 99 43a42–43 109 43b1 147 43b7 100 43b21 99 43b39 133 46a3–10 58, 99 46a17–22 99, 100 46a28–30 100 64b35 95 68b15–29 61, 63, 63 n97 68b30–37 62, 70, 127 Categoriae 1 90 12, 14a30–35

128

De Anima 408b16–17 164 416b34–35 164 II 5 184 417a20 166 n13 417a30–b1 164 417b3–7 185 417b16–26 164 418a3–6 166 n13

INDEX LOCORUM 418a12–16 173 418a16–17 174 n34 418b9–13 165 419a17–20 165 422a7 166 n13 422b8–10 174, 186 422b29–30 166 n12, 174 423b30–424a3 166 n13 424a4–6 166 n12, 174 424a18–19 165 424a28–32 166 424b2 165 425a14–b3 167 n18 425b22–23 165 425b26–426a11 164 427b12–13 173 428a11–15 174 n35 428b18–26 174, 174 n35 429a13–18 165, 185 III 4 70 n111, 185, 189, 205 429a21–22 185 429a24 185 429a29–b8 166, 185 429b31–430a2 200 n78 430a1–2 185, 213, 244 n50 430a14–15 187 430a20 187 III 5 193, 207 430a23–25 193, 199 430a26 191 430a27–29 190 430b2–23 191 III 6 159, 190 430b26–30 159, 173, 190 431a1–2 187 431a14–17 192 431a17–19 167 n19 431b2 192 431b6–8 185, 192, 207 n95 431b29–432a1 187 432a12 190, 191 433a10 249 n63 434a29 165 435a22–24 166 n13 De Generatione Animalium 721a30ff. 94 n163 724a14ff. 94 n163 De Interpretatione 16a6–9 184, 209

De Juventute 469a11–15 167 De Memoria 450a27–29 165 n10 De Partibus Animalium 648a29–35 101 n173 648b4–10 94 n163 673b4–12 94 n162 674a9–b17 95 n163 674a23–b17 67 n103 675a36–b1 94 n162 De Sensu 436a7–8 166 436b7 165 n10 439a30 165 446b28–447a12 165 448a2–5 165 n10 449a2–10 167 n18 De Somno 454a7–11 165 n10 460b1–3 175 Ethica Nicomachea 1095a30–b5 70, 127 1145b1–7 57, 58, 101 1153a14 134 Historia Animalium 489b2 100 511b31–512b12 101 n173 512b13–513a7 101 n173 566b3–26 100 664a16–17 98, 100 Metaphysica 981a5–17 183 982a4 129 982a21 129 982a25–b10 127, 129, 130 991a23 130 997a16–18 89 1010b3–10 174 1021b12–1022a3 134 n238 1051b17 191 1051b25 159 1051b32 159 1052a1–2 159, 190 1075a13–15 150 1078b28–30 20 n13, 59

319

320 Physica 184a16 142, 143, 149 184a24 130 202b6–8 164 244b11 166 Rhetorica 1356a31 39 Sophistici Elenchi 165a1–4 41, 42, 45 n61 165a35–b12 39 168a34–37 41 168b31–32 90 n152 171a1–11 41 171b22–34 40 Topica I 1 38–59 100a18–21 96 100a25–101a16 39 100a25–27 43, 45 100a27–30 39, 128, 129, 131 100b22–24 39, 53 n81, 101 100b23–101a4 40 101a26 137, 138 101a36–b5 38, 135 n240, 137 101b28–33 47 102a18–20 81 n132, 84 n138 104a5–7 47, 148 104b1–17 47, 148 104b18–24 47 105a11–19 44, 49 105b16–18 53 105b19–29 47, 148 109a34–36 52 111b5–10 60 n35 111b16–18 42 n57 112a17–24 56 II 7 48 113b27–34 48 n68 115a7–9 52 120b15–20 49 124a16–20 52, 53 124b15–18 52 138b30–37 173 141a26 128, 139 141b3ff. 70, 127, 128 VIII 1 44, 47, 48, 52 155b4–10 52 155b20–21 44, 47 n67 155b36–38 44 157a23–26 60

