Action, Inference, Belief, and Intention Bruce Aune Philosophical Perspectives, Vol. 4, Action Theory and Philosophy of Mind. (1990), pp. 247-271. Stable URL: http://links.jstor.org/sici?sici=1520-8583%281990%294%3C247%3AAIBAI%3E2.0.CO%3B2-J 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.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact
[email protected].
http://www.jstor.org Fri Jun 22 07:20:26 2007
Philosophical Perspectives, 4 Action Theory and Philosophy of Mind, 1990
ACTION, INFERENCE, BELIEF,
AND INTENTION
Bruce Aune
University of Massachusetts, Amherst
0ver.the past twenty-five years I have elaborated and defended a philosophical account of purposive behavior that might be described as a computational theory of action.' My theory has changed in various respects in recent years, and I have a much better idea now of the philosophical purposes it serves. The aim of this paper is to discuss central tenets of the theory as I now see them-tenets concerning reasoning, belief, and intention-and to expose the considerations that weigh in their favor. In attempting the latter I shall be clarifying and defending the purposes that a philosophical "theory" of action ought, in my view, to fulfill. 1. The Computational Aspect My theory can be described as "computational" because, according to it, purposive behavior is not only explainable but usefully describable by reference to the reasoning or "computations" that brought it into being.2 These features of purposive behavior stand out sharply if we consider unusual, eccentric, or bizarre actions. Suppose Jones is observed in the shrubbery one evening wearing a necklace of garlic cloves and carrying a revolver loaded with brass bullets. What is he doing? The answer might be "Hunting vampires." This answer, which describes Jones' behavior (or describes him as doing something), relates his movements, adornment, and position in the shrubbery to the thinking (the practical reasoning) that
248 / Bruce Aune supposedly explains it. If the explanation is correct, his thinking must be very unusual, involving what are today very eccentric beliefs and purposes. With the exception of Jones, people never hunt vampires. The principal reason for this is that, at least today, people don't believe in such things. If they did believe in them, they would probably still not hunt them because, according to traditional lore, they are too dangerous to deal with. Jones clearly doesn't believe the traditional lore. He thinks vampires do exist, but the stories he has heard about them have become confused in his mind with scraps of other stories and anecdotes, leaving him with the beliefs, among others, that vampires are attracted by the smell of garlic and, though dangerous, are nevertheless vulnerable to high caliber bullets, particularly when made of brass. These beliefs-together with background purposes, such as that of protecting his family, and other beliefs about, say, the sort of care his family needs-are brought together in the reasoning leading Jones to behave as he does. It is, in fact, only because of this reasoning that certain steps Jones takes and certain peculiarities of his procedure (for example, his wearing a necklace of garlic) are aspects of a complex action, or activity, reasonably ascribed to him by the predicate "hunting vampires." A thesis that distinguishes my view here from alternatives such as Anscombe's and Dennett's is that people's purposive behavior is rightly described and explained by reference to a certain line of reasoning only if they actually engaged in that rea~oning.~ Engaging in a certain line of reasoning requires two things: first, that the,people actually have the relevant beliefs and intentions and, second, that the structure of their reasoning actually conforms to the indicated pattern. We may, as Dennett says,4 occasionally succeed in predicting people's behavior by attributing standard beliefs and ordinary rationality to them, but such success is, to a degree, fortuitous. A psychology professor I once knew used to define a rationalization as a pseudo-explanation of behavior in which good reasons rather than the right reasons are given. His definition was intended to be amusing, which it is; but I believe it is conceptually penetrating. I also believe that, in a proper explanatory account, the right reasons must be arranged in the right way. Some philosophers used to argue that reasons can't be causes because reasons are "logically related" to consequent decisions or their behavioral outcomes. Davidson forcefully exposed the
Action, Inference, Belief, and Intention / 249 underlying fallacy here,5 but another point needs to be madenamely, that a given person's reasons may be fallaciously related to his or her conclusions and that a decision or its behavioral outcome may be psychologically explainable, in appropriate detail, only by reference to the agent's fallacious reasoning. Many philosophers are convinced that the errors of traditional philosophy can be avoided only if one studies formal logic and learns to think rigorously. Obviously, they believe that those who lack appropriate training have the propensity to make mistakes, to reason invalidly. When (or if) people act on such propensities, their reasons are illogically related to their conclusions, and their ensuing behavior may also be "illogically connected" to their reasons. A propensity to reason in a certain way (according to this or that formal pattern) is, as Pierce put it, a "habitw6;and since the exercise of a habit is always, broadly speaking, under the control of some causal mechanism, the traditional supposition that action based on reasons cannot be based on causes errs for a further reason (perhaps a deeper reason) than the one Davidson exposed. When 1 first sketched the elements of this computational theory, I was concerned to criticize Ryle's account of believing as (essentially) a disposition to b e h a ~ eI. introduced ~ a very unusual belief (one to the effect that one may regain one's youth by drinking from the Ucayali river in Peru) because I wanted to emphasize the very indirect relation that actually holds between belief and behavior. Subsequent reflection on the peculiarities of practical reasoning has deepened my understanding of thought and action, leading me to a better conception of the relation between desire and action and, eventually, to a more refined conception of the relation between intention and belief. I want to expand on these various points in the sections to follow. 2. Belief a s a Mental Disposition If, instead of worrying about semantical issues, we ask ourselves how behavior can be explained by reference to the agent's belief, we shall be attracted to an analysis or conception that places believing securely in the causal order. An observation to begin with is that one can believe something for years without bringing the subject of ones belief consciously to mind-without consciously
