The Two Analyses Contrasted

So much for the traditional Occurrence Analysis of belief. It is now almost universally rejected, at any rate in the English-speaking philosophical world. As we have seen, there are several different versions of the Occurrence Analysis. Locke's version differs from Newman's, and Hume's version is very different from Cook Wilson's. But the contemporary view is that all of them are mistaken. All of them start off on the wrong foot, so to speak, by trying to answer a question which should never have been asked. It is absurd to ask what kind of a mental occurrence believing is, or what kind of a mental act it is, because believing is not any kind of mental occurrence or act.

It would be admitted, I think, that entertaining a proposition is a mental occurrence and that we may call it a ‘mental act’ if we please. But (it would be said) when a man believes some proposition which he is now entertaining, there is not some special sort of event or act occurring in his mind over and above the event or act of entertaining. The additional factor, which makes the difference between bare or neutral entertaining and believing, is something dispositional.

It would also be admitted that coming to believe a proposition, and ceasing to believe it, are mental occurrences, though not necessarily introspectible ones. Or perhaps we should call them processes, for a belief can be acquired over quite a long period of time, and can also be lost gradually. At any rate, both coming to believe and ceasing to believe are happenings. But the belief itself is not a happening: any more than a taste for oysters is, though this too is something which is acquired and may later be lost, and both the acquisition of it and the loss of it are happenings.

So the Dispositional Analysis does not deny that there are mental occurrences nor that they are relevant to the analysis of belief; or at least it need not, though it has sometimes been stated in a purely Behaviouristic form. But it does deny that belief is itself an occurrence or mental act. A belief, according to this analysis, is something which we have or possess for a period, long or short, not something which happens to us, or something which we ‘mentally do’, as the phrase ‘mental act’ might suggest. In this respect (it would be said) belief is like knowledge. Both ‘know’ and ‘believe’ are dispositional words; and the ‘acts of knowing’, which philosophers used to discuss in former days, are just as mythical as ‘acts of believing’.

But the difference between the two analyses of belief is not quite so clear-cut as it looks. Even those who think that believing is a mental occurrence or mental act must admit that the word ‘believe’ is quite often used in a dispositional way. For many years I have believed, on the authority of my teachers, that Rome was founded in 753 BC. But it certainly is not true that all through those years this proposition has been continuously present to my mind in a forceful or vivid manner, or that an act of assenting which has this proposition for its object has been going on in me all the time. If there are indeed occurrent believings or acts of believing, they can only occur at times when a person is actually thinking of or entertaining the proposition believed. And if his belief still exists at other times, as some of our beliefs certainly do, we must admit that the word ‘believe’ does have a dispositional use, and is in fact used in this way very frequently. Otherwise, we should have to say that when I am not actually thinking of the proposition ‘Rome was founded in 753 BC’ I no longer believe it; and every time we fall asleep we should lose most of the beliefs which we had during our waking hours, or even all of them.

It follows that the Occurrence Analysis (or at least any plausible version of it) agrees with the Dispositional Analysis on a most important point. According to both analyses, the verb ‘to believe’ can be used in a dispositional way and must be so used quite frequently; and the same is true of such phrases as ‘suspect that’, ‘be of the opinion that’, ‘be convinced that’, which indicate the various degrees a belief may have. However devoted we are to the Occurrence Analysis, we must admit that some statements of the form ‘x believes that p’ do have to be analysed in a dispositional way.

But of course there is still a very great difference between the two analyses. According to the Occurrence Analysis, the word ‘believe’ does have an occurrent use as well as a dispositional one. Moreover, ‘as well as’ does not put the point strongly enough. According to the Occurrence Analysis, the occurrent use of ‘believe’ is primary, and the dispositional use is derivative; a belief, in the dispositional sense, is a disposition to have actual belief-experiences, occurrent believings, when and if appropriate conditions are fulfilled. Whenever during the past fifty years I have actually entertained the proposition ‘Rome was founded in 753 BC’, this proposition presented itself to my mind in a vivid or forceful way (as Hume would put it), or an act of assenting took place in me which had this proposition for its object (as Locke or Newman put it). Moreover, at any other time during those years it was true of me that if I had been entertaining this proposition at that time, it would have presented itself to my mind in a strong or solid or forceful manner, as Hume would say, or an act of assenting to the proposition would have taken place in me, as Locke or Newman would say.

