This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
E(q,p) = const.}. See figure 1.1a. Figure 1.1b represents the possible states of an ecosystem constituted by two populations of organisms, one of which preys on the other. Caution: Since the state space is abstract, so is every trajectory in it; it represents a process but not a trajectory in physical space. The generalization of the above to things with n properties, where n > 2, is immediate: SL = {
R
)
R
)
)
R
Reality and Hylorealism
19
state functions Y, which are not built property-by-property but in one piece. Each of these abstract (Hilbert) spaces is infinite-dimensional. However, the basic idea is the same: A state space is the set of all the really possible (that is, lawful) states of a material entity (particle, field, ecosystem, business, national economy, or what have you). The description of really possible facts (states, events, and processes) in terms of state spaces renders both modal logic and possible-worlds ontology redundant. The reason is that neither of these involves the central notion of a law. (See details in Bunge 1977a, 2003a.) The following table summarizes the preceding analysis. Ontological item
Conceptual counterpart
Thing Property State Event Process Law, rule, or norm
Model thing Attribute: n-ary predicate, state space Point in state space Ordered couple of points in state space Trajectory in lawful state space Restriction on state space
Note the logical order: We start with things because they are the bearers of properties; we then go to states and their changes, and end up in patterns of being and becoming (laws). This is the order of analysis, but what we find in reality are things with all their properties, and changing in lawful ways. And we model them in terms of attributes (predicates), such as functions, or state variables, that define a state space. Note also that we have avoided the common confusion between fact and datum – a confusion encouraged by ordinary language. An empirical datum is not a fact but a proposition reporting on some fact – for instance, that your age is such and such. Another difference between facts and data is this. Whereas facts ‘obey’ laws of nature or social norms, data, being propositions, abide by logic. There are negative, disjunctive, and general propositions, but not negative, disjunctive, or general facts. For example, that the hound of the Baskervilles did not bark that fateful night is not a negative fact but a proposition. Whatever fails to occur is not a fact; likewise, a hole in a chunk of cheese is not a piece of negative cheese. And that all adult dogs bark is an empirical generalization, not a general fact. All facts are singular and “positive.” Because data are propositions, they can involve theoretical concepts – that is, they can be “theory-laden.” This is the case with all the data occurring in scientific reports, such as the data concerning specific weights, DNA sequences, and income inequalities. (A ready indicator of theory-dependence is
20
Chasing Reality
the occurrence of abstract nouns.) But of course the facts reported are not theory-laden. Moreover, even the most sophisticated scientific data result not only from procedures designed with the help of theories, but also from theoryfree observations, such as those of a yellow glow, an instrument pointer, and a trace on a photographic plate. The preceding clarification is intended to refute the rather popular thesis that there is no distinction between observational and theoretical terms because all observations would be coloured by theories. Actually, only some scientific observations are theory-laden, and even these must be completed with theory-free observations, such as that of the position of a pointer reading. Even Einstein (1950a: 62), the arch-theorist and anti-positivist, emphasized the need to connect the theoretical concepts with the “primary concepts,” that is, those that “are directly and intuitively connected with typical complexes of sense experience.” 3 The World: The Totality of Facts or the Maximal Thing? Astronomers equate the universe with the system of all things. By contrast, at the start of his Tractatus, Wittgenstein (1922) famously wrote: “The world is the totality of facts, not of things” (1.1). Regrettably, he did not bother to clarify the key terms ‘fact’ and ‘totality,’ as a consequence of which he soon fell into circularity. Thus, he tells us that “[a]n atomic [simple] fact is a combination of objects (entities, things)” (2.01), only to add: “It is essential to a thing that it can be a constituent part of an atomic fact” (2.011). So, a fact is a combination of things, but in turn a thing is a part of a fact. And nowhere does Wittgenstein offer illustrations to facilitate understanding and show that his cogitations are useful to analyse scientific problems. Though still widely discussed in the philosophical literature, and recently updated by Armstrong (1997), Wittgenstein’s mini-ontology is wrong for the following reasons. 1 It is not clear how the ambiguous word ‘totality’ (Gesamtheit) is to be read in the expression ‘totality of facts’: whether as whole (system) or as collection (set). The former interpretation does not sound right, because facts do not assemble; only things do. And if the second interpretation (set) is adopted, it turns out that the universe is a conceptual item, not a concrete thing. This may be fine in an idealist ontology but is wrong elsewhere, if only because physical cosmology treats the world (universe) as a concrete and therefore changing thing, not as an idea. The same holds of course for any materialist ontology. 2 Things can interact, whereas facts may not. For instance, the fact that your
Reality and Hylorealism
21
computer lies on its table does not interact with you. But you, your computer, and its table (things all of them) interact in various ways. Another example: The fact that your watch shows that it is noon does not interact with anything. But your perceiving it is an event that may stimulate you to go out for lunch. All these are processes going on in you, your watch, your office, your street, your eatery, and their surroundings – things, not facts, all of them. 3 All of the law-statements and social norms concern (refer to) things, such as dogs and nations, not facts, such as barking and trading. For example, cosmology is about stars and other “celestial” objects. But, of course, every thing is in some state (a fact) and it may jump to another state (another fact). No things, no facts. If Newton’s laws of motion did not refer to bodies, we could not use them to describe a fact such that the Earth is currently at such and such a distance from the Sun, or the process that a certain meteorite is moving dangerously close towards our planet. 4 The world is a system of things because every one of its component things interacts with some other things. If the world were a heap of facts, or states of affairs, it would not constitute a system, because it would not be held together by interactions. However, it is not a question of opting between two ontologies, one of things and another of facts: We only need a single ontology of things involved in facts (or facting things); or, what is the same, an ontology of facts involving things (or thinged facts). However, for the sake of logic we must start with things: We need the concept of a thing before we can form the idea of a state of a thing. This is why mereology, which is about things, is basic to ontology. (Mereology is the theory of the part–whole relation and of the physical addition operation, as in “The population of territory A plus territory B equals the sum of the partial populations.”) 5 The view that the world is a heap of facts might have been accepted by Berkeley, Hume, and Kant, provided ‘fact’ were restricted to ‘phenomenon’ or ‘appearance.’ But, because appearances do not assemble through objective and lawful connections, such a heap would not constitute a world or system. However, appearances deserve a separate section. In sum, the world is not the totality of facts but that of things. And all things are changeable, and every thing is related to some other things. 4 Enter the Knower So far the sentient being, or knowing and acting subject, has not appeared. The moment we decide to include interests and points of view in our consider-
22
Chasing Reality
ations, we will have to add another three categories to those of thing and fact. These are the concepts of phenomenon (appearance), construct (such as concepts and propositions), and intention. The subject has got to be included in a comprehensive ontology, not least in an objectivist and materialist one. This is because, whereas some facts are perceptible, others are not. Thus, car collisions, unlike atomic collisions, can be experienced. And if perceptible, a fact may appear differently to different observers, because perception depends not only on external stimuli but also on the subject’s knowledge, attention, and expectations. That is, one and the same external fact may elicit different appearances or phenomena – or none. Much the same holds, mutatis mutandis, for our conceptual models of facts and for our plans of action. This is not to endorse relativism, because it comes with an important rider: Appearances can be deceiving, conceptual models inaccurate, and plans unfeasible or unfair. Hence, all subjective (subjectcentred) accounts and decisions must be regarded as, at best, preliminary sketches to be perfected or rejected. At all events, our world pictures may have components of all three kinds: perceptual, conceptual, and praxiological (action-theoretical). This is because there are three gates to the outer world: perception, conception, and action. However, ordinarily only one or two of them need be opened: combinations of all three, as in building a house according to a blueprint, are the exception. We may contemplate a landscape without forming either a conceptual model of it or a plan to act upon it. And we may build a theoretical model of an imperceptible thing, such as an invisible extrasolar planet, on which we cannot act. In other words, perceptible things elicit appearances; interesting things, whether perceptible or imperceptible, are understood through conceptual models; and useful things call for plans of action. Note the subject-related concepts in the foregoing: “perceptible,” “interesting,” and “valuable.” They bridge the subject to the object, the knower to the known, the actor to the thing he wishes to act upon – in short, the inner world to the outer one. To continue in this metaphorical vein, we add that action – in particular work, science, technology, art, and politics – builds those bridges. Thus, a single thing may elicit several appearances, various conceptual models of it, or several plans of action for it, depending on the subject’s abilities and interests. That is, we distinguish, with Kant, the thing in itself from the various concomitant things for us – but we do not follow Kant in claiming that the former is unknowable, let alone non-existent. See figure 1.2. This is not all: The naive observer may be joined or displaced by the scientist, technologist, or philosopher intent on analysing or evaluating the things for us that accompany some things in themselves. See figure 1.3.
Reality and Hylorealism
wm
f
un (a)
(b)
u
...
...
u
m1 m2
...
w1 w2
u1 u2
...
w1 w2
wp
mg
(c)
23
(d)
u
Fiction
(e)
(f)
Figure 1.2M(a) Various appearances wi of a thing u. (b) Several things uj appear the same to the untrained observer. (c) Several phenomena but no discernible underlying thing, as in illusions. (d) Alternative models mk of one and the same thing u. (e) Unknown thing: neither appearances nor models thereof, such as Spencer's Unknowable. (f) Fiction: a construct with neither real nor phenomenal counterparts, such as the Cheshire cat or the “real” line in mathematics.
Ideas about things for us
Analyses or evaluations of w or m
Things for us
Appearances w
Thing in itself
Models m
Real and knowable thing
Figure 1.3MThe things for us (appearances and models) can be analysed or evaluated (for truth, utility, beauty, etc.).
Illusions constitute a particularly interesting case in point, and one that has drawn the interest of psychologists, neuroscientists, and artists (see, e.g., Gregory and Gombrich, eds. 1973). They are facts, but they occur in the brain rather than in the external world. Consider, for instance, the out-of-body experiences that spiritualists have invoked as evidence for the separability of mind from body. These illusions can be induced at will by electrical stimulation of the brain’s right angular gyrus. The subject reports that she sees herself from above, or with her legs becoming shorter, or moving quickly (see, e.g., Blanke et al. 2002). The “trips” of drug addicts are similar: they are just brain tricks. A more sophisticated case of the distinction in question is that between property and predicate. In principle, every property of a thing may be conceptualized in alternative ways. For instance, an electromagnetic field in a vacuum may be represented either by an intensity tensor with six components, or by a vector potential with four components. Neither representation is more correct
24
Chasing Reality
or realistic than the other, but each has its advantage: the first is closer to measurement, and the second is easier to calculate with. Which is the correct representation or formalization of a predicate representing a property of a concrete thing? I submit that predicates should be construed as functions, operators, or elements of an algebra, depending on the theory. Thus, in classical physics energies are construed as functions, and as operators in quantum physics; and spins are formalized as elements of an algebra (or as matrices). Take, for instance, age: it is a property of stars, rocks, cells, rock stars, social systems, and more. After measuring, computing, estimating or guessing the age of a thing q, we write “The age of q is so many time units,” or “A(q, u) = t” for short. This shows that A is a function of the form A: Q × U ® Q+, where Q stands for the set of aging things, U for the set of time units, and Q+ for the set of positive fractions. The point of the preceding exercise is to emphasize that, whereas properties are possessed by things, predicates are rightly or wrongly attributed to them. It is not that attributes are necessarily subjective; on the contrary, in science and technology they are expected to be objective, but some of them may turn out to be mistaken. For example, we may mistake a test score for an intelligence measurement, or a stock market index for a measure of economic health. 5 Subject / Object Separability Knowledge of matters of fact has traditionally been regarded as a particular case of the subject–object relation: namely, the relation between explorer and explored. In other words, empirical cognition involves a sentient being capable of detecting signals from an object of knowledge. For example, one can see a book provided the book reflects some light that ends up in one’s retina. Thus, the acquisition of knowledge depends on the possibilities of (a) distinguishing the knower or subject from the knowable or object; and (b) using or establishing an interaction between subject and object and, preferably, altering that interaction at will, as in an experiment. The first condition, distinguishability, depends on separability for, if the two terms of the interaction are not separable, then it is not possible to ascertain what the contribution of either is. Indeed, if the constituents of the subject/ object system were joined too tightly, the subject might think either that he is creator (idealism) or helpless creature (empiricism). In the first case he would not bother to check his ideas, whereas in the second he would not dare having any ideas beyond those suggested by his sensory experience. The problem of separability, at first sight trivial, is actually quite complex.
Reality and Hylorealism
25
Newton’s theory of universal gravitation showed that the universe is the maximal system: that every component of it interacts with every other constituent. The emergence of the Faraday-Maxwell electromagnetic field theory in the mid-nineteenth century reinforced this view. This for two reasons: because it introduced a further kind of cement among things; and because a field acts as a unit, a disturbance at any point in it propagates throughout the entire field. However, it was also known, or rather assumed, that all bonds weaken with distance. This makes it possible to study nearly isolated systems, such as the solar system, light beams in a vacuum, and atoms at very low temperatures. In short, although the universe is a system, it can be analysed into nearly selfcontained parts. And even when such parcelling is impossible, as in the case of an ecosystem, it is usually possible to single out a few salient factors, such as humidity and temperature, and regard the others as roughly constant background. The existence of tightly knit (“entangled”) concrete systems poses the problem of the reality of their constituents, particularly when the strength of their bonds does not decrease with their mutual distances. Before the experimental refutation of the Bell inequalities in the early 1980s, it was presupposed that if the components of a system are distant from one another, then they are separable into mutually independent systems. Since then we know that this is not so: that “once a system, always a system.” Such entanglement is often said to refute the postulate of “local reality.” What actually happens is that the constituents of a real system are real but not independently so: what occurs in, at, or to one of them affects the state of the other(s), even if they are in a remote place. This holds at all scales, not only for atoms and photons. Thus, babies are real even though they are inseparable from their caregivers. In short, non-locality does not imply irreality; it only implies that reality does not satisfy classical physics (see Bunge 1979a). The emergence of quantum mechanics seemed to have changed all this, in showing that the constituents of a microphysical system are entangled, hence inseparable – unless strongly disturbed by the environment. Luckily for us, however, such entanglement does not hold for sentient beings, both because the things in question are macrophysical, and because they interact strongly with their environment. Thus, the popular claim that the knowing subject and the microphysical objects he studies constitute a sealed unit is false. Consequently, quantum physics does not threaten objectivity. It is still true that the world is made up of protons, photons, and the like, whether we act upon them or not. All the same, we are faced with this dilemma: No subject–object interaction,
26
Chasing Reality
no knowledge; but the interaction must not be so strong as to lead to confusing the two terms of the interaction. More precisely, this interaction must be such that the subject receives a clear signal, such that the Signal ® Object action is far weaker than the Object ® Subject reaction. The fulfilment of this condition depends critically upon the experimenter’s competence and ingenuity. His equipment may involve shielding plates, vacuum installations, micromanipulators, and other contrivances designed to avoid disturbing the object without blocking the latter’s signals. 6 Materialism Philosophical materialists and idealists have been fighting each other for more than two millennia. Whereas idealists like Plato assert the independent existence of ideas, materialists and Aristotelians deny it. Moreover, idealists either deny the independent existence of things outside the mind (the case of Berkeley); or, like Plato and Hegel, they admit that there are concrete things, but all of them deriving from ideas or ruled by them. This strife has been so bitter that philosophical materialism has been ignored or vilified as “crass” or even immoral, even in university teaching. Consequently, this ontology has been generally ignored, misunderstood, or demonized. In particular, materialism is still widely identified with the physicalism of the Greek atomists; the amoralism of some of their Indian contemporaries; the distrust of ideas inherent in medieval nominalism; Nietzsche’s vulgar (and brutal) materialism; or the genetic reductionism of the sociobiologists. In sum, it is seldom recognized that there is a vast family of materialist doctrines. But of course all of them share the principles that the world is constituted exclusively by concrete or material entities, and that thoughts are brain processes. However, the materialism/idealism strife is partly a matter of definition, because there is no consensus on the meanings of the terms ‘material’ and ‘ideal’ or their respective synonyms, ‘concrete’ and ‘conceptual.’ For example, some materialists, such as Steven Weinberg (1992: 3), claim that matter has lost its central role in physics, because they stick to the traditional definition of “material” as something characterized by mass (or inertia) – as if photons, which are massless, were immaterial. Others, notably the upholders of the computationist (or information-processing) philosophy of mind and ontology, claim that stuff does not matter: that all there is are bits and algorithms to process them (see, e.g., Barrow, Davies, and Harper 2004). As will be recalled from section 1, I submit that whatever is capable of changing in a lawful manner, from electron and gravitational field to person and society, is material. And, since being changeable is the same as possessing
Reality and Hylorealism
27
energy, the predicate “is material” turns out to be coextensive (though not cointensive) with “has energy.” By contrast, unchangeable items, such as sets, numbers, and functions, are ideal, because they possess no energy. Thus, since electromagnetic fields have energy and “obey” physical laws, they are material even though they are massless and subtle rather than solid. Contrariwise, a theory of such fields, that is, an electrodynamics, is immaterial because it has no energy, even though it can be altered by physicists. The theory, like any other construct, is a conceptual (or ideal) object. The materialist thesis is not just that there are material objects, but that the world contains only material things. More precisely, a materialist ontology includes the following two basic assumptions. Postulate 1.2 Every object is either material or conceptual, and none is both. Postulate 1.3 All the constituents of the world (or universe) are material. A corollary to the first postulate is that, contrary to the hylomorphism attributed to Thomas Aquinas, there are no compounds of matter and “forms” (ideas). Given our broad concept of matter, the second postulate is not physicalist: it does not exclude such supra-physical material things as organisms and social systems, all of which are characterized by emergent properties. In other words, this theory eschews radical reductionism. This is why it may be named emergentist materialism (see, e.g., Bunge 1977a, 2003a). This version of materialism does not eliminate the mental; it just denies the autonomous existence of ideas. Nor does it outlaw all fictions. It only locates the mind in the brain, and tries to show why sometimes we need fictions, particularly mathematical ideas, to understand facts. (More on fictions in chapter 8.) Moreover, emergentist materialism blends easily with realism to constitute what may be called hylorealism. However, before praising realism we must characterize reality. 7 Reality We have not defined ‘material’ as whatever exists independently of any minds, for this is what ‘real’ means. More precisely, we propose the following: Definition 1.1. Real things are those that exist independently of any subject. Of course, the objective idealists such as Plato, Leibniz, Hegel, Bolzano, and Dilthey have claimed that ideas exist objectively, not only in the privacy of human minds. But they have not bothered to substantiate this hypothesis. Moreover, it is empirically untestable, because the only way we can check whether an object exists really is by watching its physical actions – but, of course, ideal objects cannot exert such actions.
28
Chasing Reality
In turn, “reality” may be defined either as the set of all real things, in which case reality is unreal; or as the maximal real thing – the one constituted by all things. In the first case it turns out that reality, being a set, is unreal. Since this is not the normal usage of the term, we shall opt for equating reality with the physical (or mereological) sum of all things – that is, the universe. The term ‘exist’ that occurs in the above definition is ambiguous, for existence may be either concrete (material) or abstract (ideal). Modern logicians, from Russell to Quine and beyond, have claimed that this imprecision is remedied by using the so-called existential quantifier $. For example, “There are prime numbers” would be symbolized ‘$x (x is a prime number).’ This device certainly works in mathematics, which deals exclusively with ideal (or abstract, or conceptual) objects. But it fails wherever ideal and material objects are referred to side by side, as is so often the case in ordinary knowledge, science, and religion. For instance, an atheist has no problem with the statement that some angels are guardian angels, so long as the real existence of such supernatural beings is not stated separately. He prefers to add the explicit denial of their real existence. That is, he chooses to state something like this: “Some angels are guardian angels, but in reality there are no angels.” This statement contains the quantifier “some,” formalized by $, along with the predicate E “exists.” That is, the atheist would affirm that $x(Ax &5 ERx), where R stands for the set of all real things. The existence predicate has been defined elsewhere (Bunge 1977a: 155–6). (The existence predicate E may be defined as follows. Let W be a universe of discourse, and S a proper subset of W. Further, consider the characteristic function cS of S, that is, the function cS: W ® S such that cS(x) = 1 if x is in S, and 0 otherwise. We now define the existence predicate ES as follows: ESx = (cS(x) = 1). If S = R, we have real existence, whereas if S = C, we have conceptual (or ideal) existence. Scientists use several criteria of real existence, such as observability and reactivity (“kicking back”). In pure mathematics, utterly different existence criteria are used, such as definability, constructivity, and absence of contradiction.) Since things come with their properties and the changes thereof, these too are real. So, the expressions ‘Property X is real’ and ‘Process Y is real’ must be understood as affirming the reality of the underlying thing(s). Materialists identify reality with materiality. That is, the main assumption of materialist realism is Postulate 1.4 All and only material things, complete with their properties and changes, are real. In other words, the concepts of real existence, materiality, and possessing energy are postulated to be coextensive, even though they have different
Reality and Hylorealism
29
senses or connotations, as shown by their definitions. Postulates 1.1 and 1.3 jointly entail Theorem 1.1 All changeable items are material and conversely. To put it in somewhat paradoxical terms: To be (material, real) is to become. This identification of reality with materiality is not a philosophical extravagance but normal in factual science. For example, in criticizing Mach’s subjectivism, Ludwig Boltzmann (1979: 112) equated realism with the centrepiece of materialism, namely, the so-called identity theory of mind: “[T]he mental processes are identical to certain material processes in the brain (realism).” Today’s cognitive neuroscientists are likely to concur. However one may, like Plato and Leibniz, embrace at once realism and immaterialism. That is, one may hold that concrete things are only shadows, copies, or corruptions of self-existing ideas. That this thesis is at best untestable, and at worse false, is beside the point. My point is that realism is logically independent of materialism – but vulnerable unless united with it. However, the matter of realism deserves a separate section. 8 Realism Realism is the thesis that there are real things. However, like any other comprehensive philosophical system, realism has seven components: ontological, epistemological, semantic, methodological, axiological (value-theoretical), moral, and praxiological (action-theoretical). Besides, every one of these constituents comes in various shades: naive, critical, and scientific. Hence, one may distinguish altogether twenty-one possible kinds of realism. One way of distinguishing these various doctrines is to state and analyse the principles of scientific realism, the more restrictive and therefore most complex form of realism. This doctrine may be summarized into the seven principles conjoined by the following Postulate 1.4 Realism is the philosophical system constituted by the following seven theses: 1 Ontological realism: The external world exists independently of the knowing subject. 2 Epistemological realism: (a) The world can be known. (b) All knowledge of facts is incomplete and fallible, and much of it is indirect. 3 Semantic realism: (a) Some propositions refer to (are about) facts;
30
4 5 6
7
Chasing Reality
(b) some such (factual) propositions are approximately true; (c) in principle all approximations are perfectible. Methodological realism: The best strategy for exploring the world is the scientific method (scientism). Axiological realism: There are objective values, such as health, knowledge, security, peace, environmental protection, and fairness. Moral realism: There are (a) moral facts, such as generous deeds and selfish ones; and (b) true moral principles, such as “Rights, to be fair and respected, must be balanced by duties,” and “Solidarity and democracy favour coexistence.” Practical realism: There are objective <means-goal> pairs, such as <work, well-being>,, and <participation, democracy>.
Let us next focus on ontological, epistemological, and semantic realism. The remaining components of the system will be further examined in chapter 10. Naive realism holds theses (1), (2a) and (3a). Critical realism consists of the first three complete theses. Thesis (4), also called ‘scientism,’ is peculiar to scientific realism. Planck (1933: 82) emphasized clause (2b), which he stated as follows: “The real outer world is not directly knowable.” This was his response to the positivist thesis that sense impressions supply direct knowledge and are the only source of knowledge. A familiar counter-example to the positivist thesis is this: The light given off by a light source, such as the Sun, does not point directly to the composition and structure of the source; it only poses the problem of conjecturing these features – for instance, that our star is constituted mostly by hydrogen, that the hydrogen atom has a single electron, that the latter can be in any of infinitely many states, and that every transition between states is accompanied by the emission or the absorption of a photon whose frequency is proportional to the difference between the energies characteristic of those states. Contrary to a widespread opinion, scientific realism does not claim that our knowledge of the outer world is accurate: it suffices that such knowledge be partially true, and that some of the falsities in our knowledge can eventually be spotted and corrected, much as we correct our path when navigating in new territory. Thus, the fallibilism of thesis (2b) is balanced by the meliorism of thesis (3a). Reality checks (empirical tests) will show again and again that even the most accurate theories are at best more or less close approximations that can be improved upon. Rescher (1987) calls this thesis approximationism; I regard it as a constituent of scientific realism (e.g., Bunge 1967a). The frequent occurrence of error, perhaps even better than the occasional finding of truth, proves the existence of the real world (Bunge 1954). Indeed,
Reality and Hylorealism
31
while a subjectivist could argue that scientists construct the world as they perceive or conceive it, he would be unable to account for discrepancies from the truth, particularly since falsities and approximate truths occur more frequently than completely accurate truths. In particular intuitionists, such as Bergson and Husserl, cannot account for error because they claim to have instant access to full truths. Finally, antirealism is of course the opposite of realism. Like the latter, it consists of seven branches, every one of which comes in various shades. The most popular version of antirealism is subjectivism of the phenomenalist kind, from Kant to logical positivism. It will be examined in the next chapter. 9 Objectivity and Impartiality So far we have stressed some ontological notions, in particular that of ontological objectivity or independent existence. Let us now turn to epistemological objectivity and its kin. A factual proposition is said to be objective if it refers to real existents in an impersonal manner, and describes them to the best of its author’s knowledge (see Rescher 1997). Ideally, objective propositions are true; in practice most propositions, particularly the quantitative ones, are at best approximately true. Impartiality, though related to objectivity, is different from it: an impartial judgment is one that takes no sides in a conflict. No conflict of interests, neither partiality nor its contrary. Yet, partiality is compatible with objectivity. For instance, universal medicare may be favoured not only on moral grounds but also because a healthy population is in the interests of everyone: just think of infectious diseases and of the many resources wasted because of sickness. Max Weber (1988 [1904]) famously demanded that the social and political sciences be objective. He wanted to purge these disciplines from value judgments, all of which he (mistakenly) regarded as unavoidably subjective. Weber wished, in particular, to prevent the ideological contamination of the social studies. This goal is laudable, for scientific research is, by definition, a search for objective truths, whereas ideologies are biased and often mendacious. But are value-neutrality and impartiality possible? The postmoderns deny it: they hold that, since knowledge is power, and since research is ordinarily funded by the powers that be, power lurks behind every research project, at least in the field of social studies. In particular, Habermas (1970) and other members of the Frankfurt (or “critical theory”) school, along with a number of student activists in the 1960s, incurred two false identifications. These were “Science = Technology” and “Science (or technology) = Ideology of late capitalism.” These writers did not
32
Chasing Reality
understand that, whereas basic science is the disinterested search for truth, technologists design artefacts, and ideology is the intellectual component of a social movement. However, let us return to Weber. I submit that Weber conflated three different concepts: those of objectivity, impartiality, and disinterest. Worse, he failed to be objective whenever he put into practice the neo-Kantian dogma that the aim of social science is to “understand” or “interpret” the inner or subjective life rather than to describe the so-called material conditions of existence, such as means of livelihood, work safety, and freedom to join unions fighting for a measure of social justice. No doubt, an objective student of society will pay attention to attitudes and subjective evaluations. But emphasis on the “subjective area,” with disregard for the objective conditions of existence, is not only a methodological lapse but also a piece of ideological partisanship. It was no coincidence that Weber’s first empirical work, of 1892, was his participation in a survey of the conditions of rural workers conducted by the Verein für Sozialpolitik. “Its core was a group of university professors who were worried about the growing antagonism of German workers, organized in socialist unions, toward the German state” (Lazarsfeld and Oberschall 1965: 185). Another fact that alarmed Weber was that, by importing Polish labourers, the East Prussian landowners endangered “the German character and the national security” of the east frontier of the German Reich (ibid.: 186). Thus, on this occasion Weber put his science in the service of his liberal and nationalist ideology. Why is Weber generally regarded as more scientific than Marx and Engels, Marx’s co-worker, who described objectively the material conditions of the English working class on the strength of first-hand research as well as reports by industrial inspectors appointed by Her Majesty’s government, is yet to be explained objectively. In any event, Weber’s objectivity thesis was hardly original. In fact, it was what Thucydides and Aristotle had practised in antiquity, Ibn Khaldûn in the Middle Ages, Niccolò Machiavelli in the Renaissance, and Leopold Ranke in the early nineteenth century. Adam Smith, David Ricardo, Alexis de Tocqueville, Karl Marx, and Emile Durkheim had been no less committed to the search for truth in social matters, even though each of them had also his own axe to grind – as did Weber. In sum, objectivity, though often hard to attain, particularly in social matters, is both possible and desirable. Moreover, it is mandatory in matters of cognition. However, objectivity must not be confused with value-neutrality, because the pursuit of certain values, such as well-being, peace, and security, is objectively preferable to the pursuit of others, such as the pleasure derived
Reality and Hylorealism
33
from getting drunk or attending a public execution. Objectivity is also desirable in the fields of technology, public policy, and the struggle for power, since in these cases a false picture of reality is bound to result in practical failure. In short, objectivity is desirable and attainable everywhere except in art. In other words, realism is the philosophy of knowledge inherent in successful science, technology, and praxis. 10 Concluding Remarks We have attempted to elucidate the notions of thing and fact because they are far from clear in the philosophical literature. And, contrary to a certain tradition, we have neither confused realism with materialism nor kept them apart. After distinguishing them we have joined them in a doctrine that may be dubbed hylorealism. One reason for wedding realism to materialism is that the idea of irrealist materialism is an oxymoron, since the statement that the universe is material amounts to the claim that the non-conceptual is known to be material rather than, say, ghostly. Another reason is that materialism without realism is both pointless and toothless. Indeed, what would be the point of exploring a purely imaginary or totally unknowable material universe? And how firm would realism be without the assumption that all real things, however ephemeral or artificial, abide by laws that are either physical or rooted, however remotely, in physical laws? In general, epistemology and ontology, though distinguishable, are inseparable. For instance, radical (or dogmatic) rationalism calls for an idealist ontology, because only abstract ideas can be known (invented or learned) without the aid of experience. And radical (or dogmatic) empiricism requires a phenomenalist ontology, because experience deals in qualia, not in primary properties, which are the ones characterizing things in themselves. Whoever admits the existence of things in themselves rejects both dogmatic rationalism and dogmatic empiricism, and adopts instead ratio-empiricism (or empiriorationalism), a synthesis according to which human cognition makes use of both reason and experience. This is so for two reasons. One is that the overwhelming majority of facts are imperceptible, hence accessible only through conception. The other reason is that the normal human brain is active and inventive as well as reflexive; in particular, it looks for ideas or deeds behind words, reasons behind assertions, and mechanisms behind appearances. However, the concept of an appearance or phenomenon is so slippery, and yet so central to many influential philosophies, that it requires an entire chapter – the next one.
34
Chasing Reality
2 Phenomena, Phenomenalism, and Science
The distinction between subject and object, or explorer and territory, is commonsensical. This distinction is enshrined in naive realism, the tacit epistemology of nearly everyone. Moreover, that distinction would seem to be essential to animal life: just think of the chances a gazelle would stand if did not instinctively acknowledge the real existence of lions in its outer world. Yet, a famous Harvard philosopher (Putnam 1990: 122) once announced that “the idea of discourse-independent objects ... has crumbled under philosophical critique” – in particular that of Wittgenstein, Carnap, and Quine. So, don’t worry, gazelle, the lion is only in your mind; go see a shrink. (To his credit, shortly thereafter Putnam [1994] abandoned antirealism.) The earliest irrealist philosophy is found in the Upanishads (ca. 800 BC). According to these theologico-philosophical texts, the world is illusory: only Brahman, the godhead, would be real. Three centuries later, the Buddha taught that there is an unknowable reality hiding behind appearances, but no god. At about the same time, the view was attributed to Protagoras – alas, falsely – that there are no subject-independent things, hence no objective truths. Illusionism has remained popular in India to this day, but did not prosper in the West. In fact, the overwhelming majority of ancient Greek, medieval, and early-modern thinkers in the West have been objectivists (realists). Subjectivism became important only at the beginning of the eighteenth century, partly as both an expression of the rising individualism and a reaction against modern science. Recall Berkeley and his followers, up to the social constructivists. Only Hamlet, an idle prince, could afford to entertain the possibility that life is a dream: Shakespeare and his European contemporaries were busy constructing modernity. The ancient questions – whether all things are as they appear to be, and whether we can know anything beyond appearances – have resurfaced inter-
Phenomena, Phenomenalism, and Science
35
mittently in modern philosophy and in science: witness the epistemological discussions of Galileo, Descartes, and Locke; Berkeley, Hume, and Kant; Comte, Mill, and Mach; and their twentieth-century followers, from Russell, Bridgman, Carnap, and Reichenbach to Nelson Goodman, Thomas Kuhn, Paul Feyerabend, and David Lewis. The same questions, with similar answers, reappeared from the mid-1920s in the controversies over the interpretation of quantum mechanics. They even occur in Michael Frayn’s famous play Copenhagen. All of these ideas are involved in one of the oldest philosophical disputations: that between phenomenalism and realism. What reasons have been adduced in favour of either view, and why does it matter which horn of the phenomenalism/realism dilemma one chooses? These are some of the topics to be discussed in this chapter. 1 Phenomenon and Noumenon In ordinary language, the terms ‘phenomenon’ and ‘fact’ are synonymous. Not so in philosophy, where the Greek word ‘phenomenon’ (appearance) stands for “an observable fact or event ..., an object or aspect known through the senses rather than by thought or nonsensuous intuition” (Webster’s New Collegiate Dictionary). Thus, colours, sounds, tastes, smells, and textures are sensory or phenomenal properties, whereas wavelengths, atomic weights, chemical compositions, planetary orbits, other people’s thoughts, and political crises are non-sensory or non-phenomenal. Phenomenal properties or “raw feelings,” such as the smell of mint and the feel of the beloved’s skin, are features of sensory experience. They are also called qualia. All sentient organisms experience some qualia, whereas no machines, not even robots, have them. Nor do zombies, if they exist. Nonsentient things have and detect only physical (or chemical, biological, or social) properties. We learn these distinctions early in life, and forget them on occasion if exposed to an irrealist philosophy. Indeed, presumably we start life as spontaneous phenomenalists: the world around a newborn human is likely to look and feel like a rather chaotic mass of appearances – tactile sensations, visual images, sounds, tastes, and smells. In other words, presumably babies describe the world and themselves in terms of secondary properties such as “soft,” “wet,” “warm,” “slimy,” “smelly,” “tight,” “bright,” “loud,” and “frightening.” As we crawl away from the playpen and learn, we add primary properties to our repertory: “long,” “round,” “heavy,” “swift,” and so on. That is, eventually we become realists – first of the naive kind and later on, upon further reflec-
36
Chasing Reality
tion, critical realists. Much later, with luck and further learning, and barring encounters with subjectivist philosophies, we may end up as scientific realists. Kant famously called non-phenomenal things noumena, or things-in-themselves, by contrast to the things-for-us, phenomena, or things as perceived. However, as we shall see, he could not make up his mind on whether phenomena can exist without any underlying noumena. Many contemporary philosophers still debate the status of qualia. Physicalists (vulgar materialists) deny their existence, whereas radical empiricists claim that qualia are the source and origin of everything else. Meanwhile, biologists and cognitive neuroscientists take it for granted that phenomena are made out of noumena, and attempt to find out how qualia emerge from processes in the organism/environment interface. To the extent that they succeed in this endeavour, they confirm emergentist materialism (e.g., Sellars 1922, Bunge 1979a, Blitz 1992). This is the variant of materialism that embraces Spinoza’s central thesis: One substance, many properties. This doctrine combines substance monism with property pluralism. Hence it admits the limitation of radical reductionism (Bunge 2003a). By the same token, emergentist materialists reject physicalism or eliminative materialism, which denies qualitative novelty and therefore impoverishes both human experience and its science, namely, psychology. Freshmen face the same problem when taunted by the old chestnut, Did the tree falling in a remote forest make a noise if there was no one around to hear it? If smart, they will answer that there was sound (in fact a shock wave) without loudness: a noumenon without a phenomenon. The standard or Copenhagen interpretation of quantum mechanics makes the opposite claim: it states that electrons and the like lack properties of their own, but acquire whatever properties the experimenter decides to endow them with. We shall argue that this extraordinary claim derives from confusing reality with reality tests – the source of operationalism. This problem, like most ontological queries, has ancient roots. For example, the Buddha was a phenomenalist. The earliest Western phenomenalist seems to have been Protagoras of Abdera, and the earliest treatise on the subject Sextus Empiricus’s Hypotyposes. The former is famous for having stated the principle “Man is the measure of all things.” And Sextus is well known for having relied consistently on sense data, and for having argued in detail, alas, often sophistically, against all the scholars known to him. Many later eminent philosophers, like Hume and Kant, as well as some scientists while in a philosophical mood, like Mach and Bohr, have claimed that only appearances count. A similar thesis is this: The secondary properties or qualia – such as brightness, loudness, warmth, softness, wetness, sweetness, roughness, and coloured – are basic either ontologically or epistemologically. However, the distinction between the two kinds of properties deserves a new section.
Phenomena, Phenomenalism, and Science
37
Table 2.1 Sample of primary (or objective) and secondary (or subjective) properties. Whereas some of the former are in the outer world, all of the latter are in the brain. Primary properties
Secondary properties
Position Time Size Velocity Wavelength Sound intensity Sound frequency Temperature Viscosity Engram Activation of the amygdala Deprivation Atomic mass Spin Electric charge Entropy Valence Dissociation energy
Place Perceived succession Bulk Swiftness Colour Loudness Pitch Thermal sensation Porability Recall Fear Dissatisfaction — — — — — —
2 Primary and Secondary Properties Realists hold that properties come in two kinds: primary or subject-independent, and secondary or subject-dependent. For instance, light wavelength is primary, whereas colour is secondary because it emerges in the brain. By contrast, according to phenomenalism, all of the primary or objective properties, such as velocity, energy, entropy, chemical bond, dissociation energy, metabolic rate, population density, GDP, and the social order, would be derivative or even imaginary, because they are not apprehended by the sense organs. See table 2.1. An equivalent way of drawing the same distinction is to speak of phenomenalist and realist (or physicalist) propositions or sentences (e.g., Kaila 1979). The phenomenalist claims that the phenomenalist propositions are primary, and the realist ones secondary, in the sense that they are somehow derivable from the former. Of course, such derivation cannot be direct: the phenomenalist and physicalist propositions are separated by a chasm as long as the predicates occurring in them remain self-contained. However, this chasm can be bridged by assumptions or definitions containing concepts of the two kinds. A clear example of such bridge is Kant’s (and Mill’s, Mach’s, Carnap’s, and Husserl’s) definition of a thing as a possibility of experiences or sensations.
38
Chasing Reality
One trouble with this alleged reduction is of course that psychology is the only science that studies sensations – moreover, it attempts to explain them as brain processes, which are in turn described in terms of primary properties. Another problem is that physics is demonstrably irreducible to psychology: just try to explain electromagnetism and nuclear physics, say, in terms of perception, memory, emotion, and the like. 3 Phenomenalisms: Ontological and Epistemological The philosophical doctrine that admits only phenomena, that is, appearances to someone, is called phenomenalism. Phenomenalists have no use for noumena, such as the things or events that elicit our sensations, or the items that, though deemed to exist, no one could possibly perceive. Think, for example, of the collision of the earth’s plates that caused the latest earthquake; of the electrons that circulate in your computer; of the neuron firings involved in your watching your computer’s screen; or of the surges of greed and fear behind the latest stock-market fluctuations. All this and much more is beyond the phenomenalist’s ken – as is the universe as a whole. However, two kinds of phenomenalism must be distinguished: ontological and epistemological. The ontological phenomenalist claims that there are only phenomena, whereas the epistemological phenomenalist holds that only phenomena can be known. In other words, Ontological phenomenalism: Existent = apparent Epistemological phenomenalism: Knowable = apparent Ontological phenomenalism, such as that of Berkeley, Renouvier, Avenarius, Mach, Ostwald, Carnap, and Bohr, places man at the centre of the universe: it is anthropocentric. By contrast, epistemological phenomenalism, such as Plato’s, Ptolemy’s, Hume’s, Duhem’s, and Spencer’s, is less radical. Indeed, it makes no claims about reality, except to declare it unknowable; it just restricts knowledge to perception. (Recall Plato’s allegory of the cave: we can see only the shadows projected by real things, which dwell in the Realm of Ideas.) Kant, as will be seen below, wavered between the two kinds of phenomenalism, asserting the former on one page, and the latter on the next. Kant’s main successor, Hegel, left phenomena to science, and kept noumena for himself. Indeed, he told scientists to restrict themselves to phenomena and inductive generalizations. Among these he counted Kepler’s laws, which he misunderstood; moreover, he claimed that they entail Newton’s laws of motion (Hegel 1969: sec. 270). He, Professor Hegel, would take care of the things in themselves – revealing, for instance, that “air in itself is fire” (1969: sec. 283); and that “the magnet represents in a simple naïve fashion the nature of the
Phenomena, Phenomenalism, and Science
SUBJECT
Secondary properties
OBJECT
Secondary properties
39
Ontological phenomenalism
-------------------------------------------------------------------------------------------------------SUBJECT
Secondary properties
OBJECT
Secondary properties
Epistemological phenomenalism
-------------------------------------------------------------------------------------------------------SUBJECT
Primary & secondary props.
OBJECT
Primary properties
Ontol. & epistem. realism
Figure 2.1 The two phenomenalisms – ontological and epistemological – and double realism. Ontological phenomenalism: The knower constructs the object, which is entirely sensuous. Epistemological phenomenalism: The object, which is phenomenal, induces phenomena. Neither type of phenomenalism has any use for primary properties. Realism: Phenomena emerge at the subject-object interface, as when we perceive a grass knoll as green, or a candy as sweet. Both the object and the subject have primary properties; but of course the knower senses also secondary properties in addition to some primary ones.
concept, and in its developed form that of deduction” (ibid.: sec. 312). Hegel’s sanity is in doubt. What is certain is that he, along with Fichte and Schelling, adopted the worst components of Kant’s apriorism, and were the first moderns to successfully pass off convoluted nonsense as deep philosophy. Clearly, ontological phenomenalism entails epistemological phenomenalism. Indeed, if there are only phenomena, then only phenomena can be known. And phenomenalism of either kind opposes realism, the view that the external world exists by itself and is knowable to some extent. See figure 2.1.
40
Chasing Reality
Red
Blue
Yellow Green
0 (a)
(b)
Figure 2.2 (a) The line of light wavelengths. (b) The circle of colours.
4 Qualia in Materialism Scientific realists and emergentist materialists do not deny the existence of qualia. And they know that these are quite different from physical objects and their primary properties. For example, whereas the wavelengths of the light that causes colour sensations are ordered on a linear continuum, the corresponding colours may be arranged around a circle: see figure 2.2. Realists and emergentist materialists just hold that qualia emerge in some nervous system, usually as a result of external stimuli. Consequently they propose that qualia be studied by psychology, in particular psychophysics and cognitive neuroscience, rather than by apriorist philosophy. As Clark (1993: viii) says, the scholars who support this research project “do not argue that current explanations are complete, or even true, but merely that the approach has no conceptual flaws, and can answer the various a priori objections. In short, it could succeed. That finding would suffice to defeat the skeptic, or at least to postpone the onset of melancholia.” People like us, used to thinking of such unobservable entities as atoms, force fields, DNA molecules, neurons, dinosaurs, nations, and the universe as a whole, may find it difficult to understand how such eminent thinkers as Hume, Kant, Mill, Mach, at one time Russell, Bohr, Heisenberg, and Born, to mention only a few, could have espoused phenomenalism. This is particularly baffling in the case of sceptics like Hume, and atomic physicists like Bohr and his circle. Perhaps such incongruity can be understood by placing the doctrine in question in historical perspective. After all, phenomenalism had sometimes been used to cast doubt on the supernatural and to curb the wild imagination of metaphysicians. Let us therefore go back a few centuries. 5 From the Scientific Revolution to Locke The story of the seventeenth-century Scientific Revolution has been told many times. Here I will recall only three philosophical features of that great move-
Phenomena, Phenomenalism, and Science
41
ment that are pertinent to our subject. These are the ontologies that those revolutionaries contended with; the distinction between primary and secondary qualities; and the concept of a law of nature – or objective and necessary connection – without which the world appears, in William James’s phrase, as “a blooming, buzzing confusion.” I submit that, at the time of Galileo, Descartes, Harvey, Boyle, and the other pioneers of modern science, there were four major contending worldviews: (a) the magical view, which imagines the world teeming with supernatural beings; (b) the commonsense or ordinary-knowledge ontology, centred on the data provided by the senses; (c) Aristotelianism, which tries to account for everything in terms of perception, occult qualities, and causes of four kinds (formal, material, final, and efficient); and (d) the emerging mechanistic worldview, according to which there are only bodies and their microscopic components, along with causes of a single kind – efficient. The heroes of the Scientific Revolution hardly bothered to criticize the magical worldview: they addressed the educated public. The ontologies that the heroes of the Scientific Revolution sought to undermine were the commonsense one and the scholastic version of Aristotle’s physics. The scientists did this in two ways: by criticizing these views and by replacing them with clear and testable, if not always true, hypotheses concerning material entities in motion, such as Copernicus’s model of the solar system, Galileo’s law of falling bodies, Kepler’s laws of planetary motion, Boyle’s law of gases, Huygens’s theory of light, and Harvey’s theory of the cardiovascular system. But the Scientific Revolution was far more than a collection of scientific discoveries: it also involved a new worldview, mechanism, and a new epistemology, scientific realism. The root of both is the distinction between primary and secondary properties. The former are the geometrico-mechanical qualities inherent in things, and thus objective. By contrast, the secondary properties are subjective: they are the sensations and feelings caused by the external objects on sentient beings like us. However, this distinction was understood to be provisional, since in the end the secondary qualities were assumed to be reducible to the primary ones, as when perception was explained in terms of atoms penetrating into the sense organs (Dijksterhuis 1986: 431–3). Galileo proposed the key distinction between primary and secondary properties in Il saggiatore (1953 [1623]: 312). Here he stated that tastes, smells, colours, and the like “reside only in the sensitive body, so that, if the animal is removed, all those qualities are removed and annihilated.” For instance, if someone touches slightly our foot sole, we will feel tickled, a “condition that is all ours, and not the hand’s.” Following Galileo (though without quoting him), Descartes clarifies the primary-secondary distinction at several places. For example, he starts Le
42
Chasing Reality
monde (1664 [1633]: 7–10) with a discussion of the difference between our sensations and the things that produce them – precisely the difference that phenomenalists deny. There he attempts – alas, vainly – to account for the burning of a log in terms of the “very fast & very violent movements” of its tiny parts. He contrasts this mechanical explanation with the empty verbiage of the schoolmen in terms of the “form” of fire, the “quality” of heat, and the “action” that allegedly burns the wood – as if fire were not enough. Descartes adds that the sensations that fire causes in us – such as those of warmth and pain – are not properties of the fire although they are caused by it. He thus warns against confusing secondary with primary qualities. Let us now cross the channel and advance the calendar by six decades, to look briefly at John Locke. He too adopted Galileo’s distinction: he was an ontological realist, and moreover occasionally tempted by materialism, as when he famously wondered whether matter might think. He thought it obvious that the external bodies exist by themselves. Moreover, Locke (1690: bk. II, sec. viii, 9) asserted that the primary qualities “are utterly inseparable from the body, in what estate soever it be.” By contrast, the secondary qualities, such as colours, sounds, tastes, etc., “are nothing in the objects themselves.” However, he also held (ibid.: bk. IV, sec. iii, 28) that “we can have no distinct knowledge” of the motions of bodies beyond our experience, because we do not understand how they cause sensations in us, other than by the intercession of “an infinitely wise Agent.” So, “we are utterly uncapable of universal and certain knowledge” of the bodies around us. Little did Locke suspect that his scepticism concerning the power of science was behind the times, since the Scientific Revolution was already well advanced. In particular, he was unaware that Newton’s magnum opus (1687), which contained precisely some of the laws of motion that Locke had decreed unknowable, appeared the same year that he finished his Essay. (Both men interacted, though only briefly and about matters of government.) Luckily, not even Locke’s great intellectual authority could prevent the triumphal march of Newtonianism. However, his scepticism regarding the power of science eclipsed the important work of his near-contemporary Thomas Willis, the early-modern neuro-anatomist who regarded the brain as the organ of emotion, perception, and memory (Zimmer 2004). This is one more example of the harm that arrogant philosophers can cause. Would Locke have adopted a less sceptical stance concerning the power of the intellect to get to know “the mechanical affections of bodies” had he known of Newton’s work and its sensational success? I doubt it for the following reasons. First, Locke did not have the mathematics required to read, understand, and appreciate Newton’s formulas. Second, whoever accepts
Phenomena, Phenomenalism, and Science
43
Newtonian mechanics must give up Locke’s empiricism, for the former involves concepts that, like those of mass, acceleration, and gravitational interaction, are absent from the empirical data serving to check the theory. Ironically, Leibniz (1703) refuted Locke’s empiricism but rejected Newton’s mechanics. Be that as it may, Locke’s scepticism about science, together with his belief that only God could effect connections among things, unwittingly opened the door to Berkeley’s subjectivist philosophy – of which more anon. 6 The Counter-Revolution, Phase 1: Berkeley Not everyone accepted the mechanical worldview that replaced Aristotle’s organicism. After all, the view that the world is a clock seemed drab and grey in excluding everything that makes life worth living, from colours, tastes, textures, and scents to feelings, passions, ideas, and values. Hence most artists, theologians, and humanists were bound to react vehemently against mechanism. It is generally believed that the first anti-mechanist and anti-realist reaction was Romanticism, from Rousseau, Vico, and Burke to Goethe, Schelling, Fichte, Hegel, Schopenhauer, and Coleridge. Actually, the first reaction came much earlier and from an unexpected quarter: it was the radical phenomenalism of Berkeley, Hume, and Kant. In this regard, these philosophers were the first Romantics. But of course they were pre-postmoderns, as Merton would say, in that they upheld rationality even while defending their most outlandish doctrines. Berkeley’s animadversion to mechanism is generally acknowledged because of his explicit and clever attacks upon Newton’s physics and mathematics. But because Hume ignored Newtonianism, and Kant paid lip service to it, and because both philosophers were agnostic, we tend to forget that both were just as subjectivist as Berkeley. Worse, we are often told that Hume and Kant were the philosophers of the new science, while actually they undermined it while understanding nothing about it. Let us see why. Like Locke, George Berkeley held that to know is to perceive. But, unlike Locke, he famously contended that to exist is to perceive or to be perceived – to which he later on added “to act.” Therefore he deemed matter, with its alleged primary qualities, to be a figment of the imagination. In fact, Berkeley (1901: 260–1) claimed that “all the choir of heaven and furniture of the earth, in a word all those bodies which compose the mighty frame of the world, have not any subsistence without a mind; that their being is to be perceived or known; that consequently so long as they are not actually perceived by me, or do not exist in my mind, or that of any other created spirit, they must either have no existence at all, or else subsist in the mind of some Eternal Spirit.”
44
Chasing Reality
Without this assumption, that God can be counted on to keep the world going while the subject is absent or asleep, Berkeley’s philosophy would not have been intellectually respectable. That is, a secular version of Berkeley’s philosophy would have been even more outrageous. This is the case of Husserl’s phenomenology, according to which the world of things “is only a presumptive reality,” whereas I myself (that is, Professor Husserl) am an absolute reality (Husserl 1931: 145). Whereas to Berkeley the world is a collection of human and divine percepts, to Husserl it is “an infinite idea, ... a complete synthesis of possible experiences” (Husserl 1960: 62). And, whereas Berkeley’s prose is crystal clear and logically consistent, Husserl’s is neither. For example, Husserl states that the proof of transcendental (non-empiricist) idealism is phenomenology itself (1960: 86). And, although he claims that the real world is subject-dependent, Husserl also demands that one must “go back to the things themselves.” However, what he means by ‘thing’ is a complex of sensations. Furthermore, Husserl holds that, to grasp the essence of a thing, one must start by pretending that it does not exist: one must perform the epocheF or bracketing-out operation. Thus, Husserl closes his Cartesian Meditations with a quote from St Augustine: “Do not wish to go out; go back into yourself.” That is, look inward, not around yourself. However, let us go back to the master subjectivist, whom so many copied without giving him credit. Berkeley’s phenomenalism is ontological as well as epistemological. And he contends that his view is commonsensical because, unlike the views of the schoolmen and the new scientists alike, he invokes only sense data. Foreshadowing Hume and Kant, Berkeley also claims that there are neither objective (subject-free) connections nor laws: “There is nothing necessary or essential [about bodies]; but it depends entirely on the will of the Governing Spirit” (1901: 316). Berkeley’s subjectivism is unique in its cogency, clarity, and literary elegance. This is why it passes for being irrefutable as well as obviously false. However, Berkeley’s philosophy is hard to refute only if its premise, that knowing is limited to perceiving, is admitted, and consequently the power of scientific theory is doubted. Indeed, the success of Newtonian mechanics provides the desired refutation, for, if astronomers can use this theory to predict that Venus will appear to us first as the Morning Star, and towards the end of the same day somewhere else as the Evening Star, then we must admit that they know something about matter that goes far beyond the primitive knowledge of secondary qualities delivered by the senses. That is, conception can overcome the limits of perception: this is why theories, not data, crown modern science.
Phenomena, Phenomenalism, and Science
45
However, an even stronger argument against the identity of ideas and reality comes not from the successes of science but from its failures (Bunge 1954). Indeed, subjective idealism is falsified every time a theory fails to account for the behaviour of something. Only scientific realism makes room for factual error, that is, discrepancy between idea and fact. Such discrepancy cannot be explained by subjectivism, intuitionism, or empiricism. And yet the discovery of error is a major stimulus for engaging in scientific research: a completely true theory induces complacency and thus stagnation. This is why Rita LeviMontalcini, the great neuroscientist, titled her autobiography In Praise of Imperfection. What might have led Berkeley to his outlandish theory? Certainly not empiricism per se, because an empiricist may allow for the autonomous existence of the external world, as was the case with Sextus, Bacon, Locke, and Hume. The source of Berkeley’s theory may be the confusion of existence with existence criteria or tests – a conflation of ontology with methodology. Thus, how do I know that there is a rose bush in that garden? Because I can touch, see, and smell that plant. It would seem then that, indeed, to be is to be perceived – until the botanist examines the plant under the microscope and subjects it to tests showing that it possesses many more primary than secondary properties, such as the abilities to absorb light, synthesize sugar, and grow by cell division. Berkeley’s conflation of existence with existence tests is the source of three influential twentieth-century philosophies: operationism, the verification theory of meaning, and constructivism. Operationism (Bridgman 1927), a version of logical positivism, can be summed up in two formulas: (a) the ontological principle to be is to be measured; and (b) the semantic principle that the scientific concepts are “defined” by (or get their meaning from) laboratory operations. This was the philosophy of science dominant in the first half of the twentieth century. Amazingly, it was also the philosophy behind the so-called Copenhagen interpretation of quantum mechanics, according to which atoms and the like do not exist as long as no one “observes” them (see, e.g., Bohr 1958, Heisenberg 1958, and Bunge 1959b). The falsity of principle (a) becomes clear upon reflecting on the following counterexamples. There are many different ways of measuring times, yet there is a single (classical) concept of time. As for principle (b), not all concepts are definable in a given theory: some of them, the definers, are basic or primitive. Besides, defining is a conceptual operation, not an empirical one. For example, a speedometer measures the value of speed, which in classical physics is defined as distance over time. Incidentally, the very design of such an instrument involves fragments of physical theory, among them the said definition.
46
Chasing Reality
Conjecture about U Evidence
Reference Unobserved U
Indicator hypothesis
Observed O
Figure 2.3 A conjecture about an unobservable trait U is tested by means of an indicator hypothesis of the form U = f (O), where O is an observable feature.
The verification theory of meaning (see, e.g., Ayer 1959) states that the meaning of a non-logical sentence consists in the operations whereby it is verified. Accordingly, test would precede meaning – which of course is false, since we must understand (the meaning of) a statement before we can even imagine how to put it to the test. Just think of endeavouring to test, say, the basic law of electrostatics, namely, “Ñ2j = 4pr,” before finding out what these symbols mean, and before figuring out observable indicators, or “operational definitions,” of the potential gradient Ñj and the electric charge density r. The correct sequence is Meaning-Test, not the converse (see Bunge 1974b). And, as mentioned above, empirical tests involve indicator hypotheses, that is, bridges between unobservables, such as neural disorders, and observables, such as behavioural disorders. See figure 2.3. Finally, constructivism comes in four varieties: (a) ontological, or Berkeleyan – that is, things are bundles of perceptions; (b) social – that is, all scientific facts are social constructions rather than occurrences in the external world; (c) psychological, or Piagetian – that is, as they grow, children construct by themselves the concepts of object, time, number conservation, and so on; and (d) pedagogical – that is, the student must be allowed to learn by himself with minimal guidance. We dealt with ontological constructivism before, and concluded that it is the double delusion that the subject is the centre of the universe and that there can be observations without things observed – or phenomena without noumena. This is classical or individualist constructivism. Its contemporary successor is social constructivism (e.g., Berger and Luckman 1966, Latour and Woolgar 1986, Knorr Cetina 1981, Collins 1998). This is the opinion that (a) the scientific communities, not individual investigators, discover and invent; and (b) those groups construct not only ideas and experiments, but also the objects they study. Thus Woolgar (1988: 65): “[T]he representation gives rise to the object.” Representation of what? Of the object, of course! For want of logical rigour, the sociology of knowledge, a once-promising discipline (Merton 1973), was turned into an extravagance.
Phenomena, Phenomenalism, and Science
47
This is not to deny that scientists are embedded in social networks of various kinds, so that every one of them is indebted to several others. Nor is it to deny that scientific ideas are active and sophisticated constructions rather than automatic responses to environmental stimuli. But it is also hard to deny that only individuals can think, and that the natural and social scientists study only external items, such as earthquakes and the emotions felt by fellow humans. Among the many facts worth studying are the serious social problems of science, such as the shortage of funding, the declining enrolments in science programs, the distortion of problems and results by economic and political interests, and the popular appeal of the pseudosciences and the antiscientific philosophies. But the social constructivists overlook these problems: they are more interested in criticizing science than in defending it. Let us finally glimpse at the two remaining varieties of constructivism: psychological and pedagogical. Psychological constructivism is an important scientific theory. Its philosophical interest lies in the fact that it opposes the empiricist view that all constructs are distillates of experiences. True, when Piaget wrote about “the child’s construction of reality,” he sounded as if he had adopted Berkeley’s subjectivism. But his research shows that he was a realist. Pedagogical constructivism will be dealt with in chapter 3, section 9. Suffice it here to state that it is just as extravagant as Berkeley’s – only even more destructive, because it implies that teachers are expendable. Let us now go back to the sources of all of the above philosophical extravagances in the eighteenth century. 7 The Counter-Revolution, Phase 2: Hume Like Locke and Berkeley before him, David Hume was a radical empiricist. But, contrary to Berkeley, Hume owned that “the operations of nature are independent of our thought and reasoning” (1888 [1734]: 168). However, he denied the difference between primary and secondary qualities. Indeed, he stated that “colours, sounds, heat and cold, as far as appear to the senses, exist after the same manner with motion and solidity”; and that “the difference betwixt them is founded neither on perception nor on reason, but on the imagination” (ibid.). Consequently, Hume thought that we have no satisfactory idea of matter (ibid.: 229). Thus his phenomenalism, like Protagoras’s, was epistemological, not ontological: matter may well exist, but we cannot know it because the sense organs are the only organs of cognition – and all knowledge comes from perception. Although Hume writes about laws of nature, he regards them all as inductions from observations and, as such, as superficial and contingent: “[N]ature
48
Chasing Reality
has kept us at a great distance from all her secrets, and has afforded us only the knowledge of a few superficial qualities of objects” (Hume 1902 [1748]: sec. IV, pt. II). He goes as far as explicitly questioning Newtonian mechanics because it goes beyond sense data: “Sight or feeling conveys an idea of the actual motion of bodies; but as to that wonderful force or power, which would carry on a moving body for ever in a continued change of place, and which bodies never lose but by communicating it to others; of this we cannot form the most distant conception” (ibid.). If the laws of nature are not necessary, then anything can happen. This is what Hume asserts in the most famous passage of his Enquiry: “The contrary of every matter of fact is still possible; because it can never imply a contradiction, and is conceived by the mind with the same facility and distinctness as if ever so conformable to reality. That the sun will not rise tomorrow is no less intelligible a proposition, and implies no more contradiction, [than] that it will rise. We should in vain, therefore, attempt to demonstrate its falsehood” (Hume 1902: sec. IV, pt. I). Not surprisingly, David Lewis (1986), a champion of many-worlds metaphysics, was also a Humean. (More on this in chapter 8, section 9, and chapter 9, sections 4 and 5.) What a far cry from the thesis of the founders of modern science, all of whom swore by lawfulness, and sought constant laws and imperceptible mechanisms behind fleeting appearances! Remember, for instance, how Descartes (1664) conceived of creation: God created matter, endowed it with the laws of motion, and withdrew henceforth from it. There can be no miracles in his world or in Galileo’s, while anything can happen in Hume’s. Whereas Galileo, Descartes, and Newton conceived of the universe as the self-winding clock, Hume saw it as a patchwork of sense impressions – which is also, presumably, the way primitive humans saw it. Scientists, unlike fiction writers and many-worlds metaphysicians, distinguish clearly between real and conceptual possibility. A fact is really possible just in case its occurrence is compatible with the laws of nature (whether causal, probabilistic, or mixed). Otherwise it is really impossible except in fiction. Hence, scientists abstain from populating the universe with utterly fictitious entities. For, as Saki once wrote, “when once you have taken the Impossible into your calculations its possibilities become practically limitless.” Hume’s worldview allows for miracles: in this regard, the justly celebrated sceptic is nearly as gullible as the pious religionists he was criticizing. The peculiarity of Hume’s miracles is that they are secular, not religious. Presumably, he would have tolerated the belief that flying pigs might eventually be sighted, since they would be visible. Likewise, Hume might have encouraged
Phenomena, Phenomenalism, and Science
49
research on the spontaneous unscrambling of eggs and on telepathy, since he would have doubted that any laws of nature could “prohibit” any such processes. It is tempting to speculate that he would have been taken in by many of the pseudoscientific projects that went on in Jonathan Swift’s Academy of Lagado. True, Hume famously characterized miracles as violating laws of nature. However, he adopted the primitive conception of laws, namely, as regular successions of phenomena, such as the days-nights sequence, with no hint of mechanism, and therefore devoid of explanatory power. This is exactly how primitive and ancient man conceived of natural regularities: “Changes can be explained [by primitive and ancient man] very simply as two different states, one of which is said to come forth from the other without any insistence on an intelligible process – in other words, as a transformation, a metamorphosis” (Frankfort et al. 1949: 27). Thus, “[t]he phenomenalist program of accounting for invariable succession in a purely descriptive way, without inquiring into the ‘mechanism’ of change, is in fact characteristic of a poorly developed culture rather than peculiar to the ‘positive’ stage of mankind” (Bunge 1959a: 73). The dynamical laws underlying the fluxus formae were beyond Hume’s reach, both because they refer to facts behind appearances, and because their understanding requires some mathematics, which was above Hume’s head. The Newton-Euler equations of motion, which Hume could neither read nor accept, are a case in point. They imply, in particular, that the spin of a frictionless spinning top, such as our planet, is a constant of the motion. In turn, this invariance implies that the regular succession of days and nights is necessary, not contingent – provided of course that no large meteorite strikes our planet. Consequently, that the Sun will rise tomorrow is not just a shaky forecast on the strength of a mere inductive generalization from a finite number of observations: it is a feature of a necessary process “ruled” by laws. Hume excluded the supernatural only because it is inaccessible to the senses: “The religious hypothesis, therefore, must be considered only as a particular method of accounting for the visible phenomena of the universe: but no just reasoner will ever presume to infer from it any single fact, and alter or add to the phenomena in any single particular” (1902: sec. XI). Religion is then at best useless to understand experience. However, phenomenalism does not protect us from lay superstition. Moreover, it places forces and atoms on the same level as gods and ghosts. Thus, ironically, phenomenalism kills both science and religion with a single stone. And for this reason, religious scepticism, Hume is regarded as a member of the Enlightenment despite his scientific scepticism. The same holds for Kant, to whom we turn next.
50
Chasing Reality
8 The Counter-Revolution, Phase 3: Kant It is well known that Kant started out as a realist: so much so that he wondered at the evolution of the “nebulae” (our galaxies) – an original problem at the time. In fact, in 1755 he published his General History of Nature and Theory of the Heavens, which contained the valuable Kant-Laplace hypothesis, as well as the false conjecture that the solar system is stable because the gravitational attraction is balanced by a repulsive force that had no place in standard astronomy. (Kant, like Hume, could not read Newton because he had no higher mathematics.) However, his ambition to carve for himself an academic niche led Kant to imbibing Christian Wolff’s speculative system. This was Kant’s undoing, not because Wolff was a systematic philosopher, as has been said, but because he was a second-rate follower of the great Leibniz. In jettisoning Leibniz’s metaphysical baggage, in particular his doctrine of monads, Kant is likely to have overlooked the great man’s Nouveaux essais, published in 1756 but written in 1704, and which in my opinion constituted a definitive refutation of Locke’s empiricist epistemology. At the same time, Kant ceased to take an interest in natural history, and read only other philosophers – a custom that has been continued up to our days. Kant tells us that the reading of Hume awoke him from what he called his “metaphysical slumber.” What is less well known is that his phenomenalism was even more radical than Hume’s. In places it was, indeed, ontological as well as epistemological, and therefore closer to Berkeley’s. (Recall the difference between the two varieties of phenomenalism: section 3.) In fact, Kant (1787: B724) asserted that “die Welt ist eine Summe von Erscheinungen” – that is, “the world is a sum of appearances.” Allow me to repeat: According to Kant, the world is made of appearances, that is, facts as perceived by some subject, not facts in themselves. Consequently, the existence of the world would depend on that of sentient beings. Shorter: No sentient beings, no universe. As Norbert Elias (2000: 475) put it, the Kantian subject of knowledge, shut up in his aprioristic shell, can never break through to the thing in itself: his is the homo clausus. This fiction, central to individualism, has been pervasive in modern epistemology since Descartes, and in the social studies since about 1870, when it was used and popularized by the neoclassical micro-economists. It also occurs in the neo-Kantian philosophy of the influential sociologist Max Weber – though not in his scientific work, which is rigorously realist. However, Kant is hard to nail down, for he sometimes vacillates and backtracks. Indeed, on the same page B724 of his great work he admitted that there
Phenomena, Phenomenalism, and Science
51
are things in themselves, different from our experiences, and moreover that these are the “foundations and causes” of phenomena. As Torretti (1967: 490) puts it, the effect is sometimes pathetic, and at other times nearly comic. In any event, when Kant admits the existence of things outside us, he does so reluctantly, in thinking that such admission must be taken on faith rather than on the strength of either experience or reason. Indeed, in the preface to the second edition to his first Critique, Kant states that it is “a scandal of philosophy, and of the general human understanding, that we must admit merely on faith [bloss auf Glauben] the existence of things outside ourselves” (1902: B xxxix). But he is inconsistent, for elsewhere in the same work he tells us that space and time are in the mind; and, since he also asserts that all things are in space and time, it follows that all things are in the mind. 9 Kant Concluded: Neither Nature nor God A realist takes nature for granted, and he may ask how knowledge of nature is possible. By contrast, Kant, the subjectivist, asks how nature itself is possible, and answers in his Prolegomena (2002: 110): “by means of the constitution of our sensibility.” Consequently the laws of nature are just the laws of the connections of appearances (ibid.: 111). And “the understanding does not draw its (a priori) laws from nature, but prescribes them to it ” (ibid.: 112, Kant’s emphasis). If in doubt concerning the autonomous existence of nonsensuous things, consult Natorp (1912: 94), one of the most prominent neoKantians. He assures us that for Kant the thing in itself is just a “possible experience.” Hence, no percipient, no thing – just what Berkeley had claimed. All this is not just a funny academic eccentricity; it is scandalous, as Kant himself declared, because the independent reality of the world is taken for granted by anyone, kitten or scientist, who endeavours to explore the world around himself, whether out of sheer curiosity or just to stay alive. The uncurious animal does not learn much, and has no great chances of noticing predators or potential mates, or of discovering food or shelter. In other words, whether or not we can prove the existence of the external world, we must recognize that such existence is no less than a precondition of survival and a presupposition of inquiry. However, let us return to the phenomenalist argument. Not being perceptible, space, time, and causation are further casualties of the phenomenalist reduction. In fact, Kant renders them subjective, yet necessary for experience and therefore prior to it. Indeed, according to him, we cannot experience sounds, smells, colours, or textures except against the background of those a priori categories. As well, Kant’s scheme makes no
52
Chasing Reality
room for the laws of physics and chemistry, since these interrelate only primary properties of material entities. (True, until some decades ago chemists still listed the “organoleptic” or secondary properties of chemicals, but only as practical indicators useful to identify them quickly in the lab.) Kant thus completed the radical counter-revolution in philosophy started by Berkeley and continued by Hume – except that, paradoxically, he called it a ‘Copernican revolution.’ Note that Kant built his subject-centred philosophy one century after Newton’s great work, which crowned the view of the world as the maximal clock. That was also over one and a half centuries after Galileo and Descartes had argued that the world is composed exclusively of material things, and that the secondary qualities are in us, not in the external world. Little could Kant suspect that his subjectivism would be reinvented two centuries later in combination with sociologism, and under the name of ‘social constructivism,’ by scholars who are unlikely to ever have read him (recall section 6). Nor is modern science the only victim of Kant’s phenomenalist onslaught. Ironically, a further casualty is religion, whether theistic or deistic. Indeed, Kant asserts that “the concept of a higher intelligence is a mere idea [eine blosse Idee]” (B698). And again: The idea of a “highest being and uppermost cause” is only “a mere something in the idea, of which we have no conception of what it is in itself ” (B707). Now, anyone who, like Kant, holds that God is an idea rather than a real being, is normally deemed to be an atheist. But Kant wavered between atheism and agnosticism, just as he wavered with regard to things in themselves. Indeed, earlier in the same book (B481ff.) he had stated that we cannot decide between the thesis that the world contains a necessary Being, that is, God, and its antithesis. True, a few years later, in his Critique of Practical Reason (1788), Kant held that belief in the existence of God is a postulate of practical reason – perhaps a way of saying that, even if God does not exist, it is prudent to pretend that He does. So, God entered Kant’s philosophy through the service door, and never made it beyond the kitchen. Furthermore, Kant drew a firm line between knowledge and belief, in particular religious belief. The knowing subject, not God, was the centre of Kant’s world. Fichte took this blasphemy farther: he stated even more clearly that the self is the fountain of reality. And he added that the advantage of this view is that it guarantees the freedom of the will – provided, presumably, it is strong enough to will the creditor and the hangman away. Kant’s religious scepticism did not go unnoticed in his own day. In fact his sovereign, Friedrich Wilhelm II, King of Prussia, wrote him on 1 October 1794
Phenomena, Phenomenalism, and Science
53
that the publication of such ideas constituted an abdication of his responsibility as a teacher (Kant 1913, 3: 48–9). Consequently, the king demanded from his subject that he abstain from any further utterances of the kind. And he threatened that, were Kant to persist, he would be subjected to “disagreeable procedures.” On 12 October of the same year the professor, being a civil servant, humbly promised his king to abstain from any further utterances about religion. Kant kept his word during that king’s life. But he did not fool his contemporaries: some of them regarded him as the earliest German atheist. Only some historians of philosophy have had the impudence of holding that the adult Kant kept the fideism of his parents. 10 Concluding Remarks Phenomenalism restricts reality to a tiny portion of it, namely, the collection of appearances or pre-analytic experiences. Hence, it excludes most of the universe, either from existence (ontological phenomenalism) or from knowledge (epistemological phenomenalism). For instance, a phenomenalist will say that grass is green, whereas a realist will say that it looks green to us under white light. Besides, the realist can explain why this is so: namely, because grass reflects the light of all wavelengths except for those that cause the sensation of green in the primary visual system of a normal person. The realist can also ask certain interesting questions that the phenomenalist cannot, such as: At what stage in evolution did perception (as different from mere sensation or detection) emerge? And how is it possible for physics and chemistry to study a single noumenal (material) universe, so utterly different from the phenomenal? Furthermore, whereas the realist posits a single noumenal universe, the phenomenalist must put up with as many phenomenal worlds as sentient organisms–unless he takes the solipsistic plunge. For example, he will say that what you see as a rabbit he sees as a duck, and such ambiguity cannot be resolved because nature contains neither animal: both are only in the mind. But of course the hunter, the farmer, the butcher, the cook, and the zoologist are bound to laugh at the phenomenalist’s predicament: they know that there are real rabbits and ducks out there. An epistemological correlate of phenomenalism is descriptivism. This is the view that description is to be preferred to explanation, because (a) the latter goes necessarily beyond phenomena; and (b) there can be no law statements, because there are no objective connections and pattens – only constant conjunctions and regular sequences. According to this view, the generalizations
54
Chasing Reality
that night follows day, that drinking water quenches thirst, or that pricking hurts do not require explanation: they are well-confirmed inductions. By contrast, the realist takes such empirical generalizations as givens to be explained by higher-level laws involving transphenomenal concepts, such as those of field, gene, and nation. Since scientific research at its best is the search for objective laws, phenomenalism is inconsistent with science. Luckily, some phenomenalists failed to notice such contradiction, and produced good science. Fortunately, natural science had already attained maturity and prestige in mid-eighteenth century, so that it was not crippled by either Kant’s subjective idealism or Hegel’s objective idealism. But the social studies were then still at an embryonic state, and they were seriously distorted by the idealist philosophies of Kant and Hegel, in particular Dilthey’s. This can be seen, for instance, in the thesis, common to both methodological individualism and hermeneuticism, that the task of the social scientist is not to study social systems, such as families, schools, businesses, and polities, but to try and “understand” or “interpret” the values and intentions of unembedded and rational individual actors. (For the stealthy convergence of hermeneutics and rational-choice theory see Bunge 2003a.) A consequence of the idealist and individualist approach is of course that the large social transformations, such as industrialization, militarization, colonization, secularization, feminism, trade-unionism, the concentration of capital, and globalization, are lost sight of. It is thus no coincidence that Max Weber, who passes for being the founder of modern sociology, missed all the great social transformations of his own time, such as the rise of imperialism and democracy. He also missed the rise of science and technology, both of which happen to be the cultural engines of modern society. Instead, he was obsessed by religion, charismatic leaders, and the legitimacy of political power. By contrast, historical materialism – once purged of dialectics, economism, and Laplacean determinism – has exerted a decisive and healthy influence on historiography (see Barraclough 1979), anthropology (see Harris 1979), and archaeology (see Trigger 2003b). There are several reasons for the superiority of the realist-materialist over the subjectivist-idealist view of society. One is that people have to work in order to eat, and they have to eat before they can think. Another reason is that, as Durkheim (1901: 77) urged, “the social facts must be treated like things” – that is, objectively. A third reason is more subtle: “[O]nly one thing is absolutely sure in history: all that is particular is more or less doubtful” (Tocqueville 1985: 351). For instance, we all know that the Great Depression occurred, but no one knows for sure what started it, much less what fateful decisions its main characters made.
Phenomena, Phenomenalism, and Science
55
In sum, antirealism is not just false; it also hampers the study of reality. However, this has not prevented its diffusion, because the worth of a philosophy is seldom measured by its contribution to the advancement of knowledge. Indeed, antirealism is, if anything, more popular nowadays than three centuries ago. However, this topic deserves a new chapter.
56
Chasing Reality
3 Antirealism Today: Positivism, Phenomenology, Constructivism
The standard histories of philosophy suggest that Kant’s line died out with the now forgotten neo-Kantian philosophers, such as Lange, Vaihinger, Natorp, Cohen, Windelband, Rickert (Weber’s philosophical mentor), Cassirer, and Bauch – Carnap’s thesis supervisor. I submit that Kant’s subjectivism was also a major source of several other schools, some of which have been far more influential than orthodox neo-Kantianism. Those schools are the radical subjectivisms of Fichte and Schopenhauer; the classical positivist doctrines of Comte and Mill; Nietzsche’s combination of pragmatism and relativism; pragmatism (or instrumentalism) from William James (but not Charles S. Peirce) and John Dewey onwards; Vaihinger’s fictionism; Husserl’s phenomenology and Schutz’s phenomenological sociology; logical positivism from Mach to the Vienna Circle and beyond; and the constructivism-relativism that Kuhn and Feyerabend revived and made fashionable in the 1960s. Furthermore, Kant, along with Hegel, exerted a strong influence on the hermeneutic movement, which includes Dilthey, Husserl, Heidegger, Gadamer, Habermas, Ricoeur, and Derrida among others (see Mueller-Vollmer, ed. 1989). This movement exploited Kant’s dichotomy between the natural and the cultural. In particular, Dilthey wanted the social studies to shift focus from objective social facts to the actors in them – from trade to traders, from war to warriors, and from the polity to politicians. This is why the hermeneuticists hold that the cultural (social) world is to be understood very differently from the natural one, namely, through the Verstehen (understanding, interpretation) of individual actions. This peculiar operation would consist in grasping the “meaning” (intention, goal) of behaviour and texts. Hence, the study of social life would be a Geisteswissenschaft, that is, a science of the spirit (or cultural science). Whence, too, the indifference of hermeneuticists to everything lying outside the Word-
Antirealism Today: Positivism, Phenomenology, Constructivism
57
Kant
Classical positivism (Comte, Mill, Mach)
German Romanticism (Hegel, Fichte, Schelling)
Logical positivism Neo-Kantianism Hermeneutics (Schlick, Carnap, Reichenbach) (Cohen, Natorp, Cassirer) (Dilthey, Husserl, Heidegger)
Figure 3.1
Meaning-Interpretation triangle. This is how the environment, manual work, scarcity, poverty, oppression, segregation, and war fall beyond the purview of the hermeneuticist. Only das Höhere , the higher, is worthy of the attention of the Herr Professor. Leave the daily life miseries to the lesser beings. (More in Bunge 1996.) In sum, Kant’s philosophy spawned three main lines: neo-Kantianism, classical positivism – which eventually gave rise to logical positivism – and German Romantic philosophy, which in due course engendered philosophical hermeneutics: see figure 3.1. All of the above-mentioned philosophies rejected the hylorealist theses that the external world is material and exists by itself, and that science can grasp reality. However, of all the philosophies stemming from Kant’s, positivism was the only one that made no concessions to irrationalism, and that proclaimed its love of science. Moreover, some of the positivists, such as Mill, Mach, Pearson, Duhem, Bridgman, and Bohr, were outstanding scientists who left far deeper marks on science and its philosophy, and even on the ruling worldview, than did the self-styled neo-Kantian philosophers, who, except for Ernst Cassirer, never reached the public. Most of this chapter will be devoted to reanalysing logical positivism, also known as logical empiricism, neo-positivism, or just positivism. A scientific realist like this writer can engage in a fruitful dialogue with a positivist because both share two important principles: those of rationality and scientism. The former principle states that all ideas are rationally debatable as long as they are fairly clear; and the second holds that the best way to study facts, whether natural or social, is by using the scientific method. A good reason for re-examining positivism is that it is still the explicit philosophy of most experimental scientists. Indeed, when in a didactic mood, they claim that all research is data-driven. They forget of course that they do not pile up data mindlessly, but use always more or less explicit hypotheses.