INDEX LOCORUM 157a34–38 49 157b31–32 49 159a19–21 41 159a23–25 55 VIII 5 39, 55, 56, 87 159a33–b35 55, 56 160b17–23 49 n71 160b24–33 41 161a19–21 55 161a33–b5 40, 55 161b21 55 162b34–163a13 43 163b28–32 54 n83

Cicero Academica 1.41–42 225, 233 2.21 244 2.31 227 2.38 237 2.46 224 2.51 235 n31 2.88–90 237 2.93–94 252 2.99–101 227 2.142 224 De Divinatione 2.34 170 n28 De Finibus 1.22–23 224, 240 3.33 245, 246 De Natura Deorum 1.43 242 1.62–63 242 2.12–15 244, 247 2.19 170 n28 2.37 246 n55

Diogenes Laertius 2.92 171 n30 2.108 251 n68 2.111 252 n68 7.46 233, 236 7.49–51 235 n31, 223, 249 n62 7.53 244, 245 7.54 223 n6, 226, 233, 244 7.55–56 244 n51 7.76–81 265

INDEX LOCORUM

321

7.187 252 n68 7.192 252 7.197 252 10.31 224, 243, 255 10.32 228 10.33 238, 239 n39, 241, 244 n49 10.34 224, 232 n25, 256 10.37–38 223, 238, 239 10.39–40 261 10.46–48 232, 259 10.51 229 n19 10.56–59 262, 263 10.72 239 n40 10.76–77 240 10.80 258, 259 10.86–87 240, 258, 259 10.93–95 258 10.112 258 10.123–124 240

1.4.3 116 n202 1.4.6–12 114 1.5.3 116 n202 2.7 117, 196 9.178 117 n202 9.199 117 n202

Empedocles

On The Sects for Beginners 1, 14–2, 1 277 3, 4–9 283 3, 18–19 284 4, 8–10 284 5, 5–18 281 7, 6–8 282

DK A 86

163

Epicurus Key Doctrines (Kuriai doxai) 11 222 37 239 n40

Galen De Lociis affectis Book I, p. 25ff 278 n115 De Placitis Hippocrates et Platonis 2.2, 10–11 269 2.3.108, 26ff. 114 3.1.168, 15–16 114 3.5.8 269 5.2.49 244 n51 5.3.1 244 n51 5.566–567 115 7.5, 5–10 169 7.5, 32–33 169 7.7, 9–10 169 n25 Institutio logica 1 114 1.1 194 1.5 194 3.2 195 Methodo Medendi 1.3.8–11 116

On Medical Experience III p. 88 285 VI pp. 91–92 285 VII p. 94 279, 285 VII p. 95 279, 284 VII pp. 96–97 278 VIII pp. 97–98 285 XV 278 XVII 278 XVII pp. 115–116 254, 278, 283 n122 XVII p. 118 254, 278

Outlines of Empiricism 45, 21–26 281 46, 2–13 283 48, 2–4; 14ff. 281 51, 2–9 281 54, 13–29 286 58, 10–25 281, 283

Homer Iliad XV 187–189

251 n66

Lucretius De Rerum Natura 4.353–363 230 5.1161–1225 242

Philodemus De Signiis 11.32–12.31 262 14.17 263 n90 15.37 263 n90 21.29 263 n90

322 Philoponus (?) In Analytica Posteriora 29, 1–14 145 71, 4–13 82 n136 111, 20–31 145 112, 8–14 146 112, 25–36 146 113, 1–4 146 113, 4–114, 17 146 122, 26–123, 14 145 433, 15–21 211 433, 31–34 211 434, 3 211 435, 2–10 210 435, 23–27 211 In Analytica Priora 273, 10 148 274 146, 147 276, 10–16 148 276, 20–29 50 n73, 99, 147 276, 31–32 148 305, 12–21 148 In De Anima 6, 1–2 199 n76 520, 1–20 211, 212 523, 23–26 211 n104 526, 15–20 212 526, 29–34 212 Latin In De Anima 4, 60–65 215 5, 82–90 214 n109 5, 99 214 n109 13, 00–06 199 n76 33, 80–85 214 40, 35–43 213 45, 53–59 214 47, 96 214 n110 48, 27–49 214 48, 32–33 208, 214 51, 7–10 214 51, 11–52, 29 214 51, 95–99 214 64 215 65, 64–65 213 n108 72, 38–72 215 73, 44–53 215 75, 20–21 215 77, 83–87 213 n108