250 / Bruce Aune "entertaining" it, as philosophers used to say. This observation suggests that believing is a disposition of some kind-in traditional philosophical language, a power or propensity. Since no overt behavior is required of a belief per se (a familiar point by now8) the relevant disposition is not that of behaving overtly in some way. What sort of disposition can it be, then? The answer, I believe, is "a mental disposition-a disposition exercised in reasoning." I have claimed that purposive behavior is explainable by the agent's practical reasoning and that belief plays a fundamental role in such reasoning. My view is that belief, as a psychological state, is best understood in relation to this role. What sort of reasoning-role is distinctive of a given belief? The answer I gave in 1963, when I first wrote on the subject, was, in part, "a certain premissory role." My claim was that to have a belief is to have a disposition "to affirm something to oneself and to use the proposition affirmed as a premiss when reasoning, practically or theoretically, about a wide variety of interrelated subject^."^ My thought was that if I believe that snow is white, I am apt to premiss that snow is white when I reason about snow, when I am concerned with white things, and the like. As for "affirming something to myself," I wished to make room for the "judgments," the mental expressions of belief emphasized by traditional philosophers, that need not belong to some line of thought or reasoning. I am not really sure, today, that disconnected mental judgments actually occur; but if they do, they would not seem to differ, per se, from the mental "assertions" involved in premissing and concluding. To keep my terminology as simple as possible, I shall describe the occurrent act characteristic of believing as premissing,1° but I shall allow that premissing may occur, for all I know, in "reasoning" involving a single thought or assertion. In conceiving of believing that p as a propensity to premiss that p, I don't mean to deny that people who believe that snow is white have other, related mental dispositions. I think it is obvious, in fact, that people who have this belief usually have the disposition to avow (at least to themselves) that they have this belief. But this latter disposition is best associated with a different belief-the autobiographical belief that one has the first belief. People who have the autobiographical belief normally have the former belief, and for this reason the avowal that one has a certain belief is taken to be good evidence that one actually has it. But two beliefs seem to be
Action, Inference, Belief, and Intention / 251 involved here, not one; and the thought or premissing plausibly distinctive of the one is different from the thought or premissing distinctive of the other. I speak here of what is "plausibly distinctive" of a belief because I do not think that the word "belief" (or a related word in another language) is used according to such clear and determinate rules that one can speak of some single, determinate "concept" of belief by which an account, or definition, can be shown to be correct. I also do not think (or find any plausibility in the idea) that an essence of belief can be pinned down by a set of "metaphysical necessities" known by some kind of a priori intuition. As I see it, a philosophical account of belief (the sort of thing I am concerned with here) can be nothing more than what Carnap called an "explication" or "rational reconstr~ction."~~ When we construct such explications, we take existing usage and common knowledge into account, but we inevitably proceed beyond both in drawing distinctions relevant to philosophical concerns. In the remarks to follow, I shall take special pains to lay bare the considerations on which I base my account of the elements of reasoned behavior. I hope thereby to exhibit the purposes a philosophical "theory" in this area can, in my opinion, legitimately fulfill. J. S. Mill once remarked that in many complex subjects "three fourths of the arguments for every disputed opinion consist in dispelling the appearances which favor some opinion different from it."12 Understood in the appropriate spirit, my claim thus far is that believing that p (for an arbitrary p) is usefully understood as, essentially, a disposition or propensity to premiss that p. One obvious objection to this claim concerns the conditions under which the alleged disposition is activated. In the clearer cases in which one identifies a disposition (water-solubility, for example) one can specify the conditions under which the appropriate response will occur. But one cannot do this in the case of believing. Who can say when a person who believes that snow is white will actually premiss such a thing? I confess that I cannot do so, at least in a useful, general way. What should I conclude from this inability? Maybe only that "disposition," as it is currently used (or overused) in philosophy, is not really appropriate to belief-maybe that belief is better described by a vaguer expression such as Ryle's "proneness" to do something. I wouldn't insist that any standard philosophical term is fully appropriate here. My thought, really, is that belief is well described
252 / Bruce Aune somewhat indefinitely as an iffy mental state of premissing a certain thing if Q, where the conditions Q are, as we understand them, significantly indefinite. Jones' tendency (proneness, propensity, disposition) to premiss that p may be governed by precisely formulable laws; but if they exist, no one knows what they are. We can say that Jones will premiss that p when he thinks p is relevant, but this, though true, merely relates one propensity to another. Better to admit right off that we are thinking of a kind of response that Jones is apt to exhibit under a variety of conditions that we cannot identify in advance. What we think of this way is an iffy condition (or state) the activating circumstances for which we conceive of somewhat indefinitely. For lack of a better word, I shall describe the state as a propensity or disposition. The conception of belief that p as a propensity to premiss that p may seem objectionable for other reasons. For one thing, general beliefs may enter into reasoning as what Peirce called "guiding principles," not as explicit premisses.13 Thus, people who believe that As are Bs may have the propensity to infer that a specified or unspecified individual i is B from the premiss that i is A. The inference here "accords with" the premiss or guiding principle that As are Bs; it does not explicitly include this premiss. Considering the ubiquity of this sort of enthymemic inference, one might suppose that a belief is best regarded as a propensity or disposition either to premiss that p (for the relevant p) or to reason in accordance with this premiss. Although the point is perhaps minor, I think the supposition would be ill-advised. The key consideration is brought to mind by the question: How does one identify the guiding principle of an inference? It is not satisfactory to answer that the principle is the one on which the validity (or acceptability) of the inference depends, for there is no such unique principle. 'Bi' follows from 'Ai' given '(x)(Ax > Bx)', but it also follows given countless other premisses of which 'Ai > Bi', '-Ai V Bi', or even 'Bi" itself are particularly simple instances. If, in responding to this contention, one says that the guiding principle of an inference is the one the reasoner will volunteer in justification of it, one has in effect identified a premiss that we might accept as distinctive of (or necessary and sufficient for) the belief in the first place. If we do accept this premiss as distinctive, we can say that, owing to what Quine has called the "transitivity of conditioning,"14 people who have the propensity to use this premiss characteristically have a derivative propensity to reason merely in accordance with