In the Dispositional Analysis, on the contrary, it is held that the word ‘believe’ (and other kindred expressions such as ‘think that…’ or ‘be convinced that…’) has only a dispositional use. The disposition which we attribute to someone when we say he believes a proposition p does of course have its characteristic sort or sorts of occurrent manifestations. Various events are liable to happen in him which would not have happened if he did not believe this proposition. What sort of events they are, we shall see later. But none of these occurrent manifestations are themselves believings. According to the Dispositional Analysis there are no occurrent believings, or rather it makes no sense to say that there are.

It would follow that in this respect the verb ‘to believe’ differs from some other psychological verbs, for instance the verb ‘to fear’. It can truly be said of me that I fear adders, or have a fear of adders, though there are no adders in this neighbourhood at present. But when someone says this of me, he is saying that whenever I do encounter an adder I experience actual feelings of fear, and if I had been seeing an adder now, I should have been experiencing such feelings now. Again, we may say of someone that he is indignant about what has recently happened in Borneo, and this statement may be true of him when he is not thinking of these events at all. But if we make this statement about him, we are saying that whenever he does think of them, he experiences an actual feeling of indignation. Both the verb ‘to fear’ and the verb ‘to be indignant’ have an occurrent use as well as a dispositional use. In this respect they are unlike the verb ‘to believe’, if the Dispositional Analysis of belief is correct; the word ‘believe’, which we use for describing the disposition, cannot also be used for describing any of the actual occurrences by which the disposition is manifested. It has only a dispositional use.

The Dispositional Analysis: Some Logical Considerations

When we say that x has a disposition D, for instance that an object is elastic or soluble in water, or that a person is timid or friendly, what kind of a statement of this? The usual answer is that it is equivalent to a series of conditional statements. If x were in circumstances c₁ an event A would occur in x; if it were in circumstances c₂, a different event B would occur in x, etc. Thus when we say that x is elastic, we are saying that if it were stretched, it would return to approximately its original length when released, and if it were bent, it would return to approximately its original shape when released. We are not of course saying that it is actually being stretched or bent, or even that it ever has been stretched or bent, or that it ever will be: though if it never had been actually bent or stretched, or at least if things like it never had been, we should have no means of knowing that these conditional propositions are true of it.

Next, it is important to notice that temporal predicates apply to dispositions and not only to occurrences. We are liable to overlook this. Assuming, rightly, that there are very important differences between dispositions and occurrences (or between disposition-statements and occurrence-statements) we conclude, wrongly, that they have nothing in common at all.

Or, if we do not make this mistake, we tend to assume that the only temporal predicate which applies to dispositions is the predicate of permanence, The dispositions of x, we think, last as long as x does, whereas the occurrences which happen to it are momentary or at any rate have a brief duration. But that is a mistake too. Some dispositions, both of things and of persons, are permanent in this sense, though it is not always easy to discover which are the permanent ones; but others have a beginning and an end, as occurrences have. For example, this rubber band is elastic as present. But when it is heated to a high temperature, or is exposed for some days to strong sunlight, it loses its elasticity.

So in the conditional statements which describe a disposition we have to mention a period of time. If between t₁ and t₂ x were in circumstances c₁ an event A would occur in it. If between t₁ and t₂ it were in circumstances c₂, and event B would occur in it, etc. The period between t₁ and t₂ may cover the whole duration of x's existence; on the other hand, it may be quite short. The dispositional property of being good to eat only belongs to a plum for a few days, between the time when it gets ripe and the time when it goes rotten; and a piece of toast only has this property for an hour at the most. Or again, let us consider the dispositional property of being adhesive. A piece of sealing wax has this property for a very short time, while it is held in the candle-flame and for perhaps ten seconds afterwards. (Its adhesiveness is a dispositional property. The sealing wax has it, for that short period, even though it is not actually used for sealing up a parcel or an envelope.)