INDEX LOCORUM 98, 35–44 214 116, 75–80 214 In Physics 11, 7–8 143 11, 24–13, 4 144 12, 5–10 144 13, 3 143 17, 15–20 143 17, 23–24 145 17, 25–18, 4 143 18, 5–19, 2 144, 145

Plato Apology 27b 20 Crito 49c 19 n6 Euthydemus 293b-c 21 Gorgias 494c-d 147 n258 Meno 76c 163 86e4–87b2 98a 156

22

Phaedo 72e–75e 157, 173 76b–c 156 n1 96b 182 98b–99a 31 99c5–d1 22 99e5–100a7 22, 23, 26 100b1–9 27 100c3–7 27 101d3–8 29 101d5–e1 22, 29 101e1–6 30 Phaedrus 237b–238c 32 238e–241d 32 244a 32 245c–e 22 248b6 124 265a–266b 32, 33 277b7–9 34

INDEX LOCORUM Philebus 32a–36a 282 n120 53c–54a 135 n239 Politicus 258b–267c

4.5.3, 36–38 170 4.9.3, 1–6 170 n28 5.3.8, 20 170 5.5.7, 28–30 170 5.9.7, 5–6 126 n223

20, 59

Plutarch Protagoras 319b–c 20, 59 332c 20 Republic Book I 437a 21 Book VI 507c 172 510b–511d 22, 30 Book VII 516b–c 180 516c 31 523a–524c 173 529b–530b 172 Seventh Letter 342a–343b 180

Adverus Colotem 1109a–b 224 n10 1109d 230 1109f–1110b 231 n22 1110c 230 1120c–d 171 n20 1120e 231 1121c 231 De communibus notitiis adversus Stoicos 1059f–1060e 227, 248, 249, 250 1075e 244 1084f 226 n14 De Sollertia Animalium 961c 196 n66

Sextus Empiricus Sophist 219a–237a 32 253d 158, 178 254b–256a 178 254d 33 n41 Theaetetus 159c 169 Timaeus 29a 198 n72 39e 198 n72 45b–c 168

Plotinus Enneads 1.2.3, 27–30 199 1.3.1, 1–2 124 1.3.1, 22–35 198 1.3.4, 9–20 125 1.3.5, 1–5 126 2.4.5, 10–11 170 4.4.23 170 4.4.26, 14–16 170 n28 4.4.32, 1–17 170 n28 4.4.40 170 n28

Adversus Mathematicos 1.57 239 n40 7.24–26 254 7.151–152 221 n2, 225 7.176 227 7.184 227 7.203–204 224 n10, 229, 273 7.211 232 n25, 257 7.215–216 232 n25 7.244 233 7.247–252 233, 234 7.253–260 234, 237 7.354 171 n30 7.364–367 171 n30 7.396 254 7.401–424 234–237, 253, 279 n117 8.9 224 n10 8.63 230 8.85 235 n31 8.140 254 8.142–155 255, 256, 266, 271 8.153 267 8.185 224 n10 8.213–215 258, 260 8.223 265 8.276 254, 265

323

324

INDEX LOCORUM

8.306 267 8.316–319 255, 256 n74, 257 n79 8.397 225 9.79 170 n28 9.182–184 251 11.8–11 227 11.182 225

18, 31–19, 2 141 n249 19, 7–10 142 19, 12–17 141 24, 26–28 141 n249 63, 2–26 206 99, 30–31 208

Outlines of Pyrrhonism 1.118–119 275 1.215 171 n30 2.96–99 254–256, 267 2.110 268 n100 2.116 271 2.156 265 3.242 225

