On Thought and Feeling Bruce Aune The Philosophical Quarterly, Vol. 13, No. 50. (Jan., 1963), pp. 1-12. Stable URL: http://links.jstor.org/sici?sici=0031-8094%28196301%2913%3A50%3C1%3AOTAF%3E2.0.CO%3B2-N The Philosophical Quarterly is currently published by The Philosophical Quarterly.
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/philquar.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 07:22:02 2007
THE
PHILOSOPHICAL
QUARTERLY
VOL. 13 No. 50
JAhWARY 1963
ON THOUGHT AND FEELING I n the early years of this century empiricists commonly thought that while each person has introspective knowledge of his own states of mind, the thoughts and feelings of others have for him the status of inferred entities. This view has been under attack now for nearly two decades, and although it may appear crude and misguided to contemporary philosophers whose fear of ghosts in machines has led them to embrace a form of philosophical behaviourism, I for one still think that it is essentially sound. I n what follows I shall try to show that it actually errs less in what it asserts than in what i t seems to suggest, and that i t can be formulated in such a way that it eludes most of the criticism that has recently been levelled against it. I n order to put my argument in a clear perspective, I shall begin with a short commentary on twentieth-century empiricism.
I The empiricism of the 'twenties and 'thirties was notoriously egocentric. Nearly every empiricist philosopher had his own special epistemology, and the foundation of that epistemology was said to be the hard data of the man's private experience. When he used words like ' thought ', ' feeling ', or ' senseimpression ' with full confidence, he was typically referring to things he experienced, to things he knew directly. Sometimes, however, he cautiously used these words in relation to other people. But because he could not experience their data, he often insisted that he really had no way of knowing whether other people had private data a t all-or whether, if they did have such data, they attached the same significance to psychological words as he did. For all he knew, they might associate quite different sorts of data with these words, e.g., they might associate with the word ' thought ' what
2
BRUCE AUNE
he associated with the word ' image '. As a result of these doubts, he was sometimes tempted to say that, as he used them, first- and third-person psychological sentences were systematically different in propositional content. This temptation seemed reasonable because words like ' pain ' typically played two fundamentally different rales in his epistemological constructions. On the one hand, in sentences like " I am in pain " the word ' pain ' seemed to play the r61e of a n observation term par excellence ; on the other hand, in sentences like " He is in pain " this word apparently functioned as a theoretical or explanatory term--that is, it was supposed to refer to something, to be sure, but what it referred to was something he could never possibly observe. Now the idea that most psychological words play this dual r61e has, of course, a number of important difficulties. For one thing, it seems intuitively obvious that what I say of myself, when I say that I am in pain, is exactly the same as what I say of you, when I say that you are in pain, and vice versa. For when someone asserts that I am in pain, by saying of me " He is in pain ", I am able either to assent to this assertion by saying " Yes, I am in pain ", or to refute it by saying " No, I am not in pain ". That is, though he uses a third-person sentence in speaking of me, my reply, cast in the form of a first-person sentence, constitutes either an affirmation or denial of what he says. Yet, if the word ' pain ' did not mean the same in the two sentences, it could scarcely be maintained that we were either agreeing or disagreeing in what we said-in fact, there would be little sense in saying that we were even communicating.
