Practical Thinking, Reasons for Doing, and Intentional Action: The Thinking of Doing and The Doing of Thinking Hector-Neri Castaneda Philosophical Perspectives, Vol. 4, Action Theory and Philosophy of Mind. (1990), pp. 273-308. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281990%294%3C273%3APTRFDA%3E2.0.CO%3B2-B Philosophical Perspectives is currently published by Blackwell Publishing.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/black.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact
[email protected].
http://www.jstor.org Sat May 12 00:48:36 2007
Philosophical Perspectives, 4 Action Theory and Philosophy of Mind, 1990
PRACTICAL THINKING, REASONS FOR
DOING, AND INTENTIONAL ACTION:
THE THINKING OF DOING AND
THE DOING OF THINKING*
Hector-Neri Castalieda
Indiana University
To deliberate about what to do is to investigate one's body. To face a conflict of duties is on its surface to confront a conflict of reasons for doing this or that; but down deeply inside oneself it is for oneself, as agent, to live a conflict of bodily inclinations to do this or that. Reasons for action, taken de re are not causes, but inclinations to act. To solve the conflict is ostensibly to sort out the conflictive reasons in order to find out what one ought, everything considered, to do; but down deeply inside oneself it is to establish an order of causal dominance among the conflictive inclinations to act. The direction of that causal order is thought of as an intention to do. To deliberate is to gain reasoned beliefs about one's body. The totality of the actions one believes one can do compose one's picture of one's body. Rod Chastein To come to know what to do is to have a thought which itself consists of an awareness of its bringing about an action, or a rearrangement of one's causal powers... The causal dimension of practical thinking is the coalescence of contemplation and the causation of that contemplation, and the contemplation of that causation. (Thinking and Doing)
274 / Hector-Neri Castafieda Introduction
Our main topic here is the structure of human agency; our purpose, to deepen our understanding of human nature. The central phenomenon is the agency involved in intentional action. Enhancing our understanding of human agency is intrinsically worthwhile; but in these days of Artificial Intelligence and Robotics it has an urgent instrumental value. It delineates the framework for the facsimilization of human agency. The center piece of human agency is intentional action performed at will, that is, the causation of effects by episodes of thinking of those very effects. Human action includes bodily movements and causings of states that either come into existence in normative contexts or issue from an agent with certain frames of mind. Thus, habitual behavior may be intentional. A marvel of human agency is the doing of actions based on reasons. Its apex is deliberation. Its major achievement is the adoption of life-encompassing projects. Here we concentrate on the central phenomenon of human agency. We investigate how the characteristic contents of the thinking of human doing enter in the doing of that very thinking. I. Demarcation of Topics
What believing is to theoretical or contemplative thinking and reasoning, intending is to practical thinking and reasoning. Yet what reality, or truth, is to believing is only in part what intentional action is to intending. What one believes is true or false, and to believe it is to take it to be true. What is intended is neither true nor false, and to intend it is not to take it as true, but, rather, as something to be made true. To intend to help make it the case thatp is to ascribe to thatp some worth that imposes a legitimate claim on one's powers to make things happen. To believe that one will not be able to carry out one's intention is to posit that what one intends to do corresponds to-not, of course, is-a falsehood. In any case, intending is the fundamental state of practical reason. Thoughts of intentions, which rehearse or manifest the state of intending, are the fundamental practical operations of (human, rational) agency. The other practical states and acts presuppose and are built up on, or are complements to, intending and rehearsals of
Practical Thinking and Intentional Action / 275 intending. For instance, to want to do A is to be in a state of propensity from which intending to A may issue; wishing someone X to do A is to be in a dispositional state from which one's intending to order, request, or entreat X to A may ensue. Furthermore, a commander's order and requester's petition to X to do A is typically addressed to X as an agent and aims at creating in X an intention to A. The main topics of this essay are: the basic logical form of intentions and of the reasons for intending, the ontic nature of what is intended, and the causal diagram of intentional action. This causal diagram pivots on that structure and on that nature. For convenience, we assume that in rehearsing the state of intending to do an action A an agent has to token a representation of his intention to A. By 'representation' we mean here something semantical, which discharges two different but related functions: (i) plays a role in the computing and causal activities involved in an episode of thinking, and (ii) presents the thought of content to the consciousness constitutive of that thinking. In the present case, the representation we are interested in plays a crucial role in the agent's making up his mind about what to do and in his attempting to do something intentionally. This representation is, of course, internal to the process constitutive of thinking. It must, nevertheless, have a counterpart, a deputy representation, even if one subject to ambiguity, in the public language we use for communication. Hence, we may concentrate on sentences used to formulate intentions, or resolves, to distill from them the peculiar and characteristic logical form, semantics, and causal roles of the actual thinking representations of intended content-whatever they may be and wherever they may occur. Let us call such formulationsintentional sentences. Specifically, then, one main problem here is about what type of special representation, or representations, an intentional sentence must include to be intentional; alternatively put: what syntactico-semantic clues must a sentence contain for it to be properly interpreted as an intentional sentence? Presumably, the investigation is of interest not only to philosophers, concerned with understanding rational agency in general, and to cognitive psychologists focusing on human cognition and action, but also to experts in Artificial Intelligence working on the facsimilization of practical reasoning or, more basically, on the robotic facsimilization of intentional action. The presumption is that an examination of the conceptual framework we use in our experience of intentional action,
276 / Hector-Neri Castaiieda if correctly carried out, should deliver general constraints on, and perhaps guidance for, such facsimilizations. A useful approach to the main topic is partially pragmatic, not really holistic, but molar. We establish the need for, and describe the nature of, any peculiar representational element characteristic of our thought of intentional contents. This characteristic element must reveal itself under a scrutiny of the causal roles of intentions and intentional representations in intentional action. Conforming to a certain tradition, we may say that such a general representational element, as well as the meaning of the public linguistic deputy representation, is a concept constitutive of whatever is thought of with the assertive use of an intentional representation. This approach is analogous to what Kant called transcendental deductions. We deal with a necessary condition of the experience of intending an action, which, as observed, is the core of each practical experience. Then we must confront the complementary empirical problem of identifying the expressions, and the applications, of such a concept. For this we must approach our living language of action descriptively and analytically. 11. The Thinking of Doing 1. Fundamental Practical Thinking: Intentions and Mandates
Practical reasoning is multifarious and hierarchical. The foundation of all practical thinking and reasoning is the thinking of intentions, the thinking involved in making up one's mind about what one ought to do, or in simply adopting without reasons a given course of action. The state of intending is, of course-like that of believing-, merely a dispositional state. Thus, for instance, one normally keeps intact all of one's intentions-and one's beliefs-while resting in profound sleep. Believing is the fundamental dispositional state of so-called theoretical, or descriptive, or pure, or contemplative reason. Similarly, intending is the fundamental dispositional state of practical reason--or, better, as Kant explained, of pure practical reason. Thus, episodes of coming to intend and episodes of fully thinking believinglike to do an action A here now, that is, endorsing rehearsals of intending to A there then, are the primary episodes of the practical activity of the mind. Complex states of intending (like being all set to go on a tour of Germany following a certain intricate plan) require
hactical Thinking and Intentional Action / 277 sequences of interconnected episodes of rehearsals of intending. In last analysis, however, the success of the endorsed plan will depend on a network of endorsed conditional intentions and successful perceptions of the appropriate conditions. We discuss this in some detail below. Reason is one. However, its unity manifests itself in its different uses. First, reason is the power to judge and to make inferences and conjectures. As the power to reason, the highest inferential unity of reason is the logical structure of truth. (This harmony between the structure of truth and the structure of reason has been wrought out by evolution-or by divine providence. The origin and source of this harmony is of no consequence here.) That logical structure underlies our believings as the content of a second-order propensity to consistency and completeness. It is also the same unitary framework within which practical reasonings obtain and are validated. (For example, modus ponens and modus tollens are valid both for contemplative and for practical reasonings.) Second, beliefs function as practical premises. Thus, practical reason is a generalization of contemplative reason. To illustrate, mixed conditionals of commands (e.g., don't see him) or, intentions (e.g., I won't [shan'tlsee him), and circumstances (e.g., he comes late), are mixed commands (If he comes late, don't see him), or intentions (if he comes late, I won't [shan't] see him). In the simplest case, practical reasoning is the thinking involved in the derivation of intentions from other intentions, as, e.g., when one derives the consequent intention from both a conditional intention and the antecedent circumstances that have turned out to be true. For example: (1) If he comes late, I won't receive him.
He has come late.
Hence, I won't receive him.