In De Anima 56, 1–12 212 n107 94, 5–16 212 n107 94, 18–20 205 94, 26–27 205 95, 17–18 207 n95 97, 19–20 205 98, 29–99,4 207 99, 5–6 208 99, 9–10 209 99, 11–12 207 n54 102, 30 207 103, 36–104, 11 209 104, 7–11 208 105, 26–30 205 n90

Simplicius (?) In Phys. 15, 15–29 149 16, 7–24 150, 151 16, 31–17, 25 151 17, 17 143 n256 20, 9–11 151 20, 17 151 In De Anima 228, 17–20 168 n21

Themistius In An. Post. 6, 14–16 139 9, 21–23 141

In Phys. 1, 14–20 139 1, 20–2, 3 140 2, 5–25 140

Thomas Aquinas Summa theologiae I 1, q. 78 a. 3 166 n15

INDEX OF TOPICS

Abstraction; Aristotle 64, 181–189; Alexander 200–205 Accidental Necessities 84 Accordance (sumphônia) and Discordance (diaphônia); Plato 23–29, Epicurus 258 Active Intellect; Aristotle 193, Alexander 200–203, Themistius 207–209, Latin Philoponus 214 Alteration in the Theory of Perception; Aristotle 164–167, Philoponus (?) on Perfective Change 212 Analysis, Methods of; Plato 32–35, Aristotle 65–68, Alcinous 120–122, Epicurus 260–264; cf. also Stoic Common Notions 247–249 Analytic Truth 33, 65 Apodeixis, Apodeictic Syllogism; see Proofs, Aristotle’s theory of Aporia 18–19 Appearances; Aristotle, Saving the Appearances 57–59; Epicurus 224 n10; Stoic Notion of; 225; see Cognitive Impressions Appropriation (oikeiôsis) 246–247 Arguments for Existence; Epicureans 241–243, 260–262; Stoics 247–249 Atomism 163–164, 230–233, 260, 262–263 Axioms; Aristotle 89–93, Galen 117, 195, Alexander 131–132, Themistius 141–142, 209, Philoponus 145–146; see also Validity Axiomatic Science 93–95, 111 Begging the Question, Aristotle’s Conception of 43–45, 95 Being Better Known; see Order of Things vs. Order of Knowledge

Cataleptic Impression/Appearance; see Cognitive Impression Causes/Reasons; Plato 27–31, Aristotle 73–75, 76–78, Galen 114–115; cf. Explanation, the Methaphysical Interpretation of 46, 71–72 Clarity of Impressions 235–237 Circularity; see Begging the Question Co-affection (sumpatheia) 170–171 Cognition (katalêpsis) 225 Cognitive Impression (phantasia katalêptikê); Stoics 225–228, 233–237; Academic Criticism of 227–228 Coherence 24–26, 226; see also Consistency Common Axioms; see Axioms Common Intellect 204–205 Commonly Accepted (endoxon); 19 n10, 39–42, 53, 57–58, 95, 99–101, 128, 136, 138 Common Notions; Euclid 195, Stoics 227, 247–250, Themistius 209 Concepts; 34, 65, 155–156, 161, 177, 189, 203–204, 241–247; Vagueness of Concepts; 251–253, 277–281; see also Conceptual Thought Conceptual Analysis; see Analysis Conceptual Thought 10, 123–124, 198–199, 215; see also Propositionality Confirmation on the Basis of Perceptions, Epicurus’ theory of 255–260 Consistency 18–20, 23–26, 55, 67–68, 226, 250 Convertible Terms in Syllogisms 44, 61–63 Cosmology 31, 35–36, 119–120, 150, 158, 178 Criterion of Action 247 Criterion of Truth; Epicurus 223–225 (perceptions), 238–243 (preconceptions), Stoics 225–228 (cognitive impressions), 244–247 (preconceptions), 247–249

325

326

INDEX OF TOPICS (common notions), Academic Sceptics 227–228, Pyrrhonian Sceptics 274–275