I1 The thought that we might not really be communicating when we speak to others of our sense-impressions is very alarming, and a gregarious philosopher holding the above view might attempt to obviate this difficulty in a number of ways. I n the first place, if he has little faith in the muchcriticized analogical argument for the existence and nature of other minds, he might be tempted to reason as follows. " When Jones and I converse about the experiences of Smith, we both treat his experiences as inferred entities or states, i.e., we treat them as states of his person that result from a certain kind of stimulation and in turn dispose him to behave in a particular way. Because we speak of his experiences only in relation to his behaviour, which we can both observe, we have no difficulty understanding each other. We both know, however, that when we speak of each other's experiences we do the same thing : we treat each other's experiences as non-observable, but inferrable, events, processes, or states. When, consequently, I speak of my own experiences, I must, if I want to be sure that other people understand me, speak of myself as they would speak of me ; that is, I must use psychological words as referring to non-observable, or theoretical, states of myself. This means that I must base such statements on the sort of behavioural evidence I rely on when I ascribe psychological states to other
3
ON THOUGHT AND PEELING
people, e.g., I must be sure that I myself am disposed to jump, cry, swear, and so on, when I say of myself that I am in pain. Of course, in using ' pain ' in this way I do not convey exactly what I feel : to do this I would have to use the word in its private, ostensive sense, which I could never expect others to understand. But in order to assure myself that others can understand me, I must be content with conveying just the " structure ", not the " content ", of my feeling. There are, after all, certain limits to what can be said in an intersubjective language ; and though I cannot hope to describe the true quale of my experiences in such a language, I can perhaps express them in music or in poetry." As I have presented it, there are obviously many difficulties with this argument. But it is nevertheless quite subtle in a number of respects. Notice, to begin with, that it does not rest on the projectability of privately defined predicates. I n explicitly denying the universal applicability of such predicates, it assumes that the psychological terms capable of use in interpersonal communication have meaning because they are related to observable behaviour, not because they are tied to private experience by the tenuous bonds of individual ostension. Because of this, it suggests that certain psychological terms in common use (e.g. ' feeling ' or ' thought ') actually have the character of typical explanatory terms : they are implicitly defined with reference to a certain set of behavioural " indicators ", and they represent " inner " states or processes which are produced by external stimuli and which in turn produce a distinctive sort of behaviour or observable physical changee1 Now this part of the argument seems quite reasonable. For (i) philosophical theories aside, it is pretty clear that a pain in one's foot is neither a pattern of actual and possible behaviour nor a mere disposition, set, or propensity to behave in a certain way. Nor is it observable by other people : they must, in the end, infer from one's grimaces, etc., that one has such a pain-and in this they can always be wrong. Further, (ii) ' pain " does function, a t least in part, as an explanatory term : it is not entirely descriptive. Thus, when one is asked " Why is Jones limping ? ", it is perfectly proper-and often illuminating-to reply, " Because he has a very painful leg ". (He could be limping because of a muscular or skeletal disorder.) Or, again, if one is asked, " Why didn't Smith scream or jump when his tooth was broken off? ", it is perfectly reasonable to answer, " Because he never feels pain on that side of his mouth ; the nerves there are destroyed ". Finally, (iii) though the explanatory force of the word ' pain ' (and similar words) may be explained by a number of different theories, the above argument has the important merit of insisting that the connection between feelings, sensations, etc., and observables is not just accidental ; that is, it does justice to the intuitively sound idea that a sensation that does not a t least make one want to jump, cry, or swear, could not possibly be painful. 1Cf. H. Peigl, " The ' Mental ' and the 'Physical' osophy of Science, 11, pp. 443f.
", Minnesota
Studies in the Phil-
4
BRUCE ARNE
This is something that is very hard to explain in terms of the more usual empiricist theory, according to which the meaning of sensation-terms in actual use is entirely determined, for each speaker of the language, by a private ostensive procedure which does not relate the term to any sort of observable behaviour .