Yet a word of caution is pertinent. An action may be intentional, issuing from a conditional intention, without its causation involving an episode of inference of the nonconditioned intention from the conditional intention. (We discuss this in detail below.) Simple practical reasoning is not restricted to the first person. We also engage in communicational second- and third-person practical reasoning. For the benefit of an inattentive or obstreperous agent, we may indulge in the derivation of implied commands (or pieces
278 / Hector-Neri Castaiieda of advice, or requests) from circumstances and other commands (or pieces of advice, or requests). Consider a schoolmistress who tells her assistant: (2) Remember my order, John: Call the mother up, only if the child does not arrive by 9 a.m. But the child did arrive at 8:50. Therefore, do not call the mother up. Let's for convenience introduce the term mandate to refer to the members of the category of second- and third-person practical contents like commands, orders, requests, petitions, entreaties, pieces of advice, and their likes. Also for convenience let's speak of the category of practitions to include both the third-and second-person practical contents that are mandates and the first-person practical intended contents, which we will call intentions. Thus, intentions are first-person practitions. Palpaply, mandates are practically derivative and presuppose their corresponding first-person practitions. To begin with, an agent can understand a mandate addressed to her without that understanding exerting in her the most minimal interest in fulfilling it. When the agent acts upon the apprehension of a mandate he, first, understands it, second, transforms it as an intention (or intended content), and, third, implants this content into his own intendings. This gap between second- or third-person mandates and first-personintentions is crucial. The practical primacy of the first-person is of the essence of human agency. The practical primacy of the first-person does not, however, imply that there cannot be a conditioning in which a noise, or mark, expressive of a command can be a stimulus eliciting an automatic response described by the noise, or mark. The category of secondperson practical content is signaled phonetically and syntactically, e.g., in English by the imperative form. Hence, in certain contexts a human agent, or an animal, or a computing mechanism, may respond automatically to a stimulus consisting simply of a word, or to the imperative syntactic mechanism. (This syntactic mechanism is in most computer programs the ascii representation of the Return Key.) For instance, a man in a drill parade, and a trained dog, may respond "unthinkingly" to the noise 'Left' or the imperative 'Stop'. At the other end, even the most reflective agents must still be able to respond automatically to certain urgent noises of warning. The general characteristic purpose of practical thinking is to guide
kactical Thinking and Intentional Action / 2 79
,
behavior and influence conduct. Often this purpose is better served by people in emergencies responding automatically, but with understanding, to the appropriate noises or marks. Yet we must distinguish the mere failure of an automatic response on the part of a trained organism-or a computer-from the refusal to obey, available to an agent, grounded on the understanding, and firstperson translation, of what the agent has been ordered, or commanded, or advised to do. In sum, an agent's ability to produce intentional action rests squarely on his possession of a hierarchy of abilities to respond automatically to diverse types of stimuli. At the most basic level there are those animal reflexes to react to threatening noises or sights. Then there are the conditioned reflexes to respond to the representation of courses of bodily actions. Then there are the learned responses that include thinking and understanding of the nature of the ac.tion, which connect an action with a manifold of bodily movements. These responses presuppose some causal diagrams built in the agent's body where causal energy flows of its own once the agent decides what to do. There must be basic, normal cases in which the agent, acting at will, engages himself automatically in the action he chooses to perform, by immediately executing the appropriate bodily movements that start his doing the action in question. In these cases the agent's thinking intendily, endorsingly, to A inserts the flow of causation into the appropriate causal diagram. For instance, a lady who decides to invite a friend for dinner right now immediately starts the process of making the invitation. Her deciding causes her to walk to the telephone and call her friend up, or walk to her writing table and start writing the invitation note. Deciding is the operational episode of practical thinking: the rest is in those basic cases a mere routine: the flow of causation running smoothly through a causal channel in accordance with a diagram mobilized by the very act of deciding.
2. Deliberation, Conflicts of Duties or Desires, and Contradictions In more sophisticated cases practical reasoning is the thinking involved in important deliberations. The agent engages in a process of thinking alternative situations (or worlds), imagining her reactions in those situations, conjecturing responses, all with the aim of finding out what she ought, everything relevant being considered, to do on
280 / Hector-Neri Castaiieda that particular occasion. In general, in deliberation one faces a conflict of duties or desires, wants, or ends, and one tries to determine whatgiven the total balance of both the reasons for the duties or wants in conflict and the other facts in the world--one must in balance do, or has, ought, ultimately, everything relevant having been considered, to do. These deontic words 'ought to', 'have to', 'must' have many uses and diverse connotations. Nevertheless, in each person's idiolect there is some expression-perhaps several, which he or she uses on different occasions-by means which he or she expresses the solution to conflicts of the acknowledged duties, wants, desires, inclinations, and propensities. For convenience we shall for the most part use the italicized word 'ought', with indices, as our canonical expression. Thus a conflict of duties can be formulated as a situation in which the following obtains: (CD) Ought, (X to do A) & Oughtj (X to do B) & X is unable to perform both A and B. It is crucial to understand that (CD) is generally not inconsistent if # j. On the other hand, the Ought(X to A) & Ought(X not to A) is inconsistent, either because the same subscript is assumed or deontic operators are not conceived as adverbially modalized. To be sure a system of rules can be inconsistent. Then the problem is one of restoring consistency, not of solving a conflict of duties. A conflict demanding solution may remain after consistency has been established1. The two cases must be distinguished. (See Appendix 1.) In deliberations we confront not only conflicts of institutional duties, which duties need not be moral, but the solution sought after is generally a moral one. We confront also conflicts of wants, desires, or purposes. The logic of so-called hypothetical imperatives or of ends and means provides principles according to which wants, desires, purposes, and their ilk generate deontic structures. To be sure these are not, or need not be, moral deontic structures. Using our canonical deontic operator, one such principle can be put as follows:
i
(HI*) X has a desire (want) i that p be the case. If it is (will be) the case that p, then X performs A at time t. At t X can perform A. Hence, X Oughti to (shouldi) do A at t.
Practical Thinking and Intentional Action / 281 A moral general principle of hypothetical imperatives that combines the purposes or desires of a group of agents is this: (HI**)It won't be the case that p unless each agent Xj of a certain class C performs action Aj at time ti. Each agent Xi in C has a desire or goal gj that p be the case. Each agent Xi can at tj perform Aj. Hence, each agent Xj Oughte to do Aj at ti2. Clearly, then, we can deal with deontic structures regardless of whether they are systems of laws created by a legislative body, or systems of stipulations and agreements, or implicit commitments arising from our attitudes or inclinations. Our canonical 'Ought' with appropriate subscripts is simply that generic strong deontic modality an agent confronts, whatever its origin may be. Consequently, schema (CD) above is a general schema of all deontic conflicts whatever their nature may be. We may speak of conflicts of duties or Oughts in a general way, including not just conflicts of genuine duties created by enacted rules, but also conflicts of desires, wishes, wants, purposes, goals, promises, etc. Let's consider an agent who faces conflicting duties arising from rules or norms. Let's catch him deliberating. Patently, he can consider seriously only the rules and norms he acknowledges, not only as somehow existing, but as binding upon him, that is, as impinging on his action mechanisms and exerting a minimal effect on them: in the direction of his doing what the obligation rules prescribe, and away from what the interdicting rules proscribe. Such a normgrounded inclination to action, however minimal it may be, the agent recognizes in his very recognition of a rule as binding. (See note 4.) Qua inclination to act, this inclination is manifested in the same kinds of physicochemical changes through the same kinds of physiological processes as other inclinations manifest themselves. Hence, it is psychologically of the same type as what we customarily call "wants." In brief, normative oughts and genuine duties must be internalized for an agent to include them in his deliberation. This required internalization of relevant duties, especially the conflict-solving, is the grain of truth seized upon by those philosophers who have spoken of one's desire to do one's duty. In any case, the appropriate psychological or physiological terminology is not at issue. The point is that we may, without loss
282 / Hector-Neri Castaiieda of generality, view a piece of deliberation both as the formulation (CD) of a conflict of "duties," or indexed oughts, and as the search for a solution to such a conflict3. Means-ends reasoning often consists of seeking means that constitute also solutions to at least vaguely envisaged conflicts of duties or ends. One is engaged in ascertaining the best means, that is, the means that at the same time both: (i) contributes to one's attaining a particular goal, and (ii) conforms to, or requires acceptable modifications of, the other ends one is committed to, at least concerning one's most cherished or highest ends. In all those cases practical reasoning aims not only at finding out solutions to so-called theoretical questions about the necessary, or even sufficient, causal conditions for it being the case that certain states of affairs obtain, or for one believing what the real states of affairs are. The focal point of practical reasoning is to determine what states of affairs the relevant agents are to bring about. For these we look for reasons that can help us answer the two questions on which deliberation pivots: (Q.l) What Ought I, everything (relevant) being considered, to do? (Q.2) What shall I do? What am I to do? 3. Reasons for Doing The reasons collected, grouped, examined, pitted against each other, and then balanced in deliberation are not mere reasons for believing something, or for something being the case, but reasons for doing, i.e., for the agent to do something. Reasons for doing have two dimensions. On the one hand, they are reasons that sponsor a course of action as the candidate for what the agent Ought everything considered to do: thus, the ranking of reasons for doing determines a ranking of the conflicting actions the agent Ought prima facie to do. To deliberate is simply to attempt to come to believe what the relevant portion of that ranking is. On the other hand, a reason for doing an action A is such that its endorsing consideration, certainly by a deliberator well put together, has the effect of setting in readiness the agent's mechanisms of action oriented toward his doing A ~ . This twofold nature of reasons for doing raises some important questions: What does 'doing' mean in the locution 'reason for doing'?