Definitions 32–35, 65–67, 74, 107–109, 114–118, 238–241 Departmentalisation of the Sciences; see Axioms Dialectic; Plato 38; Aristotle 38–58; Aristotle’s Definition of 42–47; see also Validity; Dialectical topoi 51–54; Plotinus 124–125; Alexander 136–138 Discernibility of Impressions 11, 235–237 Discursive Reasoning; see Conceptual Thought; cf. Propositionality Elenchus; see Refutation Elimination and Similarity; see Analysis, Epicurus Empiricism; Aristotle 58–59, 103–110, 181–184, 188–189; Galen 114–118, 195–196; Epicurus 241; Medical Empiricism 253–254, 266, 276–287 Endoxon; see Commonly Accepted Epistêmê, the Term 68–69 Epistemic Conception of Argumentation 42, 47, 56, 69, 102, 111–112 Equivocity of Common Axioms; see Axioms; Princple of Purity Experience; Aristotle’s conception of; see Empiricism, Aristotle, Medical Empiricism Explanation; see Causes/Reasons; Proofs, Stoic Conception of Explanatory vs. Explicatory Syllogism 73–74 Evaluating the Credibility of Beliefs 40–41, 55–57 Fallacies 21–22, 39–40, 184 n54 Formation of Concepts/Preconceptions; see Concepts Forms (Intelligible Forms) 155–156, 219–221, 253, 264; Plato 24–28, 31, 33, 157–158, 176–179; Aristotle 84–85, 187, 193, 202; Galen 117, 196; Alcinous 119–122, 197–198; Plotinus 123–125, 161–162, 198–199; Themistius 205, 207–209; Philoponus (?) 212–215 Generalisation 5, 9, 12, 20, 32, 49, 54, 60, 65, 72, 79, 85, 104–111, 159, 181–184, 197 n69, 206, 263–264, 281–282; Generalisations for the Most Part 85, 281–283 Great Kinds (Very Great Kinds) 10, 33, 158, 178

Hypothesis; see Method of Hypothesis Imitative Experience 283–284 Immediacy; of the Premises of Proofs 76, 85–88, 127; of Intellectual Vision in Plotinus 123, 198 Impressions; see Appearances Indemonstrable Argument Forms 265–266 Indirect Arguments 41–42, 87, 131–132, 260–262 Infinite Regress; see Regress Argument 1, 224; Aristotle 103, 158–159; Epicurus 239–241 Innate Principles 105, 106, 161, 177, 181, 195, 211, 215, 242, 247, 265 Intellectual vision in Neo-Platonism 123–124, 198–199 Intelligible Forms; see Forms Induction 20, 44, 59–65, 67–68, 119, 128–129, 137, 260; Perfect Induction 61–62 Knowledge of a Fact vs. Knowledge of the Reason 77–78 Knowledge Terms; see Epistêmê Limit and the Unlimited 34 Literalist Reading of Aristotle’s Theory of Perception; see Alteration Location of the Commanding Faculty, Stoic Arguments about 269–271 Logical Principles; see Axioms Material Intellect, Alexander’s Theory of 200–204 Mathematical Sciences in Aristotle 79 n128, 89–93 Medical Science; in Galen 113–116; see also Methodology, Medical Science Medium, in the Theory of Perception 164–168 Methodology; Plato 23–35, Aristotle 38–68, 96–105, Alcinous 119–122, Alexander 133–138, Philoponus 146–148, Epicurus 255–263, Medical Science 276–287 Method of Hypothesis 8, 22–31, 42 n57, 121 Middle Platonism 19, 113, 118–122 Natural Conceptions 196–198, 244–247 Necessity of the Premises of Aristotelian Proofs 76, 80–85, 95 n165; Necessary Premises in Dialectic 44–48 Neo-Platonism 119, 122–126, 180, 196, 198–199, 215 Nominal Definition 87–88 Non-Deductive Inference Schemes 51–54 Non-Hypothetical Principle 30–31, 121–122

