Belief and Inference
Another important manifestation of belief is inferring, drawing inferences from the proposition believed. (We may recall Ramsey's view that when we come to believe a proposition we add it to our stock or premisses.1) The Acting-as-if Analysis does make room for this belief-manifestation. Acting as if a proposition were true amounts to using it as a premiss in one's practical reasoning. But why only in one's practical reasoning? Propositions which we believe are used as premisses in inferences of all kinds, not merely in practical ones.
But if we wished to say that believing a proposition is (among other things) a tendency to draw inferences from it, our analysis of belief would apparently be circular, since inferring itself is one way of coming to believe. Perhaps we may put it this way: the belief that p has a tendency to spread itself or extend itself to other propositions which follow from p, whether with certainty or with probability.
This ‘extensibility’ of belief is a most important property of it. If a person claims to believe a proposition, but fails to believe even its most obvious consequences (e.g. shows no surprise at all when one of them is falsified) we are inclined to doubt whether he does believe the proposition. At 2 p.m. he says he believes that the Front Quadrangle of the College is square; but two minutes later, when he walks round it, he is not at all surprised to find that its North side is three yards longer than its East side. Clearly there is something paradoxical here which needs explanation. It might be that by the time he began to walk he had lost his belief that the Front Quadrangle is square. Or perhaps he believed that it was indeed square at 2 p.m., but also believed that college quadrangles are liable to change their shapes very quickly, as clouds do. Colleges are queer places, and all sorts of strange things are liable to happen there.
The most likely explanation, however, is that he was not in full possession of his wits at the time. He was tired, frightened, flustered or half-asleep, and therefore failed to notice that the observations he made during his walk ‘had any bearing on’ the proposition he believed. No one is in full possession of his wits all the time. It is also true that the appreciation of the logical relations between propositions varies greatly between different people, even when all of them are in full possession of whatever) wits they have.
But normally, when we believe a proposition, our belief does extend itself to at least some of the consequences of the proposition. We need beliefs because we need guidance not only in our actions but in our thoughts also. We are interested in the question ‘What am I to think?’ as well as in the question ‘What am I to do?’ To put it another way, when we believe a proposition p we are interested in the question ‘p, so what?’ And ‘so what?’ does not only mean ‘so what am I to do?’ but also ‘so what else am I entitled (or obliged) to believe?’
This tendency to draw inferences from the proposition believed does have something to do with the ‘cognitive feelings’ we have been discussing (surprise, doubt and confidence). What spreads itself or extends itself from the proposition believed to its consequences, or at least to its obvious consequences, is the whole complex disposition which we call ‘believing’, and this includes the disposition to have the feelings we have discussed. If the consequences are not only obvious, but also follow with certainty from the proposition believed, the belief-disposition remains intact, as it were, when it spreads or extends itself to these consequences. In that case, a person who merely suspects that p will merely suspect that q; a person who believes with considerable confidence that p (has the opinion that p) will have an equally strong or firm opinion that q; and a person who is completely sure that p will also be completely sure that q. Moreover, whatever degree of surprise he is disposed to feel if p should turn out to be false, he is disposed to feel the same degree of surpirse if q should turn out to be false.
But if we are aware that q is only a probable consequence of p, our degree of belief is lower with regard to q than it is with regard to p. Our belief does spread itself or extend itself to q, but in a diluted or weakened form. Let us suppose that we are in the Lake District. It is raining heavily and we are quite sure that the heavy rain will continue all day. If it does, the brook at the bottom of the garden will probably be impassable by this evening. This happens quite often, though not always, when there is heavy rain in these parts. Then, if we are reasonable and in possession of our wits, we shall opine, or perhaps only suspect, that the brook will be impassable by then. We shall not feel quite sure about it, though we do feel quite sure that the rain will continue all day. It is a possibility to be reckoned with and taken seriously, but we do not feel sure that it will actually happen. If it does not, we shall be somewhat surprised, but not astonished, as we shall be if the rain stops at lunch-time.