I11 I have already remarked that this way of avoiding the difficulties of the original formulation of the empiricist theory has a number of shortcomings, but before pointing them out it is useful to consider another, more familiar move that the empiricist might naturally make. Brushing aside the sceptic's questions, he might forcefully maintain that psychological words always have a univocal meaning, that there is no difference between their use in reference to one's private data and their use in reference to the experience of others. He might argue as follows. " The univocal meaning of psychological words, for each one of us, is established by ostension, by association with certain elements of our own private experience. We of course begin applying them to others as soon as we have established their meaning, but this practice is justifiable on the grounds of analogy. We all have similar bodies, and our behaviour is essentially the same when we are stimulated in a similar way. Because of these highly significant similarities, it is perfectly reasonable to think that others experience what we experience when they behave as we behave, and that we all make the same, or a t least very similar, associations with the same word. Interpersonal communication is therefore possible : for although each person's awareness is limited to the domain of his own private experience, we are all similar enough in verifiable respects to warrant the assumption that we experience similar things in similar circumstances and make essentially the same associations with the same words." This analogical procedure, though it was a favourite move for many empiricists, has unfortunately been demolished by the efforts of R7ittgenstein, Malcolm, Strawson, and other^.^ I cannot review the entire body of their destructive arguments here, but one of them, perhaps the most important, can be stated with relative brevity. This argument concerns the impossibility of establishing the meaning of sensation-words by private ostension, and Professor Malcolm has given it an especially lucid statement. According to Professor Malcolm, the empiricist who defends private ostension is anxious to maintain that one inwardly picks out something as thinking or pain and thereafter identifies it whenever it presents itself in the soul. But the question to be pressed is, Does one make correct identifications ? The proponent of these " private " identifications has nothing to say here. He feels sure that he identifies correctly the occurrence in his soul ; but feeling =VideWittgenstein, Philosophical Investigations (London, 19fi!), sects. 256f; P. F. Strawson, Individuals (London, 1959), Ch. I V ; and N. Malcolm, Knowledge of Other Minds ", Journ. Phil., LV, 23:960-978,
5
ON THOUGHT AND FEELING
sure is no guarantee of being right. Indeed he has no idea of what being right could mean ! He does not know how to distinguish between actually making correct identifications and being under the impression that he does (p. 976). This argument might be put more abstractly, and in slightly different terms, as follows. If the mere fact that one apprehended a psychological item as were a sufficient condition of knowing it to be then either (a) one could have knowledge that one could never possibly justify or distinguish from mere opinion, or (b) one has available some independent criterion by which the infallibility of these apprehensions can be ascertained. Since alternative (a) is clearly unsatisfactory, because knowledge is vastly different from baseless opinion, an intelligible account of (b) must be forthcoming if the theory is to be taken seriously. Unfortunately, however, no such account is possible, given the special assumptions on which the theory is based. For according to these assumptions there is no way of ruling out the possibility of highly anomalous apprehensions occurring in conjunction with a given psychic item, e.g. the apprehension This is pain or I am in pain occurring in conjunction with a sensation that makes one want to smile, sing, or jump with joy. Besides, since these apprehensions are, ex hypothesi, necessarily immune from correction, no room is left for the notion of a mistaken apprehension ; that is, no matter what apprehension might occur in connection with a given psychic item, it must, according to the theory, be counted as veridical. But if a mistaken apprehension is something that could not conceivably exist, the expression ' veridical apprehension ' can mean no more than ' apprehension-simpliciter ' : the addition of ' veridical ', since i t cannot serve to distinguish any possible class of apprehensions from any other, actually adds nothing to the idea of an apprehension. Indeed, the addition of this word is not only vacuous but positively misleading, for it suggests these alleged apprehensions are to be grouped with what we normally call observations and identifications, occurrences that Ryle has suggestively called " achievements ". Some philosophers have tried to rehabilitate this theory by arguing that only those apprehensions that " cohere " with the bulk of one's current apprehensions are to be accepted as veridical ; the others, regardless of the strength of one's temptation to assent to them, are to be regarded as simply erroneous. Unfortunately, however, this move not only abandons a cardinal tenet of the theory, viz. that one has non-inferential knowledge of one's private data, but it begs the question as well-for it assumes that one can be correct in recognizing coherence. That is, if one's ability to recognize the character of these private data is in question, one's ability to recognize that two or more of them " cohere" is also in question. For this reason the resort to coherence is a resort t o question-begging. $J
IV Now if the very notion of private ostension is incoherent, it is obvious that the analogical argument, since it is based on this notion, is incoherent too. This means that the attempt to save the univocity of psychological
6
BRUCE AUNE