It is important to remember this when considering beliefs. Some of our beliefs do last for a very long time. Archibald acquired his Marxist beliefs when he was a schoolboy, forty years ago, and he has retained them ever since. But there are beliefs which are very short-lived indeed. On a warm and sunny afternoon, a motorist may believe that there is a pool of water on the road ahead. But a second or two later, he ceases to believe this, and is convinced that it was a mirage. When I got up in the morning I believed that to-day was Thursday. But I ceased to believe this as soon as I looked at the newspaper at breakfast. If beliefs are dispositions, it is important to notice that some of them are relatively short-lived dispositions, unlike many of the other dispositions of human beings such as avarice or audacity. And unless we think that there are innate beliefs, no beliefs at all are dispositions which a person has throughout the whole duration of his existence. All of them are acquired at some time or other, though once they have been acquired, some of them may continue as long as the person does who has them.

Epistemological Considerations

These logical properties of the dispositional analysis of belief have important epistemological implications, and perhaps rather disturbing ones. If ‘A believes that p’ is a dispositional statement about A, how is it possible to know or find out that A does believe that p? On the traditional occurrence analysis, there is always one person, namely A himself, who can find this out directly, by introspection. All he has to do is to consider the proposition p and notice what is going on in his mind. Does he or does he not have the experience of assenting to the proposition? If he does have this experience, he believes the proposition: if not, he does not believe it. (It does not follow that he disbelieves it. To find out that he disbelieves it, he must either have an experience of ‘rejecting’ it, or an experience of assenting to its contradictory, not-p.)

Or again, as Hume would put it, he has to consider the proposition p and notice the way in which it presents itself to his mind. Does it present itself to his mind with a ‘feel’ of solidity or force-fulness or strength? If so, he believes the proposition; if not, he does not.

In the same kind of way, by introspection of his occurrent belief-feelings, he can know directly what degree of belief he has with regard to the proposition p-whether he believes it with full conviction, or is nearly sure about it but not quite, or believes it only in a mild degree (‘I think that p’) or in a very mild degree (‘I suspect that p’). All he has to do is to consider the proposition and notice what degree of confidence he feels about it.

Other people, of course, do not have this direct knowledge of A's beliefs. They can sometimes infer what A's beliefs are, by observing his actions, his utterances, or perhaps his emotional symptoms (e.g. the indignation or discomfort he shows when the proposition p is denied by someone else, the surprise he shows when p turns out to be false). But such inferences, according to the traditional theory, are not always very reliable. Only the believer himself is in a position to know directly what his beliefs are. If others wish to find out what they are, the simplest thing is to go and ask him. He may give an insincere or dishonest reply. But at any rate he knows directly what the right answer is, and no one else does.

But if ‘A believes that p’ is a dispositional statement about him, the situation is entirely different. A himself is no longer in a privileged position. Any knowledge he can have about his own dispositions is as indirect as the knowledge which other people can have about them. A disposition, whether it belongs to an inanimate object or to a person, whether it is one's own or someone else's, is not something which can be known ‘just by inspection’ either of the introspective or the perceptual kind. The events which are occurrent manifestations of it might be ‘inspected’, or some of them might be, but that is all.

A dispositional statement, as we have already seen, is equivalent to a series of conditional statements (not just one conditional statement, but a series of them). And at least some of these conditional statements have unfulfilled if-clauses; they are what logicians call ‘counter-factual conditionals’. This rubber band was about an inch long when we first saw it. Then someone took it in his hands and stretched it to about twice that length, and when he put it down again it returned to approximately its original length and shape. We say that this rubber band is elastic, that it has been so for a considerable time and will remain so for a considerable time. But when we say this, we are saying that the same thing would have happened if it had been stretched in a similar degree a minute earlier and then released, and would be happening now if it were to be stretched and released now, and will happen again if it is stretched and released at 2 a.m. tomorrow morning, though it is almost certain that no such events are actually going to occur at such an unlikely hour.

We see now that these conditional statements are much like those to which we commit ourselves when we accept a causal law: except that in this case the conditionals (including the counter-factual ones) are about one single entity—this rubber band—whereas in a causal law they are about a class of entities. Of course, there are also universal dispositional statements, for example ‘all salt is soluble in water’, ‘No serpent is liable to catch measles’. But we are concerned at present with dispositional statements which we make about an individual entity, especially about an individual human being.