Let us now suppose that p itself is only something I suspected and that q is only a moderately probable consequence of it. I suspect, for example, that the cat has gone out for the night; and if he has, it is not unlikely that he is in the garden of the house across the road. Even here there is some spreading of my (weak) belief-attitude from p to q. But by the time it reaches q it is still more diluted, and shows itself only in my considering the question ‘whether q?’ and not in any belief about the answer. That is not much, but still it is something. It is some gain to have discovered at any rate one question which ‘arises’ from our suspecting that p. Even when the extension of belief occurs in this, its weakest and most diluted form, it is not to be despised.
If the case where q follows from p with certainty may be represented by saying p confers a probability of 1 upon q’, we may sum the matter up like this:—the manner in which our belief that p spreads itself or extends itself to another proposition q, which is a consequence of p, depends upon two factors taken together: (1) the degree of our belief that p (2) our estimate of the degree of probability which p confers upon q. If we do not or cannot estimate this degree of probability at all, our belief does not spread itself to q at all. If, or in so far as, we estimate it incorrectly, as we may, our belief does spread itself to q, but this extension of it to q is unreasonable, even though the belief that p may be perfectly reasonable.
If the Dispositional Analysis of belief is correct, we shall certainly have to say that this ‘spreading’ of belief from a proposition to its consequences is one of the most important ways in which such a disposition is occurrently manifested. This, I think, is what we have in mind when we speak of ‘relying’ on a proposition. When we believe a proposition we rely upon it. This is not quite a tautology, equivalent to ‘when we believe a proposition, we believe it.’ The word ‘rely’ makes explicit something which might have been overlooked. A proposition is relied upon when it is available to us as a premiss for inferences, whether theoretical or practical.
Our beliefs are like posts which we plant in the shifting sands of doubt and ignorance. They are fixed points or stable landmarks; and once they are there, we are able to make short journeys into the surrounding wastes, planting another post or two as we go. That is why the loss of a belief can be such a serious matter for us. We have lost something which we have been using to find our way about a wilderness. We are in a state of ‘bewilderment’.
This spreading of belief from one proposition to another may be experienced or lived through by the believing person. There is an experience called ‘inferring’ or ‘drawing a conclusion’. But sometimes we just find ourselves feeling confident of the conclusion, or feeling surprised when it is falsified, though we did not actually experience any process of inferring. The conclusion ‘drew itself’, as it were. We did not consciously draw it. For instance, if I believe that one of my colleagues went to New York the day before yesterday, I feel surprised when I meet him in Oxford this evening, though I did not consciously infer that he was unlikely to be back so soon.
Nevertheless, it is an important fact about the autonomous character of rational beings that we can, if we wish, inhibit or suspend this extension of belief from one proposition to another, until we are satisfied that q is indeed a consequence of p, and satisfied also about the strength of the logical connection between them. Does p, which we believe already, render q certain? If it does not, what degree of probability, if any, does it confer upon q? In such a case the conclusion certainly does not ‘draw itself’. We may even be tempted to say of such a self-critical and consciously supervised inference that here, at any rate, inferring is something which we do, as the phrase ‘making an inference’ might suggest. There is indeed something voluntary about it. It is an exercise of our freedom. The initial suspension or inhibition is voluntary; and so is the attention we give to the logical connection between the proposition already believed and the other proposition to which our belief would have ‘spread’ if we had allowed it to. And after that, our freedom is exercised again in our willingness to be guided by the strength, be it great or little, of the logical connection between the two propositions, and to conclude accordingly. For instance, we notice that p makes q very likely, but does not make it certain. Then, though we feel absolutely sure that p, we shall not allow ourselves to feel absolutely sure that q, but only to have a pretty confident opinion that q.
So much for inferring. We have only been concerned with it in so far as it is one very obvious way in which our beliefs are occurrently manifested. As Ramsey pointed out, every proposition which a person believes is for that person a potential premiss, so long as his belief continues; and when it actually functions as a premiss, whether consciously or not, whether in a self-critical way or not, his belief with regard to it is occurrently manifested.
