Content and Consciousness

by Daniel C. Dennett

  Now when one responds to a question about his reasons, with what sort of knowledge does he answer? If one is asked merely for the ideas that are passing through his mind, one can respond in a foolproof way, with only verbal slips to worry about. But when one is asked to give one’s reasons for an act, one is asked to give the reasons that ‘actually worked’, that led to or determined the act, and not just any plausible reasons that come to mind. There is a genuine ambiguity in the demands we make of people when we ask them to give their reasons. On the one hand we grant authority to the actor; if he says his reason for taking a drink is that it will calm his nerves, and we believe he is sincere, then that is the reason he took a drink. The psychoanalyst, however, may tell us that the real reason he took the drink was because he thinks drinking makes him masculine. What is the real reason? If the reason the psychoanalyst gives us is the real reason, this might often (or even always) be a reason of which the actor has no inkling, but if this is what we are asking for when we ask for reasons, why do we ask the actor? His best response, under these ‘rules’ of the language game, would be to say ‘I haven’t any idea; you had better ask my analyst.’ Or, if he is a fan of psychoanalysis, he may try to psychoanalyse himself, and say ‘Let’s see. I wouldn’t be at all surprised if my reason for taking the drink is that I have a death wish and am trying to drown myself.’ Surely this would be an inappropriate response to our question; we are not asking a person to psychoanalyse himself when we ask him for his reasons. On the other hand, we are not merely asking him to report his thoughts to us. Suppose I am asked why I took a drink and I think back and realize that I had just thought: ‘I am very upset and a drink would calm my nerves, so I’ll have a drink’, I may respond by announcing that I wanted to calm my nerves, and this may or may not be a sincere answer. If I know perfectly well that I am a compulsive drinker always looking for a good excuse for a medicinal nip, I would not believe I was reporting my real reasons in saying this. If I am not strong on self-knowledge, on the other hand, I may report this as my reason quite sincerely. The fact that the thought ran through my head may be undeniable; that it veridically represents the ratiocination that determined my behaviour is a matter of interpretation and fallible inference.

  If without stopping to think I pull a child away from a fire and am asked to give my reasons, I may say with a high degree of conviction, ‘because I saw he would soon be burned’, and in this case I am not relying on any remembered thoughts that may have run through my head – for none did – and my knowledge that this is the correct reason is based on my knowledge that I did not think ‘I’m going to kidnap that child’ or ‘let’s put baby in his crib for a while’. Lacking any evidence for exotic explanations, I infer (usually subconsciously) that I recognized the danger and then acted on this recognition – which is the obvious explanation.

  Of course, even when I rely on remembered thoughts in giving my answer, there is a mediation which gives rise to the possibility of error – just because I am remembering what I was aware1 of, not expressing what I am aware1 of. Any memory can be false, of course, and though I cannot be mistaken in thinking that I seem to remember being aware1 of a particular thought, I may be misremembering. When we ask a person to ‘think out loud’ while solving a problem, the protocol we get from him will thus be more reliable than the one we could get if after finding the solution he was asked to recall the steps he took, just because it eliminates one layer of inferential knowledge; it does not depend on the faithfulness of the reports of his memory. Thus in most cases of giving my reasons my report is doubly inferential: I infer that my memories of my conscious pondering are sound, and moreover that that pondering was not mere rationalizing.

  The question whether a reason is a bit of rationalizing or truly my reason for acting is the question whether I acted because of this reason, and this ‘because’ is a causal ‘because’, not a reason-giving ‘because’.9 I may have a reason for doing X, may do X, but not because I had the reason (see p. 39); when I do X because of the reason, it is not that my reason for doing X is that I have this reason, for that would lead to an infinite regress. The ordinary practice of asking for reasons is predicated on the assumption that our conscious reasoning is a reliable manifestation of the information processing that determines our actions, and psychoanalysis is predicated on the counterclaim that it is not. Our ordinary granting of authority to the actor’s reports is well supported because in so many instances it just is not to the point what one’s deepest source of direction is: when I ask why you are sawing the plank and you tell me you are making a table it is virtually inconceivable that you are mistaken and quite irrelevant that you may have deep and terrible reasons for making a table, unknown even to yourself. On the other hand, our willingness to grant authority occasionally to the psychoanalyst is well supported because we have seen many times that a person’s apparently sincere reports of reasons do not harmonize with his behaviour: for example, the proven success of advertising campaigns based on ‘sex appeal’ shows quite clearly that we do buy products for reasons other than the hard-headed practical reasons we sincerely avow.