words by resorting to this argument fails. But what about the other attempt, the one that construed psychological words in actual use as referring to inferred entities or processes ? This move seems to fare pretty well, though of course its rider, that psychological words also have a private use, is neatly trimmed off by Malcolm's argument. Aside from this, however, just what is wrong with it ? Only one thing, it seems to me : it cannot account for the peculiar character of a person's statements about his own psychological states. One does not need to observe one's behaviour in order to tell, e.g., -when one is in pain : this is something one can tell right off. Indeed, talk of observation in this connection seems wildly beside the point. There is, however, a way of avoiding this difficulty. Professor Malcolm has suggested that we can tell when we have a certain sensation, e.g. pain, because we have gradually been trained, since early childhood, to respond with words instead of tears when pains are felt. These verbal responses are not the result of " internal " observations and identifications, though ; they are simply conditioned responses. When we are properly trained and not inclined to dissemble, they spring from us as naturally and spontaneously as our cries, screams, and tears. This suggestion, though it does not tell the whole story about f i s t person statements, has considerable merit. I t explains, first, how the utterance of a first-person psychological sentence by another person can have an importance for us, though not as an identification, for in light of the above considerations it will have the same importance as the natural behaviour, e.g, crying, which we must rely on in the case of children and others incapable of articulate speech. And, second, it helps break the hold on us of the question, " How does one know when to say ' My leg hurts ' ? " ; for if our verbal utterances are indeed conditioned responses, it is easy to see that they may occur as spontaneously and as naturally as crying, laughing, or r ~ n n i n g . ~ Nevertheless, in spite of the obvious merit of this view in accounting for the spontaneity of &st-person psychological utterances, it seems to suggest (a) that we are really like highly-trained parrots that, as a result of a certain stimulation, will squawk out noises without any exercise of intelligence, and (b) that we are essentially incapable of introspection and silent deliberation. Both of these suggestions are utterly repugnant ; but fortunately they are not actually implied by Malcolm's argument. For one thing, the difference between us and parrots is not that our words are always preceded by thought or deliberation, but that they are self-consciously produced : we know what we are saying, parrots do not. For another thing, Malcolm's attack on private identifications and observations does not actually demonstrate the impossibility of introspection ; it only demonstrates the impossibility of introspection as it is described by a particular philosophical theory. This particular theory of introspection falls apart because it can provide no justification for the idea that one's thoughts or judgments aMalcolrn,p. 978.
ON THOUGHT mD
FEELING
7
correctly represent, or describe, one's current feelings, etc. Fortunately there
are other theories of introspection which do not involve this difficulty, and one of them is quite consistent with Malcolm's general position.
v I n order to appreciate the philosophical legitimacy of introspection, we should recall that much of our thinking is done out loud. When we are speaking naturally to someone, and are not concerned to weigh our words before we utter them, there are not two thought-processes going on, an interior thinking and an external saying, but only one : a coherent, thoughtful saying. When we think out loud in this way, we are of course speaking in a certain " frame of mind ", as Ryle calls it ; but this frame of mind is not a collection or bundle of thoughts, some of which duplicate the words we are uttering ; it is rather a peculiar state of intellectual readiness-readiness to speak, to answer questions, to re-express what we have just said, and so on. That is, like many of the mental states Ryle has described, a frame of mind is essentially dispositional, not episodic ; and thus when we think out loud, our speech reflects our state of mind ; it does not in any way duplicate it. Now in many instances when a person utters a first-person psychological sentence, e.g. when he says " I am in pain ", he is thinking out loud just as much as he is conveying information about himself; in particular, he is neither putting previously formulated thoughts into words nor blindly responding like a parrot. Since he is thinking as well as speaking, and since he is thinking and speaking about how he feels, there can be little objection, it seems to me, against calling this " introspection ". The fact that the man's words are not premeditated, not preceded by a silent process of thought, surely does not count against this ; for thinking has to start somewhere, and it is purely an accidental matter that this thinking is done out loud, rather than quietly to himself. I n addition, since, as Malcolm shows, a person's statements about himself are publicly confirmable, there are means by which this kind of introspection can be checked for accuracy-and this implies that introspection, a t least in its " overt " form, is quite above suspicion. VI But what about covert introspection, or introspection that is not done out loud ? Can we really make sense of such a thing ? I should say " Yes ". For if we are willing to speak of inferred entities and processes in the physical sciences, i.e., entities and processes that can neither be observed nor explicitly defined in terms of ob~ervables,~ we should be quite willing to speak of thoughts, or sayings in foro interno. Moreover, a willingness to speak this way does not require us to embrace ghostly Cartesian objects or to maintain that thoughts are " really " neurophysiological states. The ever4Cf. C. G. Hempel, " A Theoretician's Dilemma ", Minn. Studies in Phil. of Science, 11, pp. 37-98.