Belief as a Multiform Disposition
In these two lectures we have considered a good many different sorts of belief-manifesting occurrences. It should now be clear that if ‘A believes that p’ is a dispositional statement about A, the disposition we attribute to him is a multiform disposition, which is manifested or actualized in many different ways: not only in his actions, not only in his actions and his inactions, but also in emotional states such as hope and fear; in feelings of doubt, surprise and confidence; and finally in his inferences, both those in which a belief just ‘spreads itself’ from a proposition to some of its consequences (certain or probable), and those in which the inference is a self-conscious and self-critical intellectual operation. Some of these belief-manifestations are public, or public in principle, though not necessarily public de facto. But others are inward and private. They take place in the ‘inner life’ of the believer. If we wish to know what they were, we have to ask him, as indeed we sometimes have to ask him about those which are in principle public, since they are not always public de facto. Moreover, even those which are public de facto, such as the actions we see him performing, do also have their ‘inner’ or private aspect. Actions, or at any rate normal actions, have to be experienced by the person who does them. The Acting-as-if Analysis, incomplete as it is, is still more incomplete if action is ‘reduced’ to observable bodily movement, or inaction to observable bodily quiescence.
If one may dare to say so, the ‘inner life’ does matter. Certainly we cannot ignore it when we are trying to understand what it is to believe something. And if we think of these various sorts of belief-manifestations as belief-symptons, by means of which another person can find out what A's beliefs are, or what degree of belief he has concerning this or that proposition, we certainly have to consider not only overt and public symptoms but private and purely introspectible ones as well: for instance, the private feeling of slight surprise which A experienced when it did not rain after all.
It may well seem to you that the belief-manifestations we have discussed are a ‘pretty miscellaneous lot’. Indeed they are, and others might well have been added to the list. Anger, for example, can be a manifestation of belief. I am angry with you about something which I believe you said or did (or omitted to say or do). Perhaps my belief is incorrect, and then I am angry with you ‘under a misapprehension’. All the same, my anger is a manifestation of my belief. Again, belief sometimes shows itself by the distress or dismay one feels when others deny the proposition believed, or express doubt of it, or say ‘only an idiot could believe a thing like that’. For we have a certain attachment to propositions which we believe, and we may still have it even when we should much prefer them to be false. Something like the principle ‘Love me, love my dog’ applies to them.
But, after all, the miscellaneous character of belief-manifestations is one of the most interesting and important things about them. If A holds some belief, many different sorts of happenings in A's history, both overt happenings and purely private and introspectible ones, are tied together or made explicable by the fact that he holds it. If the Dispositional Analysis of belief is correct, believing must be a multiform disposition. That is why the acting-as-if version of the Dispositional Analysis is too simple and too narrow.
Belief and Assent
In this long and laborious attempt to reformulate the Dispositional Analysis, we have so far said nothing about the event which is the main topic of the traditional Occurence Analysis, the mental event (or mental act) described as ‘assenting to a proposition’. If believing is to be conceived as a disposition, is it among other things a disposition to assent to the proposition believed? Another topic which we did not discuss is the distinction between reasonable and unreasonable belief. What does the Dispositional Analysis make of this important distinction? There is a connection between these two questions, as we shall see.
There certainly is a mental event which can be quite naturally described as assenting to a proposition. Moreover, it is often a purely inward event. It need not necessarily be expressed by means of bodily behaviour, for instance by saying ‘Yes, I think so’ or ‘I am sure it is so’, or writing a sentence down on paper and underlining it in green ink.
As we have seen, a person who believes that p is disposed to use the proposition p as a premiss in his inferences. Now when the inference is of a conscious and explicit kind, it is not enough to say (though it is true) that he believes the premiss; nor is it enough to say (though this also is true) that his belief spreads itself or extends itself to the conclusion. He has to assent to the premiss and he has to assent to the conclusion, if the inference is to be a conscious and explicit one. And it is not very extravagant to describe these two events as ‘mental acts’ or ‘acts of assenting’. If believing is a disposition which we have with regard to a proposition, it certainly is (among other things) a disposition to use that proposition as a premiss. And when we do actually use a proposition p as a premiss in some conscious and explicit inference, the disposition is manifested by this important sort of mental occurrence or mental act.
The assent to the conclusion q is likewise a mental occurrence or mental act. But it is not an occurrent manifestation of a belief which we already have. It is the initiation of a new one. By assenting to the conclusion q we have acquired a new disposition, which is liable to manifest itself by various sorts of mental and psychophysical occurrences thereafter; and further acts of assenting to q may be among them. For one thing, q itself has now been added to our ‘stock of premisses’, and we may have occasion to use it, in its turn, as a premiss in some conscious and explicit inference.