Action, Inference, Belief, and Intention / 253 it. They will tend to simplify inference patterns involving that premiss into patterns that, they believe, are justified by reference to it. Another plausible objection to my identification of belief with a propensity to premiss arises from the fact a person who assumes or supposes that p in a certain context is disposed to premiss that p in that context but does not thereby believe that p. This objection is interesting because it focuses attention on two important matters relevant to the premissing associated with belief. The first concerns the formal character of such premissing: it is categorical in a way that assuming or supposing is not. A person who assumes that p in a context C uses p as a premiss from which to draw conclusions, but the conclusions thus drawn are not seen by him or her (at least in the absence of other reasoning) as assertible outside of C, except hypothetically: their premissory force is limited to C. The premissing associated with assuming or supposing is therefore significantly restricted in scope. Having said this, I must now admit that the premissing associated with believing is, in its own way, somewhat limited in scope, for a logically sophisticated reasoner would not consider premissing a believed p in every context whatever-least of all in the context of a logical or mathematical proof. If we consider the matter generally, we see that the scope of our premissing on contingent matters is intermediate between that of our mere assuming or supposing and that of our premissing on necessary (or modal) matters. Conditions of relevance aside, we will premiss what we take to be necessarily true in most contexts15; we will premiss what we take to be contingently true only in contexts pertaining to the actual world; and we will premiss "mere assumptions" only in highly restricted contexts. Considerations of relevance impose severe restrictions along a different dimension. When I said, earlier, that believing is fundamentally iffy and that one who believes that p will premiss that p only under certain (indefinitely conceived) conditions, I was calling attention to conditions that, for different reasoners, provide the occasion for premissing that p. Obviously, these occasions are partly determined by reasoner-relative conditions of relevance. The latter conditions clearly differ from (though they are related to) reasoner-relative conditions of truth or plausibility. The remarks I have made thus far expose some of the minutia that is required by a philosophically adequate treatment of belief, but there is a good deal more to be said.16 One obvious point concerns the
254 / Bruce Aune strength of a belief. Surely, it will be said, we do not simply believe, disbelieve, or have no opinion; we believe strongly, weakly, and the like. Peirce rightly emphasized that a belief is something on which a believer is prepared to act: she acts tentatively and cautiously, if her belief is weak; confidently if her belief is strong. The strength of a belief might, in fact, be measured by the relevant risks the believer will take or the relevant odds the believer will accept. The force of this point is, I think, undeniable: a careful, fine-grained explication of belief certainly ought to represent it as admitting of degrees. The question is, "How is this best accomplished?" A number of possibilities could be discussed here, but the strategy I prefer is to represent the strength of a person's belief by attaching a qualification to its content. When we express our beliefs in ordinary life, we do just this. Suppose that I strongly belief that snow is white. If I wished to indicate the strength of my belief, I could express it emphatically by saying that snow is unquestionably white. If, on the other hand, I believe with only moderate confidence that, say, the Celtics will win their next game, I could indicate this by saying "The Celtics will probably win their next game" or, if the person to whom I am speaking is knowledgeable about subjective probabilities, "The probability that the Celtics will win their next game is 0.6." Note that I am viewing these utterances as expressions of belief, not as selfascriptions of belief, which would have the form of "I believe that ...." I express my belief by producing an utterance that is a counterpart of a premiss I would use if that belief were to enter my reasoning. My proposal, then, is that we think of a firm belief as a propensity to premiss something with a firm (emphatic, weakly qualified) content, and of a weak belief as a propensity to premiss something with highly qualified or tentative content. Reflection will show that this proposal is tantamount to treating a weak belief that p as an unqualified belief that maybe p. An advantage of it is that with minimal tinkering it allows us to represent belief-premisses by subjective probability statements and relate practical reasoning to decision-theoretic inferences. As a philosophical subject, action theory belongs to metaphysics-of which ontology is a principal part. Thus far I have spoken freely of premisses and premissing, but how, ontologically speaking, are they related? In my view premissing is the more fundamental of the two. An act of premissing may be performed overtly, when one thinks out loud, or purely "mentally," in for0 interno. Semantically speaking,
Action, Inference, Belief, and Intention / 255 both overt and purely "mental" acts of premissing are episodicacts of tokening. It is useful to remind ourselves that the sentences we can see and read are records of complex assertive or (at least) tokening acts, which occur sequentially in time. (When I say "Snow is white," I utter "white" after I utter "Snow": when my utterance of "white" occurs, my utterance of "Snow" is over and done with, a thing of the past.) We think of, or at least characterize, mental premissing on the model of overt premissing, for we describe the token realized sequentially over time in mental premissing as having a subject (or subject matter), a predicate, a logical form, and the likefeatures we describe most confidently and comfortably in observable paradigms. Philosophers may argue about whether the chicken of overt utterance comes before, after, or concurrently with the egg of silent thinking; but there is no doubt that when we proceed to describe Tom as merely thinking what Mary dares to say, we are applyjng the same semantic apparatus to something silent and "mental" and to something at least largely overt: two materially different tokens are getting a common interpretation.''
3. Desire, Intention, and Action Though beliefs and desires are standardly taken to explain actions, desires (as we ordinarily speak of them, at least) are related to action only in a fairly indirect way, via the intentions that they sometimes occasion. To see this, we have only to recall some very obvious points about desire. One sober fact is that we cannot have many of the things we desire most, and we know it. Also, we are sometimes disgusted or embarrassed by other things we desire, pursuing them (as Socrates emphasized18) only in dreams. Thus we make no attempt to act on many of our desires. We are, of course, prone to daydream about, even yearn for, the salient objects of some desires; and if we think they are attainable, we consider how and where we may get them, with what likelihood of success, at what cost, and so forth. If we are passionate and impulsive, we may neglect the costs, exaggerate the likelihoods-in short, act irrationally and impulsively. But we always do some relevant thinking, however cursory. When we make up our mind what to do, we form an intention; and it is ultimately an intention, together with beliefs, that moves us to act (in the volitional sense of "act").