8
BRUCE AUNE
vexing Mind-Body problem is, after all, far from clear ; and we should not feel constrained to accept one of its famous " solutions " as a consequence of rejecting behaviourism. I have called thoughts " sayings in foro interno " because about the only thing you can say about them is that they are of or about certain thingse5 Recent discussion has shown that mental imagery has no logical connection with thought or judgment : if a man has a certain image whenever he thinks of a particular thing, this is just an accidental fact about him, and there is no absurdity in the idea that he could have the thought without the image, or the image without the thought. What is essential to a man's thought, on the other hand, is its reference or object ; and for this reason we can describe thoughts largely in semantic terms. Thoughts and judgments are, of course, occurrences-they take place in time-but nevertheless they are describable as x's that are about something or other (e.g. the thought of one's mother) or that affirm something to be the case (e.g. the thought that it will rain). Kant once argued that judgments always have logical form and that their constituent thoughts play a quasi-grammatical r81e. His way of putting the matter is no doubt a little misleading, but it does emphasize an undeniable similarity between thinking and speaking. For covert thinking does the work of speaking, a t least when one is speaking to oneself ; and the r81e of thoughts and judgments in silent deliberation is analogous to that of words and statements in overt speech. I n thinking, as in speaking, one draws inferences, makes identifications, and anticipates the course of future events. These activities would be incomprehensible unless thinking were describable as something like overt speech. But if we allow the existence of covert episodes of thinking and judging, can we make sense of the idea that they correctly represent, describe, or otherwise refer to their objects ?-that is, can we avoid the sting of Malcolm's attack on private identifications and descriptions ? Here, again, I should say " Yes ", For if we recall that a person must learn to keep his thoughts to himself, to avoid blurting out everything that comes to mind, it becomes natural to think of thoughts as covert occurrences, often barely suppressed responses, that take the place of overt speech. And just as overt speech about one's current state of mind is the product of a well-entrenched set of habits, so is the inner speech that replaces audible utterances the product of a set of habits-in fact the same set of habits. And because these habits of response can be checked for reliability, for correctness, Malcolm's objection to internal identifications does not apply here. For under the present interpretation of thought and judgment, there is no ontological barrier between thinking and speaking ; indeed, our ability to say certain things constitutes one of the most important " criteria " for the thoughts we happen to have. This is why, if a person cannot tell us what he is thinking 5My conception of thoughts, in fact the substance of this entire paper, owes a lot to the work of Wilfrid Sellars. For a discussion of thoughts and an explanation of the sense in which they are inner episodes, see his "Empiricism and the Philosophy of Mind ", Minn. Studies i n the Phil. of Science, I , pp. 253-329.