INDEX OF TOPICS Objective Starting Points; see Order of Things vs. Order of Knowledge Objects of Knowledge 7, 156, 221 Ontological Hierarchy, Plotinus’ Hypostases 122–124 Order of Things vs. Order of Knowledge 29–30, 39 n51, 62–63, 69–72, 76–77, 87, 93, 102–103, 109, 111, 127–129, 139–145, 149–151, 221, 255 Passive Intellect 199–205, 207–209, 213–214 Perception; Projective Theories of 168–169; Receptive Theories of 163–167 Perceptual Error 174, 228–233 Phantasia; see Appearance; Cognitive Impression Pneuma, Stoic Theory of 246–248 Potential Intellect; see Passive Intellect Preconceptions; see Criterion of Truth Premises of Scientific Proofs; Aristotle 72–110, Galen 114–118 Principle of Purity 90–91 Priority; see Order of Things vs. Order of Knowledge Productive Intellect; see Active Intellect Proofs; Aristotle’s theory of 72–110, Stoic Conception of 266–269, Sceptical Criticism of 271–273 Propositionality; 2, 9, 104–105, 110, 160–161, 189–193, 199, 206–207, 214, 223, 243, 247; Propositional logic 265 Psychological Approach to the Starting Points of Knowledge 4–5, 17, 103–104 n175, 126, 155–161, 175, 181, 183, 194, 285; Relation to the Premises of Argumentation 155–161 Quadration of a Circle 141–142, 145–146; see also Principle of Purity Quick Intellect (having anchinoia) 110 Realism; Metaphysical 6, 35, 72, 117 n207, 194–196, 215; Direct Realism 171–173, Representational Realism 171 n30, 231 n23 Receptivity of the Intellect 188, 200–204 Recollection; Plato 156–158, 161, 173, 176–178, 220, Aristotle’s Criticism of 105, Commentators on Recollection 208, 213–214 Refutation 8, 18–22, 38–41, 49–50, 55 n87 Reliability of Perceptions 10, 163, 169 n23, 171–174, 188, 220, 233, 238 n36

327

Saving the Appearances; see Appearances, Aristotle Self-Evidence 3, 94–95 n163, 111, 112–118, 120, 122, 149, 221, 229, 236, 238–239, 258, 261, 292 Scepticism 4–6, 19, 118, 171 n30, 175, 220, 222, 227; Academic Scepticism 11–12, 220, 223, 227–228, 234 n30, 235–236, 251–253; Pyrrhonian Scepticism 12, 222 n3, 223, 271–275, 287 Science; Aristotle’s Conception of 38–110, Galen 113–118, Medical Debate about 276–287 Signs; Syllogisms from 149–150; Hellenistic Theories of 254, 265, 267 Sorites Argument 251–253, 276–287 Suspension of Judgment 19, 253, 273–275, 287 Syllogism; Dialectical 38–59; as Opposed to Induction 59, 61–65, Proofs as Syllogisms 73–88 Synthesis 120–122, 190 Teacher’s Intellect 208, 213–214 Teleology 37, 80 n130, 84, 115, 157; Implicit Teleology 94 Theory of Forms; see Forms Truth, Temporally Unqualified Analysis of 80–81; as Opposed to Reputability 136–138; see also Criterion of Truth Truth of All Perceptions, Epicurus 228–233 Universality, of the Premises of Proofs 80–86; see also Generalisations for the Most Part Universals 64–65, 120–110, 130–131, 139, 142–145, 159, 181–184, 187–193, 196 n66, 203, 206–208, 211, 220, 226, 238, 244, 264; see also Forms; Recollection Unprovable Premises, see Premises of scientific proofs Vagueness; see Sorites Argument Validity; of Dialectical Arguments 51–54; Principle of Non-Contradiction 8, 19, 21–22, 67–68, 90–92, 112–114, 132; Entailment 23–24; see also Consistency Visual Transmission 162, 165, 168–171; see also Perception Void, Epicurean Arguments 255–256, 260–264 Witnessing and Counter-Witnessing; see Confirmation on the Basis of Perceptions