256 / Bruce Aune Although many of our intentions seem to arise as unwittingly as our desires do, some are formed as the result of a rational process. I have argued that the relevant process may be of two principal kinds.lg The first involves a choice between alternatives; the second is (or appears to be) the derivation of an intention from preexisting intentions and beliefs. These two sorts of processes may appear to differ significantly, one freer, less logically determined, than the other. A little thought shows both to be equally immersed in the causal order. In what might be called the standard case in which we are concerned to make a choice, we have some end or other and more than one available means of achieving it. Our task is to choose a preferred means, if there is one. I say "preferred means" because alternatives are assessed by beliefs and preferences. Relevant beliefs will concern (a) the likelihood of achieving the end by each available alternative and @) pertinent properties of each alternative such as the difficulty of carrying it out, its further effects, their likelihood, and so on. Relevant preferences will concern all aspects of the alternatives-their intrinsic attractiveness, their consistency with rules and policies, their consequences, and the like. In real life cases the agent cannot possibly succeed in taking explicit account of all the pertinent considerations, for they extend indefinitely into the future. At some point the agent simply chooses. The motivating preference will reflect some assessment of the alternatives, but the assessment (which is inevitably limited) will determine the preference and, indirectly, the choice only in the sense that one leads to the other. The choice-the concluding thought "So, I'll do A instead of B, C, or D"-is not, in particular, deduced from some available premiss. In an important respect the choice is a "logically free" mental act (even though it may be the result, the causal outcome, of a rational process). In decision making that does not in1:olve a choice, the agent forms an intention to do something (and so decides to do it) as the result of an inference in which the intention is (as it were) deduced from practical premisses. Consider the simple practical inference:
I will do A if P or Q. P.
So I will do A.
The inference here is clearly deductive: the conclusion is deduced
Action, Inference, Belief, and Intention / 2 57 from the preceding premisses. Though I shall attempt to justify the point only in the next section, it is reasonable to say that the first premiss and the conclusion are expressions of intentions. If a person actually draws this inference (if the sequence of formulas corresponds to distinguishable steps in someone's reasoning) he or she forms an intention from premisses representing another intention and a belief. Of course, the belief and the relevant intentions (that is, the psychological states) are not themselves the stated premisses or conclusions. The belief, I have said, is a mental disposition, and the same is true, I shall argue, of intentions. What happens, if the inference above is properly made, is that the agent, in concluding "So I will do A," thereby forms the intention of doing A. The relation between the intention and the conclusion drawn is the same, I should say, as the relation between the belief that may be formed when one draws an indicative conclusion from premisses one accepts and that conclusion itself. A philosopher who thinks of decision-making as an inferential process of the kind just considered might wish to insist that decisions are produced by reasons, not causes. As I have already observed, this would be a mistake. But there is something almost inherently misleading about the idea of forming an intention by reasoning. We reason, and (it is natural to say) we thereby form decisions, the latter being properly described as intentions formed as the result of reasoning. But though it is natural, even "correct," to say that we form an intention as the result of reasoning, it is truer to the facts to say that the reasoning forms an intention for us. The point I have in mind is easiest seen in connection with the reasoning in which we form beliefs. Though people sometimes speak of choosing to believe this or that, believing is not under our immediate voluntary control. I sometimes prove this to students by offering five dollars to anyone who actually can form the belief (make him- or herself believe for a minute) that the moon is made of green cheese. I can, on the other hand, make students believe this or that by showing them objects, giving them arguments, and the like. Of course, some people are more skeptical or more gullible than others; so the same demonstrations or the same arguments will not produce the same beliefs in everyone. Nevertheless, the beliefs my students form in these cases are formed independently of their will. I hold a frog in my hand, and those who have a good view of the contents of my hand will begin to believe that I am holding a frog. Similarly,
258 / Bruce Aune if I construct a simple, valid deductive argument with premisses they believe, they will generally begin to believe the conclusion as soon as they are convinced the argument is valid. I say "generally" because the perception that the argument is valid will sometimes make them doubtful of one of the premisses. The same holds true, basically, for the process of forming an intention as the result of reasoning. When one is presented with an argument one regards as logically compelling (perhaps deductively valid) having as a conclusion a formula the acceptance of which involves an intention to do something A, then if one fully accepts the premisses, one will accept the conclusion and thereby form (automatically, as it were) the intention of doing A. In this case as in the case of the beliefs formed by hearing the argument or seeing the frog in my hand, nature changes our mental state for us: we make no free choice. It is obvious that our training and life experiences have a lot to do with these natural responses to arguments, examples, and objects of perception. Suitable training in formal logic can often make very complex deductive arguments belief- or intentionproducing. Being duped can make us very hesitant to form firm beliefs from very clear perceptions of ostensible frogs (particularly those held in someone's hands). It is time to be more specific about intentions. What are they, or how should we think of them? And how, specifically, are they represented in thought? To get a handle on these questions, it is helpful to relate them to what I have said about beliefs. Perhaps the most elementary thing to say about intentions is that, like beliefs, they are dispositions or propensities: we can have them when we are not thinking about what we intend, but we can bring them to mind-even act on them-when the time is ripe. I have said that beliefs are best understood as propensities to premiss things. Are intentions like this too? I think yes: they must be (at last in part) propensities to premiss, for they enter into reasoning only in this way. How do these two sorts of premissing (belief-premissing and intention-premissing) differ? Superficially, this way: When we merely believe, we are prone to premiss something indicatively, and we represent what we premiss by indicative statements such as "Snow is white." When we intend, on the other hand, we premiss something "volitionally," or in a volitional mode. Such premissing is represented in literary English, at least, by sentences beginning "I (we) will ..." and "You (he, they)