58
Chasing Reality
Ordinarily, data are sought or produced in the light of some hypotheses, and these are in turn checked by pertinent evidence. So, neither data nor hypotheses have the last word. Another reason for re-examining positivism is that it is often confused with realism or even with materialism. In fact, roughly between the 1870s and the 1930s, what passed for positivism was actually a jumble of positivism proper, scientism, evolutionism (Spencer’s or Haeckel’s rather than Darwin’s), materialism, naturalism, and energetism. Anyone who admired science and rejected organized religion called himself a positivist. All this was true of México, Río de Janeiro, and Buenos Aires, as well as of London, Paris, and Berlin (see Biagini, ed. 1985). All this changed in 1927 with the foundation of the Vienna Circle, which reclaimed the heritage of Hume and Mach. (Kant was unmentionable because of his blunders over mathematics and science. But his phenomenalism was adopted wholesale.) The Circle’s members, the logical positivists, claimed to have overcome the realism/antirealism and the materialism/idealism dilemmas. In particular, they held that the question of the autonomous existence of the external world was a pseudo-problem (Schlick 1959). Little did they suspect that such epistemological agnosticism would be resurrected by the postmoderns towards the end of the century. Thus, the self-styled postmodern psychologist Gergen (2001: 806) proclaims that “arguments about what is really real are futile.” However, in practice most logical positivists sided with idealism, since they conceived of physical things as either possibilities of sensation or logical constructions out of percepts (e.g., Carnap 1967). Later on, under Wittgenstein’s influence, they adopted glossocentrism, or linguistic idealism. Thus, Carnap (1950b: 23) held that “to accept the thing world [objective reality] means nothing more than to accept a certain form of language.” That is, to be is to be talked about, or to play a “language game.” Ironically, Heidegger (1987: 11), whom Carnap rightly despised, held a similar view: that the word is the abode of being. Moral: Falsity plus logic (as in Carnap), or minus logic (as in Heidegger) equals falsity. The best-known features of logical positivism are its empiricist semantics and epistemology, the advocacy of logical analysis, an unrequited love of science, and an allegedly antimetaphysical stand. Besides, the logical positivists, like Kant, attempted to combine empiricism with rationalism. However, it is arguable that Kant combined the wrong half of empiricism, namely phenomenalism, with the wrong half of rationalism, namely apriorism. By contrast, logical empiricism retained phenomenalism but joined it with the best component of rationalism, namely modern logic. Thus, whereas Kant’s writings were
Antirealism Today: Positivism, Phenomenology, Constructivism
59
plagued by obscurities and inconsistencies – as we saw in the previous chapter – those of the logical empiricists constituted the clearest and most consistent philosophical works of their time, particularly in comparison with those of their main rivals – the intuitionists, phenomenologists, existentialists, Thomists, neo-Hegelians, and dialectical materialists. This is why one can learn far more from debating with them than from criticizing their rivals. Because of the enduring influence of the logical positivists, engaging them in debate is not a mere academic exercise. Let a couple of examples suffice. The first will be the fiasco of the American “intelligence” agencies at the dawn of this century, who spent many millions to no avail because they did not know what to do with their huge pile of data: they framed no hypotheses based on serious social studies, and consequently they did not test any either. As Charles Darwin stated, to discover anything interesting and plausible one must make some conjectures and put them to the test. But to make reasonable conjectures concerning political facts, such as the production of weapons of mass destruction or the planning of terrorist attacks, one must use the ABC of social psychology and political science in addition to gathering “intelligence” – which is mostly unconfirmed gossip anyway. Consequently, effective antiterrorist work requires the cooperation of spies with scientists on a daily basis. Our second example is the quantum theory. It is often forgotten that empiricism lurked behind the standard or Copenhagen interpretation of the mathematical formalism of quantum mechanics, particularly from 1935 on. Consequently, a realist reinterpretation of this theory must be preceded by an effective critical analysis of logical positivism. And such analysis must not ignore the phenomenalist metaphysics of logical empiricism, which is usually overlooked although it is no less than the root of its epistemology. After all, epistemology may be regarded as an application of ontology to the cognitive process: Tell me what there is, and I’ll tell you what and how we can get to know it. If the world is either an idea or a mass of appearances, just look inward; but if it is concrete, go out and explore it; and if all we can do is perceive, just record your sense impressions; but if you can also think, be prepared to think very hard. 1 Logical Positivism The original name of the Vienna Circle (1929–36) was the Ernst Mach Verein (see, e.g., Feigl 1943, Kraft 1953, and Ayer, ed. 1959). This learned society was constituted by philosophers such as Moritz Schlick and Rudolf Carnap; scientists such as Philipp Frank and Otto Neurath; and mathematicians such as Karl Menger and Richard von Mises. Its declared purpose was to update and
60
Chasing Reality
publicize the positivist philosophy of Ernst Mach, who had regarded himself as an heir to Hume and Kant with regard to phenomenalism. Mach had been a distinguished experimental physicist and physiological psychologist, as well as a popular historian and philosopher of science. He was proud of having had invented the “economy of thought,” the quaint principle that the aim of science is not to understand reality but to spare us experiences by their reproduction, colligation, and anticipation in thought (Mach 1910: 238, 1942 [1893]: 578). Mach did not care to have his name associated with the classical positivists Auguste Comte and John Stuart Mill. But in fact he took over their phenomenalism and descriptivism, as well as their rejection of metaphysics and religion. In particular Mill, following Kant, had defined a thing as “a possibility of sensations” – an idea that agrees with the popular metaphysics according to which factual existence is the same as its ordinary-knowledge criterion, namely, the possibility of seeing, touching, smelling, and the like. Mach (1914) adopted Mill’s definition of a thing, which he regarded as “a stable complex of elements,” where ‘element’ was a synonym for ‘sensation’ and his substitute for the physical atom. He warned that ‘thing’ is an abstraction, “the name of a symbol, for a compound of elements [sensations] from whose changes we abstract”; and that the concept of a thing-in-itself is an absurdity, for “[s]ensations are not the signs of things; but, on the contrary, a thing is a thought-symbol for a compound sensation of relative fixedness” (ibid.: 580). He also adopted Kant’s view that “[n]ature is composed of sensations as its elements” (Mach 1942 [1893]: 579). (In a famous book, Lenin [1908] rightly criticized Mach and others for having revived Berkeley’s subjective idealism – which, regrettably, he did not attempt to refute. Nearly half a century later Popper [1953] confirmed Lenin’s diagnosis: Mach was a Berkeleyan. True, the head of the Vienna Circle [Schlick 1959: 85] rejected this characterization, and claimed that Mach and his heirs took no sides in the realism/subjectivism controversy. But he did not quote chapter and verse to make his case.) Mach did not explain how his phenomenalism squared with his own pioneering study of bullets moving at supersonic speeds – which he was the first to photograph. Surely neither the bullets nor the shock waves they generated could have been sensations, not only because they did not occur in anyone’s brain, but also because they could not be seen directly. To be sure, the moving bullets could be touched, but the ensuing sensation would have been a datum for the surgeon, not the physicist. Nearly all the members and affiliates of the Vienna Circle, particularly Rudolf Carnap, Hans Reichenbach, and Philipp Frank, adopted Mach’s phenomenalism. This earned them an unexpected if unwelcome and silent ally: the
Antirealism Today: Positivism, Phenomenology, Constructivism
61
Inquisition. Indeed, Reichenbach (1951: 107) held that “the Copernican system differs from Ptolemy’s only in that it is another mode of speech.” Thus, the heliocentric and the geocentric models of the solar system are “equivalent descriptions,” since at the time both accounted equally well for the empirical evidence. And Frank (1957: 352) held that the Copernican theory was mathematically simpler and had the greater heuristic value. Truth was not in question. Cardinal Bellarmino, Galileo’s accuser, must have chuckled in Hades. Two reasons have been adduced in favour of the said equivalence. The first is that all the coordinate systems are equivalent; in particular, a rectangular coordinate system centred in the Sun is equivalent to a spherical coordinate system centred in the Earth. But such geometric equivalence is irrelevant, because all motions are relative to reference frames, not coordinate systems. Reference frames are physical things, and coordinate systems mirror only some geometric features of reference frames (Bunge 1967b). This point is important because (a) the Galileo and Lorentz transformations relate reference frames, not coordinate systems: the former can move, since they are physical entities, whereas the latter cannot, because they are mathematical objects; and (b) the Earth, being accelerated, is not an inertial system: that is, the equations of mechanics (whether classical or relativistic) are not satisfied exactly relative to our planet. The second reason given for the alleged equivalence was that, since all motion is relative to some reference frame, it makes no difference whether this frame is the Sun (heliocentric system) or the Earth (geocentric system). But it does. In fact, the less massive of two nearby bodies is forced to orbit around the far more massive one because it is the source of the more intense gravitational field. (Further reasons in Bunge 1961: 139-40.) In short, the planetary orbits are really, objectively elliptical or nearly so. The metaphysical phenomenalism advocated by Mach was not systematic. The first attempt to build a phenomenalist theory was Whitehead’s (1919); it was so unsuccessful that its author soon turned to his process metaphysics, which was closer to Hegel’s than to Mach’s. The next effort in the phenomenalist project was Carnap’s Logical Construction of the World (1967 [1928]). The basic tenet of this system was that in principle “all physical objects can be reduced to psychological ones” and conversely (ibid.: 92). Amazingly, this principle was a generalization from a single example – a wrong one. Consider: “Statements about physical objects can be transformed into statements about perceptions (i.e., about psychological objects). For example, the statement that a certain body is red is transformed into a very complicated statement which says roughly that, under certain circumstances, a certain sensation of the visual
62
Chasing Reality
sense (“red”) occurs” (ibid.). Carnap did not realize that he was just expanding a phenomenalist statement about seeing red. If psychological and physical objects and statements were indeed equivalent or mutually reducible, why prefer the Physicalist ® Phenomenalist translation to the converse one? Presumably, just attachment to the empiricist tradition. But one of the reasons advanced by the founders of modern science for the converse preference was that primary properties are subject-independent, and thus lend themselves to impersonal universal generalization (recall chapter 1). In any case, scientific theories do not contain concepts denoting qualia unless they happen to concern qualia, in which case they belong in psychology, not in physics, chemistry, or biology. Consider for example the following two elementary mini-theories or theoretical models: the classical model of a linear oscillator, and the standard model of a direct current metallic circuit with resistor R and inductor L connected in series. Classical linear oscillator 2
R-L circuit
2
m d x/dt = – kx
V –Ldi/dt = Ri
⇓ x = a. cos ω t, ω = √(k/m),
⇓ i = V/R + a e –(R/L)t,
where x = position
V = impressed potential difference
m = mass, k = tension
i = current intensity
a = amplitude, ω = frequency
R = resistance, L = self-inductance t = time
Although the things described by the two models – springs and circuits – may be perceptible, the predicates employed to describe them denote primary properties, not qualia. The measurement of m, k, w, V, i, and R requires instruments that embody indicator hypotheses, such as the relation between current intensity and the angle of a pointer in an ammeter. Thus, the premises of the preceding theoretical models do not entail observations – contrary to the popular philosophical accounts of scientific theories, such as van Fraassen’s (1980). We shall return to this subject in chapter 7. Four decades after Carnap’s unsuccessful attempt to eliminate theoretical terms in favour of observational ones, Donald Davidson (1970), one of the last
Antirealism Today: Positivism, Phenomenology, Constructivism
63
positivists, turned Carnap’s alleged proof into a cornerstone of his popular philosophy of mind. To “prove” that mental and physical events are mutually exchangeable, he chose a slightly different example: Two stars in distant space collide with one another at the same time that Jones notices that a pencil starts to roll across his desk. “The collision has now been picked out by a mental description and must be counted as a mental event” (1970: 211). Notice the strategy. Discuss a single case in ordinary-knowledge terms, and leap twice in succession: first to an outrageous generalization (“All physical events can be described in mental terms and vice versa.”), then from common sense to science (physics and psychology). 2 Worldmaking At first sight, positivism and post-Kantian German idealism have little in common: the former is pro-scientific and clear, whereas post-Kantian German idealism is both unscientific and often hermetic to boot. However, since both have the same sources, namely Berkeley, Hume, and Kant, their coincidences should come as no surprise. In fact, both factions coincided in rejecting realism and materialism, although they fought these in different fields. While the positivists specialized in the natural sciences, the hermeneuticists focused on social studies. In particular, while the positivists stressed phenomenalism and criticized atomism, the hermeneuticists emphasized the role of ideas in social life at the expense of macrosocial facts and the so-called material factors, such as natural resources and work. So, each faction blocked the advancement of knowledge in its own way: the positivists by banning microphysics or distorting its philosophical import; and the hermeneuticists by rejecting the macrosocial studies, or by attempting to reduce them to the “interpretation” (Verstehen) of the intentions of individual actors. The point of entry of idealism into the social studies is the ontological thesis that there is no such thing as society: there would be only individuals, and these act freely following their own interests, beliefs, and intentions. This ontological thesis generates methodological individualism. This is the thesis that the goal of the social studies should be to find out how social facts emerge exclusively from individual choices and actions. The individualist project sounds attractive by comparison with the holist thesis that individuals are but pawns of higher instances, such as God, destiny, nation, race, church, or party. But the project is unrealistic and even logically untenable, because social roles cannot be defined except by reference to social systems. For example, a warrior is, by definition, someone who is expected to fight in a war. And in
64
Chasing Reality
turn wars, far from being intimate affairs, consist in armed clashes between groups of individuals characterized by supra-individual traits, such as imperial or colonial, oil-rich or oil-hungry, right-wing or left-wing, and so on. As with social roles, so with social ranks. For instance, the landowning class is not just the collection of persons who possess land cultivated by others; those persons are also backed by a state that enforces that privilege. And a state or government is not reducible to individual items or features; it is a social system characterized by emergent properties, such as having the monopoly on taxation and legal violence. And methodological individualism denies the very existence of social systems, supra-individual entities that have emerged in the course of history to overcome the limitations of individuals. Thus, the individualist-idealist anthropologist, economist, sociologist, political scientist, or historian will attempt to “interpret” (hypothesize) individual actions exclusively in terms of personal beliefs and intentions. (In particular, if he is a rational-choice theorist, he will focus on subjective utilities and subjective probabilities.) Consequently, he will either miss or seriously misunderstand all the macrosocial events and processes that have shaped our societies, such as capitalism, imperialism, nationalism, war, democracy, secularism, socialism, fascism, feminism, technology, and science. (Further criticisms in Bunge 1996, 1998, and 1999.) Max Weber, one of the two fathers of modern sociology, is a case in point. Indeed, he missed all the above-mentioned macrosocial developments, while rightly and forcefully insisting that the social sciences should be objective (see Bunge 2005a). This incongruence was an inevitable consequence of Weber’s adoption of Dilthey’s philosophical hermeneutics, which had been at once idealist and individualist. This approach can at best shed some light on the products of culture, such as texts, while keeping the cultural communities and their societies at large in the dark. In general, most of reality is bound to escape the individualist idealist, and this for several reasons: because reality is a system of systems (see Bunge 1979a); ideas have no existence except in brains; and every thinking person is a component of several social systems that have preceded him. A moral of Weber’s failure to understand the society he lived in is that idealism stunts realism. Realism can flourish only when allied with materialism (recall chapter 1). Phenomenology constitutes an even more drastic retreat from reality. Its founder reproached Galileo and Descartes for distinguishing primary from secondary properties, and for regarding nature as devoid of mental features (Husserl 1970 [1936]). Previously he had stated that “an actual Object belonging to a world or, all the more so, a world itself, is an infinite idea ... an idea related to infinities of harmoniously combinable experiences ..., a complete
Antirealism Today: Positivism, Phenomenology, Constructivism
65
synthesis of possible experiences” (Husserl 1960 [1931]: 62). In short, the world would be subject-dependent, and we would get to know it through ordinary experience rather than through experiment, analysis, and modelling. Characteristically, Husserl offered no evidence for his subjectivist extravagance: he regarded it, like all of his opinions, as “self-evident a priori.” Nor did he mention any of his sources, in particular Berkeley, Kant, Fichte, Mill, Mach, and Dilthey. Originality and concern for evidence are not precisely the trademarks of modern idealism, in particular phenomenology. Instead, obsoleteness, dogmatism, hostility to science, and opaqueness are (see, e.g., Kraft 1957). In the previous section we recalled a parallel attempt to reduce the world to what Mach had called “a complex of sensations,” namely, Carnap’s Logical Construction of the World (1928). This work was generally ignored until Nelson Goodman unearthed and updated it with the help of Lesniewski’s mereology. His Structure of Appearance (1951) is a formally more refined system based on the assumptions that the stream of experience is both structured and constituted by indivisible (atomic) qualia. Still, unlike Carnap, Goodman admitted at that time that the problem of accounting for the physical world upon a phenomenalist basis was still open, and he offered no clue for its solution. Yet, he was unshakeable in his faith that the said reduction would eventually be accomplished (ibid.: 304–6). After trying in vain for three decades, Goodman adopted an easier strategy: that of arbitrary “worldmaking” or subjectivist constructivism (Goodman 1978). Don’t worry over how to discover new extrasolar planets, flying dinosaurs, missing links, or the organ of consciousness, since we are free to imagine them. When told by a fellow philosopher that we cannot make stars the same way as we make bricks, Goodman (1996: 145) became more cautious: “The worldmaking mainly in question here is making not with hands but with minds, or rather with languages or other symbol systems. Yet when I say that worlds are made, I mean it literally.” In other words, when pressed he admits that the “worlds” he claims to be able to make are symbolic systems. Still, Goodman refuses to distinguish them from the only real world there is. Now, if someone claims to know how to make a star, one is entitled to tell him, Show me or shut up! One is also entitled to ask, Why should the taxpayer be made to pay for astronomical observatories, if we can have any desired stars constructed by philosophers? Goodman’s seems to have been the latest, if unsuccessful, attempt to reduce primary properties (or physicalist predicates) to secondary properties (or phenomenalist predicates, or qualia). If his attempt failed even at the ordinaryknowledge level of massive bodies such as chairs, is there any reason to expect
66
Chasing Reality
that it might succeed for quanta or galaxies, cells or nations? Hint: Start by trying to reduce some classical law statements, such as Fourier’s for heat conduction, or Maxwell’s for electromagnetic induction. Warning: Don’t forget to include the translation of such elementary expressions as “sin wt” into sense data. Hint: Don’t waste your time. Besides the individualist constructivism of Berkeley, Kant, Fichte, Schopenhauer, Avenarius, Mach, Husserl, Carnap, Goodman, and a few others, there is collectivist subjectivism, or social constructivism-relativism. This doctrine, fashionable among sociologists of knowledge since the mid-1960s, holds that all facts, whether social or not, are social constructions, that is, products of social groups (see, e.g., Fleck 1979, Latour and Woolgar 1986, H.M. Collins 1981, Barnes 1983, Knorr-Cetina 1983, Knorr-Cetina and Mulkay, eds. 1983, and Fiske and Shweder, eds. 1986). There is of course some truth to social constructivism, namely, the tautology that all human social facts, from greeting and trading to child rearing and warring, are social facts. But nobody starts from scratch, and nobody constructs social reality by himself. To be sure, humans make themselves, and invent, maintain or repair social systems; but, as Marx famously added, they do so starting from a pre-existing social reality and with the help of others. Only philosophers and inmates in a lunatic asylum think that someone can create reality rather than just alter it. Correction: A senior adviser to President George W. Bush told a veteran journalist that guys like him were “in what we call the reality-based community ... That’s not the way the world really works anymore ... We’re an empire now, and when we act, we create our own reality ... We’re history’s actors ... and we, all of we, will be left to just study what we do” (Suskind 2004). The claims of the social constructivists are just as grandiose. For example, according to Fleck, syphilis was constructed by the medical community; Latour and Woolgar claimed that TRF, a certain brain hormone, was “constructed” by its discoverers; Pierre Bourdieu held that “the visible differences between the masculine and feminine sexual organs are a social construction”; the thesis of the antipsychiatry movement is that all mental illnesses are inventions of psychiatrists; and some feminist philosophers have maintained that the scientific laws, and even the concepts of objectivity and truth, are only tools of male domination. Of course, there is no evidence for any of these egregious claims. But there is evidence for the hypothesis that they originated in the confusion between fact and idea (or thing and model), typical of magical thinking. (See criticisms in Bunge 1991 and 1992, Gross and Levitt 1994, Sokal and Bricmont 1998, and Brown 2001.) In short, social constructivism is blatantly false: the overwhelming majority
Antirealism Today: Positivism, Phenomenology, Constructivism
67
of facts are independent of any minds, and ideas occur only in individual brains. This is not to deny that all thinkers, however original, are intellectually indebted to other people. Nor is it to deny that all the social systems and the norms that regulate them are constructed by social groups, albeit not always consciously. The point of realists is that, although all organisms – not just humans – construct their niches, they all employ materials, and abide by laws, that pre-exist them. 3 Phenomenalism and Quanta John Dalton put the atomic theory to work in chemistry one generation after Kant wrote his masterpiece. Most chemists embraced it eagerly because it seemed to explain the chemical compositions and reactions. By contrast, most physicists resisted atomism during most of the nineteenth century because there had been no direct evidence for the existence of atoms before 1905. An additional reason was that physicists were eager to toe the positivist line traced by Auguste Comte and John Stuart Mill, and retraced by Ernst Mach, Pierre Duhem, Karl Pearson, and Wilhelm Ostwald – a line that at the time seemed to demarcate science from non-science (in particular religion and metaphysics). The line in question was of course that enclosing the directly observable facts. In other words, the claim was that science deals only with the observable. The attempt to accommodate modern science in this procrustean bed gave rise to a funny paradox: between 1925 and 1935, Heisenberg (1947), Bohr (1958), Born (1949), Weizsäcker (1951), Pauli (1961), and a few other eminent physicists paid lip service to the phenomenalist tenet, at the same time that they were building the first successful (true) theories about such transphenomenal things as electrons, atoms, and photons. Indeed, they claimed that their theories dealt exclusively with observations. Thus, Heisenberg’s foundational paper of 1925 announced the construction of a quantum theory “based exclusively upon relations among quantities that are observable in principle.” But observations, even when automated, involve observers at some point. And one cannot observe electrons and the like; one can only observe macrophysical manifestations of microphysical events, such as the clicks of a Geiger counter. And physics does not account for perception anyway. Besides, one may ask naively, Observations of what? But one would then be rebuked, for that question presupposes the existence of things observed and distinct from the means of observation, as is the case of the atoms in the experimenter’s brain or in the Sun’s interior. Phenomenalism was not restricted to microphysical entities: Bohr, Heisenberg,
68
Chasing Reality
and their followers stated confidently that macrophysical things, such as streetcars, result from repeated observations. Even half a century later we were told that quantum physics has taught us that “the Moon is not there when nobody looks” (Mermin 1981: 405). The universe as a whole would be in an even worse predicament, since there can be no observers external to it. But if so, why do cosmologists persist in exploring it? And what do the latter-day Berkeleyans and Kantians make of the recent discovery that most of the matter in the universe is “dark”: that it is there but cannot yet be studied? Did the founders of quantum physics practise the subjectivism they preached? Of course not. In fact, when calculating energy levels, transition probabilities, scattering cross-sections, and the like, all quantum physicists assume tacitly that these are objective properties of quantum-mechanical objects; so much so that no reference to the measurement device, much less to the observer’s mind, occurs in their calculations. That this is so, that the basic laws of quantum mechanics and their main applications do not refer to any measuring instruments, let alone observers and their mental states, is best shown by axiomatizing the theory (Bunge 1967b, Pérez-Bergliaffa, Romero, and Vucetich 1993, 1996). The reason is that, when working in an axiomatic theory, one is not allowed to smuggle in extraneous items in mid-course, as is done when interpreting eigenvalues as possible measurement outcomes rather than as sharp values of objective properties, statistical variances as disturbances due to the measurement device rather than inherent fuzziness, and so on. The axiomatization of quantum mechanics shows that the assumptions of this theory, like those of any other scientific theory, are of two kinds: whereas some are mathematical formulas, the rest are semantic assumptions (the old “correspondence rules”) stipulating the factual items that those symbols stand for. And such factual items are things in themselves, such as electrons, and their (primary) properties, such as energy. For instance, the axiomatization in question, in exhibiting that quantum mechanics is really about quantum objects, helps interpret correctly Heisenberg’s so-called uncertainty relations. These inequalities state that the variance of the position is inversely related to that of the momentum. (Actually the formulas in question involve the square root of the variance, that is, the mean standard deviation.) Heisenberg himself launched in 1927 the popular interpretation of these variances as measuring the disturbances caused by the observation device. Ironically, most positivists and naive realists agree on this point. Positivists overlook the fact that such interpretation presupposes that the micro-object in question has at all times a sharp position and a sharp momentum, while they should be saying that both properties are generated by the act of observation. And the naive realists forget that quantum mechanics does not contain classi-
Antirealism Today: Positivism, Phenomenology, Constructivism
69
cal position and momentum variables with sharp values at all times. (In other words, formulas such as x(t) = f(t) and p(t) = g(t) belong in classical mechanics, not in the standard quantum theory. Only Bohm’s reformulation of the theory does contain such classical variables on top of the quantum-mechanical ones.) Another point overlooked by the two conflicting parties is that Heisenberg’s inequalities are derived in all generality, without assuming anything about measuring instruments; indeed, these are not mentioned in the premises that entail the formulas in question (see Bunge 1967b). The truth is that the Heisenberg inequalities show that, in general, the position and momentum of microsystems are blunt rather than sharp. In other words, quantons are smeared both spatially and dynamically. The limit cases, where one of the two variances is nil (point-like localization and infinite plane wave) are extreme idealizations. At all events, the general quantum theory describes noumena, that is, physical things, not phenomena, that is, mental processes. Moreover, when a measuring instrument is involved, the general theory proves insufficient precisely because it is general. Hence, it must be enriched with assumptions concerning the apparatus’s mechanism, such as condensation of water droplets around ionized particles. And, because every apparatus is macrophysical, it is describable largely if not wholly in classical terms, as Bohr (1958) rightly insisted. Even Heisenberg (1958: 55) admitted eventually that “[c]ertainly quantum theory does not contain genuine subjective features, it does not introduce the mind of the physicist as a part of the atomic event.” Two decades later, Heisenberg told me that he, like Newton, had attempted to explain things, not just to account for data (Bunge 1971). So, Planck (1933), Einstein (1950a), and Schrödinger (1935) had been right in criticizing the positivist interpretation of quantum mechanics. Where some of them went wrong was in their nostalgia for classical physics. On this point, Einstein was wrong and Bohr right. Realism does not imply classicism (Bunge 1979c). To understand why most of the creators of the quantum theories could fall for such a rudimentary philosophy as phenomenalism, it must be admitted that there is method in their madness. The point of method is that the elicitation and measurement of typical microphysical events, such as the artificial transmutation of elements, requires elaborate experimental devices – though not more so than the ones involved in optical measurements. So, it is true that, as Léon Rosenfeld (1953) once said, the experimenter “conjures up” the quantum facts. But Mother Nature does the same all the time and without philosophical obfuscation. Thus, the quantum object and the measuring instrument are intimately
70
Chasing Reality Recorder (e.g., Observer)
Apparatus Disturbing reaction
Signal
Object Figure 3.2MThe measured object, the measuring apparatus, and the recording device (whether observer or automated) are interconnected but distinguishable. So much so, that either component can be disconnected from the other two.
coupled to one another – as long as an experiment is running. And, since measuring instruments are set up and handled by experimenters, one can speak of the object-apparatus-observer supersystem. However, contrary to the Copenhagen dogma, this is not an indivisible whole: connection does not entail indistinguishability. Indeed, one can adjust an instrument without poking into the experimenter’s brain. Besides, the vast majority of quantum facts, such as nuclear reactions, atomic collisions, and chemical reactions, occur all the time outside laboratories. Furthermore, every quantum observation bears on a quantum object, such as a photon, not on an apparatus, let alone on an experimenter. For instance, physicists measure the wavelength of light emitted by an ionized gas, not its effect on the experimenter’s brain. Finally, experiments can be automated, allowing the experimenter to go home for the night and thus cease influencing the object. See figure 3.2. Let us go back to the reason that might be invoked for refusing to talk about quanta in themselves. Bridgman (1927: 150–3), a superb experimentalist and the systematizer of operationalism, reasoned as follows with regard to light. He started by reminding us, correctly, that we do not see light: all we can see are things lighted. In particular, we see light emitters, such as fires and lamps, and light detectors, such as photoelectric cells and eyes. Hence, Bridgman
Antirealism Today: Positivism, Phenomenology, Constructivism
71
reasoned, to speak of light travelling between source and sink “must be recognized to be a pure invention” (ibid.: 153). Bridgman overlooked the fact that Maxwell’s equations describe the propagation of light in intermediate space, so they would be “violated” if they were interrupted arbitrarily in mid-space. Furthermore, these equations contain the universal constant c, which happens to denote the velocity of light in a vacuum – a quantity that Rømer first measured in 1676. Clearly, if only phenomena are allowed to exist, one is forced to give up any hope of understanding how the light emitted by a star, or even by a nearby electric bulb, can reach us – even if the light source ceased to exist millions of years ago. Moreover, as Einstein stressed repeatedly, the realization that electromagnetic fields exist on their own (rather than riding on the mythical ether), and on par with bodies, constituted a milestone in the history of science, since it coincided with the definitive breakdown of the mechanical worldview. The same year that Bridgman’s book appeared, Heisenberg published his famous if flawed paper on the gamma-ray microscope, where he stated that electrons do not have continuous trajectories, but jump from one observation to the next – an instance of the operationalist confusion of evidence with reference. This led Hans Reichenbach (1944), the prominent neo-positivist, to writing about “inter-phenomena,” that is, whatever is alleged to lie between observable events. He claimed that taking inter-phenomena seriously leads to the so-called quantum-mechanical paradoxes, such as that an electron interferes with itself after passing through a double-slit device. Obviously, Reichenbach did not wonder how it was possible for photons to travel undetected all the way from the sun to his eyes. Reichenbach, like Bridgman and Heisenberg before him, overlooked the law statements – such as those involved in calculating time travels and scattering cross-sections – that describe the uninterrupted travel of a particle between source and target. If we admit these law statements as at least approximately true, we must also admit that “inter-phenomena” are just as real as phenomena. And unless we admit that there are real facts between observations, we cannot hope to understand how a television set works, since what we see are flashes caused on the screen by the impacts of the electrons emitted by the cathode-ray tube behind the screen. The TV repairmen know this, since they earn their living looking behind screens. Realism, then, is not optional for the experimental physicist and the electronic engineer: they count on the existence of electrons and photons and things even while they are not being measured. Nor is it optional for the managers of particle accelerators and other large devices, who must find billions of dollars to fund research into such problems as whether strings,
72
Chasing Reality
gravitons, and Higgs bosons exist. Only a philosopher could persuade himself that one may dismiss the issue of realism (Fine 1986, Teller 1995) – which sounds like the hermeneuticists’ denial that there is anything outside of texts. Who but scientists are expected to find out what there is? And who but philosophers of science can be expected to dig up the philosophical presuppositions of the scientific exploration of reality? In sum, the subjectivist components of the Copenhagen interpretation of quantum mechanics are inconsistent with the way the theory is actually being applied and tested. Those components are just philosophical contraband: this is why they are cheap and hard to control. (More in Bunge 1959b and 1973, Jammer 1966, and Beller 1999.) 4 Ptolemy Redux The well-known philosopher Bas Van Fraassen (1980) too embraced phenomenalism, claiming that every successful scientific theory “saves the phenomena,” that is, correctly describes observable facts. The phrase “to save the phenomena” had been coined nearly two millennia earlier by the great astronomer Ptolemy. Duhem (1908) adopted it as the devise of his own brand of phenomenalism, which he regarded as the nucleus of the “qualitative physics” that he advocated as a good Christian and hater of Galileo – whom he once contemptuously called “le mécanicien florentin.” But of course there is no hint of phenomena (appearances to a sentient being) in any of the formulas of the quantum theory, any more than in any other physical theory: none of them contains variables denoting secondary properties or qualia. A weaker empiricist claim is that the evidence for microphysical theories is constituted by perceptible macrophysical features such as angles of pointers of measuring instruments and tracks in cloud chambers. This argument is wrong because it ignores the fact that all such evidence is indirect: it works only thanks to hypotheses that bridge microphysical events (such as charged particle trajectories) to macrophysical ones (such as the condensation of water molecules resulting from the ionization caused by those particles). True, many textbooks in quantum theory discuss such experiments as double-slit diffraction and the Stern-Gerlach one. But they describe them in sketchy terms and with few if any formulas, glossing over the fact that in both cases one needs to “read” the end results, such as the impacts on a detecting screen, with the help of indicator hypotheses that bridge the micro- to the macro-level. (More on indicators in chapter 7, section 9.) Interestingly, the places of evidence and reference are often reversed in other sciences. For instance, what is observable in the social sciences is usually
Antirealism Today: Positivism, Phenomenology, Constructivism
73
individual behaviour. Thus, the data on GDP, income inequality, and birth rate are aggregates of individual data. In other cases macrosocial data, such as documents about fortresses, roads, and tombs, point ambiguously to features of unobservable individual lives. In the historical sciences, from evolutionary biology to archaeology, the only observables are such things as fossils, arrowheads, hearths, earthworks, and tracks, which are just traces left by the referents of those sciences. To state that the theories in these disciplines “correctly describe what is observable” is a gross distortion. In this case, what is observable is only evidence for hypotheses concerning unobservable referents. For example, palaeontology describes extinct organisms, whose fossil remains are described by a different discipline, namely taphonomy. In all cases, data never “speak for themselves”: they have to be “interpreted” via indicator hypotheses, such as those linking barometer readings to atmospheric pressure, or the number of new construction permits to economic prosperity. Such indicators bridge objective facts to the evidence relevant to theories about the former:
Fact
Observation Indicator hypothesis Datum Evidence
Regrettably, most philosophers overlook indicators, whereas personal experiences seldom fail to interest them, as shown by the persistent interest in qualia. Let us glimpse at the latter. 5 To Phenomena through Noumena Human knowledge comes in many kinds: from experiential or egocentric, such as “I feel hot,” to impersonal and theoretical, such as “Heat is random molecular motion.” Correspondingly, our concepts lie on a broad spectrum, from the particular and phenomenal, such as “sweet,” “painful,” and “elated,” to the universal and abstract, such as “extended,” “alive,” and “fair.” However, empiricists find the former more basic and reliable than the latter, whereas traditional rationalists regard universals as prior and immediately apprehensible. I shall contend that we need both categories, each for a different purpose: the phenomenal concepts for self-description, and the universals for thirdperson accounts – in particular, objective accounts of sensations, feelings, moods, and the like. Qualia are pre-analytic experiences, such as seeing blue or tasting chocolate, feeling lust or a toothache. Qualia are also called experiences, raw feels, and even what it’s like, as in “What is it like to be a bat?” – an unanswerable
74
Chasing Reality
question that earned Thomas Nagel instant celebrity. Qualia have baffled philosophers for more than two millennia. In particular, their mere occurrence has been regarded as refuting materialism because they are not describable in physical terms (see, e.g., Block, Flanagan, and Güzeldere, eds. 2002). A few philosophers, such as Berkeley, Kant, Mill, Mach, Avenarius, Carnap, and Goodman, regarded qualia as primary, and even as the ultimate constituents of the universe – but they failed to show how to construct physical concepts out of phenomenal ones. A few others, such as Dennett and Rorty, have denied the existence of qualia. However, most philosophers admit them, but hold that they are beyond the ken of physics, whence they render materialism untenable – which of course assumes the illegitimate equation of materialism with physicalism. To be sure, physicalism, or vulgar materialism, cannot tackle qualia because these occur in brains, which are supra-physical systems. It is likewise incapable of accounting for much more as well, such as the peculiarities of life (e.g., evolution) and society (e.g., culture). However, physicalism is only the primitive version of materialism, one that is not accepted by materialist psychologists, such as Hebb (1980), or materialist social scientists, such as Trigger (2003a). The alternative to physicalism is emergentist and systemist materialism, which has been around at least since Holbach’s time. This ontology makes room for supra-physical (though not aphysical) levels (see, e.g., Bunge 1959c, 1977c, and 2003a). Indeed, according to this ontology, a quale is just a subjective process, hence one in some brain. As such, qualia can only be studied in depth by cognitive neuroscientists. In other words, although qualia are first-person processes, they can be studied by a third person, though conceptually and experimentally rather than experientially. Of course, qualia are not transferable: for instance, though I may empathize with you, I can feel neither your joy nor your pain. But psychobiologists have made enormous strides in the description and explanation of “what it is like” to have experiences such as smelling coffee and suffering phantom-limb pain. As a matter of fact, qualia have always been among the objects of study of psychology and neuroscience. We have learned, for instance, that tasting is a long causal chain of events that starts at the taste buds and ends in various regions of the neocortex. We also know that taste can be refined as well as blunted – by, for instance, repeated exposure to very spicy food, some of which kills the taste buds. To summarize up to this point: There are no qualia in the physical world, for they are located in the subject-object interface. Indeed, every quale has two sources: the brain and something else, whether outside the body or inside it (e.g., in a tooth nerve). In other words, whereas some physical properties (such
Antirealism Today: Positivism, Phenomenology, Constructivism
75
as chemical composition) are intrinsic, all phenomenal properties (qualia) are relational: they are all for us, none exists in itself. However, subjectivity does not exclude objectivity: all it does is to challenge the neuroscientist or psychologist, asking her to try to give a third-person account of first-person experiences. However, this point deserves a new section. The first of Newton’s (1947 [1687]: 398) Rules of Reasoning in Philosophy reads thus: “We are to admit no more causes of natural things than such as are both true and efficient to explain their appearances.” Shorter: Noumena are to explain phenomena. This is why Newton’s laws concerned only primary properties of matter, such as mass and force; and that is also why his axioms were laws of motion, not descriptions of subjective experiences. Contemporary psychologists follow Newton’s advice, and attempt to explain phenomena along the lines sketched by the great Hermann von Helmholtz, a polymath and pioneer of physiological psychology. Among Helmholtz’s many contributions is the idea that perceptions are signs (indicators), not images (Abbildungen), of outer-world facts: “[W]e can only learn how to interpret these signs by means of experience and practice” (Helmholtz 1873: 274). In other words, sensory stimulation triggers complex brain processes involving cognition and memory, and sometimes expectation and emotion as well. Shorter: the ability to sense is innate, whereas the ability to perceive is learned. Leonardo said it centuries earlier: We must learn to see. Incidentally, Lenin (1947) took Helmholtz for a subjectivist. Actually, Helmholtz was a sophisticated realist, who would not have accepted Lenin’s “reflection theory of knowledge” or Wittgenstein’s “picture theory of language,” their names for naive realism. Helmholtz taught that, to understand phenomena, one must study not only external stimuli, but also the organ of perception. His prescription was lost on the behaviourists, from Watson to Skinner, who were materialists but ignored the organs that transduce inputs into outputs. Helmholtz’s injunction was also lost on the phenomenologists, who denounced realism as “absurd,” and claimed to have superseded the objective study of subjectivity with their “absolutely subjective” discipline, phenomenology, which stands in “the extremest contrast to science in the hitherto accepted sense” (Husserl 1960: 30). Unsurprisingly, no phenomenologist has found any new facts about phenomena, let alone any new laws about them. Much the same holds for cultural psychology, a closely related school inspired by Wilhelm Dilthey, a follower of both Kant and Hegel (see, e.g., Ratner 1997). Its members too reject positivism, which they mistake for the scientific method; they make much of Verstehen (arbitrary interpretation of both behaviour and utterances); and they extol qualitative and non-experimental procedures. This is why they cannot boast of
76
Chasing Reality
any findings, whereas scientific psychology, in particular the science of perception, has made sensational advances over the past few decades (see, e.g., Gazzaniga, ed. 2004). Some appearances can be explained in purely physical or chemical terms. For instance, a Sun eclipse is explained by the interposition of the Moon. And the variety in the colours of quartz, beryl, diamond, and other gems is explained by the presence in them of tiny amounts of chemical impurities. For example, emeralds are beryl stones that look green under white light because they include tiny amounts of chromium, which absorbs the light waves of all wavelengths except for those that cause the sensation of green. Emeralds can lose that colour when strongly heated or subject to acids, but not spontaneously. Hence Goodman’s (1954) infamous “grue” (or “bleen”) imaginary emerald, green before a certain arbitrary date and blue thereafter, is an idle fantasy. Indeed, emeralds cannot change colour spontaneously at any date unless strongly heated or unless their chemical composition is altered. Thus, the value of the “grue” paradox to shed light on induction is only this: Beware appearances and artificial examples concocted just to support or undermine philosophical speculations (Bunge 1973b). However, the scientific explanation of most appearances requires invoking the brain. For instance, we perceive more readily and vividly events we value (perhaps because of their survival significance) than events to which we are indifferent. This suggests that emotion strongly colours cognition, to the point that we remember best the items, even if trivial, that we first encountered in association with pleasure or pain, joy or sorrow, surprise or embarrassment. None of this should be surprising given the anatomical connections among the cortical (cognitive) and the limbic (emotive) regions of the brain. What is amazing is the persistence of most cognitive psychologists – particularly those of the information-processing persuasion – in trying to account for cognition without feeling and emotion, motivation and spontaneity, as if our brains were dry, cold, and unimaginative computers (see Damasio 1994, LeDoux 2003). The unscientific philosophers of mind claim that phenomenal qualities, or qualia, are inaccessible to scientific investigation, and perhaps even to ordinary knowledge, because they are private: I cannot feel your pain or taste strawberries in the same way as you do. Admittedly, mental experiences are not transferable; but they are describable, and some descriptions can be understood because all adults share not only roughly the same basic brain organization but also myriad similar experiences. In any event, subjective experiences are being studied objectively as neurophysiological processes influenced by cultural traditions and social circumstances. (For example, it is well known that many tastes are acquired; and it has been found that Himalayans have a
Antirealism Today: Positivism, Phenomenology, Constructivism
77
significantly higher pain threshold than Mediterraneans.) That is, phenomena are being explained by noumena. What about social attitudes, such as selfishness and altruism? The sociobiologists contend that altruism is “selfishness in disguise,” as Dawkins (176: 5) put it, copying the old utilitarian maxim. This certainly holds for reciprocal altruism (“I scratch your back and you scratch mine”). But what about authentic altruism, as exemplified by giving without prospects of reward? The sociobiologists claim that such gifts are offered, if at all, only to kin, because genes are selfish and force us to behave in such a manner as to propagate them. However, there is ample ethological evidence for the occurrence of cooperation among biologically unrelated individuals, even of different species (e.g., Cockburn 1998, Clutton-Brock 2002, Griffin 2001, Hammerstein, ed. 2003). Hence, an alternative explanation of altruism is needed. This comes from the study of primates (e.g., de Waal 1996) and people (e.g., Bateson 1991). These studies have shown that humans and the great apes are empathic, and that the performance of generous deeds makes us feel good (Rilling et al. 2002). Other studies found a significant reduction in mortality among old people who help others (Brown et al. 2003). Still other research has shown that the social emotions, in particular altruism and fear of “the other,” can be manipulated in the interests of political organizations, to the point that “the real social engineers are today’s political consultants” (Massey 2002: 21). In short, altruism is sometimes what it appears to be, namely, selflessness. The epistemological morals are obvious. One of them is that explaining a fact need not involve denying it. Another is that it is foolhardy to seek explanations of social facts at the molecular or even cellular level. All of this has certain consequences for the problem of reduction, to be examined anon. 6 Interlude: Reduction Consider the following purported examples of reduction. (1) Water = H2O (2) Evening Star = Morning Star (3) Speech understanding = Specific activity of the Wernicke brain area An anti-materialist might object to (1), arguing that this alleged identity does not account for such familiar properties of water as its ability to flow, freeze, and be pored, much less feel wet on contact. And he would be right. The well-informed materialist too will deny (1), though on different grounds. The emergentist materialist will assert instead that (1') Water = A body composed exclusively of H2O molecules. He will add that fluidity and the ability to evaporate and freeze are emergent
78
Chasing Reality
properties absent from the component molecules, though rooted in the hydrogen bonds among them. He will also contend that, whereas physics can explain the first three properties, only psychology can (eventually) explain the wetness sensation as a brain process triggered by a process in the skin and transmitted to the brain by certain peripheral nerves. Example (2) highlights the difference between physical and phenomenal objects. The physical referent of the two descriptions is the same, namely, the planet Venus. But whereas the first description points to Venus seen at dusk, the second points to Venus observed at dawn: one physical object behind two experiential objects. The two perceptions are different for two reasons. First, ordinarily the air is cooler and less polluted at dawn than at dusk, so that a Morning Star image is brighter and more distinct than an Evening Star one. Second, most people are more tired in the evening that in the morning. In short, the perceptions and descriptions of one and the same physical object seen in different circumstances are likely to be different. But this difference, far from being ineffable and mysterious, can be expressed rather clearly and demystified with the help of science. The case of the purported identity (3) is different from the previous cases, in that it is a genuine identity in the context of neurolinguistics. It is a contingent (or non-logical) identity as much as “Heat = Random molecular motion” and “Light = Electromagnetic radiation of wavelength lying between so many and so many angstroms” are reductive identities in physics. Certainly, speech understanding has further properties, such as eliciting pleasure or fright, admiration or boredom, and so on. But some of these are effects of speech processing in different brain subsystems. The identity occurring in (3) is ontological, not epistemological. That is, the materialist does not claim that a deaf-mute from birth can ever get a firstperson knowledge, or knowledge by acquaintance, of speech. All he claims is that speech comprehension is essentially a brain activity. But he also admits that only people with speech experience can do neurolinguistics, and this not because speech has an immaterial component, but because neurolinguists study the speech organ, which is defective in a deaf-mute. In other words, knowledge by description may account for knowledge by acquaintance, but neither is a substitute for the other. Neuroscience alone cannot give a comprehensive account of any speech experience – but cognitive neuroscientists hope to perform this task. (See a similar argument in Churchland 1989 and Lewis 2002.) In short, ontological psychoneural identity does not imply epistemological identity. The problem calls for the merger of neuroscience with psychology rather than for the reduction of the latter to the former (Bunge and Ardila 1987).
Antirealism Today: Positivism, Phenomenology, Constructivism
79
7 Psychological and Social Appearances It is well known that every one of us is “perceived” differently by others. It is also well known that some personal characteristics are conceptualized differently in different societies at different times. Thus, although being red-haired, left-handed, hunchbacked, or “melancholic” (depressive) are not social constructions, but objective and harmless biological characteristics, they were stigmatized until recently. Mental illness passed from a feared sign of divine or genetic curse to a manifestation of an “unresolved Oedipus complex,” or even a social construction, to a medically treatable condition on a par with the other sicknesses (see Shorter 1997). Some of what holds for persons also holds, mutatis mutandis, for social systems, institutions, and norms. Indeed, some of these are “perceived” (conceptualized) differently by different people, sometimes because of ideological bias, and at other times because they perform more than just one function. Social status is a case in point: most American blue-collar workers “see” themselves as members of the middle class, not the working class, and they perceive their wages to be significantly higher than what they really are. Religion is similar. Denounced by some sociologists as a fig leaf to cover up oppression, and exalted by religionists as the most sublime spiritual experience, actually it is both a tool of social control and a view of the cosmic and social orders. Because it may be embraced willingly by large segments of the population, religion should not be dismissed as just a tool of the ruling classes on a par with the police. In particular, in early civilizations “[t]he cosmos was conceptualized as a kingdom, but it was a kingdom in which the powers of the upper classes were based more on the consent of the governed and less on the exercise of coercion than in many later pre-industrial societies” (Trigger 2003a: 491). This is not to deny that there are genuine social constructions. All institutions and social practices are social constructions, and as such they cannot be explained in purely biological terms. So is the postmodern movement that conceptualizes everything, even nature and illness, as a social construction. In particular, it is not by chance that the antipsychiatry movement, which proclaimed that mental illness is a fabrication, and denounced psychiatry and insane asylums as tools of social control, was launched in 1960, at a time of radical questioning of the “establishment,” by such popular writers as Michel Foucault, Thomas Szasz, Ronald Laing, and Erving Goffman. Little did Foucault suspect that AIDS, which he had described as a social construction, would eventually kill him. Cognitive and affective neuroscientists attempt to explain appearances in
80
Chasing Reality
terms of primary properties – those of the central nervous system. For them, as for physicists and chemists, an appearance only poses the problem of its origin. And this problem is solved by looking for the mechanism whereby a noumenon is transformed into a phenomenon, which in turn is described in strictly noumenal terms. For instance, perception, such as colour vision, is gradually being explained in terms of processes occurring in several neuronal systems in the neocortex (see, e.g., Zeki 1993). And emotions, such as fear, are being explained in terms of processes in the cerebral amygdala and other components of the limbic system (see, e.g., Rugg, ed. 1997, Squire and Kosslyn, eds. 1998, and Gazzaniga, ed. 2000). In sum, mental events are being accounted for as processes in neural systems involving synapses, neurotransmitters, and the like – much as the electric current in a wire is explained in terms of electrons and electromagnetic fields. However, because human brains are embedded in society, neuroscience is not enough to explain the mental: this is why psychology has a social component. However, social psychology too endeavours to explain the brain– society interactions in terms of primary properties – those of society in addition to properties of the brain. A famous example is that of the dime illusion: Poor children tend to perceive coins as larger than do their rich counterparts. In social matters, the noumenon-phenomenon relation is far more complex than in other realms. This is because of the so-called Thomas theorem: we do not react to social facts but to the way we “perceive” (actually imagine, conceptualize, and evaluate) them (Merton 1976: 174-6). For instance, many of us buy well-advertised junk food and vote for friendly looking crooks. And in all early civilizations the rulers were regarded as either supernatural beings or as intermediaries between deities and commoners, and were revered and feared in consequence. The very maintenance of the cosmos was supposed to depend upon the strict observance of the social and religious conventions. This may explain why class conflict was not a major feature of early civilizations (Trigger 2003a: 671). Another case in point is self-appraisal, which depends on the reference group. For instance, an Afro-American who complains that his salary is meagre is comparing his income with that other members of his workplace, not with that of derelict Afro-Americans, let alone with the subsistence-level income of the land of his ancestors. His “perception” of relative deprivation is thus perfectly objective, so that his grievance is legitimate and may be traced back to racial discrimination. The sociologist is expected to handle his social perception as a fact just as objective as both the chosen reference group and the discrimination that originates it.
Antirealism Today: Positivism, Phenomenology, Constructivism
81
Much the same holds for the new perception that emerges when the agent adopts an alternative reference group (or acts in a different social circle), such as the personnel of another firm in the same industry. In short, one and the same agent combines with both an objective social circumstance and a reference group to yield a perception of the circumstance. Change the agent, the reference group, or both, and the perception is likely to change as well. This explains why people may react differently to the same social fact: the real chain is not the direct behaviourist two-step process circumstance ® action. Rather, it is the indirect chain:® perception ® decision ® action. Of course, every enrichment in explanation must be paid for by the uncertainties associated with the inclusion of perceptions and decisions. However, experimental psychologists have devised a whole battery of objective indicators of such mental processes. Just as the psychologist studies illusion and hallucination in addition to normal perception, so the social scientist studies the deliberate creation of misleading appearances. Mimesis is found among many animals, such as certain butterflies, fish, and cephalopods. Humans and other primates cheat in various ways, from simulation to dissimulation; but, unlike other animals, we are also capable of self-deception. What would social life be without a modicum of dissimulation and hypocrisy? What then of Durkheim’s (1988) classical methodological injunction, that sociologists should describe social facts in the same objective fashion as physicists, because they are just as real as physical facts? It stands, with the proviso that some of the processes occurring in his subject’s brain – in particular his beliefs, attitudes, evaluations, intentions, and decisions – are among the facts of interest to the sociologist. In particular, social phenomena, that is, the way an individual “perceives” the social facts around him, are to be studied from the outside as so many objective facts. After all, they occur in the student’s external world. Thus, what the subject regards as a phenomenon, the student of society views as a noumenon. Undoubtedly, this expansion of the domain of facts to be studied by the social scientist to include the agent’s subjective experience requires techniques that go beyond the mere recording of overt behaviour. But, if conducted scientifically, this study will lead to testable hypotheses, not to literary images that tell more about the student than about his subjects. (See Burawoy 2003 for a criticism of the fashionable “interpretive turn” that replaces social facts with texts, and scientific papers with literary essays.) In sum, appearances are part of the reality of many animals, particularly ourselves. This contradicts the popular view that appearance is the opposite of reality. Appearances are real, but they are only skin-deep – literally so,
82
Chasing Reality
since they occur in the subject–object interface. And, being superficial, they call for explanation in terms of unobservable (yet scrutable) things and properties, instead of being explainers. However, this point deserves a separate section. 8 Scientists in the Crib? Chomsky (1965: 27) famously attributed extraordinary linguistic abilities to newborns. According to him, every child is born knowing a universal linguistic theory that allows him “to determine which of the (humanly) possible languages is that of the community in which he is placed.” This extraordinary hypothesis of the instant linguist has recently been enlarged: Babies think, construct theories, look for explanations, and perform experiments much like scientists do. In particular, “[c]hildren create and revise theories in the same way that scientists create and revise theories ... We think there are very strong similarities between some particular types of early learning – learning about objects and about the mind, in particular – and scientific theory change. In fact, we think they are not just similar but identical. We don’t just think that the baby computers have the same general structure as the adult-scientist computers ... We think that children and scientists actually use some of the same machinery” (Gopnik, Meltzoff, and Kuhl 1999: 155). A first reaction to this amazing claim is that the words ‘theory’ and ‘experiment’ in the preceding statements cannot possibly have their standard significations. Indeed, a theory proper is a hypothetico-deductive system and, moreover, one that does not contain phenomenal (or observational) predicates – unless it happens to be a psychophysical one. And an experiment is not just any old trial and error process but a carefully designed and controlled one. True, Lashley, Tolman, and Krechevsky conjectured that rats make hypotheses and check them while exploring their environment. Kreshevsky (1932) and other scientists checked and confirmed this conjecture, but they did not claim that rats construct scientific hypotheses and perform scientific experiments, or even that they are conscious of such operations. This is why their favourite rats did not occur as co-authors of their papers. Presumably, the “theories” that Gopnik and her colleagues attribute to infants are vague and tacit (not explicit) hunches such as “Yummy stuff comes out of that when I suck it.” And the baby’s “experiments” are likely to be trials such as babbling or crying to draw the caregiver’s attention. How could the mental processes of babies be identical to those of trained adults, or even naive adults, if the respective neural architectures are so different, as revealed more than half a century ago by microscopic examinations? (See Conel 1939–67.)
Antirealism Today: Positivism, Phenomenology, Constructivism
83
Or are we to believe, with Thomas Aquinas, that the Holy Ghost grafts a fully formed soul upon the embryo or the fetus? The “theories” referred to by the “theory-theory” psychologists are even more primitive than the view of scientific theory held by the logical positivists, such as Carnap (1936–7). This influential author held that scientific theories are composed of statements of two kinds: observational, such as “this red thing is hot and hissing,” and theoretical, such as “the photon absorption caused the ejection of an electron from the atom.” The logical positivists claimed further that only the former are fully meaningful, whereas the latter are only “partially interpreted,” and acquire a vicarious meaning through their association with the former. However, even a superficial examination of scientific theories and experimental designs will show that they are made up of concepts and propositions, not percepts and feelings. And there is no evidence that a baby can form any concepts, such as that of mother, rather than a rather blurry visual, tactual, olfactory, and auditory image of its particular mother. Besides, the babble of infants has a heavy emotional and observational (or phenomenal) load. By contrast, scientific theories lack emotional and observational terms, in particular terms denoting qualia – unless they happen to be about sense perception (recall section 1). Another objection to the scientist-in-the-crib hypothesis is that it is doubtful that babies know of any properties other than phenomenal ones, such as “sweet,” “warm,” “loud,” “rough,” “prickly,” and “wet.” Nor are their inquiries disinterested like those of the basic scientist: infants are curious only about their immediate present environment, and just because they need to know in order to survive. And, being practically minded, they are likely to put together quite disparate things just because they occur jointly in their experience – such as mother and milk, or diaper and urine. The analytical skills of infants are likely to be very limited not only because of the primitive state of development of their brains, but also because their thinking is action-oriented rather than contemplative. In this regard, an infant and Rakmat, the uneducated and illiterate Central Asian adult studied by Luria (1979: 69–71), are likely to perceive their environment in similar ways. Thus, Luria once showed Rakmat a picture and told him: “Look, here we have three adults and a child. Now clearly the child doesn’t belong in this group.” Rakmat replied, “Oh, but the boy must stay with the others! All three of them are working, we see, and if they have to keep running out to fetch things, they’ll never get the job done, but the boy can do the running for them ... The boy will learn; that’ll be better, then they’ll all be able to work well together.” Rakmat’s worldview seems to have been holistic, dynamic, and above all pragmatic.
84
Chasing Reality
Besides, what infants can get to know is very limited, if only because they do not have the same brains as adults, let alone scientists. In fact, pathologists have shown that the neural circuitry in babies is very rudimentary. (In particular, the prefrontal cortex, which is critical for cognition, valuation, planning, and decision, is hardly active in early childhood, and takes two decades to fully myelinate). Furthermore, ever since the epoch-making experiments of Hubel and Wiesel (1962), it has been known that the mammalian brain develops along with experience. We are born learners, not learned. It is even likely that scientists sculpt their own peculiar brains as they learn and do science, and that consequently their brains differ in significant ways from those that musicians or managers sculpt for themselves. A third reaction to the thesis in question is this. If babies and scientists have basically the same cognitive abilities, why is science so hard to learn that the vast majority of people of all ages, even many who have taken science courses, tend to think in unscientific ways, in particular in magical terms? And why did science emerge as recently as 2500 years ago, only to submerge a few centuries thereafter, and to re-emerge precariously only 400 years ago? One reason may be because science deals in primary properties, which are not accessible to the senses. A second reason may be that science posits laws, in particular invariances, that are not found in experience. A third may be that scientific work is done in the midst of scientific communities that regulate themselves in accordance with the exacting scientific ethos first identified by Merton (1968): disinterestedness, epistemic communism, and organized scepticism. In sum, science is qualitatively different from ordinary knowledge. In particular, it is far more rational and less wasteful than trial-and-error guessing. As Wolpert (1992) put it, science is quite unnatural. If scientific research did start in the crib, it should flourish in grade school, and even more so in high school. That it does is the distinctive thesis of pedagogical constructivism, an increasingly influential pedagogical school, particularly in science education (see Fensham, ed., 2004). This doctrine, particularly as articulated by von Glasersfeld (1995), is philosophically just as wrong as its prime sources, namely, Berkeleyanism, Kantianism, and operationalism – all three of which are proudly acknowledged by von Glasersfeld. Pedagogical constructivism is also psychologically false because modern science is so counterintuitive that it is quite hard to learn, and impossible for an individual to reinvent by himself. For example, most physics teachers know that freshmen hold Aristotelian rather than Newtonian opinions about motion. In particular, beginning mechanics students, just like Kant, find inertia hard to grasp. Consequently they are baffled by the fact that the planetary orbits are perpendicular to the gravitational attraction.
Antirealism Today: Positivism, Phenomenology, Constructivism
85
Pedagogical constructivism is not only false. It is also harmful because it denies the existence of objective truth, suppresses criticism and debate, and renders teachers expendable. Can any teacher seriously maintain that normal teenagers can rediscover (or rather reinvent) by themselves the calculus or the theory of evolution? And can any serious scholar claim that one should never contradict anyone because there is no such thing as truth? Is not lack of debate the mark of dogmatism? And is not relativism a tacit confession of an inability or unwillingness to learn how to check for truth? (Further criticisms in Matthews, ed. 1998.) Still, a smart self-styled constructivist teacher may occasionally be more successful than a realist but dull one who believes that students must be spoonfed rather than motivated to study on their own. This will be the case if the teacher encourages her students to think for themselves, whereas her realist colleague demands uncritical parroting. The reason for her success is that, as María Montessori and John Dewey stated long ago, exploration prompted by curiosity is more interesting, hence far more motivating and rewarding, than the repetition of half-understood formulas. Still, science instructors are expected to provide some guidance, if only to avoid the waste of time and the accompanying discouragement that characterize blind trial-and-error endeavors. Teachers are also expected to check whether their students give correct answers, and whether they estimate correctly the measurement error. That is, the constructivist teacher, if minimally competent and responsible, will have to admit that error, and therefore truth too, matters after all. 9 Science and Technology Are Realist The phenomenalist philosopher, like the second-hand car dealer, assures us that what we get is what we see. By contrast, the realist philosopher, like the experienced car buyer, assumes that we always get far more than what we see, because the perceptible (secondary) qualities constitute only the tiny tip of the iceberg of facts. Because of this ontological presupposition of science, any nontrivial project in empirical scientific research aims at finding out what lies beneath said tip. Most of us can see, but only a few understand what they see. As Szent-Györgyi once said, “Discovery consists in seeing what everybody has seen and thinking what nobody has thought.” The assumed nature of reality – that it consists of facts and events that are mostly beyond the reach of perception – dictates the nature of the scientist’s method of inquiry. Indeed, the scientist’s first move is to step out of his subjective and pre-analytic bubble, that of qualia, feelings, inner speech, and the like. That is, the scientist takes the external world for granted, and keeps his
86
Chasing Reality
own perceptions, feelings, and reasoning for himself and under strict control; they are only the starting point. Moreover, he attempts to discover the primary qualities underneath the secondary ones – for example, the molecular and cellular processes felt as a pleasurable sensation. In sum, the scientist’s procedure is the exact opposite of the poet’s, the musician’s, and the idealist philosopher’s: where the latter may give free rein to their untutored imagination, the former must discipline his to keep in touch with reality. The same holds of course for the technologist’s tinkering and designs. No reality checks, no science or technology. The scientist’s imagination is disciplined but, at the same time, it is far richer than the artist’s because the latter, even if a devotee of “abstract” (nonrepresentational) art, is bound to sensory experience. Even a blind plastic artist fashions visible things, and even a deaf musician creates sounds, whereas a theoretical physicist, chemist, biologist, or social scientist creates conceptual systems. Moreover, the theorist, contrary to the artist, conceives of items and properties that have no experiential counterpart – such as force fields, quanta, atomic nuclei, black holes, genes, viruses, government budgets, and the decline of empires. What is more, scientists construct predicates representing primary properties and events of two kinds: superficial, like temperature and visual acuity, and deep, like quantum tunnelling and neuronal plasticity. Thus, contemporary science has enriched the primary-secondary dichotomy introduced by Galileo, to read thus: Basic (e.g., atomic number, social class) Primary Properties
Derivative (e.g., density, social status) Secondary (e.g., cold, wet, bitter)
The scientist’s procedure is also the opposite of Husserl’s phenomenological method, which allegedly dispenses with all presuppositions, and “brackets out” the world. It does this to concentrate on the stream of one’s own perceptions, memories, feelings, expectations, imaginations, and the like (Husserl 1970). How this differs from the mystic’s navel contemplation is not explained. Nor are we shown any list of the wonderful findings that this ancient “method” is said to have garnered. The phenomenologist spends all his time preparing for inquiry, promoting his program as that of a rigorous science, and denigrating standard science and objectivism. He trains for a race that he will never run.
Antirealism Today: Positivism, Phenomenology, Constructivism
87
The ultimate rationale for the scientific method is the wish to attain objective truths, that is, data and hypotheses that are adequate to their objects or referents, regardless of the mood or preference of the investigator. And using this method involves transcending sense data and constructing conceptual systems, such as classifications and theories, that contain only predicates representing primary properties – unless the theory happens to refer to phenomena such as tastes, smells, colours, and feelings, as is the case with certain neuroscientific and psychological theories. Given that scientific theories are conceptual systems rather than piles of sense data, expecting untutored children or illiterate villagers to come up with scientific theories on the sole strength of their daily-life experiences is to betray naive faith in the power of the most rudimentary of philosophies, namely, phenomenalism. Yet this is precisely what the ontological, social, and pedagogical constructivists ask us to believe: that everyone “constructs” the world by herself. In summary, contemporary subjectivism is not only preposterous and an obstacle to the exploration of reality; it is utterly unoriginal. Berkeley had said all that much earlier – in clear and elegant prose to boot. 10 Concluding Remarks The great eighteenth-century mathematician, physicist, engineer, and amateur philosopher Leonhard Euler (1846, 1: 323) was rightly called “second only to Newton.” Euler called the irrealist philosophers “clowns,” and said that their motivation was not the hope of finding truths but just to draw attention to themselves. Euler’s harsh judgment on antirealism would be regarded as bad form in contemporary academia, where all philosophical and pseudophilosophical doctrines, even the most preposterous, least original, most barren, most obnoxious, and most boring, are “given equal time” as serious schools of thought. In particular, antirealism is regarded as academically far more sophisticated and respectable than scientific realism, materialism, or scientism. Antirealism is out of step with science and technology, intent as they are on exploring or altering reality. Antirealism is not just wrong; it is utterly destructive, because it proclaims a total void: ontological, epistemological, semantic, methodological, axiological, ethical, and practical. Such integral nihilism or negativism, reminiscent of Buddhism, discourages not only objective evaluation and rational action, but also the exploration of the world. It is at best an academic game.
88
Chasing Reality
4 Causation and Chance: Apparent or Real?
Appearances do not exhibit causal connections: all we perceive are events – simultaneous or successive, contiguous or distanced, external or internal, real or illusory. For instance, we hear a clap of thunder after seeing a flash of lightning, but we do not perceive the causal connection between the two events: we only conjecture that the latter caused the former. This is why phenomenalists, from David Hume (1888 [1734]) to David Lewis (1998), have had no use for the notion of objective causation–even though, presumably, they took part in myriad causal chains. And this, in turn, is why the word ‘causation’ all but disappeared from the vocabulary of the positivist philosophers between about 1880 and the time when it was rescued from oblivion (Bunge 1959a). Fortunately, the positivist fatwah against causation did not discourage scientists from seeking causal connections and explanations. For example, in the nineteenth century the lightning-thunder hypothesis was confirmed in the laboratory by producing electric sparks and detecting the succeeding shock waves; and neuroscientists proved that insults to the brain may cause mental deficits. During the first half of the twentieth century atomic physicists showed that the cause of light emission is the decay of atoms or molecules from excited states; geneticists proved that some mutations cause phenotypic changes; social scientists proved that the cost of maintaining colonies ends up by exceeding the benefit deriving from exploiting them – and so on and so forth. The scientific evidence for causality became overwhelming. Causal determinism ruled in the sciences nearly undisputed during more than two millennia until the emergence of quantum mechanics in 1925. From then on it has had to contend with probabilistic determinism, often misleadingly called ‘indeterminism.’ Moreover, it is often claimed that causation is a particular case of randomness, namely, when the probabilities concerned equal
Causation and Chance: Apparent or Real?
89
unity. However, as will be seen below, this is a mistake: the concepts of causation and chance are mutually irreducible. Interestingly, though, sometimes they combine, as when one calculates the probability that a certain event will cause another. In any event, we still need the traditional concepts of efficient and final causation even though we now admit that their extension or applicability is limited. For example, the causal principle does not hold for particles or photons travelling in far-out space, since nothing pushes or pulls them. Nor does it hold for the spontaneous occurrence of sensations and thoughts in the absence of external stimulation. In short, not all facts call for causal explanations: only some of them do, namely, changes – and not even all of them. For instance, the presence of this book on your desk is a fact but not an event, whence it does not require a causal explanation. What does call for explanation is why the book came to occupy this unusual place. And here several causes may be invoked – perhaps because you placed it there (efficient cause) to consult it (final cause). In short, causation implies change, but the converse is false. What about chance: what if any is its ontological status? The oldest and most popular view about chance is that it is not real: that it is only a name for our ignorance of the relevant causes. This has been a fertile view, for it has encouraged the search for causal connections. This search has been so rewarding that until recently causal determinism was regarded as part and parcel of the scientific worldview (e.g., Bernard 1952: 92). George Boole (1952: 249), echoing Leibniz and Kant, went even further. He believed that “the idea of universal causation seems to be interwoven in the very texture of our minds.” Much later, Jean Piaget discovered that young children learn to attribute changes to causes, and ethologists found that the same holds for apes and corvids when they are making or using tools such as sticks and hooks. However, certain scientific advances in the nineteenth century, notably the invention of the calculus of accidental errors of observation, and statistical mechanics, suggested that chance is just as real as causation. The emergence of quantum mechanics, genetics, molecular biology, and communication engineering in the twentieth century confirmed the firm place of chance in the world, and the concomitant role of the theory of chances in factual science. Think of quantum jumps, chance mutations, and noisy communication channels. Still, none of this has proved that chance is basic or irreducible; it has only proved that the concepts in question are necessary ingredients of scientific knowledge on certain levels of analysis. We must therefore inquire whether or not chance is part of the fabric of the world, and whether probabilities can be attributed to facts, or whether they only measure the strength of our beliefs.
90
Chasing Reality
In this chapter we shall glimpse at the concepts of causation and chance, and discuss some of the philosophical problems they raise. However, chance will get here the lion’s share because it is the more slippery of the two concepts. The upshot will be that science needs both concepts, as a consequence of which metaphysicians should update determinism to include probabilistic as well as causal laws. 1 Causation Let us characterize briefly the notions of causal relation, causal principle, and causal determinism. The latter’s limitation will also be noted. (For details see Bunge 1959a and 1982.) The standard analysis of causation is in terms of necessary and sufficient conditions: X causes Y if and only if Y. Though correct, this explication is not quite satisfactory in ontology, and this for two reasons: first, because the notions of necessary and sufficient conditions occur in the explication of logical equivalence, not of real processes; second, because in the biconditional “X if and only Y,” the events called X and Y are interchangeable, whereas in reality causes cannot be exchanged for their effects: the causal relation is asymmetric. In ontology we need a concept of causation that will fit facts, such as the brain process that goes from hearing the shout “Watch it!” to looking at its source, imagining and evaluating the imminent dangerous event, making the decision to avoid it, and jumping away from it. This complex process between auditory stimulus and muscular response is a causal chain: a succession of events in a material system – a human body. We need then an explication of causation in terms of events and energy fluxes. We start by making the usual if sometimes tacit assumption that the causal relation obtains between events (changes of state in the course of time), not between things or their properties. A simple classical example is Hooke’s law: The strain or deformation of an elastic body is proportional to the applied tension or load. Because only events can cause, we must disallow such expressions as “Gene G causes trait T ” and “Brain causes mind.” We should say, instead, that the expression or activation of gene G causes it to intervene in the biochemical reactions resulting eventually in the emergence of phenotypic trait T. Likewise, we must say that mental event M is identical with brain event B – the only difference being that M is ordinarily described in psychological terms, whereas B is described in neuroscientific terms. To hold that “brain processes cause consciousness,” as Searle (1997: 7) does, is like maintaining that bodies
Causation and Chance: Apparent or Real?
91
cause motions, or that the gut causes digestion. Things do not cause processes: they undergo processes; and these in turn cause changes (events or processes) in other things. Shorter: the causal relation holds only among changes (events and processes). We are now ready for our first convention: Definition 4.1 Event C in thing A causes event E in thing B if and only if the occurrence of C generates an energy transfer from A to B resulting in the occurrence of E. (A more precise definition can be constructed with the help of the concept of a state space introduced in chapter 1, section 2 [Bunge 1977a, 1982]. Let A and B be two different things, or two different parts of a thing, such as the prefrontal cortex and the amygdala of a human brain. Call h(A) and h(B) the histories of things A and B respectively over a certain time interval and in the total state space of A and B. Further, call h(B A) the history of B when A is present or, rather, active. Then we can say that A acts on B if h(B) ¹ h(B|A), that is, if changes in A induce changes in B. The total action (or effect) of A on B is then definable as the set-theoretic difference between the forced trajectory of B, that is, h(B A), and its free trajectory h(B). In symbols, A (A,B) = h(B A) 1 h(B)c , where h(B)c is the complement of h(B). Likewise for the reaction of B upon A. Finally, the intensity of the interaction between things A and B is the set-theoretic union of A (A,B) and A (B,A).) A familiar example is the energy cascade involved in shooting an arrow: chemical energy stored in the ATP molecules of the muscle cells ® Kinetic energy of the arm tensing the bow ® Elastic (or strain) energy stored in the bow ® Kinetic energy of the arrow ® Mechanical and thermal energy absorbed by the target and the atmosphere. Another familiar example is this: the activity in the amygdala we call ‘fear’ stimulates the prefrontal cortex, which in turn “orders” the motor strip to activate the legs. Shorter: Fear causes flight. The energy transfer mentioned in definition 4.1 is negligible in two cases: when the “patient” is in an unstable state (“touch and go”), and when the energy flux is a signal, such as the movement of a neurotransmitter across a synapse, or an order to fire a missile or an employee. Social life is rife with events caused by signals. For example, consumers “signal” their intentions to the market (by buying or abstaining from buying), which “signals” back to them (just by posting prices) what they can afford to buy. This is obvious, but the direction of the causal arrow is not; so much so that there are two schools of thought on this matter. Let us look briefly into it. The neoclassical theorists claim that supply generates its own demand (Say’s “law”): that is, produce and someone will buy, because the market abhors disequilibria. The Great Depression falsified this dogma, and John )
)
)
92
Chasing Reality
Maynard Keynes (1973) suggested that the arrow actually points in the opposite direction. The reason is that demand, far from being a source fact, is an effect of expenditure decisions. This view is not only more realistic than the no-glut assumption; it also connects the microsocial level of individual decisions to the macrosocial level of mass supply and demand. At all events, the controversy on the causal arrow in question can be summarized as follows: Neoclassical theory Macrosocial level Supply Demand Keynesian theory Macrosocial level Demand
Microsocial level
Supply
Expenditure decisions
In sum, for practical purposes we distinguish two kinds of causation: energy transfer and triggering signal. However, it should be remembered that a signal, whether physical, chemical, biological, or social, is a process involving energy transfer. Besides, a signal carries information proper only if it is coded, that is, if it is coupled with an artificial code, such as the Morse code, capable of being decoded by a natural or artificial sensor. (Shorter: information is signal together with meaning.) We are now ready to state the causal principle in its simplest version: Postulate 4.1 Every event has some cause(s). This principle is true in very many cases, but not in all. The exceptions are constituted by the spontaneous events, that is, those that are not elicited by external stimuli. Examples of such changes are self-movement, such as that of a particle or a photon once it has been ejected; natural radioactivity; the spontaneous “firing” (discharge) of neurons; the spontaneous emergence of new ideas; and the emergence of social norms alongside with that of social systems such as teams or gangs that emerge spontaneously (without coercion) around a task or a leader. It might be objected that some of the preceding examples are cases where cause and effect occur in one and the same thing, such as a neuron or a society. True, but if we keep definition 4.1 for a process to qualify as causal, the thing in question must have at least two distinct parts. For example, in principle one may explain radioactive disintegration either as a case of spontaneous tunnelling through a potential barrier, or as a result of the ejected particle having attained, through random collisions with other particles in the same nucleus,
Causation and Chance: Apparent or Real?
93
the kinetic energy required to jump over the said potential barrier. Second example: The deliberate movement of a limb follows upon a decision made in the frontal lobe. Third: A political revolt is internal to a society, but it opposes two parts of it: government and insurgents. In short, a cause proper is always an event external to the thing upon which it acts. A cause may not always produce its effect: the latter may occur only with a certain probability. That is, in some cases “C causes E ” must be replaced with “C causes E with probability p,” or “Pr(E C) = p” for short, where “Pr(E C)” stands for the conditional probability of the occurrence of E given the concomitant or previous occurrence of C. In short, there are probabilistic (or stochastic) laws in addition to causal laws. The best examples of probabilistic laws are those of the quantum theory. The above formula, “Pr(E C) = p,” is the nucleus of the probabilistic theory of causation, according to which causation obtains in the particular case when p = 1 (e.g., Suppes 1970, Popper 1974). But surely we need the concepts of cause and event before we can even write down the above formula. In other words, the concepts of causation and probability are not interdefinable. What is true is something far more interesting, namely, that in many cases the two categories intertwine, even though at first sight they are mutually exclusive. Our explication of causation in terms of energy transfer is at odds with the fashionable account in terms of counterfactuals (Lewis 1973, 1983). According to the latter, if X and Y are actual events, then Y depends causally upon X if and only if, if X had not occurred, then Y would not have happened either. I submit that this analysis is flawed for two reasons. One is that it is tacitly parasitic upon the law-statement “If X happens, then Y too occurs.” The second flaw is that counterfactuals are logical outlaws because they are not propositions proper, and consequently they cannot be assigned truth-values. Let us indulge in a bit of counterfactual history: Imagine what would have happened to physics if Newton had stated his second law of motion in the following form: “A body of mass m would acquire the acceleration a = F/m if it were acted upon by force F.” Why would anybody have bothered to write, solve, or test the corresponding equation of motion for different cases? Counterfactuals are heuristic or rhetorical tricks, not law-statements belonging to scientific theories. We shall return to them in chapter 9, section 6. Besides, in the sciences and technologies, causal considerations focus on properties rather than on whole events. A standard definition is this. Let A and B denote two properties of things of a given kind, and assume that B is a certain function of A, that is, B = f(A). Then a change D B in property B is said to be caused by a change D A in property A if either theory or experiment show that change D B follows in all cases (or at all times) upon the occurrence of D A, and )
)
)
94
Chasing Reality
moreover approximately in the amount D B = f'(A). D A, where f'(A) is the value of the slope of the graph of f at A. The corresponding empirical criterion is obvious: To ascertain whether B is causally dependent upon A, wiggle A and check whether B does change as hypothesized. That is, to make sure that a hypothesized causal connection obtains, make the cause occur, and check the presumed effect. The counterfactual account of causation is qualitative, and thus provides no precise criterion to ascertain whether or not “D B = f'(A). D A” holds. So much for efficient causation, the only type of causation admitted by the founders of modern science and philosophy, all of whom rejected the other three types postulated by Aristotle (material, formal, and final). However, the news of the decease of final causation is exaggerated. Indeed, contemporary science and engineering admit goal-seeking behaviour in higher vertebrates and in artefacts equipped with control mechanisms such as thermostats and orbit-correcting devices. Still, this view of final causation is very different from the old teleological view, according to which the future somehow causes present events. In purposive behaviour, a mental representation of the future, understood as a brain process, triggers a decision in the prefrontal cortex, which in turn causes a change in the motor strip, which in turn causes a limb to move, which in turn may act on a lever or a button. This is a chain of ordinary efficient causes; so much so, that the intention-motion part of the chain can be replaced by an electromechanical prosthesis. As for automatic control, Norbert Wiener and his co-workers explained it long ago in terms of feedback (or circular causation) loops. In short, teleology is perfectly admissible provided it is analysed as emerging from chains of efficient causes. In sum, causal determinism has a wide range of application, but it does not cover all events. Not all interconnected events are causally related, and not all regularities are causal. Chance too has its place in the universe, and consequently determinism should be broadened to include probabilistic laws. However, we have already reached the next section. 2 Chance: Types The word ‘chance’ is notoriously polysemous. We should distinguish at least three distinct kinds of chance: accident or chance encounter, disorder, and spontaneous or uncaused event. As the Stoic Chrysippus explained twentytwo centuries ago, a chance encounter, or coincidence, or contingent event, consists in the crossing of two initially independent causal lines. Familiar examples are the accidental meeting of two acquaintances; finding a treasure when digging a hole to plant a tree; the involuntary collision of two vehicles;
Causation and Chance: Apparent or Real?
95
and the unanticipated consequences, beneficial or perverse, of social action (or inaction). This first type of chance was known to ancient philosophers, particularly Aristotle, who devoted to it several pages of his Physics (196b). Chance in the sense of unique or unrepeatable opportunity, or concours de circonstances, plays an important role in history as well as in individual life. This is why biological evolution has been described as opportunistic or tinkering rather than designed (Jacob 1976). There is a special term for this: a feature that has been co-opted to perform fitness-enhancing functions other than the original ones is called exaptation. This is why Gould (2002) emphasized the role of contingency, that is, accident, in biological evolution. For example, it is conceivable that certain biospecies, in particular ours, might not have emerged had it not been for the accidental combination of certain genic mutations with favourable environmental circumstances. Much the same holds for human history. Thus, it has been argued that the Florentine Renaissance was partly a side-effect of the Black Plague, which favoured the concentration of wealth, and thus rendered bolder commercial enterprises possible. Again, the world-wide influenza epidemic of 1918 was particularly lethal because it attacked populations that had been weakened by four years of war. And some scientific discoveries have been happy accidents – though, as Pasteur and others have noted, it takes well-prepared minds to seize them (see Taton 1955). The word serendipity is often used as a synonym of accidental success – the cases of the discovery of X-rays, radioactivity, penicillin, and stellar radio waves (see Merton and Barber 2004). The admission of the occurrence of coincidences marked an advance over magical thinking, according to which coincidences are indicators of inscrutable connections. This kind of thinking survived, for instance, in Carl Gustav Jung’s doctrine of synchronicity. This is an ingredient of Jung’s combination of psychoanalysis with the occult “sciences” that has become a component of New Age thinking. However, we all know that our life histories are full of chance encounters of many kinds. For instance, not even the most carefully arranged marriage might bypass the genetic features, such as recessive genes and blood type, that have no obvious phenotypic (observable) counterparts. Hence, every one of us has issued from pairings that are random to some extent. Only the original sin was presumably free from chance. The second kind of chance is disorder, such as that caused by dissolving, shaking, shuffling, scrambling, stirring, heating, rioting, and the gathering of mutually independent items, such as births and road accidents, into a statistical population. This is the kind of disorder that statistical mechanics, actuarial statistics, and the sociology of deviant behaviour deal with. It is patent in white noise, the patterns of raindrops, diffusion, the dispersal of genes and seeds, the
96
Chasing Reality
propagation of nerve impulses along an axon, the outcomes of games of chance, life expectancies, and experimental errors. Its mark is the statistical fluctuation or variance. The very existence of the life insurance companies depends on this kind of chance. As Poincaré (1908) pointed out, they would still be in business even if their physicians could foresee the date of death of every one of their policy-holders. Causation on one level may result in chance on the next, and conversely. It has been argued that the games of chance are not really such because coins, dice, playing cards, and the like move according to the laws of classical mechanics. True, but in this case chance inheres in the initial conditions resulting from deliberate randomization, as when a deck of playing cards is shuffled. Certainly, an omniscient being should have no difficulty in finding the pertinent initial conditions. But the outcome would still be random because the gist of such games is a randomization mechanism, such as the thorough shaking or shuffling that will produce an objectively random initial distribution of possibilities, leading to a matching outcome. This is only an instance of the general law that the future depends not only on the pertinent laws but also on the initial and boundary conditions. This type of chance was recognized only at the beginning of the Scientific Revolution. Of course, people had played games of chance much earlier, but only for entertainment or divination. Cardano, Galileo, Pascal, Descartes, Fermat, and the Chevalier de Méré were the first to try and discover laws of chance – a seeming oxymoron. Another century went by before Daniel Bernoulli applied the young probability theory to physics. And it took a further century for it to be applied to social statistics, which deals with the regularities found in large populations of mutually independent human events, such as births and car accidents (see, e.g., Porter 1986 and Gigerenzer et al. 1989). Epicurus introduced a third kind of chance. He speculated that atoms deviate irregularly and spontaneously from the straight line, in a motion that Lucretius dubbed clinamen. The great botanist Robert Brown was the first to discover, in 1828, a real example of this kind of chance, namely, the (Brownian) motion of pollen particles. Six decades later, Einstein explained such visible irregular motions as the effects of random molecular impacts of molecules assumed to abide by causal laws; moreover, he pointed out the variables, such as the mean free path, that had to be measured to test this hypothesis. In this case, randomness on the mesoscopic level of pollen (and dust and soot) particles resulted from the crossing of mutually independent trajectories of microscopic particles, which the theory assumed to satisfy the laws of classical mechanics. Shorter: Chance on one level may result from causation on a lower level (Bunge 1951). This does not render chance illusory: it only
Causation and Chance: Apparent or Real?
97
makes chance relative to the level of organization. (See Bunge 1959b, 1969, and 1977c for the multilevel structure of the world.) The late-nineteenth-century positivists, particularly Mach, Duhem, and Ostwald, had derided both the atomic hypothesis and statistical mechanics for postulating the existence of transphenomenal entities as well as the reality of chance. The eminent Wilhelm Ostwald (1902) proposed his “energetics,” a wild extension of classical thermodynamics, which he claimed to embrace even value theory, as well as to overcome the materialism/idealism conflict. Most physicists around 1900 shared this anti-atomistic prejudice. Fortunately, Ludwig Boltzmann and Max Planck, along with a handful of young Turks, notably Albert Einstein, Jean Perrin, Victor Henri, and Marian Smoluchowski, took the unorthodox line: they put the atomic hypothesis to work, and supported realism. In 1905, Perrin performed the first accurate measurements of Brownian motion, confirming the key theoretical formulas, and thus proved the reality of atoms. He stated that “at every instant, in a mass of fluid, there is an irregular spontaneous agitation” due to the random molecular motion (in Brush 1968: 31). Such fluctuations showed also that the irreversibility postulated by the second laws of thermodynamics holds only in the large: in fact, decreases in entropy, albeit small, infrequent, and short-lasting, are possible even in closed systems. The third type of chance, or spontaneous fluctuation, is prominent in quantum physics. Think of spontaneous radiative and radioactive decays. Even more astonishing examples are the vacuum fluctuations postulated by quantum electrodynamics, and the self-acceleration inherent in the Zitterbewegung (trembling motion) that relativistic quantum mechanics attributes to the electron. Both are reminiscent of Epicurus’s swerve. Sheer coincidences (type 1 chance events), though! It has been speculated that there might be a sub-quantum level devoid of chance. The entities on this level would be described by “hidden” variables, that is, variables with zero variance (or fluctuation). David Bohm’s extension of quantum mechanics has been the most prominent attempt in his direction. However, it did not prosper, mainly because it failed to derive the quantummechanical probability amplitude (the famous c function) from non-stochastic variables. So, the chance postulated by quantum theory is of the rock-bottom type – until new notice! However, the idea of real or objective chance is not obvious. Ironically, one way to get rid of real but unrealized possibilities is to consign them to parallel universes, the way Everett (1957) has done. Suppose a quantum theorist obtains the result that a certain thing can have infinitely many energies, each
98
Chasing Reality
with a given probability. Suppose further that you measure the energy of the thing in question: a sharp energy value emerges from the original blob (distribution). What became of the infinitely many remaining energy values? Everett’s answer is, Every one of them is realized in an alternative world, where a counterpart of yours performs a similar experiment. Although this extravagant interpretation of quantum mechanics has few adherents, it is generally treated respectfully, as if it were a piece of serious if useless science. And yet the many-worlds interpretation of quantum mechanics belongs squarely in science fiction. This is so for the following reasons. First, the parallel worlds are in principle inaccessible and therefore inscrutable, and thus beyond the reach of science. Second, a measuring act is assumed to transform the single initial thing into infinitely many copies of it, which violates all of the conservation laws. Third, the same experiment is assumed to generate infinitely many copies of the same experimenter, in defiance of all the norms of polite society. In short, it is possible to treat every possibility as an actuality in a different world, but at the staggering price of escaping the real world that one is expected to study. Let’s face it: Actuality is pregnant with real possibility. Biological evolution involves chance of the first two types: coincidence and disorder. Indeed, whereas certain phases in the evolution of the Earth favour some forms of life, they harm others. Think of continental drifts, meteoritic impacts, oceanic currents, glaciations, floods, and droughts. And once organisms emerged, they altered the composition of the soil and the atmosphere, which changes in turn induced transformations in organisms – which is why ecologists speak of the biosphere, as well as of active niche (or habitat) construction (Odling-Smee, Laland, and Feldman 2003). Then, there is often plain disorder, as when air and water currents, as well as fires and floods, disperse seeds more or less at random – as seen, for instance, in the distribution of weeds in some regions. The occurrence of chance of either kind would suffice to push conspecific organisms along different evolutionary trajectories, and thus they would end up as different species. In short, as Monod (1970) emphasized, biological evolution is half-random and halfcausal. Darwinism introduced what at first blush looks like a fourth concept of chance: “The gist of the evolutionary notion of chance is that events are independent of an organism’s need and of the direction provided by natural selection in the process of adaptation” (G.T. Eble [1999] in Gould 2002: 1036). This is the sense in which mutations are usually said to occur randomly. But this kind of chance seems to be just a particular case of the first type, since it consists in the crossing of two initially independent trajectories: those of the
Causation and Chance: Apparent or Real?
99
genome and the environment. Whether or not mutations are random in the sense of being spontaneous (type 3 chance) is another matter, and one that should eventually be solved by theoretical biochemistry. Human evolution is an instance of the combination of chance processes of all three types intertwined with causal processes. Indeed, in the case of humans, in addition to genic mutation, there is creativity – conceptual, technical, artistic, ideological, and social. Indeed, we know that original ideas in any realm are inventions rather than knee-jerk reactions to external stimuli. Moreover, some innovations are non-adaptive: think of crime, warfare, gender inequality, racism, exploitation, and the worship of cruel deities and despotic rulers. The occurrence of radical innovations renders biological and social prediction chancy. By the same token, unforeseen novelties render foolproof plans illusory. Yet of course we must plan – though making room for the unforeseen. All three kinds of chance are covered by the single formula “individual irregularity underneath aggregate order.” In other words, probabilistic laws in the small entail statistical regularities in the large. This is the gist of the Law of Large Numbers and the Central Limit Theorems of the calculus of probability. For example, although the successive high-precision measurements of a magnitude can be made to be mutually independent, their outcomes distribute ordinarily on a bell-shaped (or Gaussian) curve. That is, deviations from the mean in one sense compensate for deviations in the opposite sense. Cournot (1843), Peirce (1986 [1878]), and a few others realized early on that the admission of objective chance necessitates a revision of the traditional ontologies, all of which had been causalist (or deterministic in the narrow sense). Peirce went as far as to claim that chance is basic, so that we should adopt what he called a tychist ontology. However, his argument was flawed, because he conflated the two main concepts of law that occur in the sciences: those of objective pattern or regularity, and law-statement, or conceptual representation of the former. Peirce noted correctly that the latter are seldom if ever exact, as shown by the ever-present accidental (non-systematic) errors of measurement; but he concluded that reality itself is irregular. Boutroux (1898 [1874]) and others were led to the same mistake by the same confusion. Ampère (1834) and, much later, a few others (e.g., Bunge 1959b, Armstrong 1978, Mellor 1991) drew the pattern-statement distinction and were thus saved from the said reification. Popper (1950) and others have suggested that the admission of chance in the ontological context requires replacing determinism with indeterminism. However, the word ‘indeterminism’ suggests lawlessness, and we know that chance events have laws of their own, such as the Gauss and Poisson distributions. We
100
Chasing Reality
also know that purely negative doctrines have no heuristic power. In general, negativism is to be avoided if only because only assertions can be precise and fertile. (All of the negative mathematical, scientific, and technological statements are logical consequences of affirmative assumptions. Think, for example, of the irrationality of 2, the impossibility of a perpetuum mobile, the forbidden transitions in the quantum theory, or Arrow’s impossibility theorem in social choice theory.) What the admission of objective chance does necessitate is an extension of determinism to include not only causal laws but also probabilistic laws. Moreover, chance and causation, far from being mutually incompatible, may transform into one another. For example, quantum mechanics recovers the laws of classical particle mechanics as averages (Ehrenfest’s theorem). 3 Objective Probability For better or for worse, the three concepts of chance distinguished above – accident, disorder, and spontaneity – are handled with the help of a single mathematical theory, namely, the calculus of probability. This is an abstract theory with three basic (undefined, primitive) concepts: those of probability space, event, and probability measure. The “events” in the pure theory of probability are just measurable sets belonging in the probability space: they need not be interpreted as changes of state. And a particular probability, such as 1/3, is a value of a function Pr from that space to the unit real interval [0,1], such that, if A and B are disjoint (mutually exclusive) “events,” then Pr(A c B) = Pr(A) + Pr(B). Paradoxically, the general theory of probability does not include the concept of chance, even though it is applicable only to chance events. Indeed, the general theory is just a special case of measure theory, a semi-abstract theory where the argument of the probability function are sets that may be interpreted as line segments, areas, or volumes in some space. This is why every application of the general probability calculus requires the specification of the probability space (e.g., as the power set of the state space of entities of some kind). It also necessitates the addition of one or more substantive assumptions sketching the random mechanism in question, such as the extraction of balls from a lottery urn with or without replacement, or the spontaneous jump of an atom from one state to another. For example, quantum mechanics predicts, and spectroscopy confirms, that the probability of the 3p ® 3s transition of the sodium atom, which is manifested as a bright yellow line, is 0.629.10–8 per second. This probability is neither a frequency nor the strength of a belief: it quantitates the possibility
Causation and Chance: Apparent or Real?