Practical Thinking and Intentional Action / 283 What does 'for' mean there? To start the process of answering these questions, let's record some important contrasting parallelisms: CONTEMPLATIVE THINKING WHAT IS THE CASE? P WHY? Because r THAT IS: a) r is a reason for it BEING THE CASE that p b) r is a reason for ONE BELIEVING THAT p
PRACTICAL THINKNG
WHAT SHALL I DO?
A: I shall A
WHY?
Because R THAT IS: a) R is a reason for
[MY] DOING A
b) R is a reason for
ONE INTENDING [DECIDING] TO A.
A reason for that p being the case purports to establish the truth of that p. Because of that it justifies believing thatp. Thus, in the case of contemplative thinking both the reasons and that for which they are reasons are truth-valued thought contents. We may call them propositions. In the case of practical thinking the reasons may be truth-valued. But that for which they are reasons are intended contents, and these are as such neither true nor false. The gerund 'doing' in the locution 'reasons for doing' stands not for truth-valued propositions, but for neither-true-nor-falseintended contents, i.e., firstperson practitions. Hence, the conjunction 'for' in the phrase 'reason for doing' denotes some logical relationship that generalizes the sense of the 'for' that announces reasons for it being the case or for believing. 4. Intending as the Culmination of Deliberation
To find out what to do is to acquire a very interesting and peculiar state. To begin with, one comes to be in a position to do intentionally what one Ought to do. Typically this finding out is, again, not a mere theoretical discovering of a truth to be added to one's repertory of beliefs. To be sure one does come to believe many things in deliberation. Notwithstanding, to find out what to do is not so much to come to believe something, but to acquire the state of intending to do what one comes to believe that one should or must do. In the case of contemplative, or theoretical thinking, finding out what is the case, namely, that p, is typically to come to believe that p. Likewise, in the case of practical thinking, to find out what to do, namely, to A, is typically to come to intend to A. Thus, the creation
284 / Hector-Neri Castaiieda of a state of intending to do something specific is the state at which all deliberations and all practical reasonings aim. In brief, a fundamental train of deliberation is in the first-person by the first-person for the first-person. Its penultimate conclusion is a first-person Ought everything considered judgment, which solves the conflict of duties. The ultimate conclusion is an intention. One may, of course, deliberate vicariously in the second- or even the third-person for the benefit of another. (See, for instance, the second-person deliberation by Socrates of Thebes on behalf of Antigone in T&D, Ch. 2.) In such a case the spoken penultimate conclusion is an Ought everything considered judgment from which the deliberator may draw his or her ultimate conclusion, namely, an imperative of advice or another species of mandate. This conclusion is tendered on the expectation that the addressed agent draw the corresponding intention.
5. Summary: Deliberation, Intending, and Weakness of Will We may summarize the preceding discussion schematically by means of the following claims, where '= = >' expresses logical implication in the logic of the Ought of deliberation, and 'he*' is short for 'he himself': (1) One deliberates to ascertain what one Ought, everything relevant being considered, to do. (la) One deliberates in the second- or third-person to ascertain what to inform to others what, on one's view, they Ought everything relevant being considered to do, so that they can translate and implant one's deliberation into their correspondence first-version. (Of course, one may lie in a deliberation in which one impersonates or addresses the agent.) (2) X Ought everything relevant being considered to A = = > [the intention, or intended content] X himself to A. (3) It is not the case that (X believes that he* Ought, everything relevant being considered, to A = = > X intends to A); (4) Yet if X is a typical and attentive deliberator well put together, X may very well conclude his* deliberation (of
Practical Thinking and Intentional Action / 285 course, in the appropriate words of his own idiolect) as follows: "1 must, everything relevant being considered, to A; therefore, I shall (am going to) A." (5) If X is improperly wired (e.g., paralyzed) he may strain himself attempting unsuccessfully to A. (6) If the world around X is inhospitable X may waste his energy in an effort, thwarted by the world, to perform A. (7) If X suffers from weakness of the will, in spite of both his deliberation and perhaps also his being in the state of intending to A, X may yet attempt not to A, and may even succeed in not Aing5. (8) Believed contents and contents believingly thought of are truth-valued and may be called propositions. lntended contents or intentions are the first-person counterparts of the contents of orders, commands, requests, pieces of advice, entreaties-which we call mandates. None of these practical contents are truthvalued. We call them all practitions. Hence, intended contents and the contents intendingly thought, what one thinks to do, are first-person practitions. (9) The central elements in the whole hierarchical structure of practical thinking are both the state of intending and the thinking episodes that are rehearsals of intendinganalogous to the thinking episodes of contemplative thinking that constitute the rehearsals of belief. Hence, to appreciate fully the practical roles of deliberation and conflicts of duties or wants, we must elucidate the causal function of intentions. 111. The Doing of Thinking
I . Formulations of Intention vs Attributions of Intending We must distinguish very carefully between intentional sentences and sentences used to attribute intentions. To illustrate compare the utterances in the following dialogue:
286 / Hector-Neri Castaiieda Virginia: I can't stand it any more! I'm quitting my job. Martha: So you've decided to quit your job. Really? Well, I doubt very much that you will quit your job. You believe that you'll quit. But you won't quit. Virginia: Of course I believe that I will quit my job. But that is not the point. I intend to quit my job; I'm going to quit. You can rest assured that I will quit my job. In the above exchange the underscored locutions formulate Virginia's intention to quit her job. The fundamental expressions of Virginia's resolve are, of course, Virginia's own first-person formulations: I'm quittingmy job and I'm going to quit. The fundamental formulations of intention are so not only because they represent the first-person character of the intention, but also because by appearing in direct speech they appear naked and perspicuously. The fundamental expressions of intention are typically in English first-person futuretense sentences. But in that respect they are ambiguous, for the same sentences can express predictions. In this respect, the attributions of intention constitute a logico-ontological prism. A future-tense sentence that formulates an intention becomes embedded in sentences attributing that intention to the agent in question in the special form of an infinitive. This is what we see in Martha'syou've decided to quit your job and in Virginia's own I intend to quit my job. In contrast, a future-tense sentence formulating a prediction is embedded in sentences expressing the corresponding attributions of belief in its own indicative form, remaining unchanged if the attribution of belief is present or future. This is what we see in Martha's you believe that you'll quit and in Virginia's own I believe that I will quit my job. Thus, the problem of the special intentional representation in intentional sentences is the problem about the represented, the thought-of element that is captured by the special sense, or use, of a first-person future-tense sentence that expresses, not a prediction, but an intention. 2. The Internal Causality of Intentional Action
Obviously, an intentional action is one performed by an agent with a certain frame of mind, which for convenience we shall call an intentional frame of mind. This raises the question: What is an intentional frame of mind?