The reason why such dispositional statements cannot be established by ‘direct inspection’ is now fairly obvious. They are inductive statements. It is a defect of some of the older treatments of induction that inductive statements about individual entities were neglected: for example ‘John nearly always goes to church on Sundays’, ‘My car does about 40 miles to the gallon’, ‘Oxford is usually cold in February’. Scientists indeed are not much interested in inductive statements of this kind. But it would be a mistake to think that induction only goes on in scientific laboratories (if indeed it does go on there, which some authorities on scientific methodology deny). What we call ‘knowing a person well’ consists very largely in having made a number of fairly well-established generalizations about him in particular. We do not often make them in a conscious and reflective manner. They just grow up in us by degrees, as a result of our familiarity with him, so that we are not surprised, or are less surprised than strangers would be, by what he says and does, or by the emotions or wishes which he has in various sorts of circumstances. The same applies to ‘knowing oneself’ if we take this to mean ‘knowing what sort of a person one is’. I learn by an inductive process that I am timid or resentful and very liable to put the blame on others when things go wrong.

Now if the dispositional analysis of belief is correct, statements of the form ‘x believes that p’ are somewhat similar to these. We may describe them as inductive hypotheses. As we have seen already, their scope or subject-matter is restricted in two ways. Each of them is a hypothesis about just one individual, and each of them applies only to a limited period of time, though we may not be able to say accurately what its limits are. This period of time may extend into the future, but the more remote the future is, the more precarious our hypothesis becomes.

Believing and Acting ‘As If’

Let us begin by considering this ‘acting as if’ analysis of belief. You will find an exposition of it in Professor R. B. Braithwaite's article ‘The Nature of Believing’ in the Proceedings of the Aristotelian Society for 1932–3 (pp. 129–46). I shall make some references to this article, though it should be pointed out at once that Braithwaite's views on belief have altered very considerably since he wrote it. Moreover, the theory stated there is rather less radical than the one I wish to discuss. In this article Braithwaite makes two important concessions to the traditional occurrence analysis. First, if I interpret him correctly, he thinks that a sense can after all be given to ‘I am now believing that p’ provided that I am now actually entertaining the proposition p and it is also true of me that I have a disposition to act as if p were true (p. 132). Secondly, he admits that we do sometimes have occurrent belief feelings when we entertain a proposition, though he also thinks that no such feelings occur if our belief amounts to complete conviction (p. 142).

I propose to ignore these two concessions. The theory I wish to discuss is a pure or unmixed acting-as-if theory of belief. Anyone who holds it would indeed have to admit that entertaining is somehow relevant to the analysis of belief. However regularly one acted as if p were true, one could hardly be said to believe it unless one had actually entertained it at least sometimes. For many ages many people acted as if a diet rich in vitamins were conductive to health. But they were not in a position to believe this at a time when vitamins had not yet been discovered and the very concept of ‘a vitamin’ did not yet exist. Nor are they in a position to believe it at present if they have never even heard of this concept.

Let us suppose, however, that someone who has heard of it does actually entertain the proposition ‘a diet rich in vitamins is conducive to health’: he entertains it for ten seconds this morning, and it is also true of him that he is disposed to act as if it were true. (He eats two oranges every day.) If we accept the Acting-as-if Analysis, are we compelled to say that during those ten seconds the man is ‘engaged in believing’ this proposition about vitamins, as Braithwaite apparently would? Let us also suppose that he believes the proposition with something less than absolute conviction. If we accept the Acting-as-if Analysis, are we also compelled to say that he has, or is quite likely to have, occurrent belief-feelings about the proposition during those ten seconds? It is at any rate instructive to consider a radical version of the Acting-as-if Analysis in which no such concessions are made, and when I speak of the Acting-as-if Analysis in future, it is this radical version that I shall have in mind.

We may formulate it thus: believing a proposition consists just in being disposed to act as if the proposition were true, where the proposition is one which is actually entertained sometimes by the person who is disposed to act on it. (He need not necessarily entertain it at the time when he is acting as if it were true, nor yet at the time when he is about to act in this way.)