101
of the said event, which occurs to an excited sodium atom whether or not someone observes it. Moreover, unlike frequencies and credences, that probability is expected to approach unity (actuality) as time tends to infinity. (Notice: ‘actuality,’ not ‘certainty.’) That is, given time, a sodium atom in state 3p will necessarily decay to state 3s. Such a transformation of possibility into necessity goes against the grain of modal logic, which ignores time as well as natural laws, as a consequence of which it is irrelevant to science. In short, a probabilistic model M of a chance process consists of the said mathematical formalism F together with a set of substantive (non-mathematical) assumptions: M = FcS. (This concept of a theoretical model, namely, as a special factual theory, is unrelated to the model-theoretic concept, which is merely a mathematical example of an abstract theory: see Bunge 1983. Yet an entire philosophy of science, namely, the so-called structuralist theory proposed by Suppes, Sneed, Stegmüller, and Moulines, rests entirely on the confusion between the two concepts.) The oversight of this feature of the applications of probability theory to the study of facts is responsible for many popular mistakes. One of them is the belief that any mindless application of Bayes’s theorem will lead to correct results. This theorem states that the conditional probability Pr(B ) A) of B given A, equals its “inverse” Pr(A ) B) multiplied by the prior probability Pr(B) and divided by the prior Pr(A). In short, Pr(B ) A) = Pr(A ) B) Pr(B) / Pr(A). This formula is symmetric in the arguments: that is, A and B are mutually substitutable as long as they do not stand for facts. If they do, a mindless application of the formula may lead to ridiculous results. Here is one: Let A stand for the state of being alive, and B for that of being dead. Then, Pr(B ) A) may be interpreted as the probability of the Living ® Dead transition in a given time interval. But, since the converse process is impossible, the inverse probability Pr(A ) B) does not exist in this case: it is not even nil, because the death process is not stochastic. According to Humphreys (1985), counterexamples like the preceding prove that the probability calculus is not “the correct theory of chance.” This remark is correct but incomplete. The failure in question shows that, to be applied, the probability calculus must be enriched with non-mathematical hypotheses. At least one of these must be the assumption that the process in question has a random component. For example, the gambler assumes that the roulette wheel is not biased, and that its successive turns are mutually independent; in the kinetic theory of gases, one assumes that the initial positions and velocities of the particles are distributed at random; the measurement errors are assumed to be mutually independent and distributed at random around 0; and Shannon’s statistical theory of information assumes that the emitter, and possibly the
102
Chasing Reality
channel as well, are subject to random perturbations, which translates as white noise, the opposite of message. The axiomatic probability calculus does not involve the notion of chance because it is just a special case of measure theory, which belongs in pure mathematics. But this is no different from other pieces of mathematics. For example, the real variables x and y related by the linear function “y = a + bx” must be interpreted in factual terms if the formula is to count as a physical, chemical, biological, sociological, or economic law-statement. In other words, the reason pure mathematics does not define the concept of chance is that every instance of genuine chance is a real process, hence one that can be understood only with the help of some specific factual hypothesis, such as the weakening of the inter-atomic bonds caused by heating a solid, or the weakening of the social bonds following a natural disaster or a social revolution. If the concept of chance were mathematical, it would be possible to concoct formulas for generating random sequences. But no such formula is conceivable without contradiction, since randomness is the opposite of mathematical lawfulness; so much so that the currently accepted definition of a random string of 0s and 1s is roughly this: Such a string is random if it cannot be described efficiently other than by exhibiting the string itself (see Volchan 2002). This negative characterization is strikingly at variance with the positive ways mathematical objects are ordinarily defined, whether explicitly (by an identity) or implicitly (e.g., by an equation or an axiom system). In other words, the concept of chance (or randomness) occurs only in the applications of the probability calculus. No randomness, no probability (see also Bunge 2003a). In other words, probability is “a measure of physical [real] possibility” (Cournot 1843: 81). Hence, it is mistaken to assign probabilities to hypotheses. For the time being, the best way to construct a random sequence is to watch nature generate one – for instance, to record the clicks of a Geiger counter in the presence of radioactive material. The next best move is to try and mimic chance – for example, to construct formulas or algorithms for generating pseudorandom sequences that come close to random distributions. But in either case one would obtain a special frequency distribution, such as Poisson’s: the general concept of probability is rather abstract. By the same token, “there are no tests for [the] presence or absence of regularity in general, only tests of some given or proposed specific regularity” (Popper 1959b: 359). However, the general concept of probability allows one to define the concept of random (or blind) choice, namely, that for which all the outcomes are equally probable (see, e.g., Feller 1968: 30). The best-known example of random choice occurs in games of chance. Thus, the two possible outcomes of
Causation and Chance: Apparent or Real?
P
P
P
103
P σ
(a)
(b)
(c)
(d)
Figure 4.1M(a) Flat distribution: perfect randomness. (b) Symmetrical distribution: cancelling out of deviations from the mean. (c) Skewed distribution: simple propensity or tendency. (d) Bimodal distribution: two different propensities.
coin flipping have the same probability, namely 1/2. But the requirement of equal probability is so strong that it is unlikely to be satisfied outside casinos. Indeed, the values of the vast majority of random (or stochastic) variables are assigned different probabilities. Think, for example, of a bell (or Gaussian) curve for the distribution of a given biological feature, such as height, in a population. Similar distributions occur in the statistical theories of physics, chemistry, biology, social psychology, and social science. If one of the possible outcomes of a chance process is far more likely (probable) than the others, we may speak of propensity (or tendency, proneness, leaning, or bias). For example, in the game of craps, which is played with two dice thrown at the same time, the total 7 is three times more likely to turn up than 11, because there are six ways of getting 7, but only two of getting 11. (In gambler’s words, the betting ratio is 6:2, or 3:1.) However, this is a very special chance outcome. Hence, it is mistaken to call the objectivist (or realist) interpretation of probability the ‘propensity interpretation’ – as I myself have occasionally done following Popper (1957). This denomination is particularly wrong in the case of equiprobable outcomes – which is precisely what characterizes uniform or perfect randomness. See figure 4.1. 4 Probability in Science and Technology The concept of chance as accidental encounter is ubiquitous in science and technology. One example among many is any queuing theory, such as a model of the times of arrival of calls at a telephone exchange, or of customers at the cash of a supermarket. Another common example is random sampling, performed routinely at the quality-control phase of a manufacturing process. The simplest and most familiar procedure for building a random sequence is to flip a coin. However, this is a rather artificial example: processes with only
104
Chasing Reality
two equally possible outcomes are extremely rare in reality. A far more common scientific occurrence is that of a statistical aggregate obtained by extruding individuals, such as persons, from their systems, such as workplaces and neighbourhoods, and measuring a single feature of theirs, such as age, height, weight, ability, or wealth. It will be observed that the given feature distributes randomly, typically on a bell-shaped curve. Randomness is here a result of cutting bonds so as to produce mutually independent items. Physical analogue: As an ice cube melts, and the resulting pool of water evaporates, the water molecules become increasingly detached and disorderly. In both cases, the physical and the social, the process has three mutually dependent features: weakening of interindividual bonds, increasingly independent individual behaviour, and increasing disorder on the micro-level. These are the ideal conditions for the applicability of the probability calculus. The concept of chance as disorder occurs in the kinetic theory of gases and its generalization, classical statistical mechanics. These theories assume that a gas is constituted by comparatively distant, and therefore nearly independent, particles (atoms or molecules), and that these “obey” the laws of classical particle mechanics. But the theories in question also assume that the initial positions and velocities of the particles are distributed randomly rather than in an orderly manner. The two hypotheses, taken jointly, entail stochastic laws such as the Maxwell-Boltzmann distribution of molecular velocities. This law assigns a probability, or rather a probability density, to every value of the velocity. Thus, the latter is a random variable on the macrophysical level. In general, Classical laws & Initial randomness Þ Stochastic laws. Consider, for example, the second law of statistical thermodynamics: A closed system tends to evolve from less probable to more probable states until it reaches a state of maximum probability, which is the one of equilibrium. The probability in question quantitates the notion of objective disorder, and it is functionally related to the entropy, a measurable emergent (systemic) property of the system. A simple example is this. Take a new pack of playing cards and shuffle it many times (and honestly), to make sure that the initially ordered configuration has evolved into a thoroughly disordered mixture. The total number of possible final configurations is 52! > 1050. This figure is so huge that no card player could obtain all of the permutations in his entire lifetime. Thus, whereas the initial configuration is unique, the total number of possible disordered configurations is astronomical: the probability of getting a configuration other than the initial one has increased with every shuffling. Thus, in this case probability measures disorder.
Causation and Chance: Apparent or Real?
105
The situation in quantum mechanics is nearly the opposite. Here the basic laws are stochastic; and, whereas some of the derived laws are likewise probabilistic, others are not. For example, the formulas for the possible energy levels and the scattering cross-sections are probabilistic, whereas the formulas for the averages and variances of the dynamical variables, such as linear momentum and spin, are not. A famous example is that of a single electron, photon, or some other quanton, diffracted by two slits in a screen, and impinging on a detecting screen placed behind the first. Such a quanton has a definite probability of going through each of the slits, and it hits the screen at some “point” (very small region) on the detector. As the number of hits increases, a regular macrophysical diffraction pattern gradually emerges. Thus, individual behaviour is erratic though “governed” by probabilistic laws, whereas the resulting collective behaviour is patterned. In sum, the success of probabilistic theories in science suggests that chance is for real: that it is a basic ontological category, not a psychological or epistemological one. (See, e.g., Cournot 1843, Peirce 1935, Poincaré 1908, Du Pasquier 1926, Fréchet 1946, Bunge 1951 and 1981a, and Popper 1957). Consequently, determinism should be enlarged to accommodate probabilistic laws. This enlarged view may be termed neodeterminism (Bunge 1959a). According to this view (a) all events satisfy some laws, whether causal, probabilistic, or mixed; and (b) nothing comes out of nothing or turns into nothingness – a basic ontological principle found in both Indian and GrecoRoman antiquity (see Dragonetti and Tola 2004). The sensational success of probabilistic theories in natural science has stimulated probabilistic thinking in psychology, sociology, economics, management, and even epistemology. The idea behind this spillover is that all events, even making business decisions and evaluating truth claims, can be assigned probabilities. Shorter: it is assumed that society, and perhaps even the whole universe, is a casino – a colossal yet tacit ontological hypothesis. However, probability assignments to human actions are legitimate only in the case of trial-and-error behaviour, which is neither typical nor efficient. For example, a subject asked to guess the sequence of letters RFEZU has only one chance in 5! = 120 of succeeding. In all walks of life we follow rules, imitate, make educated guesses, or conduct research before acting. Trial and error is too irrational, slow, and wasteful to account for human behaviour. Under total uncertainty, that is, total ignorance, the wise thing to do is not to gamble but to refrain from acting on the strength of subjective probabilities. This concludes my defence of the thesis that chance is objective. Let us now examine the opposite opinion.
106
Chasing Reality
5 Chance as Ignorance An alternative to the realist view of chance and probability is the subjectivist doctrine. This is also known as Bayesianism because of its heavy reliance on a certain interpretation of Bayes’s theorem. Bayesianism is the opinion that probabilities are just opinions: that every probability value is a measure of the strength of the belief of a person in a fact or in a statement. (See, e.g., de Finetti 1972, Jeffreys 1975, Keynes 1957, and Savage 1954.) This opinion was held by no less than Jacques and Daniel Bernoulli, Pierre Simon Laplace, Augustus De Morgan, and John Maynard Keynes. More precisely, Bayesianism holds that (a) probabilities are properties of beliefs, propositions, or statements, rather than of facts of a certain kind; and (b) “probability measures the confidence that a particular individual has in the truth of a particular proposition, for example, the proposition that it will rain tomorrow” (Savage 1954: 3). Because of thesis (a), some philosophers have discussed probability with total disregard for the factual sciences of chance, such as statistical mechanics, the quantum theory, and population genetics. And because of thesis (b), Bayesianism is also rightly dubbed the subjectivist or personalist interpretation of probability. And for both reasons, Bayesianism is an alternative to the empiricist (or frequency) theory as well as to the objectivist (or realist or propensity) interpretation. For all its popularity among philosophers, and its throng of faithful among statisticians, Bayesianism is a minority view in the scientific community. Not even the upholders of the Copenhagen (or semi-subjectivist) interpretation of the quantum theory are Bayesians. They deal with facts, not beliefs. As Heisenberg excitedly told a graduate student who had started saying “I believe ...’: “We do not do belief-science here. State your reasons” (Yamazaki 2002: 31). The reason scientists have to avoid Bayesianism is that, because it is subjectivist, it invites arbitrary probability assignments to anything – hardly a scientific procedure. Besides, as Venn (1962) noted more than a century ago, any strong emotion or passion will influence our estimates of the likelihood of events. For instance, we tend to overrate the likelihood of pleasurable events, such as lottery winnings, while underrating the likelihood of disagreeable events, such as traffic fatalities. (More on this in section 5.) A major reason for adopting Bayesianism may be that, because it preserves most of the formulas of the calculus of probability, it confers a veneer of mathematical respectability on such fantasies as the infamous Drake formula for the probability of the occurrence of civilizations in extra-solar planets, as
Causation and Chance: Apparent or Real?
107
the product of a dozen or so “probabilities” of independent events, such as the emergence of planets and that of industrial civilizations. A second reason for the persistence of Bayesianism is that, since the notion of subjective probability is imprecise, it can be used to perform the facile but bogus exactification of some twenty important philosophical notions, such as those of causation, truth, evidential strength, confirmation, plausibility, and simplicity (see, e.g., Good 1967). A third reason for adopting Bayesianism is the mistaken belief that it is the only respectable alternative to the frequency doctrine proposed by such eminent empiricists as John Venn (1962) and Richard von Mises (1926). According to it, probability is definable in terms of long-run relative frequency. This view is intuitively appealing to experimental scientists because it conflates chance (or randomness) with its tests–an instance of the operationist confusions of definition with test, and reference with evidence. Indeed, to determine whether a given set of data is random, one looks for certain fairly constant global properties of the set, particularly the shape and scatter (variance) of a distribution, such as the bell curve. In particular, when observed frequencies are available, one contrasts them with the corresponding calculated probabilities. But one should remember that probabilities (and probability densities) are properties of individuals, whereas the statistical parameters are properties of aggregates. Besides, statistics are restricted to past events, the only ones on which statistical parameters, such as frequencies and variances, can be counted. (More in Bunge 1981a and 1988.) Last, but not least, the frequency theory of probability happens to be mathematically flawed (Ville 1939). So, the frequency theory of probability must be discarded. However, contrary to popular belief, Bayesianism is not the sole alternative to it. Indeed, as suggested in the previous section, there is a tertium quid, namely, the objectivist (or realist or “propensity”) interpretation, the one usually employed in theoretical physics and biology. For example, the probability distributions calculated in classical statistical mechanics are deemed to be objective properties of the systems concerned. Furthermore, those distributions depend on the system’s energy and temperature, neither of which can be estimated by counting frequencies. When such a count is possible, its serves to test probability estimates, not to assign them meanings. Finally, in addition to frequentism and subjectivism, there is probabilistic dualism. This comes in two varieties: (a) probability can be interpreted either as frequency or as credence (e.g., Carnap 1950a); and (b) probability can be interpreted either as “propensity” or as credence (e.g., Sklar 1993, Gillies 2000). However, this won’t do because, as will be seen presently, the subjectivist interpretation is wrong.
108
Chasing Reality
Bayesianism is wrong because it originates in an ontological dogma and two major confusions. The dogma in question is classical determinism or causalism, so brilliantly described by Laplace two centuries ago, and certainly justifiable at his time. This is the belief that everything happens according to laws that, such as Newton’s, have a broad causal domain. (These laws are not strictly causal because they include the self-movement due to inertia: Bunge 1959a.) If this were true, chance would indeed be but a name for our ignorance of causes, so that an omniscient being would be able to dispense with the concept of chance. However, the basic laws of quantum theory and population genetics are probabilistic, and they do not derive from causal laws. Rather, on the contrary, many a macro-law is a law of averages, and it can thus be deduced from probabilistic micro-laws. So much for the error that lies at the root of the subjectivist interpretation of probability. Let us now turn to the accompanying confusions. A first confusion is that between propositions and their referents. Suppose, for example, that V designates a random variable, such as the number of points scored in a die throw. Further, call Pr(V = v) the probability that, on a given occasion, V takes on the particular value v, such as the ace. The proposition “Pr(V = v) = p” involves the proposition “V = v,” but it should not be read as the probability of this proposition, since such expression makes no clear sense. The gambler knows that the proposition “Pr(V = 1) = 1/6” states the probability of the fact of getting an ace when throwing a well-shaken dice cup. He is interested in the outcome of a real process characterized by objective disorder – the one resulting from shaking. Likewise, the quantum physicist who writes a formula of the form “Pr(n ® n') = p,” where n and n' denote two different energy levels of an atom, states that the probability of the quantum jump n ® n' during a certain time interval equals the number p, which is analysable into parameters that describe certain features of the atom and its environment. Indeed, the expression n ® n' describes the objective transition in question, not the corresponding proposition, let alone the associated belief. This is why the scientist checks the formula against measurements on the atom in question rather than polling expert opinions. A second major source of Bayesianism is the confusion of objective chance with subjective uncertainty. This is a conflation between an ontological category and a psychological (and epistemological) one. To be sure, this confusion is rather natural, because indeterminacy implies uncertainty – though not conversely. For example, while one is vigorously shaking a dice cup, every one of the six sides of a die acquires the same chance of coming up when the die is cast. (Shaking is, of course, a randomization mechanism.) However,
Causation and Chance: Apparent or Real?
109
U
P Figure 4.2.MThe uncertainty–probability relation when there are only two possibilities, with probabilities p and 1 – p.
once the die is cast, determinacy has replaced indeterminacy, whereas subjective uncertainty remains as long as we do not look at the die. The Bayesian has no right to say that the probability that he will see an ace is 1/6, because the random process that culminated in this fact is over: alea jacta est. If an ace is what came up, the gambler is allowed to look, and his eyesight is normal, he will see an ace regardless of his expectations. Moreover, the gambler’s mental process is quite different from the random physical process that he triggers when rolling dice; so much so that the gambler who ignores the laws of chance is bound to form irrational expectations, such as the popular gambler’s fallacy (“The next throw must be an ace, since no ace occurred in the last five throws.”) That is, our expectations may not mirror objective chance. If they did, neither casinos nor lotteries would be profitable. The only way to defeat chance is by cheating. 6 Uncertainty Though inadequate for forming rational expectations, the calculus of probability includes an objective measure of uncertainty. This is the variance (or dispersion, or spread) of a distribution such as a scatter plot. The variance, or square of the standard deviation s, is a function of probability. In the simplest case of a binomial distribution, when the random variable assumes only the values 0 or 1 with corresponding probabilities p and 1 – p, the variance is s2 = p – p2. The shape of this curve is an inverted-U around the point p = 1/2; it vanishes at the extremes p = 0 and p = 1, and attains its maximum at the midpoint p = 1/2. This particular maximal uncertainty value is 1/4, whereas the corresponding improbability is 1 – p = 1 – 1/2 = 1/2. (See, e.g., Feller 1968: 230.) See figure 4.2. (There are alternative measures of the objective uncertainty of a distribution. One of them is the Shannon information-theoretic entropy H, which is the sum over – pi log pi . This famous formula, the basis for the definition of a bit, has been overinterpreted as measuring a number of features of systems of many kinds. This is not the place to criticize these extravagances. Here we are
110
Chasing Reality
only interested in the fact that, like the variance, H is a measure of objective uncertainty rather than of subjective or psychological uncertainty. Moreover, this measure is similar to the variance in the case of an experiment with only two possible outcomes, like the one discussed above. Indeed, in this case H = – p log p – (1 – p) log (1 – p), the graph of which is again an inverted U: see, e.g., Yaglom and Yaglom 1983: 49.) In short, given a probability distribution, one can calculate, among other statistics, the objective uncertainty (or indeterminacy) associated with it. This is an objective feature of the given population and the corresponding data, not of anyone’s beliefs. Whether a given human subject will “feel” or intuit the same subjective uncertainty, is a matter for experimental psychologists to find out. In any event, the upshot is clear: Improbability, or 1 – p, is not an adequate measure of uncertainty, whether objective or subjective. We all know that predictions of certain kinds are dicey. However, they should not be cast in probabilistic terms if they concern non-random events, such as collisions between tectonic plates, rainfalls, crops, the onset of sickness, the outcomes of political elections, or Armageddon. True, the weather forecasts reported in the media are usually cast in probabilistic terms. But this practice is wrong, because such forecasts are not calculated with the help of probabilistic meteorology – which so far is only a research project. The meteorological “probabilities” in question are mere likelihoods (in the nontechnical sense of the word), because they are estimated on the strength of weather records, satellite images, and measurements of the velocity of displacements of the major “weather-makers.” (See Bunge 2003a for the differences between likelihood, plausibility, and probability.) As for election outcomes, the reason that they should not be cast in probabilistic terms is that such results depend on the candidates’ track record and promises, as well as on carefully planned, smartly advertised, and wellfinanced campaigns – or even on the complicity of electoral officers and judges. Yet, two prominent academics (Kaplan and Barnett 2003) have proposed a Bayesian model for estimating the probability of winning the United States presidency. It had certainly been noted that the American political process was being privatized, but no one had suggested before that the gaming industry was about to take it over. 7 Bayesianism Is Confused So far I have argued that the idea of assigning probabilities to statements originates in both outdated metaphysics and sheer confusion. However, I have yet to prove that the idea is wrong. There are several reasons why it is mistaken. One of them is that the formula “The probability of statement s
Causation and Chance: Apparent or Real?
111
equals p,” or “Pr(s) = p” for short, does not include a variable denoting a person – which is not surprising since the calculus of probability is not a branch of cognitive psychology. Another objection is that propositions do not pop up or vanish at random. Of course, one could think of picking a proposition at random from a set of propositions, and ask what its probability is. But this would be just a parlor game, unrelated to the scientific search for truth. If the expression in question is intended to mean that the proposition of interest has a truth-value lying between 0 (total falsity) and 1 (complete truth), then this can and must be said clearly, namely thus: ‘The proposition is partially true.’ For example, “p = 3” and “Our planet is spherical” are approximately true propositions, whereas their negations are completely true (Bunge 2003a). However, partial truths are not probabilities. To make this point, the following counterexamples should suffice. First example: The proposition “Aristotle was a Mongolian philosopher” is half-true, since it is true that Aristotle was a philosopher, but false that he was Mongolian. Yet, if partial truth is equated with probability, we must regard it as false, since its probability equals the product of the probabilities of the two constituent propositions, one of which is 1 and the other 0. Second example: Let p be the proposition “p = 1.” Since the relative error incurred in making this statement is approximately 2.14/3.14 = 0.68, the truth-value of p is V(p) = 1 – 0.68 = 0.32. Now, the negation of p, that is, “p ¹ 1,” is completely true. That is, V(not-p) = 1. However, if truth-values were probabilities, we should set Pr(not-p) = 1 – Pr(p) = 1 - 0.32 = 0.68, which is significantly smaller than 1. All the calculi cast in terms of probabilities of propositions, such as those of Reichenbach (1949), Carnap (1950a), Popper (1959b), and their followers, are wrong because (a) the concept of truth is more basic than that of probability, and (b) propositions, unlike random events, cannot be assigned probabilities except arbitrarily. (Further reasons in Bunge 1963b.) 8 Beliefs Are Not Bayesian Whether probabilities measure credences, or degrees of rational belief, is an empirical question. Hence, it cannot be settled a priori – the way subjectivists claim. Let us therefore ask cognitive psychologists whether people actually think in accordance with the said calculus when reasoning about uncertain matters. The many experiments of Daniel Kahneman and his students have conclusively shown that our subjective judgments of likelihood and plausibility (or verisimilitude) are often incorrect, and do not meet the axioms of the probability calculus (Kahneman, Slovic, and Tversky 1982). To begin with, when considering a branching process, such as a decision
112
Chasing Reality
tree, there are rarely enough data to include all the possible forks – a condition for ensuring that the sum of the probabilities over all the branches equals unity. Next, we tend to exaggerate the probabilities of certain unlikely events, such as that of contracting the West Nile virus. Fear, greed, wishful thinking, superstition, strong emotion, and association with pleasurable or painful experiences are among the factors that distort our judgments of the objective likelihood and actual frequency of an event. In short, Perceived likelihood ¹ Objective likelihood. In other words, Probability ¹ Degree of rational belief. Furthermore, beliefs do not satisfy the probability laws. One of these is “Pr(A & B) Pr(A), Pr(B),” where equality holds if A and B are mutually independent. Set A = “Liberty is good” and B = “Equality is good.” Libertarians swear by A, egalitarians by B, and neither by both. In my opinion, neither liberty nor equality is by itself a social good, or even viable, because liberty is impossible among the unequal, and forced equality muffles liberty. But it is arguable that the combination of liberty with equality is both viable and good. So, if we had a reasonable doxastics, or logic of beliefs, I would state the dual of the probabilistic inequality, namely, D(A & B) D(A), D(B), where D would stand for a strength of belief function. In sum, the probability calculus is not a true theory of beliefs. Hence, Bayesianism is not a true account of beliefs. Which is not surprising, because the truth in question is objective, something Bayesians do not much care for. Something similar holds for the view that probability exactifies the vague notion of inductive (or empirical) support. This is essentially the idea that the conditional probability of hypothesis h given evidence e, or Pr(h ) e) for short, must differ from the probability Pr(h) assessed before having produced e. But how do we assess the prior Pr(h) except by fiat, hence how do we compare the prior probability Pr(h) to the posterior probability Pr(h ) e)? And what does “Pr(h)” mean anyway? It cannot mean “plausibility” or “verisimilitude,” because this is relative to context, and it is hardly a numerical concept. Let’s face it: the so-called probabilities of hypotheses are opinions on opinions – doxa squared. Yet, Bayesians claim that they hold the key to understanding the advancement of scientific knowledge, from data to hypotheses and back. For instance, hypothesis h would be confirmed by evidence e if Pr(h ) e) > Pr(h). However, to my knowledge no one has been foolhardy enough to assign probabilities to any scientific laws, such as those of classical, relativistic, or quantum mechanics. Let us contrast the scientific and the Bayesian uses of Bayes’s theorem, the centrepiece of both Bayesian statistics and inductive logic. As will be recalled, that theorem, which is a correct piece of neutral mathematics, reads thus: Pr(A ) B) = Pr(B ) A) Pr(A) / Pr(B). Let A and B denote two states of a concrete
#
$
Causation and Chance: Apparent or Real?
113
system, Pr(B ) A) the probability that, given that the system is in state A, it will go over to state B, and Pr(A ) B) the probability of the reverse process. According to non-Bayesians, at least two conditions must be met to introduce these conditional probabilities into Bayes’s theorem: (a) both processes, A ® B and B ® A, must be really possible, that is, they must be consistent with the pertinent objective laws; and (b) both processes must be random, that is, describable by a probabilistic model. Yet, Bayesians require neither condition. Therefore, as noted in section 3, a Bayesian might be tempted to assign a nonvanishing probability to the impossible transition Dead ® Alive. Furthermore, since the definition of Pr(A ) B) involves Pr(A&B), the Bayesian must admit that it is possible to be alive and dead at the same time. (Incidentally, this counterexample confutes the opinion that the transition probabilities calculated in quantum mechanics and other theories are conditional probabilities.) A third condition for the legitimate application of Bayes’s theorem is that three out of the four probabilities occurring in the theorem be known. When the prior probabilities Pr(A) and Pr(B) are unknown, as is the case when A = hypothesis h, and B = evidence e, writing Pr(h ) e) and Pr(e ) h) in terms of them amounts to scribbling squiggles. And yet this is how Bayes’s theorem is used in both Bayesian statistics and inductive logic. For instance, when estimating the probability of an event, or the plausibility of a proposition, the Bayesian consults a panel of experts. That is, he seeks “a consensus view of informed opinion,” just the way one proceeds in everyday life with regard to everyday matters – with the difference that the Bayesian assigns numbers to strengths of belief (see, e.g., Press 1989). True, the self-styled objectivist Bayesians equate Pr(h ) e) to the corresponding frequency – for example, that a positive clinical test is evidence of a certain sickness; but they make up the other “probabilities,” in particular Pr(h). Besides, in equating certain probabilities with frequencies, they violate the credo that probabilities are credences. A half-Bayesian, half-frequentist, such as Carnap (1950a), Hacking (1990), or Earman (1992), is still a Bayesian, in that he has no qualms about committing the original sin of Bayesianism: assigning probabilities to hypotheses, jury verdicts, and the like. Moreover, he takes the two wrong sides of the controversy over probability. The occurrence of unknown prior probabilities, which must be stipulated arbitrarily, does not worry the Bayesian anymore than God’s inscrutable designs worry the religionist. Thus Lindley (1976), one of the leaders of the Bayesian school, holds that this difficulty has been “grossly exaggerated.” And he adds: “I am often asked if the [Bayesian] method gives the right answer: or, more particularly, how do you know if you have got the right prior [probability]. My reply is that I don’t know what is meant by ‘right’ in this context. The
114
Chasing Reality
Bayesian theory is about coherence, not about right or wrong” (359). Thus the Bayesian, along with the philosopher who only cares about the cogency of arguments, fits in with the reasoning madman. Lastly, from a methodological viewpoint, the entire debate over subjective versus objective probabilities boils down to the following argument, the second premise of which was unwittingly supplied by no less than the statistician (de Finetti 1962: 360) who started the contemporary phase of Bayesianism: If a hypothesis is untestable, it is not scientific. Now, the Bayesians “maintain that a probability evaluation, being but a measure of someone’s belief, is not susceptible of being proved or disproved by the facts.” \ Bayesianism is unscientific. This should settle the question for scientific realists. But these are still a minority among philosophers, so that Bayesianism may survive for a while, along with mind-body dualism, many-worlds metaphysics, and other philosophical eccentricities. Meanwhile let us hope that it will be gradually discontinued in medicine, seismic engineering, policy-making, and other fields where lives are at stake. 9 Bayesianism Is Harzardous Unsurprisingly, Bayesianism can have catastrophic practical consequences. Let us note three examples. The first concerns experimental design, which is crucial in determining the efficacy and safety of drugs, agricultural techniques, social programs, and more. Since about 1930, it has been standard practise in the biosciences and the social sciences to randomize both the experimental and the control groups (see Fisher 1951). Now, Bayesians do not practise randomization because they have no use for the concept of objective randomness: for them, chance is only in the eyes of the beholder. Hence, if the U.S. Food and Drug Administration were ever to be dominated by Bayesians, it would become a public peril rather than a guardian of public health. Another example of the high risk incurred by Bayesians is the “probabilistic” risk assessment used by the NASA managers to estimate the risk of manned space flights in the cases of the ill-fated space shuttles Challenger and Columbia. I submit that juggling with probabilities is inappropriate in the case of the failure of components of a machine, because this is a causal chain, not a random one – such as a Markov chain, every link of which has an objective probability that depends only upon the preceding link. In general, disasters – from computer crash to early death, from bankruptcy to biospecies extinction, from hurricane to earthquake, and from epidemics to
Causation and Chance: Apparent or Real?
115
war – are nonrandom events, hence unaccountable in probabilistic terms. True, the extinction of species and the decline of ecosystems have often been modelled assuming that the events in question are random. But they are not, for actually the dominant ecological variables are nonrandom: think of rainfall, species rarity, body size, and the presence or absence of carnivores, keystone species (like starfish), and of aggressive invaders (such as elephant grass). Recent experiments have shown that ecosystems decline faster than expected if they happened by chance (Raffaelli 2004). The reason is that, when removing species (or rather populations) at random, one treats them all equally, while actually some are more important than others. In any event, the experiments in question have falsified the probabilistic models of ecosystem sustainability. Obviously, this result should have dramatic consequences for environmental policies as well as for theoretical ecology. The moral is that probability without randomness can be deleterious to the environment. Our third and final example will be the Bayesian approach to medical diagnosis and prognosis, adopted by Wulff (1981) and many others. Since we know that the disease-symptom association is causal rather than random, there is no reason to expect that their probabilities exist and are related the way Bayes’s theorem stipulates. Unsurprisingly, the available statistics concerning cancer detection fail to satisfy Bayes’s formula (Eddy 1982). In sum, Bayesian medicine is unrealistic, and therefore unreliable, because it does not match the actual diagnostic process, which involves plausibility judgments based on anatomy and physiology, and to a lesser extent also on epidemiology. Much the same holds for the medical application of Popper’s methodology: it, too, is unrealistic though for different reasons, namely, because (a) in medicine the only conjectures that researchers attempt to falsify are the lowly null hypotheses – which Popper and his followers have ignored – rather than high-level theories, so far non-existent in medicine; and (b) biomedical researchers pay close attention to epidemiological data and statistical inference, neither of which is countenanced by the Popperians (see Murphy 1997). The unrealistic nature of the Bayesian approach is of course part of its subjectivism. But Bayesians believe that this is a shining virtue rather than a fatal flaw. Thus, Howson and Urbach (1989: 288) state that “[s]cience is objective to the extent that the procedures of inference in science are. But if those procedures reflect purely personal beliefs to a greater or lesser extent, ... then the inductive conclusions thus generated will also reflect those purely personal opinions.” This statement contains three elementary mistakes. The first is the quaint idea that objectivity resides in inference rather than in
116
Chasing Reality
reference. (Inferences are valid or invalid regardless of reference and even truth. This is why formal logic, the theory of inference, does not contain the semantic concept of reference.) The second mistake of Howson and Urbach is the Bayesian dogma that “purely personal beliefs” have a scientific standing. If they had, religious beliefs would play a role in the finished products of scientific research. The third mistake is the statement that all scientific inferences are inductive, when in fact induction plays an insignificant role in advanced science before one reaches the stage of contrasting theoretical predictions against empirical data (Popper 1959b, Bunge 1960). No wonder that Howson and Urbach do not examine any of the scientific theories about chance events, such as quantum mechanics and genetics. Likewise, Berry and Stangl (1996: 8) write about Bayesianism in biostatistics: “although it is comforting when two [statistical] analysts give the same answer, subjectivity leading to diversity is quite appropriate. Differences of opinion are the norm in health science and in science generally, so an approach that explicitly recognizes differences openly and honestly is realistic, forthright, and welcome.” An obvious rejoinder to that extraordinary statement is this. First, biomedical models are expected to tell us something about diseases, not about the opinions that experts hold about them. Second, science and technology, including medicine, are not mere matters of subjective assessment or opinion. True, the sick Sumerians would exhibit themselves in front of their houses and solicit the opinion of the passerbys. But one millennium later Hippocrates started transforming medical doxa (opinion) into medical episteme (science). Can we afford to go back four millennia? It might be rejoined that any ideas can have bad consequences if handled incompetently. My point is that there is no way that Bayesianism (or alchemy, or parapsychology) can be handled competently, because it is radically false. Indeed, (a) probability estimates should be just as objective as length or weight estimates; and (b) probability applies legitimately only to genuine random events, such as quantum tunnelling, the drips of a leaky water tap, gene mutation, the distribution of weeds in a cultivated plot, the arrival of calls at a telephone exchange, and random sampling. To appreciate the enormity of the Bayesian attempted counter-revolution, consider, for instance, the relation between the HIV virus and AIDS. It is well known that, whereas those who have this disease test HIV-positive, the converse is not true: some individuals have lived with the virus for a decade or more without developing AIDS. Suppose then that a given individual has
Causation and Chance: Apparent or Real?
117
contracted that virus, and that we wish to ascertain the probability that he also has, or will soon develop, AIDS. Presumably, a Bayesian would set Pr(AIDS HIV) = Pr(HIV AIDS) Pr(AIDS) / Pr(HIV). Further, since the individual in question has tested HIV-positive, our Bayesian is likely to set Pr(HIV) = 1. And, since it is known that whoever has AIDS also has the HIV virus, Pr(HIV AIDS) = 1. Thus, Bayes’s formula simplifies to Pr(AIDS HIV) = Pr(AIDS). However, this is known to be false: in fact, HIV is necessary but not sufficient to develop AIDS. So, if AIDS researchers were to adopt Bayesianism, they would not try to discover the causes that, jointly with HIV infection, lead to AIDS. Talk of probability without chance is pseudoscientific and therefore imprudent, particularly if the alleged probabilities are subjective, in which case they are actually intuitive plausibilities or verisimilitudes. One should not gamble with life, peace, or truth. And one should not confuse the objective probabilities of random events with mere judgments of the likelihood of such events or the plausibility (or verisimilitude) of the corresponding hypotheses (Bunge 2003a). As Peirce (1935: 363) put it, this confusion “is a fertile source of waste of time and energy.” A clear case of such waste is the current proliferation of rational-choice theories in the social sciences, to model processes that are far from random, from marriage to crime to business transactions to political struggles (see Bunge 1996, 1998, and 1999). In short, Bayesian statistics and inductive logic are doubly wrong: because they assign probabilities to statements, and because they conceive of probabilities as subjective. Popper (1959b) was right in indicting inductive logic. Regrettably, he too assigned probabilities to propositions when proposing his various theories of corroboration and verisimilitude. (Popper also erred in asserting that all the probabilistic hypotheses are unfalsifiable. The probabilistic hypotheses of quantum mechanics and genetics are perfectly testable through their consequences for large masses of mutually independent events of the same kind, such as the typical radiative decays of zillions of sodium atoms, which we see as yellow light.) Another mistake that Popper shared with the Bayesians is that of assessing hypotheses one by one, regardless of the conceptual systems (theories, classifications, and even worldviews) to which they belong. (This was due partly to his conflating the concepts of theory and hypothesis, and partly to his severing the ties of science with the rest of culture.) As a matter of fact, scientists prefer the hypotheses that are not only the better corroborated, but also best jibe with other well-confirmed hypotheses, and even with the ruling worldview (Bunge 1959b and 1967a, Thagard 1978). )
)
)
)
118
Chasing Reality
10 Concluding Remarks Causation and chance are not just in the mind: they are in the world as well. That is, some real processes are causal, others random, and still others have causal as well as stochastic aspects. That is, the fields of causation and chance have a partial overlap. Moreover, causation on one level may emerge from chance on another, and conversely. Furthermore, causation and chance may combine with one another on the same level, as when a causal process randomizes a collection of items, or when a random process triggers a causal chain. Besides, the expressions of the forms “X is a causal process” and “Y is a random process” should be understood as abbreviations of detailed descriptions rather than as exhaustive accounts. After all, the concepts of causation and chance are so general that they are philosophical. Note also that, though vast, the field of causation is limited to events of certain kinds. It does not include spontaneous events, such as the jump of an isolated atom from an energy level to a lower one. Nor does it include nonevents, such as the permanence of a boulder on the ground: the mutual gravitational attraction of boulder and Earth does not cause the acceleration of either. The causal relation obtains only between events: No change, no causation. Chance is just as real as causation; both are modes of becoming. The way to model a random process is to enrich the mathematical theory of probability with a model of a random mechanism. In the sciences, probabilities are never made up or “elicited” by observing the choices people make, or the bets they are willing to place. The reason is that, in science and technology, interpreted probability exactifies objective chance, not gut feeling or intuition. No randomness, no probability. In short, through finding causal and probabilistic patterns, science and technology break the phenomenalist ban on unobservables. But of course they meet the scientistic imperatives to quest, guess, and test in all human endeavours.
Behind Screens: Mechanisms
119
5 Behind Screens: Mechanisms
Scientists and technologists endeavour to find out how things work, that is, what their mechanisms or modi operandi are. This is how they advance from appearance to reality, and from description to explanation. By contrast, the superstitious do not look for mechanisms. For example, some parapsychologists believe in the possibility of moving things by sheer mental power (psychokinesis). If they were to inquire into the way psychokinesis works, or rather does not, they would realize that it is impossible if only because it involves the creation of energy. A similar reasoning is used in evaluating inventions: no patent is ever granted unless the inventor succeeds in explaining how the novel device works. This is why the most effective way for a patent office to deny a patent for an allegedly revolutionary design is to point out that the proposed mechanism is incompatible with a well-known law, such as the law of conservation of energy. No law, no possible mechanism; and no mechanism, no explanation. No wonder then that the hallmark of modern science is the search for mechanisms behind facts, rather than the mindless search for data and statistical correlations among them. To elucidate these ideas, and motivate the subsequent discussion, let us begin by considering a handful of examples drawn from several fields. 1 A Handful of Examples Here is an example from elementary physics: Ohm and Kirchhoff described electric circuits, but did not know what makes electric charges move around them. This explanation was only provided by electrodynamics: the electric charges (electrons) in a metallic wire are dragged by the impressed electric field (voltage); and in turn the electric field that accompanies the electrons generates a magnetic field, which in turn opposes the electric current.
120
Chasing Reality
Our second example is taken from chemistry. There are several mechanisms for the synthesis of molecules out of atoms. The most prevalent of them are electron transfer and electron sharing. In the first case one of the atoms donates an electron to the other, as a consequence of which a positive and a negative ion are formed, which attract one another electrostatically. A textbook example is the combination of a positive sodium ion Na+ with a negative chlorine ion Cl–, to produce a sodium chloride molecule NaCl. By contrast to this (electrovalent bond), the covalent bond emerges when the precursor atoms share their outer electrons. The simplest example is the formation of the hydrogen molecule H2. This is not just the juxtaposition of two hydrogen atoms, since the two electrons from the precursor atoms now interpose between the atomic nuclei (protons). Here emergence results from restructuration. Another clear example is that of catalysis: theoretical chemists want to know how catalysts, such as enzymes, work. They have known for decades that a catalyst does not work by its mere presence, as it had been believed in earlier times. A catalyst works by combining with the substratum into a short-lived complex that subsequently dissociates into the product and the catalyst. (The reaction schema is S + C SC C + P.) But in turn why does that complex form, and why is it unstable? These problems are still under investigation. Let us now jump to the next level, that of organisms. Biological evolution had been suspected long before Charles Darwin established it. His great merit is that he explained evolution in terms of two mechanisms: inheritance with modification, and natural selection. Whereas he succeeded in describing the latter, the mechanism of the inborn modifications was discovered only many decades later. It turned out that they arise from genic mutations and recombinations, which are in turn explained in molecular terms. Focus on genes led Theodosius Dobzhansky to define evolution as a change in the frequencies of certain alleles in a population. But a frequency change – a statistical feature of a collection – is only an effect of the alterations occurring in the course of individual development. And these are changes in developmental pathways or mechanisms. Moreover, these are the roots of speciation, as recognized by the new science of evolutionary developmental biology, or evo-devo (see, e.g., Wilkins 2002). That is not all. To explain evolution we must add environmental factors, such as temperature, humidity, sunshine, soil acidity, and above all interactions with other organisms, for any of these factors may influence development. Such factors explain why, contrary to an earlier belief, the genetic composition of a biopopulation may show drastic changes in a short period rather than in
® ®
Behind Screens: Mechanisms
121
“evolutionary time.” A recent example is the overfishing of cod, which has all but eliminated the fish that grow before attaining sexual maturity: in a few decades, commercial fishing has dramatically skewed the genetic pool of the cod population. This discovery suggests the need for the merger of population genetics with ecology (Roughgarden 1979). However, organisms are not passive toys of their environment: they construct their own habitats, so that biological inheritance entwines with ecological inheritance (Odling-Smee et al. 2003), In short, bioevolution is explained by various concurrent mechanisms on several levels, from molecule to cell to organ to organism to biopopulation to ecosystem (Mahner and Bunge 1997, Gould 2002). An interesting psychological example is the mechanism of the extinction of aversive memories, such as those of fearful episodes. This is not accomplished by Freud’s mythical immaterial superego, but by the cannabioids produced by our own bodies: those molecules wreck the neuronal processes in the amygdala that store aversive memories (Marsicano et al. 2002). Another recent spectacular finding in cognitive neuroscience is the experimental induction of “out-ofbody” experiences (Blanke et al. 2002). This is achieved, not by extrasensory means or by mystical meditation, but by electrical stimulation of the somatosensory region (body map) of the cerebral cortex. Although the details of the mechanism are still hazy, the outline is clear. Whether normal or illusory, perception is a process localized in the cortex, and thus sensitive not only to external signals but also to internal stimuli. Thus, the alleged paranormal is explained by the normal. Let us finally glance at social science. In a classic paper, Robert K. Merton (1936) identified the mechanisms of unanticipated purposive social actions. One of them, perhaps the most pervasive, is this: “[W]ith the complex interaction that constitutes society, action ramifies. Its consequences are not restricted to the specific area in which they are intended to center and occur in interrelated fields ignored at the time of action.” The neoclassical economists, by contrast, ignore the very concept of a system, and consequently fail to explain the functions and dysfunctions of the economy. For instance, Ludwig von Mises (1966: 257), an influential neoconservative economist, claimed that the market is not a thing or a collective entity, but a process. But of course there are no processes except in concrete things: recall chapter 1. And a market is a thing, in particular a concrete system, whose central mechanism and raison d’être is the exchange of goods and services. The actions of central banks and stabilization funds, monopolies and oligopolies, as well as the implementation of commercial codes and government
122
Chasing Reality
regulations – though not the norms in themselves – may in turn be regarded as mechanisms for the control of the trade mechanism. They may therefore be called metamechanisms. So is the application of any free-trade agreement among unequal nations: it forces the weaker party to grant national treatment to foreign firms, hence to abstain from favouring national development – while at the same time the very same champions of free trade practise protectionism at home. Unfettered free trade is thus a mechanism for strengthening the stronger and thus perpetuating underdevelopment. Standard microeconomic theory focuses on market equilibrium, which it purports to explain as an outcome of the mechanism: Increase in supply Drop in price Rise in demand Price hike ... Regrettably, this zigzagging and self-correcting mechanism works better in textbooks than in real markets, for it ignores oligopolies and state subsidies, and it overlooks persistent market disequilibria, in particular gluts, chronic unemployment, and the irregular and sometimes enormous fluctuations of financial markets. (See Bunge 1998 for further criticisms of standard economic theory.) The economic sociologists have been baffled by the steady rise of income inequality in the United States and other countries since about 1980, despite spectacular gains in both productivity and gross domestic product – the socalled Great U-Turn. Several mechanisms, operating concurrently, have been proposed to explain this trend, notably the following ones, which are involved in globalization (Anderson and Nielsen 2002): the deindustrialization caused by the export of manufactures; the cheapening of low-skill labour; and the weakening of the bargaining power of labour consequent upon both antilabour legislation and the increased labour supply. Thus, as Hobson (1938 [1902]) observed a century ago with regard to the British empire, economic imperialism is self-destructing in the long run. Here is a clear politological example: “Democracy is a social mechanism for resolving the problem of societal decision-making among conflicting interest groups with minimum force and maximum consensus” (Lipset 1959: 92). Of course, democracy is also the mechanism allowing for political participation and civic responsibility. By contrast, dictatorship, unprovoked military aggression, and state terrorism are by far the most destructive, divisive, and irrational, and therefore also the most barbaric and immoral, of all the social mechanisms. Social mechanisms have two peculiarities: they are purposeful, and they are linked. For example, democracy may be regarded as a mechanism for favouring participation; the latter is a mechanism for reinforcing social cohesion, which favours stability, which reinforces democracy. The four mechanisms are thus linked in a self-sustaining causal chain – and so are their duals. See figure 5.1.
®
®
®
®
Behind Screens: Mechanisms Democracy
Participation
Stability
Cohesion
Dictatorship
Instability
123
Marginality
Disintegration
(a)
(b)
Figure 5.1MFeedback loops involving four social mechanisms: (a) constructive and (b) destructive.
The moral should be clear: Whereas the constructive mechanisms should be oiled and repaired whenever necessary, the purely destructive ones should be thwarted. For example, the mechanism of grass-roots terrorism is the selfsustaining cycle: Provocation Terrorist attack Retaliation Provocation Terrorist attack. We won’t understand what maintains such an irrational cycle if we keep repeating the conventional explanation, that terrorism results from a combination of poverty with ignorance. Social psychologists and sociologists have found that most kamikaze pilots and suicide bombers were well-educated members of the middle class. Only further investigation can disclose the grievances, frustrations, and religious superstitions that led them to seek martyrdom (Atran 2003). The ability to tamper with mechanisms is particularly important in technology. For instance, immunologists and epidemiologists know that vaccination protects against infectious diseases because it induces the synthesis of the appropriate antibodies. And normative epidemiologists can calculate the minimum level of vaccination that will prevent an epidemic because they know that, above a certain threshold, infection will propagate through contact until all but the naturally immune individuals have been struck. Thus, the prevention and management of infectious diseases rests on a body of theoretical epidemiology centred on the infection mechanisms, rather than on dubious inductive inferences from observed samples (see, e.g., Anderson and May 1991). In general, statistical correlation explains nothing: it is what cries for explanatory models, whether deterministic, stochastic, or mixed. And efficient control is based on knowledge of dynamics. Most epidemiological generalizations and correlations that are not backed up by solid hypotheses about mechanisms at best stimulate research, and at worse are unjustified alarm calls. For example, a strong positive or negative correlation between two features X and Y only indicates covariation, as in the case of stature and body weight. Only further research might establish that changes in X cause changes in Y or conversely, or that both depend on the changes of a third variable Z. The moral is obvious: Look for mechanisms
®
®
®
124
Chasing Reality
behind every interesting strong correlation. For example, the U-shaped relation between morbidity and body mass index (BMI) is an indicator of nutrition habits, in particular both malnutrition (for small values of BMI) and excessive food intake (for large values of BMI). A technology can be effective but not efficient: that is, it may work but only at a high cost. The classical example is Roman engineering, characterized by reliability and durability ensured by excessive bulk. Efficient technology rests on well-confirmed mechanismic hypotheses. By contrast, the inefficiency of pseudotechnology is due to the non-existence or ignorance of the relevant mechanisms. For example, it is known that magic, intercessory prayer, water dowsing, Feng Shui, homeopathy, faith healing, and psychoanalysis fail (see, e.g., Kurtz, ed. 2001). The reason is that they do not rely on any real mechanisms other than the placebo effect, that is, plain suggestion. Incidentally, we still do not know the neurocognitive mechanism whereby suggestion works. However, there are good reasons to suspect that words and placebos are effective to the extent that they trigger physiological processes that start in the cortex and are likely to involve the release or the blocking of neurotransmitters that activate or inhibit the formation of neuron assemblies. The relevance of mechanism to understanding and control is such that it is not uncommon to find in the scientific literature apologies of the form “Unfortunately, no mechanism is known to underly the fact [or the equation] in question.” For example, nobody believed in allergies until it was explained in terms of antigen-antibody reactions. Physicians do not believe in homeopathic nostrums, except as placebos, because there is no mechanism whereby a few molecules of any “active principle” could affect entire organs. And scientific psychiatrists do not believe in psychoanalytic stories, not only because they lack experimental validation, but also because they are not backed up by any known brain mechanisms. But at least we know about sex what Freud did not, namely, that the most important sex organ is the brain. In all of the above scientific examples, a mechanism was conceived of as a process (or sequence of states, or pathway) in a concrete system, natural or social. Besides, most mechanisms are concealed, so that they have got to be conjectured. This suggests the plan of this chapter: system, mechanism, mechanism guessing, and explanation. These and other concepts will only be sketched and exemplified here; they are elucidated in detail elsewhere (Bunge 1964, 1967a, 1968b, 1979a, 1996, 1997, 1998, 1999). 2 System and Systemism The Baron d’Holbach (1773), one of the major Encyclopedists, should be credited with having joined materialism with systemism. Indeed, he seems to
Behind Screens: Mechanisms
125
have been the first to state not only that all existents are material, but also that everything, both in nature and in society, hangs together. This ontology, systemic materialism, prospered in the natural sciences, all of which study material systems, whether tangible like nervous systems, or intangible like molecules and social systems. Systemic materialism has been far less popular in the field of social studies, where individualism and idealism have ruled. Still, some social scientists have realized that what they study are social systems. Thus the greatest economist of the twentieth century: “I am chiefly concerned with the behaviour of the economic system, as a whole” (Keynes 1973: xxxii); likewise Wassily Leontief, whose input-output matrices concern national economies. Kenneth Boulding (1964) preached the systemist gospel in the desert of mainstream microeconomics. Fernand Braudel’s most famous work (1972) concerns no less than the entire Mediterranean basin. And Henry Ford asserted that mass production, unlike craftmanship, is ruled by three maxims: “System, system, and more system.” However, most social scientists abstain from using the expression ‘social system’ even while studying social systems or designing policies to alter them. In short, systemic materialism has done very well in the natural sciences, and marginally well in the social and biosocial ones. On the other hand, it has not prospered at all in the humanities. Along with scientism, another candle of the Enlightenment, it was snuffed out by philosophy professors in the nineteenth century, and has remained in the hands of amateurs. What prevailed in academia were Hegel’s idealist holism and the idealist individualism of Dilthey and Rickert (Max Weber’s philosophical mentor). The Counter-Enlightenment has triumphed to such an extent that the ideas of the Encyclopedists are hardly taught in our universities. True, a few anthropologists and sociologists, in particular Radcliffe-Brown, Parsons, Luhmann, and Habermas, have written extensively about social systems. However, some of them, particularly Parsons (1951) and his follower Luhmann (1984), emphasized equilibrium – just like the neoclassical microeconomists. To them, a social system is basically in a steady state, and all change is either a departure from stasis or a return to it – as if change were abnormal rather than the rule. Besides, the static systemists regarded social systems as disembodied bundles of values, norms, actions, or communications: unlike Holbach, they were systemists but not materialists. A few other scholars, such as Giddens (1984), have conflated ‘system’ with ‘structure’ – as if a structure could exist independently of the entities that it binds. Still others, particularly systems engineers such as Ashby (1963), have called ‘system’ any black box with inputs and outputs, or even just a list of variables, with disregard for stuff, structure, and mechanism. In other words, they have adopted a functionalist point of view.
126
Chasing Reality
Because the word ‘system’ is used somewhat loosely in the social sciences, it will be convenient to adopt a definition of it. I use the following one: A system is a complex object whose parts or components are held together by bonds of some kind. These bonds are logical in the case of a conceptual system, such as a theory; and they are material in the case of a concrete system, such as an atom, cell, immune system, family, or hospital. The collection of all such relations among a system’s constituents is its structure (or organization, or architecture). This concept of a structure is borrowed from mathematics. A large number of alternative notions, none of them clear, were introduced in the 1950s, when the word ‘structure’ suddenly became fashionable in the humanities and social studies (see, e.g., Centre International de Synthèse 1957). Depending on the system’s constituents and the bonds among them, a concrete or material system may belong to any of the following levels: physical, chemical, biological, social, or technological. The semiotic systems, such as languages and diagrams, are hybrid, for they are composed of material signs or signals, some of which convey semantic meanings to their potential users. The simplest sketch or model of a material system s is the list of its composition, environment, structure, and mechanism(s), or µ(s) =. Here, C(s) denotes the set of parts of s; E(s) the collection of environmental items that act on s or are acted upon by s; S(s) the structure, or set of bonds or ties that hold the components of s together, as well as those that link it to its environment; and M(s) the mechanisms, or characteristic processes, that make s what it is and the peculiar ways it changes. Note that we distinguish a system s from its model(s) µ(s), just as the electrician distinguishes an electric circuit from its diagram(s). Note also that, contrary to Glennan (2002), we distinguish a system from its mechanism(s). All four components of the model µ(s) are taken on a given level, such as the person, the household, or the firm in the case of social systems. They are also taken at a given time. In particular, M(s) is a snapshot of the processes that go on in the system in question. In turn, a process is a sequence of states; if preferred, it is a string of events. And, whereas the net effect of some processes is to alter the overall state of the system, that of others is to maintain such state. For instance, a sailboat is moved by wind, whereas the impacts of myriad water molecules on the hull keep it afloat. And what keeps a business firm above the market water is the sale of its products. The most familiar example of a social system is the traditional nuclear
Behind Screens: Mechanisms
127
family. Its components are the parents and the children; the relevant environment is the immediate physical environment, the neighbourhood, and the workplace; the structure is made up of such biological and psychological bonds as love, sharing, and relations with others; and the mechanism consists essentially of domestic chores, marital encounters of various kinds, and child rearing. If the central mechanism breaks down, so does the system as a whole. The neoclassical economists, obsessed like shopkeepers by price competition, failed to grasp the central mechanism of the capitalist economy: innovation. Schumpeter (1950: 83) unveiled it in a single magisterial page: he saw that what “sets and keeps the capitalist engine in motion” is nearly incessant “creative destruction.” This is the introduction of qualitatively new consumer goods, new methods of production and transportation, new types of organization, and so forth – and the concomitant destruction of their precursors. This is what he called an “organic process,” that is, one that affects the entire economic system. It also has political and cultural repercussions, as when business captures political parties, and when the great literary, musical, and plastic-arts classics are displaced by mass-produced pseudo-artistic merchandise. Nor is creative destruction limited to material products; it can also affect dreams and myths. One example is the so-called New Economy of the mid1990s, centred on the illusion that e-commerce would replace snail-commerce. A related illusion was the Nasdaq bubble, favoured by the artificially low discount rate decreed by the U.S. Federal Reserve Bank – that alleged bulwark of free enterprise; the same government agency also helped form huge and vulnerable industrial conglomerates, some of which declined precipitously, while others existed only on paper. Both bubbles were punctured at the dawn of the new millennium. They might not have been formed if a materialist ontology had prevailed – that is, if it had been realized that paper bulls do not charge. Another topical subject is terrorism. Regrettably, knowledge of organized violence is poor, and as a result its prevention is inefficient. In particular, the most popular view of grass-roots political terrorism is that it is incited by some fanatical, perverse, or demented individuals. While some terrorist leaders do meet this description, it does not explain either the devotion and abnegation of numerous terrorist foot-soldiers or the persistence of their causes. In any event, a successful “war” (or rather mobilization) against group-sponsored terrorism must start by understanding it as a low-budget war from below. The first thing to understand about terrorism is that it comes in two main kinds – state-instigated and group-sponsored – and that the former is by far the more immoral and lethal of the two. State terrorism is the easier to explain because it has a single source, namely, the ruling elite; and a single goal,
128
Chasing Reality
namely, the suppression of dissent. By contrast, group-sponsored terrorism usually attracts people from different walks of life, and is a mechanism of the weak for redressing at once grievances of various kinds: economic (natural resources or jobs), political (social order), and cultural (in particular religious). Any anti-terrorist campaign that does nothing to meet genuine grievances is bound to work, at best, only in the short term and at the cost of civil liberties. In general, systemic issues call for systemic and long-term solutions, not sectoral and near-sighted measures. In particular, social violence stems from us/them barriers. Hence, it is best tackled by removing the barriers in question. For example, the Arab/Jewish confrontation is likely to be solved only by mixing both populations rather than by strengthening the barriers between them (Joseph Agassi, personal communication). This is a practical message of systemism. The twin concepts of system and mechanism are so central in modern science, whether natural, social, or biosocial, that their use has spawned a whole ontology, which I have call systemism (Bunge 1979a, 2003a). According to this view, every thing in the universe is, was, or will be a system or a component of one. For instance, the electron that has just been knocked off an atom on the tip of my nose is about to be captured by a molecule in the air. And the prisoner who just escaped from the county jail is about to be either recaptured or absorbed by a family or a gang. There are no permanent strays or isolates. Systemism is the alternative to both individualism and holism (Bunge 1979a and 1979b, Sztompka 1979). Presumably, it is the alternative that the historical sociologist Norbert Elias (2000) was looking for in the late 1930s, when he felt dissatisfied with the conceptions of the person as the self-contained homo clausus, and of society as a black box above individuals. Arguably, systemism is the approach adopted by anyone who endeavours to explain the formation, maintenance, repair, or dismantling of a concrete complex thing of any kind. Notice that I use the expression ‘systemic approach,’ not ‘systems theory.’ There are two reasons for this. One is that there are nearly as many systems theories as systems theorists. The other is that the ‘systems theory’ that became popular in the 1970s (e.g., Laszlo 1972) was another name for the old holism, and got discredited because it claimed to solve all particular problems without empirical research or serious theorizing. However, just as Monsieur Jourdain spoke prose without knowing it, so most scientists practise systemism without mentioning it. Thus, in his highly original and monumental work on early civilizations, Bruce Trigger (2003a) studies the physical environment, economy, polity, and culture (in the sociological sense of this word) of each of the seven early civilizations he examines.
Behind Screens: Mechanisms
129
Moreover, he stresses the interdependence of those various components or aspects of society, and describes the mechanisms that gave rise to the state and its functions, such as social control, public works, and state-sponsored cultural activities. Trigger also adopts explicitly a materialist ontology, though certainly not the vulgar materialism that denies the importance of ideas and rituals (Trigger 2003b). That is, Trigger has successfully adopted the materialist systemic approach, just like Fernand Braudel and his colleagues in the Annales. However, one will look in vain for the expressions ‘social system,’ ‘social mechanism,’ and ‘systemic approach’ in that work. Likewise, the ancient Egyptians, though deeply religious, had no word for religion. All of which shows once more, against Wittgenstein and his school, that ideas can thrive without the corresponding words. Systemism is just as comprehensive as holism but, unlike the latter, it invites us to analyse wholes into their constituents, and consequently rejects the intuitionist epistemology inherent in holism. For example, whereas holistic medicine claims to treat patients as wholes, without regard for the specificity of their subsystems, scientific medicine treats patients as supersystems composed of several interdependent systems. Likewise, whereas revolutionaries advocate total and instant changes of society as a whole, systemic social reformers favour gradual reforms of all the ailing subsystems of society. The systemic approach advocated here is not a theory to replace other theories. It is, instead, a viewpoint or strategy for designing research projects whose aim is to discover some of the features of systems of a particular kind. Although this approach is routinely used in science and technology, it is part of philosophy, and the latter is not equipped to tackle empirical problems. Philosophy can facilitate or block scientific research, but it cannot replace it. 3 Mechanism As stated above, mechanisms are processes in concrete (material) systems, whether physical, social, technical, or of some other kind. Biochemical pathways, electrical and chemical signals along neural networks, sexual competition, division of labour, publicity, polls, and military expeditions are mechanisms. By contrast, the conceptual and the semiotic systems, such as theories and languages respectively, have compositions, environments, and structures, but no mechanisms. The reason is that changeability (or energy) is the defining property of matter – whether physical, chemical, living, social, or technical. To coin an ambiguous slogan: To be (material) is to become. The ontological concepts of system and mechanism are central to modern science and technology. Even when they study elementary particles, they
130
Chasing Reality
inquire into the systems and mechanisms in which they become involved. The following examples should suggest that the same holds for the social sciences. Our first example is Kondratieff’s “long waves” of economic activity, which have drawn the attention of economists and historians from Schumpeter, Kuznets, and Braudel onwards. However, the very existence of such decadeslong cycles has been questioned for three quarters of a century, because it is not clear what their underlying mechanism might be. Still, one plausible mechanism hypothesis is this: Obsolescence of the dominant techno-economic system New techno-economic system & Social changes Market saturation Drop in prices (Berry, Kim, and Kim 1993). A topical example from politology is this: The flaws of American democracy, such as the high cost of running for political office, and the concomitant intimate relationship between political parties and corporations, are turning young people away from politics. In turn, this voluntary disempowerment is one of the imperfections of that democracy, and it erodes even further political participation, which is the main democratic mechanism. (Concurrent mechanisms are the application of the rule of law, education, voting, honest votecounting, and the free formation and circulation of information conveying reliable knowledge.) This is a case of feed-forward (self-amplifying) control. And it explains why political apathy leads to bad government. Indeed, when the competent and honest citizens tend to stay away from politics, the incompetent and dishonest take over – which may be called Gresham’s Law of Politics. Finally, an example from culturology. The cultural poverty of contemporary Islam, with its nearly total absence of original science, technology, philosophy, and art, stands in stark contrast to the brilliance of its culture in the Middle Ages. This fact is an aspect of a multifaceted stationary process. While today’s Islamic societies – particularly those rich in the Devil’s juice – have imported some of the trappings of modern industry, such as cars and cell phones, most of them have kept a traditional social structure. Indeed, they have discouraged or even banned the quest for novelty – economic, political, and cultural – which is precisely the quest that built modern capitalism and keeps it going. The highly complex systems, such as living cells and schools, have various concurrent mechanisms. That is, they undergo several more or less intertwined processes at the same time and on different levels. For example, a cell does not cease to metabolize during the process of division; a waking brain engages in a number of parallel processes – biochemical, vascular, cognitive, emotional, and motor-controlling; and the people who compose a school metabolize and socialize at the same time that they learn, teach, manage, or gossip. The coexistence of parallel mechanisms is particularly noticeable in biologi-
®
®
®
Behind Screens: Mechanisms
131
cal and social systems. For example, bioevolution is driven by several intertwined mechanisms, among them genetic change, natural selection, and niche construction. Human social evolution is driven by production and exchange, cooperation and competition, and much more. Think, for instance, of the mechanisms operating in a scientific community: research (the truth-finding mechanism), criticism and peer review (social quality-control mechanisms), and a combination of cooperation in the search for truth with competition in the allocation of credits and resources. Because a number of mechanisms may operate in parallel in one and the same system, and because some of them may interfere with one another, it is convenient to distinguish essential from non-essential mechanisms (see Schönwandt 2002). The former are those peculiar to the systems of a certain kind, whereas the latter may also occur in systems of a different kind. For example, contraction is essential to a muscle but inessential to a cell; and loaning money is essential to a bank but optional to a manufacturer. We are now ready to propose and refine this definition: An essential mechanism of a system is its peculiar functioning or activity. In other words, an essential mechanism is the specific function of a system – that is, the process that only it and its kind can undergo. More precisely, we propose the following stipulations, based on the concept of a specific function defined elsewhere (Bunge 1979a). Definition 1 If s denotes a system of kind S, then (1) the totality of processes (or functions) in s over the period T is p(s) = The ordered sequence of states of s over T; (2) the essential mechanism (or specific function) of s over the period T, that is, M(s) = ps(s) Í p(s), is the totality of processes that occur exclusively in s and its conspecifics during T. Definition 2 A social mechanism is a mechanism of a social system or part of it. Note that the concepts of goal and utility are absent from the above definitions. The reason is of course that some mechanisms are ambivalent, and others have unintended negative consequences. For example, free trade may make or undo a nation, depending on its competitiveness, aggressiveness, and political (in particular military) power. However, the concepts of goal and utility do occur in the characterization of the mechanisms in highly evolved brains like ours. One may also speak of the goals of artefacts, in particular formal organizations – only elliptically, though, because intentionality is a peculiarity of individual brains. In the case of social systems an essential mechanism is a process that brings about the desired changes, or else prevents the undesirable ones. Only this very particular case is
132
Chasing Reality
covered by Merton’s well-known definition of social mechanisms as “the social processes having designated consequences for designated parts of the social structure (1968: 43).” 4 Causal and Stochastic Mechanisms A causal mechanism is of course one “ruled” by causal laws, such as those of classical electrodynamics or classical ecology. An example of such a mechanism is the electromagnetic induction that drives electric motors in accordance with Maxwell’s equations. Another is the oscillation of the populations of organisms of competing species according to the Lotka-Volterra equations. A third is the cooperation between two persons, or two social systems, according to definite (though not necessarily explicit) norms, such as that of reciprocal altruism. A fourth example is a negative-feedback mechanism such as Watt’s regulator of steam pressure. However, all causal processes are affected by some random noise, as illustrated by the fluctuations in electronic circuits and by accidental errors of observation. There is a natural tendency to think of all processes, hence all mechanisms, as causal (or deterministic in the narrow sense). Thus, a chemical reaction is usually conceived of as a causal process, or mechanism, for the emergence of a product from one or more reagents. However, quantum chemistry shows that this is only an aggregate effect: an individual or microchemical reaction is a process with an important random component. For example, a reaction of the type “A + B C” is to be analysed as the random scattering of As by Bs, with a certain probability that Cs will emerge. Roughly, the expected number of molecules of kind C resulting from n collisions of As with Bs, each with probability p, is np. Caution: This is not a process of the bottom-up type, because the value of the probability depends critically upon such macrophysical variables as temperature and pressure – which in turn are emergent macroproperties of the environment in question. There are plenty more stochastic (or random) processes, such as those of random walk, random self-assembly, lottery (random extraction of balls from opaque urns), atomic collision, and genic mutation. However, arguably all of these random processes have a causal component. For example, the balls in a lottery urn are pulled by gravity; the particles in an atomic collider are accelerated by electric and magnetic fields; and genes are activated and deactivated by enzymes. It is doubtful that there are purely causal or purely random processes, and hence mechanisms (recall chapter 4). At first sight, there is a third category of mechanism besides causal and random, namely, chaotic. However, most known cases of chaos are particular cases of causation, namely, those whose outcome, like that of a roulette game,
®
Behind Screens: Mechanisms
133
depends critically upon the initial conditions. After all, most chaos-theoretic equations do not involve probabilities. 5 Mechanism and Function Sometimes mechanisms are called ‘functions.’ This conflation is not advisable when one and the same task can be performed by different mechanisms – the case of functional equivalence. For example, some quadrupeds can advance by walking, crawling, or swimming; documents can be reproduced by printing presses, mimeographs, and photocopiers; markets can be conquered by force, dumping, free-trade agreements, or even honest competition; and certain goods can be sold in markets, retail stores, department stores, or through the Internet. Because the functions–mechanisms relation is one-to-many, we should keep the two concepts distinct while relating them. Another reason is that a purely functional account, such as “cars are means of transportation,” though accurate, is superficial because it does not tell us anything about the peculiar mechanism whereby the function in question is carried out (see Mahner and Bunge 2001). Some of what holds for our knowledge of cars also holds for that of systems of other kinds, such as towns. For example, it is not enough to know that African Americans tend to self-segregate in cities because they like to live among themselves. One must add that they are actively discriminated against, and even received with hostility, if they attempt to move to predominantly White neighbourhoods. Schelling (1978: 139) notwithstanding, racial segregation is not voluntary but the result of active racial discrimination. The latter is the invisible mechanism that manifests itself as segregation. Another reason for keeping the mechanism-function distinction is that, unlike mechanisms, the functions they accomplish are ambivalent. Indeed, as Merton (1968: 115) noted, social functions can be either manifest or latent (unintended). Besides, social mechanisms generally have dysfunctions as well as functions. Thus, the manifest function of the peer-review mechanism is quality control; but one of its latent functions, or rather dysfunctions, is to entrench cliques and perpetuate their beliefs. Thus, intellectual quality control may be perverted into the control of thought and power. There are no universal mechanisms, hence no cure-all procedures. All mechanisms are stuff-dependent and system-specific. For instance, only carbon has the versatility required to enter in the vast majority of the chemical reactions that keep an organism alive. Only live brains, provided they have been properly trained and primed, can engage in original research; and only brains in certain abnormal states can hallucinate. However, mechanisms, like anything else, can be grouped into natural
134
Chasing Reality
kinds, such as those of fusion and fission, aggregation and dispersion, cooperation and competition, stimulation and inhibition, blocking and facilitating, and so on. The formal analogies among mechanisms involving substrates or stuffs of different kinds facilitates the task of mathematical modelling, since one and the same equation, or system of equations, may be used to describe mechanisms involving matter of different kinds. For example, one and the same system of equations may describe cooperation and competition among organisms, in particular people, or among chemicals (see, e.g., Bunge 1976). Likewise, mechanisms of the so-called ying-yang kind – made up of couples of entities with opposing functions, such as stimulation and inhibition, or oxidation and reduction – are “instantiated” by molecules, neurons, social systems, and more. However, the functional and structural models and simulates are unavoidably shallow because, in ignoring stuff, they overlook mechanisms. And such oversight prevents the recognition of specific differences and qualitative novelties. For instance, dust specks, germs, seeds, ideas, habits, and artefacts diffuse, but their diffusion mechanisms are very different: some are physical, others biological, and still others social. Hence, we must disclose such mechanisms if we wish to either facilitate or hamper the spread of the stuff in question. Thus, the functionalist approach, which discards stuff and therefore overlooks mechanism, is bound to fail in practice as well as to set limits on research. Functionalism has been particularly pernicious in molecular biology and psychology. Indeed, in both fields it has discouraged the search for mechanisms, and it has promoted the idea that flowcharts provide explanations, when in fact they only supply global descriptions. For example, to state that messenger RNA acts as a blueprint for (or that it “specifies”) proteins is only a metaphorical description of protein synthesis. To understand this process in depth we need to know in detail the chemical reactions that produce the proteins in question. In other words, the processes of “transcription” (DNA RNA) and “translation” (RNA Protein) can only be understood in terms of inter-molecular forces. Likewise, psychology won’t explain much as long as it does not advance from the information-processing metaphor to the neural mechanism.
®
®
6 Mechanism and Law How are the concepts of mechanism and law-statement related? John Elster (1998: 48) has claimed that “the antonym of a mechanism is a scientific law.” Accordingly, explanations by reference to mechanisms would replace expla-
Behind Screens: Mechanisms
135
nation by reference to law-statements. This opinion is mistaken. Elster seems to have been misled by his examination of only a few cases of two kinds: (a) known mechanisms with unknown laws; and (b) known laws with unknown underlying mechanisms. But the fact that the pertinent mechanismic laws are unknown in certain cases, indeed in most, does not prove that they do not exist. Mechanisms without conceivable laws are called ‘miracles.’ For instance, Thomas Aquinas held that the Holy Ghost grafts the soul onto the human embryo; and John Eccles once speculated (in the revered journal Nature!) that the mind moves neurons through psychokinesis (or telekinesis). Surely, these hypotheses are mechanismic; but they are also unscientific, because they are inconsistent with the relevant laws, none of which refers to immaterial entities. I submit that scientific research presupposes (a) materialism, or the hypothesis that the real world is material, so that it contains no autonomous (subjectfree) ideas; and (b) the principle of lawfulness, according to which all events satisfy some law(s). Trust in the first principle allows scientists to dispense with the ghostly. And trust in the second principle sustains their search for laws and the rejection of miracles. Elster’s opinion, that mechanism is the opposite of scientific law, is falsified by the following counterexamples. Statistical mechanics explains thermodynamics in assuming that the elementary constituents of a thermodynamic system abide by the laws of classical mechanics; wave optics explains ray optics in proving that light rays emerge from the interference of light waves; molecular biology explains Mendelian genetics in proving that the heredity material consists of DNA molecules, some of which “code for” proteins, and thus ultimately control such biomechanisms as metabolism, cell growth, and cell division. Incidentally, the last two mechanisms are “governed” by the logistic curve, which is also that of learning driven by external reward. What is true is that, in social studies, law and mechanism are necessary but insufficient to explain, because almost everything social is made rather than found. Indeed, social facts are not only law-abiding but also norm-abiding; and social norms, though consistent with the laws of nature, are not reducible to these, if only because norms are invented in the light of valuations – besides which every norm is tempered by a counter-norm, as Merton (1976) pointed out. In any event, mechanism and law can be uncoupled only in thought. And explanation by reference to mechanism deepens subsumption instead of replacing it. The differences between the two kinds of explanation – subsumptive and mechanismic – as well as their commonalities, emerge clearly upon analysing them logically. Let us do this in two simple cases. Consider the well-known empirical generalization “Taking Ectasy causes
136
Chasing Reality
euphoria,” which makes no reference to any mechanisms. This statement can be analysed as the conjunction of the following two well-corroborated mechanismic hypotheses: “Taking Ectasy causes serotonin excess” and “Serotonin excess causes euphoria.” These two together explain the initial statement. (Why serotonin causes euphoria is of course a separate question that cries for a different mechanism.) Incidentally, the preceding example disproves Revonsuo’s (2001) assertion that explanation via mechanism is not available outside physics, in particular in cognitive neuroscience. An example from sociology and management science could be this: “The inertia [resistance to change] of a social system is proportional to its size.” This explains why even friendly takeovers, which require quick adaptations, are hazardous to corporations. In turn, the relevance of size to inertia is explained by the need for face-to-face (or at least screen-to-screen) contacts to maintain the cohesion of the system, and thus ensure its behaving as a unit. To put it schematically, we have split the initial statement “Bulkiness Þ Inertia” into “Bulkiness Þ ¯Contacts” and “¯Contacts Þ Inertia.” (The previous argument is clarified when expressed with the help of the standard symbolism of elementary logic. We started with a law-statement of the form “x(Ax Þ Bx),” and analysed it as the conjunction of hypotheses of the forms “x(Ax Þ Mx)” and “x(Mx Þ Bx),” where M refers to a key feature of some mechanism.) All real mechanisms are lawful, but the laws–mechanisms relation is one-tomany rather than one-to-one. For example, the oscillations in the price of any stock look like a random walk, superficially similar to the motion of a molecule in a gas; and the exponential function describes both the growth of a population with unlimited resources and that of scientific papers. There are two main reasons for the laxity of the laws-mechanisms coupling. One is that any given input-output relation (or black box) can in principle be mediated by different mechanisms (or translucid boxes). See figure 5.2a. The second reason is that the macrolevel laws relate global features, such as growth or decline, concentration or dispersion, that are compatible with alternative microprocesses. See figure 5.2b. Because the patterns–mechanisms relation is one-to-many, the search for either can be uncoupled from the search for the other. However, barring miracles, there are no lawless mechanisms any more than there are mechanism-less patterns. Hence, any mechanism-free account must be taken to be superficial and therefore presents a challenge to uncover unknown mechanism(s). By the same token, any mechanism unsupported by some law(s) must be regarded as ad hoc, and therefore equally transient. In sum, satisfactory (and psychologically satisfying) explanations of both
Behind Screens: Mechanisms Input
Mechanism
Macroproperty A
137
Macroproperty B
Output
Microproperty M
(a)
(b)
Figure 5.2M(a) An input–output relation mediated by a mechanism. (b) Two macroproperties related via a microproperty.
kinds, if scientific, resort to law-statements. So, mechanismic hypotheses are not an alternative to scientific laws, but are components of deep scientific laws. In other words, ‘mechanism’ (or ‘translucent box’) opposes ‘phenomenological’ (or ‘black box’), not ‘lawfulness’ (see Bunge 1964, 1967a). 7 Guessing Mechanisms How do we go about conjecturing mechanisms? The same way as framing any other hypotheses: with imagination both stimulated and constrained by data and well-weathered hypotheses. Let us consider a few examples. Altruism and cooperation among humans are about as frequent as selfishness and competition. Why is this so, that is, what mechanism drives either behaviour? Most sociobiologists claim that, because altruism and cooperation are expensive, they are confined to relatives. Accordingly spouses, adopted children, and close friends should not expect help in emergencies. That is, the animal would have an unconscious desire to protect and spread its own genes. This is the kin-selection hypothesis, the psychological counterpart of Richard Dawkins’s “selfish gene” conjecture, according to which organisms are only gene carriers. Though popular, the kin-selection hypothesis runs counter to the empirical evidence that there is cooperation among non-relatives, as well as rivalry, sometimes even violence, among kin (see, e.g., West et al. 2002). Moreover, transgenic relationships can be more loving than same-species ones. Thus, the American president George W. Bush declared once that his dog is the son he never had. An alternative hypothesis is that cooperation occurs regardless of kinship if its benefits exceed its costs; conflict would be parallel (see, e.g., Odling-Smee et al. 2003: 298ff.). Besides, we do good deeds because it feels good to do good even if we do not expect to be repaid. In fact, recent brain-imaging studies (Rilling et al. 2002) on people playing Prisoner’s Dilemma games have shown that we feel good when behaving cooperatively toward strangers:
138
Chasing Reality
certain neuronal systems in charge of reward (pleasure) “light up” in the process. An alternative is that humans and other animals tend to cooperate among themselves because they need help and expect reciprocity. These two hypotheses are of course mutually compatible. Thus, presumably cooperation involves at least two intertwining mechanisms on many different levels, cellular and social. There is no method, let alone logic, for conjecturing mechanisms. In general, hypotheses, as Peter Medawar put it, “appear ... along uncharted by-ways of thought.” True, Peirce wrote about the “method of abduction,” but ‘abduction’ is synonymous with ‘conjecturing.’ And, as he himself stated, this is an art, not a technique. One reason is that, typically, mechanisms are unobservable, and therefore their description is bound to contain concepts that do not occur in empirical data. (This is why mathematical modelling is often used to identify mechanisms. It is not that the core of reality is mathematical, but that only mathematical thinking can cope with complexity.) Even the mechanism that makes the pendulum clock tick involves unobservables such as the gravitational field and inertia; likewise, the elastic energy stored in a tensed bow is unobservable. Social systems are epistemologically similar. For example, factories are invisible: what one can perceive is some of their components – employees, buildings, machines, reservoirs, and so on – but not the way they work synergically, which is what keeps them together and going. Even the operations of a corner store are only partly overt. For instance, the grocer does not know, and does not ordinarily care to find out, why a customer buys breakfast cereal of one kind rather than another. However, if he cares he can make guesses – for instance, that children are likely to be sold on packaging. That is, the grocer may make up what is called a “theory of mind,” a hypothesis concerning the mental processes that end up at the cash register. If the grocer were a neobehaviourist, he might reason thus: Sight of package Appetite Purchase. Observable inputs and outputs, such as publicity and consumer behaviour, explain nothing. They only pose the problem of conjecturing the mechanism(s) likely to transduce inputs into outputs. Notice that this is a typically inverse problem, of the Behavior Intention type. Once a solution has been found, it allows one to attack the direct problem: Input & Mechanism Output. See figure 5.3. When powerful and reasonably true mechanismic theories are available, as in physics, parts of chemistry, and parts of biology, most problems are direct or can be transformed into such. But this is not the usual case in social studies. Here one has to start nearly from scratch when tackling a new problem. No
®
®
®
®
Behind Screens: Mechanisms
(a) Input (b) ? Input (c) Input
Mechanism Mechanism ? Mechanism
139
? Output, Output, Output.