ON THOUGHT AXD FEELING
9
about, we have very good reason to think that he really has nothing definite in mind. A thought is, by definition as it were, the sort of thing that can always be expressed in words. VII The idea that we have thoughts, which are neither overt speech episodes nor mere dispositions to behave, has historically seemed too obvious to require justification ; yet in recent years, because of a fondness for the Verifiability Criterion and its helpmate, Occam's Razor, many philosophers have come to feel that there is really no need to postulate such mysterious entities. " If we can infer future behaviour from the behaviour a man exhibits now, then " their argument goes, " there is no need to risk metaphysical difficulties by postulating intervening processes ". I t seems clear to me, however, that this attitude is seriously misguided-not only for a but also for the logical analysis of our programme of scientific psy~hology,~ everyday concepts. For we do ordinarily speak of thoughts-and we speak of them as inner episodes, not as dispositions or patterns of observable behaviour. This means that an attempt to do away with thoughts as inner episodes is an attempt to reform, not analyze, our ordinary manner of speech. More important, though, is the fact that covert processes of thought are needed in order to explain certain kinds of human behaviour. Peter Geach, in Mental Acts (London, 1957), p. 8, illustrates this nicely. When Dr. Johnson did penance in Uttoxeter market-place, he may have begun standing around bareheaded ; yet merely by observing his behaviour one would not think that he was waiting for the threatened shower to fall, that his behaviour was really rain-expecting behaviour. I n order to understand what he was trying to do, one must know (a) that he wished to do penance, and (b) that he thought standing bareheaded in the rain would constitute an act of penance. That is, in other words, if one wants to understand the full significance of Dr. Johnson's behaviour on that occasion, one would not only have to know what he wanted to do, but also what line of thought led him to do what he did. If one tries to argue that Dr. Johnson was disposed to get wet because he was disposed to do penance, one is quickly stopped by the question, Why did a disposition to do penance bring about the disposition to stand hatless in the rain '1 Clearly, this question is to be answered in terms of the thinking Dr. Johnson must have done ; that is, the answer must mention the fact that he thought standing in the rain would constitute an act of penance. The need for postulating silent acts of thought can also be brought out by reference to problem-solving. Most of us can solve fairly complicated mathematical problems only by writing out the steps that lead to the solution, but some people can work them out wholly in their head. Yet if you ask one of these exceptional persons how he arrived a t a certain solution, he can generally tell you what his strategy was, what steps he took. I n 6Cf. Hempel and, especially, M. Scriven, " A Study of Radical Behaviourism Minn. Studies in Phil. of Science, I , pp. 88-130.
",
10
BRUCE AUNE
fact, in telling you this, he might say, " I remember thinking . . .". Unless we accept his statement a t face value and assume that he actually did think through a series of steps, that he did something analogous to what we do when we write out our steps on paper, we could never explain how he got his solution or why he tells us, " I remember thinking . . .". VIII At the beginning of this paper I expressed my opinion that the empiricists of the 'twenties and 'thirties were basically correct in insisting (a) that one can have introspective knowledge of one's current state of mind, and (b) that the thoughts and sensations of others have, for each person, the status of inferred episodes and states. I n what followed I tried to show that this view can be formulated in such a way that it eludes a good deal of contemporary criticism ; in particular, I worked hard to show that a place can be made for the notion of introspection, and that we should not feel compelled, in the light of today's flourishing behaviourism, to deny the " interior " and non-observable character of thoughts and feelings. So far, however, I have focused most of my attention on the concept of thought ; and there are philosophers, no doubt, who could accept what I had to say about thoughts, but who might nevertheless find it difficult to accord a similar status to feelings and sensations. Still, I find it hard to see how anyone could honestly deny that most feelings, e.g. headaches, are " inner " states. For they obviously are not just dispositions or patterns of observable behaviour. They have, of course, " external criteria ", as Wittgenstein has shown ; but this does not make them wholly dispositional or merely behavioural. Besides, the " external criteria " for a feeling cannot be understood as sufficient or necessary conditions ; they function, rather, as indicators-but indicators which provide logically impeccable evidence for the feeling in question.' I have already pointed out, in section 11,that the word ' pain ' does much of the work of a typical explanatory term of the theoretical sort (though it has, to be sure, the important peculiarity that it can also be used in introspective reports) ; and in view of the fact that pains, unlike moods, wants, or tendencies, are always assigned some sort of location, e.g. in one's leg, it seems that no purely dispositional, or behavioural, account of them can possibly succeed. I must, however, anticipate an interesting objection often advanced by philosophical behaviourists, an objection which brings out an undeniable oddity in my statement that the thoughts and feelings of others have the status of inferred processes and states. It might be expressed as follows. " If the pain in someone else's knee is something whose presence you must infer, how is it that in point of fact you don't normally make an inference when you see that a child is crying and holding his knee ? Usually, you simply see the child and, without any inference whatever, you know that 7 1 have discussed this point in detail in sections X-XIV of my " Feelings, Moods, and Introspection ", which will appear shortly in Mind.