Such an analysis of belief seems very plausible and very easy to understand. But when we consider it, we soon find ourselves confronted by questions which are difficult to answer. Let us begin with the most obvious one.

The Meaning of ‘As If p Were True’

How are we to interpret the crucial phase ‘as if p were true’? The subjunctive ‘were’ might suggest that p is actually false. Your dog is not at all dangerous really, though I habitually act as if it were whenever I encounter it. But when we say ‘he believes that the dog is dangerous’ there is no suggestion that the proposition believed is false. It might perfectly well be true. Again, ‘he acts as if…’ might suggest that he is pretending or shamming. For example ‘he sometimes acts as if he were a lunatic, but in fact he is as sane as you or I’. But in ‘he believes that p’ there is no suggestion that he is pretending.

‘He acts as if p were true’ does not entail that p is false, nor yet that he is pretending. Nevertheless, it is liable to suggest one or other of these two thoughts to a person who hears this statement or reads it; and we must resolutely ignore both of them if we are to understand the Acting-as-if Analysis of belief. No doubt we can ignore them, and we will now proceed to do so. But we should be told that we have to.

A more serious difficulty is this. When it is said that someone acts as if a proposition p were true, for example the proposition ‘it is going to rain’, we might quite well take this to mean ‘he acts as a person would who believes that p is true’. (He takes his umbrella with him when he goes out of the house.) This is quite a natural way of explicating the somewhat mysterious phrase ‘as if p were true’. But according to that interpretation, the acting-as-if analysis of belief would of course be circular.

Are we saying, then, that he acts as if p were in fact true? But this will not do either. Let p be the proposition ‘Vladivostok is a large town’. I have long believed this proposition, and doubtless most of you believe it. Do we have a disposition to act as if it were in fact true? If we have, it does not prevent us from acting on many occasions exactly as we should act if the proposition were in fact false. If Vladivostock were a tiny village, or even if there had been no such place as Vladivostock at all, many of us would be doing exactly what we are doing now.

Why is this? It is because the proposition ‘Vladivostok is a large town’ has no practical relevance for us, and never has had any. Still, it would have practical relevance for us, if we ever had occasion to go to that part of the world, and would have had practical relevance for us in the past, if we ever had had occasion to go there. So when we say of someone ‘he has a disposition to act as if p were true’, perhaps we mean that he would act in this way, if the question whether p is true or false were practically relevant to some action of his? Not quite. For it is not enough that the proposition p is in fact relevant, in this practical way; to the circumstances in which we are. We might be unaware that it is relevant.

For example, a man believes firmly that toadstools are poisonous. Yet he eats a toadstool which is served up to him for breakfast, and it makes him very ill. He did not want to be ill. Far from it. He ate a larger breakfast than usual in order to do a good morning's work afterwards. Surely he acted as if the proposition ‘toadstools are poisonous’ was false, and he acted in this way in a situation to which that proposition was highly relevant? He did, but he was not aware that it was relevant. He believed that the object on the plate was a mushroom.

It seems, then, that a person who does act as if a proposition p were true has to believe that the question ‘is p true or false?’ is relevant to the achievement of his purpose in the situation in which he is, or rather, in the situation as he believes it to be. But in that case, the Acting-as-if Analysis is open to a very serious objection. To explain what is meant by ‘acting as if a proposition p were true’ we have to introduce the concept of belief over again. If so, the Acting-as-if Analysis of belief is circular.

‘Acting-As-If’ and Practical Reasoning

Let us see whether we can avoid this difficulty by a reformulation of the Acting-as-if Analysis. The source of the difficulty is fairly obvious. When a man is said to be acting as if p were true, it is not only the belief that p which is manifested in his action. His action is a manifestation of several beliefs at the same time, and the belief that p is only one of them.

Perhaps it will be helpful to consider some remarks which F. P. Ramsey makes about the relation between belief and inference in his essay ‘General Propositions and Causality’ (The Foundations of Mathematics, pp. 237–55) especially as he himself accepted a form of the Acting-as-if Analysis. The view which he suggests might be formulated thus: when we come to believe a proposition p, we add it to our stock of premisses.2 So long as our belief continues, this proposition is available to us as a premiss in our inferences, and our belief that p is occurrently manifested whenever we do actually use p as a premiss in some inference which we make: and not only when we use it as our sole premiss, but when we use it as a premiss, along with some other premiss or premisses also contained in our ‘stock’.