If the inference from p to q is a valid one, the belief that q which we have thus acquired is a reasonable belief. But what shall we say if the belief that p is itself unreasonable? Conceivably, it might be just a piece of ‘wishful thinking’. For instance, I believe (on no evidence at all) that I am going to win £1000 in next week's lottery; and from this I infer that by the end of next week I shall have more than enough money to buy a new car which costs £800. It might be said that in believing this conclusion I am just as unreasonable as I am in believing the premiss. And yet, so long as I do believe the premiss, it would be unreasonable of me not to believe the conclusion. Perhaps we may say that it is ‘conditionally reasonable’ to believe the conclusion.
But conditionally reasonable believing is not the only sort of reasonable believing. If it were, every reasonable belief would have to be acquired by means of inference from something we believe already. There is, however, another way of acquiring reasonable beliefs. We can acquire them by considering evidence. (The four most important types of evidence are the evidence of sense-perception, of self-consciousness, of memory, and of testimony.2) This way of acquiring reasonable beliefs is the primary one. It provides us with the premisses which we may subsequently use for acquiring other beliefs by means of inference.
Nevertheless, this reasonable procedure of considering evidence, and believing accordingly, does resemble inference in one important respect. When we infer q from p in a self-conscious and self-critical way, our belief that q is initiated by a conscious act of assent. The same thing happens when we acquire a new belief by examining the evidence. We acquire it by assenting to that proposition which is supported (on balance) by the evidence which we have. Here too, our assent is an introspectible mental event and can be quite naturally described as a ‘mental act’. Here too, it initiates a disposition which we did not have before. For instance, as a result of this assenting, we are now disposed to feel surprise if the proposition to which we have assented turns out later to be false, to use the proposition thereafter as a premiss, and to feel doubt of any other proposition which would be very improbable if the proposition assented to is true.
Here, then, is the connection between assenting and reasonable believing. Both in the primary sort of reasonable believing which depends on the examination of evidence, and in the derivative or inferential sort which we called ‘conditionally reasonable’, a belief is acquired by means of a special sort of mental event or act. We need a name for this mental event or act; and ‘assent’ (or ‘assenting’) is the most suitable one.
We cannot just call it ‘coming to believe’. For one thing, we may come to believe a proposition in an unreasonable way; and then, this multiform disposition which we call ‘believing’ is not initiated by any conscious mental act. Unreasonable beliefs come into being in a ‘behind the scenes’ manner. The believer does not notice himself acquiring them. He acquires them without knowing what is happening to him.
Secondly, assenting also occurs, when the disposition has already been established; and not only when the proposition believed is being used as a premiss in some inference, theoretical or practical, but on other occasions too. For instance, if someone asks me (or I ask myself) whether I do believe the proposition p or whether I ‘really’ believe it, I may consciously and attentively entertain the proposition p; and then I may find myself assenting to it. It is true that this does not completely settle the question whether I do believe that p or do ‘really’ believe it. It might be that I am in the curious state called ‘half-belief’ which we shall discuss later.3 (Some events in my history, both overt and introspectible, may suggest that I do believe the proposition, while other events in my history suggest that I do not.) But still, the fact that I consciously assent to the proposition here and now when I entertain it attentively is some evidence—and quite strong evidence—that I do believe the proposition. If we wish to find out whether a person believes that p, there really is something to be said for going and asking him.
The Occurrence Analysis Reconsidered
The traditional Occurrence Analysis of belief was certainly mistaken when it described belief as an introspectible mental event or mental act. Nevertheless, the mental events or acts which its exponents refer to do occur, and they are relevant to the analysis of belief. Only, it is a mistake to say that these mental events or acts are beliefs or believings. I suggest, therefore, that we ought to be more indulgent to the traditional Occurrence Analysis than we are. If we were willing to be very indulgent indeed, we might even say that its mistake was mainly a mistake of idiom. Philosophers do use language in odd ways, especially when they are trying to say something which is very difficult to say at all. But still, they may manage thereby to draw our attention to some important question which we should not have considered otherwise. In this case the question is ‘In what way are introspectible events relevant to the analysis of belief?’