ON TlIOUGHT AND BEELING
11
he is in pain and you do something to help or comfort him. Yet if you can tell that he is in pain without going through a process of inference, how can you maintain that his pain is less observable than his crying or his tears ? " Plausible as this objection may appear to be, it actually involves a confusion between two different senses of the word ' observe '. For while there is nothing wrong with saying that one can observe that a child is in pain-and that one can often observe this without making any conscious inferences-it does not follow that the child's pain is itself observable. Indeed, if one's ability to observe that p automatically conferred observability on what ' p ' describes, we would have to conclude that the man who observes that it will rain observes not only some threatening clouds but also a future rainstorm. But surely this would be nonsense. Observables, properly so called, are quite different from feelings, pains, and future rainstorms ; they are things that can be seen, felt, or heard. When we observe a hurt child, we observe a sobbing, trembling child ; we do not observe his feelings. True, we do not always have to think about, or consciously identify, his tears and cries in order to know how he feels ; but this does not mean that we therefore observe his feeling. What we strictly observe-and what we must recall if we want to justify our opinion about the child-is his behaviour. Of course, by the time we have become adults we have become so adept a t recognizing the feelings of others that conscious attention to their overt behaviour is often unnecessary. But our ability to do this, as Quine e.g. has s u g g e ~ t e d is , ~ easily explained by the " transitivity of conditioning " : our recognition may be directly elicited by changes in our sense organs (or nervous system), and the conscious step from He i s crying to He is in pain may well be skipped. Nevertheless, though we may in this way fail to heed the sobbing we hear, the tears we see, and the trembling we feel, we must recall and refer to these things when our opinion is challenged. And for this reason we may legitimately say that the feelings of others are inferred, not observed.
IX I can now summarize the main strands of my argument. Thoughts and feelings are inner episodes or inner states ; they are neither Cartesian objects nor mere dispositions to behave in a certain way. They are, nevertheless, properly ascribable on the basis of observable behaviour : they have, as Wittgenstein insisted, " external criteria ". I n spite of this, however, one can have introspective knowledge of one's own feelings and states of mind. This knowledge is not based on observation and it is not the result of an inference : it is immediate, or spontaneous. One can, of course, observe that other people have certain feelings, and sometimes one's observationsthat may appear to be wholly spontaneous, or not based on conscious inference. Nevertheless, the feelings of others are not, properly speaking, observable : statements about them can be justified only with reference to 8W. V. 0. Quine, Word and Object (New York, 1960), pp. 9-13.
12
BRUCE AUNE
their behavioural indicators, to the behaviour that counts as their " external criteria ". For this reason, together with certain facts about the logic of psychological explanation, both feelings and thoughts may legitimately be called " inferred states or processes ". And although words for feelings and thoughts may also be used in introspective reports, they still have a univocal meaning ; for even in introspective reports, when one is speaking of one's own experience, one is speaking of states or conditions which have " external criteria ".
BRUCEAUNE Oberlin College.