Action, Inference, Belief, and Intention / 259 shall...".20 Such sentences contrast with the corresponding indicatives beginning "I (we) shall ..." and "You (he, they) will ...." To get a more fine-grained understanding of the relation between beliefs and intentions, it is important to attend to the logic of volitional reasoning, that is, practical reasoning involving volitional statements. 4. The Logic of Volitional Expressions What I propose to call the logic of volitional expressions is a proper part of what is generally known as practical reasoning. This latter subject may be traced back to Aristotle, for it was he who first drew the distinction between practical and nonpractical (or "theoretical") reasoning. As it is now understood, practical reasoning clearly includes assertoric reasoning of all sorts, deductive, probabilistic, and inductive; it does so because all sorts of reasoning may enter into a line'of practical reasoning (without, of course, being distinctive of it). The question I want to consider here is "What, if anything, is logically special about the part of practical reasoning that involves volitional expressions?" In considering this question I shall be ignoring those derivative forms of practical reasoning that relate "I will ..." to "I ought ..." or "x is better than y", and also those forms (which I mentioned earlier) that deal with choices between alternatives and that explicitly take account of personal preferences and subjective probabilities. One answer often given to the question deserves immediate notice: Volitional reasoning is distinctive in containing formulas (premisses or conclusions) that lack truth values.21 This answer is generally based on the view that volitional statements express what might be called "objectual intentions," these being quasi-propositional entities supposedly referred to by singular expressions such as "what Tom intended." These alleged entities are quasi-propositional because they are comparable to the propositions (or objectual beliefs) supposedly expressed by indicative statements and referred to by singular expressions such as "what Mary believed." Although objectual beliefs (sometimes described as the "contents" of belief-states) are held to be true or false, objectual intentions are said to take some other semantic values. This opinion is usually defended by the observation that, whereas we specify objectual beliefs by propositional clauses to which "is true" or "is false" may properly be added (as in "It is
260 / Bruce Aune true that snow is white"), nothing similar holds for objectual intentions. If we answer the question "What did Mary intend?", we may offer a verbal expression such as "to run for president," the implied subject of which is "Mary." Yet it makes no sense to add "is true" or "is false" to an expression of this kind. In many contemporary discussions it is taken for granted that "belief is a relation to a proposition"; and those who accept this view might naturally agree that an intention is a relation to a similar object, a quasi-proposition such as the one allegedly referred to by "Mary to run for ~ r e s i d e n t . "The ~ ~ view that belief is a relation to a proposition is based, as I implied above, on a semantic analysis or interpretation of belief-statements such as "Tom believes that snow is white" and on the interpretation of relative clauses such as "what Tom believes." I had nothing to say about such statements and clauses when I outlined my reasons for viewing belief as a propensity to premiss something, and I had nothing to say about the corresponding statements and clauses when I outlined my reasons for viewing intentions as propensities to premiss related things. It is important to say something about these matters now. The first thing I want to say is that my account is actually compatible with the slogan that belief is a relation to a proposition. I have been concerned with belief as a psychological state; I have attempted to explain how I think this state is best understood, given the way it is described in everyday life. In explaining this, I was specially concerned to indicate how believing enters into (or qualifies) a subject's reasoning and thereby affects his or her subsequent behavior. My view was that believing is a disposition to premiss something, premissing being an act of tokening. Abstractly considered, such a disposition is describable as a relation to some token or other; and if one believes that every token is related to a type and that the type corresponding to a sentential token is a proposition, one could add that a person who is belief-related to a token is belief-related* to a proposition, where a belief-relation* is a compound relation, the relative product of two simpler relations.23 As it happens, I do not believe in propositions as irreducible objects, but my view on this matter is not really crucial for the claims I wish to make in this paper about beliefs and intentions as psychological states. The second point I want to make is that, even if expressions such as "Mary to be president" do denote entities, the fact that those
Action, Inference, Belief, and Intention / 2 6 1 entities are not plausibly describable as true or false has no obvious implications for the logic of such statements as "I will wear a tie," which enter into the sort of reasoning that concerns me here. I will concede, as an added point, that it sounds odd to speak of intentions as true or false but not odd to speak of beliefs as true or false. But the relevance of this point to the logic of volitional statements is equally remote. If we focus our attention squarely on the question at issue herenamely, what semantic values are appropriate to volitional thoughts and statements such as "I will shave a little laterv-the natural answer would appear to be "truth and falsity." This answer is natural because ordinary speakers of English cannot reliably distinguish "I will..."from "I shall...," and few would see a problem in describing the latter as true or false. It is true that "shall" and "will" are etymologically distinct, "shall" deriving from the Anglo-Saxon sceal, meaning "oblige" or "compel," and "will" from the Anglo-Saxon willan, meaning "will" or "intend."24 But these verbs are now very close in meaning; in fact, as auxiliaries in the context "I shall (will)..." they are often described (according to a discerning usage now becoming extinct) as different moods of the future "to be": "I shall ..." is the declarative future and "I will ..." the purposive future.25 These different moods of what is now a common verb may carry different connotations or implications; but it is very natural to say that the sentences in which they occur are true or false. A natural inclination is not, of course, a decisive consideration, particularly in logic. But it ought to make us very scrupulous about finding an argument for a contrary opinion that we are expected to take seriously. Finding no compelling arguments for a contrary opinion, I accept the position that in a regimented dialect where volitional statements are scrupulously distinguished from merely indicative ones, the auxiliary verbs in the contexts "I shall/will do A" are different moods of "to be" in the future tense. So understood, "I will ..." differs from the corresponding "I shall..." not in truth-value but in the attitude associated with it. In this respect, "will" and "shall" are comparable to "and" and "but" or "if" and "when": they are different in total meaning, but (as it were) tokens of the same logical (or formal) type. The relations between them would be logically perspicuous if "I will do A" were invariably written as "I shall do A in red ink-the color showing the attitude involved, the word "shall" showing the assertion made. A similar observation is applicable to appropriate "shal1"s and
'
262 / Bruce Aune "wil1"s occurring (with volitional connotations reversed) in secondand third-person discourse and in the sentential prefix, "It will/shall be that ...." There is one fundamental problem with this strategy that I wish to discuss. In a regimented dialect where the distinctions I have been discussing are carefully preserved (but the red-ink transformation is not practiced), the following arguments would count as valid: (I) I will do A.
(11) I shall do A.
So, I shall do A.
So, I will do A.