Hitherto, I have been emphasizing the differences between the traditional Occurrence Analysis and the modern Dispositional Analysis, as if we had to choose between the two. The situation is not quite so bad as that. The differences are there, and if we do have to choose, we must prefer the Dispositional Analysis. But still, up to a point, we may have it both ways. Much of what is said in the traditional Occurrence Analysis can be incorporated into the Dispositional Analysis. Mental events which can quite properly be described as ‘assents’ or ‘assentings’ really do occur; and they certainly are relevant to the analysis of the complex and multiform disposition which we call belief, whether we are considering the initiation of such a disposition, or the occurrent manifestations of it when it has been acquired. For once in a way, let us rejoice in complexity.
Finally, our discussion of assenting may help us to understand how there can be ‘unconscious’ beliefs. When we say that A has an unconscious belief that p, we mean that all or most of the other manifestations of a belief that p do occur in him, but he does not assent to the proposition p when he entertains it and attends to it, and perhaps he even rejects it. He is surprised when presented with evidence which suggests that p may after all be false, and quickly forgets that adverse evidence. He acts as a person would act who used p as a premiss in his practical reasoning, but does not notice that he is acting as such a person would. He may also show the emotional symptoms of a belief that p, for instance a fear of some unpleasant event which would be unlikely to happen unless p were true. But when he actually entertains and attends to the proposition p (as he has to, when we ask him ‘Do you believe it?’) he does not assent to the proposition and his answer is ‘No, of course not’. And he makes the same answer to himself, when he asks himself that question.
Assenting to a proposition is certainly a very important symptom of belief, as we have seen; and we might be inclined to say ‘since he does not assent to the proposition when he entertains it, he obviously does not believe it’. Yet he does show many of the other symptoms of believing it, including some of the introspectible ones. It cannot be denied that he is at any rate in a belief-like state with regard to this proposition. It is better, then, to use a terminology which emphasizes the striking resemblances which there are between this man's state and the state of the normal believer. We do this by saying ‘he unconsciously believes that p’. The other alternative, of refusing to say that he believes it, has the defect of drawing our attention away from some interesting and puzzling phenomena which we need to understand.
For similar reasons, it seems perverse to deny that there can be unconscious wishes or unconscious fears (as philosophers sometimes do) by insisting on terminological rules which would make it nonsensical to use such expressions. If we insist on such a narrow and rigid terminology, we deprive ourselves of insight into the complex facts of human nature, by depriving ourselves of the linguistic tools which are the most handy ones for describing them and discussing them.
I conclude then that there are unconscious beliefs; ‘A believes that p’ can still be true, even though A does not assent to the proposition p when he entertains it and attends to it. We may, if we please, regard this as no more than a terminological recommendation. But I hope it has been shown that there are reasons for making it.
It is not very surprising that we should find the term ‘unconscious belief’ useful. Everyone has many beliefs about other human beings, and some of them are closely connected with his moral dispositions. Among these beliefs of his, there may well be some which suggest pretty strongly that he has an uncharitable attitude towards some other person or persons. So he has a strong motive for concealing from himself the fact that he holds such beliefs. The same is true of a person's beliefs about himself. For example, it has been said that no one is indispensable, and most of us would agree with this remark. Nevertheless, it is quite easy for a person to believe that he is an exception to this general rule; to believe, for instance, that he is himself an indispensable member of some institution to which he belongs, even though no one else is. If he were to acknowledge that he does hold this belief, he might well have to admit that he over-estimates his own mertis or capacities, and also perhaps that he underestimates the merits or capacities of others.
No one likes to admit that he has such morally-reprehensible attributes as conceit or unfairness, and therefore we are all reluctant to admit that we have the kind of beliefs which a person with these attributes is likely to have. If we were to recollect some of our past sayings and doings and some of our own past introspectible states, and I ask ‘what do all these add up to?’ it would become obvious that we do hold some belief which it is painful or shameful to acknowledge. But either we fail to recollect these phenomena which have occurred in our own past history; or if we ever do, we fail to consider them together, and therefore fail to notice ‘what they all add up to’.