Practical Thinking and lntentional Action / 287 We must be prepared to find that intentional frames of mind are such, not by virtue of one uniform property common to all of them, but that they form a hierarchical family built on a network of relationships to some basic frames of mind. However this may be, it is rewarding to explore different types of cases. In the simplest case of intentional action an agent simply chooses to perform a rather simple bodily movement, e.g., flexing her right index, blinking, jumping a ditch, grabbing a moving object, picking up a book. Let us scrutinize one such reasonably simple case under our logico-philosophical microscope. Let's suppose that our agent Agens has decided to jump a certain ditch D. It is immaterial for our present purpose whether Agens has acquired the state of intending to jump ditch D as a result of a protracted deliberation, or as a result of an inference from his immediately previous conditional intention to jump D if p, and its being the case that p, or whether Agens simply, so to speak, out of the blue, chooses (i.e., comes to intend) to jump ditch D. Patently, Agens jumps ditch D intentionally only if his intending to jump is somehow involved in his jumping. But how? Dispositional states are not causes, although they can be causal conditions; only events (as Leibniz taught us) are the bearers of energy and, hence, causation. Since Agens' intending to jump is a dispositional state, we need a particular event that both (a) serves as the trigger of the causal train ending in ditch D being jumped by Agens, and (b) involves Agens' intending to jump D. Obviously, an event that is a rehearsal of Agens' intending to jump ditch D can fulfill both conditions (a) and (b). Clearly, as a cause, or as a causal factor, of an intentional jumping by Agens of ditch D, Agens' rehearsing his intending to jump ditch D has to be characterized by its content, by what it represents. The causal connection has an internal dimension. Agens' thinks intendingly to jump ditch D and this thinking causes, in the appropriate causal setup, a causal train that is oriented towards Agens' jumping ditch D. The internal representation of his jumping ditch D makes the thinking in question effective by guiding the thinking's effect. What kind of representation is it, then? To appreciate the force of the question better let's consider a natural theory6: Int=Bel Theory. To think intendingly to A is to think
288 / Hector-Neri CastaReda believingly that one will A when this thinking believingly causes one's Aing. This theory captures in part the internal causality of intentional action. But it does not seem to me to characterize well the intentional content. We may imagine that Agens has been so conditioned by his early training that whenever he thinks believingly, believes occurrently, as other philosophers like to say, that he will jump, or to jump a ditch, his so thinking causes him to jump. Agens' conditioning may be so well-entrenched that regardless of how he resists his inclination he just can not help but jump. Agens may in fact decide not to jump and in his very deciding not to jump he may believe that he will jump, and, lo and behold!, he jumps. Consider a stronger theory than the Int=Bel Theory, one that requires that the agent believe that his thinking believingly that he will A causes his Aing. This theory also fails even for more occasional reasons than deeply-seated conditioning. Agens may be standing before ditch D in the process of some neuro-psychological experiments of which he is a subject. His brain is wired in such an external, but efficient way that his thinking "I will jump ditch D" will cause him to jump. He knows of the nature of the wiring, and believes, when he sees a red light, that the mechanism is operating and hence that he will jump ditch D, and he also believes that his so thinking will cause him to jump ditch D. Yet Agens does not intend to jump ditch D. Clearly, an agent well-put together must be able both to intend to A and to consider, and to believe, that he will not A. Furthermore, an agent well put together must be able both to think intendingly to A and to think believingly that he will not A. Moreover, an agent must at least have a feeling that he is thinking intendingly one content and thinking believingly the other. In fact, a mature agent must be able to tell the difference, so to speak, from inside, from the nature of the contents he is thinking, whether he is thinking an intentional content or a doxastic one. A mature agent well-put together must be able to know immediately and in advance of anypossible effects, without an empirical investigation concerning the effects of his thinking episodes, whether he is attempting to carry out an intention or not. If the difference between intending and believing, or between thinking intendingly and thinking believingly, were not an internal difference in thought contents, the agent could only tell which one
Practical Thinking and Intentional Action / 289 is his mental state, or his mental act, by an empirical investigation about the effects of his mental acts. The immediate knowledge of one's intentions is required for one to be able to acquire intentions in advance, as when one makes promises, and for one to have control of one's actions and accept responsibility for them. I conclude, therefore, that the thinkable content that is intended is not a doxastic content; a fortiori, it is not a proposition. If, as in some recent accounts by David Lewis and Roderick Chisholm claim7, doxastic contents are not propositions, but attributes, then my conclusion is that we must distinguish the propositional attributes that function as doxastic contents from thepractitional attributes that function as contents of intending. I have argued against this Attribute View of Believing and its counterpart Attribute View of Intending8, but I won't press the issue here. For the time being I am content to argue that intentional and doxastic contents belong to different ontico-semantic categories.
3. The Basic Practical Representation: Its Causal Role The intended content has some special element that makes the intendingly thinking of it at least initially causally efficacious. We say "initially" advisedly. A rehearsal of intending to A may turn out to be unsuccessful, (a) because the agent is not well put together: there is a break in the causal paths from the point where the rehearsal impinges to the bodily parts whose movements are required for the agent's Aing, or @) because the environment is inhospitable: there is no causal passage from bodily movement to the state characteristic of Aing. Weakness of the will is a very special case under (a). In any case, because of the basic practical element of intended content, thinking intendingly this intended content is precisely to move the will; indeed, an episode of such a thinking is an act of will, a primary volition. Obviously that element has to be represented in the token that embodies the thinking episode in question. But what sort of element is it? Let us for brevity use the schema 'I to A' to represent a basic intended content, where the infinitival connection located at to represents the special practical element. Thus, a primordial act of will is the tokening of a representation of an intended content, say, of the form I to A now, provided that tokening is constitutive of an
290 / Hector-Neri CastaReda episode of intendingly thinking oneself to A. Obviously, in some sense one is thinking of the first-person proposition I will A. But one is thinking more than that. One is thinking in some unspecified and even inarticulated way of the causality that the very thinking of that content can have, and that one can mobilize in executing intentional action, through which action, if successful, the proposition I will A is made true. The basic practical element, the TO-element, is a signal of that causality. We have seen that the fundamental practical content is a firstperson content. But obviously, the first person appears in predictions and in statements about one's own past. Hence, the first-person subject is not the peculiar basic practical element. Similarly, intended contents are future-tense. The future envisioned in an intention can be so immediate that it may lie within a specious present, as when one expresses a resolve to do something right away: "I'm leaving now." Yet all varieties of futures are included in doxastic contents. Hence, the future itself is not the basic practical element. Patently, any action whatever can be intended by an agent. Thus, the basic practical element cannot be an action. Furthermore, the basic practical element must be systematically connected to the structures of intending and to the causal setup involved in the production of intentional action. It can, at most, be an operator or modality that applies to actions to make them suitable for intending. In brief, the basic practical element in intended content must be a purely formal element that can apply to any first-person subject and to any action whatever. The representation of that element, the basic practical representation, must, therefore, be a general syncategorematic symbolic mechanism applicable to the representation of any action whatever. The occurrence of the basic practical representation in the total representation of an intended content is, precisely, what makes the tokening of that total representation insert the requisite energy in the agent's appropriate Action Center, from where the energy is channeled in the direction of the parts of the agent's body whose movement, if followed by an appropriate string of changes, counts as the agent's doing the intended action. To sum up: i) The representation of the first-person subject of an intended content indicates the total system where the causation of intentional action takes place;
Practical Thinking and Intentional Action / 29 1 ii) The actional predicate or representation of the action in a basic intended content indicates the direction, within the agent's body in which the energy introduced by the intendingly thinking of an intended content is expected to run; thus, the representation of the intended action guides the flow of volitional causation; iii) The basic practical representation, not by itself, but syncategorematically, in the context of the total representation of intended content, occurring at the appropriate Action Center, is the catalyst that bestows upon intendingly thinking its volitional, causal, role. 4. The Syntax of the Basic Practical Representation
We must, then, carefully distinguish between (1) and (2): (1) Agens' intended content: I will jump ditch D. (2) Agens' doxastic content: I will jump ditch D. As we have observed, intentions, first-person practitions, are not propositions. As the intention-prediction ambiguity of first-person future-tense sentences show, an intention and its corresponding prediction have the same first-person referent as agent-subject and the same action type as predicate. Perhaps the intentiodprediction ambiguity of the sentence can be located in the future tense. Then first-person practitions can be considered to differ from their corresponding predictions by two different contentual elements represented, alternatively, by the future tense: we can, thus, speak of the theoretical or contemplative future tense and of the practical future tense. On this conjecture we can for clarity represent (1) as follows: (1A) Agens' intended content: I will-P jump ditch D. On the conjecture that the future-tense is practical/contemplative ambiguous, we may say that an intention and its corresponding proposition differ in a copula: the intention has the practical copula will-P,whereas the proposition has the propositional copula will. Yet we may analyze further the practical copula will-P as a value of the copular modality P applied to the propositional copula will. On the other hand, we can parse our intention (1) differently:
292 / Hector-Neri Castaheda (1B) Agens' intended content: I will P-jump ditch D or I will Poump ditch D). That is, we may at least initially consider the basic practical representation to be an operator on ordinary (contemplative or theoretical) properties, yielding practical properties. These properties are, then, the suitable contents for intentions and other practical thought contents. We may alternatively take the representation P as a modality of the whole predictive proposition. Thus, intended content (1) can be viewed as having a sentential modal form, as follows: (1C) Agens' intended content: P(1 will jump ditch D). A little reflection reveals that the practical modality P of (1C) has to be treated as a modality that applies to some primary actional propositions. The main point is that there are mixed practical contents (and sentences), e.g., conditional intentions or commands: If it rains, close the windows and Only if he asks me will I send him the book. Clearly, mixed practitions are practitions, but the practitional modality remains confined to one component. It is not my plan to defend any one of the preceding logical views of the basic practical representation. Obviously, a logic of practical reasoning has to make a choice as to how to view the fundamental, irreducible representation of intentional causality with triggering forceg. Furthermore, a facsimilization of practical reasoning has to adopt a logic of practical reasoninglo.