The question is, "Are these arguments really acceptable? Should they count as logically valid, comparable to "P but Q; so, P and Q" and "P and Q; so P but Q"? I believe the answer is a clear yes, but before developing an argument for my belief, I want to remind the reader of what is at issue here. According to the semantical values I am attributing to volitional statements and their corresponding pure indicatives, the arguments in question are clearly valid. The question at issue is whether my choice of semantic values for these statements is satisfactory. To make a plausible case for my answer, it is sufficient to show that the practical inferences that come out valid (or invalid) according to my semantical treatment are ones that we judge, on reflection, to be valid (or invalid), and vice versa. The arguments I shall give should be seen in this light. Consider the conversation: "That tree will be cut down by tomorrow." "How do you know?" "It is my tree, and I will cut it down later today." The claim to know here is supported by an argument, which ought to be considered valid. It is: The tree is mine & I will cut it down later today. Therefore, the tree will be cut down by tomorrow. My semantical treatment of volitional statements allows us to prove that it is valid. The proof is as follows: From the premiss we conclude "I will cut it (the tree) down later today," which clearly entails "I will cut it down by tomorrow." From this we draw the purely indicative conclusion, "I shall cut it down by tomorrow." But this clearly implies
Action, Inference, Belief, and Intention / 263 the weaker passive assertion, "It will be cut down by tomorrow." Since the key inference here conforms to the pattern (I) above, we have plausible support for that pattern. What about pattern (II)? The case just described can be adapted to this pattern, too, if we reason byreductio. To employ this indirect form of reasoning, we have to form the negation of the volitional statement "I will do A." In the vernacular the explicit negation of "I will do A" is "I won't do A," a statement that is equally volitional. Pattern (11) is valid just in case "I won't do A and "I shall do A" are inconsistent. But "I won't do A" is merely a contracted form of "I will not do A"; and if pattern (I) is valid, "I will not do A" must imply "I shall not do A," which means that "I will not do A" and "-(I shall not do A)" are inconsistent. But surely "-(I shall not do A)" is equivalent to "I shall do A." Therefore, "I won't do A" is inconsistent with "I shall do A," and pattern (11) must be valid. If ','I will do A" and "I shall do A" are both future-tensed assertions about what will be, the former differing from the latter in having an extra expressive (or "emotive") property, then the validity of pattern (I) would be too obvious to require explicit defense. The converse is not equally obvious if, as I say, "I will do A" does more than the merely indicative "I shall do A." But the "more" that is done by the volitional form is purely expressive (or emotive) and not pertinent to the inference. It is vital to see that statements commonly do more than make (or convey) assertions. Consider the form "P even if Q." What is the logical form of this? The answer, I should say, is "P & (Q > P)," which is logically equivalent to "P." Given this fact about logical form, "I will go to the party" logically entails "I will go to the party even if Mary is there." But the latter (as normally used) expresses an attitude that the former does not express-a negative attitude toward being where Mary is. This fact is irrelevant to logic, however; it casts no doubt on the inference. The same is true, if I am right, of inferences conforming to pattern (11). Since I have shown, in a recent paper, that my approach gives the right results for a representative sample of practical inferences and that it can also accommodate practical inferences of very unusual c o m p l e ~ i t y I, ~will ~ not deal with further examples here. On the other hand, in accordance with Mill's contention that defending a view often consists in dispelling appearances favoring an alternative, I want to say something about objections, or worries, of two further kinds.
264 / Bruce Aune The first has recently been expressed to me (in conversationz7)by Wilfrid Sellars; it is that if I were right, there really wouldn't be such a thing as genuine practical reasoning. The objection is mistaken. I am in no way denying that there is such a thing as practical reasoning; what I am denying is that practical reasoning involves a special logic, or that formulas (sentences, thoughts) peculiar to it have some special semantic value. To this I might add that tokenings peculiar to volitional inferences, by virtue of the attitudes associated with them, move us to think and do things that we are not moved to do by tokenings of other sorts. In this respect they are particularly significant, practically. This sort of significance is in no way aided or abetted by a special logic. The second objection I want to mention is to the effect that "I shall do A" can't entail "I will do A" because believing one will do something does not entail intending to do it. The objection is confused. Although the state of believing that one will do something A is expressed by (in the sense, roughly, of giving rise to) utterances of "I shall do A" (if one habitually speaks the relevant English dialect) and the state of intending to do A is expressed (or represented) by utterances of "I will do A" (if the same condition holds), the fact that utterances of the first sort imply corresponding utterances of the other sort means only if the first sort have a positive semantic value (or are true), the second (if they are produced) also have such a value. It does not follow that if a person has the relevant belief, he or she has the relevant intention-nor does it follow that if one produces an utterance of the first kind, one will produce an utterance of the second kind. It doesn't even follow that those who think that they shall do A (an indicative thought) realize that they are logically entitled to think they will do A (a volitional thought). It might be objected that if, as I seem to be claiming, practical reasoning does not change beliefs or form intentions, it cannot have the importance philosophers have always ascribed to it. Again the objection is confused. I have said that implications (both ways) between shall- and will-formulas do not require or insure that people who have a certain belief (or intention) must also have a certain intention (or belief). I have not said that reasoning does not produce beliefs and intentions, or change them. It does. But reasoning is a psychological process, not a logical relationship between formulas. If we are logically sophisticated and fastidious, our reasoning will conform to logical patterns in the sense, mainly, of not exemplifying
Action, Inference, Belief, and Intention / 265 invalid ones. (We can't form the conclusions logically entailed by our premisses because every set of premisses entails infinitely many conclusions.) As I mentioned earlier, anyone who reasons is guided by habits, and these habits (together with current interests and aims) provide the rational mechanism by which conclusions are selected. Those who actually affirm a volitional premiss in their reasoning will thereby be expressing or forming an appropriate intention, but the fact that such a premiss-type is implied by some formula they do affirm does not insure that a token of that type is ever produced. A final point about practical inference. I originally arrived at my view about the appropriate semantical values for volitional statements and their corresponding indicatives by developing a mistaken hypothesis. The hypothesis was that volitional statements have the semantic values of what I described vaguely as intentions. I reasoned that intentions are realized or not; so volitional statements should have these values. Yet "I will do A" will be realized just in case the indicative "I shall do A" is true. This close, systematic connection between the values realized and true renders one value logically otiose: everything will work out fine, logically, if volitional statements are identified with their indicative counterparts and said to be true or false. Though I soon learned that my initial hypothesis here was erroneous, the connection I perceived between truth and realization was helpful for seeing the point of practical reasoning as I now understand it. When we engage in practical reasoning, we typically have some intention or other and seek some means of realizing it. Since an intention is realized just in case the corresponding belief is true, conclusions about how to realize intentions are logically equivalent to conclusions about what will make corresponding beliefs (which are about the future)true. The latter sorts of conclusions are logically equivalent to what will make certain formulas (statements, thoughts) true. The aims, in purely logical respects, of practical reasoning are thus met by reasoning involving nothing but indicatives-that is, ordinary assertoric logic. To say this is to say that no special logic is needed for practical reasoning. This is not to say, again, that practical reasoning does not require volitional formulas. It does, but not because these formulas have special values. These formulas are needed because of their connection with motivating attitudes. Such attitudes move me to reason (in ways governed by my conscious habits) and to form new attitudes, which eventually move me to act
266 / Bruce Aune (to intervene in the physical world). The habits often select volitional conclusions when their purely indicative counterparts could be drawn with equal validity.