  In some cases I may say to my interlocutor that I do not know for what reasons I did something, and in these cases there still may be reasons for what I did. In other cases I may reply that there was no reason for what I did, and I can be right or wrong about this. I may believe that what I did was sheer doodling, or pointless motion, but this motion might have some ‘deeper significance’, might have a reason. In some cases of bodily motion we know enough about the actual mechanisms of control to say with a high degree of certainty that there was no reason, as in the case of the reflex kick. In these cases we are apt to say that the behaviour was caused but had no reason. The same verdict is often reached in more speculative cases. Consider Anscombe’s example: ‘ “Why did you jump back suddenly like that?” “The leap and loud bark of that crocodile made me jump”.’10 What sort of knowledge do we have of this sort of ‘mental cause’? Not non-inferential knowledge. It is true that there is something I know about my jump that another person cannot (at this time) know, and so in a way I have ‘privileged access’ to such mental causes. I know that I saw (or seemed to see) the crocodile leap and bark, and I know (by ‘proprioceptive feedback’ from joints and muscles) that I jumped (or seemed to jump), and so far as I know nothing else entered into the situation. What I can know that another person cannot is what was missing from the experience. I know I did not think to myself, just before the crocodile barked, ‘I think I’ll just jump back for the fun of it’, and I know I am not afflicted with some malady that makes me jump every now and then. So I conclude (consciously or subconsciously) that it was the sight of the barking, leaping crocodile that made me jump, and this conclusion is fairly safe. But I do not have non-inferential or immediate knowledge of the cause of my jump, and of course it is only contingent that another person cannot know what I do about my jump. Neurologists might someday know just as well – in fact better – what caused my jump. I have no access, private or otherwise, to my cerebral processes, but only to my awareness and the succession of messages arriving there; having no other explanations of the jump, and having seen others jump when presented with sudden, strange sights, I infer (consciously or subconsciously) that the startling sight caused the jump.

  Suppose I am crying, and someone asks why. I say ‘because Smith just died’, and this is a cause, not a reason, for my crying, for I am not crying on purpose or deliberately. In making this report I am assuming, again, that there is a causal relation between learning the sad news and crying, since the occasion is similar to other occasions on which people have cried. The regularity with which the receipt of sad news is followed by crying suggests that there is a causal relation between the two, and neurologists may someday provide detailed confirmation of this hypothesis. But all that I, the crier, may know that another person may not, is that in this case nothing else of conceivable relevance, such as an onion or directions in the script: �
��cry here’, has entered into the case.

  The fact that such knowledge of causes and reasons is inferential is obscured when one looks only at the personal level account of what is going on. I, the person, do not make an inference (consciously), so it is tempting to say that I just know what these reasons or causes are, and then the case is easily confused with the cases of genuine experiential certainty, such as my infallible, non-inferential knowledge that I am in pain, which have a different evidential status altogether. Perhaps in cases of inferential knowledge we should not say that I make the inference, since it is made subconsciously, but it is made, and this is enough to give my knowledge of these reasons and causes a mediated evidential status.

  Before leaving reasons and causes I want to dispose of a common misunderstanding to the effect that where there are reasons there are no causes, and vice versa. Anscombe, for one, if I understand her, does not wish to speak of an action being caused if it occurred for a reason. She says:

  … how would one distinguish between cause and reason in such a case as having hung one’s hat on a peg because one’s host said ‘Hang up your hat on that peg’? … Roughly speaking – if one were forced to go on with the distinction – the more the action is described as a mere response, the more inclined one would be to the word ‘cause’; while the more it is described as a response to something as having a significance that is dwelt on by the agent in his account, or as a response surrounded with thoughts and questions, the more inclined one would be to use the word ‘reason’. But in very many cases the distinction would have no point.11

  There is, of course, causation in both cases – and reasoning in both cases. Reasoning, or its subconscious counterpart, must be going on even when one ‘unthinkingly’ hangs one’s hat on the peg; the behaviour is appropriate to the stimulation because it is mediated by organizations established by stored information – about manners, pegs, hats, and so forth. The ‘unthinking’ response is leagues beyond the Pavlovian conditioned response (people are not trained to hang their hats up on pegs upon hearing verbal cues), and leagues more beyond the knee-jerk, which is genetically wired in. How very strange it would be that a person should hang up his hat in response to a verbal cue, unless it were a swiftly reasoned response.

  Support for the view that what we do for a reason does not have a cause is often found in the claim that what we do for a reason we do intentionally, and part of what we mean when we say that an action is intentional is that it is not caused. This claim refers us to an ‘ordinary’ question and answer sequence:

  (1) Were you caused to do that (e.g., spill the coffee)?

  (2) No, I did it intentionally.

  I doubt that this is ordinary. I doubt that anyone would ever speak just this way, but even if it is granted that we do speak this way, what of it? How do I know I was not caused to spill the coffee? Have I non-inferential knowledge that I was not caused to do it? Would it not be better to say that I have inferential knowledge that at least certain sorts of causes were absent? That is, I know I did not feel anyone bump my arm, I know that I am not an epileptic, and I know moreover that I just had the malicious thought: ‘Let’s make a mess of Smith’s carpet.’ Question (1) asks if anything like a bump or a twitch or a startling sight caused me to spill the coffee and, as far as I know, nothing like that did cause me to do it. It would be absurd to suppose that when one asked (1) he intended to cover all physical and metaphysical eventualities with regard to causes, and that (2) is anything like a firm assertion of the absence of causes. Thus (1) is, plausibly, an ellipsis for

  (3) Did you do that because some external object or internal malfunction moved your body?

  and (2) is an ellipsis for

  (4) No cause of that sort operated – to the best of my knowledge. I did it intentionally (and I really have no idea what sort of causes if any that might involve).