Figure 5.3MDirect or forward problem (a) and inverse or backward problems (b) and (c). In principle, given a theory, solving (a) and (b) are matters of computation; by contrast, (c) calls for a theory that may not yet be available, and is therefore the hardest to solve.
general equations of social motion are known that can help predict what an individual or a social system will do when acted upon by certain stimuli, or figure out what were the stimuli and the internal processes that caused the observed reaction. In particular, the time variable does not occur in the vast majority of the formulas in standard mathematical economics, obsessed as it is by equilibrium. 8 Explanation: Subsumptive and Mechanismic The standard account of explanation in the philosophy of science, from Mill to Popper, Hempel, Braithwaite, and Nagel, is the so-called covering-law model. According to it, to explain a particular fact is to subsume it under a generalization, according to the schema: Law & Circumstance Fact to be explained. For instance, one may say that Aristotle died because he was human, and all humans are mortal; or that the price of soap went up because all merchandises became more expensive, and soap is a merchandise. All this is logically valid and sound (true). But it does not elicit understanding, and consequently does not qualify as explanation proper. I submit that the covering-law model fails to capture the concept of explanation used in the sciences, because it does not involve the notion of a mechanism (Bunge 1967a). For instance, one explains the drying of wet clothes exposed to sunlight by the absorption of light, which increases the kinetic energy of the water molecules in the wet cloth, to the point that they overcome the adhesive forces, and fly off. Senescence and death are explained by wear and tear, inflammation, oxidation, telomere shortening, and other mechanisms that operate concurrently. Cognitive neuroscience explains learning by the formation of new neuronal systems that emerge when they fire jointly in response to certain (external or internal) stimuli (Hebb’s law). Genuine explanations in the social sciences are similar. For example, unemployment of a certain kind is accounted for by the spread of labour-saving devices, which is in turn driven by the search for decreasing waste and
Þ
140
Chasing Reality
increasing profits. Alejandro Portes explained the growth of the informal or underground economy as a perverse effect of the labour legislation designed to protect workers. Mark Granovetter explained getting jobs by “the strength of weak ties,” that is, information about openings supplied by friends of friends. Wars are explained either by the desire of governments to retain or expand territories, natural resources, or markets, or to win the next election by making the innocent rally around their Great Leaders at a time of a National Emergency – namely, the very same emergency engineered by the very same patriots. In all such cases, to explain is to exhibit or assume a (lawful) mechanism. This is the process – whether causal, random, or mixed – that makes the system work the way it does. Of course, a mechanism need not be mechanical. There are thermonuclear, thermomechanical, electromagnetic, chemical, biological (in particular biochemical and neurophysiological), ecological, social, and many other mechanisms as well. I call this kind of explanation mechanismic. Use of this neologism is advisable given that most mechanisms are nonmechanical. Let us consider briefly two comparatively simple cases of non-mechanical mechanisms: those of demographic change and social cohesion. To a first approximation, the changes in the numerosity N(t) of a human group are represented by the rate equation dN/dt = kN. The solution to this equation is the exponential function N(t) = N0 exp (kt), where N0 is the initial value of N, and k the rate of growth. If k > 0, the population grows exponentially; if k = 0, it remains stagnant; and if k < 0, it declines exponentially. So far, we have only a description – as with all rate equations. The previous description is easily transformed into an explanation if the rate of change k is analysed thus: k = birth rate – death rate + immigration rate – emigration rate. This may be regarded as the overall demographic mechanism of a social system s. That is, we may set M(s) = k. (Incidentally, the sociobiologists are so obsessed with reproduction that they underrate the three remaining demographic mechanisms. And yet evolution involves not only reproductive success – a large birth rate – but also physiological fitness, adaptation to the environment, and active alteration of the latter, that is, niche construction. All of these processes show up jointly as a low death rate – in addition to migration.) Our second example is one of the oldest problems in social studies, namely, What holds society together in spite of the different, and sometimes conflicting, interests of its individual components? There have been many answers to this question, and some of them, though different, are mutually compatible.
Behind Screens: Mechanisms
141
For instance, according to some, the cement of society is reciprocal altruism (quid pro quo); other scholars claim that exchange is the glue of society; still others hold that homophily breeds cooperation – that people with similar interests, traditions, values, and customs are likely to get together and stick together; game theorists design Prisoner’s Dilemma models where people learn to cooperate – or, on the contrary, to defect; finally, Hobbesians only believe in conflict and coercion. Every one of these views has a grain of truth, but none of them is fully satisfying. We might learn something more by asking, What are the roots of social inequality, disunity, and marginality? Some of the most obvious answers are violence and exploitation; gender, race, class, and ideological discriminations; and residential mobility (see Tilly 1998). How do these causes or motives work, that is, what are the mechanisms that transform them into the observed segregations? It would seem that the mechanism common to all of them is exclusion or non-participation, whether or not deliberate. For instance, women are excluded from most top management positions, medical faculties, and golf clubs; Blacks, Catholics, and Jews from Wasp clubs; the poor from leafy streets and good schools; and agnostics and labour organizers from high political office. If we now turn to the original question, we realize that the key mechanism of social cohesion is participation – of people of all kinds in social networks of various types, of citizens in campaigns and polls, of women in jobs, of workers in the way their workplace is run, of teenagers in gangs, of seniors in charities, and so on. The notion of participation can be quantitated as follows. The degree or intensity of the participation of As in Bs (e.g., of women in academia, of youngsters in politics, or of employees in management) may be set equal to the numerosity of the intersection of the sets A and B, divided by the numerosity of the guest social circle A. Thus, a numerical index p of social cohesion (and its dual, social marginality) can be set up (Bunge and García-Sucre 1976). Since cohesion is sufficient for stability, we may set M(s) = p. Note that, far from being empirical, like most other social indicators, p is based upon systems-theoretical assumptions about social structure and social cohesion. The same systemic view suggests different indicators of further social features. For instance, we get a handy indicator of community attachment by asking the question, What factors, other than segregation, contribute decisively to social disorganization, that is, to the weakening or even dismantling of mechanisms of social cohesiveness? The answer is that one such factor is residential mobility: nomads do not fit easily in local social networks, hence they make no significant contribution to their strengthening. In other words, length of residence indicates level of participation in local social activities,
142
Chasing Reality
which in turn contributes to social cohesion (Kasarda and Janowitz 1974, Sampson 1988). In general, individual features explain collective ones, which in turn contribute to explaining individual behaviour. The systems-theoretic researcher moves from micro to macro and conversely. In other words, he employs two mutually complementary research strategies: top-down and bottom-up. 9 Realism versus Descriptivism Scientists have always known that to explain the behaviour of a system is to exhibit or conjecture the way it works, that is, its mechanism(s). Thus, William Harvey accounted for the circulation of the blood by conceiving of the heart as a pump; Descartes explained the rainbow in terms of the refraction of sunlight by the water droplets suspended in the air after a rain; Newton explained orbits in terms of forces and inertia; Berzelius (to Hegel’s horror) explained chemical reactions in terms of electrostatic forces; Tocqueville explained the fall of the ancien régime as a delayed result of the aristocrats’ neglect of their properties and counties, which in turn followed upon their concentration at Paris and Versailles, one century earlier, under pressure from Louis XIV; Darwin explained evolution by descent with modification cum natural selection; Marx and Engels explained history by both economic change and class struggle; Einstein explained the bending of light rays in the vicinity of massive bodies by the space curvature induced by the latter; Bohr accounted for light emission by the decay of atoms from excited to lower energy levels; Hebb explained learning as the formation of new neuron assemblies – and so on. We do not understand adequately the things whose mechanisms are still unknown. For instance, nothing but illusory understanding is gained by stating that the mind, or the brain, has “computed” this movement or that emotion. The computer metaphor is seriously mistaken, for (a) the most interesting mental processes, such as questioning, inventing, and problem-finding are spontaneous rather than rule-directed; and (b) algorithms are artificial rules for performing computations on symbols, not natural and lawful processes (Bunge and Ardila 1987, Searle 1980, Kary and Mahner 2002). This is why computers imitate some (not all) of the global features of some (not all) cognitive processes, and not the other way round. A consequence of the computationist fad is that some psychologists prefer doing computer simulations to engaging in brain research intent on finding neural mechanisms. A consequence of such neglect of the minding organ is that depression, schizophrenia, and other disabling mental disorders are not yet being well treated because their mechanisms have not yet been fully unveiled.
Behind Screens: Mechanisms
143
It is not enough to know that mental diseases are correlated with the imbalance of certain neurotransmitters. To know how such molecular facts translate into mental experiences, such as joy and sorrow, we must find out the effect of the excess or lack of dopamine, serotonin, and so forth, on neurons and neuronal systems. We must climb up the whole staircase, from molecule to brain and environment, and back. No knowledge of mechanism, neither understanding nor control. The phenomenalist and empiricist rules “Don’t explain: just describe” and “Describe only phenomena (appearances)” led the behaviourists and computationists, as well as the positivist philosophers, such as Comte, Mill, Mach, Duhem, Kirchhoff, Ostwald, and the members of the Vienna Circle, to reject the search for mechanismic hypotheses. Yet, ironically, the builders of modern atomic physics, a non-phenomenalist theory, paid lip service to those same positivist dogmas. Thus, in his epochmaking paper of 1925, Heisenberg stated that theoretical physicists should use only observable variables. But at the same time he introduced position and momentum operators without classical and therefore measurable counterparts. (Incidentally, the search for suitable position and velocity operators is still on: see Bunge 2003c.) In time, however, Heisenberg realized this inconsistency, and complained about his young colleagues who only wished to describe and predict facts. In 1969 he told me: “I am of a Newtonian cast of mind. I wish to understand facts. Therefore, I appreciate the theories that explain the working of things, more than any phenomenological [descriptive] theories” (Bunge 1971). Finally, we note that the quantum theory explains much but not everything. For example, it explains light emission as the decay of an atom or molecule from one energy level to a lower one. But it suggests no mechanism for this decay when not excited by a collision. Although most physicists are satisfied that this is a spontaneous (uncaused) process, a few wonder whether a mechanism for it might eventually be uncovered; which is an instance of the theses that calculation is not the same as explanation and that there is no final explanation. 10 Concluding Remarks Philosophical realism postulates that the aim of science is to explain reality. This thesis may easily be conceded, but differences are bound to emerge when it comes to explaining explanation. Mill, Popper, Hempel, and Braithwaite adopted the so-called covering-law model, according to which to explain a fact is to subsume it under a regularity. Others claimed that to explain is to propose
144
Chasing Reality
a metaphor. Some were led to this view by a superficial understanding of Bohr’s model of the atom and Darwin’s tree of life – none of which make sense separately from the theories they are part of. Still others adopted the metaphor view of explanation because it is part of the hermeneutic or linguistic “turn,” the postmodern view that the world turns around words. Rationalists reject metaphorism, because even good analogies can play only heuristic or pedagogical roles. Worse, they can be grossly misleading if taken literally. For example, the linguistic analogy in molecular biology, and the computer analogy in psychology, are misleading because they induce the illusion that the processes in question can be understood even though the corresponding mechanisms are largely unknown. Even the famous particle-wave duality in quantum mechanics is misleading, in suggesting that electrons and their kin can be described in classical terms. Actually the referents of quantum physics are neither particles nor waves: they are sui generis entities. Hence they deserve a name of their own; I have proposed to call them quantons (Bunge 1967c). The confusion over quantons is harmless compared with the hermeneutic view that is gaining currency in the cultural (social) sciences. According to it, these studies are “metaphors on metaphors” since culture (society) would pivot around language, and the latter is, at least in part, “a metaphorical construction of the world” (e.g., Tilley 1999). Please, keep talking, for we want the world to keep going. Rationalist philosophers dismiss metaphorism as an irrationalist extravaganza and an excuse for not soiling one’s hands in handling real things. And a few of them have realized the importance of mechanism, and thus the superiority of mechanismic explanation over subsumption (e.g., Bunge 1967a, 1968b, 1983, 2004a, 2004b; Kitcher and Salmon, eds. 1989; Machamer, Darden, and Craver 2000). And so, there is progress in philosophy. It may be slow because of conservation mechanisms – such as respect for tradition, willful ignorance, obscurity worship, ideological censorship, and a peer-review process dominated by philosophical conservatives. Nevertheless, philosophical advances do occur once in a while due to the operation of countervailing mechanisms, such as grappling with new problems posed by society, science, or technology, institutionalized scepticism, and, above all, curiosity about what goes on behind screens.
From Z to A: Inverse Problems
145
6 From Z to A: Inverse Problems
It is well known that engaging in research of any kind is to tackle cognitive problems. This is why a well-written paper starts by stating the problem(s) it tackles, and ends up by listing some open problems. The general concept of a problem should therefore be central to the study of knowledge. Yet the philosophical literature about problems in general – their logic, semantics, epistemology, and methodology – is scandalously poor. In particular, nearly all philosophers, social scientists, and policy-makers have ignored the very existence of inverse (or backward) problems. The exception has been the so-called problem of induction (Data Hypothesis), though it has seldom been realized that it is an inverse problem. This failure is likely to have nurtured the illusion that there must be an inductive “logic” just as formal, algorithmic, and rigorous as deductive logic. Yet, inverse (or backward) problems are pervasive, as well as the toughest and most interesting of all. Think of Newton’s problem, of “inferring” (in fact guessing) the laws of motion of planets from data concerning some of their successive positions; of jumping from sample to population; of diagnosing a sickness on the strength of its symptoms; of guessing a past event from its vestiges; of designing a device to perform certain functions; or of figuring out a plan of action to achieve certain goals. Likewise, to guess the intentions of a person from her behaviour; to discover the authors of a crime knowing the crime scene; to “image” an internal body part from the attenuation in intensity of an X-ray beam (computed tomography); to identify a target from the acoustic or electromagnetic waves bouncing off it (sonar and radar); or to guess the premises of an argument from some of its conclusions (axiomatization) – all of these are inverse problems too. And all of them are far harder than the corresponding direct problems. In general, going downstream is easier than going upstream.
®
146
Chasing Reality
Direct or forward problems call for analysis, or progressive reasoning, either from premises to conclusions, or from causes to effects. By contrast, inverse or backward problems require synthesis, or regressive reasoning from conclusion to premises or from effects to causes. In other words, work on direct problems is basically one of discovery, whereas the investigation of inverse problems calls for radical invention. No wonder then that the inventors of X-ray crystallography (William and Lawrence Bragg, 1915), the doublehelix model of the DNA molecule (Francis Crick and James Watson, 1953), and computed tomography (G.N. Hounsfield and Allan M. Cormack, 1979) were awarded the Nobel prize. Besides, a hallmark of most inverse problems is that, if nontrivial and soluble at all, they have multiple solutions. Just think of planning an activity designed to attain a given goal, as opposed to describing any such activity; of “reading” someone’s mind off his behaviour; of detecting landmines in a field, as opposed to planting them; or of the multiple interpretations of a postmodern text, by contrast to scribbling it. We all face inverse problems some of the time. For instance, not only humans but also apes do so when they form “theories of mind” to account for the observable behaviour of conspecifics. However, not even Polya (1957), who devoted an admirable book to the analysis of problems, realized that inverse problems constitute a distinct category of problems. The widely used collection of problems in analysis, by Polya and Szegö (1925), contains only direct problems. What Polya and others have done is to analyse, with a primarily didactic aim, indirect methods for solving direct problems, such as reductio ad absurdum. Widespread interest in inverse mathematical problems emerged only in the last few years. The vast majority of philosophers have been remarkably laconic on inverse problems, with the sole exception of the induction, or Data Hypotheses, problem. Since this is an inverse problem, it is likely to have either multiple solutions or none. The reason for this is as follows. By definition, a hypothesis goes beyond the data relevant to it. It does this in at least one of two ways: either because the hypothesis involves a leap from some existents to all possibles; or because it includes concepts that, like those of causation, mass, intention, and national sovereignty, do not occur in the data because they are not experiential. In sum, since data do not exude hypotheses, these have to be invented. And, of course, once invented they have to confront both old and new data. We will return to this problem in chapter 7. To emphasize the sharp contrast between the two kinds of problem, direct and inverse, and to better understand the nature of inverse problems, let us look at a sample of problems occurring in various branches of learning. The
®
From Z to A: Inverse Problems
147
goal of this exercise is not to attain truth by induction, but to become familiarized with the beast. After all, philosophers of science should expect more from analysing real cases of scientific research than from rereading and reinterpreting classical philosophical texts. 1 Preliminary Sample To find the shadow projected by a solid body on a given surface is a direct (or forward) problem with a single solution. Projective geometry and its application, descriptive geometry, provide precise rules for solving this problem familiar to machine designers and cartographers. The corresponding inverse problem is to reconstruct the shape of a body from one or more of its infinitely many projections or plane maps; like nearly all soluble backward problems, this one has multiple solutions. For example, when staring at a Necker cube – an ambiguous plane representation of a cube – one sees now a cube facing to the right, and about thirty seconds later a cube facing to the left, or conversely. The early spectroscopists faced an even harder task: that of “inferring” (actually guessing) the composition and structure of a light source, such as the Sun, from the line or band spectrum it emits. This inverse problem was a major motivation for constructing the quantum theory. With the help of this theory one can solve, at last in principle, the corresponding direct problem: Given (assuming) the composition and atomic or molecular structure of a source, find the frequencies and intensities of the light it can emit. Again, finding the traces left in the world by a well-defined deity – such as one indulging in planting bogus fossils – is a trivial direct problem, since every feature of the world will be attributed to its Creator. By contrast, “inferring” the profile of the deity from a study of Creation was the goal of the natural theologians of the first half of the nineteenth century – as well as of the contemporary professors funded by the Templeton Foundation. Somehow, no definitive identikit of the Creator has resulted so far. The preceding set of problems constitutes only a minute sample of the vast family of inverse (or backward) problems that pop up in all fields of research, policy, and action. A larger sample is shown in table 6.1. 2 The Direct–Inverse Relation: Generalities The general concept of an inverse problem is familiar to mathematicians, physicists, and engineers – to the latter under the names of synthesis and reverse engineering. By contrast, the concept is unfamiliar to most social scientists, policy-makers, and philosophers. A major reason for this gap is of
148
Chasing Reality
Table 6.1 Sample of forward vs. backward problems. The arrow symbolizes the research process, from givens to solution(s). Direct or forward
Inverse or backward
®
Algebraic equation Roots Antenna Wave form Artefact Function (task) Author Text Axiom system Theorems Basic science Applied science Brain in society Mind Cause Effect(s) Computer program Computation Creator Creation Crystal structure Diffraction pattern Dynamics Kinematics Disease Symptoms Earthquake centre Seismogram Equation of motion Trajectories Field equations Field intensities Force(s) Movements Gene mutation Mutant phenotype Generalization Particular case Grammar Sentences Ideas Speech Initial date Radioactivity Input Output Intention Behaviour Legislation Social behaviour Mathematical function Derivative Mathematical function Integral Means Goal Organ Role Past Present Plaintext Cyphertext Plan Activity Population Sample Potential Field intensity Probability Statistics Question Answer Reagents Chemical compound Scatterer Scattering Source (e.g., atom) Spectrum Stimulus Response Technical design Description Theory & data Predictions Transaction Budget
ÿ®ÿ ® ®
® ® ®
® ®
®
® ® ®
® ® ® ® ® ®
® ® ® ® ® ® ® ® ® ® ® ®
®
® ® ®
®
® ®
® ®
®
®
®
®
Roots Algebraic equation Wave form Antenna(s) Function (task) Artefact(s) Text Author Theorems Axiom system(s) Applied science Basic science Mind Brain in society Effect Cause(s) Computation Program Creation Creator Diffraction Crystal structure Kinematics Dynamics Symptoms Disease Seismogram Quake centre Trajectories Equation Field intensities Field equations Movements Force(s) Mutant phenotype Mutation Particular(s) Generalization Sentences Grammar Speech Ideas Radioactivity Initial date Desired output Requisite Input Behaviour Intention Social behaviour Legislation Derivative Function Integral Function Goal Means Role Organ Present Past Cyphertext Plaintext Activity Plan Sample Population Field intensity Potentials Statistics Probability Answer Question(s) Chemical compound Reagents Scattering Scatterer Spectrum Source (e.g., atom) Response Stimuli Specification Technical design Data Theory Budget Transaction
®
® ®
® ®
®
®
®
®
® ® ® ® ® ® ® ® ® ® ® ® ® ® ® ® ® ® ®
® ® ® ® ® ® ® ® ® ® ® ® ®
®
From Z to A: Inverse Problems
149
course that the latter fields of studies are mathematically underdeveloped. A second reason is the popular belief that a sufficiently powerful computer can handle any problem. A third reason is that inverse problems are far tougher than the corresponding direct problems. By the same token, inverse problems are also more intriguing, more demanding in ingenuity, experience, and labour, and often also more rewarding than the corresponding direct problems. Think, for instance, of guessing equations of motion from trajectories; of conjecturing the forces that scatter a beam of particles; of imagining the genotype-environment combination that produces a given phenotype (visible appearance); of hypothesizing the situationintention combination behind a given action; of ferreting out the grammar underlying a text; of guessing the way of life of the people who left behind certain archaeological remains; of imagining the artefacts that might perform a desired task; or of designing the industrial and commercial process that might yield a given return. A highly developed research field contains plenty of special methods (techniques and algorithms) for tackling direct problems of various kinds. By contrast, there are no special rules, in particular algorithms, for solving the vast majority of inverse problems. The usual way to go about them is to proceed by trial and error: to figure out and try out different hypotheses until the correct one is found. In turn, the latter is found by scanning and varying the existing set of similar solutions to the corresponding direct problems. By proceeding in this way, an inverse problem is rendered equivalent to a family of direct problems. In particular, an inductive problem is thus transformed into a set of deductive problems. However, before setting out to solve a problem we had better find out whether it is direct or inverse. And how do we distinguish an inverse problem from a direct one? To my knowledge, so far no generally accepted general and precise definition or criterion has been proposed. The following rules of thumb are tacitly used in different fields: (1) Mathematics: All the problems soluble with the help of well-defined techniques (in particular algorithms) are direct, whereas the converse process, that of recovering such problems from their solutions, is inverse. Examples: (a) Roots of an algebraic equation Equation; (b) Dots Curve; (c) Theorem Postulate(s) & Definition(s). (2) Natural and social science: The inverse problems are of the forms: Effect Cause, Sample Probability distribution, Property Thing, Behaviour Mechanism, or Macro-level Micro-level. Examples: (a) figure out the bond(s) that holds the constituents of a given system together; (b) discover the neuromuscular circuit that performs a given behavioural task;
®
®
®
®
®
®
®
®
150
Chasing Reality
(c) guess the decisions that triggered the individual actions that gave rise to a given social event. (3) Technology: The inverse problems are of the forms Function Mechanism and Dysfunction Defect in mechanism. Examples: (a) design a process for the mass production of a given drug; (b) design or redesign a formal organization to carry out a given task; (c) find out the causes of the malfunction of an artificial system, such as a computer or a bank. These and similar examples suggest the following pair of conventions: Definition 6.1 A direct or forward problem is one whose research goes down either the logical sequence or the stream of events; that is, from premise(s) to conclusion(s), or from cause(s) to effect(s). Definition 6.2 An inverse or backward problem is one whose research goes up either the logical sequence or the stream of events: that is, from conclusion to premise(s), or from effect to cause(s). Two caveats are in order. The first is that simplicity is irrelevant here. Although most inverse problems are harder than the corresponding direct problems, there are exceptions. For example, given a finite set of numbers, together with the fundamental theorem of algebra, we can easily find the algebraic equation that those numbers solve. By contrast, there may be no algorithms to solve the given equation. The second caveat is that sometimes more than one inverse problem corresponds to a direct one. For instance, consider the equation A f = g, where f and g are functions, and A is the operator that transforms f into g. The direct problem is: given A and f, find g. The same equation raises two inverse problems. The simplest of them is: Given A and g, find f. If A has an inverse, the solution is f = A–1g. The second inverse problem is much tougher, and it has an uncounted number of solutions: Given g, find A and f. For example, given the spatial distribution g of some stuff, find the law that relates g to the field f generated by g. The problem encountered by the founders of the field theories of electromagnetism and gravitation was precisely of this kind. Let us now proceed to examining the peculiarities of inverse problems in some research fields. In the present chapter we will look at mathematics and physics; some of the other sciences and technology will be tackled in the next chapter.
®
®
3 Logic and Mathematics Wittgenstein (1978), that great trivializer, regarded mathematics as a game. Indeed, he believed that doing mathematics is a matter of “following rules.” This is, presumably, how he taught at elementary school the mathematics he
From Z to A: Inverse Problems
151
knew. Paul Bernays (1976), Hilbert’s closest co-worker, aptly called it “household mathematics.” When the rules are known, mathematical work is not highgrade research but rather routine work, if often demanding and time-consuming. Such work can often be entrusted to a student or even a computer. When algorithms are available, the student is spared the effort of imagining any tricks; he tends to memorize rules and to go about problem solving in a rather mindless way. For instance, he may calculate integrals even after having forgotten the definition of an integral. This is why the study of Euclidean geometry, which is bereft of algorithms, can be more formative than that of analytic geometry, which resorts to algorithm-rich algebra and the calculus. Original mathematical research, like research in any other field, starts by spotting, inventing, or tackling research problems, that is, interesting open problems. And inverse mathematical problems happen to be research problems par excellence, because there are few if any rules (in particular algorithms) for solving them. In particular, the endeavour to invent a rule or method to solve mathematical problems of some kind is itself an inverse problem of the Goal Means type. This endeavour is so daunting in the case of inverse problems that the bulk of what is presumably the longest and most recent treatise in the field – Methods for Solving Inverse Problems in Mathematical Physics, by Prilepko, Orlovsky, and Vasin (2000) – is constituted by theorems on the solubility of inverse problems, in particular on the conditions for the existence and uniqueness of solutions to them. To a physicist or engineer this is like leaving a restaurant without having eaten anything, but having paid the cover charge for the privilege of reading the menu. To get a feel for mathematical inverse problems, let us examine a couple of elementary examples. The first is to decompose a given natural number into integers, which is harder than adding given integers. Thus, 1 + 2 = 3, whereas 3 = 1 + 1 + 1, and 3 = 1 + 2. The first problem is of the form (?x) (1 + 2 = x), whereas the second is of the form (?x)(?y)(?z) (3 = x + y + z). This second problem, unlike the first, consists of one equation with three unknowns. In general, when the number of unknowns is greater than the number of equations, the problem is indeterminate or mathematically ill posed. (In the preceding example, this happens because the addition function has no inverse.) Hence a problem of this kind has, at best, multiple solutions. We shall next meet an insoluble problem. The capital accumulated over a given number n of years at a given rate r is given by the familiar formula: Cn = C0(1 + r )n, where C0 is the principal (initial capital). To compute Cn from C0, r, and n is a direct problem that any calculator can handle. By contrast, the problem of finding out both the number of years
®
152
Chasing Reality
and the rate of interest that will earn a desired sum has as many solutions as couples, that is, infinitely many. Hence, from a practical point of view this problem, so easy to state, is insoluble. No wonder that investment counselling is (black) art rather than rigorous technique. 4 Interlude: Induction To compute a value of a well-defined mathematical function is a direct problem. By contrast, figuring out a function given some of its values is an inverse problem, namely, that of curve fitting (dot joining). There are standard techniques, such as the venerable Gregory-Newton formula, for performing such interpolations. However, most of these techniques yield functions of the same dull family, namely, polynomials that can be made to pass as closely as desired to the given empirical points. The existence of such algorithms falsifies Popper’s opinion that induction is a myth. Induction does occur in science, both in suggesting low-level generalizations and in evaluating the empirical support of hypotheses (Bunge 1960). Moreover, induction is sometimes “mechanizable,” as in the case we just saw. What is true is that induction does not yield high-level and amazing (counterintuitive) results. Indeed, in general the resulting data-fitting polynomials that compress and expand data differ from the true laws, which ordinarily involve far more complicated functions. For example, the elementary law of the simple (idealized) pendulum, deduced from Newton’s second law of motion, is T = (1/2p)(l/g)1/2, an approximation that can in turn be approximated by any of infinitely many polynomials in l/g. In this case, which is rather typical, the problem of induction is an inverse problem with infinitely many solutions, none of which is the correct one. The quest for high-level laws from data is hopeless unless the corresponding direct (deductive) problem has been solved in very many similar cases. More on this below. That induction is confined to low-level generalizations has been known for a long time, not so the reason for this limitation. The reason that induction cannot lead to the upper rungs of the deductive ladder is this: An inductive reasoning is semantically “horizontal,” in that it leaps from particular statements to a generalization containing exactly the same specific (non-logical) concepts involved in the data. For instance, n experimental dots on the x-y plane can be joined by a polynomial of degree n – 1 in x, namely, y = a0 + a1x + a2x2 + ... + anxn. Thus, if y stands for a spatial coordinate, and x for time, the function may represent the motion of a point mass, such as the tip of a harmonic oscillator. Yet, no matter how large n may be – that is, regardless of the bulk of the database – the said function is bound to be a poor and clumsy
From Z to A: Inverse Problems
153
approximation to the true solution of the linear oscillator problem, namely, y = a sin w t. This solution involves the oscillation frequency w, a parameter missing in the data but occurring in the equation of motion, which is one logical and semantic rung above the data. This, then, is why there can be no “vertical” inference from data to highlevel laws: because the latter contain concepts absent from the former. Since experience cannot generate any high-level concepts or hypotheses, these must be invented. And invention is anything but a rule-directed process, one subject to algorithms that could be fed into a computer. Calling such invention ‘abduction,’ following Peirce (1934: 5.181), only suggests the illusion that it is a mode of conclusive inference. It is preferable to think of theoretical concepts as “free creations of the human mind” (Einstein 1950a). Such constructs are free in the sense that they are not necessitated by sense data. But, of course, contrary to the “free fancies” that the artist may indulge in, and that Husserl (1931: 199ff.) commended for doing phenomenology, the scientific constructs are constrained by logic and scientific observation. 5 Mathematical Problems to Find and Problems to Prove Most mathematical problems are, in Pappus’s (A.D. 320) terms, either problems to find or problems to prove (Thomas, ed. 1941: 567). With luck, the former are well-posed direct problems, such as drawing a polygon, solving an equation, integrating a function, expanding a function in series, or just checking a calculation. By contrast, problems to prove are generally harder and more important than problems to find. The first strategy a mathematician will try to prove a conjecture of the form “If A, then B” is to assert it, yet negate its consequent B, and check whether this entails some contradiction. If it does, the theorem is pronounced true; if not, it stands falsified. In either case the task will have been accomplished. Clearly, such indirect reasoning is strictly deductive. It is also clear that, in case of success, one may summarize the process with the formula “conjectures and confirmations,” since proving a theorem amounts to confirming the initial conjecture. Popper’s (1963) slogan “conjectures and refutations,” adopted by Lakatos (1978), is only good for weeding out falsities through exhibiting counterexamples. Spotting mistakes does not make up for constructing guesses and corroborating them. The farmer who did nothing but weeding would have nothing to eat. If the theorem to be proved does not yield to indirect reasoning, the mathematician may have to resort to first principles. In this case, he tackles an inverse problem: that of guessing the assumptions A (axioms, other theorems, lemmas, or definitions) that jointly entail the statement t to be proved. The
154
Chasing Reality
form of such a problem is (?A)(A £ t). Regrettably, theorem proving is all too often presented as a direct problem like this: Find the logical consequences of a given set of assumptions. This is not a realistic description of mathematics in the making, because (a) not all the consequences of a set of assumptions are equally interesting; and (b) some new assumptions, such as auxiliary constructions in geometry, and lemmas in analysis, may have to be introduced or revised to prove a conjecture – and it takes knowledge and imagination to hit on the pertinent constructions or lemmas. Inverse problems are also called ‘ill posed.’ Surprisingly, Jacques Hadamard (1952) proposed the first definition of a well-posed mathematical problem as recently as in 1902. Regrettably, calling inverse problems ‘ill posed’ may have discouraged their investigation, for nobody wants to waste his time tackling wrong questions. Not surprisingly, most of the literature on inverse mathematical problems is rather recent and comparatively modest, although it has been growing exponentially in recent years. For instance, the journals Inverse Problems, Journal of Inverse and Ill-posed Problems, and Inverse Problems in Engineering were born as recently as in 1985, 1989, and 1995 respectively; the first international conference on the subject took place in Hong Kong in 2002; and the year after the first handbooks (Uhlmann 2003, Woodbury 2003a) appeared. However, some ancient mathematicians did attack successfully a few tough inverse problems. The best known of them is Euclid, who among other things was the first to propose the postulate system for what has since been known as Euclidean geometry. Some historians of mathematics have claimed that Euclid did nothing but collect the geometrical knowledge that had accumulated until his time. In truth he accomplished this, not a mean task in itself, and much more. Indeed, he organized that corpus in the most rational way, namely, axiomatically. (In fact, he invented the so-called axiomatic method – which is actually a format rather than a method, since there are no rules for finding axioms.) Euclid guessed a postulate system encompassing all that geometric knowledge. His was an inverse problem, because it is the dual of proving theorems from given premises with the help of given rules. Given a set of axioms and definitions, finding what they entail is a straightforward task by comparison with going upstream, from conclusions to premises, in particular from special cases to generalizations. True, Euclidean geometry is synthetic, not analytic, in the sense that it contains no algebra and therefore no algorithms; hence, the proof of nearly every theorem in it requires inventing a special construction, or importing a lemma from an adjoining field. However, given all the premises (and the rules of inference), the conclusions follow unambiguously, so much so that there are algorithms for proving
From Z to A: Inverse Problems
155
theorems in Euclidean geometry on computers. This is not the case with guessing postulates, an inverse problem with multiple solutions. In fact, there are several postulate systems for Euclidean geometry, among them those proposed by Hilbert in 1899, Huntington in 1913, and Tarski in 1959. Two of the twenty-three famous problems posed by Hilbert in 1901 were inverse, namely, to axiomatize the calculus of probability and the main physical theories of the time. The first task was attempted by various mathematicians, until in 1933 Kolmogoroff proposed the standard solution. As for physical axiomatics, Hilbert himself gave one example, namely, the simple phenomenological (descriptive) theory of radiation. Regrettably he did not axiomatize any of the basic physical theories of his day, although he was conversant with them. The first axiomatization of non-relativistic quantum mechanics, as well as of classical and relativistic mechanics, classical electrodynamics, special relativity, and Einstein’s theory of gravitation, was performed by Bunge (1967a). These axiomatizations included not only the mathematical formalisms, but also their realist physical interpretations via semantic postulates of the form “Mathematical concept M represents physical thing or property P.” This work is being updated by a handful of physicists (see, e.g., Covarrubias 1993 and Pérez-Bergliaffa et al. 1993, 1996.) 6 Astronomy and Microphysics Aristarchos may have been the first to tackle an inverse astronomical problem, namely, that of guessing the real and presumably regular planetary orbits from data about their apparent positions. This was a particularly intriguing problem at the time, given that the planetary trajectories, observed from Earth, were anomalous by comparison with those of the fixed stars. Indeed, “planet” means wanderer. Aristarchos hit on the right answer: he hypothesized the heliocentric system. However, he was unable to calculate any predictions, because the requisite mathematics had not yet been invented. Ptolemy mastered these formal tools, but his explicit goal was to describe appearances rather than to explain them. In fact, he may have been the first positivist (Duhem 1908). His ingenious geocentric system, involving the proverbial cycles and epicycles, succeeded in describing and predicting with remarkable accuracy the apparent motions of the then known planets. This was an astronomical feat, but at the same time a setback for physics, because the planets were attributed motions very different from those of the rest of the bodies. Besides, the planets and the Sun were considered as a motley assemblage of idiosyncratic celestial bodies rather than the system postulated by
156
Chasing Reality
Aristarchos. Ptolemy’s achievement was also a setback for ontology, because it widened the theological (and Platonic) chasm between the terrestrial and the celestial. Finally, his victory was a defeat for epistemology too, since it seemed to show that science favours phenomenalism and conventionalism over realism – a mistake repeated by the Inquisition against Galileo, as well as by Hume, Kant, and the positivists and neopositivists, from Comte and Mill to Nelson Goodman and David Lewis (recall chapter 1). The ancient Greek and Indian atomists were even more ambitious than either Aristarchos or Ptolemy. Indeed, they wished to explain everything visible in terms of invisible atoms hurtling in the void – from the drying up of sails and the manufacture of bronze to mental processes. They transformed the inverse problems Effects Causes and Phenomena Noumena into the corresponding direct problems. Ancient atomistics was, moreover, the first thoroughly naturalistic ontology. However, it was also a grand fantasy, since its upholders, who claimed to explain everything, did not account in detail for anything in particular. The transmutation of atomistics from a metaphysics into atomic physics took two millennia. Indeed, Daniel Bernoulli (1738) was the first to explain a macrophysical law (Boyle’s law of ideal gases) in atomic terms. And Dalton (1803) was the first to explain (approximately) some simple chemical compounds in like manner. These and other accomplishments did not suffice to persuade the majority of nineteenth-century physicists, who clung to classical (phenomenological) thermodynamics up until Planck’s foundational paper of 1900. As usual, philosophers upheld tradition: thermodynamics matched the positivist philosophy popular in the philosophical community between the mid-nineteenth century and mid-twentieth century. This philosophy – like that of Ptolemy, Hume, Kant, d’Alembert, Mach, Pearson, Duhem, and Ostwald – and contrary to the realism of Galileo, Huygens, Newton, Euler, Ampère, Faraday, Maxwell, and Boltzmann, enjoined scientists to stick to the phenomena (appearances). It involved a ban on the investigation of the inverse problem of going from phenomena to noumena, in particular the microphysical realities behind the former. Mach went so far as to claim that sensations, such as those of hotness and loudness, were the bricks of the world. The same phenomenalist thesis resurfaced later on in some of the works by Bertrand Russell, Rudolf Carnap, Hans Reichenbach, and Philipp Frank. Meanwhile, physicists were busy exploring such non-phenomenal entities as electrons and fields. Ironically, most of the founders of contemporary atomic theory – in particular Bohr, Born, Heisenberg, Jordan, Pauli, and Dirac – kept that obsolete empiricist philosophy. This they did while performing the Herculean (or rather Newtonian) feat of solving the inverse problem of going from macrophysical
®
®
From Z to A: Inverse Problems
157
description to microphysical explanation. For example, magnets were explained as assemblages of atoms with parallel magnetic momenta, and light rays as streams of photons. Still, positivism, in particular phenomenalism, was smuggled into textbooks and philosophical essays. For example, the basic dynamical variables, such as linear and angular momentum, spin, and energy, were dubbed ‘observables’ even though they are not directly measurable. 7 Reading Diffraction Patterns At about the time that atomic and nuclear physics were being built, crystallography underwent a breakthrough thanks to the use of X-rays. These were employed to discover the atomic configurations underlying the 230 pure crystal structures that had earlier been described morphologically. The method used to solve this inverse problem is as follows. A beam of X-rays (or of electrons) is made to diffract by a crystal. The diffracted waves show up on a photographic plate as spots forming an intricate geometrical pattern – such as parallel bands or concentric circles. These patterns are strikingly different from the atomic array that diffracts the rays. Hence they do not “speak for themselves”; they have to be “interpreted.” And this “interpretation” is actually a scientific explanation, not a hermeneutic intuition. It is carried out as follows. The crystallographer assumes a plausible crystal structure, and solves the direct theoretical problem of calculating the corresponding diffraction pattern. More precisely, he builds a theoretical model, or mini-theory, comprising hypothetical crystal structure, wave optics, and Fourier analysis. The coefficient of each term in a Fourier series for each hypothesized crystal structure is found by measuring the intensity of the corresponding X-ray beam. The theoretical pattern is then compared with the experimental one to check whether the theoretical model (or mini-theory) fits the experimental data to within the tolerable error. If it does not, the scientist keeps changing the hypothesized crystal structure until he reaches a correct result. Molecular biologists use the same method to discover the structure of proteins and nucleic acids. This is in fact how Crick and Watson unveiled the so-called genetic code in 1953: by the reciprocal adjustment of a sheaf of clever hypotheses with piles of X-ray diffraction patterns obtained by other researchers. Nowadays the crystallographer or molecular biologist need not proceed from scratch. Indeed, he can avail himself of a catalogue of crystal (or molecule) structure-diffraction diagram pairs. If the solution occurs in the catalogue, he looks no further. Otherwise he chooses the pattern closest to his own, wiggles some of the parameters in the corresponding crystal structure, and
158
B
Chasing Reality
T
D
Figure 6.1MX-ray and particle diffraction. A beam B of waves or particles is diffracted by a target T of unknown nature, and a diffraction pattern D is photographed. The problem is to guess T from D. The way to proceed is to assume several alternative Ts, calculate the corresponding patterns using Fourier analysis, and check which of them agrees best with the observed D. The latter may be concentric rings of dots, or parallel bands, that do not resemble at all the structure of T – a neat example of the poverty of appearances.
tries again, until he solves his problem. He proceeds thus by successive approximations to the truth. See figure 6.1. The problems in quantum physics are similar. Here, the forward problems are of this type: Assuming the forces (or the potentials from which they derive), calculate the possible energy levels, transition probabilities, scattering cross-sections, and other genuine observables. The corresponding backward problem is, of course, to figure out the potential from the said measurable properties. In particular, a physicist working with particle accelerators (“atom smashers”) faces the following problem. He knows what goes into a box (namely, a particle beam), and what comes out of it (scattered particles or photons). He knows the former because he has made it; and he knows the latter because his measuring instruments have recorded it. The physicist’s ambition is to find out what goes on inside the box. That is, he seeks to unveil the mechanism mediating the incoming to the outgoing particle beams. In short, his problem is: Observed facts Likely mechanisms. Until recently, next to nothing was known about nuclear forces except that they are strong, short-range, saturable, and non-classical. In the early days of theoretical nuclear physics, a rather simple-minded method was adopted to tackle problems such as that posed by the scattering of neutrons by protons. Typically, one used to posit a potential well (for the unknown attractive force) with a simple shape, such as a dromedary’s humps. One then adjusted the depth (force strength) as well as the width (force range) of the well to fit the experimental values of the energy levels and the effective scattering crosssection. In recent years, some algorithms have been devised for performing this task, which may be carried out on a computer (e.g., Zakhariev and Chabanov 1997,
®
From Z to A: Inverse Problems
159
Hall 1999). Moreover, by deforming the potential and noting the effects on either the energy levels or the scattering cross-section, the scientist is said to develop an intuition different from that gained in solving direct problems; even new vistas and new problems may emerge when focusing on inverse rather than direct problems. However, some of the properties of the said potential have to be assumed – for instance, that it is piece-wise continuous, bounded below, and symmetric. Besides, the algorithms in question, though numerically powerful, do not solve the essential problem, which is that of the nature of the forces at play. After all, computers are not concerned with physical interpretation. Nowadays there are refined theories of nuclear forces, so that in principle one could solve the direct problem, that is, compute the scattering they cause. However, the computational difficulties are formidable. Therefore, inverse problem solving is still prominent in the entire field. The difference with the old days lies in the fact that by now a rather extensive catalogue of solutions of a state (or wave) equation for different potentials (or forces) is available. Hence, an examination of any new set of empirical data may point a trembling finger to the corresponding potential. 8 Invertibility If the solution to a direct problem is a 1:1 map, such as the exponential function, then the solution to the corresponding inverse problem exists and is unique. For example, if it is known that a given process decays exponentially in the course of time, as in the cases of radioactivity and the destruction of a bacterial colony by an antibiotic, then the time at which the process started can be calculated from the residual intensity of the process. This is how carbondating works. Indeed, assuming the ratio of Carbon 14 (radioactive) to Carbon 12 (stable) at the time of death of a fossilized organism, the date of that event is obtained by taking logarithms. (Recall that y = a exp(–bt) is equivalent to t = – (1/b) ln (y/a).) Another typical case is that of economic planning on the strength of a production model (see, e.g., Gale 1960). The gist of this model is the equation for the net production of a firm: p(I – A) = g, where p is the production vector, A is the input-output matrix, I is the unit matrix, and g the demand or goal. The direct or forecast problem is to find g given p and A. By contrast, the inverse or planning problem is to find p, the intensity of the activity, given A and g. This problem consists in inverting the matrix I – A to obtain p = g(I – A)–1. However, this problem need not have a solution: that is, the matrix in question may not have an inverse. In this case, then, Solubility Invertibility. And even if the
Û
160 C
Chasing Reality M
R
Figure 6.2MA black-box (or externalist) model of a system specifies the input or circumstance C and the output or response R, but not the internal process or mechanism that transduces C into R.
matrix does have an inverse, the solution vector p, though mathematically unique, stands for any member of a whole set of different activities (or mechanisms) with the same intensity. Such indifference to specific mechanism is of course characteristic of black-box models. A black-box theory models the system of interest as a black box with two terminals: the input C (circumstance, situation, stimulus) and the output R (overt response), both observable (e.g., Bunge 1963). See figure 6.2. Typically, both input and output are indicators, that is, observable manifestations of imperceptible events in the system or its environment. (The particular cases of null input, or spontaneous activity, and null output, or indifference to environmental stimuli, are supposed to be included.) That is, the low-level law characterizing the system in the given environment is of the form R = B . C. The mathematical forms of C, R, and B depend on the kind of problem. For example, C and R may be functions, and B may be the operator relating them; or all three may be matrices. Problems of two kinds can be tackled with the help of a black-box model like the one we have just sketched: one direct problem and two inverse problems. The direct problem is one of forecast: Given B and C, compute R. The corresponding inverse problems are these: (a) Hindcast problem: Given R and B, find out C, that is, calculate C = B–1 . R. This problem is insoluble if B lacks an inverse, as in the case of a rectangular matrix. And when B does have an inverse, the computational task may require a high-powered computer, as the macroeconomists, who deal with huge input-output (Leontief) matrices B, know well. (b) Explanation (or synthesis) problem: Given R and C, find out B. In other words, given the data, guess the process or mechanism responsible for their particular pairing. This is not a mere matter of computation: the task calls for potent theorizing, as made plain by crystallography (Diffraction spots Crystal structure) and atomic and nuclear physics (Scattering cross-sections Force). The practical ideal is of course to be able to use algorithms, that is, precise and “mechanical” (mindless) computation rules or procedures that can be entrusted to computers. However, the vast majority of algorithms work only
® ®
From Z to A: Inverse Problems
161
for well-posed (forward) problems, or for certain narrow classes of backward problems occurring in theories with powerful mathematical formalisms. For instance, as mentioned in section 4, there are now algorithms to tackle the Scattering data Scattering force problem in microphysics. Still, these methods fall short of the empiricist’s dream, which is to leap from data to theory, since nothing can be inferred about the nature of the forces at play. Indeed, any given potential shape is consistent with forces of many different kinds. For example, although the shapes of the gravitational and the electrostatic (or Coulomb) potentials are the same, the corresponding fields (or forces) are very different, as shown by the different ways of detecting them and measuring their intensities. So, strictly speaking, the algorithms in question do not solve the given inverse problem. What they do is to supply the theorist with constraints on the possible forces. The rest, that is, guessing the forces, is up to the theorist’s ingenuity. Still, in most cases even the most astute theorist will fail. For instance, the mass of a body is uniquely determined by its mass density; on the other hand, any given value of the total mass is compatible with infinitely many mass densities. Hence, the Mass Density problem is insoluble. The same holds for all the other densities. In particular, given a probability density, one can compute any desired statistic, such as the average and the variance. Yet the inverse problem, namely, Statistics Probability, is just as insoluble as the Trajectory Equation(s) of motion problem. Equally insoluble is the problem of recovering the initial state of a branching process of the stochastic kind: any given final state is compatible with very many different beginnings. Think of the thermal equilibrium state attained by a system composed of two or more bodies, at different temperatures, that come into contact. Better yet, think of the Boltzmann-Planck equation “S = k ln W,” which relates the macrophysical property S (thermodynamic entropy) to the number W of different microphysical configurations consistent with a given value of S. What if the inverse problem proves intractable? In this case one may try cheating a little, namely, by transforming the given ill-posed problem into a well-posed one. This somewhat irregular procedure is called regularization. “The idea is to make the problem ‘regular’ by changing the original problem slightly. Essentially, one exchanges the old ill-posed problem for a new ‘regular’ or well-posed one whose solution (hopefully) lies near that of the original ill-posed problem” (Woodbury 2003b: 57). This procedure is admissible if the solution is stable, that is, if small changes in the data map into small changes in the “regularized” solution. Thus, the cheating is kept under control. So much for the formal aspect of the invertibility problem. Let us now
®
®
®
®
162
Chasing Reality
glance at its substantive aspect. In all of the factual sciences, inverse problems are tackled in either of two ways. One is to transform the given ill-posed problem into a well-posed one (regularization). The other approach is to attempt to guess the unobservable mechanisms (processes) that mediate between the observable stimuli and responses. The first strategy is purely formal, and therefore it can be employed only when precise mathematical models are available. Otherwise one must take the bull by the horns, hoping to come up with a more or less precise model of the system or process in question that describes the essential traits of the mechanism that gives rise to the observed facts. Here is a random sample of examples of the second kind: the generation of seismic waves by the collision of tectonic plates; the cathode-ray tube behind the screen in a TV set; the process whereby insulin regulates the breakdown of sugar; the action of alcohol on mood; the combination of gene changes with natural selection that drives evolution; the genome-environment combination that results in a phenotype; the combination of cooperation and competition that holds a social system together; the big business–politics connection that corrupts democracy; and such quality-control mechanisms as random product sampling and peer review. (More on mechanisms in chapter 5.) 9 Inverse Probabilities Let us finally make a quick remark on the old problem of inverse probabilities, which we met briefly in chapter 4. This problem arises in applications of the Bayes theorem Pr(B A) = Pr(A B) . Pr(B) / Pr(A). In this formula, the arguments of the probability function P are sets, and the conditional probabilities Pr(B ) A) and Pr(A ) B) are said to be mutually inverse. Actually they are mutually dual rather than inverse, because neither of them is logically prior to the other. The direct-inverse distinction is epistemological and it arises in applications. Let us peek at this problem. As in any other formula in the probability calculus, the arguments A and B can be interpreted as random states or events. And yet some statisticians, epidemiologists, rational-choice theorists, and philosophers have used the formula to calculate either the probability Pr(C ) E) of the cause C given the effect E, or the probability Pr(H ) D) of hypothesis H given data D. These popular applications are open to three fatal objections. A first objection is that such interpretations contradict the tacit assumption in science that, with reference to any real situation, the arguments of a probability function must denote random states or events rather than either causally related events or propositions. For instance, one does not compute or measure )
)
From Z to A: Inverse Problems
163
the probability that a rainfall will wet the ground or the probability that the Schrödinger equation is true. Popper claimed that law-statements have zero probability, and this is why he regarded improbability as a virtue. My point is that it makes no sense to assign probabilities to propositions: recall chapter 4. A second objection is that ordinarily the prior probabilities Pr(A) and Pr(B) are unknown, so they are made up – hardly a scientific procedure. Indeed, an arbitrary assignment of numbers to unknowns yields no knowledge; it only creates a false impression of rigour. It may even lead to disastrous practical consequences, as when Bayes’s formula is applied to medical diagnoses or business or military decisions. (More in Bunge 1985b, 1988, and 2003a.) Probability applies legitimately only to random events, such as quantum tunnelling, gene mutation, or random choice. My third and final objection to using Bayes’s formula to infer causes from effects, or initial states from final states, is that it does not fit actual scientific practice. Thus, the quantum theory of radiation allows one to calculate the probability Pr(n n') that an atom in a state n will decay to a state n' and emit a photon. The same theory yields a different formula for the probability P(n' n) that an atom in a state n' will absorb a photon and jump to the energy level n. These two probabilities are not related by Bayes’s theorem. Nor could they be, because the corresponding mechanisms are quite different. Indeed, spontaneous light emission occurs even in a closed system, and is thought to be influenced by the zero-point fluctuations of the vacuum field. By contrast, light absorption can only occur if the system is open enough to let at least one photon in.
®
®
10 Concluding Remarks Whether epistemic or practical, problems can be direct (forward), inverse (backward), or neither. The third category includes such problems as those of evaluating actions or artefacts, interpreting formalisms, teaching new subjects, and finding new problems. Research on direct problems follows either the stream of events or the logical sequence. By contrast, research on inverse problems works backward. And research on problems of the third kind proceeds typically in a zigzagging fashion. Obviously, a problem of either kind is either soluble or not. If soluble and inverse, its solution is either single (first kind) or multiple (second kind). The problems of the first kind are characterized by an invertible mapping, whose domain is a set of unobservables, and whose image is a set of data. For better or for worse, such problems are less numerous than the far more unwieldy but more interesting problems of the second kind. An inverse problem of the
164
Chasing Reality
second kind, that is, one with multiple solutions, is equivalent to a whole family of direct problems. That is, in this case one proceeds by imagining different scenarios (or mechanisms) that might produce the same result, or different premises that might entail the same conclusions. In sum, some of the most challenging and rewarding scientific problems are of the inverse type. And the most promising strategy for handling a tough inverse problem is to try and transform it into a direct problem. This is what Newton did to account for the observed trajectories of terrestrial and celestial bodies. Indeed, despite his protestation that he did not feign hypotheses, he certainly posited some and proceeded deductively rather than inductively. In fact, Newton crafted the earliest successful theory in history in response to the most intriguing and topical inverse problem of his time. Thus, he tackled the so-called Hume’s problem in technical terms long before Hume posed it in ordinary-knowledge terms. Furthermore, Newton solved that problem. Why then not call it “Newton’s inverse problem”?
Bridging Fact and Theory
165
7 Bridging Fact and Theory
In this chapter we will carry on the investigation of inverse problems begun in the previous chapter. But here we will concentrate on the dual problems of how to go from facts to hypothesis or theory and how to subject the latter to reality checks. In other words, we will study the problems of how to move from “the given” to the far wider and deeper sought-for and how to force a theory to face facts. We shall also tackle the converse problem: How does one go down from high-level theories that make no direct reference to any data to empirical evidence for or against them? In particular, what counts as an observable indicator of an unobservable fact? 1 Induction Again Much, perhaps too much, has been written about the leap from data to hypothesis. And yet this kind of inference still puzzles philosophers and cognitive psychologists – as it should. There are two main myths about induction: that it is everything (e.g., sometimes Bacon and Russell; Carnap and Reichenbach always), and that it is nothing or nearly so (e.g., Leibniz, Hume, and Popper). The former myth belongs to the core of empiricism, whereas the latter is at the centre of rationalism. Francis Bacon is usually regarded as the inductivist par excellence. This he certainly was in his New Atlantis (1624), the utopia where he depicted scientists as data gatherers and packers. But Bacon was no simplistic inductivist in his Novum Organon, published four years earlier. Here he dismissed induction proper, or generalization from particulars, as trivial. He advocated instead framing alternative hypotheses and checking them, if possible experimentally; and he introduced the concept of a crucial experiment, by analogy with a crossroad.
166
Chasing Reality
Of the various examples Bacon gave, perhaps the clearest and most interesting concerns the nature of weight, a much-discussed problem in his time. Is heaviness an intrinsic property of bodies, or is it caused by the Earth’s attraction? He proposed the following ingenious experiment to settle the question (1905: xxxvi). Take two clocks with different mechanisms, one moved by a spring and the other by a pendulum, synchronize them, and carry them to the top of a tall steeple. If weight is an intrinsic property of bodies, the two clocks will continue to mark the same time; otherwise the pendulum clock will tick slower than the other, because the gravitational attraction must be weaker at a higher altitude. Bacon’s experiment was actually not performed until a couple of centuries later, disproving conclusively the hypothesis that weight (unlike mass) is an intrinsic property of bodies. At the same time, he showed the centrality of hypothesis in science – in spite of which he is usually regarded as a radical inductivist. Many philosophers have been rightly sceptical about the power and reliability of induction proper, but different philosophers have adduced different reasons for doubting it. Hume’s (1734) is the best known. He argues that the fact that the sun has “risen” every day in the known past is no guarantee that it will “rise” again tomorrow. Hume’s tacit assumption is that there are no laws of nature, that nature has only certain habits that it may quit without warning, much as an individual may suddenly quit smoking. Astronomers think differently. They predict confidently that the sun will “rise” every day as long as it lasts, provided our planet is not hit by a large asteroid or comet. This prediction is based on the law of conservation of spin, or intrinsic angular momentum. And this law “applies” to our planet because this body is similar to a frictionless spinning top. (Hold a rotating spinning top in your hands, and you will find out how hard it is to alter the rotation axis.) Likewise, human biologists predict confidently that human immortality is impossible, not because so far all humans have had a finite lifespan, but because the human body ages unavoidably through various mechanisms until finally all the more critical subsystems break down. Thus, we affirm the truth of “All humans are mortal” not just because of past experience, but because we can explain ageing. In both cases, those of sunrise and mortality, scientists reason on the strength of lawful mechanisms, not mere empirical generalizations. In the case of the succession of days and nights, the mechanism is the occultation of the Sun resulting from the diurnal rotation of the Earth. And human mortality results from the concurrent operation of several mechanisms: wear and tear, stress, oxidation, inflammation, mutations and the consequent inability to synthesize certain proteins, telomere shortening, and so on.
Bridging Fact and Theory
167
The limitations of induction derive from the shallowness of its products, all of which concern only observable properties, with no reference to hidden mechanisms. By contrast, direct reasoning from law-statement to fact presupposes that nature is lawful rather than capricious. Hume was obviously right in holding that the leap from “some” to “all” is logically invalid. But he was wrong in denying the objectivity of connections, in particular laws, and thus in denying the existence of natural (or nomic) necessity along with logical necessity (see Bunge 1959a). Two centuries later Kripke (1980) perpetrated a similar fallacy when asserting that, since the hypothesized identity “Mental function = Brain process” is contingent rather than logically necessary, it may not hold in worlds other than our own. But of course if such “worlds” are accessible from ours, then they are not worlds proper but only remote regions of our universe; and if they are inaccessible, then we can only fantasize about them – a task best left to fiction writers. To conclude this section, all the deeper and therefore more interesting scientific hypotheses and theories are conjectured. And this process, from data to hypothesis or theory, is neither inductive nor deductive; it is as little ruledirected as falling in love, composing a poem, or designing an artefact. 2 Abduction Again Many students agree that the way to account for observable facts is to frame alternative hypotheses, and then weed out the false ones, ideally by means of crucial (conclusive) experiments. This was well known to Francis Bacon (1620), William Whewell (1847), and John R. Platt (1964), among many others. Regrettably all three authors, along with many others, called this procedure “induction,” while strictly speaking induction is the formation of a single lowlevel generalization from the data in hand – as when a dog learns to associate leash with walk, or a baby learns to call all men Dad. Calling this procedure “inductive method” or “strong inference” suggests mistakenly that it is a safe and quick recipe for devising hypotheses from data, when in fact there is no such method – except for Pavlovian conditioning, which obviously does not help frame scientific hypotheses. The only strong inference is deduction, as it occurs in the derivation of a scientific prediction from law(s) and data. This is so because deduction, unlike induction, analogy, and guesswork, is the only inference subject to precise, stringent, and general logical rules. In turn, these are general because they are formal, whereas non-deductive inference depends essentially on content. For example, the analogies “1:2 :: 2:4” and “Individual is to society as atom is to molecule” are both content-
168
Chasing Reality
dependent and logically independent from one another: neither is deducible from the other. Some philosophers favour what the great Charles S. Peirce (1934: 5.181) called “abduction.” He rightly regarded this as an act of fallible insight rather than a methodical procedure. Harman (1965) termed it “inference to the best explanation,” a phrase that occurs frequently in the contemporary philosophical literature. However, it is still being debated whether this is either a form of induction or anything other than the choice of the hypothesis that best accounts for the data, where in turn “best” is equated with “most accurate,” “explaining the most (or least ad hoc),” “simplest,” or all three (see, e.g., Flach and Kakas, eds. 2000). However, exact empirical adequacy and conceptual simplicity are neither necessary nor sufficient in the advanced sciences. Here, a reasonably wellconfirmed hypothesis that belongs to a well-corroborated theory, on top of which it coheres with the ruling worldview, ranks much higher than a stray empirical generalization, even if the latter fits somewhat better the data in hand. This condition, which I have called ‘external consistency’ (Bunge 1961, 1967a), is roughly what Thagard (1978) terms ‘explanatory coherence.’ However, the word ‘systemicity’ may be more suggestive than either of those names. The reason conceptual systemicity is highly desirable is this: When entrenched in a coherent (systematic) body of knowledge, a hypothesis enjoys the support of other members of that body in addition to the relevant empirical evidence. If only for this reason, the frequent resorting to crime detection in the literature on the scientific method is somewhat unfortunate. Surely good detectives make alternative conjectures and check them, but they do not construct or even use scientific theories. And in some fields, such as psychology, anthropology, medicine, and management science, theorizing is still frowned upon, so that the hypotheses are scattered and therefore they hardly enrich and control one another. To get a “feel” for this creative process, let us analyse some representative examples taken from the life sciences, social studies, and technology. 3 Biology: Evolution Evolutionary biology, like cosmology, geology, archaeology, and historiography, is a historical science. Hence its practitioners face a large family of inverse problems of the Present ® Past type. In particular, the reconstruction of any lineage (or phylogeny) is tentative if only because of the large gaps in the fossil record. However, qualitative novelties emerge in the course of
Bridging Fact and Theory
169
individual development, which can be monitored and altered in the laboratory. Therefore, some of those novelties can be caused deliberately on modern organisms. This is why some inverse problems in evolutionary biology and genetics can be transformed, at least in principle, into direct problems. Actually, this is how evolutionary biology became an experimental science between the two world wars: by tampering with the genome, first with X-rays, and nowadays chemically as well. Recall from section 1 that inductive generalization from a bunch of data works only for hypotheses involving exclusively observable features, such as “All mammals are hairy,” but it fails miserably for high-level hypotheses, such as “All mammals descend from reptiles.” How does one go about framing such high-level hypotheses? The short answer is that one invents them, though not out of the blue but with the help of data, analogies, further assumptions – and above all a sensitive and experienced “nose.” For example, a comparison of certain well-known phylogenies with the history of the Earth suggested to Elizabeth Vrba (in Gould 2002: 918ff.) her “turnover-pulse hypothesis.” This is the conjecture that the massive speciations and extinctions that have “punctuated” biological evolution over three billion years were triggered by drastic and rather quick environmental changes, such as large variations in global temperature, meteoritic impacts, and continental drift – to which we must now add the alterations caused by pollution, logging, overcultivation, overfishing, and the like. The hypothesized mechanism is obvious: such swift and large-scale environmental disruptions destroy some habitats while opening up others. For instance, when a landmass splits into two, the area of shallow water, propitious to marine life, increases roughly by half. Certainly, evolutionary biology is not the sole abode of inverse biological problems. Every attempt to find the unknown organ that discharges a known function (or performs a certain role) requires research into an inverse problem. This holds, in particular, for the task of the cognitive neuroscientist, said to be that of “mapping the mind onto the brain.” However, here too many an inverse problem can be transformed into a direct one. For example, by tampering with the brain, the neuropsychologist can cause mental disorders or deficits in experimental subjects. (More in the section 5.) What holds for organismic and evolutionary biology holds, a fortiori, for the sciences that investigate lower levels of organization, in particular genetics. Thus, the problem of identifying the gene(s) “responsible” for a given phenotypic trait is of the inverse type. For example, if an adult mammal does not tolerate dairy products, it is because it cannot synthesize lactase, the enzyme involved in the digestion of milk; and in turn lactase deficiency is due to the
170
Chasing Reality
lack of the gene involved in its synthesis. The researcher is thus faced with the inverse problem: Metabolic disorder ® Enzyme deficiency ® Genetic disorder. Once the suspect genes have been fingered, the problems of finding out the corresponding enzymes can be tackled. The solution to these direct problems should solve the original inverse problem. The practical payoff is obvious though still distant, namely, gene therapy. 4 Medicine: From Symptoms to Diagnosis Something similar holds, mutatis mutandis, for medicine. A few medical diagnostic problems are direct. Open bone fracture is one of them: In this case there is no need to guess invisible mechanisms to account for the cause of tissue ripping, bleeding, and pain. However, all of the diagnostic problems in internal medicine are of the inverse type. Indeed, they are of the form “Given such and such signs, guess the underlying causes.” In the old days, such guesses were educated only by experience and post-mortem examinations. Nowadays basic anatomical and physiological knowledge, along with biological and biochemical assays, X-rays, CT scans, fMRI images, and other diagnostic tools often help spot the trouble. Using such diagnostic tools, the physician transforms the original inverse problem into one or more direct problems. For example, if the patient complains of fatigue even with adequate nourishment and rest, the physician may suspect anemia, and order a leukocyte count to check this hypothesis. But a look at the patient’s family history may suggest diabetes, in which case the physician will order a sugar-level measurement among others. These are direct problems because experiment has found that the said suspects, and no others, cause the symptoms in question. As with inverse problems in other fields, the most promising biomedical research strategy is to imagine and test mechanismic hypotheses. Actually this is how contemporary biomedical research proceeds. The starting point is a conditional of the form “Ailment Þ Syndrome,” or “x(Ax Þ Sx)”; read “For all individuals, if anyone suffers from ailment A, then she exhibits syndrome S.” The direct problem of inferring S from A is trivial, but it seldom occurs in the internal medicine clinic. Here the problem is inverse: Symptom ® Ailment. And any solution to it is only tentative, unless supported by research findings concerning mechanisms. These extra hypotheses are likely to be of the forms “A if and only if M” (A Û M ), and “M if and only if S” (M Û S). Here M summarizes the description of some mechanism, such as artery clogging, TB infection, or work-related stress. Now the strong inference A Û S can proceed confidently (more in Bunge 2003a).
Bridging Fact and Theory
171
5 Psychology: Behind Behaviour In daily life we all face the problem of figuring out the beliefs and intentions that motivate other people’s actions. In other words, we attempt to “read” other people’s minds off their overt behaviour. There are two ways in which humans, chimpanzees, and some pets do “mind-reading.” One is by empathizing with others (e.g., Preston and de Waal 2002). The other is by framing various “theories of mind,” that is, by attributing to others certain beliefs, preferences, or intentions (e.g., Premack and Woodruff 1978, Whiten 1997). Clearly, consistent success in empathizing and “mind-reading” confers an advantage. By contrast, “mind-reading” inability, as in egotistic individuals, and autistic and Asperger patients, is a serious handicap in social intercourse. Methodologically speaking, the “mind-reader” transforms the inverse problem Behaviour ® Intention into a family of direct problems of the Intention ® Behaviour kind. But of course any such “theory,” or rather guess, may turn out to be false. (Worse yet, the observer may be unable to craft one because his medial prefrontal cortex is deficient or damaged, like that of an autistic patient.) Hence, our guesses about other minds are tentative and subject to falsification. However, judging from the assurance of his pronouncements, the believer in the infallibility of introspection, mind-reading, and Verstehen (hermeneutic “interpretation”) does not seem to realize how formidable his task is, and how dicey his “conclusions” (conjectures) are bound to be. (This holds in particular for the psychoanalyst, the parapsychologist, the phenomenological psychologist, and the follower of the Verstehen or interpretation “method.”) Indeed, going from behaviour to intention is incomparably harder and riskier than going the other way round. In this process, Missverstehen (misunderstanding) is almost as frequent as Verstehen (understanding),. One reason for such uncertainty is that different people are likely to react differently to the same stimulus: some from self-interest, others from reason; some from habit, others under coercion; some rationally, others under the sway of passion; some out of fear, others out of greed; some for moral reasons, others from hypocrisy; some lucidly, others under self-deception – and so on and so forth. In short, the behaviour–motivation relation is one-to-many. Cognitive neuroscientists have been tackling the Behaviour ® Intention problem for nearly half a century. They have been using such techniques as implanting electrodes on the prefrontal cortex to record the neuronal discharges that are part of the decision-making process. More recently they have also started to study the brain mechanisms that ensue in social behaviour as well as in its inhibition – such as the role of the amygdala and other brain
172
Chasing Reality
“centres” in causing both prosocial and antisocial behaviour (see, e.g., Ochsner and Lieberman 2001, Blakemore et al. 2004). To accomplish this task, the cognitive neuroscientist resorts not only to classical electrophysiological procedures, but also to such non-invasive brainimaging techniques as CT scans and fMRI. Thus, he outreaches not only the social scientist but also the prebiological social psychologist, because he treats mental processes as brain processes. By comparison, the claims of the believer in the power of Verstehen (hermeneutic interpretation) sound like wishful thinking or even shamanism. More on this in the next section. Evolutionary psychologists face a far more daunting task, because emotions and ideas do not fossilize. They proceed in two very different ways: whereas some of them fabricate entertaining stories, others engage in evolutionary neurobiology and comparative psychology research. The former way is unconstrained by hard data, and is therefore popular. It consists in explaining every contemporary pattern of behaviour, social norm, and deviance from the norm as having been “designed” by natural selection (see, e.g., Barkow, Cosmides, and Tooby, eds. 1992). For instance, allegedly we find relativistic mechanics hard to understand because our mind was shaped about 100,000 years ago by the “Pleistocene environment,” when our ancestors had to contend only with slow-moving things such as stones and leopards. The fact that many people can learn and even invent counterintuitive ideas remains mysterious. Serious evolutionary psychology is composed of two main intertwining threads: evolutionary neurobiology, and comparative ethology and psychology. So far, the former has not found much, perhaps because it is basically limited to ascertaining changes in brain size, which is a poor indicator of mental ability. For instance, the brain must certainly grow beyond a certain threshold for certain mental processes to emerge. However, actually human brain size has decreased, not increased, over the past 35,000 years. We still don’t know why. Comparative psychology has been somewhat more successful. For instance, Cabanac (1999) has suggested that emotion is likely to have emerged with reptiles, not before, because lizards, and of course mammals and birds, but not fish or frogs, exhibit signs of emotion when stroked. And de Waal (1996) has conjectured that empathy emerged with the immediate ancestors of both hominids and apes, because all of them empathize, whereas monkeys hardly do so. If these conjectures were to hold up, emotion is likely to have emerged ca. 400 million years ago, and empathy (and cruelty along with it) only about 8 million years ago.
Bridging Fact and Theory
173
6 Social Studies: From Individual to Society and Back The social sciences tackle inverse problems of two main kinds: “inferring” (actually guessing) individual behaviour from systemic social features (such as cultivation, stratification, and dictatorship); and figuring out beliefs, preferences, and intentions from ostensible individual behaviour. The two problems concatenate thus: Collective features ® Individual behaviour ® Belief & Intention Whereas collective features can sometimes be gathered from archaeological remains or social statistics, individual behaviour calls for either direct observation or conjecture; and in turn the attribution of intention demands a questionnaire or conjecture – as well as experiment whenever possible. Again, these are only special cases of problems of the following forms, which occur in all of the sciences and technologies: Observable ® Unobservable, Output ® Mechanism, and Effect ® Cause. There are at least two reasons for the difficulty, ubiquity, and importance of problems of this kind. A first reason is that social scientists have a hard time producing data related to collective features, such as social strife, inflation, and the dominant morality. In particular, “[a]rchaeology is the only discipline that seeks to study human behavior and thought without having any direct contact with either” (Trigger 2003b: 133). A second reason for the difficulty of the problems in question is that to explain a fact – state, event, or process – is to describe how it happens, that is, to unveil its mechanism; and mechanisms are typically unobservable (recall chapter 5). Hence, the inverse problems should not be written off in the name of the empiricist (in particular behaviourist) dogma that all that matters is observable behaviour. All the same, it would be foolish to ignore that problems of this kind are hard, so much so that, in attempting to solve them, we often mistake fiction for fact – as when taking altruism for enlightened self-interest. As we saw in chapter 5, physics contains powerful and generally accepted principles that allow one to solve many direct problems, and that can then be used as guides for solving the corresponding inverse problems. Regrettably, no such guides are available in the social sciences, which still lack theories at once general, precise, and well confirmed. The social scientist must nearly always hazard hypotheses and build theoretical models (specific theories) from scratch. To highlight the predicament of the social scientist, let us propose a few elementary examples.
174
Chasing Reality
First example: Country X has very few hospital beds. This circumstance can be “interpreted” (attributed hypothetically to) either the neglect of public health care or the good health of the population. Second example: There are no peasant revolts in country Y. This non-fact may be “interpreted” as the effect of either harsh repression or light taxes and corvées. Third example: Only half of the electorate bothers to vote in country Z. This fact may be “interpreted” as either contentment or disillusion with the political system. Only further research, making use of additional indicators, could resolve the three ambiguities. A mainstream social scientist is likely to investigate either socially embedded individual actions, such as weddings and burglaries, or entire social systems, such as schools or businesses. By contrast, a doctrinaire methodological individualist is bound to focus on the subjective experiences of the actors, such as the choices and decisions of a businessman. If a hermeneuticist, he will focus on either empathy, like Dilthey, or on intention-guessing, like Weber. (In either case he will engage in psychological “interpretation,” or attribution of mental states, not in semantic interpretation, or pairing of symbols to concepts.) And if a rational-choice theorist, the student will centre his attention on the (alleged) probabilities of events and their attending subjective utilities – on the assumption that all persons act so as to maximize their expected utilities. That is, in any case the individualist will psychologize. He will do it whether, like Comte, Durkheim, Weber, or Popper, he denies emphatically that psychology is relevant to social science; or whether, like Menger, Simmel, von Mises, or Boudon, he invokes a psychology that, being commonsensical, “abstract,” or “rational” (a priori) rather than experimental, is anything but scientific. In either case, the methodological individualist will tend to exaggerate the importance of individual motives and decisions at the expense of natural resources, work, and social structure (see Bunge 1996, 1998, and 1999). And he cannot be taken seriously if he adopts or concocts an “abstract” (or folk) psychology, which is bound to be simplistic, hence unrealistic, because it is centred on cartoons, such as homo œconomicus, homo loquens, homo aleator, or homo ludens, or even on outrageous myths, such as that the behaviour of Jews is characterized by resentment (Nietzsche and Weber). The methodological individualist claims to be able to find out what is usually supposed to be private, namely, the hopes and fears, as well as the goals and intentions, of the decision-makers. In particular, if he is a hermeneuticist, the student will claim to read such subjective features off the actor’s behaviour: he would thus be the privileged owner of a special faculty, Verstehen (understanding or interpretation), unknown to the natural scientist. But how does the hermeneuticist know that he has “interpreted” correctly the
Bridging Fact and Theory
175
data in hand, that is, that he has hit on the true hypothesis? He does not, because he lacks precise and objective standards or criteria to choose among alternative “interpretations” (guesses). Nor is there any chance that he will ever come up with such criteria, because the very concept of an objective and universal standard goes against the grain of any subject-centred philosophy. The hermeneuticist’s choices are arbitrary, because he trusts his intuitions and consequently does not put them to the test. However, this objection won’t worry the hermeneuticists, because they hold that the cultural sciences, unlike the natural ones, do not practise the scientific method, which is centred on the requirement that hypotheses pass some empirical tests before being declared true in some degree. In sum, the hermeneuticist asks us to take him at his word – which is natural enough for an intuitionist but not for a rationalist, who knows that intuition is an unreliable friend. The rational-choice theorist, for his part, does not bother to guess the particular motives of his actors; he claims them to be the same for everybody under all circumstances. Indeed, he decrees that all motives are basically one, namely, to maximize expected utilities. Where the hermeneuticist cooks à la carte, the rational-choice theorist offers a single tourist menu. (The moral philosopher will note the parallel between act-utilitarianism and ruleutilitarianism.) Whether rational-choice theorist or hermeneuticist, the methodological individualist faces the problem of figuring out the beliefs and intentions that may motivate his characters’s actions. In psychological jargon, as we saw above, he attempts to “read” other people’s minds off their behaviour. In other words, he has to invent “theories of mind.” (Methodologically speaking, he transforms the inverse problem Behaviour ® Intention into a family of direct problems of the type Intention ® Behaviour.) But of course “mind-reading” may and often does fail. However, judging from the assurance of his pronouncements, the believer in accurate mind-reading does not seem to realize how formidable and dicey this task is. And yet married people know how many misunderstandings arise daily from the naive belief that they can read their partner’s intentions off their gestures or stray words. In short, the behaviour–motivation relation is one-to-many. Claiming that it is one-to-one is to defy overwhelming adverse empirical evidence, and to oversimplify a huge pile of extremely complex problems. 7 Figuring Out Social Mechanisms Of all the inverse problems in social studies, the toughest are those of figuring out and checking the mechanisms underlying the observed social facts. For
176
Chasing Reality
example, nobody seems to have found the mechanisms of the business cycle. In particular, the neoclassical microeconomists do not have a clue about it, because they focus on free markets in equilibrium operating in a political vacuum – a triple idealization. Two additional reasons that those theorists cannot find the mechanism in question are that their theories seldom if ever involve what physicists call ‘the independent variable,’ namely time, and that they contain only observable variables. This limitation to observable variables is the reason Marx called them ‘vulgar economists.’ Marx’s failure to find what he called ‘the laws of motion’ of the capitalist economy is beside the point. His methodological point, that such laws must exist beneath the phenomena (quantities and prices), was sound. Logic and semantics explain why there can be no straightforward leap from single observable events to general laws. The logical reason is that valid (deductive) reasoning may go from generalities to particulars, not the other way round (except in the trivial case This Þ Some). And the semantic reason is that the high-level concepts occurring in the high-level laws – such as those of equilibrium, efficiency, knowledge, and marginal propensity to save – do not occur in the data relevant to them. (Recall chapter 5, section 2.) The hermeneuticist seems not to realize this elementary methodological point; he believes that he can leap safely from observable behaviour to the wishes or intentions behind it. By contrast, the rational-choice theorist seems to realize that point, since he claims to be able to deduce behaviour from the alleged universal law about maximizing expected utilities. In other words, whereas the hermeneuticist claims to be able to discover causes from their effects, the rational-choice theorist claims to know a priori the putative mother of all effects, namely, self-interest. One of the many problems with the rational-choice theorist is that he takes that alleged law for granted. However, the “law” in question is not universally true (see, e.g., Hogarth and Reder 1987). Indeed, most of us cannot afford the best: ordinarily we must settle for a far more modest goal – for “satisficing” (March and Simon 1958). Moreover, most of us tend to be risk-averse, and so try to avoid losses rather than aim for great gains (Kahneman, Slovic, and Tversky 1982). Consequently, the revealed preferences seldom coincide with the secret ones. Worse, sometimes we are the victims of self-delusion; at other times we do not know what we want; and we seldom know exactly why we act as we do. Knowledge of self is so imperfect, that the ancient Greek adage, “Know thyself,” is still topical. Thus, the claim of both the rational-choice theorist and the hermeneuticist, that they can attain perfect knowledge of other people’s innermost motivations without performing psychological experiments, is comparable to the psychoanalyst’s conceit.
Bridging Fact and Theory
177
Risk of sanctions (Deterrence) Offence Dispositions & social situation
Perceived options & choices Other Temptations & provocations
Figure 7.1MProximate causal factors of unlawful acts (modified from Wikström and Sampson 2003: 122)
The alternative to both hermeneutics and rational-choice theory is the scientific method. A quick study of the familiar yet poorly understood problem of the causes of delinquency should help clarify this point; it may also help design effective crime-prevention policies. We need two models to account for criminal offences: one for individual behaviour and the other to account for criminality as a regular feature of a whole social group, such as the proverbial inner-city inhabitants of the industrialized world, or of the shanty town of the Third World. In other words, we need one model for the proximate causes of crime, and a different one for its distal causes – those that drive an individual to commit repeated offences, or even to adopt crime as a career. I adopt Wikström’s model for the former case. Figure 7.1 exhibits a simplified version of it. This model may be supplemented with one of the distal causes of crime. The basic idea is that criminal behaviour is largely a result of marginality – economic, political, or cultural. Two salient features of marginality are unemployment and physical segregation. Besides, there are two micro-variables that must be reckoned with: anomie – the psychological counterpart of marginality – and solidarity, which compensates to some extent for anomie. I assume that these four variables are related as shown in figure 7.2. The arrows in this diagram symbolize causation. In all cases but the last, an increase in one of the variables causes an increase in the dependent one; by contrast, stronger solidarity decreases both anomie and criminality. Both models can easily be quantitated (see Bunge 2005b). What holds for the basic social sciences also holds, with all the more reason,
178
Chasing Reality
Macro-level
Marginality
Criminality
Ostensive variables
– Micro-level
Anomie
–
Solidarity
Hypothetical variables
Figure 7.2MDistal causal factors of unlawful acts (modified from Bunge 2003a: 209)
for the social technologies, from pedagogy and the law to management science and the so-called policy sciences. In fact, here one faces practical issues of this kind: Given the present state of a social system, and a set of desiderata (or desirable final state), find out the best means for forcing the Initial state ® Final state transition. However, technology deserves a separate section, not only because it is the engine of modern industry and government, but also because the philosophy of technology is even more backward than the philosophy of science. 8 Reverse Engineering The expression ‘reverse engineering’ has spilled over to psychology and other fields. However, technological problems, whether direct or inverse, are very different from problems in other disciplines. They are typically design or maintenance problems, not problems to find something out there or to prove some conjecture. There are two reasons for this difference. One is that technology deals with concrete artefacts such as machines and organizations. (Even software is ultimately concrete since, by contrast to pure mathematics, it guides a sequence of states of a machine.) The second reason is that nowadays the typical technologist is not a self-employed inventor but an employee or a consultant. His employer or customer, far from giving him carte blanche, requests the invention or maintenance of a device performing a given task; he only specifies the desired output and the budget. Indeed, technologists are expected to invent, perfect, or maintain artefacts of some kind, whether inanimate like roads and computers, living like cultivars and cows, or social like corporations and government departments. The more creative technologists invent new artefacts to either meet needs or create wants. Think of designing a bridge capable of taking a given load and subject to given budgetary constraints; or of designing a firm capable of producing or selling a given good or service, as well as of yielding a certain return on a given capital. Both are inverse problems, and they have either many solutions
Bridging Fact and Theory
179
or none, depending on the state of the art, the available resources, and the employer’s goals. However, the inventor of cybernetics asserted that problems of invention may be direct as well as inverse (Wiener 1994: 91ff.). I submit that, whereas the former are problems of improvement on an existing artefact or process, the latter require radical innovation. An improvement problem looks like this: Perfect an existing artefact with a known defect in design or operation. By contrast, a radical invention problem is of either of two types: Come up with a radically new device or find a new use for a given device. (Wiener’s own examples were inventing the vacuum tube for radio transmission, and using it in one of the early computers.) From an economic viewpoint, the difference is this: Whereas improvements are often market-driven, radically new inventions may help create new markets. More precisely, a technological problem of the second type looks like this: Invent a device that will meet such and such specifications. Ordinarily, a problem of this type is harder the broader the specification. A broad specification lists only the essential functions or tasks that the artefact is expected to perform. (For the concepts of function occurring in technology and elsewhere see Mahner and Bunge 2001.) Such specification states nothing about materials or processes. Now, in principle any technological function can be realized by several alternative mechanisms (processes in the system concerned). Just think of the variety of time pieces, calculators, bridges, means of transportation, communications media, health-care systems, and business firms. In other words, the function–mechanism relation is one-to-many, not one-to-one. This leeway gives the designer some freedom. By the same token, it imposes compromises and hard choices. Indeed, typically, the technologist asked to solve an inverse inventionproblem is presented with an input-output pair, and asked to design the box that will transduce the unknown input into the desired output. In electrical engineering, this class of problems has traditionally been known as ‘circuit synthesis.’ This is the converse problem of the analysis of circuits, a comparatively straightforward direct problem that is solved using standard physics and the mathematics inherent in it. By contrast, an inverse or synthesis problem calls for more than physics: it also requires technological imagination, tinkering, and trial and error. Of course, all creative scientists are imaginative. For instance, they try to imagine the constituents of a given or assumed natural thing, and the way such constituents are held together. The engineer’s daydreaming is necessarily different, because he is not given anything, except for a specification and
180
Chasing Reality
perhaps also a box of possible constituents of a system. His task is to devise a new system with existing constituents, or even to invent new constituents, such as nanotechnological circuits and motors. Take applied chemistry for instance. The task of the analytical chemist is to find out the chemical composition of a given lump of matter. This is certainly an inverse problem, but one quite different from that of the pharmacologist intent on designing a new molecule with a given biological effect, such as lowering high blood pressure. Another clear example is this: A physicist can calculate the electromagnetic field irradiated by the electric current in an antenna of a given shape. By contrast, a radio engineer has got to devise a directional antenna (or antennas array) capable of irradiating waves of a given bandwidth to a given place on Earth or in outer space. The biomedical researcher faces a similar problem: Given a diseased organism, design a therapy that will improve the patient’s health. Here again, the problem is to find the most suitable means to attain a given goal. Could reverse-engineering problems be entrusted to a suitably programmed computer? Hardly, because machines cannot invent new ideas; they can only combine given ideas, or rather their physical tokens, in accordance with given rules. For example, from the mid-1970s on, some computer programs have been designed, starting with DENDRAL, that come up with possible molecular structures when fed certain empirical data, such as chemical composition, along with certain constraints, such as valences (see Feigenbaum and Lederberg 1968). Yet Herbert Simon and other computer enthusiasts have claimed that future computers will solve problems and construct theories of all kinds, perhaps with the help of the mythical logic of discovery first announced in the seventeenth century. For instance, it has recently been claimed that the so-called genetic programs can come up with new electronic circuits (Koza, Keane, and Streeter 2003). However, analysis of these achievements show that they have four limitations. First, the programs in question are far less valuable to science than to industry, since they do not contribute to understanding anything. Second, they do not come up with radical (non-combinatorial) novelties. Third, they could not possibly fit such specifications as “A molecule that will affect mood” or “A computer that will solve finite-difference equations.” It still is up to the pharmacologist or the engineer to conduct bench tests to find out what the functions of the novel thing in question are. The fourth fact is that all the programs in question do is to examine or to put together, either haphazardly or according to rule, a large number of given elements, such as resistors, inductors, and capacitances. They do not tell us which configurations are the most suitable for certain goals. Likewise, a calculator can tell us, say, that ten guests
Bridging Fact and Theory
181
can be seated at a table in more than three million ways; but only the experienced hostess will know how to pair off congenial people. In short, combinatorial chemistry and combinatorial electronics are useful but not enlightening. Only theory can enlighten. Let us consider now an example from social technology. In principle, any public-policy goal, such as the containment of an epidemic or population control, can be achieved by different means. These go from quarantine to education to economic disincentives to forced sterilization to massacre. This is a problem of reverse social engineering, as is that of the central banker intent on using fiscal means to tame inflation, enhance consumer confidence, decrease unemployment, win an election, finance a military aggression, or what have you. The case of terrorism is tougher than the above-mentioned cases, because it is less well understood, and what is not well understood cannot be controlled effectively, let alone efficiently. One obvious reason for our deficient knowledge of terrorism is that terrorists are elusive and, when caught, are at best studied only by government psychologists. And these tend to overlook the political context, which is macrosocial. Besides, usually terrorism is handled only by politicians obsessed by the next election and by the military, and the individuals of both classes only know the “tit-for-tat” tactics, the biblical recipe for an unending spiral of revenge. Popular platitudes, such as that the sources of terrorism are poverty and ignorance, are at variance with the evidence: most terrorist are well-educated members of the middle class (Sageman 2004). The practical moral is that raising the standards of living and education of the population where the terrorists are recruited is not the solution. Political terrorism is not only multilevel (micro- and macrosocial) but also multifactorial: political, economic, and cultural. Hence its understanding calls for a systemic approach. In any case, far more serious research is needed, and only a concerted effort by social psychologists and politologists might explain suicide bombing and kamikaze-type attacks. In the meantime, I suggest a combination of two concurrent mechanisms, involving the micro (or individual) and the macro (or institutional) levels to explain that sinister combination of self-immolation and the murder of innocents in the name of a losing cause: Macrolevel
Oppression & Segregation
Microlevel
Anger
Retaliation
Terrorist attack
182
Chasing Reality
9 Bridging Theory to Fact The popular account of the way scientific theories are contrasted to facts is this: Deduce observable consequences, and compare these with the pertinent outcomes of observations. That is, proceed as follows: Deduce observable consequences from theory. Obtain empirical data. Confront prediction(s) with data. Evaluate theory. This schema, called the hypothetico-deductive method, is common to both empiricists (inductivists), in particular logical positivists, and rationalists (deductivists), in particular Popper and his followers. Undoubtedly, it contains an important grain of truth. But it is also severely flawed, because theories do not imply observations without further ado, and consequently they cannot be contrasted directly with the relevant empirical data. A few examples should suffice to make this point. Celestial mechanics cannot predict eclipses by itself; quantum mechanics by itself cannot predict the wavelengths of the light emitted by atoms; psychology remains speculative without behavioural and physiological indicators; and economics can neither explain nor predict anything without a battery of indicators, such as ammonia production, an indicator of agricultural production. Much additional assumption and information are needed in all cases. For example, Newton’s equations of motion do not imply any features of the simple pendulum. To describe the latter we must start by modelling it, for instance, as a point mass hanging from a rigid frame and subject to gravity. Likewise, the calculus of probability is insufficient to make probabilistic predictions; we must add to it a randomness hypothesis, such as that of random mating (recall chapter 4). In general, the actual test procedure is this: Enrich the given theory with a more or less idealized model of the object under study, whether planet, molecule, cell, or social organization. This model is defined by a number of subsidiary assumptions, that is, assumptions not contained in the general theory though compatible with it – for example, that the surface of a solid is smooth, that a wave is plane, that a bacterial colony has an infinite supply of food, that a market is perfectly competitive, or that only the parties directly involved in an international conflict matter. The next step is to relate some of the concepts occurring in the model to the pertinent observable indicators of the unobservable items referred to by the theory – angle of deviation of a pointer in the case of an electric current, bones
Bridging Fact and Theory
183
Table 7.1 Sample of indicators Indicated
Indicator
Gravity Temperature Current intensity Acidity Brain activity Thought Hatred Primate sociality Welfare Height Economic activity Income inequality Political participation
Pendulum period Height of thermometric column Deviation of ammeter pointer pH Glucose consumption Linguistic expression Pupil contraction Grooming time Longevity Nutrition GDP Gini index Voter turnout
and artefacts in the case of an archaeological site, GDP in the case of a national economy, and so on. An indicator is a relation, preferably functional, between an unobservable property U and an observable one O. That is, U = f(O). Reading O on a meter or in a statistical table we can calculate U (see Bunge 1967a and 1973). The function that bridges the observable to the unobservable is hypothetical, and it can be empirical or theoretical. If empirical, it is bound to be unreliable because isolated from the rest. If theoretical, the indicator hypothesis is part of the theory of the measuring instrument – for instance, the theory of the galvanometer. But in either case the observable indicator fact (such as the deviation of a pointer) is likely to be qualitatively different from the indicated fact (such as the electric current). See table 7.1. The third and final step in the testing process is to set up and read the pertinent observation and measuring instrument – scale, microscope, spectroscope, chromatograph, Geiger-Müller counter, posted prices, or what have you. The typical laboratory operation involves measuring instruments with pointers and dials. Such instruments translate unobservable facts, such as electric currents, into observable dial readings: they embody indicator hypotheses. This is when the experimenter starts looking, listening, touching, and perhaps even smelling and tasting, in order to obtain authentic observational data that could not possibly be derived from theory alone. In short, the actual procedure of theory testing is this sequence:
184
1 2 3 4 5
Chasing Reality
General theory & Subsidiary hypotheses Þ Theoretical model Theoretical model & Indicator(s) Þ Testable prediction(s) New empirical data obtained by observation or experiment Confrontation of predictions with empirical findings Evaluation of theoretical model
Figure 7.3 tells all of this graphically. Naive view
Realistic view
Theory o
Theory o
Evidence o
Data o
Indicator o
o Subsidiary Hypothesis
o Theoretical Model
o Prediction
o Observable consequence
o New data
o Observation
* Objective fact Figure 7.3MTwo accounts of the theory-fact confrontation. (a) Naive: direct comparison; (b) realistic: comparison via subsidiary hypotheses, indicator hypotheses, etc. Note that the naive account involves neither objective facts, nor models, nor indicator hypotheses.