5. Intention and Belief My avowed purpose in discussing the logic of volitional expressions was to obtain a more fine-grained understanding of the relation between belief and intention. I am now in a position to discuss this latter relation. My initial accounts of belief and intention (as psychological states) were these: Believing that one will do A is a propensity to premiss, or think, "I shall do A" (or something translatable as such); and intending to do A is a propensity to premiss "I will do A" (or something so translatable). Now, I have claimed that these premisses are very closely related: they are logically equivalent, differing only in the "volitional attitude" associated with the latter. Insofar as one uses a metaphysically nonperspeciuous language, beliefs and intentions are distinct states: Since "I shall do A" and "I will do A" are different formulas, the propensity to token one formula is different from the propensity to token the other. If, on the other hand, we reasoned in a philosophically perspicuous notation in which "I will ..." formulas differed from "I shall..." formulas only in some nonlexical property of the former (corresponding to red ink), then every intention would be a belief, for it would be a propensity to premiss "I shall...", that is, to premiss something lexically indicative or assertoric. Although intentions would be beliefs if this latter possibility were realized, they would be very special beliefs because of the volitional feature they would possess. This volitional feature is not, I repeat, a "propositional" attitude; in fact, it is best described as a property of a tokening-a causal property. It is describable this way because volitional thoughts (apart from the propositional content they share with corresponding pure indicatives) have the "motive force" that Hume attributed to our passions.28This "force" is shown in two ways. First, if a volitional thought "I will now do A" concerns an action within an agent's immediate voluntary control, the occurrence of the thought will normally result in the agent doing A (unless some obstacle intervenes). Second, if a volitional thought is concerned with
Action, Inference, Belief, and Intention / 267
,
an action or end not within an agent's immediate voluntary control, the occurrence of the thought will normally dispose the agent to consider (or at least seek) means of bringing it about. In this latter case the effect of the thought is a propensity to scan the environment, to think up alternatives, and to engage in practical reasoning (where habits of drawing volitional conclusions from appropriate premisses are activated). I cannot think, at least offhand, of anything else that I would regard as distinctive of volitional thoughts qua volitional, and since the features I have mentioned are fundamentally causal, I say that the volitional feature of a thought is a purely causal one. I mentioned earlier that, given the notation we actually use in formulating our practical premisses, volitional thoughts are not purely indicative ones and that, as a result, intentions do not count as beliefs. The distinction is purely terminological, however, and we could certainly stipulate that intentions shall constitute a special class of beliefs, beliefs being propensities to premiss in one or the other of two modes. Since intentions would invariably involve premissing in the volitional mode, many beliefs about the future would not amount to intentions. Some philosophers might object to this degree of overlap between intentions and beliefs on the ground that one can fully intend to do things one knows one will not do-such as continue to hang on to a rope dangling over a precipice. But the existing usage of "belief" and "desire" is insufficiently determinate to support such a counterexample. One who rejects the example need only insist that, as he or she uses the words, the relevant belief or intention must be redescribed. One can know one will probably let go; or one may fully intend to hang on as long as one can, or to do one's best to hang on. Someone who wished to accept the counterexample can, of course, offer a rival explication of belief or intention. There is no fact of the matter that can render one approach right and the other wrong. As this last paragraph makes clear, I would not dream of contending that the distinctions I have been drawing-and, therefore, the relation between belief and intention that I have described-are precisely there in ordinary or extraordinary speech and that I have unearthed them by mere analysis, thus exposing something previously hidden in the phenomena. As I in effect mentioned earlier, I have not attempted a 60's style analysis of ordinary language. I have also not attempted to analyze some "concepts" (perhaps eternal and Platonic) supposedly reflected in English and other languages. Nor have I tried
268 / Bruce Aune to expose key "metaphysical necessities" to which our psychological states conform; that would amount to a priori psychology, which I could not take seriously. Finally, to conclude this litany of denials, I have not attempted to contribute to natural language semantics, specifying "truth conditions" for specimen psychological sentences.29 What I have attempted to do is to "make philosophical sense" of a familiar practice of describing and explaining human behavior by reference to reasoning in which beliefs and intentions are principally involved. Making sense of this practice inevitably involves "rationally reconstructing" it- developing requisite implications and drawing distinctions not clearly drawn already. But I have not pursued an ideal of explicitness and precision that is far removed from the idiom I am concerned with. This last point is particularly important, if my remarks about beliefs and intentions are to be adequately appreciated. I want to conclude, therefore, with a final remark about it. I have claimed that beliefs and intentions are best understood as propensities or somewhat indefinitely conceived dispositions-more exactly, as propensities to premiss things under indefinitely conceived conditions. So understood, beliefs and intentions are usefully describable as incompletely specified conditional properties. Many philosophers would prefer not to speak of such properties because doing so is speaking vaguely and imprecisely-something they object to.30 But if one is dealing with a nontechnical locution (or "concept") that is vague or imprecise in certain respects, a sharp and determinate reconstruction may move one too far away from one's intended target and thus fail to illuminate the right subject. One who insists that all clarified (or explicated) locutions be rendered determinate and precise reminds me of Socrates' fancied reply to the man who objects to painting the eyes of statues anything but purple, the most beautiful of colors: "Don't think we must make the eyes so beautiful that they no longer appear to be eyes at all".31So with the locutions "believes" and "intends": if they are explicated according to some ideal standard of determinacy and precision, they will no longer appear to describe the believing and intending we speak of in everyday life.