The Acting-as-if Analysis can now be formulated without circularity as follows: when we say that a person believes a proposition p we mean (1) that p is a member of his stock of premisses (2) he is disposed to use it as a premiss in his practical reasoning or practical inferences, inferences whose conclusions, if put into words, would be of the form ‘let me therefore do x’ as opposed to ‘therefore q is true’. It may well be that he cannot ever draw a practical conclusion from any one of these premisses alone. He may always have to use two of them, or more, in combination, and one of them may only have been added to his stock of premisses a moment ago (e.g. ‘there is ice on the road’). But still any one of the propositions in his stock of premisses has the status of a potential premiss in some possible piece of practical reasoning; and that is what we should have to mean by saying that he believes it, if the Acting-as-if Analysis of belief is correct. So far as I can see, there is no circularity in the Acting-as-if Analysis when it is reformulated in this way.

We speak sometimes of ‘acting upon’ a proposition, or acting ‘in the light of it’, or ‘taking account of it’ in deciding what to do; and this revised version of the Acting-as-if Analysis could be put in an untechnical way by saying that a proposition which we believe is one of those propositions which we are disposed to take account of in our practical deliberations. ‘Taking account of it’ does not entail that it is ever the only proposition which we take account of. To answer the practical question ‘what am I to do?’ we might always have to take account of several propositions; and in any action which we perform, we might have to act ‘upon’ or ‘in the light’ of several propositions, and not just of one.

Assuming that this is the best way to state the Acting-as-if Analysis of belief, we may now consider an example which supports it. This afternoon you intend to go to a village ten miles away, and you believe that the bus will be crowded. Then you will do your best to arrive early at the bus station. Or you will make the journey on a bicycle. Or you will get a lift from a passing motorist. Or you will start your journey very much earlier and walk all the way. In doing any of these actions you are using the proposition ‘the bus will be crowded’ as a premiss in a piece of practical reasoning or inference-guided conduct. You are acting ‘in the light of’ this proposition or acting ‘upon’ it. You are not just making bodily movements of one sort or another. You are acting thoughtfully or intelligently.

‘Acts Speak Louder than Words’

There is another type of case which appears to support this analysis of belief. We sometimes find that a person claims to believe a proposition p or professes to believe it; and yet, in circumstances to which the proposition is obviously relevant, he acts as if it were false. For instance, he says he believes that no one can do a good day's work unless he has had a good night's sleep. Yet on Monday night he sits up playing Bridge until 1.30 a.m., although he knows that he has a very heavy day's work on Tuesday; and he repeatedly acts in this way. We are inclined to say of such a person that he does not really believe what he says he believes. He does not act ‘upon’ or ‘in the light of’ the proposition which he claims to believe.

Obviously there is a good deal of truth in such cynical remarks about ‘not really believing’; all of us are very familiar with this situation in which someone says he believes a proposition p, but fails to act as if it were true. And we do quite often judge of a man's beliefs—his real beliefs as opposed to his professed ones—by observing the way he acts. ‘Acts speak louder than words.’ ‘It is what a man does, not what he says, which shows what he really believes.’

But though these observations about ‘not really believing’ do in some respects support the Acting-as-if Analysis, in other respects they do not.

Anyone who insists in this way that ‘acts speak louder than words’ will get himself into difficulties if he also holds that ‘acting as if p were true’ includes speaking as if p were true.

For as these examples are supposed to show—indeed, that is the whole point of them—that one may speak as if p were true when one does not believe it, or does not really believe it. There is no difficulty about ordinary liars, perhaps. They do speak publicly as if a proposition p were true when they do not believe it. But it could be argued that they speak privately, to themselves, as if it were false; that it is private or inward speech (sub-vocal speech or verbal imagery) which counts; and that someone who speaks privately, to himself, as if p were true really does believe it. The difficulty is not so much about liars as about self-deceivers: people who say to themselves, as well as to others, ‘I believe that p’ when their actions show that they do not believe it. Probably there is some degree of self-deceit in most of us.