10 Concluding Remarks What can we learn from the discussions in this and the preceding chapter? The first thing we learn is that the very existence of the sadly neglected category of inverse problems raises some interesting logical, semantic, epistemological,
Bridging Fact and Theory
185
methodological, and ontological problems. Regrettably, the logic and semantics of problems has hardly attracted the attention of philosophers (see, however, Bunge 1967a and Rescher 2000). In particular, the questions of the order and equivalence of problems should be further clarified. For instance, which comes (logically) first, an algebraic equation or its roots? In scientific practice usually the equation comes first, since one starts by assuming that a given polynomial has zeroes, and proceeds to finding them out. However, the fundamental theorem of algebra allows one to go either way. Hence, in this exceptional case the inverse problem is logically equivalent to the direct one. Moreover, the concept of problem duality (or complementarity) applies here better than the direct-inverse couple. Hence the general problem: What is the difference between problem duality and problem inversion? And under what conditions is one of them reducible to the other? The epistemological problem is to clarify the relation between the directinverse and the observable-unobservable distinctions. Finding the solution to a direct scientific or technological problem boils down to constructing a map from unobservables, such as forces or intentions, to data, such as positions or pieces of human behaviour; and to solve the inverse problem is to find the inverse of this map. This inverse will exist only if the map in question is oneto-one. However, this is not enough, since the empirical data are likely to have errors that may correspond to a large difference in the domain of the map. (That is, a mere empirical error in the measurement of the effect might suggest the wrong cause.) To prevent such instability, the unobservables-observables map must be continuous in addition to being one-to-one (e.g., Newton 1989). Note how closely the mathematical discussion of the invertibility conditions fits the epistemological distinction, and how the innocent-looking observableunobservable distinction can translate into fancy mathematics. The complications in question constitute a particular case of the so-called underdetermination of the theory by the data, which has puzzled inductivists for half a century. The same complications suggest why the “automated discovery systems” project currently under way can yield at most rediscoveries. One of these is that of the quarks, first postulated in 1964, and rediscovered in 1990 by the GELLMANN computer program when fed a large number of theoretical models (Giza 2002). No reason to crow: Once one has found out that the criminal is hiding in the house, and has procured a detailed identikit of him, one is likely to “discover” (recognize) him upon searching the house. The methodological problem is this: When dealing with entities on different levels of organization, which problem-solving strategy is more convenient, top-down (analysis) or bottom-up (synthesis)? The answer is that ordinarily
186
Chasing Reality
Macrofact
Explanandum
Macrofact
Explanans
Microreduction
Deduction
Macroreduction
Deduction
Microfact
Explanans
Microfact
(a)
Explanandum (b)
Figure 7.4MTwo mutually dual and complementary research strategies when only two levels of organization, micro and macro, are involved: (a) top-down and (b) bottomup.
one starts from what one knows best. In other words, the explanandum– explanans relation depends on the state of knowledge. Whereas in some cases the direct problem is of the Macro ® Micro type, in others it is of the converse kind. These dual strategies can be summarized as in figure 7.4 (Bunge 1996: 145). The top-down and bottom-up strategies should eventually be combined. For example, if we wish to explain a known macrofact, we should attempt to reduce it to microfacts; whereas if microfacts are better known, then we should attempt to aggregate them into macrofacts. Thus, at the present time neurons are far better known than neuronal systems. Hence, theoretical cognitive neuroscientists will be well advised to imagine neuronal systems, circuits, or networks capable of performing certain mental functions known from psychology. However, their experimental colleagues will continue to observe and alter multi-neuronal systems to find out what mental functions (processes) they “instantiate” (carry out). Note that the direct/inverse distinction is not a dichotomy, because many problems are neither direct nor inverse. Among these we find the following: the problems of “finding” a new problem; of exactifying a coarse or intuitive notion; of interpreting in factual terms a mathematical concept or a symbol; and of evaluating an item in the light of certain desiderata and rules. Finally, our discussion of the confrontation of a scientific theory with the relevant empirical evidence has involved a brief examination of the indicator hypotheses “embodied” in the very construction and reading of measuring instruments. It is a sad comment on the state of the philosophy of science that such indicators are seldom if ever mentioned, let alone analysed, in the philosophical literature: that most philosophers still believe that scientific theories imply observational data and that these are the direct product of perception. A
Bridging Fact and Theory
187
mere look at a domestic electric meter should have sufficed to explode this myth. If art critics look at artworks, why cannot science critics look at real-life scientific projects instead of relying on second-hand reports? In sum, the old problem of the theory–fact relation still needs much investigating. To be successful, this research must start by recognizing that going from facts to theories is an inverse problem, and that theories are never confronted directly with the relevant observations.
188
Chasing Reality
8 To Reality through Fiction
At first sight, materialists cannot countenance abstract objects, such as theories and myths, because these are immaterial; and realists could not admit them either because they are not out there. In fact, the medieval nominalists, who were at once materialists and realists (in the modern sense of this term), admitted only one kind of entity, the concrete individual – in addition to God and the angels. In particular, Buridan conflated propositions with sentences, and identified these with single utterances or inscriptions, believing that these are just physical objects (see Scott 1966). This is why the modern nominalists use the expressions ‘sentential calculus’ instead of ‘propositional calculus,’ and ‘language’ instead of ‘theory.’ One problem with such rechristening is that linguistic expressions, unlike mere physical entities, are supposed to convey meanings, which are obviously immaterial. A second problem is that sentences are studied by linguistics, which presupposes logic, an eminently non-physical science. As for the reduction of theory to language, it is impossible because theories have non-linguistic properties, such as consistency and explanatory power, just as texts have non-logical properties, such as persuasiveness and style. Besides, we need the concept of a proposition to explicate translation as a meaning-preserving linguistic transformation. Indeed, everyone uses tacitly the convention that two signs are mutually substitutable just in case they designate constructs with the same meaning. In other words, symbols are of course physical objects, but also more than this, since they represent other objects. And a construct cannot be identified with any of its symbols because, by definition, a symbol names something other than itself. Besides, the vast majority of objects must remain nameless: think of continua such as the set of real numbers and the set of points in a
To Reality through Fiction
189
region of spacetime. (The naming or symbolizing function maps designata or denotata into names or symbols, which constitute a denumerable set.) Besides, sentences have structures, such as the SVO word order, which are abstract; and they convey meanings, which are likewise transcendental items. Indeed, symbols and other semiotic objects differ from physical objects in that they come together with explicit or tacit conventional codes or interpretation rules that assign them meanings. This is why we do not discover the signification of a term such as ‘ignis’ by examining any of the ways in which it can be inscribed or uttered, but by consulting a Latin lexicon. And that is also why computers, which process symbols but not their designata or denotata, cannot grasp meanings, let alone create them (see, e.g., Searle 1980 and Bunge 2003a). So, what are materialists and realists, and a fortiori hylorealists, to do about abstract objects, in particular concepts and propositions? Some nominalists, such as Quine, have held that there are signs but not concepts, sentences but not propositions, and behavioural reactions to verbal stimuli rather than meanings. However, this doctrine fails to account for the elementary fact that the two sentences ‘I love you’ and ‘Je t’aime’ are equivalent because they designate the same proposition. In other words, a translation is faithful if and only if it preserves meaning. So, meaning is basically a property of constructs, and only derivatively a property of the signs that designate concepts or propositions – in which case it had better be called ‘signification’ (Bunge 1974a, 1974b, 1975). In short, there are concepts, propositions, and systems of such – though of course as abstract objects not physical ones. The nominalist fears that talk of concepts, propositions, and other abstract objects implies a concession to idealism. The goal of this chapter is to allay this fear, showing that there is nothing wrong with admitting abstract objects as long as they are not assigned independent existence. But before doing this let us see why we need abstract objects. 1 The Need for Abstraction In the preceding chapters we have argued for scientific realism, or the thesis that there is a real outer world, and that it can be known, particularly through scientific research. However, the scientific models of concrete things are symbolic rather than iconic: they are systems of propositions, not pictures. Besides, such models are seldom if ever completely accurate, if only because they involve more or less brutal simplifications, such as pretending that a metallic surface is smooth, a crystal has no impurities, a biopopulation has a single predator, or a market is in equilibrium. These are all fictions. However,
190
Chasing Reality
they are stylizations rather than wild fantasies. Hence, introducing and using them to account for real existents does not commit us to fictionism, just as defending the role of experience need not make us empiricists, nor is admitting the role of intuition enough to qualify as intuitionist. In other words, in science and technology we often use as ifs. That is, we feign that something has a certain property even knowing that it lacks it. When pretending something of the sort we do not lie, and we do not assume that the real nature of the item in question is unknowable. We just make a simplifying assumption, knowing full well that further research or action may disclose previously overlooked factors, and thus force us to adopt a more complicated assumption. For instance, we may conceptualize the surface of a solid body as perfectly smooth, but are ready to concede that actually it is ragged and porous. That is, we do not condone fictionism or instrumentalism – the common ground of conventionalism and pragmatism, according to which all scientific constructs are only fictions allowing us to correlate and organize experiences. The scientific realist merely admits tacitly that neither the knowledge nor the efficient management of a concrete item is likely to be instantaneous. The same holds, with all the more reason, for the “worlds” of mathematics, such as set theory, Boolean algebra, and the theory of differential equations. Indeed, these objects are not out there, ready to be picked up like pebbles; they are made, not found. That is, they are artificial, not natural. Moreover, the mathematical objects are imperceptible and they satisfy purely formal laws, not laws of nature or social norms. For example, it is a theorem in number theory, not a physical or an economic statement, that there is no greatest number. In other words, the statements in pure mathematics are not ontologically objective: they do not refer to the real world. But of course they are not subjective either: they do not report on the speaker’s state of mind. Thus, they are neither objective nor subjective in an ontological sense, even though they are impersonal and asocial. The same holds for myths, stories, and the like. All such statements are fictions. That is, for the sake of analysis we may feign that their referents – sets, ghosts, gods, or what have you – exist independently of the subject. But no sane person will mount an expedition to search for them in the wilderness. Still, no realistic person can deny the power of some fictions, such as those of mathematics in exactifying and unifying ideas, and those of ideology in mobilizing or freezing people. Do you remember the myth of the Iraqi weapons of mass destruction made up by the Bush administration in 2001? Plato and Nietzsche might have approved of this “noble lie” invented to lead millions of politically naive people by the nose. At first sight, the admission of fictions constitutes a serious blow to scien-
To Reality through Fiction
191
tific realism, since fictions, considered in themselves – that is, regardless of the way they are constructed and used – are neither things in the outer world nor brain processes. Is this true? Whichever the answer, we must face this practical question: Which of the various philosophies of mathematics accords best with the thesis that mathematical objects are fictions, without compromising the thesis that only concrete (material) objects are real, or the freedom so dear to creative mathematicians? These questions challenge us to sketch a philosophy of mathematics to supplement our realist philosophy of science and technology (see details in Bunge 1985b). 2 Fictionism If mathematical objects are not real in the same way as stars and birds and schools, then scientific realism does not apply to them, and therefore it cannot help us understand their nature. Could idealism help? Let us see. Platonism, the paragon of objective idealism, claims that ideas are just as real as stars and birds and schools, and actually even more real than the latter. The trouble is that there is no evidence for this conjecture. Nor can the Platonist offer any such evidence, since there are no mathematical quarries or warehouses; all of the mathematical objects are made up. This explains why mathematical discoveries are not made in the field or in the lab. It also contributes to explaining why mathematics, like literature and the law, is only a few millennia old. Still, there is no doubt that mathematical objects are ideas rather than real entities. Could they be fictions, just like Aesop’s stories? Let us see. The thesis that mathematical objects, and perhaps other ideas as well, are fictions, is called fictionism. But of course fictionism, like all the other philosophical isms, comes in two main strenghts: radical and moderate. Radical fictionism (or fictionalism) is the doctrine that all discourse is fictive, so that there is no truth of any kind – mathematical, factual, or other. According to it, the most we can assert is that our ideas work as if they were true. It is thus a combination of conventionalism with pragmatism. Like other epistemological doctrines, fictionism has ancient roots. One of these is ancient radical scepticism, or Pyrrhonism (“Nothing can be known with certainty”). Another is medieval nominalism, a variety of vulgar materialism (“There are no concepts; there are only things and their names”). However, fictionism only flowered in Vaihinger’s monumental book Die Philosophie des als ob of 1911, which owed much to Hume, Kant, Lange, and Nietzsche. (In fact, Vaihinger is usually regarderd as a neo-Kantian.) Fictionism has had few orthodox adherents, but it has left some vestiges in the literature, though seldom under the same name. One of those vestiges is
192
Chasing Reality
found in Milton Friedman (1953), the famous monetarist and apologist of unregulated capitalism. He held that the assumptions of an economic theory need not be true; all that should concern us is whether their consequences match the relevant facts. But of course one may deduce true propositions from false ones, in particular contradictions. (Indeed, a contradiction entails any number of propositions.) Besides, the point of checking for the truth of a logical consequence is to evaluate the truth claims of its premises. Another vestige of fictionism is the view that theories and historical accounts are metaphors rather than representations of real facts (e.g., Hesse 1966, Ricoeur 1975, Hayden White 1978, McCloskey 1985). This opinion is quite popular in the postmodern crowd. An even more popular variety of fictionism is constructivism-relativism, according to which “science is a form of fiction or discourse like any other, one effect of which is the ‘truth effect,’ which (like all literary effects) arises from textual characteristics” (Latour and Woolgar 1986: 184). If truth were just a literary effect, it would hardly matter whether fictionism is true or false. But it does matter everywhere, particularly in science and technology, and this for the following reason. If fictionism were true, there would be no point in checking scientific theories and historical narratives for truth. Nor would it be possible to design empirical tests and artefacts with the help of scientific theories, unless these were at least approximately true. All laboratories, museums, archaeological sites, and the like could be closed down to the applause of all obscurantists. In sum, fictionism is false with regard to factual science. Still, I submit that fictionism, while utterly false regarding factual science, is quite true concerning pure mathematics. In other words, I submit that the mathematical objects, such as numbers, figures, sets, functions, categories, groups, lattices, Boolean algebras, topological spaces, vector spaces, differential equations, manifolds, and functional spaces, are not only entia rationis; they are ficta. Consequently, the concept of existence occurring in mathematical existence theorems is radically different from that of real or material existence. (We shall come back to this subject in section 4.) This is why mathematical existence proofs – and all other mathematical proofs as well – are purely conceptual procedures. In short, mathematicians, like abstract painters, writers of fantastic literature, “abstract” (or rather uniconic) painters, and creators of animated cartoons, deal in fictions. To put it into blasphemous terms: ontologically, Donald Duck is the equal of the most sophisticated nonlinear differential equation, for both exist exclusively in some minds. This, in a nutshell, is the kind of mathematical fictionism to be sketched and argued for in the balance of this chapter.
To Reality through Fiction
193
As will be seen below, our fictionism, unlike Vaihinger’s, is of the moderate rather than the radical kind. This is so because (a) it does not embrace factual science; and (b) it regards mathematics as a science, not as a game, let alone an arbitrary fantasy. Indeed, it distinguishes between mathematical fictions on the one hand and myths, fairy tales, theological speculations, abstract paintings, parapsychological and psychoanalytic fantasies, as well as many-worlds philosophical theories, on the other. 3 Four Kinds of Truth I follow Leibniz (1703: 394) in distinguishing propositions de raison, such as “2 + 2 = 4,” from propositions de fait, such as “Fire burns.” The former refer exclusively to entia rationis, and they can be proved or falsified by purely conceptual means, namely, argument (deduction and criticism) or counterexample. By contrast, the propositions de fait refer at least partly to putatively real (concrete) entities. Consequently, if testable at all, they are checked for either truth or efficiency with the help of direct or indirect empirical operations, such as observation, counting, measurement, experiment, or mere practical trial. The formal/factual distinction calls for distinguishing formal from factual truths or falsities (Grassmann 1844, Einleitung). In particular, we distinguish mathematical theorems, on the one hand, and scientific (e.g., biological) or technological data and hypotheses, on the other. The difference between the two kinds of truth is so pronounced that a factual theory, such as classical electrodynamics, contains some mathematically true formulas – such as those for advanced potentials – that fail to match the facts, that is, that are factually false. A simpler example is the equation “p . q = const.,” which relates prices to quantities: the branch for negative values of p and q is excluded for being economically meaningless. Besides, most mathematical theorems are yet to be employed in factual science or in technology I have just smuggled in the distinction between formal and factual sciences. We define a formal science as an exact science that contains exclusively formal propositions, or propositions de raison. On the other hand, at least some of the propositions of a factual science must be factual: they must describe, explain, or predict things or processes belonging to the real (natural or social) world. Logic, philosophical semantics, and mathematics are formal sciences. By contrast, the natural, social, and biosocial sciences are factual. So are all the technologies, from engineering and nursing to management science and the law. The formal/factual distinction leaves out all the propositions and fields that
194
Chasing Reality
are neither formal nor factual – hence, it is not a dichotomy. Among them the artistic fictions stand out. When reading about Don Quixote’s misadventures, we may feign, and may actually feel, that he exists along with the figments of his own sick imagination. And when attending a performance of Othello we may believe for a moment that in fact the Moor kills Desdemona. However, when reflecting critically upon these and other works of fiction, we do not mistake them for factual accounts – unless of course we happen to be mad. We group them together under artistic fiction. Moreover, occasionally we are justified in talking about artistic truth and falsity, as when we say that Don Quixote was generous and Othello’s suspicion false. In order to establish the artistic truth or falsity of an artistic fiction we only need resort to the work of art in question. That is, artistic truth, like mathematical truth, is internal and therefore context-dependent. In other words, it only holds in some context and it need bear no relation to the external world. To an atheist, the so-called religious truths are parallel: they only hold in holy writ. That is, they call for a deliberate attitude of assent in the absence of any clear independent evidence. Thus, the statement that Christ was human is blasphemy in some theologies and dogma in others. The only evidence accepted by the religious believers is a set of reputedly sacred texts. No data or theories coming from science or technology are regarded as pertinent to religious dogma. In other words, a religion together with its theology is a closed system – unlike a science, which overlaps partially with other sciences as well as with mathematics and philosophy. Allow me to repeat a platitude: Mathematical truth is essentially relative or context-dependent. For example, the Pythagorean theorem holds for plane triangles but not for spherical ones; and not all algebras are commutative or even associative. On the other hand, factual statements such as “There are photons” and “Breathing fresh air is healthy” are absolute or context-free. By contrast, social norms are context-dependent because they are social constructions. Finally, let us admit that the problem of truth, though central in factual science and philosophy, is peripheral in mathematics. As Mac Lane (1986) writes, it is not appropriate to ask of a piece of mathematics whether it is true. The appropriate questions are whether a piece of mathematics is correct, “responsive” (that is, solves a problem or carries further some line of research), illuminating, promising, or relevant (to science or to some human activities). In sum, we distinguish between formal, factual, and artistic truths and falsities. (We may even add moral truths, such as “Racial discrimination is unfair” and “Poverty is morally degrading” but these are irrelevant to our present subject; see chapter 10, sections 6 and 7.) Moreover, from the above
To Reality through Fiction
195
discussion it is clear that, in our view, mathematics is closer to art than to science as regards its objects and its relation to the real world, as well as regards the role of truth. However, as will be argued in section 7, there are important methodological differences between mathematics and art. 4 Mathematics Is Ontologically Neutral To say that logic and mathematics are formal sciences is to say, to borrow Quine’s idiom (though contrary to his claim), that they have no ontological commitment, that is, that they do not assume the existence of any real entities. In other words, logic and mathematics, and a fortiori metalogic and metamathematics, are not about concrete things but about constructs – predicates, propositions, and theories. For example, predicate logic is about predicates and propositions; category theory is about abstract mathematical systems; set theory is about sets; number theory is about integers; trigonometry is about triangles; analysis is about functions; topology and geometry are about spaces – and so on. (Warning: Quine and others misuse the word ‘ontology’ when they equate it to the universe of discourse, or the reference class of a construct. An ontology is not a set of items but a theory about the world.) The thesis that mathematics is about mathematical objects seems selfevident, yet it may not be proved in a general way. What we can do is to support it in two ways: by methodological and semantic considerations. The former is in two stages. A first step consists in recalling that all known mathematical objects are defined (explicitly or implicitly) in purely conceptual ways, without resorting to any factual or empirical means – except occasionally as heuristic devices. The second step consists in recalling that mathematical proofs (and refutations) too are strictly conceptual processes making no reference to empirical data. In this regard, computer-assisted proofs are no different from pencil-assisted ones. As for the semantic consideration, it consists in identifying the referents of mathematical constructs, that is, in finding out what they are about. (For example, set theory is about nondescript sets, whereas number theory refers to whole numbers.) This task requires a theory of reference, as distinct from a theory of extensions. (Unlike the latter, the former makes no use of any truth concept. For example, the predicate “even and not-even” refers to integers while its extension is empty, for it is not true of any number.) To prove the above assertions, I shall presently sketch my own axiomatic theory of reference, which is couched in elementary set-theoretic terms (Bunge 1974a). Let us begin by analysing the general concept of a predicate, such as the one occurring in the proposition “America is populated” or Pa for short. I
196
Chasing Reality
submit that P is to be construed as the function that pairs the individual a to the proposition Pa. This construal differs from Frege’s, of a predicate as a function from individuals to truth-values. I find this absurd because it yields the result that predicating property P about an object a yields not the proposition Pa, but its truth-value. In other words, unlike Frege and his followers, I define a predicate P as a function from individuals, or n-tuples of individuals, to the set S of statements containing the predicate in question. In standard symbols, an n-ary predicate P is to be analysed as a function P : A1 × A2 × ... × An S, with domain equal to the Cartesian product of the n sets Ai of individuals concerned, such that the value of P atin that domain is the atomic statement Pa1a2 ... an in S. For example, the unary predicate “is prime” is a function from the natural numbers to the set of propositions containing “is prime,” that is, P: N S. And the binary predicate “is greater than,” when defined in the same numerical set, is analysable as >: N × N T, where T is the set of propositions involving “>.” I next define two reference functions, Rp and Rs , the first for predicates and the second for statements, through one postulate each: Axiom 1 The reference class of an n-ary predicate P equals the union of the sets occurring in the domain A1 × A2 × ... × An of P, that is, Rp (P) = cAi . Axiom 2 (a) The referents of an atomic proposition are the arguments of the predicate(s) occurring in the proposition. That is, for every atomic formula Pa1a2 ... an in the set S of statements, Rs (Pa1a2 ...an) = {a1, a2, ... an} . (b) The reference class of an arbitrary propositional compound (such as a negation, a disjunction, or an implication) equals the union of the reference classes of its components. (Corollary: A proposition and its negation have the same referents. By the way, the insensitivity of reference to the differences among the logical connectives is one of the differences between reference and extension – two concepts that are usually conflated in the philosophical literature.) (c) The reference class of a quantified formula (that is, one with the prefix “some” or “all”) equals the union of the reference classes of the predicates occurring in the formula. With the help of this theory, one can identify the referents of any construct (predicate, proposition, or theory). For example, since the logical operations (such as conjunction) and relations (such as that of entailment) relate predicates and propositions, they are about them and nothing else. The formal proof for disjunction goes like this: Disjunction can be analysed as a function from
®
®
®
To Reality through Fiction
®
197
pairs of propositions to propositions, that is, w: S × S S. Now, by Axiom 1 above, Rp (w) = ScS = S. That is, disjunction refers to arbitrary propositions – formal, factual, moral, artistic, religious, or what have you. In general, logic refers to arbitrary predicates and their combinations, namely, propositions, classifications, and theories. Obviously, our theory of reference is about arbitrary predicates and propositions. It is then a formal theory just like logic. The same applies to our theories of sense and meaning and, indeed, to our entire philosophical semantics (Bunge 1974a). Of course, no theory of reference can prove that every single mathematical formula refers exclusively to constructs. But ours serves to test any particular claim concerning the reference class of any well-defined predicate. Such test is necessary not only to prove that mathematics is not about the world: it is also crucial in the factual sciences, when their objects or referents are the subject of spirited controversies, such as those elicited by relativistic and quantum mechanics, evolutionary biology, and microeconomics. True, Quine (1969) claimed that reference is “inscrutable.” Donald Davidson (1984) concurred, and added that no theory of reference is needed. But this only shows that their negative semantics cannot be used to find out, say, what the function in quantum mechanics refers to, or whether “thinks,” when used in cognitive psychology, refers to brains, souls, or computers. What about the well-known thesis that “Existence is what existential quantification expresses” (Quine 1969: 97)? This claim is false, as one realizes upon recalling that, unless the context is indicated, the expressions of the forms “ xPx” and “ x (x = a)” are ambiguous, since they do not tell us whether the x’s in question are real or imaginary, that is, whether we are talking about real or ideal existence. Worse, as Loptson (2001: 123) notes, Quine’s view implies that abstract objects are real – which contradicts the standard conception of abstraction as irreality, and thus the opposite of concreteness. To avoid such confusions and contradictions, the ill-named “existential” quantifier should always be completed by indicating the set over which the bound variable in question ranges. The once-standard notation “( x)DPx,” where D names the universe of discourse, will do. (The bounded “existential” quantifier can be defined thus: ( x)DPx = df ( x)(x 0 D & Px).) A modicum of conceptual analysis, used to clarify the ambiguity of the word ‘existence,’ would have spared us Quine’s influential but false thesis of the ontological commitment of the “existential” quantifier, hence of logic and everything built upon the latter. For the same reason, such analysis would have spared us Meinong’s paradox: “There are objects of which it is true that there are no such objects” (Meinong 1960: 83). Much the same holds for the egregious claim that “there are more things than actually exist” (Lewis 2001:
Y
$
$
$
$
$
198
Chasing Reality
86). The same is true of the attempt to revive the nominalist thesis, that “there are” no mathematical entities (Field 1980). All this imprecision and extravagance would have been avoided upon distinguishing explicitly the two kinds of existence, conceptual and real. An appropriate analysis would also have shown that, outside mathematics, the proper interpretation of that quantifier is not “existence” but “someness.” More precisely, in pure mathematics “($x)Px” can be read either as “Some individuals are P’s” or as “There are Ps,” with the understanding that such individuals are conceptual, that is, they exist in some mathematical universe. But outside mathematics it is best to avoid the ambiguity, and read “($x)Px” as “Some individuals are Ps.” In other words, only in mathematics “there are” amounts in practice to “some.” In alternative contexts, for instance, in physics and in ontology, when speaking of existence we may have to make two distinct statements: one of real existence (or inexistence) and the other of someness. (See Bunge 1977a: 155–6 for a formalization of the existence predicate ER and its combination with the someness quantifier $. Such combination allows one to construct well-formed sentences such as ‘Some of the particles imagined by theoretical physicists exist (really)’ – which cannot be constructed with the sole help of $. In self-explanatory symbols: ($x)(Px & ERx).) In conclusion, pure mathematics, in particular logic, is ontologically neutral (more in Bunge 1974c). Moreover, it is a gigantic (though not arbitrary) fiction. This explains why pure mathematics (including logic) is the universal language of science, technology, and even philosophy, and why it is the most portable and serviceable of all sciences. (Take note, science policy-makers and bureaucrats reluctant to fund pure mathematical research.) It also explains why the validity or invalidity of mathematical ideas is independent of material circumstances such as the state of the brain and the state of the nation. However, this subject deserves a new section. 5 Mathematics, Brains, and Society How does mathematical fictionism differ from Platonism? The difference is that the Platonic philosophy of mathematics is part and parcel of objective idealist metaphysics, which postulates the reality of ideas in themselves – that is, the autonomous existence of ideas outside of brains – and even their ontological priority and primacy. By contrast, mathematical fictionism is not included in any ontology, because it does not regard mathematical objects as real but as fictions.
To Reality through Fiction
199
When introducing or developing an original mathematical idea, the mathematician creates something that did not exist before. And, as long as he keeps the idea to himself, it remains locked in his brain – for, as a cognitive neuroscientist would say, an idea is a process occurring in someone’s brain. However, the mathematician does not assign his idea any neurophysiological properties. He pretends that the idea in question has only formal properties: for example, that the theorem he has just proved holds even while he is asleep and that, if made known to others, it will continue to hold long after he is gone. This is of course a fiction, since only wakeful brains can do (correct) mathematics. But it is a necessary fiction because, although proving is a neurophysiological process, a proof in itself, regarded as an abstract metamathematical object, does not contain any neurophysiological data or assumptions. In short, even though theorems are human creations, we may pretend that their form, meaning, and validity are independent of any human circumstances. We are justified in adopting this fiction because mathematical ideas are not about the real world: every mathematical idea refers to some other mathematical idea(s). Mathematics (including logic) is the self-reliant science, even though some mathematicians, such as Archimedes, Newton, Euler, Fourier, von Neumann, Wiener, and Turing were occasionally motivated by problems in physics or in engineering. What holds for brains also holds, mutatis mutandis, for societies. While it is undeniable that there is no (sustained) mathematical activity in a social vacuum, but only in a community, it is equally true that pure mathematics has no social content. If it did have any, then mathematical theories would include socialscience predicates such as “commodity,” “competition,” “social cohesiveness,” “crime,” and “political conflict.” Furthermore, they would double as social-science theories, perhaps to the point of rendering the latter redundant. Our thesis of the factual (in particular social) neutrality of mathematics is at odds with Quine’s semantic and epistemological holism, itself an echo of Hegel’s dictum “Das Wahre is das Ganze” (“Truth is in the whole”). It is certainly true that mathematics permeates, at least potentially, all of human knowledge, which in turn is a system rather than a mere set. However, one can spot the formal components in every discipline, and treat them separately. This is why mathematicians without any specialized extra-mathematical expertise are often consulted by workers in the most varied disciplines. The neutrality view is also at variance with the fashionable social constructivist-relativist sociology of science (Bloor 1976, Restivo 1992, Collins 1998). This school contends that “mathematics is through and through social.” This would be so not just because all mathematical research is conducted in a
200
Chasing Reality
scholarly community, and because the mathematical formulas are communicable, but also because all of the mathematical formulas would have a social content. Needless to say, no evidence has ever been proffered for this thesis; it is just an opinion. However, we can check it in every particular case with the help of the theory of reference sketched in section 3. Take, for instance, the two formulas that define (implicitly) the factorial function, namely, “0! = 1” and “(n+1)! = (n+1) n!”. Obviously they refer to natural numbers: there is no trace in them of the social circumstances surrounding their origin. The formulas in question belong in combinatorics, the discipline that studies the permutations and combinations of objects of any kind. Likewise, Thales’ theorem is about plane Euclidean triangles, not about ancient Greece. And the Taylor series concerns functions, not the events that occurred the year that Brook Taylor published it – among them the death of Louis XIV and the expulsion of the Venetians from the Peloponnese. Because the above mathematical formulas are correct, and neither of them describes any social circumstances, they will hold for as long as there remain people interested in mathematics. (For criticisms of the social constructivist-relativist sociology of knowledge see Boudon and Clavelin, eds. 1994, Bunge 1999 and 2000b, and Brown 2001.) 6 How to Make Ontological Commitments Pure mathematics, then, is not about concrete or material things such as brains or societies, although mathematical objects can only be invented by living brains under favourable social circumstances. Mathematics is exclusively about conceptual or ideal objects. One may prefer to say, as Plato first realized, that mathematics is about changeless or timeless objects, not about events or processes. The ontological neutrality of mathematics explains why this discipline is the universal language of science, technology, and even philosophy – that is, why it is portable from one intellectual field to the next. Yet when looking at any work in theoretical physics, chemistry, biology, or social science, a mathematician may be tempted to regard it as a piece of mathematics. There is a grain of truth in this belief. After all, rate equations, equations of motion, field equations, and more are mathematical formulas, and they are solved using mathematical techniques such as those of separation of variables, series expansion, and numerical integration. This is why Pierre Duhem, a conventionalist, famously claimed that classical electromagnetism is identical with Maxwell’s equations. It is the same reason that many mathematicians used to hold that rational mechanics is part of mathematics, and Gerard Debreu that his theory of general equilibrium is a chapter of mathematics.
To Reality through Fiction
201
However, this is only a partial truth, as can be seen in the case of any formula in elementary logic. Thus, the formula “For all x: if Px, then (If Qx, then Rx)” does not state anything about the real world or even about mathematical objects: it is an empty shell. In order to “say” something definite – true, false, or half-true – the formula must be interpreted. A possible interpretation is this: Int (P) = is human, Int (Q) = thinks, Int (R) = is alive. These semantic assumptions turn the above formula into “Any human who thinks is alive” – a mere generalization of Descartes’s famous Cogito, ergo sum. Any change in the interpretation yields a formula with a different content. Thus, we confirm a conclusion drawn in section 4: that Quine (1953: 12) was mistaken in stating that “the only way we can involve ourselves in ontological commitment” is “by our use of bound values [i.e., quantification].” Actually, the only way to make an ontological commitment is to state a factual interpretation of the signs (words or symbols) in question. This obvious semantic thesis applies to all of mathematics. Indeed, a mathematical formula does not become part of a factual science unless enriched with a factual content. Such enrichment is achieved by pairing the formula to one or more semantic assumptions (or correspondence rules, as they used to be called). Such assumptions state that the formula refers to such and such concrete things, and that at least some of the symbols occurring in it denote certain properties of such things. For example, the equation “d2x/dt2 + ax := 0” is the equation of motion of a linear oscillator in classical mechanics, provided x is interpreted as the instantaneous value of the elongation, t as time, and a as the ratio of the elastic constant to the mass of the oscillator. The concepts of linear oscillator, elongation, elastic constant, mass, and time, which confer a physical interpretation upon the mathematical symbols, are extra-mathematical. In other words, to bridge mathematics to the real world we must enrich it with semantic assumptions (or “correspondence rules”). Thus, the factual counterpart of a predicate is a property, that of cardinality is numerosity, that of continuity is smoothness, that of gradient is slope, and that of Laplacean is the slope of a slope. But of course there are plenty of mathematical constructs without factual counterparts: just think of negation, contradiction, tautology, entailment, 0, and ¥. In sum, mathematics does not suffice to describe or explain the real world. But of course it is necessary to account for it in a precise and deep manner. Indeed, mathematics supplies one of the two components of any theory in advanced theoretical factual science or technology, namely, the mathematical formalism. The other component is the set of semantic assumptions, or “correspondence rules,” that “flesh out” the mathematical skeleton. More precisely, a mathematical theory or model of a domain of factual items
202
Chasing Reality
is a triple domain-formalism-interpretation, or M =for short. Here D denotes the factual domain or reference class, F the union of some fragments of theories in pure mathematics, and Int is a partial function from the formalism F to the power set of D that assigns some of the predicates and formulas in F to sets of factual items in D. (Int is a partial function because not every item in F need be interpreted in terms of facts.) One and the same formalism F may be paired off to any number of different factual domains. (Think, for example, of the multiple uses of linear algebra and the infinitesimal calculus.) A mathematician can check the formal correctness of F, but only empirical tests can tell whether any given theoretical model M matches the domain D, that is, is factually true. In other words, the mathematical truth of the theorems in F does not guarantee the factual truth of M . (On the other hand, any important mathematical flaw guarantees factual falsity.) And yet it is easy to fall into verbal traps, taking literally such expressions as ‘dynamic logic’ and ‘dynamical systems theory,’ which suggest to the unwary that certain mathematical theories deal with time and change after all – whence they are worth being studied and supported by the taxpayer. In fact, those theories are just as timeless as number theory and geometry. What happens is that, when applied, some of the concepts occurring in them get interpreted in factual terms. (For instance, the independent variable is routinely interpreted as time – an eminently nonmathematical concept.) This is how dynamic logic is applicable to computer programs, and dynamical systems theory is applicable to the analysis or design of concrete systems of many kinds – even though it is often but an excuse for studying systems of ordinary differential equations. Likewise, elementary logic can be applied to analysing reasoning processes such as the arguments we construct in real life, by assuming that the steps in a logical sequence match the temporal sequence of our real thoughts. In sum, mathematics supplies ready-made formal and timeless skeletons, some of which scientists and technologists see fit to flesh out (interpret) in alternative ways in order to map concrete changing things. To change the metaphor: From a purely utilitarian viewpoint, mathematics is a huge warehouse of ready-to-wear clothes that scientists, technologists, and humanists can help themselves to. When none of the “existing” clothes fits, the user has got to do the tailoring himself, thus becoming a mathematician for a while. Ptolemy, Newton, and Euler come to mind. So do Einstein, Heisenberg, and Dirac – though only up to a point, because Einstein reinvented Riemannian geometry, Heisenberg reinvented matrix algebra, and Dirac’s famous (or infamous) delta motivated the creation of the mathematical theory of distributions.
To Reality through Fiction
203
7 Responding to Some Objections Let us now address some of the possible objections to mathematical fictionism. One of them is that mathematicians do not invent but discover – or at least this is what they ordinarily say. Some of them may do this out of humility, others for being either Platonists or empiricists. In any event, if that were correct, either Platonism or empiricism would be true, and hence fictionism false. I share the commonsensical view that there are mathematical inventions as well as discoveries. That is, the original mathematician sometimes posits and at other times finds. He posits definitions, assumptions – in particular axiomatic definitions – and generalizations; and he discovers logical relations between previously introduced constructs. In particular, mathematical “entities,” such as categories, algebras, number systems, manifolds, and functional spaces, are invented; there are no mathematical quarries where one can find them ready-made. Even theorems are ordinarily first conjectured, then proved or disproved – sometimes much later and by other people. But the proof process consists in discovering that the new theorem follows from previously known assumptions – axioms, lemmas, or definitions. (However, sometimes such discovery necessitates inventing further items, such as the auxiliary constructions of elementary geometry.) For example, using a convergence test one discovers that a certain infinite series is divergent. But all convergence tests have been invented, not discovered. Likewise, by expanding a function in series, and integrating term by term, one may compute the integral of the function (provided the series converges uniformly). But the given function, the concepts of integral, and much more had to be invented before they could be handled. The general rule is: First invent, then discover. And if the discovery is negative – for instance, that a bunch of assumptions is inconsistent, that they do not entail an alleged theorem, or that a series is divergent – then alter some of the assumptions; that is, revise the invention process. Since in mathematics there is invention as well as discovery, Platonism is false. But evidence for mathematical invention is not enough to substantiate fictionism. Yet we only need to recall that we handle an infinite totality, such as a line or a hypothetico-deductive system, as if it were an individual, and as if it were “there” all in one piece – without of course saying where “there” is. (Obviously, mathematicians are in spacetime, but they assume that their own creations are out of spacetime.) A second possible objection goes as follows: If mathematics is a work of fiction, why don’t librarians group mathematical works together with novels
204
Chasing Reality
and, in particular, with fantastic literature? In other words, what if any are the differences between mathematical fictions and artistic ones? For example, how does the fundamental theorem of algebra differ from the claim that Superman can fly or Mickey Mouse can speak? Because mathematical fictions are unlike all others: they are strictly disciplined, and they can be used to think about any epistemic subject. More precisely, I submit that the crucial differences between mathematical fictions and all others are the following (adapted from Bunge 1985a: 39–40): 1 Far from being totally free inventions, mathematical objects are constrained by laws (axioms, definitions, theorems); consequently, they cannot behave “out of character” – for example, there can be no such thing as a triangular circle, whereas even raving mad Don Quixote is occasionally lucid. 2 Mathematical objects exist (ideally) either by postulate or by proof, never by arbitrary fiat. 3 Mathematical objects are either theories or referents of theories, whether in the making or full-fledged, whereas myths, fables, stories, poems, sonatas, paintings, cartoons, and films are non-theoretical. 4 Mathematical objects and theories are fully rational, not intuitive, let alone irrational (even though there is such a thing as mathematical intuition). 5 All mathematical statements must be justified in a rational manner–either by their premises or by their fruits – not by intuition, revelation, or experience. 6 Far from being dogmas, mathematical theories are based on hypotheses that must be repaired or given up if shown to lead to contradiction, triviality, or redundancy. 7 There are no strays in mathematics: every formula belongs to some system (theory), and in turn theories are linked together forming supersystems, or else they are shown to be alternative models of one and the same abstract theory; thus, logic employs algebraic methods, and number theory resorts to analysis. On the other hand, artistic or mythological fictions are self-sufficient: they need not belong to any coherent system. 8 Mathematics is neither subjective like art nor objective like factual science. It is ontologically noncommittal; but the process of mathematical invention is subjective, and that of proof (or disproof) is intersubjective. What is real (concrete) about mathematics is that it is produced by living mathematicians embedded in mathematical communities. 9 Some mathematical objects and theories find application in science, technology, and the humanities.
To Reality through Fiction
205
10 Mathematical objects and theories, though produced and consumed in society, are socially neutral, whereas myth and art are sometimes used either to support or to undermine the powers that be. 11 Because it deals in timeless objects, correct mathematics does not age, even though some of it may go out of fashion. 12 Mathematics is a science. That is, its practitioners employ the scientific method: Background knowledge Solution candidate Checking Re-evaluation of either problem or background knowledge.
®
®
®
A practical upshot of the preceding is that librarians have good reasons for placing mathematical fiction and literary fiction in different sections. After all and above all, mathematics is a science – nay, the old Queen of the Sciences. A third possible objection to fictionism is this: If mathematics is made up exclusively of ficticious and timeless objects, how can it represent real things and processes? The answer lies of course in the concept of symbolic representation. Even ordinary language allows us to form sentences designating propositions that may represent ordinary things and processes, even though such sentences do not resemble their denotata, as in the case of books and the word ‘books.’ The semantic relations involved are those of designation (D ) and reference (R ): Sentences
D
Factual propositions
R
Facts
The key to the reference component of meaning is of course a system of linguistic (in particular semantic) conventions, mostly tacit. In the case of scientific theories, the semantic components are not conventional and therefore irrefutable and changeable at will; instead, they are hypothetical. (This is why the name ‘semantic assumption’ is more adequate than the traditional ‘correspondence rule.’) Thus, the semantic assumption in Einstein’s theory of gravitation, that the value of the metric tensor at a given point in spacetime represents the intensity of the gravitational field at that point, is not a convention, or even a rule: it is a hypothesis that can be put to the experimental test – and, moreover, one that makes no sense in action-at-a-distance theories of gravitation. However, the matter of convention deserves a separate section. 8 Conventionalism and Physicalism Fully fledged fictionism is conventionalist: it holds that all assumptions are conventions, and none are true. By contrast, the moderate fictionism I am advocating is not conventionalist, because it retains the concepts of formal
206
Chasing Reality
truth, as well as the distinction between assumption and convention (in particular definition), which conventionalism rejects. Mathematicians use the concept of mathematical truth when claiming that a certain formula (other than an axiom) is true either because it is satisfied in some model (or example), or because it follows validly from some set of assumptions in accordance with the inference rules of the underlying logic. Likewise we use its dual, namely, the concept of mathematical falsity, when disproving a conjecture by exhibiting a counterexample. In other words, ordinarily theoremhood equals either satisfiability or provability, neither of which is conventional. (The adverb ‘ordinarily’ is intended to remind ourselves of Gödel’s proof that a formal axiomatic theory may contain formulas that cannot be proved with the sole resources of the theory in question.) The dual of the concept of truth, namely, that of falsity, is equally important in mathematics. It occurs whenever a mathematical idea is criticized as flawed for some reason, and it is central in approximation theory and numerical analysis. Think of approximating an infinite convergent series by the sum of its first few terms. (In particular, one can make true statements about quantitative errors – for instance, that they are bounded, or that they fit some distribution.) Conventionalists have no more use for the concept of error than for that of truth. What about the axioms of a mathematical theory: can one say that they are true? One might be tempted to claim that they are “true by convention.” But this expression strikes me as an oxymoron because conventions, such as definitions, are checked for well-formedness and convenience, not truth. In any event, there is no need to assign truth-values to the axioms of a mathematical theory. (By contrast, we need to know whether the axioms of a factual theory are true or false to some extent.) In practice, the most important piece of knowledge about a postulate system is that it entails the standard theorems and, preferably, some new and interesting ones as well. (To be sure, it would also be important to be able to prove the consistency of the system. But here again Gödel has taught us some humility.) As noted above, a second difference between moderate mathematical fictionism and conventionalism is that the former keeps the difference between assumption and definition, whereas conventionalism holds, in Poincaré’s famous thesis, that “axioms are disguised definitions.” The falsity of this view is best seen in the light of the theory of definition and, in particular, in the light of Peano’s thesis that definitions are identities – which most mathematical assumptions are not. Indeed, an axiom may be an equality, such as a differential equation, but not an identity, such as “1 = the successor of 0.” (Identities are symmetrical, whereas equalities are not. This difference is sometimes indi-
To Reality through Fiction
207
cated by the symbols ‘=’ (identity) and ‘:=’ (equality) respectively. Examples: “For any real number x: (x+1)(x–1) = x2 – 1” and “For some complex number(s) x: ax2 + bx + c := 0” respectively. A particularly shallow version of mathematical conventionalism is Carnap’s view that mathematical statements are “empty linguistic conventions.” Actually the only conventions of this kind that occur in mathematical texts are the notational conventions, such as “Let n designate an arbitrary natural number.” The well-formed mathematical statements are at once meaningful and testable. Indeed, they refer to definite mathematical objects, such as numbers, figures, functions, spaces, or what have you, and they have precise contents or senses even when their referents are abstract or nondescript. If the (well-formed) mathematical formulas were meaningless there would be no way of checking them, and no point in doing so. But they are checked in several ways: definitions for non-circularity, axioms for fertility, conjectures for theoremhood, and so on. (Further objections to Carnap’s conventionalism in Quine 1949, and Gödel’s in Rodríguez Consuegra 1992.) In conclusion, mathematical conventionalism won’t do. In any event, mathematical fictionism differs from it and has no use for it. Vulgar materialists, whether physicalists or nominalists, particularly if they are mathematically illiterate, are bound to reject mathematical fictionism for regarding it as a variety of idealism. For instance, Vitzthum (1995: 146), who favours the nineteenth-century mechanistic materialism popularized by Karl Vogt, Jacob Moleschott, and Ludwig Büchner, regards fictions as constituting an “antiworld.” He claims that my view is “a perfect vacuum, null, a mirage,” and holds that I fail to see “the idealist, countermaterialist quicksand” into which the doctrine leads. Presumably, the nominalists would react in a similar manner, for they reject all concepts: they only admit inscriptions – which is why they prefer to speak of numerals rather than numbers, and of sentences rather than propositions. Neither physicalists nor nominalists offer a viable philosophy of mathematics; the former because they overlook the fact that mathematical objects have no physical, biological, or social features. And nominalism is false if only because there are not enough names to call even a tiny part of the “world” of mathematics, such as the real numbers lying between 0 and 0.001, which constitute a non-denumerable infinity. Besides, constructs, unlike their symbols, have formal structures. Actually mathematical fictionism is neutral in the debate between materialism and idealism. In fact, mathematical fictionism is only about mathematical objects; it makes no assertions about the nature of the world. As far as mathematical fictionism is concerned, one may hold the world to be material,
208
Chasing Reality
spiritual, or a mixture or combination of material and ideal objects. Materialists will reject fictionism out of hand only if they confuse materialism with realism. A materialist should not feel uneasy about the thesis that mathematical objects are ideal and therefore timeless, as Plato first observed, as long as he subscribes to the thesis, shared by mathematical intuitionists like Brouwer and Heyting, that mathematics is a human creation. All the reassurance the materialist needs is that mathematical ideas do not exist by themselves in some Platonic Realm of Ideas both immaterial and eternal. (If he so wishes, he may regard any construct as an equivalence class of brain processes occurring in different brains or in the same brain at different times: see Bunge 1983.) Yet it might be argued that mathematics is “ultimately” about the real world for, after all, arithmetic originated in counting concrete things like shells and people, and geometry originated in land surveying and, in particular, in the need to allot land lots to farmers. No doubt, such were the humble origins of the parents of mathematics. But mathematics proper is not about such empirical operations as counting and surveying. In fact, mathematics freed itself from its empirical origins about three millennia ago, when the Sumerians proposed the first general mathematical propositions and proofs. A cognate argument for both vulgar materialism and empiricism is that mathematicians often make use of analogy or of ordinary (incomplete) induction to find patterns, as Polya (1954) showed persuasively. True, but the result of any such plausible reasoning is a conjecture that has got to be proved (or disproved) by purely mathematical means – for example, by reductio ad absurdum or using the principle of complete induction, neither of which is suggested by ordinary experience. In short, analogy and induction by enumeration have at most a heuristic value: they prove nothing, and proving happens to be the main job and sole privilege of the mathematician. Moreover, incomplete induction and analogy may lead us to error unless we check their outcome. Undoubtedly, factual science has sometimes stimulated mathematics by posing new problems. For example, dynamics encouraged, even required, the invention of the infinitesimal calculus and the theory of differential equations; Einstein’s gravitation theory stimulated the growth of differential geometry; and quantum mechanics that of functional analysis, group theory, and the theory of distributions. But none of this proves that mathematics is about the real world, because a mathematical formula can be given alternative factual interpretations – or none. Besides, nowadays mathematics is far more useful to science and technology than the latter are to the former. In the modern intellectual production line the main arrow goes from the abstract to the concrete.
To Reality through Fiction
209
Finally, another familiar objection to the autonomy of mathematics is the idea that some mathematical theories are more “natural” than others, for being closer to human experience. For instance, according to mathematical intuitionism in its radical version, no mathematical formula is “meaningful” unless it is somehow related to the natural numbers (Dummett 1977). But in fact there are plenty of non-numerical mathematical fields, such as logic, category theory, set theory, much of abstract algebra, and topology. Besides, the mathematical intuitionists have yet to produce a suitable semantic theory elucidating the notion of meaning, and thus helping us test formulas for meaningfulness (in their sense). Hence, strictly speaking, what they say about meaningfulness is meaningless. In conclusion, mathematics is semantically and methodologically self-sufficient. But at the same time it feeds all of the factual sciences and technologies, and is occasionally stimulated by them, so much so that mathematics is at the centre of the system of human knowledge, which may be pictured as a rosette of partially overlapping petals. In other words, far from being independent from the rest of knowledge, mathematics is at the very centre of it. But this does not entail the thesis – held consistently by Quine, and at times also by Putnam and others – that there are no important differences between mathematics and the rest, and that in principle mathematics might be refuted by experiment. Like spouses, mathematics and science are neither identical nor separate. So much for serious fictions: those that constitute knowledge and help advance knowledge. Let us now turn to a fashionable philosophical jeu d’esprit, namely, many-worlds metaphysics. 9 Metaphysical Fictions: Parallel Worlds Folklore and religion are full of impossibilia, from pathetic ghosts to awesome deities and fearsome devils. Fiction writers, painters, cartoonists, and filmmakers have imagined further impossible beings, such as Aristophanes’ reasoning birds, and worlds where the laws of physics and biology do not hold. Metaphysicians too have imagined impossible entities, such as Descartes’s demon and Kripke’s parallel universes – though never in such minute and delightful detail as artists such as Hieronimus Bosch or Maurits Escher, Anatole France or Italo Calvino. In particular, the contemporary philosophers who have written about possible worlds have been deliberately vague about these, perhaps because they were only interested in very general statements. However, they all agree in showing little interest in the only world that science regards as possible, that is, the one that satisfies the (known) laws of nature or
210
Chasing Reality
society. Let us therefore start by distinguishing the two concepts of possibility in question: conceptual and real (see also Bunge 2003a). It is often assumed that, whereas science accounts only for the actual, artists and metaphysicians can afford to speculate about possibilities. The first part of this opinion is mistaken, because every scientific law covers not only actuals but also possibles – possible things, events, processes, and other factual items. Thus, the great physicist Ampère (1834: 198) stated that mechanics applies not only to terrestrial bodies and machines, but also “to all possible worlds.” However, these are of course exclusively those that satisfy the laws of mechanics. By contrast, the possible worlds referred to by many-worlds metaphysics and semantics are quite different from those of science and technology: the former need not be constrained by natural laws. Amazingly, one of the wildest fantasies about parallel worlds is the brainchild of a physicist, Hugh Everett III (1957), and his thesis adviser, John A. Wheeler. This is the many-worlds interpretation of quantum mechanics. According to it, every calculated possibility is realized in some physical world. Thus, if an electron can have three different energies, each of them with some probability, you will measure one of those energy values, a copy of you in an alternative world measures the second value, and your second copy the third value. The most obvious objections to this ingenious fantasy are that it violates all the conservation laws, and that it postulates inaccessible worlds, each of which is created by an experimenter, earthly or other-worldly. No wonder that the many-worlds interpretation is generally regarded as a jeu d’esprit that does no work. Yet, at the same time it is treated as a piece of serious science, rather than pseudoscience, because it is formulated in mathematical language and it avoids certain weird features of the standard theory, such as the instantaneous collapse (or reduction) of the state function upon measurement – which is like preferring leprechauns to the ghosts said to haunt Scottish castles, because they do not drag noisy chains. Another curious development has come recently from the discovery that the universe is globally flat rather than curved. According to Einstein’s theory of gravitation, this result implies that the universe is spatially infinite (see, e.g., Tegmark 2004). Now, in an infinite universe “ruled” throughout by the same laws, any known material system is likely, if not bound, to occur at infinitely many places. In other words, there must be infinitely many nearly identical copies of the known part of the universe. For instance, there must be someone like you, though perhaps with a different haircut, reading a nearly identical copy of this very book. This is not the place to discuss the merits of this view. But one thing is clear: the alleged parallel universes cannot be other than very
To Reality through Fiction
211
distant regions of the universe, namely ours. If reality were multiple rather than one, it would be unknowable. But of course fantasy worlds are far more attractive to idealist philosophers than to scientists. Thus, the father of phenomenology, and grandfather of existentialism, held that the real world is only “a special case of various possible worlds and non-worlds,” all of which are in turn nothing but “correlates of consciousness” (Husserl 1931: 148). The plurality of worlds follows of course from the basic assumption of subjective idealism, that the world is somehow secreted by the subject. And we are supposed to know that this is so by intuition, which, according to Husserl (ibid.: 145), is infallible. However, the current vogue of possible worlds does not stem from phenomenology – so much so that David Lewis (1986), the most famous of contemporary possibleworlds metaphysicians, called himself a materialist. According to possible-worlds metaphysics (e.g., Kripke 1980), all of the conceivable universes, however weird, are equally possible, and moreover equally real, just because they are conceivable. Modal logic condones this fantasy, because it does not distinguish between logical and real possibility, and because its modal operators “possible” and “necessary” operate on propositions, not facts. By contrast, in science, technology, and everyday life possibility and necessity are predicated exclusively of factual items, as when we say that it is possible for humans to be vegetarians, but necessary for tigers to be carnivorous. Since modal logic does not distinguish between conceptual and real modalities, it is only a game with fake diamonds (logical possibilities) in paper boxes (logical necessities). Yet, as I write, modal metaphysics is mainstream metaphysics (see, e.g., Laurence and Macdonald, eds. 1998, Lowe 2002, and Loux and Zimmerman 2003). Still, this theory is open to a number of obvious objections. One of them is that, unless we are prepared to fail utterly in real life, we must be able to distinguish between dreamed and real possibilities. If anything were really possible, then nothing would be impossible, and thus the word ‘possible’ would be emptied – just like the word ‘normal’ in psychoanalysis, according to which nobody is normal. A second objection is that in science “really possible” is coextensive with “lawful,” whereas in mathematics it amounts to “noncontradictory,” in technology to “feasible,” in the law to “legal,” and so on. Since the ontological concept of lawfulness and the praxiological concept of feasibility are alien to modal logic and to the many-worlds fantasies, this theory is useless for discussing the prospects for scientific, technological, or social advancement. Besides, many-worlds metaphysics is not intended as a serious ontology, for it tells us nothing, or at least nothing both interesting and correct, about being
212
Chasing Reality
or becoming, space or time, causation or chance, life or mind, society or history. It cannot even be used by experts in the supernatural. In fact, it has been ridiculed by some of them, because it “distort[s] the notions of being and creation” and is “stuffed with infinities of abstracta that no one seriously believes to be real” (Ross 1989). The theories in question are only intended as tools for semantics. Indeed, they are supposed to allow one to state, for instance, that tautologies are true in all possible worlds (that is, regardless of what happens), whereas counterfactuals hold only in some of them. Let us glimpse at the latter. A counterfactual (or contrary-to-fact) sentence has the form “If A were the case, then B would be the case.” The occurrence of the words ‘were’ and ‘would’ hints that neither A nor B is actually the case. Hence, the given sentence is quite unlike the conditional “If A is the case, then B is the case,” or “A B” for short. Whereas ordinarily the latter may be true or false to some degree, the corresponding subjunctive conditional can be neither. Hence, it falls beyond the reach of logic. However, the modal metaphysicians have invented a trick for transforming a counterfactual into a declarative sentence designating a proposition. The trick consists in pretending that the non-fact in question occurs in a dream world. Thus, it is posited that the sentence If A were the case, then B would be the case. (1) is equivalent (in some unspecified sense of ‘equivalent’) to If A is the case, then B is the case in world W. (2) Of course this is not a logical equivalence, because sentence (1) does not stand for a true or even a false proposition, since it neither states nor denies anything. Moreover, whereas in affirming (1) one suggests that neither A nor B occur in the real world, when asserting (2) one may take refuge in an unreal “world,” much as a madman escapes the harsh facts of the real world by pretending to be God or at least Napoléon Bonaparte. In other words, if something is either impossible or false in the real world, one invents a topsyturvy (yet otherwise non-descript) “world” where the impossible is possible, the false true, and evil good. Such emigration from the real world to Wonderland is the gist of an entire semantic theory, called possible-worlds semantics, as well as of an entire metaphysics, namely, possible-worlds metaphysics. These theories are said to elucidate a number of notions, such as those of disposition, lawfulness, and causation (see, e.g., Kripke 1980, Lewis 2001 and 1986, Lowe 2002, and McCall 1994). The advantage of such elucidations, is that they exempt the philosopher from learning how those important philosophical concepts are actually used in science and technology. Let us glance at some of those
Þ
To Reality through Fiction
213
brave attempts to face philosophical dragons with the handy if unreal sword, or rather scabbard, of conceptual possibility. Consider this well-worn example: This portion of cyanide has not killed anyone yet, but it would kill anyone who were to swallow it. That is, cyanide is potentially lethal: it has the dispositional property of being capable of killing, much as that heap of table salt has the disposition of dissolving in water, or that powerful government of conquering an oil-rich country. It is only natural for the ordinary-language philosopher to elucidate such dispositions in two steps: This is toxic = This would kill anyone who were to swallow it. (1) This would kill anyone who were to swallow it = There is a World W such that in W this kills anyone who swallows it. (2) What has been gained? Nothing, since we already knew that the world W in question is the real world. We have no idea of any “worlds” where cyanide is not toxic, fundamentalists are tolerant, empires do not wage wars, and so on. Besides, what we really need to know is, rather, why cyanide is lethal in the only world there is. Metaphysicians cannot solve this problem, but toxicologists can: they can unveil the biochemical reactions that kill an organism poisoned with cyanide. What they do is to analyse the disposition “toxicity” in terms of actual properties of the organism-cyanide system. They tell us that cyanide is toxic because it binds strongly with hemoglobin, thus depriving the rest of the organism of oxygen, which in turn is necessary for life. Physicists elucidate in like manner such dispositions as elasticity, potential energy, solubility, and refractivity. For instance, they do not state that the gravitational potential of a body suspended at a certain height above the surface of the Earth is the kinetic energy that it would acquire if it were released. Instead, they assert that the said gravitational potential equals the actual body weight times its height. They also tell us that, as the body falls, this potential energy is transformed into kinetic energy. Besides, they tell us precisely what the value of the latter is: they can calculate and measure it. Furthermore, physicists caution that the formulas in question only hold for small bodies (idealized as point particles) and small heights (compared with the Earth’s radius). Compare the precision, richness, and testability of this bit of knowledge with the poverty of the possible-worlds analysis. We shall return to this subject in chapter 9. This is not to deny that we need fictions to do realistic science; on the contrary, we need plenty of them. Scientific fictions concern entities, properties, or experiments. For example, the atomic hypothesis was a grand fiction of the first type (about things) before any empirical evidence for it was produced more than two millennia after it was imagined. And the most common scientific fiction of the second type (about properties) is approximation, such as the
214
Chasing Reality
assumption that a given system is closed, that the interaction among its components is weak, or that its laws are linear. Weber called such idealizations ‘ideal types.’ Their occurrence necessitates the use of the concept of partial truth. For instance, Ohm’s law is only partially true at low temperatures. Literally, the truth-value of this law-statement decreases with temperature – a thought that is likely to induce a chill in anyone who believes that propositions are eternal objects, or at least born fully true or false. Finally, a number of imaginary (or thought) entities and experiments have been invented with several purposes: dialectical, heuristic, critical, didactic, or rhetorical. For example, Descartes imagined a malicious demon intent on deceiving him. The demon succeeded in everything except concerning Descartes’s own existence: “There is no doubt that I exist, if he is deceiving me.” Maxwell illustrated the second law of thermodynamics imagining a “being” capable of directing the molecular traffic from a cold body to a warm one. This would have falsified the said law – but at the price of introducing the supernatural. Popper (1959b: 442–52) performed a magisterial analysis of some of the most influential thought experiments in the history of physics, from Galileo’s to Bohr’s. He emphasized correctly that imaginary experiments prove nothing, though they may illustrate, disprove, suggest, or “sell” hypotheses. For example, Bohr’s clock-in-a-box imaginary contraption had only what Popper called an apologetic use. In addition, it induced Bohr to postulate a false if often quoted formula, namely, the so-called fourth indeterminacy relation between energy and time. Mathematicians and philosophers too need fictions. For example, to use the truth-table decision procedure one pretends that truth-values are intrinsic and eternal, while actually the truth value of any proposition about facts only emerges (or submerges) from tests (Bunge 2003a). And to introduce the general concept of a property, we may have to start off with the concept of a nondescript substantial individual or substrate, onto which we may subsequently stick such labels as “massive” and “inhabited” (Bunge 1977a). But these fictions are not idle, and they can be removed when no longer required. Which is not the case with the fictions that occur in speculative metaphysics. 10 Concluding Remarks We have outlined and defended one version of mathematical fictionism. This is our alternative to Platonism, nominalism (formalism), intuitionism, conventionalism, and empiricism. According to moderate fictionism, all mathematical objects are fictions, whence they are neither empirically compelling nor logi-
To Reality through Fiction
215
cally necessary (even though they are constrained by logic). Those constructs are fictions because, although they are human creations, they are deliberately detached from physical, personal, and social circumstances. We pretend that those timeless ideal objects exist in a “world” (conceptual system) of their own, along with other fictions, such as myths and fables – which, however, have no ontological standing. In asserting the ideal existence of such fictions, we do not distance ourselves from reality; we just construct propositions that do not refer to the real world. Even children learn early on the difference between story and fact. Our version of fictionism is not only confined to mathematics. It is also moderate, because it is at variance with conventionalism, and it regards mathematics as a science, not as a grammar, much less as a game or pastime on a par with chess. It differs from conventionalism in assigning conventions a rather modest role in mathematics by comparison with hypotheses, proofs, and computations. And it regards mathematics as a serious activity that enriches our stock of ideas, and puts them at the disposal of factual science, technology, and even philosophy. In particular, mathematics helps us hone ideas of any kind, and discover structures and patterns, as well as pose and solve problems of all kinds in all fields of knowledge and rational action. Think of the central role that mathematics plays in modern engineering. Thus, far from being escapist, mathematical fictions are necessary to understand and control reality. If moderate mathematical fictionism is true – that is, if it fits mathematical research – then mathematical Platonism, conventionalism, formalism (nominalism), empiricism, and pragmatism are false even though they are the most popular of all the philosophies of mathematics. Platonism is false, or at least untestable, because fictions are human creations, not self-existing eternal objects. And mathematical empiricism is utterly false because most mathematical fictions go far beyond experience, and none are tested by it. The same holds, a fortiori, for mathematical pragmatism – the philosophy of mathematics tacitly espoused by those who demand that mathematics and the other basic sciences should be responsive to the market. As for mathematical intuitionism, it is inadequate because it restricts severely the invention of mathematical fictions on doubtful philosophical grounds, such as constructivism and the Pythagorean worship of natural numbers. (Caution: This criticism concerns mathematical intuitionism, not intuitionist logic or mathematics – see Bunge 1962.) Nominalism (or fomalism) is false too if only because – as Frege pointed out – it rejects concepts and confuses signs with their designata, for instance, numerals with numbers. (Not even the self-styled nominalist, or formalist, can help using notational conventions,
216
Chasing Reality
such as “Let N designate the set of natural numbers,” that pair concepts to symbols.) Finally, conventionalism is false for a different reason, namely, because it conflates assumption with definition, and it jettisons the concept of mathematical truth. This concept is needed in at least three circumstances: when checking a calculation, proving a theorem, and showing that an abstract theory is satisfied by a “concrete” one (or model). Furthermore, if mathematics (including logic) is the most exact and comprehensive of fictions, then we need no special logic of fictions, such as those of Woods (1974) and Routley (1980). In particular, the free logics (that is, logics with empty domains), deliberately invented to handle fictions, turn out to be unnecessary. This remark may be useful to curb the inflationary growth of deviant logics, the vast majority of which are mere academic pastimes “only good to get promotions,” as Hilbert might have said. (See Haack 1974 for a superb survey.) What about the holy trinity of foundational studies, namely, logicism, formalism, and intuitionism? It may be argued that, contrary to received wisdom, these are not philosophies of mathematics but foundational strategies (Bunge 1985a). Mathematical fictionism has little to say about any of them. Indeed, in principle one may couple mathematical fictionism to either logicism, formalism, or moderate (post-Brouwerian) intuitionism. Alternatively, one may latch it onto the combination of all three strategies, which seems to be tacitly favoured nowadays by most working mathematicians (Lambek 1994). Moderate mathematical fictionism ought to exert a liberating influence on the mathematical researcher in reminding him that he is not out to discover a ready-made universe of ideas, or to explore the real world, or even to latch on to natural numbers. His tasks are to create (invent or discover) mathematical concepts, propositions, theories, or methods, and to discover their mutual relations, subject only to the conditions of consistency and conceptual fruitfulness. As Cantor said, freedom is essential to mathematics. I submit that the main duties of a philosopher of mathematics are to defend the freedom of mathematical creation from philosophical strictures; to see to it that mathematical fictions are not reified; to use some of them to elucidate, refine, or systematize key philosophical ideas – that is, to do some exact philosophy; and, above all, to build a philosophy of mathematics matching actual mathematical research as well as a science-oriented general philosophy. And as semanticists and metaphysicians, our duty is to face the fact that the conceptual and practical problems of the only world there is cannot be tackled successfully by asking for asylum in fantasy worlds. In sum, we invent fictions in all walks of life, particularly in art, mathematics, science, and technology. Whereas some fictions are disciplined, others are
To Reality through Fiction
217
wild. A disciplined fiction abides by the laws contained in an exact theory. Science, technology, and serious philosophy use only disciplined fictions, particularly mathematical ones, such as the concepts of set, equivalence relation, function, and consistency. The rightful place for wild fictions is art – or incontinent philosophy.
218
Chasing Reality
9 Transcendentals Are Of This World
Anything not encountered in experience is usually called ‘transcendental.’ Like all negative definitions, this one is ambiguous. Indeed, according to it, intuitions and a priori ideas, along with universals and theological items, qualify as transcendental. For example, the idealist philosophies of Plato, Leibniz, Hegel, Fichte, Bolzano, and Husserl are transcendental because they do not rely on experience. By contrast, Berkeley’s, Hume’s, Comte’s, Mill’s, and Mach’s philosophies are not transcendental because of their empiricism. And Kant’s philosophy is only semi-transcendental because, though aprioristic and intuitionist, it is also semi-empiricist on account of its phenomenalism. From a practical viewpoint, transcendental items, or transcendentalia, may be partitioned into two classes: useful and useless. Thus, the concept of class, kind, sort, or species is necessary in all walks of life, whereas that of parallel universe has at best a heuristic function. This chapter deals with a miscellany of transcendentals of both kinds, particularly the concepts of universal, species, possible world, disposition, spacetime, and liberty. No attempt will be made to organize this collection. The point of the present exercise is to clarify the concepts in question from the point of view of scientific hylorealism and, by the same token, to clear away a number of misconceptions about them. The net result will be that, whereas some transcendentals are indispensable to account for reality, others only block the view of it. 1 Universal The word ‘universal’ designates a general predicate, such as “concept” and “number,” as well as a frequently occurring property, such as “material” and “live.” Idealists, from Plato onwards, have ignored the predicate-property difference; they have also postulated that universals exist by themselves and
Transcendentals Are Of This World
219
precede particulars, which they regard as mere “instantiations” (examples) of the former. This view, idealist (or Platonic) realism, is still popular. For example, it is inherent in contemporary statements that brain systems “instantiate,” “mediate,” or “subserve” mental abilities, and that the latter have “multiple realizations” (in brains, computers, ghosts, and deities), as Fodor (1981) and other philosophers of mind have claimed. According to materialism, in particular nominalism, all real existents are singular. As the putative father of nominalism put it, “every thing outside the soul is singular” (Ockham 1957 [ca. 1320]). This is why we can only experience particulars. In other words, the phenomenal world, just like the physical one, is composed exclusively of individuals. The nominalists (or vulgar materialists), from Ockham and Buridan to Quine and David Lewis, reject universals altogether, not only as concepts but also as features of material individuals. They postulate that the only real existents are bare individuals, that is, concrete entities devoid of properties other than the ability of joining with other individuals. Hence, they dream of a science without properties (e.g., Woodger 1952). But that would be no science, because scientific research is essentially the search for laws, and laws happen to be invariant relations among properties (see Bunge 1967a). Furthermore, nominalists equate properties with classes of such individuals. Thus Lewis (1983: 164): “To have a property is to be a member of the class ... The property of being a donkey, for example, is the class of all the donkeys.” In obvious symbols: D = {x Dx}. But of course this definition is circular. Besides, when forming a (natural) class or kind, we start with the defining properties. In the example at hand, we may set (in ordinary-knowledge terms) D = Equid & Smaller than an Arabian horse & Long-eared & Stubborn. Since nominalists go from individuals to classes to properties, they are bound to form artificial classes rather than natural kinds. Hence, they are unlikely to understand such expressions as ‘body,’ ‘field,’ ‘chemical species,’ ‘biospecies,’ ‘origin of species,’ or ‘social class.’ And, since they countenance neither predicates nor propositions, the nominalists cannot help us analyse them either. Although we cannot sense universals, we could not possibly think or speak without them. This is why even the nominalist Ockham admitted “universals by convention,” that is, words that are universal because they can be predicated of several particulars. As Russell (1912) pointed out, it is impossible to form even the simplest sentence without employing words designating universals. Thus, the trivial egocentric statement “I want it” is constituted by words designating universals. Indeed, there are as many “I”s or selves as human beings, every one of whom wants a number of things. Likewise, “it” is a blank )
220
Chasing Reality
or index to be filled by different items in different circumstances. Therefore, nominalism does not even hold for languages although it attempts to replace concepts with their names. Nominalism fares even worse in mathematics and in the factual (“empirical”) sciences, for all of them attempt to subsume particulars under universal laws, which are the universals par excellence. These laws are either universal generalizations, or statements asserting the existence or non-existence of individuals of a given kind, such as “zero,” “black hole,” and “hominid.” Moreover, no physical or chemical law-statement concerns observations, whereas all of these are observer-bound and couched with the help of phenomenal predicates. And the basic law-statements in physics are required to be the same relative to all reference frames (hence observers in particular) of a certain kind (i.e., inertial), whereas the results of observations of certain properties, such as particular positions, frequencies, and energies, are frame-dependent. Hence, the nominalist program, of either reducing universals to particulars or dispensing with them altogether, cannot be carried out. Rather, on the contrary, many particular concepts and words can be characterized in terms of universal concepts or words. For example, “my daughter” describes a unique individual in terms of two universals. However, the latter are not regarded as independently existing ideas. Modern science too rejects the independent existence (ante rem) of universals. However, scientists look for the universal and lawful in the particular and contingent. As Levins and Lewontin (1985: 141) famously put it, “Things are similar: this makes science possible. Things are different: this makes science necessary.” The philosophical problem is of course that of the nature of universals and their relation to particulars. Are universals self-existing ideas (Platonic “realism”), properties shared by many individuals (scientific realism), or just sounds or inscriptions (nominalism)? In scholastic jargon, are universals ante rem, in re, or post rem? This problem comes from antiquity and is still with us. One of the most interesting, yet least well-known, solutions to it is that of the Jaina school, which has been in existence for two and a half millennia. According to the Jains, everything has a general as well as a particular aspect (Hiriyanna 1951: 65). For example, a particular cow is characterized by cowness, which it shares with all the other cows; but it also has certain properties that are special to it. In other words, the universals inhabit concrete individuals. At about the same time, Aristotle defended the view that the universals, far from being ante rem, as his teacher had held, are in re. Nearly one and a half a millennium later, Averroës expressed essentially the same view (Quadri 1947: 209-11). In this, as in other regards, modern science has vindicated Aristotle’s moderate position concerning universals. Indeed, it is a tacit postulate of
Transcendentals Are Of This World
221
modern science that all properties are properties of something or other: there are neither properties in themselves nor entities devoid of properties. The same problem resurfaced in the nineteenth century in relation to the social sciences: are they confined to particulars (idiographic), or do they also attempt to uncover patterns (nomothetic)? Arguably, the social sciences, just like the natural ones, are nomothetic as well as idiographic (Bunge 1996). For example, the historian looks for the common features of all military aggressions – and finds them in the wish to appropriate something, such as territory, natural resource, labour force, or trade route. And criminologists look for patterns behind particular cases of law-breaking – for instance, the weakening of social bonds (social control theory), anomie (social strain theory), association with delinquents (social learning theory), marginality (systemism), or what have you. According to scientific realists, the word ‘universal’ denotes both (a) a property that, like mobility, life, and being yours, is possessed by several items; and (b) the concept(s) representing that property. That is, universals are assumed to exist both through concrete individuals and as abstract ideas – in neither case in and by themselves. In Hegelian jargon, there are universals of two kinds: abstract and concrete. The former are post rem, the latter in re. For example, logical consistency is an abstract universal because it only applies to sets of propositions, in particular theories. By contrast, income inequality is a concrete universal because it is inherent in all stratified societies. However, there are several measures of the degree of that inequality, every one of which is an abstract universal. (Recall the difference between properties and the predicates representing them: chapter 1, section 4.) This view of universals, in particular laws, is close to Aristotle’s, and is part of hylorealism, or materialist realism. Not surprisingly, this view is unpopular among contemporary philosophers. In particular, Popper, Quine, Goodman, and Gellner called themselves nominalists just because they opposed Platonic “realism,” and agreed that only individuals exist – without however caring to clarify whether they meant abstract or concrete individuals. As hinted above, even particulars may be regarded as instances of universals. For instance, 7 is a number, Fido is a dog, and being 30 years old is a particular value of age. In general, functions subsume particulars under universals. More precisely, a particular property may be regarded as just a special value of a general property represented by a mathematical function. For example energy, the most universal of all non-conceptual properties, is classically representable as a function from concrete things to numbers. Even proper names may be regarded as particular values of the naming function. This function maps people into names; that is, N : P N, where N designates the
®
222
Chasing Reality
set of names, complete with surnames, patronymics, matronymics, and toponymics to make them unique. Here, the members of the set N are particulars, whereas N and P are universals. (More precisely, in classical physics the energy of a concrete thing c relative to a reference frame f, at time t, and calculated or measured in the unit u, can be written E(c, f, t, u) = e, where e is a real number. This formula makes it clear that energy is a property representable by a function, not a thing designated by a name. In the previous example, the function E maps the Cartesian product C × F × T × U into the real line. Drop c or f, the bearers of energy, and no energy remains. And dispense with energy, and no physics remains.) In mathematics, both classes and their members exist, albeit only fictitiously (recall chapter 8). By contrast, the chemical, biological, and social classes exist only as concepts (or fictions), whereas their members are material existents. However, far from being arbitrary, the classes in question are the extensions of universalia in re, such as bivalence, bipedalism, and poverty. Thus, Humankind (H ) is the class of all the individuals possessing the complex property H of being human: H = {x Hx}. How do we know that these are universalia in re rather than either ante rem or post rem? Because they are representable as predicates whose domains contain classes of concrete things. Equivalently: because the corresponding predicates refer to material entities. That is, the function H maps the set M of material items into statements of the form “x is human.” (See Bunge 1974a for this non-Fregean construal of predicates.) The above functional analysis of the universal–particular relation helps analyse the thesis that there can be no historical laws because history does not repeat itself. In fact, although no two historical events are identical, any two historical events have something in common – for instance, that they are basically economic, political, cultural, or all three. For example, although the two world wars were quite different, both were military conflicts embracing nearly the entire world. In other words, although taken as a whole every historical event is unique, and thus defies classification in all respects, when analysed it is seen to share some features with other events: it is a combination of concrete universals. The same holds, mutatis mutandis, for all facts. We may surmise that even free will is lawful, as suggested by the fact that it can be removed surgically (lobotomy). In other words, scientists assume that all facts are lawful. This explains why we find laws in all domains – provided we try hard enough. We tend to take laws for granted, both as objective patterns and as the conceptual representations of the latter. We do this to the point of adopting, at least tacitly, the lawfulness principle, according to which all facts satisfy some laws. However, the ontological status of laws is far from clear, largely because of the idealist confusion between objective patterns and the formulas whereby )
Transcendentals Are Of This World
223
we attempt to represent them. Take, for instance, the chemical laws, that is, the laws that “govern” the formation and dissociation of molecules. Did these laws exist before the first chemical compounds emerged? A Platonist, such as a medieval “realist,” would answer in the affirmative: the first chemical compound was just an instantiation of a chemical universal that literally governed the process in question. By contrast, a nominalist would hold that the laws in question appeared only post rem, in the mind, by way of abstraction from phenomena. That is, he would claim that nature is contingent, and that patterns are only in the mind. Finally, a scientific realist is likely to argue that the chemical laws inhere in chemical processes, so that they first emerged along with the earliest chemical compounds, when the planets cooled sufficiently. In general, the natural laws are objective patterns of being and becoming (Bunge 1959b). To conclude this section: We cannot dispense with universals anymore than we can dispense with phenomenal concepts. We need the latter to describe experience, and universals to explain it. Only a divine intellect could dispense with qualia and their corresponding concepts; and only lowly organisms make do without universals, such as “reachable” and “edible.” Presumably, a puppy forms gradually the universal “meal” as it finds food in its dish, and the universal “people” as it meets individual humans. No dog could survive on a nominalist diet, for it would be unable to (a) generalize individual experiences; (b) learn rules, such as “Bark at strangers,” compliance with which puts food in his dish; and (c) distinguish the different properties of one and the same thing, such as “round,” “soft,” “rollable,” and “bouncing” applied to a ball. Besides having some universal concepts, puppies and all other things have universals in re. For example, they and all the other multicellular organisms satisfy the logistic law of body growth. In short, universals are indispensable because they are both in the world (as properties) and in theories (as predicates). The nominalist’s world of bare (property-less) individuals is just as fictitious as Plato’s Realm of Ideas. Neither helps understand, say, that buoyancy is a property of boats but not of electrons, let alone of numbers. Failing to see the universality inherent in every particular is just as mistaken as failing to identify what makes an individual unique. In other words, in trying to understand things and ideas, we cannot dispense with the concept of a kind or species. Which brings us to the next section. 2 Kind Contrary to Plato, Aristotle held that only individuals may be real, that all classes or species are in the mind. The medieval nominalists, such as Ockham
224
Chasing Reality
and Buridan, went further: they had no use at all for the concept of a species or natural kind, because they rejected universals and regarded the cosmos as a collection of mutually independent items (see, e.g., Gilson 1947). Consequently, for a nominalist all classifications, even the periodic table, are purely conventional. The fathers of evolutionary biology, from Bonnet and Buffon to Lamarck and Darwin, concurred: to them, in nature there are only individuals, species are constructs, and classifications conventional. But Darwin contradicted himself in stating at once that species are conventional, but phylogenies, which are relations among species, are natural. Indeed, since humans and monkeys have common ancestors, all three classes in question, though not material entities, must somehow represent real evolutionary processes. Whereas some classes are artificial, others are natural, that is, they represent objective commonalities among existents. For example, there is nothing arbitrary about the periodic table of the elements. And, while both the Linnean and cladistic classifications are natural, the latter is generally deemed to be more adequate than the former. Nevertheless, whether artificial or natural, every class is a construct. Hence, the expression ‘origin of species’ must be interpreted elliptically, as short for “origin of organisms belonging to new species.” In other words, speciation, the emergence of new species, is a real process on the organismic level, and extinction is parallel, but in both cases what emerges or submerges are individuals. What happens before speciation and after extinction is that the corresponding species are empty collections. Scientists ignore the nominalist ban on species. Regardless of his philosophical allegiance, the very first thing a scientist does is to heed Aristotle’s advice: Identify the object of study in terms of species and specific differences, as in “unary predicate,” “continuous function,” “higher mammal,” “business enterprise,” and “developing country.” That is, scientists take it for granted that all the objects they study group into kinds or species that are natural rather than artificial or conventional. As the gruesome idiom has it, they attempt to “carve the universe at its joints.” Moreover, scientists formulate generalizations involving the species concept, such as “Birds of a feather flock together” and “All metals are opaque.” What holds for things also holds for facts, such as states and changes of state. Every real fact is singular, but at the same time it belongs to some class or other, namely, the one “covered” by the law(s) it abides by. Shorter: Facts are particular but not unique, because they all fall under some law or other. Even the origin of our solar system, though unrepeatable in details, is but an instance of the class “origin of planetary systems,” a class characterized by certain astrophysical laws. The main philosophical problem about species is, What is a species? That is,
Transcendentals Are Of This World
225
how is it defined and what kind of object is it: conceptual, factual, or – as Leibniz thought – semi-mental? To answer these questions, let us start by recalling how a species is ordinarily defined. Depending on the richness of our background knowledge, we can resort to either of two methods of definition: by property or by equivalence relation. In the first case, we pick on a salient property, such as atomic number, which we represent by a predicate P. Next we find out the extension of P, that is, the collection of all the Ps, or members of the universe W of objects, that possess the property represented by the predicate P: S = E (P) = {x Î W Px}. The defining predicate P is usually the conjunction of a number of simpler predicates. When a single predicate is at stake, one gets the rather weak notion of a kind, such as age or income group. The far stronger concept of a natural kind is defined in terms of laws: it is the collection of all the things that possess properties that are lawfully related to one another. Example: “Electromagnetic field = Whatever satisfies Maxwell’s equations or their quantum-theoretical generalization.” Shorter: A kind is a natural kind if all its elements meet at least one law. However, we do not always know the relevant laws, particularly at the start of a new research line. In this case we resort to mere similarities, equalities in certain respects, or equivalences, such as having the same sex, age, occupation, or income bracket. Thus, we state that, although Nelson Mandela is unique, in one respect he is in the same league as all the other freedom fighters. These individuals constitute the class F = {x Î H x ~f m} where ~f designates ‘is equivalent to with respect to f’ (freedom fighting), and m abbreviates ‘Nelson Mandela.’ The two methods of class formation falsify, among others, the following rather popular theses. One is the opinion that it is impossible to pigeonhole complex things, such as people, mental states, and social facts. Actually it is perfectly possible to do so in some respects, though not in all. Another related misconception, currently popular among philosophers, is that biospecies are concrete individuals on a par with their constituents, that is, individual organisms (e.g., Ghiselin 1974, Hull 1976). The reason usually given is that species are said to evolve. But of course biopecies, being sets, not aggregates, are unchanging: only their members (organisms) and aggregates of them (populations) can and do change. Still, a species may be regarded as a higher-order individual, or system, if taken together with its structure, which includes the relation of common descent. Thus, one stipulates that any two )
)
226
Chasing Reality
Ecosystem o
Things
Population o
oS
o S
System of species
Species
Constructs
Organism o Figure 9.1MConcrete individuals on 3 different levels: organism, population, ecosystem. Abstract individuals on 2 levels: species, and system of species.