5. Other Intentional Frames of Mind We have seen that the background intentional frame of mind from which intentional action flows is composed of a dispositional state of intending to A here now and a rehearsal of such a disposition, such rehearsal being a volition to A. Such an intentional frame of mind is the fundamental one. More complex intentional frames of mind are constructed on varying relations to it. For instance, in adopting a plan of action we create in ourselves right there and then frames of mind within which further subplans and specific intentions will be adopted. Some actions will be intentional by virtue of that intentional background, even if they were not conceived as such,
Practical Thinking and Intentional Action / 293 but have certain determined places in the causal chains issuing from that intentional background. Plans have a holistic character. In the case of a well organized routine, an agent may create the appropriate intentional frame of mind, from which the intentionality of each segment of the routine ensues, by simply entering intentionally into the routine. In such a case each movement he performs will be intentionally performed, even if the agent did not conceive of it when he decided to engage in the routine, because the movement is, so to speak, monitored by the initial volition to engage in the routine here now. This is illustrated by a very skillful clown who once he decides to perform he just unfolds his show. There may, of course, be additional volitions, as, for instance, when the agent comes to a rough spot in his routine, or when the environment introduces a disruption. We also adopt plans that are general or schematic. Here the initial volition creates a general causal diagram within the agent's body, where energy will have to run as it is introduced by other sources, even by further comings to intend either of subplans or specific courses of actions. This may, of course, be well entrenched habitual routines, and can, therefore, be executed by a detailed train of actions that are performed intentionally because of the holistic background created by a comprehensive initial volition. In general, the more schematic a plan of action is the more need it has to be complemented, in fact embodied, through the agent's particular actions that he chooses to perform, i.e., actions that are intentional in the fundamental sense, by virtue of an intentional background of the fundamental type, or by virtue of a holistic monitorial volition. We have no time here to consider the intentionality of actions that are performed in the process of carrying out plans, which can be regarded as hierarchies of intentions. But we must, for contrastive enrichment, consider at least the case of intentional action where there need not be a particular volitional episode, yet there is an operative background intentional frame of mind. This is illustrated by conditional intentions.
6. Conditional Intentions The holistic character of an intended course of action, we have seen, requires that a given volition cover, i.e. start and monitor, a whole sequence of acts. This suggests that plans that depend on future
294 / Hector-Neri Castaiieda circumstances may be governed by an initial monitorial volition. Such plans are actually composed of condition intentions: intentions of the form I f p , then I will A. Therefore, in principle an agent can fulfill intentionally a conditional intention without having to go through the event of coming to intend the consequent. That is, conditional intentions can be adopted by an agent and then put into realization by one overall volition. If the volition that starts the realization of a conditional intention maintains its Gestalt character, then there would be no need for the insertion in the agent's agency of a special volition that brings about the realization of the conditioned intention. Let us consider an example. As a special show of friendship, Bob Rosthal has made a decision that he expresses thus: (1) If Jay Rosenberg visits me tonight, I will offer him my most expensive French wine. Rosthal is, therefore, in a dispositional state of intending, which itself is not conditional but whose content is a conditional intention or intended. (I use the word 'intention' to refer to what is intended, not to the state of intending.) The condition is internal to his intending, and we must be careful to describe his state as having a conditional content, for instance, by saying: (2) At time t Bob Rosthal intends to do the following: if Jay Rosenberg visits him tonight, offer him his most expensive French wine. Clearly, for Rosthal to carry out his intention (I), it is not sufficient that he offers his most expensive French wine to a visitor who, unbeknownst to him, happens to be Jay Rosenberg. Indeed, it may even be the case that the appearance of the visitor in his house is what causes Rosthal to offer him his new French wine. Yet for the offering to be the intentional offering that fulfills the conditional intention (I), it is not sufficient that it be caused by the realization of the condition. This causation would certainly conform to the holistic character of intentional action we have established above. Yet we need more. We need, besides, that Rosthal believe that his visitor is Jay Rosenberg. Some persons will even say that Rosthal carried out his conditional intention (1) if he offers his most expensive French wine to a visitor who he mistakenly believes to be Jay Rosenberg. Such persons distinguish between Rosthal's carrying out intention (1) and Rosthal's doing intentionally the action of offering
Practical Thinking and Intentional Action / 295 Rosenberg his new expensive French wine; for this doing to occur Rosenberg must be around. In a literal sense for Rosthal to carry out his conditional intention (1) it is not enough that he believes that Jay Rosenberg is visiting him. He must actually think that Rosenberg is there visiting him. The condition of conditioned intention (1) must be thought believingly to obtain, and that episode of thinking must be causally involved in Rosthal's offering the wine to whom he (correctly) takes to be Jay Rosenberg. Some philosophers, however, even philosophers who hold a holistic view of intentional action, may want to argue that for Rosthal to carry out his conditional intention (1) there must be another fact, or event, namely, Rosthal's going on to infer, from his conditional intention (I), and his minor premise (that Rosenberg is there), the unconditioned consequent intention:
(3) I am (now) going to offer Jay Rosenberg my most expensive French wine. The view here is this: (V) If an agent has adopted a conditional intention If c, then I
shall A, then he adopts the unconditioned intention I shall A, only if he adopts this intention directly or he infers it from his previous intention and his belief that condition c obtains. Of course, nothing prevents Rosthal from adopting, independently of all his previous decisions, the unconditioned intention to offer Rosenberg his most expensive French wine. With this part of view (V), there is no quarrel. Nor is there any quarrel with the claim that Rosthal can infer his consequent, unconditioned intention to offer Rosenberg his new French wine from both his conditional intention (1) and the condition (as he believes it to obtain) that Rosenberg is visiting him then. The quarrel lies with the onlyif.Rosthal need not make such an inference. He can be moved to act by his thinking, believingly, that the condition of his conditional intention (1) obtains. For that Rosthal's psychological background must include his not having changed his mind about (I), and his intention either explicitly in the focus or in the penumbra of his consciousness, or in available memory beneath consciousness. I claim that the Gestalt character of intending explains how Bob Rosthal can proceed, having adopted conditional intention (I), from his thinking believing (not merely entertaining) that the antecedent
296 / Hector-Neri Castaheda of (1) obtains, directly, without the mediation of any inference and without a new volition to do the action mentioned in the consequent of (I), to the performance of such an action. My claim is based, partly on the following general principle of implication that bridges the logical distance between practical and contemplative reason: (Int. gel.*) If at time t X rehearses his intention to (A at t, if c), then if at t X rehearses his belief that c, then at t X intends to A at t. This principle also bridges with episodes of thinking the distance between the dispositional state of intending to do a conditional action and the dispositional state of intending to do the action simpliciter (without the condition). (Int. Bel. *) is actually a weak principle in two respects. First, although it does not require that there be an occurrence of a thinking that rehearses the unconditioned intention, it requires a rehearsal of the conditioned intention. Second, (Int. Bel. *) requires that the conditioned intention be rehearsed, i.e., be in or beneath consciousness. Thus, we have also: (Int. Bel. **) If during an interval of time d X intends to (A at t, if c), then if time t is included in d and at t X has assertively in the penumbra or in the focus of or beneath his consciousness his intention to (A at t, if c) and at t X thinks believingly that c, then at t X intends to A then*. These principles connect intending and believing along evident logical connections-by modus ponens none other-that relate their contents. The best support for them lies in the holistic insight we gained in the preceding sections. Recall that because of the divisibility and the additivity of realized actions, any action can be considered as a sequence of smaller actions, or as a part of a larger action. We also noted that, regardless of how much divisible an action may be, intended actions must be considered as unitary wholes, whose divisions are junctures included in their very conception. We noted that it is utterly absurd to postulate a special volition for each possible segment that any division of a realized action can yield. Thus, any action especially the more complex ones, can be treated as sequences of unitary actions. But here comes something truly significant. Seldom is a unitary act sequence we intend to perform merely a juxtapositive
Practical Thinking and Intentional Action / 297 succession of acts: acts have a Gestalt, as recorded in this principle: (Act. Str.*) Seldom (if ever) an intended action, A, which can be decomposed into the act sequence al..., ai...,an is such that each act ai is intended by itself. Characteristically, the structure of the intention to A is of the form: intention to ai & intention to (do a2, if a l is performed) & ... & intention to (do a,, if anVlis performed)ll. In short, every possible way of dividing an action that one intends to do yields a sequence of conditionally related actions in one's intending. Recall that an agent does not think, or cannot even think, of each of the possible divisions of his actions into sequences. The agent has only a very schematic conception of the process that would obtain in the world were he to fulfill his intention. The present claim is that if the agent thinks of some possible division of his action, especially an action that he thinks of as an undivided unit, he would consider the latter acts of the sequence as intended conditionally under his realization of the preceding acts. For illustration consider a situation in which you simply decided simply to move your right index finger and you moved in through a distance d. You moved it through d intentionally, but you also moved it intentionally through each distance d' less than d-even though you did not think of d' , let alone of the motion of your finger through d'. The unthought of intentionality of your moving your finger through distance d' + d' ' equally can be analyzed as the unthought-of intentionality of your moving your finger through d" if you have moved it. If the above sequentialization of any action whatever is correct, then the issue, whether or not a sequence of acts conceived as one unitary action is matched by an isomorphic sequence of volitions, is precisely the issue whether or not a sequence of intended conditional acts is to be matched on a one-one basis with a sequence of volitions, or rehearsals of intending. Perhaps the bridging implications recorded in (Int. Bel. *) and (Int. Bel**) may be considered too rationalistic. Perhaps the reader may want to reject both principles and accept causal counterparts that include the additional condition that the agent X is rational. These weaker versions will certainly suffice for many purposes. Yet I believe that the stronger (Int. Bel. *) and (Int. Bel. **) hold also for irrational agents. These principles of implication determine neither which volitions the agent will have nor which actions he will perform. The
298 / Hector-Neri Castaiieda principles merely tell that (dispositional) intending conditional intentions and occurrent believing that the conditions obtain imply the dispositional state of intending the unconditioned intention. The principles are non-empiricistic in that they allow that intentions can be adopted without having to go through consciousness. And this has to do with rationality only to the extent that it involves the rational power to think the believed and intended contents under consideration. But for the main causal thesis about intentional action I am proposing here, the weaker "rationalistic version" suffice. If the reader too desires he may read both (Int. Bel. *) and (Int. Bel. **) as having a built-in suitable antecedent, e.g. "X is at t rational with respect to his intentions."