The position is, then, that sometimes a man who speaks as if a proposition were true does believe it, and sometimes he does not; it is even possible that he does not believe the proposition although he speaks privately, to himself, as if it were true. And this is the difficulty we get into, if we maintain (as some acting-as-if theorists might) that acting as if p were true includes speaking as if it were true.

‘Natural’ or Unstudied Speech

It is then maintained that one of the actions which manifests your belief that p is speaking in a natural or spontaneous or unstudied manner as if p were true. (This could perhaps be stretched to cover what one may call automatic speech: what a person says when he is drugged or drunk or half asleep, or what he says in slips of the tongue, or slips of the pen, when he is awake.5) In all these cases one is speaking without self-concern, if I may put it so—without asking oneself what impression one is making on other people. Perhaps when a person speaks in this way, what he says does always accord with what he ‘really’ believes?

Sometimes it does, but not always. There is such a thing as unconscious self-deceit, pretending, even to oneself, that one's beliefs are different from what they actually are, without being aware that one is pretending. When this happens, other people may be able to discover, by observing one's conduct, that one is thus deceiving oneself. But it is much more difficult for them (and for oneself too) to make the converse discovery, that is, to establish conclusively that one is speaking ‘naturally’ or ‘spontaneously’ or ‘without self-concern’. And this is what matters in the present argument.

In actual fact, I think, we are inclined to argue the other way round—from the beliefs which a man is already known to hold, to the ‘naturalness’ (or otherwise) of the way he speaks. In other words, we have an inclination to define ‘natural speech’ or ‘spontaneous speech’ in terms of belief: speaking naturally or spontaneously is saying what you do really believe. But if so, it would of course be circular to define belief itself, wholly or partly, in terms of ‘natural’ or ‘spontaneous’ speech.

There is another difficulty in the doctrine that speaking as if p were true is just a special case of acting as if p were true. It is this. Although speaking, even speaking privately and sub-vocally, is indeed a kind of action, it is very different from other kinds of action; and consequently the phrase ‘as if’ has a different meaning, when it is ‘speaking as if’ that we are discussing, from the meaning which it has when we are discussing other sorts of action. When I say (in a natural or unstudied manner) ‘it is going to rain soon’ I am indeed speaking as if the proposition ‘it is going to rain soon’ were true. And when I take my umbrella or my mackintosh down from the peg, I am moving a material object from one place to another as if this same proposition about the weather were true. But in the second case I am using this proposition as a premiss in a piece of practical reasoning or drawing a practical inference from it (that to avoid getting wet I must take an umbrella or a mackintosh with me when I go out); whereas in the first case, when I merely say to others or to myself, ‘it is going to rain soon’ I need not be using the proposition as a premiss at all. Of course, I may be using it as a premiss in an argument to show that the meteorological forecast in the morning paper is not infallible. But I need not be using it as a premiss in any piece of reasoning, either theoretical or practical. I may be just expressing my belief. And expressing a belief is very different from acting upon it, as I do when I take the umbrella down from the peg.

A philosopher of ‘Ordinary Language’ might also draw our attention to another point about the phrase as if. If we used this phrase about someone's utterances—e.g. ‘he talked as if it were going to rain’—we should usually mean not that he actually said ‘it is going to rain’, but that he said something else, such as ‘the barometer has been falling all day’ or ‘you will need an umbrella when you go out’. That is, he asserted some other proposition q, which would be rendered true, or highly probable, if p were true, but he did not assert p itself. In short, in this (ordinary) use of the phrase ‘speaking as if p’ or ‘speaking as if p were true’ it is implied that the speaker was using p as a premiss for an inference, without actually uttering the sentence ‘p’ itself. He was not actually expressing his belief that p, though we can infer that he holds it.

‘Not Really Believing’

So much for the difficulties which arise when ‘acting as if p were true’ is taken to include speaking as if p were true. We may now return to the inference from ‘he does not act as if p were true’ to ‘he does not really believe p’. As we have seen, the fact that we do often make this inference is one of the most persuasive arguments in favour of an Acting-as-if Analysis of belief. But the inference is often a good deal more precarious than it looks.