  Thus not only is it the case that when I do something for a reason, what I do is caused, but what makes a reason my real reason for doing something is that the events of information processing which cause what I do have among them an event with the content of my real reason, whether or not I am aware1 of this content.

  9

  ACTIONS AND INTENTIONS

  XXI INTENTIONAL ACTIONS

  The concept of personal action is an essential adjunct to the concept of a person, for, as we have seen, it is only on the personal level that explanations proceed in terms of the needs, desires, intentions and beliefs of an actor in the environment. Beyond this, the concept of action plays a critical role in our notions of responsibility and punishment. It is well worth detailed elucidation, therefore, and all the more so because once again the traditional views of intentional action will be seen to founder on a failure to make clear the distinction between the personal and sub-personal levels of explanation. The first step is to characterize the class of intentional actions, and since this task has been brilliantly executed by Miss Anscombe in Intention, I can do no better than to give a précis of her analysis, making a few alterations along the way and then wedding the results to our emerging picture of awareness1.

  First, she points out, it is not bodily motions, but motions under particular descriptions that are intentional or unintentional. I may be sawing a plank, and it may be one of Smith’s oak planks, so that I am sawing a plank, sawing one of Smith’s planks, and sawing an oak plank. These are not, however, different motions; in doing all three at once I am not performing three separate feats of motion. The action of sawing the plank (the motions considered under that description) can be intentional, while the action of sawing one of Smith’s planks (the same motions under a different description) is not.

  A necessary condition for membership in the class of intentional actions is that the actor be aware of the action, i.e., aware of the motion under a particular description. If I am typing, and someone asks me ‘Why are you tapping out the rhythm of “Rule, Britannia”?’ and I reply ‘Oh, am I doing that? I was not aware of it’, I show (if I am truthful) that the action of tapping out the rhythm was unintentional. I am aware of the motions of typing, but not under that description, and so, not being aware of tapping out the rhythm, I cannot be doing it intentionally. Hence, intentional actions exhibit Intentionality, or, more circumspectly, intentional action ascriptions are Intentional contexts.

  Awareness, however, is not a sufficient condition. I may be aware that I am doing one thing in the course of doing something else, and yet not be doing the former intentionally. I may notice that I happen to be neatly not stepping on the cracks of the pavement, and yet not be intentionally missing them; they just happen to match my normal stride. I may, in fact, pay particular attention to this phenomenon to see how long I can keep to my normal stride before my foot lands on a crack, and in such a case I am still not missing the cracks intentionally, but only keeping to my normal stride intentionally – however much I want to keep on missing the cracks.

  Anscombe takes account of this by distinguishing intentional actions as members of a subclass of the class of actions of which one is aware: the class of one’s actions of which one is aware without observation. That is, one denies that an action is intentional if one says ‘I only observed that I was doing that’.1 When I say I only observed that I was doing something, I mean I saw with my eyes, or heard with my ears, or felt that I was doing something (e.g., inadvertently making scratches on the table with my pen), or even that I knew proprioceptively what I was doing: I knew my knee jerked without looking, because I had the kinaesthetic sensation of a jerking knee. If these are the only ways I know or am aware that I am doing X, then doing X is not intentional. In what other way, then, could I have knowledge or awareness of my action, so that it could be intentional? The question can be reformulated in the terms developed in previous chapters for explaining awareness: an action falls into the class of actions ‘known without observation’ if a signal with a content descriptive of the action crosses the awareness line, and this signal is neither a proprioceptive
signal from muscles or joints nor a signal mediately from the sense organs. Now could any signals fulfil these conditions?

  It was argued earlier that last-rank motor signals can be ascribed only uninteresting content, such as ‘contract, muscle! ’, but that the higher-level directing efferents would, if they could be pinned down at all by the investigator, be amenable to more interesting content ascriptions, viz., commands to perform actions. For example, ‘open the door!’ could be the content of a relatively high-level efferent event or state which controlled a number of different sub-routines all ‘designed’ to get the door open in one way or another. Then if such an efferent command were to send a signal also across the awareness line, one would be aware1 of its content; one would be aware1 that one was trying to do X, and one’s awareness1 would be of a particular description under which the resulting motions were to be subsumed. The sort of aware-ness1 one had of intentional actions would then differ from other sorts of awareness1 in having an efferent rather than afferent source. Then knowledge without observation can be construed as knowledge of efferent controls, not knowledge of afferent input. That is, one would be aware1 that one was opening the door (or trying to) and not via any afferent route, and hence one could say what he was doing. Anscombe says several things to support this characterization of our knowledge of intentional actions. This knowledge, earlier characterized only as non-observational, becomes ‘practical knowledge’, and is compared with the knowledge of a man ‘directing a project, like the erection of a building which he cannot see and does not get reports on, purely by giving orders.’2 The knowledge we have in these cases is knowledge of the orders we give, and our beliefs about the state of the building at any time will be correct, provided our orders are carried out. As Anscombe says, if they are not, ‘Theophrastus’ remark holds good: “the mistake is in the performance, not in the judgment”.’3