7. Causation and the Intending of Conditional Intentions Principles (Int. Bel*) and (Int. Bel**) are not causal principles. They pertainto the structure of the one mind or reason of an agent, which functions both contemplatively and practically. Those principles do not tell us which action or indeed that any action will ensue; they merely describe how certain structures of the mind create other structures to which they are logically related. Now, those structures have to do-again, not with the causation of action, since this has to do with the flow of energy-with the structures of the routes for the flow of energy. This is precisely what the dispositional state of intending amounts to. To acquire the state of intending to do an action A is to create within one's body a network of routes for energy to travel, routes oriented toward the place where the relevant muscles and nerves are activated. But the dispositional state of intending is not itself the activation of anything. Thus, when Bob Rosthal made up his mind to offer Rosenberg his most expensive French wine if he visited him, he literally made up his mind: he created channels for energy to go through. He arranged his mental structure. Now, where is the energy required to move Rosthal's body toward his wine collection coming from? Here I am assuming a generally Humean account of causation to the effect that causation is fundamentally a relationship between events. Hence, we have to find an event that is to furnish the energy that can move Rosthal's limbs and muscles. The energy need not, of course, come from outside Rosthal's body. But there must be an event to at least mobilize energy already available potentially.
Practical Thinking and Intentional Action / 299 Clearly, if, as the inferential view (V) has it, Rosthal will have to make an inference to an unconditioned intention, then the event of inferring as follows could certainly provide or mobilize the required energy: (1) If Jay Rosenberg visits me tonight, I will offer him my most expensive French wine. (2) Here is Rosenberg visiting me.
Hence,
(3) I am going to (will, shall) right now offer him my most expensive French wine. Again, I most emphatically accept the doctrine that it is possible, and, in some cases, very likely, that Rosthal reasons "(I), (2), therefore (3)." The issue isnot whether he can make the inference, but whether he must make the inference in order to fulfill his initial conditioned intention (1). Of course, we all agree that Rosthal can very well forget his initial conditional intention (I), and upon seeing Rosenberg decide anew to offer him his most expensive French wine. This is not germane to our present concern. My reason for rejecting the view (V) that Rosthal must infer the unconditioned intention (3) from his intending (1) and his believing (2) is that this has been shown not to be required in the preceding section. If when he comes to intending conditional intention (1) Rosthal creates in himself a structure of routes for energy to go through, it seems redundant to require that he recreates that structure by rehearsing intention (1) as a major premise. Sometimes that may be precisely what he has to do. A reasonable long time may have elapsed since he came to intend conditional intention (I), and he has been enjoying a very rich train of experiences, including the doing of many intentional actions of different sorts, and he may have, thus, blurred those routes. A process of clearing up the network of actual routes may be required, and an act of inferring "(I), (2); therefore, (3)" may do just that. On the other hand, Rosthal made up his mind to do what conditional intention (1) formulates; he has not been enjoying experiences that touch the structure of energy paths that that state of intending consists of. Thus conditional intention (1) is in the penumbra of consciousness, or just beneath consciousness. Now, Rosthal being in that frame of practical mind, Rosenberg, with his characteristic flare, with no disguises or hats, obviously himself as
300 / Hector-Neri Castaiieda he can be, and enjoys being, comes in and greets Bob Rosthal with his unmistakable voice and style. Rosthal perceives Jay Rosenberg. Perceiving is an event, and one that brings in energy to Rosthal's brain. That event creates in Rosthal the dispositional state that Rosenberg is visiting him, but that dispositional state is not involved in the mobilization of energy, although it is involved in the guidance of the energy to be mobilized. Why could not the event of Rosthal's perceiving Rosenberg, locating Rosenberg in his visual field through its demonstrative references, mobilize energy (whatever its ultimate source) and place it at the position where it will go through the clear routes that constitute the dispositional state of Rosthal's intending conditional intention (I)? Why indeed could it not do so? It is purely an empirical matter whether or not the event of perceiving Rosenberg can mobilize enough energy within Rosthal's body for him to rise up, go to his wine cellar, choose from his French collection a rare Rothschild Lafitte, open it, and offer it to Jay Rosenberg. That is my point: it is an empirical matter. The empirical circumstances both within Rosthal's body and without his body can be such as to provide the clear channels for the requisite energymobilized by his perceptual thing-to flow. Rosthal's agency mechanisms may be well put together with no relevant circuit broken; the immediate environment must be hospitable: there are no unsurmountable obstacles in the way to the wine cellar, etc. The case of Bob Rosthal is relatively simple. We have here an isolated conditional intention (or so it seems), and his act does not seem terribly complex. The nature of the phenomenon is perhaps clearer in the case of more complex projects. The analysis we provided above for the sequentialization of an action into conditional intentions should make the whole thing obvious. Let us briefly apply that analysis schema (Act.Str. *) to Rosthal's action. Clearly, once he perceives Rosenberg (as Rosenberg), whether or not he derives intention (3) from intention (1) and (2) being immaterial, Rosthal will go to the wine cellar, etc. These actions through which he carries out his intention (3), form if you wish, a train of acts, each of which Rosthal performs intentionally. He goes to his wine cellar intentionally, and intentionally walks down the stairs, and so each step he takes he takes it intentionally. As long as there are no obstacles in the way, Rosthal performs-like Ryle's clown-a piece of habitual behavior. His total intention covers holistically the whole sequence of movements. But he will not have failed to fulfill his
Practical Thinking and Intentional Action / 30 1 intention to take step 15, if his course of events is forced to abort at step 5 by the invincible obstacles. Step 15 enters in his plan, in his global intention, as one to be taken if step 14 is taken, and so on. In general, habitual intentional action must be understood as a unitary pattern of acts that enter into the agent's intentionality in one piece, globally. The agent goes through each of the steps as if he had only contemplative consciousness of it. But that there is practical, volitional consciousness underlying it, undergirding the step to the previous steps, is manifested perspicuously when obstacles appear. An obstacle breaks the tranquillity of consciousness and demands on the agent's part a fresh new volition to continue his planned course of action, or to find a detour, or even to cancel the project. The greatest economy of intentional action requires that it must be possible for an agent to adopt sometimes a conditional intention and be moved to action by the thought that rehearses the belief that the condition obtains-without the mediation of an inference from the conditioned intention and the belief in the condition to the unconditioned consequent. Thinking takes timeas Plato taught us-and consciousness is too complex and too precious a commodity to be squandered away in unnecessary inferences. In an admirable universe-and I believe ours is, as I learned from Leibniz-neither thinking nor consciousness is squandered away. That is why we need consciousness and hard thinking when we are learning practices, acquiring habits, and overcoming obstacles. Thereafter our behavior is habitual, efficient with great economy of consciousness, a precious commodity. Yet habitual behavior is still intentional and voluntary.