members a and b of S, however different in details, are equivalent in that they have a common ancestor. That is, we state that a ~ b = df $c (Dac & Dbc). We next construct the system of akin species S = < S, ~ >. Thus, we have introduced two concepts, S and S . Only the members of S, and the populations of S, are concrete. The former are first-level, and the latter second-level systems. In turn, every population is a part of some community or ecosystem. See figure 9.1. If species were concrete individuals, it would be possible to gather or hunt, cook, and eat some of them rather than some of their members. Biospecies are not individuals but collections; and, at any given instant, a collection is a set proper (see Bunge 1979a and 1981c and Mahner and Bunge 1997). Proof: All biospecies are defined as sets defined by predicates. (For example, the standard if flawed definition used in biology is this: “Biospecies are groups of potentially interbreeding organisms, which are reproductively isolated from other such groups.”) The same holds for chemical species, such as Hydrogen and Carbon, as well as for social species, such as Rural and Urban, or Employer and Employed. In all cases, the individual–species relation is the 0 (membership) relation, not to be mistaken for the part–whole relation that obtains, for instance, between cell and organ, or word and text. This is why one can copyright texts but not words, nor even entire vocabularies: whereas the former have owners, the latter belong to linguistic communities. 3 Possibility Possibility is the most slippery of ontological categories, for it lies between being and non-being, present and future, datum and conjecture. No wonder that some philosophers have denied real (or ontological, or objective) possibility. For instance, Kant restricted possibility to experience: there are only possible experiences; moreover, this is what things would be – namely, bundles
Transcendentals Are Of This World
227
of possible experiences. (Hence, no percipients, no things.) Likewise, modal logic treats “possibly” as an operator acting on propositions, not on facts (see, e.g., Hughes and Cresswell 1968). And possible-worlds metaphysicians analyse “p is possible” as “there is a world at which p is true.” However, they do not bother to characterize such “worlds” in a precise manner. Since some of the “worlds” in question violate the known laws of nature, the corresponding conceptual possibility may be really impossible. We looked at these unscientific fantasies in chapter 8, section 9, and will examine them further here in sections 4 and 5. Let us now focus on the serious problem of the concept, or rather concepts, of real possibility. Let us start by recalling briefly how the concept of real possibility is used in the sciences and technologies. To begin with, the reality of possibility is made quite clear in the state-space representation of things of any kind. Indeed, every point in such an abstract space is deemed to represent a really possible (that is, lawful) state of the thing in question, whether atom, organism, ecosystem, social system, or artefact (recall chapter 1, section 1). However, even when no such representation is available, the following varieties of real possibility are generally admitted outside mainstream metaphysics, which is predominantly speculative rather than scientific Natural possibility = Compatibility with the laws of nature Social possibility = Compatibility with the norms prevailing in a given society Technical possibility = Feasibility Economic possibility = Profitability Political possibility = Accessibility of public office Moral possibility = Not infringing on any basic rights of anyone Legal possibility = Compatibility with the ruling legal codes Epistemological possibility = Knowability with the means in hand Alethic possibility = Plausibility in the light of the background knowledge Methodological possibility = Testability None of these ten concepts is elucidated by modal logic, because all of them involve non-logical concepts; hence the irrelevance of modal logic to real life, science, and technology. Worse, mixing the various concepts of possibility in the same context involves confusion and it may lead to error or paradox. For example, “You can (physically) kill me, but you may (morally) not kill me.” Formalizing factual and conceptual possibility with the same possibility operator (the diamond symbol of the modal logics) leads to trouble. One solution would be to introduce ten diamonds with the same syntax but different semantics. However, in this case one should also introduce forty-five bridges between these different possibilities, such as “Whatever may be done can be
228
Chasing Reality
done (but not conversely).” Moreover, one could then speak of ten different “worlds,” and of the forty-five bridges between them. But this solution would drive Ockham up the wall. Why not dispense with modal logic altogether, since it can solve no philosophical problems because it is too coarse to distinguish among utterly different kinds of possibility? (More in Bunge 1977a, 2003a.) To sum up, since we admit the category of real possibility, we must enlarge the concept of reality, and with it our worldview, to include possible facts along with actual facts, which in turn can be either necessary or contingent. That is, we must adopt the following partition (Bunge and Mahner 2004: 109):
Causal (or conditional) possibility Real possibility Chance (or unconditional) disposition
Reality
Necessary facts Actuality Contingent facts 4 A Surfeit of Worlds Nearly everyone agrees that there are possibilities and impossibilities in addition to actualities. For example, it is possible that you will make a new friend, but impossible that your health will improve upon smoking. On a more technical level, quantum mechanics specifies that certain transitions are possible and others “forbidden,” and thermodynamics proves the impossibility of engines of certain kinds. In addition to asserting such real possibilities and impossibilities, physicists resort to virtual items such as imaginary points, virtual displacements, and their kin. In fact, in physics, “virtual” means conceptual possibility, as opposed to real (or physical) possibility. This distinction occurs in the very formulation of the highest-level principles in physics, namely, the variational ones. The form of all such principles is “dA = 0,” where d designates a conceptual variation, such as an imaginary displacement, and A names a basic physical property, such as an action (or energy times time). The formula states that A is an extremal (maximum or minimum); or, equivalently, that A is conserved under the variations in its independent variables (such as coordinates and velocities); or that the real trajectory of the thing of reference is the one, among all the conceivable trajectories, along which A is extremal, usually minimal (see, e.g., Lanczos 1949). That this is the correct interpretation is seen by
Transcendentals Are Of This World
229
examining the way dA is calculated. Indeed, this calculation obeys the laws of the calculus of variations, a branch of pure mathematics, where nothing ever happens. Interestingly, the condition “dA = 0” implies the equations of motion (or the field equations), which describe the real trajectories (or the field propagations). That is, a statement of conceptual possibility (though one concerning a physical magnitude) implies a statement of actual occurrence. All that is standard if somewhat heady scientific fare. What about pigs flying, people communicating through telepathy, or government “of the people, by the people, and for the people”? Here is where the paths of philosophers diverge. Science-oriented thinkers regard such events as impossible, for violating certain natural or social laws. By contrast, the modal metaphysicians, such as Meinong (1960), Kripke (1980), Routley (1980), David Lewis (2001, 1986), and their numerous followers regard them as possible, or even necessary, in alternative worlds – much as the theologians claim that bliss, while impossible in this world, is bound to be experienced in the next. Modal metaphysics is not just an ignorable extravagance; it has become part of mainstream philosophy (see, e.g., Laurence and Macdonald, eds. 1998; Lowe 2002; Loux and Zimmerman, eds. 2003; and Melia 2003). We glimpsed at this theory in chapter 8, section 9. Let us now take a closer look at it. Obviously, the choice between the two alternatives noted above depends critically upon the meanings assigned to the words ‘world,’ ‘possible,’ and ‘true.’ On a broad construal of these terms, the above-mentioned possibilia are such in some sense (logically or physically) and in some world or other, whether real or imaginary. For example, in Cockaigne, the land of plenty, everyone lives in luxury without working. Obviously, this is impossible in the actual world, and in the particular sense in which the expression ‘real possibility’ is used in science and technology, namely, as compatibility with the known laws of nature. But the whole point of speculative philosophy is to work with the sole restriction of logic. To mark the contrast between conceptual and real possibility, consider the ways that existence hypotheses are proved (or just confirmed) and disproved (or just undermined). In mathematics, existence proofs, if correct, are definitive. Thus, it has been proved once and for all that there is no such thing as the greatest number, and that there are infinitely many prime numbers. By contrast, real existence is ordinarily proved by exhibiting a specimen of the hypothesized entity, or at least by showing reliable indicators of it – such as brain waves in the case of neural activity and fossils in the case of extinct vertebrates. In some cases, a theory suggests the existence or non-existence of real things or events of certain kinds. Thus, classical electromagnetism suggested the existence of electromagnetic waves long before they were produced
230
Chasing Reality
in the laboratory; and according to Einstein’s theory of gravitation there must be gravitational waves. Physical existence can be suggested by theory, but it can only be proved by experiment or fieldwork. Characteristically, the so-called modal metaphysicians are not interested in the real world that scientists and technologists study; they enjoy imagining worlds where nearly anything is possible. Still, there are a handful of different schools of modal metaphysicians, some more tolerant of unbridled imagination than others (see Lycan 1998). The radical wing of modal metaphysics is constituted by people who assert the real existence of multiple parallel universes inaccessible from one another. Paradoxically, they call themselves ‘modal realists.’ In fact they are ontological relativists, and as such irrealists. By contrast, thinkers in the moderate wing are content with asserting that “things could have been different in countless ways,” and call them misleadingly ‘possible worlds.’ Still other metaphysicians, notably David Lewis, waver between the two wings. Thus, he would have claimed that a “world” in which Hitler won the war is very different from the world we know. But he also wrote about a world “at” which heat is the defunct caloric rather than random molecular motion (Lewis 1983). An exact philosopher would regard the first case as an equivocation, while a scientist would dismiss the second as idle extravagance. Regardless of the wing to which they belong, all modal metaphysicians start from the semantics for modal logic proposed by Kripke four decades ago. The reason is that the only justification for talking of possible worlds is the need to equip modal logic with a semantics. A realist philosopher would have thought that modal logic is not worth saving at the staggering price of adding uncounted and barren dream worlds to the real world. He would have wondered why anyone would mourn the loss of a theory that had failed to deliver the goods that Clarence I. Lewis promised nearly one century ago, namely, the simultaneous elucidation of the notions of possibility and logical necessity, both de re and de dicto (see Bunge 1977a). At all events, modal semantics purports to explicate the modalities “possible” and “necessary” by reducing them to the concepts of world, existence at a world, and truth. Thus, it is posited that, for any proposition p, “Possibly p = df there is at least one world at which p is true,” and “Necessarily p = df p is true at every world.” Thus, “Everyone lives without working” is deemed possible because somebody dreamed up Cockaigne. A roughly equivalent formulation of the same idea is counterpart theory (Lewis 1986). It states that individual x possibly has property P if and only if x has a “counterpart” in a universe spatiotemporally isolated from ours. For
Transcendentals Are Of This World
231
example, though unknown to you, you may be an alchemist if in an alternative universe there is an individual closely resembling you, except that he engages in alchemy. These counterparts are real according to some modal metaphysicians, notably Lewis, whereas they are abstract or “semi-abstract” according to others. (I confess that the notion of semi-abstraction is above my head.) The historical root of all these elucidations is Leibniz’s (1703) important and correct distinction between vérités de raison and vérités de fait, and his assertion that the former, unlike the latter, are true in all possible worlds. This famous metaphor makes literal sense in Leibniz’s theodicy. (When pondering about the kind of world He should create, God could choose whatever laws of nature He pleased, but He could not go against logic; in particular, He could not contradict Himself without risking Leibniz’s contempt.) This metaphor is pedagogically effective. But, if taken literally, the possible-worlds elucidation of the modalities is open to the following technical objections. To begin with, this allegedly sophisticated explication of modality is actually coarse, because it is unrelated to the concepts of real (nomic) possibility and necessity used in science, from physics to population genetics and the sociology of networks. In particular, the above construal makes no contact with the scientific concepts of disposition (such as solubility) and chance. Worse, none of the three definientia – the concepts of world, existence, and truth – is clearer than the corresponding definienda. It is an obscurum per obscurius explication, like defining the number 1 as 00, or as the negative of the square root of the imaginary unit. It is also objectionable to make possibility in our world dependent upon actuality in a parallel world. The reason is that, since that fantastic world is inaccessible from ours, it cannot be inspected to check whether, in fact, it contains the desired “counterparts.” Hence, we could never ascertain whether or not a real individual has possibly a given property. Thus, to find out whether in our world substance X is soluble in liquid Y, the counterpart theorist need not make any calculations or experiments; all he needs is to dream up a world where Y actually dissolves X. Counterpart theory evokes the doctrine of predestination: my lot in this world is already specified in the Celestial Record kept in the Great Beyond. (More in Merricks 2003.) I submit that the basic philosophical concepts in question – those of world, existence, and truth – are highly problematic. Indeed, each of these words actually designates a whole bunch of concepts. In particular, there are imaginary “worlds,” such as Lewis Carroll’s, Jorge Luis Borges’s, and Isaac Asimov’s, as well as those of mathematics, in addition to the real “worlds” of business and sports. Consequently, there are at least two very different kinds of exist-
232
Chasing Reality
ence, namely, material and conceptual (recall chapter 1). Accordingly, there are at least two kinds of truth, factual and formal – in addition to artistic and moral. Modal metaphysicians overlook or deny all these huge differences. Moreover, in nearly all fields of knowledge and action, truth comes in degrees. In fact, in all walks of life we find and use many more near-truths and near-falsities than full truths and full falsities respectively. For example, even the most precise measurements are affected by errors or discrepancies from truth; and even the best theories are only approximately true. What do we get when substituting “approximately true” or “half-true” for “true” in the above definitions of the modalities: half-possibility and half-necessity, or else halfworlds? The modal metaphysician does not raise these questions; he is content with coarse notions of truth, world, and existence. I submit that this imprecision, inherent in ordinary knowledge and everyday language, is a root of the various systems of modal metaphysics. More on this anon. 5 Many-Worlds Metaphysics Is Inexact The best way to find out the presuppositions and basic assumptions of any theory is to organize it axiomatically, that is, to force it into the axiomdefinition-theorem format. Since modal metaphysics is inexact, its axiomatization should help pinpoint the roots of its troubles. I will therefore axiomatize a fragment of David Lewis’s theory, which is currently the most widely discussed of all modal metaphysics. To begin with, the primitive (undefined) notions of this theory are those of world and truth, both of which are actually problematic. The underlying logic is modal logic. (But we are not told which one of the 256 possible systems of modal logic.) The first specific assumption is this: Axiom 9.1 To be is to be a value of a bound variable. (Equivalently: Existence = $.) Remark 1 This is of course Quine’s famous formula – a typical piece of panlogism, although Quine thought of himself as a physicalist (or nominalist, or vulgar materialist). Remark 2 Although at first sight the above is a convention, actually it is an assumption, for it equates two radically different modes of existence, namely, logical and physical, or conceptual and material, or abstract and concrete respectively. Indeed, Axiom 9.1 entails informally Corollary 9.1 There is no difference between real (or material) and logical (or imaginary) existence. (Equivalently: Whatever is conceivable exists simpliciter, and conversely.) Corollary 9.2 There is no difference between facts and propositions.
Transcendentals Are Of This World
233
This identity is of course suggested by ordinary language, from Sanskrit to English, where one seldom distinguishes between “This happened” (or “This is a fact”) and “The proposition that this happened is true.” But of course the difference between facts and propositions is basic in any self-respecting ontology and epistemology. Refined nominalists, like Ockham, admitted the difference, since they located all fictions in the mind. Axiom 9.2 There is only one concept of possibility, and it is exhaustively elucidated by any of the systems of modal logic. Corollary 9.3 There is no difference between real and logical possibility. (Equivalently: Whatever is really possible is logically possible, and conversely.) Corollary 9.4 Whatever could (conceivably) happen does happen in at least one of the possible worlds. Definition 9.1 A possible world is similar in kind to ours. Remark: Lewis (1973: 85) asks us to admit that “we know what sort of thing our actual world is,” and goes on to “explain that other worlds are more things of that sort, differing not in kind but only in what goes on at them.” This is a paragon of inexact philosophy, and as such it invites multiple arbitrary “interpretations.” Definition 9.2 Object x is real = There is at least one world that contains x. Definition 9.3 Two individuals are world-mates if and only if they are related spatiotemporally. Definition 9.4 The actual world is the one of my world-mates. Theorem 9.1 There are infinitely many possible worlds. Theorem 9.2 “Our actual world is only one world among others” (Lewis: ibid.). Proof: By Axiom 9.1 and Definition 9.2, neither the possible worlds nor their constituents are subject to any constraints, in particular laws. This completes my construction of the axiomatic minisystem that encapsulates Lewis’s metaphysics. There are two ways of judging this theory, or any other for that matter. One of them is to examine the axioms and definitions themselves; the other is to assess their logical consequences. If the former or the latter are found to be either untestable or false, then the system as a whole will be inadmissible in a science-oriented ontology. It would seem that the least damning judgment on Lewis’s metaphysics is that it does not help explore the actual world, because it does not even state any interesting falsities about it. The next negative evaluation is that, because the system says nothing specific about the actual world, no empirical evidence can be adduced either in its favour or against it. In fact, Lewis’s system is worse than either trivially true or obviously false: it is utterly irrelevant because it is
234
Chasing Reality
essentially fuzzy. To substantiate this charge let us examine the above axiomatic reconstruction. To begin with, Axiom 9.1 is inadmissible in any domain, because it conflates two radically different modes of being: logical (or conceptual) and real (or material). More precisely, it asserts that material (or real) existence is reducible to conceptual existence, which in turn equals someness. Such conflation is justifiable only in a system of objective idealism, such as Plato’s, Hegel’s, or Frege’s. But it is untenable elsewhere. In logic and mathematics something exists if it is well defined, whether explicitly or tacitly. For instance, an equation defines implicitly the numbers or functions that satisfy it; and a consistent system of postulates defines tacitly the object(s) it describes. By contrast, to prove the real existence of a hypothetical particle or field we must have it interact with some observation or measuring instrument: we must show that it possesses energy of some kind. (Recall chapter 1, section 7.) For these reasons, the “existential” quantifier should be read “for some” rather than “there exists.” (Recall chapter 8, section 4.) Existence proper, the most important of all properties, is not to be formalized as a quantifier but as a predicate. There is good reason for this, namely, that existence is the most important property that anything can have. If in doubt, ask Hamlet. Consider Meinong’s puzzling albeit famous statement “There are objects of which it is true that there are no such objects” (Meinong 1960). According to van Inwagen (2003: 141), what Meinong seems to have had in mind is elucidating the second “there are” as an existential “quantifier” like this: ExFx = df $x (x has being & Fx). Regrettably, Meinong left “has being” undefined, and van Inwagen did not fill this hole. I submit that the concept we need to dissolve Meinong’s paradox is that of existential predicate (Bunge 1977a: 155–6), explained in chapter 1. That is, the problem is solved upon replacing the first occurrence of “there are” with “some,” and the second with “exists in the world” (or its equivalents “is material,” “is changeable,” or “has energy”). The result is “Some individuals do not exist really,” or ($x)¬ERx, where ER is read “exists in R,” R being the name of a collection (“world”) of objectively real (material) objects. Axiom 9.2, on possibility, is likewise untenable. First, strictly speaking it is false, because modal logic does not involve a unique concept of possibility, since there are 256 possible systems of modal logic. (It would be a long, boring, and unrewarding task to find the sense or content common to all those different notions of possibility.) Second, as everyone knows, not everything thinkable is really possible; and, as historians of science and technology know, we now admit many entities and events that were earlier regarded as impossible, or were not even thought about. Third, and most importantly, modal
Transcendentals Are Of This World
235
logic cannot handle the concept of real (or physical) possibility, which scientists characterize as compatibility with the natural laws or social norms, because that theory does not contain that notion of an objective law or pattern. If only for this reason, the modal logics are utterly irrelevant to science. This is why they are never used in science, whereas the exact concept of probability is ubiquitous. Corollary 9.4 says that the difference between possibility and actuality is world-relative. This result is more than puzzling: it is self-defeating, for it shows that the ‘possible’ in ‘possible-worlds metaphysics’ is basically indistinguishable from ‘actual.’ As for the Definition 9.1 of a possible world, it is pathetically imprecise, because it specifies neither the nature nor the interrelations of the entities that may inhabit it. In particular, neither the Lewisian worlds nor their inmates are constrained by any laws. For example, they might contain perpetual-motion machines of the first and the second kinds, in violation of the first two laws of thermodynamics respectively. This places modal metaphysics on the same bookshelf with fantastic literature, New Age thinking, and theology. According to Definition 9.2, reality would be world-relative. This is not so in science, where only concrete objects can be said to be real, and reality is absolute even though some properties and changes thereof are frame-dependent. For instance, a light wave emitted by an atom is absolutely real, although the value of its wavelength is relative to the frame of reference: it is velocitydependent (Doppler effect). Besides, sometimes there are mismatches between reality and its perceptions. For example, when a fingertip is stimulated at two adjacent points, it is felt as a single stimulus in the middle. And one and the same political event, such as the passing of a bill or a government change, is likely to be “perceived” (evaluated) differently by members of different parties, but the event itself is single. In both cases, although the world is one, it is likely to be “perceived” somewhat differently by different subjects. Definition 9.3 is reasonable but, by the same token, it performs no work in a system designed to allow for any old “world.” It is like decreeing that dragons won’t be taxed. The Definition 9.4 of “actual world” suggests that the existence of the latter depends on that of myself. It is not clear whether this is a case of subjective idealism à la Fichte or Husserl, or at least of operationism. At all events, it is useless in science and technology. Finally, Theorem 9.1 prompts two remarks. First, it is not clear whether the infinity in question is actual or potential: that is, whether the infinitely many worlds are already “there” (where?), or are in the making (by what mechanism?), or else are just conceivable (by whom?). Second, the theorem is trivial
236
Chasing Reality
Table 9.1 Contrast between Lewis’s fantastic metaphysics and scientific metaphysics Key concept
Modal metaphysics
Scientific ontology
World Possibility Real Existence Truth Truth-values
Any old collection Single concept: logical Thinkable Single concept: $ Single concept: vulgar 0 and 1
System of either things or constructs Two concepts: conceptual and real Material = Changeable Two concepts: material and conceptual At least two concepts: factual and formal All the values in the [0, 1] real interval
since, where “anything goes,” there is no limit to the type or number of worlds. In fact, the theorem is so trivial that it is possible to program a computer that will design a potential infinity of “worlds” consisting of only two material points moving inside a box starting with different positions and velocities. Admittedly, the proposed rational reconstruction of a fragment of David Lewis’s modal metaphysics is unlikely to satisfy any of Lewis’s admirers, because my prose is far less elegant if more precise than Lewis’s. But I hope it does what I wanted, which is to exhibit the roots, and the rot in them, of the metaphysical nightmare dreamed up by Lewis. Table 9.1 summarizes the main differences between Lewis’s metaphysics and my own (Bunge 1977a, 1979a, 1981a). This table suggests that modal metaphysics is mired in confusion as well as being utterly at variance with science. In fact, it amounts to folk ontology plus modal logic. Lewis (1973: 88) himself stipulated that the philosopher’s task is to systematize his pre-philosophical (that is, uncritical, non-scientific) opinions. Still, the modal metaphysician will claim that his theory has a practical application: it allows one to make sense of counterfactuals, which in turn can be used to elucidate the key notions of causation and law. Let us examine these claims. 6 Counterfactuals A counterfactual is a sentence of the form “If A were the case, then B would occur.” (Recall chapter 8, section 9.) Example: “If pigs could fly, we could eat leaner ham.” It is tacitly understood that the antecedent of this subjunctive conditional is false. Everyone makes counterfactual statements, for instance, when drawing plans, expressing regrets, or assigning blames, as in “If you had arrived in time, the accident would not have happened.” But only some philosophers believe that counterfactuals can be true. And a few of them, starting with Chisholm and Goodman, have looked for truth-conditions for
Transcendentals Are Of This World
237
counterfactuals. Still, the various attempts to solve this problem met insurmountable difficulties, until it was announced that the solution was found. As McCall (1994: 170) wrote: “[F]or many years those who studied counterfactuals wandered in the wilderness, until at last Stalnaker and Lewis showed us the promised land. This was the land of possible worlds semantics.” The alleged solution boils down to this in the case of flying pigs: If you want leaner ham, travel to the land of flying pigs. There, “at” that imaginary land or world, the indicative conditional “If pigs fly then their hams are lean” is true because, by stipulation, “there” is where pigs fly. I submit that the possible-world solution to the problem of the truth-value of counterfactuals is not only cost-inefficient, but also rests on the presupposition that such sentences can have a truth-value. This assumption is dubious for several reasons. The first is that counterfactuals are sentences that do not designate propositions without further ado. In this regard, counterfactuals are in the same league as wishes, interrogatives, and imperatives. A second reason that counterfactuals are neither true nor false is that no datum could serve to either confirm or undermine a counterfactual, since it fails to refer to real facts. A third reason is that counterfactuals, unlike ordinary conditionals, entail nothing. For example, “If pigs flew they would be slender” can be analysed as the conjunction of the following two indicative sentences: “If pigs fly then they are lean” and “Pigs do not fly.” But valid reasoning stops right here, for nothing follows from these premises. Still, one may learn something from analysing counterfactuals in the way just indicated, which I proceed to render more explicit (Bunge 1968c). We are given the rhetorical expression If A were the case, then B would be the case. (1) and translate it into the proposition If A, then B. & Not-A. (2) Next, we check whether the conditional in (2) is plausible or not. If the conditional is a law, a norm, a robust trend, or is compatible with either, it can be said to be nomic, otherwise anomic. A nomic counterfactual only suggests a non-fact, whereas an anomic one suggests the violation of a law, norm, or trend – that is, either a miracle or the non-occurrence of a necessary condition for the law to hold. For this reason the nomic counterfactuals may be said to be reasonable (though not true), and the anomic ones unreasonable (though not false. A nomic counterfactual can be meek to the point of ridicule; or it may have some heuristic power, as when the assumption that an event did not really happen shows how important, or else unimportant, it was. An early example of the latter is Herodotus’s evaluation of the Athenian
238
Chasing Reality
contribution to repelling the Persian invasion: What if the Athenian fleet had not been ready? Weber (1906) too used counterfactuals to gauge the importance of certain events. A more recent example is Robert Fogel’s research on the role of railways in the American economy at the time of the Gilded Age: What would have happened without them? His interesting conclusion is that the American economy would have attained roughly the same present level through the exclusive use of canals and highways (Fogel 1994). But of course there is no way of knowing for sure. Though unreasonable, some anomic counterfactuals can lead to discoveries. For example, “If the second law of thermodynamics did not hold, life might have emerged spontaneously” suggests removing the condition of system closure. Indeed, the said law does not hold for open systems, which of course are far more permissive than closed systems, and thus more propitious to qualitative novelty. Something similar holds for social norms, such as legal restrictions. For instance, “Were it not for this norm, we could lay our hands on that piece of property” suggests either changing the law or breaking it. In short, whereas some counterfactuals are meek, others are subversive. (Yet, ironically, the many-worlds metaphysicians are hardly interested in alternative social worlds: they are politically aseptic.) And still others, those that involve impossible worlds, are idle parlor games. Regrettably, the above distinction is lost on the many-worlds philosophers who have attempted to analyse causation in counterfactual terms (see, e.g., Collins, Hall, and Paul 2004). This strategy is understandable in a subjectivist perspective, but wrong in a realist one, where a causal link is an objective energy transfer (recall chapter 4, section 1). This is why scientists and technologists use counterfactuals only as heuristic, critical, or rhetorical devices, as when a politician states that everyone would be better off if the income tax were either cut or increased. In these cases one assumes a causal (in general, non-logical) analysis of counterfactuals rather than a counterfactual analysis of causal (in general, non-logical) relations. Proceeding in the reverse manner is like putting the cart before the horse. We shall return to this subject in section 7. In sum, counterfactuals are neither true nor false, since they cannot be put to the test. However, some counterfactual questions, such as those involved in thought-experiments, do have a heuristic power. But, since this power depends on the truth of the pertinent law-statements, norms, or trends, counterfactuals cannot elucidate, let alone validate, scientific hypotheses. This is why no scientific theory contains counterfactuals. These belong either in the heuristic scaffolding or in the rhetorical packaging; which is why their place, albeit a modest one, is in epistemology, not ontology. Serious ontology deals only with the real world, and it does so in the light of the sciences of this world. In other
Transcendentals Are Of This World
239
Table 9.2 A sample of manifest and dispositional properties Manifest or actual
Dispositional or potential
Actuality Relative frequency Mass Kinetic energy Refraction Radioactive decay Sickness Performance Intellectual output Election Action Production Good deed Social division Saving Test Truth
Real possibility Probability Weight Potential energy Refractivity Probability of radioactive decay Predisposition to a sickness Competence (ability) Intelligence Eligibility Feasibility Productivity Good will Social mobility Propensity to save Testability Plausibility, verisimilitude
words, serious metaphysics, by contrast to its speculative counterpart, is scientific, the way Peirce (1935) had envisaged. 7 Disposition Whereas fragility is a dispositional property, being broken is a manifest one. Crystal shops would not exist if there were no difference between dispositional and non-dispositional properties. In general, the features of concrete things, and their corresponding predicates, have traditionally been split into manifest (such as mass, age, and population) and dispositional (such as solubility, sociability, and carrying capacity). A property of a thing may be said to be actual or manifest if the thing possesses it, and potential or dispositional if emerges under suitable circumstances. Table 9.2 exhibits a sample of familiar properties of both kinds. Dispositions have elicited philosophical puzzle and blunder for over two millennia. For example, Aristotle and his followers liked dispositions (or “powers”) because they are central to their simplistic explanation of change as the transition from potency to act – as when Michelangelo claimed to see and extract the forms hiding in a marble block. By contrast, empiricists mistrust dispositions because they are unobservable. In fact, one may observe change, not mutability; bending, not elasticity or plasticity; speech, not linguistic
240
Chasing Reality
ability; creations, not creativity; dying, not mortality; production, not productivity; and good deeds, not good will. According to empiricism, fragility is not an actual property of a glass pane, but only a construct: we say that a glass pane is fragile only because, if it were struck or dropped, it would be likely to shatter. We shall come back to this intrusion of the troublesome subjunctive conditional, which we first met in the previous section. Scientists take the manifest/dispositional distinction for granted, and attempt to relate dispositions to one another as well as to manifest properties. For example, the fragility of glass is explained in terms of the buildup of internal stresses; the solubility of table salt is explained by the sharp weakening of the ionic bond due to the high dielectricity of water, whereas sugar dissolves by bonding with water; predisposition to diabetes and other genetic diseases is explained by the occurrence of certain genes; sociability is explained in terms of the inability of individuals to fend for themselves; the nature/nurture controversy is solved by posing that we are born able, not accomplished: our intellectual dispositions may or may not be actualized, depending on the circumstances; political instability is accounted for in terms of conflicts of interest between different groups – and so on. Consider, for example, the bending of a light beam when striking a transparent medium such as water or quartz. The refractivity of a transparent body is placed on the same footing as its mass and shape, even though it becomes manifest only when light shines on it. Actual refraction, a process, is described by optics, in particular by the Snell-Descartes law. This law relates the refractive index n, a dispositional property of all transparent media, to the angles of incidence i and refraction r: “n = sin i / sin r.” Elementary optics treats the refractive index n as an empirical parameter. By contrast, classical electromagnetic optics analyses n in terms of the electric permittivity e and the magnetic permeability m of the transparent medium: The formula “n2 = em” relates all three dispositional properties in question. (Electric permittivity is the propensity of a body to form electric dipoles under the action of an electric field. Magnetic permeability is parallel.) The quantum-theoretical explanation of refraction, though still electrodynamical, is far more detailed and accurate because it involves the interaction between the incoming photons and the outer electrons in the crystal’s atoms. (Incidentally, most transparent crystals, including ice, quartz, and calcite, exhibit birefractivity. That is, a light beam incident on such a crystal splits into two, so that the crystal has two refractive indices – two dispositions. Actually, multiple dispositions occur in many fields. For example, the sex of sea turtles is determined by the air temperature. And every one of us assumes different personalities and enacts different roles in different circumstances.)
Transcendentals Are Of This World
241
Interestingly, a disposition at the individual level may show up as a manifest property of an aggregate. For example, the probability of a radioactive decay is a property of individual atoms (or rather their nuclei), whereas a mean halflife, like any average, is a manifest property of a large collection of radioactive atoms of the same species. Likewise, genetic predisposition to longevity is a property of individual organisms, whereas average longevity (or life expectancy at birth) is a statistical or collective property. In general, real probabilities are dispositional properties of individuals, whereas statistical parameters, such as average, median, variance, and skewness, are manifest properties of collections. Let us now look into the explications of dispositionals proposed by some influential philosophers. Take, for example, carbon monoxide, or CO, well known for its toxicity. A naive empiricist is likely to say that CO is toxic as shown by uncounted fatalities. But of course this only confirms the assertion that CO is toxic. On the other hand, a sophisticated empiricist, particularly if influenced by the recent literature on counterfactuals, might say that CO is toxic because, if it were inhaled, it would poison. Again, this amounts to resorting to experience without explaining anything. A toxicologist will say something quite different, namely, that CO is toxic because, when inhaled, it combines with hemoglobin with a bond that is about 200 times stronger than with oxygen, and thus prevents the transport of oxygen to the brain and other organs, which die as a consequence. In general, we may say that a disposition, or dispositional property, is a property actually possessed by a thing that, under appropriate environmental conditions, generates another property. Ordinarily, the latter property is manifest or ostensive, whereas the former is not. The logical form of the scientific analysis of a dispositional property D is this: Thing x possesses the dispositional property D = df x has the actual property A & If thing x interacts with another thing y, then x acquires the relational property P. We rephrase this as Definition 9.5 x [Dx = df Ax & $y $P (y ¹ x & y is a thing & P is a property & Ixy & Pxy)]. This analysis shows that a disposition, potentiality, or propensity, is reducible to two actual properties, one (A) intrinsic and the other (P) relational, the former represented by a unary predicate and the latter by a binary one. No reference to parallel worlds is made: as usual in science and technology, it is taken for granted that one is dealing with the only world there is. However, potentiality or possibility has not been eliminated in favour of actuality, as it would have been if we had mistakenly equated D with A. Possibility has just been shifted from the thing in question to the thing-in-its-environment. For
242
Chasing Reality
example, sugar is soluble in water if and only if water dissolves sugar. (More deeply: Dissolution happens because sugar bonds with water–a clear case of interaction.) Likewise, democratic political leaders or parties are powerful if they can mobilize a large sector of the population–which amounts to saying that a large sector of the population will rally around them. (More deeply: Democratic politicians will be able to influence public opinion to the extent that they are perceived, rightly or wrongly, as defending the interests of a large sector of the population.) Besides, dispositions are deemed to be just as real as manifest properties. Consequently, nothing need be said about fantastic worlds. The preceding analysis applies to the dispositions that may be called conditional, because they actualize only under suitable circumstances. The quantum theory introduces unconditional dispositions that get actualized regardless of environmental conditions. Thus, the probability that an excited atom, atomic nucleus, or molecule, will decay spontaneously (without external stimulation) within the next time unit, is an unconditional property of the thing in question in the given state. The quantum conditional dispositions are similar to the classical ones in that they actualize when the suitable environmental conditions arise. But they are non-classical in that they involve blunt (or distributed) as well as sharp values of the dynamical variables, such as momentum and energy. Einstein believed that this feature of quantons, namely, that their dynamical properties only exceptionally have sharp values, contradicts realism. This is not so, because realism is not committed to any particular feature of things except their existence. Whether certain physical properties are manifest or dispositional, sharp or blunt, and so on, is for science, not philosophy, to determine. (More in Bunge 1979c and 1985a.) Let us take a closer look at this question. Ordinarily a quanton, or quantummechanical object, is in a superposition (sum) of two or more (perhaps infinitely many) states corresponding to sharp energy values or eigenvalues. The same holds for the other dynamical variables. Every one of the sharp values has a certain weight or probability. Interaction with the environment (e.g., a measuring device) will destroy the superposition or coherence: only one of the sharp possible values will get actualized. In other words, the quanton has a conditional disposition to acquire a definite energy value, and much the same holds for the remaining dynamical variables, such as momentum and spin. The quantum theorists know how to calculate the probabilities in question, but only a few admit that we do not yet know the details of the mechanism of the said “collapse” or projection of the state function. (More in Bunge 2003d and 2003e.) So much for the concept of a disposition. Let us now peek at the way scientists test for dispositional properties – on the non-empiricist assumption
Transcendentals Are Of This World
243
that meaning precedes test, not the other way round. To test for a disposition, such as refractivity, we need to relate it to a manifest property, such as actual refraction. But not every manifest property will do because, whereas a few manifest properties are directly observable, most are not. (For instance, atomic mass, basal metabolism, family cohesiveness, and economic output are not directly observable.) In each case we must look for a suitable observable property that can reliably be taken to proxy for (or indicate) the corresponding disposition. That is, we need to invent and confirm at least one indicator hypothesis of the form: D = f (M), where M is an observable manifest property. In the case of refractivity, one such indicator hypothesis is the Snell-Descartes law recalled a while ago. In general, one uses the following Criterion 9.1 A thing possesses a property, whether manifest or dispositional, if the given thing, or a different thing connected with it, exhibits an observable property lawfully related to the former. Note the difference between the Definition 9.5 of a disposition and the Criterion 9.1, which parallels the existence-test and the ontology-methodology differences. Whereas the definition tells us something about the object of interest, the criterion instructs us how to check whether the object of interest actually possesses the property elucidated by the definition. For example, the ability of litmus paper to turn colour when immersed in an acid solution can be used to tell whether a liquid is acid; and the dielectricity of bone marrow is used to diagnose leukemia (because the value of that disposition doubles in leukemia patients). The distinction between definition and criterion is typically blurred by empiricism, in particular operationism (or operationalism), as first described by Bridgman (1927). The gist of this doctrine is the thesis, reminiscent of Berkeley, that to be is to be measured. An example is Carnap’s (1936–7) classical explication of a disposition, such as solubility, in terms of a subjunctive conditional referring to test conditions. Thus, thing b is pronounced soluble in water just in case, if b were put in water, b would dissolve. This empiricist account of dispositions is good enough for everyday life, but useless in science. First, like all operationist accounts, it conflates a property with the way it is tested for. That this is a confusion between ontology and methodology becomes even more patent in familiar cases such as the tuberculine test: Human subjects are attributed tuberculosis if they exhibit a positive reaction (inflammation) to a tuberculine injection. But of course such inflammation is only an indicator that the patient has been invaded by Koch bacilli. The tuberculine test makes no reference to these germs. Likewise, the litmus test of acidity makes no reference to pH, or hydrogen-ion concentration. Second, the positivist explication of dispositionals involves subjunctive
244
Chasing Reality
conditionals, which are anything but logically clear items. Indeed, they are not propositions, hence they neither satisfy the propositional calculus nor have a truth-value (recall section 6). No wonder that such conditionals play an ambiguous and therefore indecisive role in argumentation. On the one hand, they invite considering hitherto overlooked possibilities, as in counterfactual historiography, or what-if speculation. On the other hand, subjective conditionals offer cheap and untestable excuses, such as “If Hitler had not been mistreated by his father, he would not have become a mass-murderer.” We conclude that there is nothing mysterious, let alone unreal, about dispositions, as long as they are handled as real possibilities, namely, in terms of laws. We also conclude that nothing is gained by attempting to elucidate problematic concepts, such as that of disposition, in terms of logical and epistemological outlaws such as counterfactuals. Let the latter remain where they belong: in the conceptual wilderness. 8 Space and Time One usually thinks of space as the Great Container, and of time as the Silent River. However, when pressed, everybody seems to agree that these are just metaphors. Space and time are not concrete things, since they possess no energy – the peculiarity of material things (chapter 1, section 1). Furthermore, space, not being a material thing either, cannot contain anything; likewise time, not being fluid, cannot flow. This is why we have no direct experience of space and time. We can certainly feel spaced things, but not space; likewise, we can sense successive events, but not time. Leibniz (1956 [1715?] 2: 1083) put it into a nutshell: Space and time “are orders [relations], not things.” More precisely, according to Leibniz, space is the “order” of coexistents, and time that of successives. Hence, the scientific materialist adds, if there were no things there would be no space; and if nothing changed there would be no time. Moreover, for either to exist there must be at least two distinct items: two things in the case of space, and two events in that of time. (This, the simplest case, can be described by the trivial metric space <S,d>, where S = {a,b} and d(a,b) = 1 if a ¹ b, whereas d(a,b) = 0 if a = b.) Thus matter, space, and time, though conceptually distinguishable, in reality constitute a single block – the universe. As we shall see below, Einstein’s theory of gravitation confirms this view. Hence, Newton’s “absolute” (selfexisting) space and time are only fictions. But they are useful fictions, since they allow one to sketch real things and their changes against a rigid spatiotemporal grid that does not alter along with any changing things – except in the vicinity of massive bodies.
Transcendentals Are Of This World
245
At first blush, the immateriality of space and time defies materialism; and their lack of independent reality poses a challenge to realism. However, I submit that a suitable relational (or “adjectival”) theory of space and time, such as the one proposed earlier (Bunge 1977a), meets this challenge to both materialism and realism. Indeed this theory, hinted at in antiquity and revived by Leibniz and a few others, postulates that spacetime is the basic structure of the collection of all material things. Space is rooted in the separation between things, and time in the separation between events (relative to the same reference frame). So, spatiality and temporality are vicariously just as material, and therefore just as real, as the properties of the material objects that generate them; only, they have no independent existence. But then, things and the changes in their properties have no independent existence either: there are only mutually spaced things and successive changes in things–an excellent approximation far from black holes. (For the relevance of these ideas to contemporary physics, see Greene 2004.) However, empiricists have trouble accommodating space and time in their doctrine. Because space and time are imperceptible, a consistent empiricist should either deny space and time, or try to construct them out of experiential terms. The first move, actually proposed by Sextus Empiricus, would be too extravagant even for most philosophers – the only noteworthy exception being the Hegelian John McTaggart (see Jammer 1954). The second empiricist project was attempted several times, notably by Whitehead (1919) and Nicod (1923), the latter under Bertrand Russell’s supervision. But the theories of space and time resulting from this empiricist approach are subject-centred, and consequently useless in physics and engineering. (Imagine a railway engineer letting himself be guided by the psychological finding that subjects perceive parallel rails as meeting in the horizon.) And those theories are of no interest to psychology either, because they make no use of the experimental investigations into the perception of spatiality and temporality. And of course Kant’s view, that space and time are intuitions rather than features of the real world, cannot be taken seriously by anyone who has ever measured distances or periods. Nowadays we distinguish three different concepts of space: 1 Abstract spaces of many kinds and dimensions, such as topological, projective, Cartesian (like the state spaces), Euclidean, and Riemannian. These are the objects of so many mathematical geometries. Experience and measurement are irrelevant to these theories. 2 Physical space and time regarded as objective features of the world. They are the object of the physical geometries contained in certain physical
246
Chasing Reality
theories, notably Einstein’s theory of gravitation. Unlike the mathematical geometries, the physical ones are expected to be validated by precision measurements. 3 Phenomenal (or experiential) spaces, such as the visual and auditory spaces of humans and other animals. They are studied by cognitive psychology, which has found that the human visual and auditory spaces are not Euclidean. The abstract spaces pose no special philosophical problems beyond the ones raised by any other mathematical abstractions. By contrast, the physical and phenomenal (or psychological) spaces, and the corresponding theories, do raise philosophical perplexities. The oldest and most important of them are the following three. First, what kind of object are space and time: stuff or systems of relations? Second, are they related to matter, and if so how? Third, are space and time real or unreal, objective or subjective? All these questions were addressed above. Our answer was, in a nutshell, a materialist version of Leibniz’s relational theory: space, time, and matter do not exist by themselves. The universe is composed of spaced things, and time is the rhythm of their change. Thus, space and time are vicariously material and real (objective), just as much as change. This view is not nearly as popular as its opposite, namely Kant’s. When Kant abandoned Leibniz’s system as codified by Christian Wolff, he adopted a subjectivist view of space and time: these two would be intuitive preconditions of experience, which in turn would make up the world. Thus, Subjective space and time Experience The world. In 1770 Johann Heinrich Lambert, that remarkable polymath, wrote to Kant (in Kant 1913, vol. 1: 101–2) that it cannot be denied that real existents change; and, since change is real, so is time. Kant did not listen. Eleven years later he repeated the subjectivist thesis in his first Critique. The controversy over the reality or objectivity of space and time remained unresolved until 1915, when Einstein invented his theory of gravitation (general relativity). According to this theory, space and time are fused (though not confused) into spacetime. And in turn the properties of the latter are determined by the distribution of bodies and fields other than the gravitational one, in accordance with the central equation of the theory: “G = kT.” Here G designates the geometric tensor, and T the matter tensor. If T = 0 everywhere, that is, for a totally hollow “universe,” physics steps aside, and “G = 0” describes a family of purely mathematical spaces. In other words, “T ¹ 0 somewhere” describes real matter-cum-spacetime. So, neither matter nor spacetime exists by itself. If preferred, both matter
®
®
Transcendentals Are Of This World
247
and spacetime are real though not independently from one another. This then is the contemporary answer to the millennia-old question whether space and time are real: They are real, though not in themselves but as features of matter (in the broad sense of the set of changing things). Once again, science answered a key philosophical question. 9 Free Will and Liberty Free will is of course the ability to do what one wants. More precisely, it is the ability to have feelings and thoughts, as well as to make decisions and take actions that, though constrained by external circumstances, are not caused by them. In psychological jargon, free will is behaviour (internal or overt) that is not stimulus-bound. A clear example is the refusal of a prisoner, soldier, or priest to obey orders at grave risk. Another case of free will is creativity: the invention of ideas that go beyond or even against sensory stimulation. Einstein (1949: 49, 1950a: 60) understood the theoretical concepts required to account for reality as “free creations of the human mind.” They are free in the sense that they are independent of sense perceptions even when exceptionally triggered by the latter. (Incidentally, this was one of Einstein’s main objections to the positivists: their prejudice against such free creations.) Free will violates of course the central tenet of Pavlov’s reflex theory, the Watson-Skinner stimulus-response behaviourism, and Gibson’s ecological psychology. By the same token, free will violates the narrow version of causal determinism, according to which only external stimuli count as causes. But it is consistent with causality lato sensu, since the implementation of any decision involves causal links such as those connecting events in the prefrontal cortex with events occurring in the neuromuscular system. Hence, free will is not to be defined as overcoming causation. Nor is free will definable as unpredictability, which is an epistemological and methodological category. Theologians and philosophers have debated for two millennia whether free will is real or fictitious. The problem is of great theoretical and practical importance, because it is about whether humans can take initiatives, break away from some environmental constraints, and even revolt against the terrestrial and celestial powers that be. Some theologians need free will to be able to legislate on sin and evil, as well as to justify the cruel punishments raining from above; and moral and legal philosophers need free will to make sense of autonomy and personal responsibility. However, regardless of these needs, the ontological and scientific question is whether free will is really possible. Idealists have no problem admitting free will, because the immaterial soul is assumed to escape the laws of nature. By contrast, vulgar materialists (physi-
248
Chasing Reality
calists, nominalists) reject the free-will hypothesis because they conceive of humans as complex systems abiding by the laws of physics and chemistry, none of which seems to make room for spontaneous or self-initiated processes. On the other hand, emergentist materialists admit the possibility of free will, and the concomitant self-consciousness, as results of evolution. Indeed, evolution involves the emergence of new kinds of things, satisfying laws absent from the lower levels of organization. For example, neurons can fire and associate spontaneously, not just in response to external stimuli. Since volition is a mental ability, and since everything mental happens in the brain, to find out whether free will is for real we must shift the focus, from theology and speculative philosophy to the study of the human brain. In particular, we must look at its prefrontal cortex, which is known to perform the so-called executive functions. The first modern scientist to attack the free-will problem, and to state that it can be solved by neuropsychology, was Donald O. Hebb (1980). He argued that Cajal’s anatomical investigations, as well as electroencephalographic data, had showed that the brain is more than a relay between receptors and effectors: that it is continuously active, even during sleep, and it always adds something to the input signal. Hebb’s own research on sensory deprivation confirmed those studies: it showed that external stimulation distorts the ongoing brain activity, but is not its only source. We can experience desires and images, and form intentions and plans, in a spontaneous fashion, that is, in the absence of external stimulation. Since “free will is control of behavior by thought” (Hebb, ibid.: 139), and since not every thought occurs in response to external causes, free will is biological fact, not illusion. Of course, it can be either attenuated or enhanced by education and social circumstances, but not more so than other mental abilities and processes. Which brings us to our next subject: the possibility of liberty. The social counterpart of free will is liberty. This is the feature of a social order that enables individuals to do what they want. However, liberty is bounded, and what can be had of it comes at a price. It is not only that, as Thomas Jefferson said, the price of liberty is eternal vigilance. It is also the fact that not everyone can afford to be totally free, nor should anyone so feel, to do exactly what he wants. In fact, even in the freest of societies, individuals have obligations, particularly those of respecting other people’s liberties, and of sharing the burden of maintaining the social order that ensures them. Besides, in deeply divided societies only the members of the ruling minority have the wherewithal required to enjoy security, clean air and water, beautiful landscapes, and access to higher culture. Moreover liberty, like any other value, is part of a package or system: no
Transcendentals Are Of This World
249
value can be realized in isolation from all other values. In particular, no one can be free in a community of utterly selfish individuals unwilling to lend a hand when needed, and this for the simple reason that no one, not even the autocrat, is self-sufficient and omnipotent. The revolutionaries of 1789 had it right: Liberté, égalité, fraternité. I venture to add idoneité, that is, technical competence, for without it even the best intentions are unlikely to come to fruition. Sadly, everyone knows that none of the societies currently in existence realizes the Liberty-Equality-Solidarity-Competence ideal. Does this prove that we should stop dreaming up utopias? Hardly. It only goes to prove that so far we have been unable to educate and mobilize people to work (rather than fight) for that ideal. 10 Concluding Remarks The word ‘transcendental’ has earned a bad reputation because it has all too often designated otherworldly items. However, it is a legitimate name for anything that transcends (overreaches) experience. So, it applies not only to deities but also to fictions, electrons, nations, and things. Moreover, some transcendental items, such as possibility, disposition, and spacetime are indispensable to account for concrete things and their changes. They are needed because they have real counterparts. For example, biological species are collections, not concrete individuals, but they are not arbitrary or conventional sets; likewise, although the law-statements are abstract objects, they represent real patterns. In general, the universals are in re rather than ante rem or post rem.
250
Chasing Reality
10 From Plato’s Cave to Galileo’s Hill: Realism Vindicated
Although philosophical realism is practised by all sane people, antirealism flares up once in a while, even in such unexpected quarters as marketing and management. In particular, realism is one of the bêtes noires of postmodernism, that bulldozer of everything that is good about modernity – in particular, trust in reason and the search for the true, the good, and the right. Hence, if we wish to stop that bulldozer, we must defend philosophical realism among other things. The philosophical realism I advocate is a comprehensive doctrine. Indeed, it is a system with seven components: ontological, epistemological, semantic, methodological, axiological (or value-theoretic), ethical, and practical. Here is an abbreviated formulation of the seven constituents of integral philosophical realism (R): Ontological R = The external world exists by itself. Epistemological R = The external world can be known. Semantic R = External reference and factual truth. Methodological R = Reality checks and scientism. Axiological R = Objective as well as subjective values. Ethical R = Moral facts and moral truths. Practical R = Efficiency and responsibility. Note the absence of two realisms from the above list: political and aesthetic. The former, Realpolitik, is a synonym for the unprincipled struggle for power. It is political cynicism and expediency leading to barbarism. As for aesthetic realism, it is so broad a category that it embraces the best literature as well as the conservative (“academic”) visual arts. The confusion of philosophical with aesthetic realism led Lyotard (1980: 77), the inventor of the word ‘postmodern,’ to proclaim the “lack of reality” of reality – a postmodern gem. Integral philosophical realism is a system (or “organic whole”) rather than a
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
251
set of disconnected opinions. Thus, practical realism is the thesis that action should be efficient – as well as consistent with a modicum of morality, if only because immoral action is ultimately self-defeating. Ethical realism promotes addressing real-life moral problems with the help of the relevant scientific or technological knowledge. An axiology is realist insofar as it deals with both objective and subjective values, and justifies the former with the help of solid knowledge and cogent arguments. A methodology is realist if it adopts the strategy that scientists and technologists actually use to study or alter reality. A semantic theory is realist if it contains a theory of reference that allows one to find out what predicates and propositions refer to, and adopts the correspondence view of factual truth. An epistemology is realist if it makes contact with cognitive psychology and shows that the scientific exploration of the universe presupposes ontological and epistemological realism. Finally, the realist ontologies assume the independent (i.e., subject-free) existence of the universe. The reverse stream, from theory to praxis, is equally strong. Indeed, realist ontologies support an epistemology that encourages the study of the real world, distinguishes primary from secondary properties, and discourages fruitless speculations about impossibilia such as disembodied spirits and parallel worlds. In turn, a realist epistemology elicits a semantics that starts with semantic theories of reference and factual truth. Such semantics encourages a methodology centred in the scientific method, and consequently it admits the subject/object dichotomy, and encourages an account of secondary properties in terms of primary properties. The conjunction of realist methodology, semantics, epistemology, and ontology suggests a realist axiology that regards valuation as a process occurring in a brain immersed in a social network. In turn, such an axiology invites the realist construal of moral norms as rules to tackle objective moral issues. And a realist ethics suggests a practical philosophy that helps tackle practical problems both efficiently and in accordance with moral principles. Let us now proceed to arguing for all seven branches of philosophical realism. 1 Ontological Realism: Brain and History Ontological realism is the thesis that the universe, or reality, exists in se et per se, in itself and by itself. Only a tiny part of reality, namely, the human social world, emerged, subsists, and changes because of us and for us – but its existence does not depend on the knowing subject. In other words, the subject, knower, or explorer, is a real thing surrounded by real things, most of which have pre-existed him and did not require his assistance to come into being.
252
Chasing Reality
And, once they arise, the social inventions or constructions are just as real as the mountains. We shall privilege four arguments in support of ontological realism: those from physics, biology, cognitive neuroscience, and history. The first argument is that all the basic physical laws, such as Newton’s second law of motion, Maxwell’s triplets, and Schrödinger’s equation, are invariant under (certain) changes of reference frames, in particular observers. Only the derived lawstatements, such as Galileo’s law of falling bodies and the law of the Doppler effect, may be frame-dependent. But reference frames are special physical systems, not necessarily manned ones (see Bunge 1967b). The argument from biology is that all organisms, even bacteria and subjectivist philosophers, draw nourishment and energy from their environment, and are equipped with sensors of external signals. This is why organisms die if totally isolated from their environment, and face high risks when their sensors malfunction. In short, we are constitutional realists – tenured professors of philosophy excepted. The third argument, first proposed by Condillac in 1754 with the help of his famous imaginary statue, boils down to this. Odour comes from smelling some external thing or from remembering such experience. The other four classical senses work similarly. (Proprioception was unknown at the time.) Condillac’s argument may be updated as follows. The mental functions are brain processes, some of which represent external events. For example, olfaction is a long and complex causal chain that can be summarized thus: Volatile molecules in the subject’s environment ® Chemical receptors on the cilia of olfactory sensory neurons ® Olfactory bulb ® Olfactory cortex ® Other brain regions (in particular the organs of emotion). The first and last links of the olfaction chain may be missing: olfactory hallucinations may occur, and we may not be aware of subliminal olfactory stimuli. Likewise, the blind may detect shadows without perceiving them consciously. In short, sometimes the brain detects external stimuli, though not always consciously. And at other times the brain generates illusions or hallucinations: it hears “voices,” feels phantom-limb pain, has religious and out of body experiences, and so on. Only a confrontation with harsh reality can correct our wrong representations of it. The brain could not exist without the world surrounding it, from which it draws not only nourishment but also the stimulation required for its normal development and functioning. The role of the environment in the development of the normal brain was shown experimentally by Hubel and Wiesel (1962) in a famous experiment that earned them the Nobel Prize. This experiment consisted in surgically closing an eye of a newborn kitten, and removing the
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
253
stitches after twelve weeks. The result was that the cat never developed depth vision because the stripes of the primary visual cortex corresponding to the blind eye were invaded by neurons in the same area stimulated by the intact eye. In short, the genome does not suffice for normal development: stimulation by the external world is needed too. Another well-known case is that of individuals who become blind after birth: their hearing improves because some of the neurons formerly attached to vision are drafted by the auditory area. In short, environmental stimuli contribute powerfully to the development of the brain. When such stimuli are screened off, the brain ceases to develop or function normally. Indeed, Hebb’s classic experiments on sensory deprivation in humans have shown that, in the absence of external stimuli, the subject hallucinates and loses count of time. In other words, bracket out the world, the way Husserl recommends to capture the essences of things, and you are likely to go crazy. Thus, Hilary Putnam’s mythical “brain in a vat” would not only be solipsistic: it would also be quite deranged. Certainly, no two individuals see the world in exactly the same way, for no two brains and no two life histories are identical. Still, all persons, even the severely autistic, agree that there are certain things around them, such as trees, buildings, and other people, that exist on their own. And emotionally salient and surprising events, such as gunshots and plane crashes, are likely to be perceived correctly by everyone within earshot. Moreover, in perceiving such events, different brains have been found to tick synchronically. Thus objectivity, or at least intersubjectivity concerning sensory stimuli, seems to be hardwired in the brain. Most of the individual differences concern intellectual and moral matters. So much for the argument from cognitive neuroscience. Finally, the argument from history consists in pointing out that all of the historical sciences, from cosmology and geology to evolutionary biology and historiography, take the past for granted. Moreover, they assume that no study of the past can alter it. True, every generation of historians rewrites historiography. The same holds for students of the evolution of the universe and its constituents, from molecules to rocks, organisms, and social systems. The reasons for scientific (not ideological) historical revisionism are well known: the historical record is incomplete; new evidence has been unearthed; new methods have been invented; new hypotheses have been suggested; or whole new philosophies of history have been proposed. In short, historians cannot alter the past; they only try to fill lacunae or correct mistakes in previous accounts of what happened. Only a social constructivist could affirm that the past (not just its study) is a social construction – but he does not bother to offer any evidence for this
254
Chasing Reality
bizarre claim, other than confusion between the past, history, and its descriptions, historiographies. (For example, after canvassing the opinions of a number of American historians, Novick [1988] concluded that historical objectivity is nothing but “a noble dream.”) Travel to the past is physically impossible, and history is irreversible. If we could tamper with the past, we would alter the present without moving a finger – something that not even the occultists dare promise. 2 Epistemological Realism: Kicking and Exploring Epistemology deals with the subject–object relation: it is expected to tell us some generalities about cognition and its object (see, e.g., Bunge 1983a and 1983b). Actually, epistemology is multiple rather than one: there are as many epistemologies as ontologies – realist and irrealist, materialist and idealist, and so on. Epistemological realism may be compressed into one definition and six theses. The definition is this: An item is real if it exists independently of the subject or knower. Note that this definition cannot double as a reality criterion, for it does not tell us how to find out that the object persists while we are not looking. Ordinarily, to find out whether a given item is real, or else a figment of our imagination, we watch whether it influences something else, or whether we can act effectively upon it. For instance, Boswell tells us that Samuel Johnson claimed to have proved the reality of a stone by kicking it and feeling pain in his foot. However, Berkeley might have rejoined that the kick just proved that the stone was there only because Doctor Johnson had perceived it. Kicking, then, is not decisive, except for scoring soccer goals. Besides, in most cases we cannot act upon a thing to find out whether it exists really. For example, we cannot kick the Sun, let alone the Milky Way or the universe as a whole. Seeing the Sun is certainly a mighty clue, but not one that would satisfy Berkeley. We accept this sensory evidence because we can hide from sunshine, and also because we can explain the process that ends up in our perception of the Sun: Photons emitted by the Sun ® Retina ® Optic nerve ® Visual cortex. That is, we accept a visual sensation as evidence for the independent existence of external visible things because we can explain the causal chain that goes from the Sun to the brain. In sum, kicking and observing a reaction is highly suggestive but not sufficient to certify the reality of an entity. So much for the definition and the criterion of reality. Let us now state the theses of our scientistic version of epistemological realism.
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
255
1 Reality is scrutable: Or, better, it is possible to know some facts, though usually only partially and gradually. Realism does not require exact truth: our knowledge may be only rough and sketchy. Moreover, accuracy is not always more valuable than insight, tractability, and testability. Thus, a rough but deep (mechanismic) model is a better exploratory tool than a highly accurate but shallow black box. 2 Indirect knowledge is the deepest: It is attained via theories and indicators rather than through mere perception or instant intuition. This thesis opposes direct realism, that is, the empiricist opinion that we do not need reason to get to know anything. This opinion does not even hold for knowledge of everyday matters, since all cognition depends upon past experience, expectation, attention, and conjecture. For example, the paleontologist sees fossils strewn among rocks that are likely to escape the layman; time “passes” more slowly when we pay attention to external events than otherwise; and understanding someone’s behaviour requires “reading” his mind, that is, conjecturing some of his thoughts. 3 Fallibilism: We are bound to err sometimes – we may employ deficient techniques, make false assumptions, simplify excessively, or elicit false data. Fallibilism is of course the sceptical ingredient of epistemological realism. This ingredient, absent from naive realism, is the mark of critical realism. 4 Meliorism: Given any piece of knowledge, in principle it is possible to improve on it – to make it more comprehensive, accurate, or deep. Meliorism moderates the impact of fallibilism: it prevents us from falling into the barren trap of radical scepticism. And it is not an article of faith: an analysis of the simplifying assumptions involved in constructing any hypothesis and in designing any experiment suggest that, by removing any such fictions, more accurate results should be obtained. 5 Moderate pluralism: In principle, any set of facts can be represented by alternative hypotheses or theories. And, whereas some rival constructs are equivalent in some respects, others differ in accuracy, generality, or depth. 6 Objective knowledge, backed by solid evidence and sound theory, is far superior to subjective hunch. For instance, knowing that the defendant in a murder case was abroad at the time of the murder, demolishes whatever suspicion his police record or the colour of his skin may have elicited in a juror. Incidentally, objectivity or impersonality must not be confused with intersubjectivity, or consensus about a matter of knowledge. The reason is that consensus can be reached through manipulation or coercion, regardless of
256
Chasing Reality
evidence or argument. In science and technology, intersubjectivity, or even consensus, is a bonus of objectivity and rational discussion. Typically, phenomenology, feminist philosophy, Bayesianism, pseudoscience, and postmodernism replace objectivity with either subjectivity or intersubjectivity. For example, two prominent Bayesian theorists announce: “We can find no real role for the idea of objectivity except, perhaps, as a possibly convenient, but potentially dangerously misleading, ‘shorthand’ for intersubjective communality of beliefs” (Bernardo and Smith 1994: 237). Objectivity differs also from both value neutrality and impartiality – three concepts that Max Weber (1988b) conflated in his influential position paper. Indeed, one can be objective yet at the same time have certain preferences – for, for instance, truth over falsity, democracy over dictatorship, or equity over inequity. Likewise, objectivity is compatible with partiality. As Rescher (1997: 43) put it, “Being objective about determining the facts does not demand being prepared to welcome them as is and refusing to try to change the conditions of things that they represent.” Partisanship is to be avoided only if it interferes with either fairness or the search for truth, as when the free-trade mantra is sung as the panacea for national development, even though the relevant statistics show that it only favours the economically and politically powerful. What may be called the Explorer’s Argument is this: Anyone who endeavours to explore a territory or investigate a concrete thing, event, or process assumes that it exists or may exist. Thus, the physicists who attempt to find Higgs bosons with the help of high-energy particle accelerators, assume that it is really possible for such particles, postulated by a theory, to exist. Likewise, the evolutionary biologist who digs for a fossil intermediate in age between two known fossils assumes that the missing link may exist. And the archaeologist who looks for human remains or traces of work near a ruin assumes that the people who made them are likely to have existed. All explorers hope to confirm their hunches: they are not falsificationists à la Popper. In particular, miners dig for gold, not to falsify the hypothesis that there is none in the plot they have staked. What will an explorer think if he fails to find what he was looking for? Either that the hypothesized thing never existed, or that it is somewhere else, or that it got destroyed or stolen. In the first case he will have falsified his hypothesis; but in the other two cases he may entertain the hope that someone else, equipped with better instruments or truer ideas, will confirm his hypothesis. In either case, the explorer assumes that some of the things he took for granted or assumed are out there. This is why archaeology, evolutionary biology, and geology are not armchair occupations. As for cosmology, it is certainly a theoretical discipline, but one motivated and checked by astronomi-
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
257
cal data, and these in turn are about “celestial” objects that exist or have existed in reality. 3 Semantic Realism: Reference and Correspondence I submit that semantic realism is the view that (a) some propositions refer to (are about) facts; and (b) some factual propositions are true to some extent. Thesis (a) is rejected by the textualists, in particular the deconstructionists, such as Derrida, according to whom every symbol refers to other symbols. And radical sceptics (Pyrrhonists) reject thesis (b). Everyone else subscribes to both theses. But not everyone realizes that the key concepts the theses involve, those of reference and truth, are highly problematic. Most semantic theories, in particular possible-worlds semantics, fail to elucidate the concept of reference: they do not help finding out what a speaker is talking about. They also confuse reference with extension, as when Ptolemy’s epicycles are said to be “non-referential,” whereas Kepler’s orbits are regarded as “referential.” As a matter of fact, or rather logic, all propositions are referential even if they are undecidable or about nothingness. Thus, fairy tales refer to fairies. And tautologies (logical truths) refer to anything at all. Since they are content-free, they do not depend on specific reference (Bunge 1974c.) Note the difference between reference and representation (see Bunge 1974a). Although tautologies, conventions, and statements in pure mathematics have referents, whether precise or non-descript, they do not represent anything in the real world. Thus, a definition of velocity refers to a moving thing but does not represent any feature of it. On the other hand, the statement that particles cannot attain the speed of light represents a property of things endowed with mass. See table 10.1. My theory of reference (Bunge 1974a) posits that the referent of a simple statement, of the form “b is a P,” is b; the referents of a proposition of the form “b and c are R-related” are b and c; and the reference class of a molecular statement, such as “p and q,” and “if p, then q,” equals the union of the partial reference classes. This semantic theory allows one, in particular, to identify the referents of theories of uncertain referents, such as relativistic and quantum mechanics. It can be shown that these theories refer exclusively to physical things: that they make no reference to observers (Bunge 1967b). (The way to proceed is to analyse the key concepts in the postulates, such as the Schrödinger equation. This formula specifies the effect of the Hamiltonian, or energy operator H, on the state function Y. By hypothesis, H represents the energy of the thing or things in question, such as a proton and an electron joined by their electrostatic interaction in the case of a hydrogen atom. Neither experimental devices nor observers are mentioned in H. Hence H refers
258
Chasing Reality
Table 10.1 Some conceptual objects represent items in the real world in a more or less faithful (true) fashion, whereas others refer but do not represent. Still others, such as string theory, are still in limbo at this time. Conceptual items
Real world
Model things Attributes Singular propositions Law-statements Conventions Logic Mathematical fictions Metaphysical fantasies Falsities
Things Properties Facts Patterns – – – – –
exclusively to the microphysical entities assumed in writing it. The same holds for the state function. Anything else is philosophical contraband.) As for truth, most philosophers agree that it can only be predicated of propositions, that is, the objects designated by sentences. (It may be argued that diagrams and pictures too may be more or less true, but we shall not pursue this possibility here.) However, philosophers are notoriously divided on this question. Besides radical scepticism (“There are no truths”) and relativism (“All truths are local or tribal social constructions”), the main views on truth are the following: 1 Redundancy (or deflationary) theory: The concept of truth is redundant. That is, to state p is the same as to assert the truth of p. 2 Coherence theory: A proposition is true in a body of propositions if and only if it is consistent with the remaining constituents of that body. In particular, an abstract proposition, such as “The operation • is associative in S,” is true in a system if and only if either (a) it is satisfiable (or has a model or example), or (b) it is deducible from the basic assumptions (postulates) of the system. Obviously, the concept of truth as coherence applies only in mathematics. 3 Correspondence theory: A factual proposition is true if and only if it corresponds to (represents) the facts it refers to. Peirce (1986: 282) put it thus: “Truth consists in the existence of a real fact corresponding to the true proposition.” More precisely, a factual proposition is (factually) true if and only if it has been empirically confirmed, or if it is entailed by propositions that have passed the empirical tests. This concept of truth as correspondence applies to ordinary knowledge, factual science, and technology.
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
Proposition
Test
259
Truth-value
Represents Fact(s) Figure 10.1 Factual truth-values derive from truth tests. Hence, untested propositions have no truth-values.
The correspondence theory is usually taken to assume that every factual proposition has a precise truth-value, and is moreover true or false, whether or not we know it. This claim strikes me as unrealistic, because factual truthvalues can be legitimately attributed only on the strength of truth tests. If no such test can be performed, the corresponding proposition lacks a truth-value. Examples: (a)“It was snowing in Montreal when Christ was born”; (b) “Christ’s last thought was that all his disciples scuttled away when he was arrested”; (c) “Some Corinthians must have replied to the epistle that St Paul addressed them.” See figure 10.1. Unlike stars and cells, whose properties are subject-independent, propositions are constructs, not givens with objective properties: we construct, evaluate, and modify all propositions. Hence, a pinch of constructivism is justified with respect to constructs, provided it is not of the voluntarist kind. Since propositions are evaluated in the light of more or less rigorous tests, their truth-value, if any, may change in the course of research rather than being innate and constant (see Bunge 1967a). Only Platonists, naive realists, and those who confuse truths with facts are entitled to pretend that propositions are true or false from birth and remain so forever. So far we have dealt with the propositions of ordinary knowledge, factual science, and technology. How about moral propositions? If we grant that there are moral facts, such as wars, and therefore moral truths, such as “It is wrong to wage war if unprovoked,” then we may admit the following definition. A moral proposition is true if (a) it follows from higher moral principles, such as “Enjoy life and help live”; and (b) its implementation contributes to lessening misery. (The moral postulates meet trivially condition (a), because every proposition entails itself.) Let us now evaluate the three views of truth sketched above. According to the redundancy “theory,” saying that p is true does not add anything to p. For instance, it is obvious that snow is white is true if and only if snow is white. The redundancy thesis has been defended by such famous philosophers as Frege,
260
Chasing Reality
Wittgenstein, Ramsey, Ayer, Strawson, and Quine, in addition to many younger scholars (see Horwich 1990). The redundancy thesis about truth is false for the following reasons: (a) it does not distinguish between a truth claim, of the form “p is true because of q,” and a groundless pretence; (b) it fails to distinguish between different kinds of truth, such as formal and factual; (c) it only concerns low-level statements that can be confronted directly with reality, such as “the cat is on the mat,” but it fails for high-level statements, such as Newton’s laws of motion, which require deductions, indicator hypotheses, and sophisticated empirical procedures; (d) it draws no difference between making a statement and justifying it; (e) it does not distinguish between making a statement for the sake of argument (in which case one does not assign it a truth-value) and asserting it either as postulate or as theorem; (f) it makes no room for partial truths, such as “The Earth is spherical,” hence for statements of the form “p is truer than q,” much less for any of the standard procedures of successive approximations; and (g) it does not allow us to state methodological norms such as “Abstain from attributing truth-values to untested propositions” and “Prefer the truest of all hypotheses.” In conclusion, we should keep the distinction between a statement and the various metastatements that can be made about it, in particular those of the form “The truth-value of p is such and such.” This distinction is particularly important when we mistakenly assert a falsity or deny a truth. Such distinction becomes a matter of life and death when interpreting the outcomes of medical tests, as in the case of “false negatives.” Next in our list comes the coherence view (usually dubbed “theory”). This thesis is obviously correct with regard to logic and mathematics, particularly whenever truth can be equated with provability. But the thesis leaves us in the lurch in the cases of ordinary knowledge, factual science, technology, and the humanities. For example, the great majority of scientific theories, though presumably consistent, have proved to be at variance with the facts – which is why they have been corrected or rejected. This shows that consistency is not enough. In any domain dealing with facts we need “agreement” with (or “adequacy” to) the pertinent facts in addition to internal consistency and external consistency, or compatibility with other pieces of the background knowledge. The correspondence “theory” of truth refers only to factual statements, such as empirical data and scientific hypotheses. It captures the intuition that factual truth, unlike formal truth, consists in adequacy to reality. In everyday life, science, and technology we first assert tentatively a proposition, then check it for truth; and finally, with luck, we pronounce it true. This will be the case if the proposition in question represents correctly the facts it refers to. However, the correspondence “theory” has the following flaws: (a) it is a vague thesis
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
261
rather than a theory proper or hypothetico-deductive system; (b) it does not tell us what negative and general propositions correspond to; (c) it makes no room for partial truths; and (d) it leaves external consistency (systemicity) out. Still, all of the above defects can be corrected, at least in principle (see Bunge 2003a). The corrections result in a fourth and last candidate: the synthetic theory of truth. This embryonic theory may be compressed into a definition and a criterion of factual truth. The definition is this: “The proposition p asserting fact f is true if and only if f occurs.” This definition is the socalled correspondence theory of truth, adopted tacitly by nearly everyone. Of course, it is rejected by the subjectivists, because they deny the very existence of objective facts to be represented in the brain. Finally, the criterion of factual truth is this: (a) p is compatible with (and exceptionally equivalent to) the relevant evidence; and (b) p is consistent with the bulk of the pertinent background knowledge. Notice the above distinction between the definition and the criterion of truth: the former tells us what truth is, and the latter suggests how to identify it. This difference underlies the use of indicators, such as the frequency of Geiger clicks as a measure of radiation intensity, and corticosterone level as a measure of stress. Failure to distinguish between truth and evidence for it is the source of some varieties of antirealism. One of them is the holistic and pragmatist claim that there is no bridge between our theories and the world they claim to describe, because those theories would only describe the relevant evidence (Davidson 1984, Clough 2003). Such confusion between reference and evidence amounts to conflating stars with telescopes. The same conflation is also inherent in the so-called verification theory of meaning, characteristic of the Vienna Circle and operationalism, and also recommended by Davidson. This theory is obviously false: measurements and experiments can tell us whether a hypothesis is true, not what it means. Moreover, meaning precedes test, if only because we must know what a hypothesis is about before designing a test for its truth. (Further criticisms in Bunge 1967a, 1974a, and 1974b.) A more popular variety of antirealism derived from the failure to distinguish the concept of truth from its criterion is the neo-positivist replacement of “true” with “confirmed.” Putnam (1981: 64) once adopted this doctrine and rechristened it “internal realism.” According to it, a true statement “is a statement that a rational being would accept on sufficient experience.” For example, every morning the Aztecs confirmed their belief that the previous day’s human sacrifices made the gods bring back the Sun. Clearly, Putnam confused the concept of (factual) truth with the empirical truth criterion.