Action, Inference, Belief, a n d Intention / 269
Notes 1. My first effort, composed in 1963, appeared in Aune (1967), pp. 21 1-218; a more ambitious statement occurred in Aune (1977); and some recent thoughts can be found in Aune (1986a) and Aune (1989). 2. According to Jerry Fodor, who has perhaps done the most to make the notion of a computational process fashionable in philosophy, such processes are both symbolic and "formal," the latter expression indicating that computations are carried out in accordance with (roughly speaking) the syntax of the ingredient representations. Though I do not share Fodor's views regarding "the language of thought," I can agree that a person's reasoning or computations may be viewed in this way by that person. The so called contents of the ingredient representations (or premisses and conclusions) are, as I shall explain, interpretations that reflect the perspective of an interpreter. Such interpretations of (or assignments of content to) a person's thoughts are vitally important for the explanation of her behavior because those who require the explanation and those who give it have an other-person, interpreter's perspective. For Fodor's description of a computational process, see Fodor (1981), pp. 226f. 3. See Anscombe (1957) pp. 57-88, and Dennett (1987), ch. 2 andpassim. 4. Ibid. 5. Davidson (1980), pp. 3-20. 6. See, e.g., Peirce (1934), 5.367. 7. Aune (1967), pp. 213ff. 8. See ibid. 9. Ibid. 10. The suggestion that I use the verb "to premiss" here was made to me years ago by Roger Rosenkrantz. I hereby express my thanks. 11. See Carnap (1949), p. 330. 12. John Stuart Mi11 (1978), p. 35. 13. Peirce (1934), 5. 365. 14. Quine (1960), p. 12. 15. An exception is the context of a proof. In addition to being necessarily true, the premisses of a strict proof must be simple or elementary-or demonstrated consequences of such premisses. 16. What about so called de r e beliefs? In my view there are no such beliefs. One may speak of de r e belief sentences, an example of which may be "Jones believes of Smith that he is wise"; but "de re" modifies "sentence," not "belief." We often use such sentences when we wish to ascribe a belief to a person but cannot specify, even in indirect discourse, the premiss that she is disposed to use. 17. The fact that in saying "Tom believes that he is wise" we are interpreting his belief from our point of view should remove the sting from the arguments that Tyler Burge, in Burge (1979), offers in support of his "anti-individualism." I may be said to ascribe "content" to Tom's belief when I use the clause "that he is wise," but the content I thus ascribe is not an intrinsic feature of his belief; it is my interpretation of it from
270 / Bruce Aune my point of view. I discuss the semantic interpretation of beliefs and relate it to Quine's thesis of translational indeterminacy in Aune (1986b). 18. See Plato's Republic, 571c. 19. See Aune (1977), ch. 4. 20. The auxiliaries "shall" ("should"), "will" ("would") are discussed fully in Follet et a1 (1966), pp. 369-391. 21. This has been emphasized by H.N. Castenada, among others; see Castaneda (1975). 22. Castaneda takes this line; see ibid. 23. This is not, of course, the only way of relating a belief to a proposition. For a critical discussion of some other possibilities, see Schiffer (1987), Chs. 2 & 3. 24. See Opdycke (1965), p. 111. 25. lbid, p. 112. 26. See Aune (1986a). 27. Sellars has also expressed this view in an unpublished paper, "On Reasoning About Values." 28. See David Hume (1888), Bk. 11, section 111, pp. 413-422. 29. 1 have criticized these various means of supporting a metaphysical theory in Aune (1989). 30. Some philosophers would say that dispositions are actually structural properties-the physical properties causally responsible for the relevant behavior. I don't doubt that many dispositions are causally explainable by reference to such structural properties, but I know that not all dispositions are so explainable. The obvious exceptions are the dispositions of fundamental objects such as photons. Such objects have no micro structure and, therefore, no structural properties of the kinds in question. Since we must acknowledge irreducible dispositions at the fundamental level (whatever that level happens to be) there can be no a priori objection in acknowledging them at higher levels either. My claim is that beliefs, as we ordinarily speak of them, are best understood as mental dispositions, or incompletely specified conditional properties. Other conceptions are, in principle, possible, but none, to my knowledge, has actually been developed. If a "structural" conception were developed, it would no doubt have the feature of the painted eyes that Socrates noted in the Republic. See note 31. 31. Republic, 420d.
References Anscombe, G.E.M.: 1957, Intention, Blackwell, Oxford. Aune, B.: 1967, Knowledge, Mind, and Nature, Random House, New York (reprinted by Ridgeview Publishing Company, Atascadero, CA). Aune, B.: 1977, Reason and Action, Reidel, Dordrecht, Holland. Aune, B.: 1986a, "Formal Logic and Practical Reasoning," Theory and Decision, 20 (1986) 301-320.
Action, Inference, Belief, and Intention / 27 1 Aune, B.: 1986b. "Other Minds After Twenty Years," Midwest Studies in Philosophy, X , pp. 559-574. Aune, B.: 1989, "Action and Ontology," Philosophical Studies, 54 (1988) 195-213. Burge, T.: 1979, "lndividualism and the Mental," Midwest Studies in Philosophy, IV: 73-12 1. Carnap, R.: 1949, "Two Concepts of Probability," in Herbert Feigl and Wilfrid Sellars, eds., Readings in Philosophical Analysis, Ridgeview Publishing Company, Atascadero, CA. Castaneda, H.N.: 1975, Thinking and Doing, Dordrecht, Holland. Davidson, D.: 1980, "Actions, Reasons, and Causes," in Davidson, Essays on Actions and Events, Oxford Univ. Press, Oxford, England. Dennett, D.: 1987, The Intentional Stance, Cambridge, Mass. Fodor, J.: 1981, "Methodological Solipsism Considered as a Research Strategy in Cognitive Psychology," Representations, Cambridge, Mass. Follet, W. et al: 1966, Modern American Usage,eds. Hill & Wang, New York. Hume, D.: 1888, Treatise of Human Nature, ed. L. A. Selby-Bigge, Oxford University Press, Oxford. Mill, J.S.: 1978, On Liberty, ed. Elizabeth Rapaport, Hackett, Indianapolis. Opdyke, J.: 1965, Harper's English Grammar, Harper & Row, New York. Peirce, C.S.: 1934: Collected Papers of Charles Sanders Peirce, eds. Charles Hartshorne and Paul Weiss, Harvard Univ. Press, Cambridge, Mass. Quine, W.V.O.: 1960, Word and Object, M.I.T. Press, Cambridge, Mass. Schiffer,S.: 1987, Remnants of Meaning, Bradford, Cambridge, Mass.