8. Reasoned Intentional Action We have argued above that an agent can be moved to do an action he conditionally intends to do upon his thinking that the condition obtains. Such an action is intentional, even if it is not caused by a volition to do it. It is a special case of what we may call reasoned intentional action. Another species is that in which the agent infers the unconditioned intention from both his conditional intention and his belief that the condition obtains. In the latter case there is a volition to do the (unconditioned) action. Reasoned intentional action contrasts with mere intentional action. This is an action that arises from a volition to do it, but this volition
302 / Hector-Neri Castaneda has no bases or reasons for its adoption. Mere intentional actions need not, of course, issue from uncaused volitions. Whether a volition has a (sufficient) cause is the problem of metaphysical determinism about the will. Clearly this problem cuts across the distinction between reasoned intentional action and mere intentional action. If a reasoned intentional action issues from the agent's inferring the unconditioned intention from the other premises, then the derived volition may still be causally underdetermined. The logical fact that the intended conditional intention and the believed condition imply the unconditioned intention does not guarantee that the agent experiences the volition to perform the unconditioned action. Recall that principles (Int. Bel. *) and (Int. Bel. **) only guarantee that the agent will have the dispositional state of intending the unconditioned intention. But from the possession of a dispositional state, or a capacity, even a propensity, to the occurrence of an exercise or manifestation of that disposition, capacity, or propensity, there is an enormous chasm. In any case the chief point we are making at this juncture is that a reasoned intentional action contrasts with mere intentional action. A sudden determination to do some action right away-whether the determination is caused or not-without the consideration of any reasons, would, if fulfilled, deliver a mere intentional action. Reasoned intentional action is the genus under which falls intentional action that arises from deliberation. In the case of deliberation there are conflicts of duties, wants, and intentions recognized by the agent. His problem is both to establish some ranking among those items in conflict and to connect that ranking to his motivational profile, so that he can ascertain the intentions which he is both to give preference, and, ultimately, to think volitionally.
Conclusion In this essay we have investigated the contents and the causation of practical reason. We have seen how the contents of practical thinking must have a unique element that allows episodes of practical thinking to have their unique causal role. In that element lies the ultimate autonomy of practical reason, and, a fortiori, the derivative autonomy of morality. That causal element of practical thinking can
Practical Thinking and Intentional Action / 303 be represented in many different logical ways. In the Indo-European languages it seems characteristically to appear as a form of copulation, e.g., in imperative sentences, in practical infinitive or subjunctive clauses. In Artificial intelligence it can be represented by means of different logical mechanisms. (See Note 9.) But the progress of Robotics, in its attempts at facsimilization of intentional action-not to mention the facsimilization of complex behaviors that carry out complex plans-will have to involve some representation of that peculiar practitional element both as content and as the nucleus of the causation by the facsimiles of intentional thinking, or volition-like episodes, of what is thought in such episodes. Notes
* An earlier version of this essay was written for, and presented at, an Interdisciplinary Conference Workshop on Practical Reasoning held at Stanford University during June 18-23, 1984. 1. See Hector-Neri Castaiieda, Thinking and Doing: The Philosophical Foundations of Institutions (Dordrecht: D. Reidel Publishing Co., 1975), Ch. 7, Sections 5-6 and especially 14. 2. Principle (HI**) is a revised version of the principle of collective hypothetical imperatives (H,N) formulated in my "On the Conceptual Autonomy of Morality," Nous 7 (1973): 67-77, pp. 74f. As far as I know (H.N)is the first proposal of a social hypothetical imperative. For another principle corresponding to (HI*) see Hector-Neri Castaiieda, "Reply to Michael Bratman: Deontic Truth, Intentions, and Weakness of the Will" in James E. Tomberlin, ed., Agent, Language, and the Structure of the World (Indianapolis: Hackett Publishing Company, 1983),to be cited as Tomberlin 1983. 3. The indexes of deontic operators I have variously called "qualifiers," "adverbial modalities," and "institutional modalities." The last one I like best. See T&D Chs. 2 and 7. However, it should not be assumed that each institution determines or is characterized by just one deontic structure. Most institutions are actually hierarchies of deontic structures. A simple promise is perhaps just a system composed of a single deontic structure. Morality, for instance, is a two-storied institution: a general and vague conception of ideal of human nature and its fulfillment on top and a concrete moral code. A moral code is, as remarked in the Appendix 1 below, a system of rules composed of three interlocking deontic structures: the euergetic system (dealing with interpersonal relations outside institutions),the ethical system (dealing with institutional duties and their conflicts), and the metaethical system (dealing with the needs to correct logical or moral errors in the moral code behind the ethical and the euergetical systems). Each of these structures yields an adverbial sub-institutional modality. Hence, there is no monolithic moral
304 / Hector-Neri Castaheda obligation. The question "What ought I morally to do here now?" allows of (at least) three possible different answers: I ought morally euergetically to do A; I ought morally ethically to do B, and I ought morally metaethically to do C. For details see The Structure of Morality (Springfield, Illinois: Charles C. Thomas Publisher, 1974), Ch 8. For a preliminary study of some of the main adverbial deontic modalities characterizing a legal system, see Hector-Neri Castaneda, "The Logical Form of Legal Systems: A New Perspective" in Antonio Martino, ed.,
Deontic Logic, Computational Linguistics, and Legal InformationSystems, Vol. 2 (North-Holland Publishing Company, 1982). A detailed application to an article of a New Jersey law together with a proposed rich secondorder deontic logic appears in Hector-Neri Castaneda, "The Basic Logic for the Interpretation of Legal Texts," in Charles Walter, ed., Computer Power and Legal Language (Westport, Connecticut: Greenwood Press, Inc, a Quorum Book, 1988)(forthcoming).For an extension of this deontic framework to epistemic deontology, see Hector-Neri Castaneda, "Knowledge and Epistemic Obligation," in J.E. Tomberlin, ed., Philosophical Perspectives, 2, Epistemology (1988): 2 11-233. 4. This is a generalization to.all Ought thinking of what Kant called "respect for the law" (Achtung furs Cesetz). Kant actually used this phrase to refer to the internal re-arrangement of inclinations wrought out by thinking a judgment with an ought (must) everything considered. As I see it, he confused the in-balance Ought with the moral ought-because he focused on moral agents, for whom, if they reason correctly, the two ought's are indeed convergent. As Kant so beautifully put it in Foundations of the Metaphysics of Morals, Prussian Academy, Vol. IV, p. 40111, about the conflict-solving Ought we should say: What I recognize directly as a law for myself I recognize with respect, which means merely the consciousness of the submission of my will to a law without the intervention of other influences on my mind. The direct determination of the will by the law and the consciousness of this determination is respect. Here Kant is misplacing the causal determination of the will on the deontic operator. It belongs rather to the intended content that underlies the deontic operand. More specifically, as we discuss below, that causal role belongs to the peculiar and characteristic representation of the intentionality of, an intended action. See T&D, Ch. 11. 5. For an account of weakness of the will, built upon the thesis that intended contents are not propositions, but first-person practitions, see the exchange between Michael Bratman ("Castaneda's Theory of Thought and Action") and Castaneda (paper mentioned above in Note 2 above) both in Tomberlin 1983. Because of the proposition/practition distinction that account is richer than the customary ones. 6. For a theory of intentions akin to Int=Bel, but with an emotivist twist, see Bruce Aune's "Castaiieda on Believing and Intending" in Tomberlin 1983. See also Wilfrid Sellars's "Conditional Promises and Conditional
Practical Thinking and Intentional Action / 305 Intentions (Including a Reply to Castaiieda)," and my rejoinders to Aune and Sellars, all in Tomberlin 1983. See also T&D, Ch. 6, and Ch. 10 Section 3. 7. See David Lewis, "Attitudes De Dicto and De Se," The Philosophical Review 88 (1979): 513-543, and Roderick Chisholm, "Review of Thinking and Doing," Nous 12 (1979): 385-396, and The First Person (Minneapolis: The University of Minnesota Press, 1981). This view has been adopted by Myles Brand, first in "Intending and Believing" in Tomberlin 1983, and later in his Intending and Acting (Cambridge, Massachusetts: MIT Press, 1984): 123-175, 272ff. 8. 1 have argued against the Attribute View for intending in "Reply to Myles Brand: Intentions, Properties, and Propositions," and against the Attribute View for believing in "Reply to Ernest Sosa: Self-Reference and Propositions," both in Tomberlin 1983, and more fully in "SelfConsciousness, Demonstrative Reference, and the Self-Ascription View of Believing," in James E. Tomberlin, ed., Philosophical Perspectives I (1987): 405-454. For related discussion of the Attribute view see the following studies by James_E.Tomberlin: Critical Review of Myles Brand, Intending and Acting, Nous 2_1 (1987); Critical Review of David Lewis, On the Plurality of Worlds,Nous 2 111989); Critical Review of R.J. Bogdan, ed., Roderick M. Chisholm, Nous (forthcoming); and "Belief, SelfAscription, and Quantification," presented at the second conference of SOFIA on consciousness in Buenos Aires, Argentina, July 1989. 9. In T&D Chs. 3-9 practitionhood is represented as a copula. In Ch. 2 it is shown that treating practitionhood as a form of copulation is equivalent to treating it as a predicate operator. In Ch. 10 it is considered as an operator on the propositional (indicative)form of predication. In HectorNeri Castaiieda, "Ought, Time, and Deontic Paradoxes," The Journal of Philosophy 74 (1977): 775-791, practitionhood is treated as a proposition operator. 10. For rich deontic logics see T&D Chs. 7-9, the papers on legal systems mentioned in Note 3, and Hector-Neri Castafieda, "The Paradoxes of Deontic Logic: The Simplest Solution of All of Them in a Fell Swoop," in Risto Hilpinen, ed., Deontic Logic and the Foundations of Ethics (Dordrecht: D. Reidel Publishing Company, 1980).The richness of those deontic logics goes hand in hand with their ability to solve the deontic paradoxes in one swift move. This move is the introduction of the proposition/practition distinction. (See Appendix 1.) 11. This principle is what ultimately explains the patent force of Gilbert Ryle's attempt at proscribing volitions from reality by means of ridicule, when he asked: "How many volitions did the skillful clown execute during a certain difficult performance?" See his The Concept of Mind (London: Hutchinson, 1949). The propriety of the question is obvious. It does not show, however, that there are no volitions. The answer may be: "Just one: His monitorial and incipient volition to start his performance off." But the aging clown may have needed to intersperse several monitorial volitions not to falter. For a broader discussion of the issue of volitions and time see Hector-Neri Castaiieda, "Intentional Action, Conditional
306 / Hector-Neri Castafleda Intentions, and Aristotelian Practical Syllogisms,"Erkenntnis, 18 (1982): 239-260.