262
Chasing Reality
So much for the statement of our version of semantic realism. Let us now propose some arguments for it. The most popular of them is this: There must be something out there if a theory about it turns out to be true. This argument is particularly persuasive if the existence of the thing in question was predicted by the theory in question, as in the cases of Neptune, electromagnetic waves, positive electrons, neutrinos, and hominids and other extinct organisms. What if the evidence pertinent to a theory is unfavourable to it? This may be either because the theory is false while the evidence is firm, or because the theory is true while the evidence is not. Only further empirical or theoretical work may settle the question. The postulated entity may turn out not to exist. This will be a tragedy for a beautiful theory, but not for realism: the existence of the universe is always presupposed rather than questioned. Paradoxically, an even stronger support for semantic realism is the argument from error (Bunge 1954). Indeed, the very concept of a scientific error, whether conceptual or empirical, presupposes the real existence of the entity or feature in question. For example, we know that the figures for unemployment rates are only approximately true, because many unemployed get discouraged and stop going to labour exchanges in search for jobs. In other words, there are jobless out there who have not been counted. At other times we count twice what is actually a single thing, as is the case with the double images created by gravitational lenses. Realism thus accounts for both error and its reduction: it explains scientific and technological progress as a product of our effort to improve the description and explanation of the outer world. True, some irrealist philosophers admit the possibility of improving scientific theories, but they are likely to hold that “it is meaningless ... to inquire into the absolute correctness of a conceptual schema as a mirror of reality” (Quine 1953: 79). However, they do not bother to elucidate how ‘improved theory’ differs from ‘truer theory.’ Worse, they cannot perform such elucidation because, lacking the concept of objective truth, they cannot employ that of error as discrepancy between theory and reality. Some realists, such as Popper (1972), make much of scientific progress, but declare the success of science to be miraculous, and therefore unexplainable. Putnam (1975) has correctly argued that the success of science would be miraculous if its theories were not at least approximately true. But phenomenalists like van Fraassen (1980), and pragmatists like Laudan (1981), have claimed that we should not talk of miracles or even of truth: success would suffice. But what does ‘success’ mean in science other than “truth”? The Nobel Prize is not awarded to saints for performing miracles; nor is it awarded to successful engineers, businessmen, or politicians. It is only awarded to scien-
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
263
tists who have “found” (discovered or invented) some important truths about a part or feature of reality. Science does not and could not possibly prove realism, because every scientific proposition, whether datum or hypothesis, refers only to facts of a particular kind. Science does more for realism than confirming it: it takes realism for granted. In other words, scientists (and technologists) practise semantic realism even if they occasionally pay lip service to an antirealist philosophy. They presuppose both the independent existence of the world and the possibility of producing objective truths about some of it. So do laymen when going about their daily business, mindless of the subjectivists, conventionalists, pragmatists, and constructivist-relativists, all of whom deny the possibility of objective truth. (More on this in Durkheim 1988 [1901], Weber 1988b [1904], Lenin 1947 [1908], Einstein 1950a [1936], Popper 1956, Sellars 1961, Smart 1968, Keuth 1978, Trigg 1980, Newton-Smith 1981, Agassi 1990, Stove 1991, Brown 1994, Siegel 1987, Vacher 1991, Niiniluoto 1999, and Hunt 2003.) To appreciate the centrality of truth in daily life, imagine two lands: Analetheia, whose inhabitants deny truth, and Anapistia, whose citizens deny falsity. (Both place names have Greek roots: alef theia means “truth” and apistia “falsity.”) The Anapistians hold that “anything goes,” whereas the Analetheians maintain that “nothing goes.” The former are too open-minded, that is, gullible, whereas the Analetheians are excessively closed-minded, that is, radical sceptics. Of course, the natives of both lands contradict themselves when holding that their own views are right, but they do not realize this. Nor can they debate one another: the Analetheians, because they have no convictions; and the Anapistians because they see nothing wrong in whatever their counterparts may claim. Worse, the Analetheians cannot use any truths to shape their own lives; and the Anapistians unwittingly use falsities galore. Consequently life in both lands, like that of Hobbes’s primitives, is nasty, short, and brutish. Not to worry, though, because in the real world all Analetheians and all Anapistians are college professors who earn sheltered lives teaching radical scepticism, constructivism, or relativism, instead of working hard to try and find new truths. 4 Methodological Realism: Reality Check and Scientism Methodological realism amounts to both the demand for “reality checks” (empirical tests) and the adoption of scientism. A reality check is of course the confrontation of a proposition, in particular a hypothesis, with the pertinent
264
Chasing Reality Hypothesis
Confrontation
Reference Fact
Evidence Reference
Indicator
Observation
Figure 10.2 A factual hypothesis refers to a fact that, when under observation, is somewhat different, and is the referent of a datum or piece of evidence.
empirical data and pertinent theories. Note that this is not a direct confrontation of a proposition with the fact that it refers to, but with a datum about it. Whereas a proposition cannot be compared with a fact, it can be contrasted to another proposition. See figure 10.2. As for scientism, it is the thesis that the scientific method is the best strategy for attaining the more objective, more accurate, and deepest truths about facts of any kind, natural or social. This was the central thesis of the famous speech delivered by Condorcet (1976) in 1782 to the Académie Française. Scientism has since been a key principle of both positivism and scientific realism. In fact, it is the main bone of contention between the pro-science and the anti-science camps. True, Hayek (1955) famously claimed that scientism is something quite different, namely, the attempt on the part of some social scientists to ape their colleagues in the natural sciences, in ignoring the inner life of their referents. But this arbitrary redefinition involves confusing naturalism, or reductionist materialism (as practised, e.g., by the sociobiologists), with scientism. (For the standard definitions of both terms, see, e.g., Lalande 1938 and Bunge 2003b.) The opponents of scientism hold that subjectivity cannot be studied scientifically. This is not true: psychology is the scientific study of the mind (Hebb 1980). The proponents of scientism admit of course the difference in both subject matter and special techniques between the sciences of the inner and the outer worlds. But it also holds that these differences are no obstacle to using the same general method in all of the sciences, and this for the following reasons. All existents are material, all investigators have similar brains, and they all share the goal of finding objective truths about patterns as well as particulars. This thesis is sometimes called methodological monism. A merit of methodological monism is that it allows social scientists to use some of the techniques and findings of the natural sciences – as when social psychologists make use of neuroscience. Another is that it renders the biosocial sciences (demography, anthropology, psychology, linguistics, etc.) pos-
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
265
sible, whereas the natural/cultural dichotomy stemming from Kant overlooks them (see, e.g., Bunge 2003a). The usual argument for methodological realism is the success of the scientific method. This is certainly a strong argument. But what about the failures of the scientific endeavour, particularly in psychology and the social studies? It might be argued that any particular failure only suggests that there was some flaw in the application of the scientific method – for instance, that the sample was either too small or not random; that the experimental design overlooked certain variables; or that a single line of evidence was used when the complexity of the matter required employing multiple lines. However, such excuses are unlikely to persuade the scientific sceptic. The sceptic about the reach of the scientific method might be swayed by the hermeneutic argument that the cultural (or social) sciences, unlike the natural ones, attempt to “understand” social facts in terms of “meaningful” behaviour rather than to explain them. Regrettably, the hermeneuticists do not elucidate their slippery notion of meaning, although from the context one suspects that they mean “intention” or “goal.” If this is the case, then the hermeneutic thesis in question is false, since cognitive neuroscience studies intentions as processes in the prefrontal cortex; social psychologists and sociologists use the scientific method to investigate goal-seeking behaviour; and social technologists, such as management scientists, social workers, and legislators, attempt to steer behaviour. Nor is it true that the social sciences are bound to explain the familiar by the familiar, as the intuitionists, phenomenologists, ethnomethodologists, and the doctrinaire partisans of the Verstehen (understanding) approach hold. The serious social scientist is likely to go beyond commonsense knowledge, and is thus bound to cast doubt on much conventional wisdom, as well as to come up with some counterintuitive hypotheses. Let the following examples suffice to shore up the thesis that social scientists have found some counterintuitive generalizations: “The more similar, the fiercer the competition” (ecological law), “Rebellion occurs not when oppression peaks, but when it starts to slacken” (Alexis de Tocqueville), “Free trade favours only the powerful,” “Colonialism ends up by impoverishing the colonial powers as well as the colonies” (John Hobson), and “The spread of higher education lengthens the jobless queues” (Raymond Boudon). Each of these counterintuitive hypotheses can be supported not only by statistics but also by unearthing the pertinent mechanism. For example, the more similar two groups of organisms, the more needs they will share, and thus the closer their niches. Counterintuitiveness is of course even more pronounced in the natural sciences: just think of the “paradoxes” (counter-
266
Chasing Reality
intuitive hypotheses) of relativistic and quantum physics. To hold that science does not differ radically from everyday knowledge, just because both make use of empirical evidence (Haack 2003), is beside the point. The point of engaging in scientific research is to build and check theories about factual items inaccessible to common sense. Ordinary knowledge, even enriched with logic and empirical evidence, equals ordinary knowledge. To sum up, scientific hypotheses are expected to be testable, and moreover to undergo tests if found interesting. And scientism, far from being an ideology – as Hayek and Habermas have claimed–is no less than the epistemic engine that has been driving all of the sciences since the early 1600s. 5 Axiological Realism: Objective Values The conventional wisdom about values is that all of them are subjective: that value lies only in the mind of the evaluator. Accordingly, there would be no objective and true value judgments. This is the opinion of the axiological nihilists, such as Nietzsche; of the emotivists, such as Hume and the logical positivists; and of the intuitionists, such as G.E. Moore and Max Scheler. By contrast, axiological realists hold that, whereas some values are indeed subjective, others are objective because they are rooted in biological or social needs (see Bunge 1989 and Boudon 2001). For example, whereas beauty may well be only in the eye of the beholder, security and peace are objective social goods. That the preceding is not a dogma but an objective truth is suggested by the fact that insecurity and war put life at risk and therefore drag all the other values down. In other words, axiological realism rejects the dogmatic view that values must be accepted blindly either because they have a supernatural source or because they are purely a matter of feeling or intuition. Axiological realists are not intimidated by G.E. Moore’s branding of the attempt to join value to fact as perpetrating the “naturalistic fallacy.” Rather, realists regard the fact/value dichotomy as a supernaturalist or irrationalist fallacy. However, value objectivism does not entail value absolutism. First, one and the same item may be objectively valuable in some respects but disvaluable in others. For instance, riding a good bicycle is good for the body but it tempts thieves; and giving alms may make one feel good but it postpones social justice. Second, some values have to be restricted in order to secure others. For example, not all knowledge is valuable; thus, knowing the sex of a human fetus may lead to aborting female fetuses. Besides, some values are bound to conflict with others. For instance, privacy calls for limiting the right to know. Values may be individual or social. Individual values may be defined in
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
267
terms of individual needs and desires. Anything that contributes to meeting a basic need, such as food, shelter, or company, may be said to have a primary value. A secondary value is anything that contributes to satisfying a legitimate desire – that is, one whose satisfaction does not prevent anyone else from meeting his basic needs (for a finer partition and a quantitation see Bunge 1989). The social values, such as security, social cohesion, democracy, progress, and peace, promote individual welfare and peaceful social coexistence. However, sometimes they conflict with individual values. For example, security may conflict with liberty. However, such conflicts are best solved through negotiation and compromise rather than by coercion. In any case, the social values are just as objective and important as the individual objective values. Finally, scientism suggests that values, whether subjective or objective, can be investigated and defended or attacked on scientific grounds instead of being handed over to ideologues. For instance, artistic appreciation has been investigated experimentally (e.g., Berlyne, ed. 1974, Zeki 1999); basic needs are being investigated by physicians and social psychologists among others; and policy scientists are expected to design policies aiming at defending certain objective values. Thus, axiology should be regarded as straddling philosophy, science, and technology. 6 Ethical Realism I: Moral Facts and Moral Truths Ever since Hume, most philosophers have adopted ethical subjectivism or relativism. Consequently, they have held that there are no moral facts, and that therefore moral statements are neither true nor false. That is, the ethical tradition is irrealist. Let us put this tradition to the test in two typical cases: those of theft and altruism. I submit that the actions of both kinds are moral because they affect the persons at the receiving ends. And at first sight, the corresponding norms, “Don’t steal” and “Help the needy,” can be neither true nor false. However, they are true if the moral norm “Try to help others” is postulated So far, all may look rather clear and simple. But it is not quite so, because the norms in question have duals, or counter-norms, regarding mitigating circumstances – as is the rule with all social norms (see Merton 1976). Thus, we usually postulate that lying and stealing are wrong – except when human life is at stake. Likewise, we set limits on altruism: we do not normally require helping others at the risk of one’s own life. In both cases the facts of the matter are placed in context, and so are the corresponding moral principles. In a situational or systemic perspective, then, lying, stealing, and helping others without expecting reward are moral facts;
268
Chasing Reality
and the norms and counter-norms attached to those facts are true because they match the maximal moral principle “Enjoy life and help others live enjoyable lives” (Bunge 1989). In other words, the moral norms should be placed in social context rather than be separated from it, since they are to be used in real life. This is a requirement of moral realism, the view that there are moral facts and therefore moral truths as well. Though at variance with mainstream ethics, in particular with Kantianism, emotivism, and standard (subjective) utilitarianism, moral realism has been gaining currency in recent years (see, e.g., Lovinbond 1983, McGinn 1985, Bunge 1989, and Rottschaefer 1998). However, in many cases this view has been presented as a byproduct of the philosophy of language, or rather linguistic philosophy. Hence, this variety of moral realism is unlikely to appeal to any thinkers who regard words as conventional. Moreover, some moral realists, such as Platt (1979), are intuitionists – hence distant from rationalism as well as from empiricism; and others, such as Wiggins (1976), write about “the meaning of life.” But no one who takes science and technology seriously, and regards them as inputs to philosophy, can condone intuitionism; and genuinely analytic philosophers distrust musings about the meaning of life. Still other moral realists, particularly Brink (1989), profess ethical naturalism and objective utilitarianism. However, naturalism (unlike emergentist materialism) is untenable, since moral rules are social, hence partly artificial, as well as biological (or psychological). And utilitarianism does not account for the well-known fact that we often make sacrifices without expecting reward. For these reasons, I will presently sketch a different variant of moral (or rather ethical) realism. According to the correspondence theory of factual truth, the thesis that there are moral truths and falsities presupposes that there are moral facts. In turn, this presupposes that moral truths are just as factual as the truths of physics, biology, or history. Now, the vast majority of philosophers deny that there are moral facts. (Harman 1977, Nino 1985, and Rottschaefer 1998 are exceptions.) They base this denial on the fact-norm dichotomy usually attributed to Hume. Let us peek at the latter. Hume was certainly right in asserting that what ought to be is not the same as what is: that norms are not of the same kind as factual propositions. However, I submit that he was wrong in holding that the value/fact gap is unbridgeable. He was mistaken because we move daily from the one to the other, namely, when taking action. For instance, if I tell myself that I must pay my debt to someone, and proceed to paying it, I cross the gap between ought and is. Likewise, when I take note of an unjust situation and attempt to remedy it, I go the other way.
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
269
In other words, is and ought are undoubtedly distinct: they are separated by a conceptual or logical gap. But in practice this is just a ditch that we can jump over through action. Moreover, it is a ditch that not only conscious beings can cross. It can also be crossed by any artefact endowed with a negative feedback device capable of leading the system from its current state to the final state judged valuable by the operator or user. Furthermore, I suggest that every fact involving rights or duties, whether met or not, is a moral fact. For instance, poverty is a moral fact, not only a social one, because poverty involves unnecessary suffering and degradation. This is why it elicits moral emotions, such as compassion and shame. Involuntary unemployment is a moral fact too, because it violates the right to work, the source of all legitimate income necessary to meet basic needs and legitimate desires. Analogously, job creation is a moral fact, not only an economic one, because it satisfies the right to work. Another case of moral (or rather immoral) fact is the dumping of toxic waste, and this because it violates the right we all have to a clean environment. For the same reason, dragging a water body to remove toxic waste is a moral fact. In short, life in human society is packed with moral facts, and therefore with moral problems that can be tackled with the help of factual and moral truths. How should moral facts be identified or individuated? One option is to use this definition: “A fact f is moral = f poses a moral problem to some person in some culture.’’ In turn, a moral problem is an issue whose examination and solution calls for, among other things, the invention or application of moral norms. And of course the latter pose ethical and meta-ethical problems, which may have solutions leading to altering some moral precept. In turn, such change may have an impact on individual behavior and therefore on the original moral problem, for example, in contributing to either solving or worsening it. The moral facts and our moral, ethical, and meta-ethical reflections on them are then components of a feedback loop: Moral fact
Moral problem
Modified moral principle(s)
Moral principle(s)
Meta-ethical problem(s)
Ethical problems
In the end, then, moral truths resemble the truths of science and technology. But of course the two differ in an important respect: unlike the latter, the former are contextual, situational, or relative. Indeed, since moral truths concern in the last analysis rights and duties, and since these are relative to a culture and its moral code, moral truths are contextual. In this regard, and only
270
Chasing Reality
in this, moral truths resemble mathematical truths, in that they hold in some theories but not in others. However, this context-dependence of moral rules is not absolute: it does not involve the total relativity defended by the anthropological relativists and the social constructivists. This is so because all viable moral codes share certain principles, such as those of respect of the other and reciprocity. In other words, some rights and duties are basic, hence cross-cultural and non-negotiable, while others are secondary, and therefore local and negotiable. The right to enjoyable and non-harmful life, and the duty to help the needy when no one else can, are absolute and universal, whereas property rights and religious duties are not. The natural-law theorists, whether secular like the ancient Stoics or religious like Thomas Aquinas, deny of course this partial relativity of the moral norms. In fact, they hold that all moral rights and duties are natural and therefore universal or absolute, that is, context-free. But since as a matter of fact some moral codes do not admit certain rights and duties enshrined in alternative codes, the natural-law theorist cannot hold that the corresponding propositions are universally true. And if he nonetheless insists on his claim, then he cannot hold that such moral truths correspond to any facts. Instead, he must either state that they are just as formal as the mathematical theorems, or he must adopt a non-realist theory of truth. Given these difficulties, the natural-law theorist will perhaps take refuge in the traditional opinion that moral norms are neither true nor false, but only effective or ineffective, just like cooking recipes or medical prescriptions. But if he adopts this instrumentalist (or pragmatist) stand, he relinquishes the right to call himself a natural-law theorist. In short, moral facts are social, not natural: they belong to the fabric of society, not to that of nature. Moreover, those facts are artificial, that is, made, not found: they do not occur in the absence of moral agents. For example, there are still people among us, even Harvard law professors, who justify torture, the death penalty, and military aggression. This is why moral naturalism, or the view that morals are in our genes, is false. Morality is made and learned, not found or inherited. This is why it can be perfected or degraded, as well as be made the object of legislation. Like any other social facts, moral facts can be “perceived” (evaluated) differently by different people, or by the same persons under different circumstances. Interestingly, all such “perceptions” can in principle be objectified: that is, they can be detected by impartial observers. For example, Rilling and co-workers (2002) found that the reward (or pleasure) brain centres of experimental subjects playing a Prisoner’s Dilemma “light up” when they cooperate
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
271
with the experimenter. That is, it feels good to do good deeds–unless one has been educated in the hard school of mainstream economics, for which selfishness is the supreme virtue. That is partly why we often help others without expecting reciprocity. Experimental economists found long ago that small-scale business transactions involve a moral factor ignored by standard microeconomic theory. Thus Kahneman, Knetsch, and Thaler (1986: S299) wrote: “A realistic description of transactors should include the following traits. (1) They care about being treated fairly and treating others fairly. (2) They are willing to resist unfair firms even at a positive cost. (3) They have systematic implicit rules that specify which actions of firms are considered unfair.” Another empirical finding that falsifies the dogma that people act always only in their own best interests, that is, “rationally,” is the following. From a “rational” (selfish) viewpoint, it is obvious that whoever punishes someone else should not expect altruistic cooperation on the victim’s part, regardless of the fairness of the sanction. However, this is not the case: that is, fairness can make all the difference. Indeed, the experiments of Fehr and Rockenbach (2003) suggest that sanctions perceived as fair make no dent on altruism, whereas sanctions revealing selfish or greedy intentions all but destroy altruistic cooperation. In short, the utilitarian philosophers, neoclassical economists, and sociobiologists have been wrong in disregarding the moral factor or in construing altruism as enlightened selfishness. Experience and experiment show that common decency, fairness, and loyalty do matter in transactions of all kinds, even in business deals among strangers. So does unilateral generosity: Most of us spend time giving directions to strangers, tip waiters even when abroad, give to charities, prefer to deal with firms that treat their employees fairly, and boycott firms, products, and entire countries suspected of being morally tainted. I submit that all such facts are moral facts. If there are moral facts, then there should be moral truths. Certainly, most philosophers do not even entertain the possibility of moral truths corresponding to moral facts. In some cases the overlooking of this possibility is ultimately due to their thinking of moral principles or norms as imperatives, of the type of “Thou shalt not kill.” And although an imperative may or may not be pertinent, efficient, or good, it can be neither true nor false, because it neither affirms nor denies any matters of fact. This objection overlooks the fact that the linguistic wrapping of an idea is superficial because conventional. Indeed, any ordinary-knowledge idea may be expressed in different ways in any language and, a fortiori, in different languages. (The rider ‘ordinary’ excludes mathematical, scientific, and techno-
272
Chasing Reality
logical ideas: these require symbols that, though conventional, are fairly universal.) Since there are several thousand living languages, it is possible to give thousands of translations of any non-technical statement. In particular, any moral imperative may be expressed by a sentence in the indicative mood, with hardly any loss of content or practical value, in particular psychological or legal. For example, “Thou shalt not kill” can be translated into “Killing is evil,” “Murder is the worst sin,” “It is forbidden to kill,” “If you kill, expect to be punished,” and so on. The first two translations designate a proposition that holds (is true) in any moral code that affirms the right of all persons to life, and false in any code that does not admit such a right. (Note that the principle in question holds tacitly only for persons, not for human embryos or for non-human animals.) As for the legal counterpart of the moral principle in question, that is, “It is forbidden to kill,” it holds (is true) in the criminal codes that do not equate justice with revenge. In short, it is possible to propositionalize any given imperative. Such propositionalization is necessary to find out the truth-value (or the truth conditions) of moral maxims. Once a translation of an imperative into an indicative has been performed, one should proceed to a second translation: one in terms of rights and duties, or else virtues and vices, or good or bad consequences of an action. For example, “Help thy neighbour” can be translated into “It is good to help your neighbour,” or “Helping one’s neighbour is good.” The truth of this maxim becomes obvious upon expanding it into “If you take your neighbour’s interests into account, or wish to count on her in an emergency, or wish to be at peace with yourself, you’ll help her.’’ Likewise, the precept “Saving is virtuous while waste is not” is true in a world suffering from increasing scarcity. (The fact that almost all economists recommend increasing spending during recessions, while decreasing it during periods of expansion, only suggests that they are as amoral as shortsighted.) The methodological rule that holds for moral truths also holds, of course, for moral falsities. For instance, the thesis that what is good for “the economy” is good for the individual is false because “the economy” can be stimulated in the short term by spending on superfluous or even noxious commodities, such as sports cars and weapons. Likewise, the maxim “What is good for the government (or the Corporation, Church, or Party) is good for all” is false every time the government (or the corporation, the church, or the party) overlooks the welfare of the majority. Now, if there are moral falsities there must also be moral lies, that is, sentences expressing what the speaker knows to be moral falsities. For example, the claim that there are inferior human races is not only scientifically
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
273
false; it is also a moral falsity because it has been utilized to justify the exploitation or oppression of certain ethnic groups. And, whereas the statement of a moral falsity may be a mere error, that of a moral lie is nearly always sinful. The exception is of course the white or pious lie told to avoid unnecessary suffering. 7 Ethical Realism II: Testability of Moral Norms If all the moral norms, whether universal or local, concern moral facts, then it must be possible to subject them to empirical tests. I submit that a moral norm other than the maximal moral rule is justifiable if, and only if, (a) it fits a moral fact; (b) it is consistent with the supreme postulate of the moral code in question; and (c) it is effective in promoting prosocial conduct. The maximal moral norm is the benchmark against which all the other components of the code are gauged. And it is justifiable by its logical and practical consequences. Much the same holds for the axioms of a mathematical or scientific theory: they can only be justified by their consequences. I submit that the moral norms can be tested in three different though mutually complementary ways. The first test is that of coherence, that is, compatibility with the higher-level principles. The second is that of compatibility with the best available relevant knowledge (ordinary, scientific, or technological). The third test is that of the contribution to individual or social welfare. Let me explain. The moral norms, just like the scientific hypotheses and techniques, ought to be compatible with the highest-level principles, in this case the moral and meta-ethical maxims of the system in question. In the case of agathonism, the maximal principle is “Enjoy life and help others live an enjoyable life.” Any rule that facilitates the implementation of this principle will be deemed to be morally correct, otherwise incorrect. Example of the former: “Take care of yourself, your kin, and your neighbour.” An example of the latter kind is any rule leading to negative discrimination on the ground of age, sex, race, class, or religion. By contrast, positive discrimination, as in “affirmative action,” is justifiable when it corrects unfair imbalances. The second of our tests is that of compatibility with the best available knowledge. For example, we should discard any rule that ignores positive moral feelings, such as compassion, or negative moral feelings, such as shame and envy. The same holds for any rule that overlooks the fact that moral rules are social constructions, and that their implementation is likely to affect individual action and therefore human relationships. For instance, a rule that
274
Chasing Reality
commands human sacrifices to a divinity is not only cruel; it is also inconsistent with modern knowledge about myth and social convention. The third and last of our tests is that of efficiency. Evidently it consists in finding out whether the given norm facilitates or hinders the realization of the basic underlying values, such as those of survival and self-government, and the freedoms to love and learn. In this regard, the moral rules resemble the rules of modern technology: both are based on laws, and both are tested by their efficiency in reaching certain goals. Furthermore, the efficiency of moral rules, like that of the technological ones, can be estimated by the scientists and socio-technologists who study or control human conduct. A moral rule is good only if it is efficient on top of aiming at contributing to individual or social welfare. But, in order to be efficient, a rule must start by being viable, which in turn presupposes epistemological realism. For example, “If threatened, levitate” is not a viable rule. However, even if a moral rule were to pass all three tests, it should be regarded as perfectible. Centuries of scientific and technological experience, not to mention social experience and philosophical critique, ought to dissuade us from chasing infallibilist mirages, in particular that of a perennial ethics fashioned for perfect humans in the perfect society. We ought to have learned by now that no moral code and no ethical theory can guarantee right conduct. We may only hope to improve on the existing moral codes and ethical theories through conceptual analysis and comparison with empirical data. (For example, we must reject moral positivism, as well as legal positivism, for being both relativist and conformist.) Such ethical fallibilism is far from natural law, moral intuitionism, and the ethical doctrines attached to religious dogmas. Still, unless accompanied by meliorism, fallibilism is destructive. I suggest that we can and must seek moral and ethical progress through theoretical work and social action, voluntary as well as institutional. The abolition of slavery, torture, and the death penalty, as well as other progressive reforms of the penal code, are instances of real moral progress. Ethical meliorism may be justified as follows. First, morals coevolve with society, and moral philosophy evolves along with the rest of philosophy as well as with the social sciences and technologies. Second, if we try we may discover and correct moral deviations and ethical mistakes, although sometimes either calls for some intellectual or even civic courage. Take for instance the slogan Freedom or death, shouted since ancient times by uncounted freedom fighters, ideologists, and nationalist politicians. At first sight all friends of liberty ought to embrace it. But on closer inspection one realizes that the precept in question ought to be optional. To begin with, even a slave’s life may be worth living, particularly if it allows for some hope of
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
275
emancipation. Second, no one, not even the state, has the moral right to force anyone to fight till death for any cause. Third, the price exacted by any war without quarter can be staggering, not only because of the number of casualties but also for making the life of the survivors more difficult. The above slogan is justified only in two cases: when there is solid evidence that the enemy won’t give quarter and when the price of personal liberty is treason. What is the status of moral axioms? Are they wishful thoughts, commandments, conventions, or testable hypotheses? Our preceding discussion points to Einstein’s view (1950b: viii): “Ethical axioms are found and tested not very differently from the axioms of science. Die Wahrheit liegt in der Bewährung. Truth is what stands the test of experience.” In the cases of moral norms and ethical principles, the relevant empirical data concern human welfare. For this reason, the biological and social indicators, such as life expectancy, infant mortality rate, number of school years, and median disposable income, are more relevant to moral norms and ethical principles than are academic discussions on moral mini-problems or pseudoproblems, such as those posed by blasphemy, masturbation, homosexuality, abortion, in-vitro fertilization, sex change, gay marriage, clonation, or suicide. All of these pale by comparison with war, misery, and tyranny (see Waddington 1960). Assuming that we have solved the problem of validating moral rules, what can we say about testing ethical theories, that is, systems of ideas about the nature, root, and function of moral norms? I hold that such theories are testable much in the same way as scientific and technological theories, that is, by their agreement with the relevant facts and their compatibility with other theories. Let me explain. I submit that any ethical theory should meet the following conditions: 1 Internal consistency: non-contradiction. 2 External consistency: compatibility with the bulk of scientific and technological knowledge about human nature and institutions. 3 Ability to account for viable (livable) moral codes. 4 Usefulness in suggesting social reforms necessary for the exercise of free and enlightened moral judgment. 5 Usefulness in analysing moral and ethical concepts and principles. 6 Usefulness in identifying and tackling moral problems, and in settling moral conflicts. The condition of internal consistency does not apply to stray ethical opinions, such as Machiavelli’s or Nietzsche’s immoralism. The condition of
276
Chasing Reality
external consistency disqualifies ethical emotivism (such as Hume’s and the Vienna Circle’s) and ethical intuitionism (such as Moore’s and Scheler’s), since these views overlook the rational and empirical inputs to all moral deliberations. The same condition disqualifies utilitarianism, for it ignores the reality of moral feelings, such as empathy and shame, and it makes use of subjective values and subjective probabilities. It also disqualifies all the nonconsequentialist ethical theories, in particular Calvin’s and Kant’s. The fourth condition disqualifies naturalism (or vulgar materialism), for it ignores the social root and social efficiency of morals. The fifth disqualifies all of the moral doctrines associated with philosophies hostile to logical analysis, such as intuitionism, phenomenology, existentialism, and dialectical materialism. And the sixth disqualifies all of the moral doctrines that are but arbitrary inventions of intellectuals far removed from social reality–which, alas, has been the case with most moral philosophers. Let us wind up our discussion of moral realism. The thesis that there are moral facts and moral truths has at least three consequences. First: If there are moral facts, then the moral principles are not dogmas but hypotheses, as Ingenieros (1917) and Einstein (1950b) proposed. And, being hypotheses, they must be confronted with the relevant facts, and revised if they do not match them. However, it does not follow that moral facts are just as inevitable as astronomic events. Whether or not such facts happen depends on us, moral agents. Second: If there are moral facts, then moral judgments are not totally subjective and relative. Rather, on the contrary, it is possible to evaluate such judgments in the light of experience, as well as to discuss them rationally. Moreover, it is possible to adjust action to morals and conversely. In particular, it is possible and desirable to reform society so as to minimize the frequency of destructive moral facts such as crime, war, oppression, and exploitation. Third: If the moral and ethical principles are empirically testable hypotheses, as well as guides for individual and collective action, then it is possible and desirable to reconstruct moral codes, as well as ethical theories, in a rational manner. In particular, it is possible and desirable to reconstruct morality and its theory in such a manner that the following conditions be met. 1 Realism: Adjustment to the basic needs and legitimate aspirations of fleshand-blood people placed in concrete social situations. 2 Social utility: Ability to inspire prosocial conduct and progressive social policies, as well as to discourage antisocial ones. 3 Social plasticity: Adaptability to new personal and social circumstances. 4 Equity: Efficiency in the task of decreasing social inequalities (though without imposing uniformity).
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
277
5 Compatibility or coherence with the best available knowledge of human nature and society. In sum, there are moral facts and moral truths. The former are part of the fabric of reality, and moral truths intertwine with other factual truths. Consequently, moral realists are expected to assume social commitments, that is, to practise morality in addition to teaching it. 8 Practical Realism: Efficiency and Responsibility Praxiology, or action theory, is the branch of practical philosophy concerned with deliberate human action. In its narrow construal, praxiology seeks to determine the conditions for efficient action regardless of moral considerations (von Mises 1966, Kotarbinski 1965, Gasparski 1993). I submit that, since our actions may affect others, we should try to foresee their consequences and assume responsibility for them – a duty that the ethical non-consequentialists exempt themselves from. Not to do so is unrealistic, because it involves leaving out an important component of the outcome of anyone’s actions. The obvious consequence for praxiology is that it should include responsibility and accountability along with efficiency. In other words, a realist praxiology presupposes a realist morality. Further, it is doubtful that there can be general efficiency conditions. Indeed, maximizing the Output/Input ratio of any artefact, whether physical, biological, or social, requires the study of the specific Actor/Object interface. Thus, it is not the same to design or construct an efficient engine, an efficient factory, or an efficient school. This is why the study of efficiency is a task for engineers, management scientists, ergonomists, and other technologists, rather than a task of action theorists. The most advanced and popular of all action theories is rational-choice theory. According to it, every action, or at last every rational action, is undertaken freely and in the light of a cost/benefit calculation, and it obeys the utilitarian imperative “Maximize your expected utilities!” (see, e.g., Becker 1976). This theory, born in neoclassical microeconomics, has spilled over other social sciences, and even ethics. Regrettably, rational-choice theory is conceptually fuzzy because neither of its two key concepts, those of subjective probability (perceived likelihood) and subjective utility (pleasure), is mathematically well defined (see Eichner, ed. 1983 and Bunge 1996). Besides, the theory is empirically untenable because (a) most events in ordinary life are not random, hence they cannot be assigned probabilities proper (recall chapter 4); (b) most problems in daily life are solved by following ready-made rules, without resorting to sophisticated cal-
278
Chasing Reality
culations; (c) real-life actors are seldom free, and they are never omniscient; and (d) normal people are constrained by social and moral norms. Consequently, rational-choice theory is not a realistic guide to life. To go back to the scope of praxiology, I submit that philosophers are not equipped to design efficiency rules. What philosophers can do is to analyse the system of concepts centred in the notion of action, as well as to link praxiology to ethics. Here we shall only sketch a few such ideas insofar as they relate to the realism requirement (see Bunge 1998 for details). I suggest that a good action is one that satisfies two conditions: the technical condition of optimal efficiency, and the moral condition of being more beneficial than harmful. As a rule, optima lie midway between minima and maxima. The reason is that the variables that describe a system are interrelated, so that when one of them increases others are bound to decrease. In particular, we should not aim for maximal efficiency, because it is bound to sacrifice other values, such as welfare and environmental protection. But, of course, we want to limit overall waste. Utilitarians disregard the moral aspect of action, but of course this aspect is taken into consideration by anyone who respects other people’s rights. Now, every human action is bound to affect other people, as a rule positively in some respects and negatively in others. Hence, before doing something important we should estimate the costs and benefits of the action in question to self and others. For example, before preaching the virtues of globalization (or free trade) we should find out how to correct or compensate for the increasing imbalances that it generates. Optimally efficient design and implementation implies ontological, epistemological, and semantic realism. Indeed, an irrealist approach to action guarantees practical failure because of flawed design, incompetent execution, or both. And right action implies moral realism, the view that there are moral (and immoral) facts and rules. In fact, moral antirealism, in particular the view that morals are subjective, leads to overlooking the rights of others; hence the harm that action may cause them – and the actor by reaction. Let us finally glance at one aspect of praxiology: political philosophy. This is the branch of practical philosophy that studies the merits and demerits of the various historical and possible social orders, and the ways to perfect or undermine them. A realist political philosophy is expected to have two roots: a realist praxiology and a realist social science. Both roots are necessary to avoid costly failure. Political science teaches that political action can only succeed if its underlying strategy is based on a realist study of the polity in question: one that, through statistics, opinion polls, and public debates, exhibits flaws, aspira-
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
279
Ethics Social issues
Social policies
Science & Technology Figure 10.3 Social issues should be tackled in the light of both the relevant knowledge of social reality and moral principles.
tions, and possibilities. It also teaches that, while the vast majority of people favour reform, only elites opt for revolution. There are two reasons for this preference of ordinary folks for evolution over revolution. One is that most people wish to get on with their lives rather than devote themselves to a risky cause. The other reason is that all social action is likely to have unanticipated consequences, some good but others harmful – and small mistakes are easier to repair than big mistakes. Social reform, however, is not the same as piecemeal social engineering. Indeed, sociology suggests that sectoral social reforms are bound to fail, because society is constituted by many interdependent sectors or subsystems rather than by a single one, such as the economy, the polity, or the culture. In other words, only systemic (or integral) political programs may succeed – and this provided they win the support of ample sectors of society rather than appealing only to a sect. Thus the correct slogans: Study reality before attempting to change it, and Try to improve everything at once – though gradually rather than abruptly. (More in Bunge 1998.) In short, to improve on a ghastly a social reality while avoiding both utopia and dystopia, we should design social policies and plans based on both realist social studies and moral means-goal pairs. See figure 10.3. 9 Scientific Hylorealism Realism is essential to the empirical exploration and transformation of reality. The reason is that no one can investigate the inscrutable or alter the inaccessible. Now, realism can be wedded to idealism, as in the case of Plato; materialism, as in Democritus; or to dualism, as in Aristotle and Descartes. (For materialism see chapter 1, section 6.) Which combination is likely to be the more fertile? An objective idealist, or Platonic realist, is bound to make claims that are at best untestable and at worse false, such as that the laws of nature precede the law-abiding things (Heisenberg); that the real plants are but imperfect copies
280
Chasing Reality
of the ideal Urpflanze or primeval plant (Goethe); that “in the beginning was the Word” (St John); that “the word is the abode of being” (Heidegger); or that disembodied ideas can move neurons (Eccles). A dualist or hylomorphist, like Aristotle, would make only half of the claims of this kind, and so he would fail to satisfy the upholders of the other half. Only a consistent materialist will insist that solely material objects, that is, changeable things, are objectively real. By the same token, he will look for universals, particularly laws, in re rather than ante rem. Realism, then, is not enough. Indeed, what is the point of professing realism while admitting the reality of supernatural beings or extrasensory perception, the independent existence of ideas, Dilthey’s “objective mind” (or Popper’s “world 3”), or parallel universes? All this idealist fog burns in the strong light of materialism, which postulates that there are no ideas without brains (recall chapter 1, section 6). In short, realism without materialism is vulnerable. Realism can only stay sober, testable, and effective, as well as open to new ideas, provided it is combined with materialism and scientism. Materialism without realism and scientism is dogmatic because only the investigation of reality can corroborate it. Worse, vulgar materialism can be obnoxious: think of Nietzsche’s joint adoption of vulgar materialism (physicalism), pragmatism (in particular fictionism and vitalism), anti-scientism, and immoralism. Remember also that, partly because of his epistemological and ethical nihilism, Nietzsche helped discredit rationality and pave the way to postmodernism and fascism (see, e.g., Wolin 2004 and Vacher 2004). Finally, materialism, not even combined with realism, suffices to guide the search for truth and smoke out pseudoscience. Think of the sociobiological attempt to explain everything human – even politics, crime, morals, and religion – exclusively in genetic and evolutionary terms. Surely, all this is both materialist and realist; but it fails the empirical tests: it is sheer fantasy. Scientism, the thesis that scientific research is the best cognitive strategy, must be blended with materialism and realism if we wish to keep pseudoscience at bay. (Caution: Far from involving radical reductionism, scientism is compatible with emergentism: see Bunge 2003a.) In sum, to help understand and control reality, materialism must be combined with realism and scientism. This triad may be called scientific hylorealism. 10 Concluding Remarks Integral philosophical realism is a system that embraces all the branches of philosophy except logic: it covers ontology, semantics, epistemology, methodology, value theory, ethics, and action theory. I submit that it is the philosophy
From Plato’s Cave to Galileo’s Hill: Realism Vindicated
281
Table 10.2 Integral philosophical realism vs. antirealism (nihilism), or Galileo and Einstein vs. Nietzsche and Heidegger Discipline
Item
Realism
Antirealism
Ontology Epistemology Semantics Methodology Axiology Ethics Praxiology
Universe Knowledge Truths Best strategy Values Morals Actions
Real and single Possible Some Scientific method Some Ego-altruism Right
Unreal or plural Impossible None Intuition None or evil Egoism None or antisocial
that nearly everyone practises when trying to solve everyday problems. Only philosophers can profess antirealism, and this solely while writing or teaching. See table 10.2. Mathematicians do not face the choice between realism and antirealism because they do not study reality: they deal only in fictions. By contrast, scientists and technologists behave like realists – even if some of them pay lip service to antirealist philosophies – because their task is to study or design real things. Likewise, serious philosophical investigation is realist: it faces facts, tackles genuine problems, and incites us to go beyond appearance, idle fiction, and knee-jerk action. Only philosophical realism encourages us to chase reality in order to understand or control it. However, realism is not enough: to be deep and efficient, it must fuse with scientism. The moment it does so, realism attracts materialism. This is so because science shows that all the constituents of reality are material, ideas being only processes in highly evolved brains. Thus, scientism somehow entails realism and materialism. Therefore, any attack on either component of this triad is bound to cripple the other two. Furthermore, scientism suggests the possibility of building a scientific practical philosophy. This would include a moral code and a political philosophy designed for real persons in real societies, rather than for angels in utopias. That is, such a code would be designed for individuals facing real moral dilemmas in real social systems: persons with needs and aspirations, as well as with rights and duties. I suggest that scientists and social technologists are best equipped to find out which needs are real, which aspirations legitimate (compatible with meeting other persons’ needs), and which types of social organization are beneficial to nearly everyone, as well as viable and sustainable. Obviously, this project goes beyond the limits of the present book (see however Bunge 1989 and 1996 for previews).
282
Chasing Reality
In sum, we have defended not only realism but also materialism, scientism, and the project of a scientific ethics. Far from constituting an unstructured set, these four theses constitute the coherent system of research projects shown in this diagram: Scientism
Realism
Practical philosophy
Scientific philosophy
Materialism
Fact and Pattern
283
Appendix: Fact and Pattern
There is an amazing confusion in the contemporary philosophical literature concerning the basic ontological concepts of thing, system, property, state, event, process, fact, and law. For example, some logicians, echoing Plato, state that a fact is whatever makes a proposition true. Wittgenstein (1922) held that the world is the totality of facts, not things – as if facts were above concrete things. Likewise, David Armstrong (1999) has proposed to start ontology with the concept of a state of affairs, as if this could exist separately from things. Whitehead (1919) had famously defined a thing as a bundle of events – as if an event were anything other than a change in a concrete thing. Kim (1998) defined an event as a property exemplification, rather than a change in one or more properties. And so on. Never mind that ontological concepts can only be elucidated in an ontological theory rather than one by one. And never mind what the people who study and manipulate facts, namely, the scientists and technologists, mean by ‘fact’ and its kin. The situation concerning patterns, regularities, or laws is just as bad. For example, very few philosophers distinguish objective laws or patterns from the hypotheses representing them – a distinction familiar to every physicist at least since Ampère (1834). This is why, when confronted with a law-statement that proves to be only approximately true, some philosophers have concluded that nature is inexact (Boutroux 1898). And, of course, many-worlds metaphysicians feel free to imagine “worlds” satisfying quaint laws or even none – as if metaphysics were a branch of fantastic literature. (Recall chapter 9, section 6.) Worse, nowadays “scientific facts” – meaning facts studied by science – are often said to be social constructions or conventions on par with road signs and table manners. According to this collectivist variety of subjectivism, there would be no difference between facts and data, laws and rules, or even sitters
284
Appendix
and portraits. All this, they claim without offering any evidence, should be lumped into a single category, that of “social construction.” Accordingly, the notions of reality, objectivity, and truth would be redundant. Scientists would not study facts: they would make them up. They would even restrict themselves to chatting, “making inscriptions,” and “negotiating” with colleagues (e.g., Rorty 1979, Latour and Woolgar 1986). These extravagances have been so ridiculed (e.g., by Sokal and Bricmont 1998, Bunge 1999, and Brown 2001) that one of the main offenders (Latour 2004) has recently recanted. Given the above-mentioned confusions, it may be apposite to re-examine the ideas of fact, pattern, and their kin that are utilized in the scientific and technological literature. They might prove useful to understand what scientists, technologists, businessmen, and policy-makers are up to. 1 Thing, Property, and Predicate The most basic or general concept of a thing is that of a bare substantial individual. This is defined as anything that can join another individual to form a third individual. More precisely, we stipulate that x is a bare substantial individual if and only if x can associate, join, or concatenate with other individuals, forming further individuals. Shorter: x belongs to a collection S having the structure of a semigroup. (For a detailed formalization and discussion of this concept and its kin, see Bunge 1977a.) Of course, real things have other properties, such as energy, in addition to the ability to join with other individuals. A real thing is a substantial individual endowed with all of its properties. In turn, a property of a thing may be conceptualized or represented by an attribute or n-ary predicate. And a predicate F may be analysed as a function from a certain domain A to some codomain B, or F: A ® B, where A may be the Cartesian product of a number n of collections (Bunge 1974a). In particular, if F is a qualitative unary predicate representing an intrinsic property P of substantial individuals, or members of the collection S, F may be analysed as the function F : S ® Propositions containing F. Thus, if b is a substantial individual, that is, if b is in S, then the value of F at b is F(b), read ‘b is attributed F-ness.’ (Note that this analysis of predication differs from Frege’s, who characterized predicates as functions from individuals to truth-values.) All of the qualitative properties, such as cohesiveness and stability, are representable by such unary predicates. If G is a binary attribute representing a relational property Q of pairs of things, such as separateness or interaction, then
Fact and Pattern
285
G : S × S ® Propositions containing G. For example, if b and c are in S, G(b,c) is read ‘G is predicated of the ordered pair of individuals .’ The generalization to higher-order predicates is obvious. Note that we have tacitly distinguished properties from functions, in particular predicates or attributes. This distinction is unnecessary in logic and mathematics, where the two concepts coincide, because all the mathematical objects are of a single kind, namely constructs. But the distinction is indispensable everywhere else, for one and the same property of a substantial individual may now be conceptualized as a given attribute and later on, in the light of new knowledge, as a different attribute. Remember the historical changes undergone by the concepts (or rather the words) ‘energy’ and ‘quantity of motion.’ Realists assume that the physical world took no notice of such conceptual changes. Hence the need to distinguish attributes predicated of things from properties possessed by things. Another rationale for the distinction in question, and one that necessarily escapes the irrealist, is that not all predicates represent properties of real things. In particular, negative predicates, such as “mindless,” and disjunctive predicates, such as “conducting or working,” represent no properties of things even though they may occur in our discourse about things. (Disjunctive predicates are legitimate provided they define genera, that is, logical sums of species. Examples: Herbivores c Carnivores = Omnivores, and Blue collar c White collar = Worker.) A third reason for drawing the predicate-property distinction is that, whereas predicates satisfy some system of predicate logic, and the set of all properties is a Boolean algebra, the properties of things satisfy natural or social laws. More on this below. The following rather typical examples (taken from Bunge 1977b) should suggest the rich variety of property-representing functions we must cope with, and will prepare the ground for a general definition. Example 1. Dichotomic global property. Let F : A ® B represent stability. Then, A = Collection of all concrete systems (physical, chemical, biological, social or technical), B = Set of all propositions of the form “x is stable,” where x is in A. Example 2. Qualitative global property. Let F : A ® B represent social structure. Then A = S × T, with S = Set of all human societies, T = Set of all time instants, B = Family of all collections of persons,
286
Appendix
and, for s in S, and t in T, F(s,t) = Family of all social groups included in society s at time t. Example 3. Quantitative global property. Let F: A ® B represent the electric charge on a body. Then A = C × T × Uc, where C = Collection of all bodies, T = Set of all instants, Uc = Set of all electric charge units, B = Set of all real numbers, so that ‘F(c, t, u) = r,’ for c in C, t in T, u in Uc , and r in , abbreviates ‘The electric charge of body c, at time t, expressed in unit u, equals r.’ Example 4. Quantitative global stochastic property. Let F : A ® B represent the momentum probability distribution of a quanton. Here A = Q × F × T × R3, where Q = Collection of quantons (quantum-mechanical entities) of a certain species, F = Set of all reference frames, T = Set of all time instants, B = R = Real line, so that ‘F(q, f, t, p) dp’ abbreviates ‘The probability that quanton q, relative to frame f, at time t, has momentum lying between p and p + dp.’ (In standard notation, F = j ) 2, where j is the Fourier transform of the state function y.) Example 5. Quantitative local property. Let F : A ® B represent the gravitational potential. Then A = G × F × E3 × T× Ue, where G = Collection of all gravitational fields, F = Collection of all reference frames, E3 = Euclidean three-space, T = Set of all time instants, Ue = Set of all energy units, B = R = Real line, so that ‘F(g, f, x, t, u) = r’ abbreviates ‘The (scalar) gravitational potential of field g in G, relative to frame f in F, at point x in space E3, and at time t in T, expressed in energy unit u in Ue , equals the real number r.’ Further, the field intensity is defined as the gradient of the field potential. Both magnitudes represent the same thing, namely, the field. The choice between them is a matter of convenience. Example 6. Population is a conspicuous biological and sociological property. It may be conceptualized as a state function P: S × T ® N from pairsto the natural numbers. Every value P(s,t) = n, for n in N, represents a possible individual property of s at t, whereas the function P itself represents a general property, or a property of
R
R
)
Fact and Pattern
287
all the members of the collection S. (Shorter: P is universal in S, or P is a universal of S.) In each of the above cases the central predicate F represents a property of entities of some kind, or a general property, such as age. And the value of F for some particular entity or thing is an individual property, or property possessed by the individual in question, such as being thirty years old. Individual properties are called ‘tropes’ in contemporary ontology. 2 State and State Function Some properties, or rather property-representing functions, are of special interest to us: they are those describing the states a thing can be in. For example, the mass, stress, and force densities in a mechanical system determine its dynamical properties. They are therefore called state variables or, better, state functions. By contrast, stability, though a property of systems, is not a state function but a sort of outcome of the interplay of certain properties. Nor are t (time), f (reference frame), or u (unit of some kind) state functions – nor, indeed, are they functions of any kind. They are arbitrary members of certain sets, and they are not possessed by any thing in particular. They are rather “public,” in the sense that they can be applied to a number of things. (Caution: In any relational theory of time, t must be a value of a function defined for pairs of events.) In Example 2 above the function Fs: {s} × T ® B is a state function for the individual system s. In Example 3, the function Fcu : {c} × T × {u} ® is a state function for c. In Example 4, the function Fqf : {q} × {f} × T × 3 ® is a state function for q relative to f. And in Example 5, the function Fgfu : {g} × {f} × E3 × T × {u} ® is a state function for g relative to f. The preceding remarks suggest the following preliminary characterization. A function is a state function for a thing of a given kind only if it represents a property possessed by the thing. Whether this representation is faithful (true) is not essential to the function’s qualifying as a state function. What is decisive is that the function should refer to the thing and be interpretable as representing or conceptualizing the intended property. The reason is that we must build theories before we can put them to the test to find out whether they are reasonably true. Nor is it necessary for a state function to occur in a theory. State functions are also useful in empirical inquiry, to display the current state of a system and follow its course. For example, a state function for a nation could be a rather long list of values of environmental, biological, economic, political, and cultural indicators, such as average rainfall, life expectancy, GDP, voter turnout,
R
R
R R
288
Appendix
and literacy rate. But, of course, it is best if such empirical values can be plugged into theoretical formulas such as law-statements, for in this case one can hope to explain and predict the facts in question. Every scientific theory refers to concrete things of some kind or kinds. And every such theory, whether general such as Maxwell’s electrodynamics or specific such as a model of a circular antenna, involves a finite number of state functions to describe its referents. Since some scientific theories, such as Einstein’s theory of gravitation and quantum electrodynamics, are true to a remarkable degree, it is reasonable to assume that the things they concern do in fact possess only a finite number of general properties. Let us now consider the whole bunch of state functions for things of a given kind K. In principle they have little in common except for their common reference to things of the given kind. In particular, they need not even be defined on exactly the same domain. (For example, one of them may be defined on K, another on K × T, and a third on K × F × T.) However, a harmless trick will assign them all the same domain. Thus, if Fl : A ® B, and F2 : C ® D, where A ¹ C and B ¹ D, we may adopt the new state functions G1 = A × C ® B, such that G1(a,c) = F1(a) G2 = A × C ® D, such that G2(a,c) = F2(c) for all a in A and all c in C. But we seldom need to resort to this trick because, as a matter of fact, many state functions for things of a given kind are defined on a common domain. Think for instance of the collection of property-representing functions (such as mass density, velocity, and field potential) concerning a continuous medium such as a fluid or a field: they may all be construed as sharing a single domain, namely, some four-manifold. Likewise, the set of dynamical variables (“observables”) of a quantum-mechanical thing are operators in the Hilbert space of the thing in question. Still, whether or not this is in fact the case in a given instance, the above-mentioned procedure will do the trick of regimenting the domain of the state function for the given thing. Therefore we may adopt Definition 1. Let Fi : A ® Vi, with i in N, be a state function for things of a given kind. Then the function F =: A ® V1 × V2 × ... × Vn, such that (a) = , for a in A, is called a (total) state function for the things of the kind concerned. (If all the Vi are vector spaces, F is called a state vector.) Notice the cautious indefinite article in the above phrase ‘a state vector.’ The reason is that there is no such thing as the state function for things of a given kind. Indeed, there are as many state functions as representations (or models) of the thing can be conceived of, that is, any number of them. (Analogue:
Fact and Pattern
289
There can be any number of portraits and photographs of one and the same person.) For example, whereas Lagrangian theories employ generalized coordinates and velocities as the basic state variables (functions), Hamiltonian theories use generalized coordinates and momenta. Furthermore, one and the same representation is compatible with infinitely many choices of reference frame, every one of which will ensue in a different state function. (More on this below.) This is an instance of the principle of scientific realism, according to which any concrete thing may be represented in different ways. Hence, the antirealist will be unable to make any sense of the formulas that relate different representations of one and the same fact – such as the Lorentz transformation formulas occurring in special relativity. How do we test for the suitability of a given state function? Ultimately by testing for the adequacy (factual truth) of the model or theory as a whole, in particular that of its key formulas. These are the formulas interrelating the various state functions, that is, the law-statements of the theory – on which more below. Still, there may be alternative though basically equivalent formulations of one and the same theory. (For example, most field theories can be formulated using either field strengths or field potentials, and the latter are mutually equivalent modulo certain arbitrary constants or functions.) In the case of equivalent theories, there are no preference criteria other than those of computational convenience, ease of interpretation, heuristic power, or even sheer beauty or fashion. (All of these criteria were tacitly adopted when wave mechanics was preferred over matrix mechanics.) To put it negatively: The choice of state function is not uniquely determined by experimental data, but depends partly upon our total knowledge, as well as upon our abilities and goals, and even our inclinations. This consideration will play an important role in any talk of states and state spaces, on which more in the next section. Lest I have given the impression that the choice of state functions is utterly arbitrary, and therefore strictly a matter of taste, let me hasten to state that, whichever state function one chooses, it is expected to satisfy the law-statements included in the theory – and this is far from being a matter of convention. Neither is the choice of law-statements arbitrary: such statements are expected to be reasonably true to fact. Shorter: Every scientific theory contains some conventions, but the choice of theory is not conventional. More on this in the next section. The concept of a state function may be used to elucidate the notion of a concrete system such as a political party, in contradistinction to an aggregate, such as a constituency. Indeed, an analysis of a state function for an entity can
290
Appendix
help us discover whether or not it constitutes a system, that is, whether or not it has parts that act upon one another and are thus held together. Indeed, the state function of an aggregate of non-interacting things is uniquely determined by the partial state functions. Typically, the state function of a whole of this kind equals either the sum or the product of the state functions of the parts. Not so in the case of a system proper: here the state of every system component is determined, at least partly, by the states of the remaining components. Hence, the total state function is no longer separable either additively or multiplicatively. The contributions of the parts become inextricably entangled. (This should suffice to wreck the program of methodological individualism in social science.) A couple of examples should bring this point home. Think of a system composed of a population of foxes and a population of rabbits occupying the same territory. The rate of growth of each population depends on the numerosities of both populations. (Each of them varies sinusoidally, though out of phase, in the course of time.) The two populations are so entangled that the state function of the whole cannot be decomposed into two partial state functions. A methodological individualist could not even pose this familiar ecological problem, because he does not admit the very concept of a system. If it were up to him, the foxes would die of hunger and the rabbits would multiply without bounds. Actually, the methodological individualist could not even approach the problem of modelling a system containing two electrons close enough to interact appreciably, as is the case of those in a helium atom. This system is described by a Schrödinger equation (classically, by two coupled equations of motion) jointly with the Pauli exclusion principle. The latter is not derivable from the former, and it only applies to entire systems of a certain kind; it selects those total state functions that are odd or antisymmetric in the coordinates of the electrons. (That is, the additional law-statement is “y(x,y) = –y(y,x),” where x and y are the electron coordinates.) This principle expresses an emergent systemic property, that is, one that the individual components do not possess, whence it cannot be represented by the partial state functions. Therefore, the construction of a state function for the system must proceed from scratch rather than on the basis of the state functions of the individual electrons. The individualist strategy fails once again. (More on individualism in Bunge 1996, 1998, and 2003a.) 3 State Space and Event We are now in a position to elucidate the notions of state and change of state. Let F : A ® V1 × V2 × ... × Vn be a state function for things of a certain kind.
Fact and Pattern
291
Every value F (a), for a in A, represents a state of a thing of the kind in question. (In particular, if A = T = the set of all instants of time, the value F (t), at t in T, will be said to represent the state of the thing in question at time t. Note, however, that the concept of time has not been presupposed: so far, t is just a dummy.) Since F spans the lawful state space SK for things of kind K, every state of such things is a point in SK. Note also that there are no states in themselves, any more than there are properties in themselves – except of course in a Platonist ontology. This is no dogma: it is shown by the way state functions are constructed. In fact, every one of these functions represents, by hypothesis, the states of a concrete thing. This remark should suffice to dispose of the ontologies that take states (or states of affairs) as the building blocks of the universe. The “life” of abstract ideas, such as the concept of a set, is utterly uneventful. Only concrete things can change. Now, change can be instantaneous or nearly so, such as a quantum jump or a car collision; or it may take some time, such as changing places and learning. In the first case one speaks of events, and in the second of processes. Furthermore, as Aristotle taught us, a change may be merely quantitative, as in the case of motion, or qualitative as well, as in the case of the assembly or the breakdown of a system. A light flash, the dissociation of a molecule, a storm, the growth of a bud, the learning of a new trick, and a change of government are events. Actually, these particular events are processes, for – unlike the elastic collision of two atoms – they are complex, that is, they can be analysed as strings of events or, better, as sequences of states. Being changes of state of things, events are representable as pairs of states, and processes as trajectories in a state space of the things where they occur. By contrast, the emergence of a new property is representable as the budding of a new axis in the state space, and the submergence of a property by the pruning of an axis. Because states are relative to the reference frame and the representation (in particular the choice of state functions), their changes, or rather their conceptualizations, are relative in the same sense. Hence, one may well construct a state space on which the representative point is stationary, and alternative state spaces in which the point is moving. Every conceptualization of change, then, is relative to frame and representation. (Caution: Every observer is a possible frame of reference, but the converse is not true. Hence, relativity to a frame is not the same as relativity to a subject, or subjectivity.) Example 7. Suppose a thing that, in the respect of interest, can be in either of three states – for instance, on, off, and transient, as in the case of a switch. Calling these states a, b, and c, we have S = {a, b, c}. Form now all the ordered pairs in S × S:
292
Appendix
The thing stays in state a, that is, the identity event at a, The thing jumps from state a to state b, and so on. We have thus altogether nine elementary (not composite) events in S × S: e1 = = ua , e4 = , e7 = e2 = = ub , e5 = , e8 =e3 = = uc , e6 = , e9 = That is, S × S = {ua, ub, uc, e4, e5, e6, e7, e8, e9}, where the u’s are the identity events, which actually are non-events. Suppose now that all of these events are lawful, that is, really possible. In this very particular case, then, the event space is E = S × S. (In general, E is properly included in S2.) A standard representation of finite event spaces like this one is a Moore graph, familiar from computer science. Certain events can compose to form more complex events, others cannot. For example, event e4 = above can be followed by event e5 = , but not by e9 = . The composition of the events e4 and e5 amounts to the net event e6 = . In other words, event e6 can be analysed as the composition of events e4 and e5. To symbolize the composition of events we use the asterisk and write: e4 * e5 = e6, or, explicitly, * = . (Intermediate states do not show up in the net change: they are absorbed.) On the other hand, the complex “event” * , which is the reverse of the former, cannot occur, and so it is left undefined. In other words, * is a partial (not everywhere defined) operation in the event space E. More precisely, * is an operation such that, for all a, b, c, d in S, iff b = c * = not defined iff b ¹ c . If e and f are two events in E, then e * f = g is another event in E, consisting of event e followed by event f. In general, not all state transitions are possible. For example, the transition <dead, alive> is impossible. Moreover, if a state transition is possible at all, it may occur in more than one way. That is, different trajectories in a state space may have the same end points. In other words, two different processes, one along a curve g and the other along a different curve g¢, may result in the same net change. The two functions must be lawful if we are to allow only lawful events and discard lawless ones, such as miracles. In other words, g and g' must be compatible with the laws of the thing under study. (They need not be law functions. In general they will be laws cum circumstances, for instance field laws together with specifications of the field sources and the boundary conditions.) This suggests introducing Definition 2. Let S be a state space for things of some kind, and let s and s'
{
Fact and Pattern
293
be two points in S. Then a lawful event (or process) with end points s and s’ will be representable by a triple <s, s', g>, where g: S ® S is compatible with the laws in question. For a particular transformation g we may focus on the end points of the net change, that is, on the initial and final states of the process. The ordered pairs <s, s'> in Sg × Sg constitute the space of events of the things considered for the transformation g, or Eg for short. In general, Eg is a proper subset of Sg × Sg, because g may exclude certain conceivable but physically impossible changes. 4 Process A process may be defined as a string of events, for instance, p = e1 * e2 * e3 * ... * en . However, this definition only works for finite state systems such as digital computers – or rather the idealized models of computers that occur in mathematical computer science, as different from the models used by computer engineers. Indeed, even the simplest physical thing, such as an electron or a photon, can be in any of a non-denumerable set of states, and it can undergo continuous changes in addition to quantum jumps. This is one of the many reasons that the computer metaphor of the brain and its mental functions is so shallow and misleading. A more realistic definition of a process, applicable to all state spaces, is this: A process is a lawful sequence of states. More precisely, we adopt Definition 3. LetF : A ® V be a state function for things of a certain kind, and SL the set of lawful states that those things can be in. Then the sequence p =of states is a really possible process occurring in the things in question if it proceeds along a trajectory g: SL ® SL compatible with the laws of the things of the kind in question. If A = T = Time, then p = is a temporally ordered sequence of states. Upon reversing the sign of t we get p– = , which is the time-reversed image of p. It would be mistaken to interpret p– as representing a thing plunging into the past, or as involving “the backwards flow of time.” (See Bunge 1959b for a critique of the literal interpretation of the Feynman diagrams, which involve such fictitious travel backwards in time.) The exchange of -t for t, and consequently the substitution of negative velocities for positive ones, and conversely, is only a paper-and-pencil operation that need not represent a real process. When reference to a real process is intended, what is meant by ‘time reversal’ is not operating Wells’s time machine, but just inverting the velocities (and spins if any) of the entities concerned. (More on this and the following in Bunge 1968a.) It is well known that Newton’s equations of motion for point particles are T)
)
294
Appendix
invariant, that is, they do not alter under the substitution of –t for t. By contrast, Fourier’s equation for the flow of heat in a body is not T-invariant. The trajectories of ideal (perfectly elastic) billiard balls on an ideal (frictionless) pool table are T-invariant – as long as the balls do not fall into any of the pockets of the table. In other words, their time-reversed images are physically possible. By contrast, scrambled eggs do not unscramble, and alpha particles do not re-enter the nuclei that emitted them. There are, then, laws and processes that are T-invariant and others that aren’t. A process or history is said to be T-invariant, or reversible, if its timereversed image is really possible. Otherwise it is said to be irreversible. A lawstatement is T-invariant if it does not alter upon the substitution of –t for t. The relation between T-invariant processes and T-invariant law-statements is this: If a process is T-invariant, then its laws are T-invariant as well. The converse is false, whence the two kinds of T-invariance are not equivalent – contrary to popular belief. In other words, some irreversible processes, such as radioactive disintegration, can be described with the help of T-invariant law-statements. There are two reasons for the inequivalence of T-invariant processes with Tinvariant laws. The first is that a process, or history, is usually accounted for by some logical consequences of the basic law-statements concerned, and the former may lack the symmetry properties of the latter. For example, Maxwell’s equations are T-invariant but, whereas the electric field intensity does not change upon time reversal, the magnetic induction is inverted. A simpler example: a sinusoidal oscillation is not T-invariant even though it solves a Tinvariant equation of motion. The second reason for the inequivalence in question is that processes are described by logical consequences of basic law-statements together with constraints, boundary conditions, and constitutive equations, any of which is bound to further restrict the set of really possible trajectories. A simple example is that of a rain droplet falling onto the middle of a sliding roof. The reversed motion, resulting from the inversion of the velocity, would take the droplet past its point of impact, to the top of the roof and beyond. We exclude this conceptually possible trajectory by adding the condition of continuity of the velocity (not only the trajectory). A spatial analogue is this. Maxwell’s field equations have solutions representing incoming (convergent) electromagnetic waves as well as outgoing (divergent) waves. Since no convergent waves have been observed, one discards such solutions by imposing an additional condition on the asymptotic behaviour of the solutions (Sommerfeld’s Ausstrahlungsbedingung). A moral of this story is that, to judge whether a conceptually possible event or process is really possible or else impossible (miraculous), we must check
Fact and Pattern
295
whether it satisfies its laws together with all of the pertinent subsidiary conditions. Interestingly enough, the latter are often afterthoughts prompted by a negative result of the comparison of the theory with reality. An antirealist would never hit on such subsidiary conditions and would insist on making contrary to fact predictions. Being averse to distinguishing theory from reality, he is bound to condone miracles. Realism, then, is a necessary condition for doing genuine science and avoiding bogus science. 5 Objective Pattern and Law-Statement Realists distinguish objective patterns from the propositions (e.g., equations) representing them. This allows them to explain a typical endeavour of scientists: that of attempting to uncover regularities. The same distinction helps explain a good portion of the history of science as the history of successive approximations to the truest possible representation of objective patterns. Objective patterns, be they natural, social, or mixed, are supposed to be regularities or constancies with broad scopes, namely, entire species of concrete things or even genera of such. Every such pattern can be conceptualized in a number of ways – in fact, in as many as there are choices of state functions. Such conceptualizations are called law-statements. In short, we distinguish (objective) laws or patterns from their conceptual representations. And we assume that the objective patterns do not change when their conceptual representations alter. In particular, scientific revolutions may impact society, but they do not change the universe. In addition to objective laws (patterns) and law-statements, there are higherlevel principles concerning either patterns or law-statements. Instances of these are the requirement of Lorentz-covariance, and the philosophical principle that all facts satisfy some laws (Bunge 1959b). In sum, the word ‘law’ denotes three different concepts, which we identify by a subscript each: L1 = Objective pattern. L2 = Law-statement = Proposition representing an L1. L3 = Metanomological statement = Proposition about some L1 or L2. Let us focus on L2. A law-statement may be regarded as a restriction on certain state functions for things of a certain kind. Such a restriction is neither stray nor arbitrary: it must belong to a system (theory), and it must have been confirmed to a reasonable degree with the help of observations, measurements, or experiments. The first condition disqualifies empirical generalizations, and the second untested or false formulas. Law-statements may take a number of forms, depending not only on the things they refer to but also on the state of knowledge and even on the
296
Appendix
mathematical ability and the goal of the scientist who proposes them. All of the following are conspicuous simple forms of law-statements. Example 8. RF = V, where RF is the codomain of state function F, and V some well-defined set. Thus, nothing can travel faster than light, and the prices and quantities of goods cannot be negative. Example 9. ¶F/¶t ³ 0, where t is in T, and T is a subset of the real line occurring in the domain of F and representing time. Example 10. dF/dt = g(F,t), with g a specific function of t. t2
Example 11.
* F(q,dq/dt,t)dt = extremum, with t1 and t2 two designated t1
elements of T. Example 12. F2(x,y) = * du dv F1(u–x, v–y), with F1, F2: E3 × E3 ® ¶2F
R
C.
Example 13. = F2 , with F1, F2: ® . 1 The preceding considerations suggest adopting the following Definition 4. Let F be a state function for things of a given kind. A restriction on the possible values of the components of F , or a relation between two or more such components, is called a law-statement if (1) it is included in a consistent factual theory, and (2) it has been satisfactorily confirmed (for the time being). If only condition (2) is met, the restriction is called an empirical regularity. Consider now the collection LK of law-statements for things of kind K. Calling L an arbitrary member of that collection, L(x) may be regarded as the value that the predicate “law function” L takes at x Î K. This function has the form L: K ® LK. Since at each stage in the history of science we only know a finite subset of LK, this collection is variable, it is not a set proper. For example, in the elementary theory of electric networks, a batteryresistor circuit with a single loop and at room temperature (an individual of class K) satisfies a single law, namely Ohm’s. (This law-statement is not true for all conductors, and it is only a first approximation for conductors at very low temperatures. Further, the particular law for the ohmic resistance of conductors of a given kind is called a constitutive equation.) Ohm’s law can be written as follows: For every x in K: L(x) = “e(x) = R(x) ? i(x),” where e, R, and i represent the electromotive force, the resistance, and the current intensity respectively. Thus, in this case the law predicate L is the function L: K ® LK such that, for every x in K, L(x) equals Ohm’s formula. This manner of writing shows clearly that laws interrelate properties, and are themselves properties of things. That is, they are “possessed” by things instead of hovering above them. That is, laws are universalia in re (recall chapter 9). If /¶x2
E3
Fact and Pattern
297
a more explicit evidence for this claim is called for, one may write Ohm’s law in even more detail: e, i, R : K ® , and L: K ® LK, such that the above condition (Ohm’s) holds. This shows also what Ohm’s law is about, namely, physical things of a certain kind K. Which suffices to refute the views that physical (or other) laws are subjective or else social constructions. Occasionally one reads about natural objects, such as brains, whole organisms, or even molecules, behaving in accordance with algorithms, such as “evolutionary algorithms” and algorithms for “computing” limb movement, perception, or even emotion. This idiom betrays ignorance of both science and algorithm. Indeed, algorithms are, by definition, rules for computing something, such as the squares or the square roots of numbers. By contrast, the natural laws are thoroughly natural: they emerged along with the things they inhere in. This is why the expression ‘natural algorithm’ is an oxymoron. So much for laws or natural regularities. Let us now deal briefly with norms or artificial regularities, such as the moral and legal norms. We propose Definition 5. A norm or rule is a man-made restriction on the possible values of the components of a state function, or a man-made relation between two or more such components, compatible with the law-statement(s) satisfied by the state function. This definition incorporates the idea that, no matter how stupid, brutal, and powerful humans may be, they cannot violate the natural laws. In other words, every norm or rule involves in some way or other the pertinent natural laws. In particular, the efficient norms for assembling or operating a machine or a social organization must satisfy the laws possessed by the components of either system.
R
6 Lawful State Space The point of constructing mathematical models of things of a certain kind is to represent as accurately as possible the really possible (lawful) states of the things in question, and perhaps their really possible (lawful) changes of state as well. Hence, every such model is centred on a state space for the things referred to. A few typical examples should give us a feel for this matter. Example 14. In the elementary theory of the ideal gas, the state function is the triple consisting of the pressure, volume, and temperature functions. The domain and codomain of each of these functions are the set of (ideal) gas bodies and the positive real line respectively. Hence, the corresponding state space is a box contained in ( +)3. Example 15. In the genetics of populations, three state functions are often
R
298
Appendix
employed: the size N of a population, the frequency (or rather probability) p of occurrence of some particular gene, and the adaptive value v of the latter. Each of these is a real-valued function. Hence, for a system composed of two interacting populations, A and B, the state space is the region of ( +)6 spanned by the vectorin the course of time. Example 16. In my theory of social structure (Bunge 1974d), the instantaneous state of a community may be construed as the distribution of its population among the various social groups in the community. Hence, the component Fi of the total state function may be taken to be the column matrix i Nij i for a fixed i, the elements of which are the populations of the (mutually disjoint) social groups resulting from the partition of the total population (at the given time) by the ith equivalence relation with a social significance, such as equal occupation or similar educational level. Example 17. In chemical kinetics, the instantaneous state of a chemical system is described by the values of the partial concentrations of reactants and products. Therefore, the state space of the system is inside ( +)n, where n is the number of system components (reactants, catalysts, and reaction products). Example 18. In electrostatics, the state function is F = , where r represents the electric density and j the potential of the electric field. The domain and codomain of both functions are E3 and respectively. Hence, the local state of the given field is the value of F at x Î E3, and the entire state space is the set of ordered pairs { Î 2 ) x Î V}, where V denotes the region of E3 occupied by the field. Example 19. In quantum mechanics, the state of a thing is represented by a ray in the Hilbert space associated with the thing in question. Since a thing of this kind typically is not attributed a point-like location, but is assumed instead to be spread over some spatial region V in three-space, with a definite probability distribution, the state of the thing is the set of all the values its state vector y takes in V. Before attempting to draw general maxims let us emphasize a point of method made earlier. We may certainly assume that, whether we know it or not, every isolated thing is, at each instant, in a definite state relative to some reference frame. (That the state may be a superposition of eigenstates is true but beside the point.) Yet our representation of such a state will depend upon the state function chosen to represent the thing, and this choice depends in turn upon the state of our knowledge as well as on our goals. What holds for every single state holds a fortiori for the entire state space for a thing. That is, far from being something out there, like stars and people, a state space for things of a certain kind stands with one leg on the things it refers to, a second on a reference frame, and a third on the theorist or modeller. To
R
R
R
R
Fact and Pattern
299
persuade oneself that this is so, suffice it to take another look at Example 18 above, where a reference frame at rest relative to the field source was tacitly assumed. If now the same system of electrically charged bodies is considered relative to a moving reference frame, a four-vector current density will have to replace the single charge density, and a four-vector potential will take the place of the single scalar potential. (“Seen” from a moving frame, a charge is a current surrounded by a magnetic field.) Alternatively – and this is where the scientist’s freedom comes in – the four-vector potential can be replaced by an antisymmetric tensor representing the electric and magnetic components of the field relative to the new frame. (Incidentally, the problem of finding an adequate vector potential is of the inverse kind. For instance, given A, H equals the curl of A; but there is no rule for finding A from H.) Having emphasized the conventional ingredient of every representation of the states a thing can be in, let us now stress that every such representation has an objective basis as well – as long as the representation has a grain of truth. For one thing, a state function may not take values in its entire codomain, but may be restricted to a subset of the latter, and this by virtue of some law. Indeed, recall that law-statements are restrictions on state functions (section 5). Therefore, for every component of a state function for things of some kind, the focus of our concern will be the range of the function rather than its entire codomain. If we now take all the components of a state function for things of a given kind, and form the Cartesian product of their respective codomains (in tune with Definition 1), we obtain the conceivable state space for those things. This is precisely what we did in the examples that led off the present section. However, this restriction is insufficient to identify the really possible states of a thing, as should become apparent from the following examples. The total population of organisms of a given kind, in a given territory, is constrained not only by the carrying capacity of the latter but also by the birth and death rates, as well as by such additional factors as sunshine, rainfall, and plagues. Again, although the codomain of the speed function for a body is the entire interval [0,c), the speed of an electron travelling in a transparent medium will not come close to the upper bound c (the speed of light in a vacuum), for such a body is subject to further laws, such as those concerning the radiation emitted by the electron travelling through a transparent medium. In general: Only those values of the components of a state function that are compatible with the laws, constraints, and initial and boundary conditions will be really (not just conceptually) possible. In other words, because the law-statements impose restrictions upon the state functions and their values, hence upon the state spaces, only the states in certain subsets of the latter are
300
Appendix
accessible to the things in question. We shall call the accessible region of a state space the lawful state space for the things in question (in the given representation and relative to a given reference frame). To say that a thing of a certain kind behaves lawfully amounts then to saying that the point representing its (instantaneous) state cannot wander beyond the bounds of the lawful state space chosen for the things of the kind in question. The preceding remarks can be summarized into Definition 6. Let F =: A ® V1 × V2 × ... × Vn be a state function in a theoretical model for things of kind K, and call LK the (variable) collection law-statements of the Ks. Then the subset of the codomain V1 × V2 × ... × V of F restricted by the conditions (law-statements, constraints, and n initial and boundary conditions) in LK is called the lawful state space of such things, or SL for short. Example 20. In Example 18 above, the conceivable state space of an electrostatic field was S = { Î 2 ) x Î V}, where V is included in E3. Since the two components of the state function F = < r , j> are linked by the Poisson law-statement “Ñ2f = 4pr” and the boundary condition that the potential vanishes at infinity, the lawful space state for the thing (electrostatic field) is SL = {< r(x), f(x)> Î 2 ) x Î V & [Ñ 2f(x) = 4pr(x)] & f(¥) = 0}, a set that is properly included in S. What holds for the state spaces of natural and social things holds also for the state spaces of technical systems such as machines, factories, and armies. Indeed, all artefacts satisfy not only laws but also norms or rules, which amount to new restrictions on the state functions for the things concerned. (Recall Definition 5 in section 5.) However, artefacts (including formal social organizations) have (emergent) properties that their natural components lack – this being why we care to design, build, and use them. Hence, the work of technologists, managers, and planners does not boil down to shrinking state spaces for natural things. It involves setting up new state spaces not found in the natural sciences.
R
R
7 Concluding Remarks The category of facts is very broad: it includes a property having a given value, a thing being in a given state, a change of state, whether a point event (or instant transition from one state to another) or a protracted process (or sequence of states), and a fact fitting a given pattern. Things, by contrast, are not
Fact and Pattern
301
facts: they are the supports or bearers of facts, as in “this car started” and “that car passed by.” Whatever involves a thing is a fact, and nothing that fails to involve concrete things is a fact. For example, it is a fact that this is a printed page, but it is not a fact that twice two equals four. All facts are singular: there are no general facts. Some facts are composite, but there are neither negative nor disjunctive facts: negation and disjunction are de dicto, not de re. A pattern, regularity, or constancy can be of either of the following kinds. It can be objective like the law of gravity, or conceptual like the associative law. If objective, a pattern can be a law of nature or a norm, custom, or convention in force in some social group. A conceptual pattern can be either purely logical or mathematical, such as “p or not-p”; or it may represent an objective pattern – in which case we call it a ‘law-statement.’ It is a tacit ontological principle of scientific research that all concrete things behave in accordance with laws. This, the principle of lawfulness, cannot be proved, but it animates all modern scientific research, which to a large extent is the search for or application of laws and norms (see, e.g., Bunge 1967a). Lawfulness is often mistaken for uniformity. Actually, these are different though related concepts, and so are the corresponding principles. The lawfulness principle states that every state or change of state of an arbitrary thing satisfies some laws. By contrast, according to the principle of uniformity, the laws are the same across space and at all times. Put negatively: No laws would ever emerge or disappear anywhere. In a stronger version, this principle states that the same events recur everywhere and at all times: that there is never anything new “under the sun.” This strong uniformitarianism is false. It was falsified in natural science in the mid-nineteenth century, with the emergence of evolutionary geology and biology, but it is still going strong in social science. In particular, it is inherent in all of the rational-choice models, for they assume the constancy of both human nature and personal preferences. Every law-statement is about concrete things of some kind or other. It consists in a restriction on the possible values of the state functions of the things in question, or on the possible relations among components of such state functions. The same holds for norms or rules. Since the values of a state function represent possible states of the corresponding things, a law-statement tells us which are the possible states a thing can be in, and which are the changes (events or processes) it can undergo. Since every concrete thing is in some state or other, is engaged in some process or other, and is assumed to satisfy some set or other of laws (or else norms), the concepts of thing, property, state, event, process, fact, law, and norm are intimately connected with one another. Yet they are usually treated, if at all, in separation from each
302
Appendix
other. Worse, most of the definitions of these concepts proposed by philosophers are at variance with their meanings in science and technology, which is where they are used every day, if mostly in a tacit fashion, and in the most precise and consistent way. When in Rome, do as the Romans do. When in scientific territory, speak the language of science.
References 303
References
Agassi, Joseph. 1990. “Ontology and Its Discontent.” In P. Weingartner and G. Dorn, eds., Studies on Mario Bunge’s Treatise, 105–22. Amsterdam: Rodopi. Albert, Hans. 1994. Kritik der reinen Hermeneutik. Tübingen: J.C.B. Mohr (Paul Siebeck). Ampère, André-Marie. 1834. Essai sur la philosophie des sciences. Paris: Bachelier. Anderson, Arthur S., and François Nielsen. 2002. “Globalization and the Great U-turn: Income Inequality Trends in 16 OECD Countries.” American Journal of Sociology 107: 1244–99. Anderson, Roy M., and Robert M. May. 1991. Infectious Diseases of Humans: Dynamics and Control. Oxford: Oxford University Press. Armstrong, D[avid] M. 1978. Universals and Scientific Realism. Cambridge: Cambridge University Press. – 1997. A World of States of Affair. Cambridge: Cambridge University Press. Ashby, W. Ross. 1963. An Introduction to Cybernetics. New York: John Wiley. Athearn, Daniel. 1994. Scientific Nihilism: On the Loss and Recovery of Physical Explanation. Albany: State University of New York Press. Atran, Scott. 2003. “Genesis of Suicide Terrorism.” Science 299: 1534–9. Ayer, A[lfred] J[ules], ed. 1959. Logical Positivism. Glencoe, IL: Free Press. Bacon, Francis. 1905 [1620]. The New Organon. In Philosophical Works, ed. John M. Robertson. London: George Routledge & Sons. Bacon, G[eorge] E[dward]. 1966. X-Ray and Neutron Diffraction. Oxford: Pergamon Press. Barkow, Jerome H., Leda Cosmides, and John Tooby, eds. 1992. The Adaptive Mind: Evolutionary Psychology and the Generation of Culture. Cambridge, MA: MIT Press. Barnes, Barry. 1983. “On the Conventional Character of Knowledge and Cognition.” In Knorr-Cetina and Mulkay, eds., 1983: 19–51.
304
References
Barraclough, Geoffrey. 1979. Main Trends in History. London: Holmes & Meier. Barrow, John D., Paul C.W. Davies, and Charles L. Harper, eds. 2004. Science and Ultimate Reality: Quantum Theory, Cosmology. Cambridge: Cambridge University Press. Bateson, C.D. 1991. The Altruism Question. Hillsdale, NJ: Lawrence Erlbaum. Becker, Gary. 1976. The Economic Approach to Human Behavior. Chicago: University of Chicago Press. Beller, Mara. 1999. Quantum Dialogue: The Making of a Revolution. Chicago: University of Chicago Press. Berger, Peter L., and Thomas Luckman. 1966. The Social Construction of Reality. Garden City, NY: Doubleday. Berkeley, George. 1901 [1710]. Principles of Human Knowledge. In Works, vol. 1, ed. A. Campbell Fraser. Oxford: Clarendon Press. Berlyne, David, ed. 1974. Studies in the New Experimental Aesthetics. Washington: Hemisphere. Bernard, Claude. 1952 [1865]. Introduction à l’étude de la médecine experimentale. Paris: Flammarion. Bernardo, José M., and Adrian F.M. Smith. 1994. Bayesian Theory. Chichester: John Wiley & Sons. Bernays, Paul. 1976. Abhandlungen zur Philosophie der Mathematik. Darmstadt: Wissenschaftliche Buchgesellschaft. Berry, B.J.L., H. Kim, and H.-M. Kim, “Are Long-Waves Driven by TechnoEconomic Transformations?” Technological Forecasting and Social Change 14: 111–35. Berry, Donald A., and Dalene K. Stangl, eds. 1996. Bayesian Biostatistics. New York: Marcel Dekker. Biagini, Hugo E., ed. 1985. El movimiento positivista argentino. Buenos Aires: Editorial de Belgrano. Blakemore, Sarah-Jayne, Joel Winston, and Uta Frith. 2004. “Social Cognitive Neuroscience: Where Are We Heading?” Trends in Cognitive Science 8: 216–22. Blanke, Olaf, Stéphanie Ortigue, Theodor Landis, and Margitta Seeck. 2002. “Simulating Illusory Own-Body Perceptions.” Nature 419: 269. Blitz, David. 1992. Emergent Evolution: Qualitative Novelty and the Levels of Reality. Dordrecht, Boston: Kluwer. Block, Ned, Owen Flanagan, and Güven Güzeldere, eds. 2002. The Nature of Consciousness: Philosophical Debates. Cambridge, MA: MIT Press. Bloor, David. 1976. Knowledge and Social Imagery. London: Routledge & Kegan Paul. Bohr, Niels. 1958. Atomic Physics and Human Knowledge. New York: John Wiley & Sons.
References 305 Boltzmann, Ludwig. 1979 [1905]. Populäre Schriften. Ed. E. Broda. Braunschweig: Vieweg. Boole, George. 1952 [1854]. An Investigation of the Laws of Thought. New York: Dover. Born, Max. 1949. Natural Philosophy of Cause and Chance. Oxford: Clarendon Press. Boudon, Raymond. 2001. The Origin of Values. New Brunswick, NJ: Transaction. Boudon, Raymond, and Maurice Clavelin, eds. 1994. Le relativisme est-il irrésistible? Paris: Presses Universitaires de France. Boulding, Kenneth E. 1964. “General Systems as a Point of View.” In Mihajlo D. Mesarovic, ed., Views on General Systems Theory, 25–38. New York: John Wiley & Sons. Boutroux, Émile. 1898 [1874]. De la contingence des lois de la nature. Paris: Alcan. Braudel, Fernand. 1972 [1966]. The Mediterranean and the Mediterranean World in the Age of Philip II. 2 vols. New York: Harper Colophon Books. Bridgman, Percy W. 1927. The Logic of Modern Physics. New York: Macmillan. Brink, D.O. 1989. Moral Realism and the Foundations of Ethics. Cambridge, New York: Cambridge University Press. Brown, James Robert. 1994. Smoke and Mirrors: How Science Reflects Reality. London: Routledge. – 2001. Who Rules in Science? Cambridge, MA: Harvard University Press. Brown, Stephanie L., Randolph M. Nesse, Amiram D. Vinokur, and Dylan M. Smith. 2003. “Providing Social Support May Be More Beneficial than Receiving It: Results from a Prospective Study of Mortality.” Psychological Science 14: 320–7. Brush, Stephen G. 1968. “A History of Random Processes. I, Brownian Movement from Brown to Perrin.” Archive for History of Exact Sciences 5: 1–36. Bunge, Mario. 1951. “What Is Chance?” Science and Society 15: 209–31. – 1954. “New Dialogues betwen Hylas and Philonous.” Philosophy and Phenomenological Research 15: 192–9. – 1959a. Causality: The Place of the Causal Principle in Modern Science. Cambridge, MA: Harvard University Press. Repr. New York: Dover, 1979. – 1959b. Review of Popper 1959. Ciencia e investigación 15: 216–20. – 1959c. Metascientific Queries. Springfield, IL: Charles C. Thomas. – 1960. “The Place of Induction in Science.” Philosophy of Science 27: 262–70. Repr. in A.P. Iannone, ed., Through Time and Culture: Introductory Readings in Philosophy, 239–46. Englewood Cliffs, NJ: Prentice-Hall, 1994. – 1961. “The Weight of Simplicity in the Construction and Assaying of Scientific Theories.” Philosophy of Science 28: 129, 149. – 1962. Intuition and Science. Englewood Cliffs, NJ: Prentice-Hall. – 1963a. “A General Black-Box Theory.” Philosophy of Science 30: 346–58.
306
References
– 1963b. The Myth of Simplicity: Problems of Scientific Philosophy. Englewood Cliffs, NJ: Prentice-Hall. – 1964. “Phenomenological Theories.” In Mario Bunge, ed., The Critical Approach: Essays in Honor of Karl Popper, 234–54. New York: Free Press. – 1967a. Scientific Research, 2 vols. Berlin, Heidelberg, New York: Springer-Verlag. Rev. ed., Philosophy of Science, 2 vols. New Brunswick, NJ: Transaction, 1998. – 1967b. Foundations of Physics. Berlin, Heidelberg, New York: Springer-Verlag. – 1967c. “Analogy in Quantum Mechanics: From Insight to Nonsense.” British Journal for the Philosophy of Science 18: 265–86. – 1968a. “Physical Time: The Objective and Relational Theory.” Philosophy of Science 35: 355–88. – 1968b. “The Maturation of Science.” In I. Lakatos and A. Musgrave, eds., Problems in the Philosophy of Science, 120–37. Amsterdam: North-Holland. – 1968c. “Problems and Games in the Current Philosophy of Science. Proceedings of the XIVth International Congress of Philosophy 1: 566–74. Vienna: Herder, 1968. – 1969. “The Metaphysics, Epistemology and Methodology of Levels.” In L.L. Whyte, A.G. Wilson, and D. Wilson, eds., Hierarchical Levels, 17–28. New York: American Elsevier. – 1971. Review of Werner Heisenberg’s Der Teil und das Ganze. In Physics Today 24(1): 63–4. – 1973. Philosophy of Physics. Dordrecht, Boston: Reidel. – 1974a. Treatise on Basic Philosophy. Vol. 1: Sense and Reference. Dordrecht, Boston: Reidel. – 1974b. Treatise on Basic Philosophy. Vol. 2: Interpretation and Truth. Dordrecht, Boston: Reidel. – 1974c. “The Relations of Logic and Semantics to Ontology.” Journal of Philosophical Logic 3: 195–210. – 1974d. “The Concept of Social Structure.” In W. Leinfellner and W. Köhler, eds., Developments in the Methodology of the Social Sciences, 175–215. Dordrecht: Reidel. – 1975. “¿Hay proposiciones?” In Aspectos de la filosofía de W.V. Quine, 53–68. Valencia: Teorema. – 1976. “A Model for Processes Combining Competition with Cooperation.” Applied Mathematical Modelling 1: 21–3. – 1977a. Treatise on Basic Philosophy. Vol. 3: The Furniture of the World. Dordrecht, Boston: Reidel [Kluwer]. – 1977b. “States and Events.” In W.E. Hartnett, ed., Systems: Approaches, Theories, Applications, 71–95. Dordrecht: Reidel. – 1977c. “Levels and Reduction.” American Journal of Physiology 233(3): R75-82. – 1979a. Treatise on Basic Philosophy. Vol. 4: A World of Systems. Dordrecht, Boston: Reidel [Kluwer].
References 307 – 1979b. “A Systems Concept of Society: Beyond Individualism and Holism.” Theory and Decision 10: 13–30. – 1979c. “The Einstein-Bohr Debate over Quantum Mechanics: Who Was Right about What?” Lecture Notes in Physics 100: 204–19. – 1981a. “Four Concepts of Probability.” Applied Mathematical Modelling 5: 306–12. – 1981b. Scientific Materialism. Dordrecht, Boston, Lancaster: Reidel, 1981. – 1981c. “Biopopulations, Not Species, Are Individuals and Evolve.” Behavioral and Brain Sciences 4: 284–5. – 1982. “The Revival of Causality.” In G. Floistad, ed., Contemporary Philosophy, 2: 133–55. The Hague: Martinus Nijhoff. – 1983a. Treatise on Basic Philosophy. Vol. 5: Exploring the World. Dordrecht, Boston: Reidel. – 1983b. Treatise on Basic Philosophy. Vol. 6: Understanding the World. Dordrecht, Boston: Reidel. – 1985a. Treatise on Basic Philosophy. Vol. 7, part 1: Formal and Physical Sciences. Dordrecht, Boston: Reidel. – 1985b. Treatise on Basic Philosophy. Vol. 7, Philosophy of Science and Technology, part 2: Life Science, Social Science and Technology. Dordrecht, Boston: Reidel [Kluwer]. – 1988. “Two Faces and Three Masks of Probability.” In E. Agazzi, ed., Probability in the Sciences 27–50. Dordrecht, Boston: Reidel. – 1989. Treatise on Basic Philosophy. Vol. 8: Ethics: The Good and the Right. Dordrecht, Boston: Reidel. – 1991. “A Critical Examination of the New Sociology of Science. Part I.” Philosophy of the Social Sciences 21: 524–60. – 1992. “A Critical Examination of the New Sociology of Science. Part II.” Philosophy of the Social Sciences 22: 46–76. – 1996. Finding Philosophy in Social Science. New Haven, CT: Yale University Press. – 1997. “Mechanism and Explanation.” Philosophy of the Social Sciences 27: 410–65. Rev. version repr. in Bunge 1999. – 1998. Social Science under Debate. Toronto: University of Toronto Press. – 1999. The Sociology-Philosophy Connection. New Brunswick, NJ: Transaction Publishers. – 2000a. “Energy: Betweeen Physics and Metaphysics.” Science and Education 9: 457–61. – 2000b. “Physicians Ignore Philosophy at Their Risk – and Ours.” Facta philosophica 2: 149–60. – 2000c. “Philosophy from the Outside.” Philosophy of the Social Sciences 30: 226–45.
308
References
– 2001. Philosophy in Crisis: The Need for Reconstruction. Amherst, NY: Prometheus Books. – 2003a. Emergence and Convergence: Qualitative Novelty and the Unity of Knowledge. Toronto: University of Toronto Press. – 2003b. Philosophical Dictionary. Amherst, NY: Prometheus Books. – 2003c. “Velocity Operators and Time–Energy Relations in Relativistic Quantum Mechanics.” International Journal of Theoretical Physics 42: 135–42. – 2003d. “Twenty-five Centuries of Quantum Physics: From Pythagoras to Us, and from Subjectivism to Realism.” Science & Education 12: 445–66. – 2003e. “Quantons Are Quaint but Basic and Real, and the Quantum Theory Explains Much but Not Everything: Reply to My Commentators.” Science & Education 12: 587–97. – 2004a. “How Does It Work? The Search for Explanatory Mechanisms.” Philosophy of the Social Sciences 34: 182–210. – 2004b. “Clarifying Some Misunderstanding about Social Systems and Their Mechanisms.” Philosophy of the Social Sciences 34: 371–81. – 2005a. “Did Weber Practise the Philosophy He Preached?” In L. McFalls, ed., The Objectivist Ethic and the Spirit of Science. Toronto: University of Toronto Press. – 2005b. “A Systemic Perspective on Crime.” In P.-O. Wickström and R.J. Sampson, eds., Contexts and Mechanisms of Pathways in Crime. Cambridge: Cambridge University Press. Bunge, Mario, and Rubén Ardila. 1987. Philosophy of Psychology. New York: Springer-Verlag. Bunge, Mario, and Máximo García-Sucre. 1976. “Differentiation, Participation, and Cohesion.” Quality and Quantity 10: 171–8. Bunge, Mario, and Martin Mahner. 2004. Ueber die Natur der Dinge. Stuttgart: S. Hirzel. Burawoy, Michael. 2003. “Revisits: An Outline of a Theory of Reflexive Ethnography.” American Sociological Review 68: 645–79. Cabanac, Michel. 1999. “Emotion and Phylogeny.” Japanese Journal of Physiology 49: 1–10. Carnap, Rudolf. 1936–7. “Testability and Meaning.” Philosophy of Science 3: 419–71; 4: 1–40. – 1950a. Logical Foundations of Probability. Chicago: University of Chicago Press. – 1950b. “Empiricism, Semantics, and Ontology.” Revue internationale de philosophie 10: 20–40. – 1967 [1928]. The Logical Structure of the World. Berkeley and Los Angeles: University of California Press. Cartwright, Nancy. 1983. How the Laws of Physics Lie. Oxford: Oxford University Press.
References 309 Centre International de Synthèse. 1957. Notion de structure et structure de la connaissance. Paris: Albin Michel. Chisholm, Roderick M, ed. 1960. Realism and the Background of Phenomenology. Glencoe, IL: Free Press. Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Churchland, Paul M. 1989. A Neurocomputational Perspective. Cambridge, MA: MIT Press. Clark, Austen. 1993. Sensory Qualities. Oxford: Clarendon Press. Clough, Sharyn. 2003. Beyond Epistemology: A Pragmatist Approach to Feminist Science Studies. Lanham, MD: Rowman & Littlefield. Clutton-Brock, Tim. 2002. “Breeding Together: Kin Selection and Mutualism in Cooperative Vertebrates.” Science 296: 69–72. Cockburn, Andrew. 1998. “Evolution of Helping Behavior in Cooperatively Breeding Birds.” Annual Review of Ecology and Systematics 29: 141–77. Coleman, James S. 1964. Introduction to Mathematical Sociology. Glencoe, IL: Free Press. Collins, Harry M. 1981. “Stages in the Empirical Programme of Relativism.” Social Studies of Science 11: 3–10. Collins, John, Ned Hall, and L.A. Paul, eds. 2004. Causation and Counterfactuals. Cambridge, MA: MIT Press. Collins, Randall. 1998. The Sociology of Philosophies: A Global Theory of Intellectual Change. Cambridge, MA: Belknap Press of Harvard University Press. Condorcet [Marie-Jean-Antoine-Nicolas, Caritat, Marquis de]. 1976. Condorcet: Selected Writings. Ed. K.M. Baker. Indianapolis: Bobbs-Merrill. Conel, J.L. 1939–67. The Post-Natal Development of the Human Cerebral Cortex. Cambridge, MA: Harvard University Press. Cournot, Antoine-Augustin. 1843. Expositions de la théorie des chances et des probabilités. Paris: Hachette. Covarrubias, G.M. 1993. “An Axiomatization of General Relativity.” International Journal of Theoretical Physics 32: 2135–54. Damasio, Antonio. 1994. Descartes’ Error. New York: Putnam. Davidson, Donald. 1970. “Mental Events.” In Essays on Action and Events, 2nd ed.: 207–25. Oxford: Clarendon Press. – 1984 [1974]. Inquiries into Truth and Interpretation. Oxford: Clarendon Press. Dawkins, Richard. 1976. The Selfish Gene. New York: Oxford University Press. Debreu, Gerard. 1991. “The Mathematization of Economic Theory.” American Economic Review 81: 1–7. De Finetti, Bruno. 1962. “Does It Make Sense to Speak of ‘Good Probability Appraisers’?” In I.J. Good, ed., The Scientist Speculates: An Anthology of Partly-Baked Ideas. London: Heinemann.
310
References
– 1972. Probability, Induction, and Statistics. New York: John Wiley. Descartes, René. 1664 [1634]. Le Monde ou Traité de la lumière. In Oeuvres, vol. 10, ed. C. Adam and P. Tannery. Repr. Paris: Vrin, 1974. d’Espagnat, Bernard. 1981. A la recherche du réel. 2nd ed. Paris: Gauthier-Villars. Deutsch, David. 2004. “It from qubit.” In Barrow, Davies, and Harper, eds., 90–102. Dietzgen, Joseph. 1906 [1887]. “Excursions of a Socialist into the Domain of Epistemology.” In Philosophical Essays. Chicago: Charles H. Kerr & Co. Dijksterhuis, E[duard]. J[an]. 1986 [1959]. The Mechanization of the World Picture from Pythagoras to Newton. Princeton, NJ: Princeton University Press. Dragonetti, Carmen, and Fernando Tola. 2004. On the Myth of the Opposition between Indian Thought and Western Philosophy. Hildesheim: Georg Olms. Duhem, Pierre. 1908. SWZEIN TA FAINOMENA: Essai sur la notion de théorie physique de Platon à Galilée. Offprint from the Revue de philosophie chrétienne. Paris: Hermann. Dummett, Michael. 1977. Elements of Intuitionism. Oxford: Clarendon Press, 1977. Du Pasquier, L.-Gustave. 1926. Le calcul des probabilités: Son évolution mathématique et philosophique. Paris: Librairie scientifique J. Hermann. Durkheim, Émile. 1988 [1901]. Les règles de la méthode sociologique. 2nd ed. Intro. by J.-M. Berthelot. Paris: Flammarion. Earman, John. 1992. Bayes or Bust? Cambridge, MA: MIT Press. Eddy, Charles. 1982. “Probabilistic Reasoning in Clinical Medicine: Problems and Opportunities.” In Kahneman, Slovic and Tversky, eds.: 249–67. Eichner, Alfred S., ed. 1983.Why Economics Is Not Yet a Science. Armonk, NY: M.E. Sharpe. Einstein, Albert. 1949. “Autobiographical Notes.” In Paul Arthur Schilpp, ed., Albert Einstein: Philosopher-Scientist. Evanston, IL: Library of Living Philosophers. – 1950a [1936]. “Physics and Reality.” In Out of My Later Years. New York: Philosophical Library. – 1950b. “The Laws of Science and the Laws of Ethics.” Preface to P. Frank, Relativity: A Richer Truth. Boston: Beacon Press. Elias, Norbert. 2000 [1939]. The Civilizing Process: Sociogenetic and Psychogenetic Investigations. Oxford: Blackwell. Elster, Jon. 1998. “A Plea for Mechanisms.” In Hedström and Swedberg, eds.: 45–73. Engels, Frederick. 1954 [1894]. Anti-Dühring. Moscow: Foreign Languages Publishing House. Euler, Leonhard. 1846 [1768]. Letters of Euler on different Subjects in Natural Philosophy addressed to a German Princess. Ed. David Brewster. 2 vols. New York, London: Harper & Brothers. Everett, Hugh, III. 1957. “‘Relative State’ Formulation of Quantum Mechanics.” Reviews of Modern Physics 29: 454–62.
References 311 Fehr, Ernst, and Bettina Rockenbach. 2003. “Detrimental Effects of Sanctions on Human Altruism.” Nature 422: 137–40. Feigenbaum, E.A., and J. Lederberg. 1968. “Mechanization of Inductive Inference in Organic Chemistry.” In B. Kleinmutt, ed., Formal Representation of Human Judgment. New York: Wiley. Feigl, Herbert. 1943. “Logical Empiricism.” In Dagobert D. Runes, ed., Twentieth Century Philosophy, 371–416. New York: Philosophical Library. Feller, William. 1968. An Introduction to Probability Theory and Its Applications. 3rd ed., vol. 1. New York: John Wiley. Fensham, Peter J., ed. 2004. Defining Identity: The Evolution of Science Education as a Field of Research. Dordrecht: Kluwer. Field, Hartry H. 1980. Science Without Numbers. Princeton, NJ: Princeton University Press. Fine, Arthur. 1986. The Shaky Game: Einstein, Realism, and the Quantum Theory. Chicago: University of Chicago Press. Fisher, Ronald A. 1951. The Design of Experiments. 6th ed. Edinburgh and London: Oliver and Boyd. Fiske, Donald W., and Richard A. Shweder, eds. 1986. Metatheory in Social Science: Pluralisms and Subjectivities. Chicago: University of Chicago Press. Flach, Peter A., and Antonis C. Kakas, eds. 2000. Abduction and Induction: Essays on Their Relation and Integration. Dordrecht, Boston: Kluwer Academic. Fleck, Ludwik. 1979 [1935]. Genesis and Development of a Scientific Fact. Foreword by T.S. Kuhn. Chicago: University of Chicago Press. Fodor, Jerry A. 1981. “The Mind-Body Problem.” Scientific American 244(1): 114–23. Fogel, Robert W. 1994. Railroads and American Economic Growth: Essays in Econometric History. Baltimore, MD: Johns Hopkins University Press. Frank, Philipp. 1957. Philosophy of Science. Englewood Cliffs, NJ: Prentice-Hall. Frankfort, H., H.A. Frankfort, J.A. Wilson, and T. Jacobsen. 1949 [1946]. Before Philosophy: The Intellectual Adventure of Ancient Man. London: Penguin. Fréchet, Maurice. 1946. “Les définitions courantes de la probabilité.” In Les mathématiques et le concret, 157–204. Paris: Presses Universitaires de France. Friedman, Milton. 1953. “The Methodology of Positive Economics.” In Essays in Positive Economics, 3–43. Chicago: University of Chicago Press. Frisch, Uriel, Sabino Matarrese, Roya Mohayaee, and Andrei Sobolevski. 2002. “A Reconstruction of the Initial Conditions of the Universe by Optimal Mass Transportation.” Nature 417: 260–2. Gale, David. 1960. The Theory of Linear Economic Models. New York: McGraw-Hill. Galilei, Galileo. 1953 [1623]. “Il saggiatore.” In Opere, ed. Ferdinando Flora. MilanNaples: Riccardo Ricciardi.
312
References
Gasparski, Wojciech W. 1993. A Philosophy of Practicality. Acta Philosophica Fennica, vol. 53. Gazzaniga, Michael S., ed. 2004. The New Cognitive Neurosciences. 3rd ed. Cambridge, MA: MIT Press. Gergen, Kenneth J. 2001. “Psychological Science in a Postmodern Context.” American Psychologist 56: 803–13. Ghiselin, Michael T. 1974. “A Radical Solution to the Species Problem.” Systematic Zoology 23: 536–44. Giddens, Anthony. 1984. The Constitution of Society: Outline of the Theory of Structuration. Cambridge: Polity Press. Gigerenzer, Gerd, Zeno Swijtini, Theodore Porter, Lorraine Dast, John Beatty, and Lorenz Krüger. 1989. The Empire of Chance: How Probability Changed Science and Everyday Life. Cambridge: Cambridge University Press. Gillies, Donald. 2000. Philosophical Theories of Probability. London: Routledge. Gilson, Étienne. 1947. La philosophie au Moyen Age. Paris: Payot. Giza, Piotr. 2002. “Automated Discovery Systems and Scientific Realism.” Minds and Machines 12: 105–17. Glennan, Stuart. 2002. “Rethinking Mechanistic Explanation.” Philosophy of Science, supplement to vol. 69 (3): S342–S353. Good, I[rving]. J[ohn]. 1967. “Corroboration, Explanation, Evolving Probability, Simplicity, and a Sharpened Razor.” British Journal for the Philosophy of Science 19: 123–43. Goodman, Nelson. 1951. The Structure of Appearance. Cambridge, MA: Harvard University Press. – 1954. Fact, Fiction, and Forecast. London: Athlone Press. – 1978. Ways of Worldmaking. Indianapolis: Hackett. – 1996. “On Starmaking.’’ In McCormick: 143–7. Gopnik, Alison, Andrew N. Meltzoff, and Patricia K. Kuhl. 1999. The Scientist in the Crib: Minds, Brains, and How Children Learn. New York: William Morrow. Gould, Stephen Jay. 2002. The Structure of Evolutionary Theory. Cambridge, MA: Belknap Press of Harvard University Press. Grassmann, Hermann. 1844. Ausdehnungslehre. In Gesammelte mathematische und physikalische Werke, vol. 1, part I. Repr. Bronx, NY: Chelsea, 1969. Greene, Brian. 2004. The Fabric of the Cosmos. New York: Alfred A. Knopf. Gregory, R[ichard].L., and E[rnest].H. Gombrich, eds. 1973. Illusion in Nature and Art. London: Duckworth. Griffin, Donald R. 2001. Animal Minds: Beyond Recognition and Consciousness. Chicago: University of Chicago Press. Gross, Paul R., and Norman Levitt. 1994. Higher Superstition: The Academic
References 313 Left and Its Quarrel with Science. Baltimore, MD: Johns Hopkins University Press. Haack, Susan. 1974. Deviant Logics. Cambridge: Cambridge University Press. – 2003. Defending Science – Within Reason: Between Scientism and Cynicism. Amherst, NY: Prometheus Books. Habermas, Jürgen. 1970. Toward a Rational Society. Boston: Beacon Press. Hacking, Ian. 1990. The Taming of Chance. Cambridge: Cambridge University Press. Hadamard, Jacques. 1952. Lectures on Cauchy’s Problem in Linear Partial Differential Equations. New York: Dover Publications. Hall, Richard L. 1999. “Constructive Inversion of Energy Trajectories in Quantum Mechanics.” Journal of Mathematical Physics 40: 699–707. Hammerstein, Peter, ed. 2003. Genetic and Cultural Evolution of Cooperation. Cambridge, MA: MIT Press. Harman, Gilbert. 1965. “The Inference to the Best Explanation.” Philosophical Review 74: 88–95. – 1973. Thought. Princeton, NJ: Princeton University Press. – 1977. The Nature of Morality. New York: Oxford University Press. Harris, Marvin. 1979. Cultural Materialism. New York: Random House. Hayek, F[riedrich].A. 1955. The Counter-Revolution of Science: Studies on the Abuse of Reason. Glencoe, IL: Free Press. Hebb, Donald O. 1980. Essay on Mind. Hillsdale, NJ: Lawrence Erlbaum. Hedström, Peter, and Richard Swedberg, eds. 1998. Social Mechanisms: An Analytical Approach to Social Theory. Cambridge: Cambridge University Press. Hegel, Georg Wilhelm Friedrich. 1969 [1830]. Enzyklopädie der philosophischen Wissenschaften im Grundrisse. Hamburg: Felix Meiner. Heidegger, Martin. 1987 [1953]. Einführung in die Metaphysik. Tübingen: Max Niemeyer. Heilbron, J.L. 1986. The Dilemmas of an Upright Man. Berkeley: University of California Press. Heisenberg, Werner. 1947. Wandlungen in den Grundlagen der Naturwissenschaften. 7th ed. Zurich: Hirzel. – 1958. Physics and Philosophy. New York: Harper & Brothers. Helmholtz, H[ermann von]. 1873. Popular Lectures on Scientific Subjects. London: Longmans, Green and Co. Hesse, Mary. 1966. Models and Analogies in Science. Notre Dame, IN: University of Notre Dame Press. Hilbert, David. 1901. “Mathematische Probleme.” In Gesammelte Abhandlungen, 3: 290–329. Berlin: Julius Springer, 1935.
314
References
Hiriyanna, Mysore. 1951. The Essentials of Indian Philosophy. London: George Allen & Unwin. Hobson, John A. 1938 [1902]. Imperialism: A Study. London: Allen & Unwin. Hogarth, Robin M., and Melvin W. Reder. 1987. Rational Choice: The Contrast between Economics and Psychology. Chicago: University of Chicago Press. Holbach, Paul-Henry Thiry, Baron d’. 1773. Système social. 3 vols. Repr. Hildesheim, New York: Georg Olms, 1969. Horwich, Paul. 1990. Truth. Oxford: Basil Blackwell. Howson, Colin, and Peter Urbach. 1989. Scientific Reasoning: The Bayesian Approach. La Salle, IL: Open Court. Hubel, David H., and Thorsten N. Wiesel. 1962. “Receptive Fields, Binocular Vision, and Functional Architecture in the Cat’s Visual Cortex.” Journal of Physiology 160: 106–54. Hughes, G.E., and M.J. Cresswell. 1968. An Introduction to Modal Logic. London: Methuen. Hull, David L. 1976. “Are Species Really Individuals?” Systematic Zoology 25: 174–91. Hume, David. 1888 [1734]. A Treatise of Human Nature. Ed. L.A. Selby-Bigge. Oxford: Clarendon Press. – 1902 [1748]. Enquiry concerning Human Understanding. 2nd ed. Oxford: Clarendon Press. Humphreys, Paul. 1985. “Why Propensities Cannot Be Probabilities.” Philosophical Review 94: 557–70. Hunt, Shelby D. 2003. Controvesy in Marketing Theory: For Reason, Realism, Truth, and Objectivity. Armonk, NY: M.E. Sharpe. Husserl, Edmund. 1929. “Phenomenology.” In Encyclopaedia Britannica, 14th ed. Repr. in Chisholm, ed. 1960, 118–28. – 1931 [1913]. Ideas: General Introduction to Pure Phenomenology. London: Allen & Unwin. – 1960 [1931]. Cartesian Meditations: An Introduction to Phenomenology. The Hague: Martinus Nijhoff. – 1970 [1936]. The Crisis of European Sciences and Transcendental Phenomenology: An Introduction to Phenomenological Philosophy. Evanston, IL: Northwestern University Press. Ingenieros, José. 1917. Hacia una moral sin dogmas. Buenos Aires: L.J. Rosso y Cía. Jacob, François. 1976. The Logic of Life. New York: Vintage Books. James, Patrick. 2002. International Relations and Scientific Progress: Structural Realism Reconsidered. Columbus: Ohio State University Press. Jammer, Max. 1966. The Conceptual Development of Quantum Mechanics. New York: McGraw-Hill.
References 315 Jeffreys, Harold. 1975. Scientific Inference. 3d ed. Cambridge: Cambridge University Press. Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler. 1986. “Fairness and the Assumptions of Economics.” Journal of Business 59: S285–S300. Kahneman, Daniel, Paul Slovic, and Amos Tversky. 1982. Judgment under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press. Kaila, Eino. 1979. Reality and Experience: Four Philosophical Essays. Dordrecht, Boston: Reidel. Kandel, Eric R., James H Schwartz, and Thomas M. Jessell, eds. 2000. Principles of Neural Science. New York: McGraw Hill. Kant, Immanuel. 1787. Kritik der reinen Vernunft. 2nd ed. [B]. Hamburg: Felix Meiner, 1952. – 1913. Briefwechsel. 3 vols. Ed. H.E. Fischer. Munich: George Müller. – 2002 [1783]. Prolegomena to any Future Metaphysics. In Theoretical Philosophy after 1781, The Cambridge Edition of the Works of Immanuel Kant. Cambridge: Cambridge University Press. Kaplan, Edward H., and Arnold Barnett. 2003. “A New Approach to Estimating the Probability of Winning the Presidency.” Operations Research 51: 32–40. Kary, Michael, and Martin Mahner. 2002. “How Would You Know If You Synthesized a Thinking Thing?” Minds and Machines 12: 61–86. Kasarda, John, and Morris Janowitz. 1974. “Community Attachment in Mass Society.” American Sociological Review 39: 328–39. Keuth, Herbert. 1978. Realität und Wahrheit. Tübingen: J.C.B. Mohr (Paul Siebeck). Keynes, John Maynard. 1957 [1921]. A Treatise on Probability. London: Macmillan. – 1973 [1936]. The General Theory of Employment, Interest and Money. In Collected Writings, vol. 7. London: Macmillan and Cambridge University Press. Kim, Jaegwon. 1998 [1976]. “Events as Property Exemplifications.” In Laurence and Macdonald, eds.: 310–26. Kirchhoff, Gustav Robert. 1883. Vorlesungen über mathematische Physik: Mechanik. Leipzig: Teubner. Kitcher, Philip, and Wesley Salmon, eds. 1989. Scientific Explanation. Minneapolis: University of Minnesota Press. Knorr-Cetina, Karen D. 1981. The Manufacture of Knowledge: An Essay on the Constructivist and Contextual Nature of Science. Oxford: Pergamon Press. – 1983. “The Ethnographic Study of Scientific Work: Towards a Constructivist Interpretation of Science.” In Knorr-Cetina and Mulkay eds.: 115–39. Knorr-Cetina, Karen D., and Michael Mulkay, eds. 1983. Science Observed: Perspectives on the Social Study of Science. London: Sage. Kolmogoroff, A.N. 1956 [1933]. Foundations of the Theory of Probability. New York: Chelsea.
316
References
Kotarbinski, Tadeusz. 1965. Praxiology: An Introduction to the Sciences of Efficient Action. Oxford: Pergamon Press. Koza, John R., Martin A. Keane, and Matthew J. Streeter. 2003. “Evolving Inventions.” Scientific American 288 (2): 52–9. Kraft, Julius. 1957. Von Husserl zu Heidegger: Kritik der phänomenologischen Philosophie. 2nd ed. Frankfurt am Main: Oeffentliches Leben. Kraft, Victor. 1953. The Vienna Circle: The Origin of Neo-Positivism. New York: Philosophical Library. Krechevsky, I. 1932. “‘Hypotheses’ versus ‘Chance’ in the Pre-Solution Period in Sensory Discrimination Learning.” University of California Publications in Psychology, vol. 6 no. 3. Kripke, Saul A. 1980. Naming and Necessity. Cambridge, MA: Harvard University Press. Kulp, Christopher B., ed. 1997. Realism/Antirealism and Epistemology. Lanham, MD: Rowman & Littlefield. Kurtz, Paul, ed. 2001. Skeptical Odysseys. Amherst, NY: Prometheus Books. Lakatos, Imre. 1978. Mathematics, Science and Epistemology. Cambridge: Cambridge University Press. Lalande, André, ed. 1938.Vocabulaire technique et critique de la philosophie. 3 vols. Paris: Librairie Félix Alcan Lambek, Joachim. 1994. “Are the Traditional Philosophies of Mathematics Really Incompatible?” Mathematical Intelligencer 16: 56–62 Lambert, Johann Heinrich. 1764. Neues Organon. 2 vols. Repr. in Philosophische Schriften, vols. 1 and 2. Hildesheim: Georg Olms, 1965. Lanczos, Cornelius. 1949. The Variational Principles of Mechanics. Toronto: University of Toronto Press. Laszlo, Ervin. 1972. Introduction to Systems Philosophy. New York: Gordon and Breach. Latour, Bruno. 2004. “Why Has Critique Run Out of Steam?” Online at www.ensmp .fr/latour/articles/article/089.html. Latour, Bruno, and Steven Woolgar. 1986. Laboratory Life: The Construction of Scientific Facts. Princeton, NJ: Princeton University Press. Laudan, Larry. 1981. “A Confutation of Convergent Realism.” Philosophy of Science 48: 19–49. Laurence, Stephen, and Cynthia Macdonald, eds. 1998. Contemporary Readings in the Foundations of Metaphysics. Oxford: Blackwell. Lazarsfeld, Paul F., and Anthony R. Oberschall. 1965. “Max Weber and Empirical Social Research.” American Sociological Review 30: 185–99. LeDoux, Joseph. 2003. Synaptic Self. New York: Penguin. Leibniz, Gottfried Wilhelm. 1703. Nouveaux essays. Paris: Flammarion. – 1956. Philosophical Papers and Letters. Ed. L.H. Loemker. 2 vols. Chicago: University of Chicago Press.
References 317 Lenin, V[ladimir]. I[lich]. 1947 [1908]. Materialism and Empirio-Criticism. Moscow: Foreign Languages Publishing House. Levins, Richard, and Richard Lewontin. 1985. The Dialectical Biologist. Cambridge, MA: Harvard University Press. Lewis, David. 1973. “Causation.” Journal of Philosophy 70: 556–67. – 1983. “New Work for a Theory of Universals.” Reprinted in Laurence and Macdonald, eds.: 163–97. – 1986. On the Plurality of Worlds. Oxford: Blackwell. – 2001 [1973]. Counterfactuals. Rev. ed. Oxford: Blackwell. – 2002 [1990]. “What Experience Teaches.” In Block, Flanagan, and Güzeldere, eds.: 579–616. Lindley, D.V. 1976. “Bayesian Statistics.” In W.L. Harper and C.A. Hooker, eds., Foundations of Probability Theory, Statistical Inference, and Statistical Theories of Science. Vol. 2, Foundations and Philosophy of Statistical Inference, 353–62. Dordrecht, Boston: Reidel. Lipset, Seymour Martin. 1959. “Political Sociology.” In Robert K. Merton, Leonard Broom, and Leonard S. Cottrell Jr, eds. Sociology Today: Problems and Prospects, 81–114. New York: Basic Books. Locke, John. [1690.] An Essay Concerning Human Understanding. London: George Routledge & Sons. Loptson, Peter. 2001. Reality: Fundamental Topics in Metaphysics. Toronto: University of Toronto Press. Loux, Michael J., and Dean W. Zimmerman, eds. 2003. The Oxford Handbook of Metaphysics. Oxford: Oxford University Press. Lovinbond, Sabina. 1983. Realism and Imagination in Ethics. Minneapolis: University of Minnesota Press. Lowe, E. J[onathan]. 2002. A Survey of Metaphysics. Oxford: Oxford University Press. Luhmann, Niklas. 1984. Soziale Systeme: Grundrisse einer allgemeinen Theorie. Frankfurt: Suhrkamp. Luria, A[lexander]. R[omanovich]. 1979. The Making of Mind: A Personal Account of Soviet Psychology. Cambridge, MA: Harvard University Press. Lycan, William G. 1998. “Possible Worlds and Possibilia.” In Laurence and Macdonald, eds.: 83–95. Lyotard, Jean-François. 1980. The Postmodern Condition. Minneapolis: Minnesota University Press. Mach, Ernst. 1910. Populär-wissenschaftliche Vorlesungen. Leipzig: Johann Ambrosius Barth. – 1914 [1900]. The Analysis of Sensations and the Relation of the Physical to the Psychical. Chicago: Open Court.
318
References
– 1942 [1893]. The Science of Mechanics. La Salle, IL: Open Court. Machamer, Peter, Lindley Darden, and Carl F. Craver. 2000. “Thinking about Mechanisms.” Philosophy of Science 67: 1–25. Machamer, Peter, Rick Grush, and Peter McLaughlin, eds. 2001. Theory and Method in the Neurosciences. Pittsburgh: University of Pittsburgh Press. Mac Lane, Saunders. 1986. Mathematics: Form and Function. New York: SpringerVerlag. Mahner, Martin, ed. 2001. Scientific Realism: Selected Essays of Mario Bunge. Amherst, NY: Prometheus Books. Mahner, Martin, and Mario Bunge. 1997. Foundations of Biophilosophy. New York: Springer-Verlag. – 2001. “Function and Functionalism: A Synthetic Perspective.” Philosophy of Science 68: 75–94. March, James G., and Herbert A. Simon. 1958. Organizations. New York: John Wiley. Marsicano, Giovanni, et al. 2002. “The Endogenous Cannabioid System Controls Extinction of Aversive Memories.” Nature 418: 530–4. Massey, Douglas S. 2002. “A Brief History of Human Society: The Origin and Role of Emotion in Social Life.” American Sociological Review 67: 1–29. Matthews, Michael, ed. 1998. Constructivism in Science Education: A Philosophical Examination. Dordrecht: Kluwer Academic Publishers. Maurin, J. 1975. “Simulation déterministe du hasard.” Paris: Masson. McCall, Storrs. 1994. A Model of the Universe: Space-Time, Probability, and Decision. Oxford: Clarendon Press. McCloskey, Donald M. 1985. The Rhetoric of Economics. Madison: University of Wisconsin Press. McCormick, Peter J., ed. 1996. Starmaking: Realism, Anti-Realism, and Irrealism. Cambridge, MA: MIT Press. McGinn, Colin. 1985. Ethics, Evil and Fiction. Oxford: Oxford University Press. Meinong, Alexius. 1960 [1904]. “The Theory of Objects.” In Chisholm, ed., 76–117. Melia, Joseph. 2003. Modality. Montreal and Kingston: McGill-Queen’s University Press. Mellor, D.H. 1991. Matters of Metaphysics. Cambridge: Cambridge University Press. Mermin, N. David. 1981. “Quantum Mysteries for Anyone.” Journal of Philosophy 78: 397–408. Merricks, Trenton. 2003. “The End of Counterpart Theory.” Journal of Philosophy 100: 521–49. Merton, Robert K. 1936. “The Unanticipated Consequences of Purposive Social Action.” American Sociological Review 1: 894–904. Repr. in Social Ambivalence and Other Essays, 145–55. New York: Free Press, 1976. – 1968. Social Theory and Social Structure. 3rd ed. New York: Free Press.
References 319 – 1973. The Sociology of Science: Theoretical and Empirical Investigations. Chicago: University of Chicago Press. – 1976. Sociological Ambivalence and Other Essays. New York: Free Press. Merton, Robert K., and Elinor Barber. 2004. The Travels and Adventures of Serendipity. Princeton, NJ: Princeton University Press. Meyerson, Émile. 1921. De l’explication dans les sciences. 2 vols. Paris: Payot. Mills, C. Wright. 1959. The Sociological Imagination. New York: Oxford University Press. Monod, Jacques. 1970. Le hasard et la nécessité. Paris: Éditions du Seuil. Mueller-Vollmer, Kurt, ed. 1989. The Hermeneutics Reader. New York: Continuum. Murphy, Edmond A. 1997. The Logic of Medicine. 2nd ed. Baltimore, MD: Johns Hopkins University Press. Natorp, Paul. 1912. Die logische Grundlagen der exakten Wissenschaften. Leipzig: B.G. Teubner. Newton, Isaac. 1947 [1687]. Principles of Natural Philosophy. Ed. F. Cajori. Berkeley: University of California Press. Newton, Roger G. 1989. Inverse Schrödinger Scattering in Three Dimensions. New York: Springer. Newton-Smith, W.H. 1981. The Rationality of Science. London: Routledge & Kegan Paul. Nicod, Jean. 1962 [1923]. La géométrie dans le monde sensible. Preface by Bertrand Russell. Paris: Presses Universitaires de France. Niiniluoto, Ilkka. 1999. Critical Scientific Realism. Oxford: Oxford University Press. Nino, Carlos S. 1985. Etica y derechos humanos. Buenos Aires: Paidos. Novick, Peter. 1988. That Noble Dream: The “Objectivity Question” and the Rise of the American Historical Profession. Cambridge: Cambridge University Press. Ochsner, Kevin, and Matthew D. Lieberman. 2001. “The Emergence of Social Cognitive Neuroscience.” American Psychologist 56: 717–34. Ockham [William]. 1957 [1320s]. Philosophical Writings. Ed. P. Boehner. Edinburgh: Nelson Odling-Smee, F. John, Kevin N. Laland, and Marcus W. Feldman. 2003. Niche Construction. Princeton, NJ: Princeton University Press. Ostwald, Wilhelm. 1902. Vorlesungen über Naturphilosophie. Leipzig: Veit. Pappus. 320? Collection, The Treasury of Analysis. In Ivor Thomas, ed., Greek Mathematics, vol. 2. London: William Heinemann; Cambridge, MA: Harvard University Press, 1941. Parsons, Talcott. 1951. The Social System. New York: Free Press. Pauli, Wolfgang. 1961. Ausätze und Vorträge über Physik und Erkenntnistheorie. Braunschweig: Friedrich Vieweg & Sohn. Peirce, Charles S. 1934. Collected Papers. Vol. 5. Ed. Charles Hartshorne and Paul Weiss. Cambridge, MA: Harvard University Press.
320
References
– 1935 [1898]. Scientific Metaphysics. In Collected Papers, vol. 6, ed. C. Hartshorne and P. Weiss. Cambridge, MA: Harvard University Press. – 1986 [1878]. “The Doctrine of Chances.” In Writings of Charles Sanders Peirce, 3: 276–89. Bloomington: Indiana University Press. Pérez-Bergliaffa, S.E., G.E. Romero, and H. Vucetich. 1993. “Axiomatic Foundations of Nonrelativistic Quantum Mechanics: A Realistic Approach. International Journal of Theoretical Physics 32: 1507–25. – 1996. “Axiomatic Foundations of Quantum Mechanics Revisited: The Case for Systems.” International Journal of Theoretical Physics 35: 1805–19. Pickel, Andreas. 2001. “Between Social Science and Social Technology.” Philosophy of the Social Sciences 31: 459–87. Planck, Max. 1933. Where Is Science Going? Ed. James Murphy. Preface by Albert Einstein. London: George Allen & Unwin. Platt, John R. 1964. “Strong Inference.” Science 146: 347–53. Platt, Mark, ed. 1979. Ways of Meaning: An Introduction to a Philosophy of Language. London, Boston: Routledge & Kegan Paul. Poincaré, Henri. 1908. Science et méthode. Paris: Flammarion. Polya, George. 1854. Mathematics and Plausible Reasoning. 2 vols. Princeton, NJ: Princeton University Press. – 1957 [1945]. How to Solve It. 2nd ed. Garden City, NY: Doubleday Anchor Books. Popper, Karl R. 1950. “Determinism in Quantum Physics and in Classical Physics.” British Journal for the Philosophy of Science 1:117–33, 173–95. – 1953. “A Note on Berkeley as Precursor of Mach.” British Journal for the Philosophy of Science 4: 26–36. – 1956. “Three Views concerning Human Knowledge.” Repr. in Conjectures and Refutations (1963), 97–119. – 1957. “The Propensity Interpretation of the Calculus of Probability and the Quantum Theory.” In S. Körner, ed., Observation and Interpretation, 65–70. London: Butterworth Scientific Publications. – 1959a. “The Propensity Interpretation of Probability.” British Journal for the Philosophy of Science 10: 25–42. – 1959b [1935]. The Logic of Scientific Discovery. London: Hutchinson. – 1963. Conjectures and Refutations: The Growth of Scientific Knowledge. New York: Basic Books. – 1972. Objective Knowledge: An Evolutionary Approach. Oxford: Oxford University Press. – 1974. “Autobiography.” In P.A. Schilpp, ed., The Philosophy of Karl Popper, 1: 3–181. La Salle, IL: Open Court. Porter, Theodore M. 1986. The Rise of Statistical Thinking: 1820–1900. Princeton, NJ: Princeton University Press.
References 321 Premack, David, and Guy Woodruff. 1978. “Does the Chimpanzee Have a Theory of Mind?” Behavioral and Brain Sciences 1: 515–26. Press, S. James. 1989. Bayesian Statistics: Principles, Models, and Applications. New York: John Wiley & Sons. Preston, Stephanie D., and Frans B.M. de Waal. 2002. “Empathy: Its Ultimate and Proximate Bases.” Behavioral and Brain Sciences 25: 1–20 Prilepko, Aleksey I., Dmitry G. Orlovsky, and Igor A. Vasin. 2000. Methods for Solving Inverse Problems in Mathematical Physics. New York: Marcel Dekker. Putnam, Hilary. 1975. Philosophical Papers. Vol. 1: Mathematics, Matter and Method. Cambridge: Cambridge University Press. – 1981. Reason, Truth, and History. Cambridge: Cambridge University Press. – 1990. Realism with a Human Face. Cambridge, MA: Harvard University Press. – 1994 “The Dewey Lectures.” Journal of Philosophy 91: 445–517. Quadri, G[offredo]. 1947. La philosophie arabe dans l’Europe médiévale. Paris: Payot. Quine, W[illard]. V[an Orman]. 1949 [1936]. “Truth by Convention.” In H. Feigl and W. Sellars, eds., Readings in Philosophical Analysis, 250–73. New York: AppletonCentury-Crofts, 1949. – 1953. From a Logical Point of View. Cambridge, MA: Harvard University Press. – 1969. Ontological Relativity and Other Essays. New York: Columbia University Press. Raffaelli, David. 2004. “How Extinction Patterns Affect Ecosystems.” Science 306: 1141–2. Ratner, Carl. 1997. Cultural Psychology and Qualitative Methodology: Theoretical and Empirical Considerations. New York: Plenum Press. Reichenbach, Hans. 1944. Philosophic Foundations of Quantum Mechanics. Berkeley and Los Angeles: University of California Press. – 1949. The Theory of Probability. Berkeley and Los Angeles: University of California Press. – 1951. The Rise of Scientific Philosophy. Berkeley and Los Angeles: University of California Press. Rescher, Nicholas. 1987. Scientifc Realism: A Critical Reappraisal. Dordrecht: Reidel. – 1997. Objectivity: The Obligations of Impersonal Reason. Notre Dame, IN: Notre Dame University Press. – 2000. Inquiry Dynamics. New Brunswick, NJ: Transaction Publishers. – 2003. Imagining Irreality: A Study of Unreal Possibilities. Chicago: Open Court. Restivo, Sal. 1992. Mathematics in Society and History. Dordrecht, Boston, London: Kluwer. Revonsuo, Antti. 2001. “On the Nature of Explanation in the Neurosciences.” In Machamer, Grush, and McLaughlin, eds., 45–69.
322
References
Ricoeur, Paul. 1975. La métaphore vive. Paris: Éditions du Seuil. Rilling, James K., David A. Gutman, Thorsten R. Zeh, Giuseppe Pagnoni, Gregory S. Berns, and Clinton D. Kiets. 2002. “A Neural Basis for Social Cooperation.” Neuron 35: 395–405. Rodríguez-Consuegra, Francisco. 1992. “Un inédito de Gödel contra el convencionalismo: Historia, análisis, contexto y traducción.” Arbor 142 nos. 558–60: 323–8. Rorty, Richard. 1979. Philosophy and the Mirror of Nature. Princeton, NJ: Princeton University Press. Rosenfeld, Léon. 1953. “L’évidence de la complémentarité.” In André George, ed., Louis de Broglie: Physicien et penseur, 43–66. Paris: Albin Michel. Ross, James F. 1989. “The Crash of Modal Metaphysics.” Review of Metaphysics 43: 251–79. Rottschaefer, William. 1998. The Biology and Psychology of Moral Agency. Cambridge: Cambridge University Press. Roughgarden, Jonathan. 1979. Theory of Population Genetics and Evolutionary Ecology: An Introduction. New York: Macmillan; London: Collier Macmillan. Routley, Richard. 1980. Exploring Meinong’s Jungle and Beyond. Interim edition. Canberra: Australian National University. Rugg, Michael D., ed. 1997. Cognitive Neuroscience. Hove, East Sussex: Psychology Press. Russell, Bertrand. 1912. The Problems of Philosophy. London: William & Norgate. Sageman, Marc. 2004. Understanding Terror Networks. Philadelphia: University of Pennsylvania Press. Sampson, Robert J. 1988. “Local Friendship Ties and Community Attachment in Mass Society: A Multilevel Systemic Model.” American Sociological Review 53: 766–79. Savage, Leonard J. 1954. The Foundation of Statistics. New York: John Wiley. Repr. New York: Dover, 1972. Schelling, Thomas C. 1978. Micromotives and Macrobehavior. New York: W.W. Norton & Co. Schlick, Moritz. 1959 [1932/3]. “Positivism and Realism.” In Ayer, ed., 82–107. – 1974 [1925]. General Theory of Knowledge. Trans. A.E. Blumberg and H. Feigl. New York, Vienna: Springer-Verlag. Schönwandt, Walter L. 2002. Planung in der Krise? Theoretische Orientierungen für Architektur, Stadt-und Raumplanung. Stuttgart: W. Kohlhammer. Schrödinger, Erwin. 1935. “Die gegenwärtige Situation in der Quantenmechanik.” Die Naturwissenschaften 23: 807–12, 823–8, 844–89. Schumpeter, Joseph A. 1950. Capitalism, Socialism and Democracy. 3rd ed. New York: Harper & Row.
References 323 Scott, Theodore K. 1966. Introduction to John Buridan’s Sophisms on Meaning and Truth. New York: Appleton-Century-Crofts. Searle, John R. 1980. “Minds, Brains, and Programs.” Behavioral and Brain Sciences 3: 417–24. – 1995. The Construction of Social Reality. New York: Free Press. – 1997. The Mystery of Consciousnesss. New York: New York Review of Books. Sellars, Roy Wood. 1922. Evolutionary Naturalism. Chicago: Open Court. Sellars, Wilfrid. 1961. Science, Perception, and Reality. London: Routledge & Kegan Paul. Shorter, Edward. 1997. A History of Psychiatry. New York: John Wiley & Sons. Siegel, Harvey. 1987. Relativism Refuted. Dordrecht: Reidel. Simmel, Georg. 1977 [1982]. The Problems of the Philosophy of History. New York: Free Press. Sklar, Lawrence. 1993. Physics and Chance. Cambridge: Cambridge University Press. Smart, J[ohn]. J.C. 1968. Between Science and Philosophy. New York: Random House. Sober, Elliott, and David Sloan Wilson. 1998. Unto Others: The Evolution and Psychology of Unselfish Behavior. Cambridge, MA: Harvard University Press. Sokal, Alan, and Jean Bricmont. 1998. Fashionable Nonsense: Postmodern Intellectuals’ Abuse of Science. New York: Picador USA. Spencer, Herbert. 1937 [1900]. First Principles. 6th ed. London: Watts & Co. Squire, Larry R., and Stephen M. Kosslyn, eds. 1998. Findings and Current Opinion in Cognitive Neuroscience. Cambridge, MA: MIT Press. Stove, David. 1991. The Plato Cult and Other Philosophical Follies. Oxford: Basil Blackwell. Suppes, Patrick. 1970. A Probabilistic Theory of Causality. Acta Philosophica Fennica 24. Amsterdam: North-Holland. Suskind, Ron. 2004. “Without a Doubt.” New York Times Magazine, 17 October: 45–51, 64, 102, 106. Sztompka, Piotr. 1979. Sociological Dilemmas: Toward a Dialectic Paradigm. New York: Academic Press. Taton, R[ené]. 1955. Causalités et accidents de la découverte scientifique. Paris: Masson. Tegmark, Max. 2004. “Parallel Universes.” In John D. Barrow, Paul C.W. Davies, and Charles L. Harper, eds., 459–91. Teller, Paul. 1995. An Interpretive Introduction to Quantum Field Theory. Princeton, NJ: Princeton University Press. Thagard, Paul. 1978. “The Best Explanation: Criteria for Theory Choice.” Journal of Philosophy 75: 76–92.
324
References
Thomas, Ivor, ed. 1941. Selections Illustrating the History of Greek Mathematics. Vol. 2: From Aristarchus to Pappus. London: Heinemann; Cambridge, MA: Harvard University Press. Tilley, Christopher. 1999. Metaphor and Material Culture. Oxford: Blackwell. Tilly, Charles. 1998. Durable Inequality. Berkeley and Los Angeles: University of California Press. – 2001. “Mechanisms in Political Processes.” Annual Reviews of Political Science 4: 21–41. Tocqueville, Alexis de. 1985. Selected Letters on Politics and Society. Ed. Roger Boesche. Berkeley: University of California Press. Torretti, Roberto. 1967. Manuel Kant: Estudio sobre los fundamentos de la filosofía crítica. Santiago: Ediciones de la Universidad de Chile. Trigg, Roger. 1980. Reality at Risk: A Defence of Realism in Philosophy and the Sciences. Brighton: Harvester Press. Trigger, Bruce G. 2003a. Understanding Early Civilizations: A Comparative Study. Cambridge: Cambridge University Press. – 2003b. Artifacts and Ideas. New Brunswick, NJ: Transaction Publishers. Uhlmann, Gunther. 2003. Inside Out: Inverse Problems and Applications. Cambridge: Cambridge University Press. Vacher, Laurent-Michel. 1991. La passion du réel. Montreal: Liber. – 2004. Le crépuscule d’une idole: Nietzsche et la pensée fasciste. Montreal: Liber. Vaihinger, Hans. 1920. Die Philosophie des als ob. 4th ed. Leipzig: Meiner. van Fraassen, Bas C. 1980. The Scientific Image. Oxford: Oxford University Press. van Inwagen, Peter. 2003. “Existence, Ontological Commitment, and Fictional Entities.” In Loux and Zimmerman, eds., 131–57. Venn, John. 1962 [1866]. The Logic of Chance. New York: Chelsea Publishing Co. Ville, Jean. 1939. Étude critique de la notion de collectif. Paris: Gauthier-Villars. Vitzthum, Richard C. 1995. Materialism: An Affirmative History and Definition. Amherst, NY: Prometheus Books. Volchan, Sérgio B. 2002. “What Is a Random Sequence?” American Mathematical Monthly 109: 46–62. von Glasersfeld, Ernst. 1995. Radical Constructivism: A Way of Knowing and Learning. London: Falmer Press. von Mises, Ludwig. 1966 [1949]. Human Action: A Treatise on Economics. Chicago: Henry Regnery. von Mises, Richard. 1926. Probability, Statistics, and Truth. London: Allen and Unwin. Waal, Frans B.M. de. 1996. Good Natured: The Origin of Right and Wrong in Humans and Other Animals. Cambridge, MA: Harvard University Press. Waddington, Conrad Hall. 1960. The Ethical Animal. London: Allen and Unwin.
References 325 Weber, Max. 1906. “Kritische Studien auf dem Gebiet der kulturwissenschatlichen Logik.” In Weber 1988a: 215–90. – 1988a. Gesammelte Aufsätze zur Wissenschaftslehre. Tübingen: J.C. Mohr (Paul Siebeck). – 1988b [1904]. “Die ‘Objektivität’ sozialwissenschaflicher und sozialpolitischer Erkenntnis.’’ In Weber 1988a: 146–214. Weinberg, Steven. 1992. Dreams of a Final Theory. New York: Random House. Weizsäcker, Carl-F. von. 1951. Zum Weltbild der Physik. 5th ed. Stuttgart: Hirzel. West, Stuart A., Ido Pen, and Ashleigh S. Griffin. 2002. “Cooperation and Competition between Relatives.” Science 296: 72–5. Wheeler, John A. 1996. At Home in the Universe. New York: Springer-Verlag. Whewell, William. 1847. The Philosophy of the Inductive Sciences. 2 vols. 2nd ed. London: John W. Parker. Repr. London: Frank Cass, 1967. White, Hayden. 1978. Tropics of Discourse. Baltimore, MD: Johns Hopkins University Press. Whitehead, Alfred North. 1919. An Inquiry Concerning the Principles of Natural Knowledge. Cambridge: Cambridge University Press. Whiten, Andrew. 1997. “The Machiavellian Mindreader.” In Machiavellian Intelligence II: Extensions and Evaluations, 144–73. Cambridge: Cambridge University Press. Wiener, Norbert. 1994. Invention: The Care and Feeding of Ideas. Cambridge, MA: MIT Press. Wiggins, David. 1976. “Truth, Invention and the Meaning of Life.” Proceedings of the British Academy 62: 331–78. Wikström, Per-Olof H., and Robert J. Sampson. 2003. “Social Mechanisms of Community Influences on Crime and Pathways in Criminality.” In Benjamin B. Lahey, Terrie E. Moffitt, and Avshalom Caspi, eds., Causes of Conduct Disorder and Juvenile Delinquency, 118–48. New York: Guilford Press. Wilkins, Adam S. 2002. The Evolution of Developmental Pathways. Sunderland, MA: Sinauer Associates. Wittgenstein, Ludwig. 1922. Tractatus Logico-Philosophicus. London: Routledge & Kegan Paul. – 1978. Remarks on the Foundations of Mathematics. Oxford: Blackwell. Wolin, Sheldon S. 2004. Politics and Vision: Continuity and Innovation in Western Political Thought. Princeton, NJ: Princeton University Press. Wolpert, Lewis. 1992. The Unnatural Nature of Science. London: Faber and Faber. Woodbury, Keith A. 2003a. Inverse Engineering Handbook. Boca Raton, FL: CRC Press. – 2003b. “Sequential Function Specification Method Using Future Times for Function Estimation.” In Woodbury 2003a: 41–101.
326
References
Woodger, Joseph Henry. 1952. “Science Without Properties.” British Journal for the Philosophy of Science 2: 193–216. Woods, John. 1974. The Logic of Fiction. The Hague: Mouton. Woolgar, Steve. 1988. Science: The Very Idea. Chichester: Ellis Horwood. Wulff, H.R. 1981. Rational Diagnosis and Treatment. 2d ed. Oxford: Blackwell. Yaglom, A.M., and I.M. Yaglom. 1983. Probability and Information. Dordrecht and Boston: D. Reidel. Yamazaki, Kazuo. 2002. “100 years: Werner Heisenberg – Works and Impact.” In D. Papenfuss, D. Lüst, and W.P. Schleich, eds., 100 Years Werner Heisenberg: Works and Impact. Weinheim: Wiley-VCH. Zakhariev, Boris N., and Vladimir M. Chabanov. 1997. “New Situation in Quantum Mechanics (Wonderful Potentials from the Inverse Problem).” Inverse Problems 13: R47–R79. Zeki, Semir. 1999. Inner Vision: An Exploration of Art and the Brain. Oxford: Oxford University Press. Zimmer, Carl. 2004. Soul Made Flesh: The Discovery of the Brain and How It Changed the World. New York: Free Press.
Index of Names
Agassi, Joseph 128, 263 Alembert, Jean Le Rond d’ 156 Ampère, André-Marie 99, 283 Anderson, Arthur S. 122 Anderson, Roy M. 123 Aquinas, Thomas 135, 270 Ardila, Rubén 142 Aristarchos of Samos 155–6 Aristotle 14, 41, 43, 94–5, 220–1, 223–4, 239, 280 Armstrong, David M. 17, 20, 99, 283 Ashby, W. Ross 125 Atran, Scott 123 Augustine, St 44 Averroës 220 Ayer, Alfred Jules 46 Bacon, Francis 165 Barber, Elinor 96 Barkow, Jerome H. 172 Barnes, Barry 66 Barnett, Arnold 110 Barraclough, Geoffrey 54 Bateson, C.D. 77 Bauch, Bruno 56 Becker, Gary S. 277 Bellarmino, Cardinal 61
Beller, Mara 72 Bergson, Henri 31 Berkeley, George xi, xiii, 34, 43–7 passim Berlyne, David 267 Bernard, Claude 89 Bernardo, José M. 256 Bernays, Paul 151 Bernoulli, Daniel 96, 106 Bernoulli, Jacques 106 Berry, B.J.L. 130 Berry, Donald A. 116 Berzelius, Jöns Jacob 142 Blakemore, Sarah Jayne 172 Blanke, Olaf 23, 121 Blitz, David 36 Bloor, David 199 Bohm, David 69, 97 Bohr, Niels 36, 45, 67, 69, 142, 214 Boltzmann, Ludwig 29 Bolzano, Bernard 27 Boole, George 89 Born, Max 67 Boudon, Raymond 200, 266 Boulding, Kenneth E. 125 Bourdieu, Pierre 66 Boutroux, Émile 99, 283
328
Index of Names
Braithwaite, Richard Bevan 139 Braudel, Fernand 125 Bridgman, Percy W. 45, 70–1, 243 Brink, D.O. 268 Brown, James Robert 66, 200, 263, 284 Brown, Robert 96 Brown, Stephanie L. 77 Brush, Stephen G. 97 Buddha 34, 36 Burawoy, Michael 81 Buridan, Johannes 188, 224 Bush, George W. 66, 137 Cabanac, Michel 172 Cajal, Santiago Ramón y 248 Calvin, Jean 276 Cantor, Georg 216 Carnap, Rudolf 5, 57, 58, 61–2, 65, 83, 107, 111, 113, 207, 243 Cassirer, Ernst 57 Chabanov, Vladimir M. 158 Chisholm, Roderick M. 236 Chomsky, Noam 82 Chrysippus 94 Churchland, Paul M. 78 Clark, Austen 40 Clough, Sharyn 261 Clutton-Brock, Tim 77 Cockburn, Andrew 77 Cohen, Hermann 57 Collins, Harry M. 66 Collins, Randall 46, 199 Comte, Auguste 57 Condillac, Étienne Bonnot de 252 Condorcet, Marquis de 264 Conel, J.L. 82 Copernicus, Nicolaus 5, 61 Cosmides, Leda 172 Counter-Enlightenment 125 Cournot, Antoine-Augustin 99, 102, 105
Covarrubias, Guillermo, M. 155 Craver, Carl F. 144 Cresswell, M.J. 227 Crick, Francis 157 Dalton, John 156 Damasio, Antonio 76 Darden, Lindley 144 Darwin, Charles 59, 120, 142, 224 Davidson, Donald 62–3, 197, 261 Dawkins, Richard 77, 137 Debreu, Gerard 5, 200 de Finetti, Bruno 106, 114 De Morgan, Augustus 106 Dennett, Daniel 74 Derrida, Jacques xii, 257 Descartes, René xi, 10, 41–2, 48, 142, 201, 214 d’Espagnat, Bernard 5 Deutsch, David 15 de Waal, Frans B.M. 77, 171, 172 Dewey, John 6, 85 Dietzgen, Joseph 11 Dijksterhuis, Eduard Jan 41 Dilthey, Wilhelm 27, 54, 57, 64, 75, 174 Dirac, Paul Antoine Marie 202 Dobzhansky, Theodosius 120 Dragonetti, Carmen 105 Duhem, Pierre 72, 143, 155, 200 Dummett, Michael 209 Du Pasquier, L.-Gustave 105 Durkheim, Émile 9, 54, 81, 263 Earman, John 113 Eble, G.T. 98 Eccles, John C. 135, 280 Eddington, Arthur 14 Eddy, Charles 115 Einstein, Albert 20, 69, 153, 202, 210, 242, 246, 247, 263, 275, 276
Index of Names Elias, Norbert 50, 128 Elster, Jon 134–5 Engels, Friedrich 12, 32, 142 Epicurus 97 Euler, Leonhard 87, 202 Everett, Hugh, III 97–8, 210 Faraday, Michael 25 Fehr, Ernst 271 Feigenbaum, E.A. 180 Feigl, Herbert 59 Feldman, Marcus W. 98 Feller, William 102, 109 Fensham, Peter J. 85 Feyerabend, Paul K. 34, 56 Feynman, Richard 12 Fichte, Johann Gottlieb 57, 235 Field, Hartry H. 198 Fine, Arthur 71 Fisher, Ronald A. 114 Fleck, Ludwik 66 Fodor, Jerry A. 219 Fogel, Robert W. 238 Ford, Henry 125 Foucault, Michel 79 Frank, Philipp 61 Frankfort, H. 49 Frankfurt School 31 Friedman, Milton 192 Friedrich Wilhelm II 52–3 Frayn, Michael 35 Fréchet, Maurice 105 Frege, Gottlob 196, 215, 234, 284 Friedman, Milton 192 Gale, David 159 Galilei, Galileo 41, 61 García-Sucre, Máximo 141 Gasparski, Wojciech W. 277 Gellner, Ernest 221
329
Gergen, Kenneth J. 58 Ghiselin, Michael T. 225 Gibson, J.J. 247 Giddens, Anthony 125 Gigerenzer, Gerd 96 Gillies, Donald 107 Gilson, Étienne 224 Giza, Piotr 185 Glennan, Stuart 126 Gödel, Kurt 207 Goethe, Johann Wolfgang 280 Goffman, Erving 79 Good, Irving John 107 Goodman, Nelson 4–5, 65, 221, 236 Gopnik, Alison 82 Gould, Stephen Jay 95, 121 Granovetter, Mark 140 Grassmann, Hermann 193 Greene, Brian 245 Griffin, Donald R. 77 Gross, Paul R. 66 Haack, Susan 216, 266 Habermas, Jürgen 31 Hacking, Ian 113 Hadamard, Jacques 153 Haeckel, Ernst 58 Hall, Richard L. 159 Hamlet 34, 234 Harman, Gilbert 168, 268 Harris, Marvin 54 Harvey, William 142 Hayek, Friedrich A. 264 Hebb, Donald O. 74, 142, 248, 253, 264 Hegel, Georg Wilhelm Friedrich 16, 27, 38–9, 57, 221, 234 Heidegger, Martin xii, 6, 58, 280 Heilbron, J.L. 4 Heisenberg, Werner 45, 67–9, 106, 202, 279
330
Index of Names
Helmholtz, Hermann von 75 Hempel, Carl G. 139 Henri, Victor 97 Herodotus 237 Hesse, Mary 192 Hilbert, David 155 Hippocrates 116 Hiriyanna, Mysore 220 Hobbes, Thomas 141, 263 Hobson, John A. 122 Hogarth, Robin M. 176 Holbach, Paul-Henry Thiry, Baron d’ 124 Horwich, Paul 260 Howson, Colin 115–66 Hubel, David H. 84, 252–3 Hughes, G.E. 227 Hull, David L. 225 Hume, David xi, xii, xiii, 47–9, 58, 88, 164, 166, 267, 276 Humphreys, Paul 101 Hunt, Shelby D. 263 Huntington, Edward V 155 Husserl, Edmund 15, 31, 44, 75, 86, 153, 211, 235, 253
Kahneman, Daniel 111, 176, 271 Kaila, Eino 37 Kant, Immanuel xi, xiii, 6, 22, 36, 50–3, 56–8, 60, 226, 245–6 Kaplan, Edward H. 110 Kary, Michael 142 Kasarda, John 142 Keuth, Herbert 263 Keynes, John Maynard 92, 106, 125 Kim, H. 130 Kim, H.-M. 130 Kim, Jaegwon 283 Kirchhoff, Gustav Robert 143 Knetsch, Jack L. 271 Knorr-Cetina, Karen D. 46 Kolmogoroff, Alexander N. 155 Kotarbisnki, Tadeusz 277 Koza, John R. 180 Kraft, Julius 65 Kraft, Victor 59 Krechevsky, I. 82 Kripke, Saul A. 167, 211, 229 Kuhl, Patricia 82 Kuhn, Thomas S. 34, 56 Kurtz, Paul 124
Ingenieros, José 276 Inquisition 61
Laing, Ronald 79 Lakatos, Imre 153 Laland, Kevin N. 98, 121, 137 Lalande, André 264 Lambek, Joachim 216, 246 Lambert, Johann Heinrich 7, 246 Lanczos, Cornelius 228 Lange, Friedrich 56 Laplace, Pierre Simon 106, 108 Lashley, Karl 82 Laszlo, Ervin 128 Latour, Bruno 46, 66, 283 Laudan, Larry 262
Jacob, François 5 Jains 220 James, William 41 Jeans, James xiv Jefferson, Thomas 248 Jeffreys, Harold 106 John, St 280 Johnson, Samuel 254 Jordan, Pasqual 156 Jung, Carl Gustav 95
Index of Names Lazarsfeld, Paul F. 32 Lederberg, J. 180 Leibniz, Gottfried Wilhelm 27, 29, 43, 193, 225, 231, 244–6 Lenin, Vladimir Ilich 12, 60, 75, 263 Leonardo da Vinci 75 Leontief, Wassily 125 Levi-Montalcini, Rita 45 Levins, Richard 220 Levitt, Norman 66 Lewis, Clarence I. 230 Lewis, David 4, 48, 78, 88, 93, 197, 211, 219, 229–36 Lewontin, Richard 220 Lieberman, Matthew D. 172 Lindley, D.V. 113–4 Lipset, Seymour Martin 122 Locke, John 42–3 Loptson, Peter 197 Lovinbond, Sabina 268 Lowe, E. Jonathan 211, 229 Lucretius 96 Luhmann, Niklas 125 Luria, Alexander Romanovich 83 Lycan, William G. 230 Lyotard, Jean-François 250 Mac Lane, Saunders 194 Mach, Ernst 36, 57, 58, 66, 143 Machamer, Peter 144 Machiavelli, Niccolò 275 Mahner, Martin 121, 133, 142, 179, 266 Mandela, Nelson 225 March, James G. 176 Marsicano, Giovanni 121 Marx, Karl 12, 32, 142 Massey, Douglas S. 77 Matthews, Michael 85 Maxwell, James Clerk 25, 214
331
May, Robert M. 123 McCall, Storrs 212, 237 McCloskey, Donald M. 192 McGinn, Colin 268 McTaggart, John 245 Medawar, Peter 138 Meinong, Alexius 197, 229, 234 Melia, Joseph 229 Mellor, D.H. 99 Meltzoff, Andrew N. 82 Menger, Karl 59 Mermim, N. David 68 Merricks, Trenton 231 Merton, Robert K. 46, 80, 84, 121, 132, 133, 135, 267 Michelangelo 239 Mill, John Stuart 57, 60, 143 Monod, Jacques 98 Montessori, María 85 Moore, G.E. 266 Murphy, Edmond A. 115 Nagel, Ernest 139 Nagel, Thomas 74 Natorp, Paul 57 Neurath, Otto 59 Newton, Isaac xii, 25, 42, 75, 164 Newton, Roger G. 185 Newton-Smith, W.H. 263 Nicod, Jean 245 Nielsen, François 122 Nietzsche, Friedrich 56, 275 Niiniluoto, Ilkka 263 Nino, Carlos S. 268 Novick, Peter 254 Oberschall, Anthony R. 32 Ockham, William 219, 223 Ochsner, Kevin 172
332
Index of Names
Odling-Smee, F. John 98, 121, 137 Ostwald, Wilhelm 143 Pappus 153 Parsons, Talcott 125 Pasteur, Louis 95 Pauli, Wolfgang 67 Pavlov, Ivan 247 Peano, Giuseppe 206 Pearson, Karl 67 Peirce, Charles Sanders 99, 105, 117, 138, 153, 168, 239 Pérez-Bergliaffa, Santiago E. 68, 155 Perrin, Jean 97 Piaget, Jean 46, 89 Planck, Max 3–4, 30, 69 Plato 5, 10, 14, 27, 29, 200, 223, 234 Platt, John R. 167 Platt, Mark 268 Poincaré, Henri 96, 105, 206 Polya, George 146, 208 Popper, Karl R. 60, 93, 99, 102, 103, 105, 111, 115–16, 117, 153, 163, 214, 221, 256 Porter, Theodore M. 96 Portes, Alejandro 140 Premack, David 171 Press, S. James 113 Preston, Stephanie D. 171 Prilepko, Aleksey 151 Protagoras of Abdera 34, 36 Ptolemy xii, 155–6 Putnam, Hilary 6, 34, 209, 253, 261 Quadri, Goffredo 220 Quine, W.V. 15, 28, 195, 197, 199, 201, 207, 209, 221, 232, 262 Raffaelli, David 115 Ramsey, Frank P. 260
Ranke, Leopold 32 Ratner, Carl 75 Reichenbach, Hans 57, 61, 71, 111 Renouvier, Charles 38 Rescher, Nicholas 30, 31, 185, 256 Restivo, Sal 199 Revonsuo, Antti 136 Ricardo, David 32 Rickert, Heinrich 56 Ricoeur, Paul 192 Rilling, James K. 77, 137, 270 Rockenbach, Bettina 271 Rodríguez-Consuegra, Francisco 207 Romero, Gustavo E. 68 Rorty, Richard 6, 74, 284 Rosenfeld, Léon 69 Ross, James F. 212 Rottschaeffer, William 268 Roughgarden, Jonathan 121 Routley, Richard 216, 229 Rømer, Olaf 71 Russell, Bertrand 28, 219, 245 Sagemore, Marc 181 Saki 48 Sampson, Robert J. 142 Savage, Leonard J. 106 Scheler, Max 266 Schelling, Friedrich Wilhelm 57 Schelling, Thomas C. 133 Schlick, Moritz 57, 58 Schopenhauer, Arthur 56 Schönwandt, Walter L. 131 Schrödinger, Erwin 69 Schumpeter, Joseph A. 127 Schutz, Alfred 56 Scott, Theodore K. 188 Searle, John R. 17, 90, 142 Sellars, Roy Wood 36 Sellars, Wilfrid 263
Index of Names Sextus Empiricus 36, 245 Shannon, Claude 101, 109 Shorter, Edward 79 Siegel, Harvey 263 Simon, Herbert A. 176 Sklar, Lawrence 107 Skinner, Burrhus F. 75, 247 Slovic, Paul 176 Smart, John J.C. 263 Smith, Adam 32 Smith, Adrian F.M. 256 Smoluchovski, Marian 97 Sokal, Alan 66, 284 Spencer, Herbert 58 Spinoza, Benedict 36 Stangl, Dalene K. 116 Stove, David 263 Suppes, Patrick 93 Susskind, Ron 66 Swift, Jonathan 49 Szasz, Thomas 79 Szent-György, Albert 85 Sztompka, Piotr 128 Tarski, Alfred 155 Taton, René 95 Taylor, Charles 12 Tegmark, Max 210 Teller, Paul 71 Thagard, Paul 117, 168 Thaler, Richard M. 271 Thucydides 32 Tilley, Christopher 144 Tilly, Charles 141 Tocqueville, Alexis de 32, 54, 142 Tola, Fernando 105 Tolman, Edward C. 82 Tooby, John 172 Torretti, Roberto 51 Trigg, Roger 263
333
Trigger, Bruce G. 54, 74, 79, 128–9, 173 Tversky, Amos 176 Uhlmann, Gunther 154 Urbach, Peter 115–16 Vacher, Laurent-Michel 263, 280 Vaihinger, Hans 56, 191 van Fraassen, Bas C. 4, 62, 72, 262 van Inwagen, Peter 234 Venn, John 106 Vienna Circle 5, 59–60, 143 Ville, Jean 107 Vitzthum, Richard C. 207 Volchan, Sérgio 102 von Glasersfeld, Ernst 84–5 von Mises, Ludwig 121, 277 von Mises, Richard 59, 107 von Weizsäcker, Carl F. 67 Vrba, Elizabeth 169 Vucetich, Héctor 68 Waddington, Conrad Hall 275 Watson, James 157 Watson, John B. 75, 247 Weber, Max 31–2, 50, 64, 174, 214, 238, 256, 263 Weinberg, Steven 26 West, Stuart A. 137 Wheeler, John A. 15, 210 Whewell, William 167, 210 White, Hayden 192 Whitehead, Alfred North 245, 283 Whiten, Andrew 171 Wiener, Norbert 94, 179 Wiesel, Thorsten N. 84, 252–3 Wiggins, David 268 Wikström, Per-Olof 177 Wilkins, Adam S. 177
334
Index of Names
Willis, Thomas 42 Wittgenstein, Ludwig 5, 6, 9, 20, 58, 150, 283 Wolff, Christian 50, 246 Wolin, Sheldon S. 280 Wolpert, Lewis 84 Woodbury, Keith A. 154, 161 Woodger, Joseph Henry 219 Woodruff, Guy 171 Woods, John 216
Woolgar, Steven 46, 66, 284 Wulff, H.R. 115 Yaglom, A.M. 110 Yaglom, I.M. 110 Yamazaki, Kazuo 106 Zakhariev, Boris N. 158 Zeki, Semir 80, 267 Zimmer, Carl 42
Index of Subjects
Abduction 138, 153, 167–8. See also Guessing Abstraction 188–91, 197 Accident 94–8. See also Chance Action 91 Agathonism 273 AIDS 79, 116–17 Algorithm 142, 160–1, 297 Altruism 77, 137, 141, 270–1 Analogy 134, 144, 167–88 Analysis 16, 19 Antipsychiatry 79 Antirealism 56–87, 281. See also Constructivism, Idealism, Irrealism, Phenomenalism, Phenomenology, Positivism, Subjectivism Appearance 3–4, 6, 21, 23. See also Phenomenon Approximation 30, 158, 213–14 Approximationism 30 Apriorism 39, 58 Archaeology 173 Artefact 300 As if 188, 203. See also Fictionism Astronomy 155–6 Atomism xiii, 97, 156 Axiology 266–7
Axiom 206, 275 Axiomatization 68, 155, 232 Babies 35, 82–3 Bayesianism 5, 7, 106–17. Bayes’s theorem 112–13 Belief 112 Bell inequalities 25 Biconditional 90 Biology 134, 168–70 Biostatistics 116–17 Bit 15, 26 Black box 125, 136, 160 Bond 104. See also Structure Brownian motion 96 Buddhism 87 Capitalism 127 Carbon dating 159 Causal principle 92 Causality 88, 247 Causation 88–94, 118; counterfactual analysis of 238; criterion of 94; as energy transfer 91; probabilistic theory of 93 Cause efficient 88–94; final 89, 94 CESM model of a system 126
336
Index of Subjects
Chance 89, 94–118. See also Accident, Disorder, Probability, Randomness Changeability 10, 26–7, 129. See also Energy, Material Chemistry 120, 132, 180 Classicism in quantum physics 69 Class 222. See also Kind, Species Cognition 24 Cognitive neuroscience 169, 171–2, 186 Cohesion, social 140–1 Commitment, ontological 200–2 Composition of a system 126 Computationalism 142–3 Computer 142, 180–1, 189 Conditional 237 Consistency, external 117, 168 Construct 10, 188–9, 195. See also Concept, Proposition, Theory Construction, social 79 Constructivism: ontological 43–7, 66, 253–4, 283 – see also Subjectivism; pedagogical 47, 84–5; psychological 47; social 46–7 Constructivism-relativism 5, 65, 199– 200, 253–4 Contingency. See Accident Convention 289 Conventionalism 156, 191, 206–7 Cooperation 137–8 Copenhagen interpretation 5, 36, 45, 59 Correlation, statistical 123–4 Correspondence theory of truth 260–1 Counter-Enlightenment 125 Counterfactual 212–14, 236–9; analysis of causation 93–44, 238; history 93, 237–8; question 238 Counter-intuitive 84, 265 Counter-norm 267 Counterpart theory 98, 230–1
Counter-Revolution, scientific 43–53. See also Berkeley, Hume, Kant Covering-law model 7, 139. See also Subsumption Criminality 177 Criterion 94, 243 Data 19 Deception 81 Deduction 167 Definition 206 Democracy 122, 130 Denotation. See Reference Descriptivism 53, 142–3. See also Positivism Design, experimental 114 Designation 205 Determinism 88, 99, 108 Diagnosis, medical 170 Diffraction 157–8 Discovery 146, 203 Disinterestedness 32 Disorder 95–6, 104. See also Chance Disposition 213, 239–44; conditional 242; unconditional 242 Distinguishability 24 Doxastics 112 Drake formula 106 Dualism 11 Ecology 98, 115 Economics 5, 92 Economy of thought 60 Effect 91 Efficiency 274, 277 Emergence 14, 78, 92, 120, 224, 248, 290 Emotion 80 Emotivism, ethical 276 Empiricism: classical 33, 45, 58, 245;
Index of Subjects logical 59–63 – see also Positivism, logical Energetism 14 Energy 12, 91, 119, 222 Engineering, social 279 Entanglement 25 Entropy: information-theoretic 109–10; physical 161 Environment of a system 126 Epidemiology 123 Epistemology 6, 33, 59, 251, 254–7 Epoche 44 Equality 206–7. See also Identity Equivalence relation 225 Error 30–1, 99, 262 Ethics 267–77 Event 16, 90–1, 291–3 Evidence 46, 72–3, 261 Evolution: biological 98–9, 120, 131, 224; social 131 – see also History Evolutionary biology 168–9, 224 Evolutionary developmental biology 120 Exactification 107 Exaptation 95 Existence 28–9, 197–8; criterion 28; formal 28, 197, 234; hypothesis 229– 30; mathematical 192; predicate 28; real 28–9, 45, 51, 197, 234 Existential quantifier 28, 197–8 Experiment 69–72, 82, 98; crucial 165–6 Explanation 160; causal 89; mechanismic 139–42; subsumptive 139 – see also Covering-law model Explorer Argument 256 Extension of a predicate 222, 225 Extinction 224 Fact xii, 9, 15–17, 224, 283–302; moral 268–71 Fact-appearance-fiction triad xii
337
Factual/formal distinction 193, 202 Fact/value gap 268–9 Fallibilism 30, 255 Falsificationism 256 Feedback loop 94, 123 Feynman diagram 293 Fiction 9, 188–217; artistic 194; mathematical 196–209; metaphysical 209–13; scientific 213–14; tame 10 – see also Mathematics; wild 10 – see also Many-worlds metaphysics Fictionism 190–3 Fluctuation 97 Forecast 160 Formalism in a theory 202 Foundations 216 Free trade 122 Free will 247–8 Frequentism 107 Function 133–4, 179, 221 Functionalism 125, 133–4 Gambling 96, 102–3, 108–9 Geisteswissenschaft 56. See also Cultural science, Social science Genetics 136, 169–70 Globalization 122 Glossocentrism 58. See also Hermeneutics, Wittgenstein Goal 131–2 Gravitation, theory of 25, 61, 155, 205 Grue 76 Guessing 137–8. See also Abduction, Inference to the best explanation, Theory of mind Heisenberg’s inequalities 68, 69 Hermeneutics xi, 54, 56–7, 63–4, 174– 5, 265 Hindcast 160
338
Index of Subjects
Hinduism 34 History 73, 91, 168, 222, 253–4 Holism 128–9 Homo clausus 50, 128 Hume’s problem 7 Hylomorphism 27 Hylorealism xiii, 27, 33; scientific 279–80 Hypothesis 117, 167 Hypothetico-deductive: method 182–4; system – see Theory Idealism, philosophical 191, 218–19; objective xiv, 27, 198 – see also Platonism; subjective xiv, 211 – see also Subjectivism Ideal object 27–8 Identity 78, 206–7. See also Equality Ideology 31–2 Illusion 23 Illusionism 34 Imagination 86 Immunology 123 Impartiality 32 Imperative 272 Imperialism 122 Indeterminism 99 Indicator 8, 46, 72–3, 141, 182–6, 243 Indirect proof 146, 153 Individual 284 Individualism, methodological 50, 63–4, 174–5, 290 Induction xii, 7, 116, 145, 146, 152–3, 165–7 Inductivism 165–6 Inference: horizontal 152; vertical 153 Information 92; theory of 101–2, 109 Informationism 15. See also Computationalism Instrument, measuring 69–71, 183 Intelligence (spying) 59
Inter-phenomena 71 Interpretation: psychological 63, 81, 174 – see also Theory of Mind, Verstehen; semantic 201–2 Intersubjectivity 255–6 Intuition 175 Intuitionism: axiological 266; ethical 267–8; mathematical 208–9, 215; philosophical 31, 129 Invariance 252, 294 Invention 99, 146, 169, 178–81, 203 Invertibility 159–62, 185 Irrealism 34. See also Antirealism Islam 139 Kicking 28, 254 Kin selection 77 Kind 223–6. See also Species Knower 21–3. See also Subject Knowledge 24; by acquaintance 78; by description 78 Law 14, 51, 134–7, 166, 219–20, 222–5, 235 Lawfulness 48, 136, 222, 293, 301 Law-statement 21, 220, 252, 295–7 Learning 84, 139 Level of organization 96–7 Liberty 248–9 Likelihood 110 Linguistics 78, 82, 188 Logic: deductive 58, 197, 202; deviant 216; dynamic 202; free 216; inductive 117, 145; modal 19, 211, 227 Long waves 130 Macroproperty 137 Management science 136 Many-worlds: interpretation of quantum
Index of Subjects mechanics 97–8, 210; metaphysics 209–14; semantics 212 Market 121–2, 177 Material 10–12, 27, 129 Materialism 26–7; emergentist 27; historical 54; systemic 125; vulgar 26, 207 – see also Physicalism Mathematics 138, 149–55, 188–217 Matter 11–12, 129 Meaning: psychological 56–7, 265 – see also Goal; semantic 188–9, 209 – see also Reference, Sense Measure 100 Measurement 68–71, 183 Mechanics 49 Mechanism 7, 119–44, 126, 170; causal 132; essential 131; latent 133; manifest 133; parallel 130–1, 166; social 122, 131, 175–8; stochastic 100–4, 132 Mechanismic 140 Mechanist worldview 43, 48 Medicine 129, 170 Meliorism 30, 255 Mereology 21 Metaphysics. See Ontology Method, scientific 205, 265 Microeconomics 122, 176 Microproperty 137 Micro-reductionism 77–8 Mind, theory of 138 Miracle 48–9, 262 Model: probabilistic 101–2; theoretical 182, 201–2 Model theory 101 Monism: methodological 264; ontological – see Idealism, Materialism Naming 188–9, 221–2 Naturalism 268. See also Materialism Natural-law theory 270
339
Necessity, nomic 101, 167 Negativism 100 Neo-determinism 105 Neo-Kantianism 50, 56–7 Niche construction 98 Nihilism 87 Nominalism 188, 207, 219–24 Non-referential 257 Nonsense 39 Norm 135, 267–72, 207, 273–5 Noumenon 36. See also Thing-in-itself Novelty. See Emergence Objectivism. See Realism Objectivity 32, 256 Observable 72–3 Observation 67 Ontology 6, 33, 59, 211–12, 238–9, 251 Operationism 36, 45 Opinion 106, 115 Out-of-body experience 23 Parapsychology 121 Particular 221 Pattern. See Law Pauli principle 290 Perception: psychological 75, 78–81; social 78–81 Phenomenalism xi, 6, 38–9, 156–7 Phenomenology 7, 44, 64–5, 86 Phenomenon 33, 35–6. See also Appearance Physicalism. See Materialism, vulgar Physics 12, 38, 119, 158–9 Planning 159–60 Platonism 191, 198, 203, 215, 223, 279–80. See also Idealism, objective Plato’s cave 38 Plausibility 117
340
Index of Subjects
Plausible reasoning 208. See also Abduction, Analogy, Induction Political science 278 Positivism 30, 156–7; classical 57; logical 57–63 Possibility 210–14, 226–39; conceptual 227–9; real 19, 227 Possible world: metaphysics 19, 210– 14, 229–39; semantics 212–14, 257 Postmodernism 250 Pragmatism 191, 215 Praxiology 277–9 Predicate 23–4, 194–6, 219, 284. See also Property Probability 7, 100–18; of causes 162; conditional 93, 101, 112; inverse 162–3; objective 100–5, 241–2; prior 113; of propositions 106–9, 110–11; subjective – see Bayesianism Problem 145–64; direct or forward 150; to find 153; Hume’s 164; ill-posed 154 – see also inverse; insoluble 161–2; inverse or backward 7, 145–64, 185– 6; Newton’s 145, 164; to prove 153 Process 16, 293–5 Proof 199 Propensity 103 Property 11–14, 23–4, 219–22, 284–6; absolute – see Invariant; accidental 13; actual 239; basic 86; blunt 69, 242; cluster or bundle 14; conditional 242; derivative 86; dispositional 239 – see also Disposition; essential 13; general 286; individual 286; invariant 13, 14; manifest 239; phenomenal 35; primary 13, 37–8, 41–2, 86; secondary 13, 37–8, 41–2, 86 – see also Quale; sharp 69, 242 Proposition 188–9, 193, 237 Propositionalization 272
Pseudoscience 280 Psychiatry 79 Psychoanalysis 124 Psychology 75–6, 111; cultural 75; evolutionary 172; folk 174 Pyrrhonism 191, 257 Quale, qualia xi, xii, 35, 40, 73–5. See also Secondary property Quanta 67–72 Quanton 144 Quantum chemistry 132 Quantum mechanics 25, 67–72, 88, 97–8, 105, 108, 158–9, 210, 242, 290 Random choice 102 Randomization 96 Randomness 102, 162–3. See also Chance Ratioempiricism 33 Rational-choice theory 175–6, 277 Rationalism 33 Reaction, chemical 120, 132 Realism xii; axiological 30, 266–7; critical 30; epistemological 29, 251–4; integral 250; internal 271; methodological 30, 263–6; naive 30, 75; ontological 29, 251–4; Platonic – see Idealism, objective; practical 30, 277–9; scientific 29–30; semantic 29, 257–63 Reality: local 25; test 36, 263–4 Realpolitik 250 Reduction 77–8, 186 Reductionism 36. See also Level of organization Reference 16, 72–3, 108, 195–8, 257; class 196–7, 202; frame 13, 61, 235, 252 Reform, social 279 Relativism 6, 270
Index of Subjects Relativity 291 Religion 49, 52, 79, 194 Representation, semantic 205, 288–9 Research 135, 145 Reverse engineering 178–81 Risk 114 Romanticism 43 Rule 151 Say’s law 91 Scepticism xii, 42, 49 Science 30–1; biosocial 264–5; cultural 56–7; factual 193; formal 193; social 64, 173–8, 221 Scientific Revolution 40–1, 96 Scientism 30, 264, 280 Semantic assumption 155, 201 Semantics 212 Semigroup 284 Sensation 60, 75, 156 Sense 207 Sentence 188–9, 237. See also Proposition Separability: subject/object 24–6 Serendipity 95 Signal 91–2 Simplicity 150 Someness 198 Space 51, 244–9 Speciation 120, 224 Species 224–6. See also Kind Spectroscopy 147 Spontaneity 92, 97, 248 State 16, 287–91; of affairs 17, 283; function 287–90; space 17–19, 91, 290–300 Statistical mechanics 104, 107, 136 Statistics 96, 107 Structuralism 101 Structure of a system 126
341
Subject 21–6. See also Knower Subjectivism 34, 66 Subsumption 135–6. See also Coveringlaw model Supervenience. See Emergence Symbol 188 Syndrome 170 System 21, 121, 124–9; hypotheticodeductive – see Theory; social 63–4, 125–8, 138 Systemicity, conceptual 168 Systemism 124–9 Technology 30–1, 123–4, 150, 178–81; social 178 Terrorism 59, 123, 127–8, 181 Theology 147, 231 Theory 8, 62, 82–3, 173, 188; confrontation with data 182–6; ethical 275–6 Theory of mind 146, 171 Thermodynamics 104, 156 Thing 9–11, 27, 37–8, 60; thing for us 22–3 – see also Phenomenon; thing in itself 22–3, 60 – see also Noumenon Thomas theorem 80 Thought experiment 214 Time 51, 244–9; reversal 293–4 Transcendental 218–49 Translucent box 137 Trope 15, 286 Truth: artistic 194; criterion of 261; factual 192, 193–4; formal 193–4, 206; moral 194, 259, 267–73; partial 111, 232, 260; theories of 258–63; value 237, 259 Tychism 99 Uncertainty: objective 109–10; relations – see Heisenberg’s inequalities
342
Index of Subjects
Underdetermination of theory by data 185 Uniformitarianism 301 Universal 12, 218–23 Universe 20, 210 Utilitarianism 276 Value 31–2, 266; individual 267; social 267; theory – see Axiology Variance 109 Variational principle 228–9 Venus 78 Verification theory of meaning 46
Vérité de fait 231. See also Factual truth Vérité de raison 231. See also Formal truth Verstehen 56, 63, 171, 174–5. See also Interpretation, psychological Virtual 228 Word 219 World 21, 64–5; parallel 97–8, 167, 209–14, 227–36 Worldmaking 63–7. See also Subjectivism