Appendix 1 MORAL DILEMMAS, CONFLICTS OF DUTIES, AND CONTRADICTION. There is a growing literature on moral dilemmas that sees moral dilemmas as contradictions. This has promoted the extraordinary idea that a deontic logic that is consistent is not morally neutral! It also has promoted the idea of a minimal deontic logic with so little structural requirements so that unindexed obligation statements of the form Obligatory (A)& Obligatory (-A) are consistent. To me it seems that understanding of our experience of rules and conflicts of rules is not encouraged by such views. Rules have a systemic nature. They belong into different systems with their own brand of obligatoriness, forbiddenness (wrongness), and permittedness. For instance, each promise is a normative system in its own right; each law, each chapter of a code, indeed each article of a statute determines a normative system. Obligations are indexed to systems of rules. Conflicts of duties are collisions of deontic systems. This is what (CD) formulates: (CD) Oughti@ to A) & Oughti@ to B) & Cannot (X do both A and B). The manifolds of purposes, needs, abilities, and actions of individuals and groups in each society are of tremendous complexity. And each enacted law, ordinance, contract, promise makes a posit about the future of human needs and purposes, technological developments, and constancy of the environment. But these posits, regardless of how ambitious they may be, are very limited in scope. Therefore, the more rules we need the more conflicts of rules we are bound to live. Again, these conflicts are collisions of deontic systems in given circumstances. The possibility of conflicts of duties-or the possibility of contradictions in a given normative system-should not prevent us from appreciating the need to have a rich deontic logic, a logic capable of accounting for our compliance and our violation of rules. Within each system we must be able to reason so as to distribute obligatorine~s~ across logical implications. of the components of Otherwise we could not recognize the obligat~riness~ a complex plan of action, or the additive obligatorine~s~ of certain compound actions made up of obligatoryj components. Patently, every rule R makes certain actions obligatoryR contingently upon certain circumstances obtaining. We must, clearly, be able to apply modus ponens from the circumstances and rule R to the unconditioned obligatorine~s~ of the action. Otherwise, we will not be in a position to apply rule R to our actions. (This trivial requirement of normative life is the breaking point of many a deontic system.) Hence, we need a good deal of deontic logical structure to live our institutional lives. The deliberate limitation to very simple deontic logics aim
Practical Thinking and Intentional Action / 307 at preserving the consistency of Ought (X to do A) & Ought (X not to do A) is well motivated; nevertheless, it is obscurantist in that it precludes understanding of our normative experience. The problem of contradiction is real, and must be solved by revising the moral code, the laws, the statutes, or whatever normative system reveals itself contradictory. The procedures for revision must be determined within each institution-with morality supervising them all. (For a detailed discussion of all the above points see T&D, Ch. 7.) So-called paraconsistent logics allow a system to have contradictions and develop mechanisms for stopping the flow of inference so as to minimize the damages arising from contradictions. The idea is admirable; the project is, however, misplaced. Contradictions are to be avoided not merely as premises, but eliminated. When we find a contradiction those among us who are rationally constructed have a twofold second- order propensity toward consistency: not to make inferences from the contradiction-except by reduction ad absurdum and deny one assumption-, and to restore consistency by revising our beliefs. Evolution has implanted such secondorder propensity. Here,are a few of the most interesting papers on the bearing of moral dilemmas on the consistency of moral codes or on the minimal morallyneutral deontic logic. E. J. Lemmon, "Deontic Logic and the Logic of Imperatives,"Logique et analyse 8 (1965): 39-71; Bas C. Van Fraassen, "Values and the Heart's Commands," The Journal of Philosophy 70 (1973): 5-19; Terrance McConnell, "Moral Dilemmas and Consistency in Ethics," Canadian Journal of Philosophy 8 (1978) 269-287; Ruth Barcan Marcus, "Moral Dilemmas and Consistency," The Journal of Philosophy 77 (1980): 121-136; Earl Conee, "Against Moral Dilemmas," The Philosophical Review 91 (1982): 87-97; Geoff~eySayreMcCord, "Deontic Logic and the Priority of Moral Theory," Nous 20 (1986): 179-197. I have expounded my view of conflicts of duties and inconsistent laws in Thinking and Doing: The Philosophical Foundations of Institutions (Dordrecht: D. Reidel Publishing Company, 1975) Chs. 2 and 7. I discuss moral conflicts and the moral duty to revise a defective moral code in The Structure of Morality (Springfield, Illinois: Charles C. Thomas Publisher, 1974) Ch. 8. This chapter presents a threefold compartment of moral codes, very roughly divided along the following lines: (i) the euergetical system of duties pertaining to the solution of interpersonal conflicts outside institutions;(ii) the ethical system containing principles for ranking institutions and establishes duties aimed at solving conflicts of institutional duties; (iii) the metaethical system, which establishes duties to change the agent's own moral code. For a defense of paraconsistent logic in deontic reasoning see Newton C. A. da Costa and Walter A. Carnieli, "On Paraconsistent Deontic Logic," Philosophia 16 (1986): 293-305. For a paper arguing that some sentences are both true and false see Graham Priest, "The Logic of Paradox," Journal of Philosophical Logic, 8 (1979): 219-241. For a descriptive study of actual reasonings involving contradictory theories in physics see Joel Smith, "Inconsistency and Scientific Reasoning," Studies in History and Philosophyof Science (forthcoming). This study shows how paraconsistent logic plays
308 / Hector-Neri CastaReda no role in those reasonings. Yet those physicists were reasoning consistently from premises d r a m from an inconsistent body of theories and networks of auxiliary hypotheses. Of course, the best solutions to the Liar Paradox and kindred perplexities may very well require the recognition of an isolated region of propositions that are both true and false. That region has to be outside the standard paths we must travel to solve conflicts of duties, tensions in empirical data, or even collisions between empirical theories. James E. Tomberlin has chastised the deontic systems of Bas Van Fraassen, Asizah Al-hibri, Peter L. Mott, David K. Lewis, Fred Feldman, and others for not allowing the necessary inference from the obligatoriness of an action in a certain circumstance and the obtaining of the circumstance, to the straight obligatoriness of the action. This structural weakness originates in weakenings of deontic logic to solve some of the deontic paradoxes. The remedy seems to me to be suicidal. The purpose of proposing deontic logics is presumably to give us understanding of how our deontic reasonings, especially the most basic ones, are valid. Because of the fruitfulness of the patent proposition/practition distinction in solving the paradoxes, that structural weakness issues from neglecting that distinction. See TomberFn's papers: "Contrary-to-Duty Imperatives and Conditional Obligation," Nous 15 (1981): 357-375; "Good Samaritan and Castaneda's Deontic Logic," in James E. Tomberlin, ed., Hector- Neri Castaiieda (Dordrecht: D. Reidel Publishing Company, 1986); "Obligation, Conditionals, and the Logic of Conditional Obligation," Philosophical Studies 73 (1989); "Deontic Paradox and Conditional Obligation", Philosophy and Phenomenological Research (forthcoming); and_his Critical Review of Lennart &vist, Introduction to Deontic Logic, Nous (forthcoming).