Let us first consider a complication which Braithwaite himself points out. The purpose of the agent must be taken into account (his ‘needs’ as Braithwaite calls them6) before we decide that he does not really believe what he claims to believe.

For example, a man says he believes that coal gas and air form an explosive mixture. One evening, as he is passing the cellar door, he notices a peculiar smell. He goes in there and lights a match, and then there is an explosion. We are inclined to say ‘You can see that he did not really believe this proposition about coal gas and air: for when it came to the point, he did not act as if it were true. Acts speak louder than words.’

But if we do say this, we may easily be mistaken. It depends what his purpose was. It may be that he wanted to commit suicide, or to blow up the house in order to get the insurance money or because he so much disliked its Ruskinian architecture. Supposing he did want to do any of these things, he acted as if the proposition about coal gas and air was true; and in that case he ought to infer not that he ‘did not really believe’ the proposition, but that he did believe it.

Let us assume, however, that he did not want to commit suicide or to do anything else which an explosion would facilitate. Even so, his action fails to prove that he did not believe the proposition. For the proposition ‘coal gas and air form an explosive mixture’ is a universal proposition. The sentence formulates a general law. And though he believed the general law and had not at all lost this belief, he may have failed to recognize that it applied to this particular case.7 Perhaps he had a heavy cold and could not smell very well. Or he may have failed to recognize what sort of smell it was. Having had a purely literary education, he has never learned to distinguish this sort of smell from others. It was just ‘a stink’ to him.

‘Losing One's Head’

But we must also consider another and perhaps more interesting possibility. When the man struck the match and caused an explosion, he may have been flustered or frightened. Perhaps when he smelt the smell in the cellar he ‘lost his head’, as we say. He did not at all wish or intend to cause an explosion. But because he lost his head, he did just the thing which was bound to have that result.

The very existence of this phrase ‘losing one's head’ seems to refute the doctrine that a person's actions are infallible signs of his beliefs. For the phrase was invented to cover the case where a man does believe a proposition, but fails to act on it—owing to fear, anger, anxiety, astonishment or some other ‘disconcerting’ emotional state (another is excessive eagerness to ‘do something about’ the situation at once).

A person may believe a proposition p; he may also believe, and believe correctly, that he is in a situation to which p is relevant. And yet, under stress of emotion, he fails to act upon this proposition. He fails to use it as a premiss in his practical inferences. He even acts as if it were false, or as if it were not relevant to the situation in which he is. When other people hear that he has acted in that way, some of them are sure to say that he did not ‘really’ believe the proposition p at all, even though he has spoken sometimes as if he did. But instead of saying that his action shows the absence of the belief, one might equally (or better) say that it shows the presence of another disposition in him, an emotional disposition such as excitability, or a tendency to get into a panic in sudden emergencies. By this account of the matter, the man's belief was really there. He did not just claim or profess to believe that coal gas and air form an explosive mixture. He really did believe it, and he still believed it even at the moment when he failed to act upon it. The belief was still there, but it was inhibited, prevented from manifesting itself, by the emotions which the situation aroused; and this, surely, is just what ‘losing one's head’ amounts to.

So if we must make some cynical comment on such a person, one who loses his head in circumstances of this kind, the appropriate comment is that his character is weak or unstable. Indeed, part at least of what we mean by saying that a person's character is weak or unstable is that in emotion-arousing situations he is likely not to act in accordance with propositions which he does quite genuinely believe.

Perhaps it will now be agreed that the relation between belief and action is more complicated than a simple Acting-as-if Analysis of belief might suggest. Many of our beliefs certainly do manifest themselves in our actions; and certainly this is a very important fact both about belief and about action. But before we can infer from a man's actions to his belief or absence of belief that p, we must take into account the following factors:

(1) his purposes, which may possibly be queer or eccentric ones, quite different from the purposes we ourselves should have if we were in his situation;

(2) his belief, or failure to believe, that the situation is one to which the proposition p is relevant. He may believe that p is relevant to the situation, when we believe or even know that it is not. He may believe that it is not relevant to the situation, when we believe or even know that it is;

(3) the strength or weakness, the stability or instability, of his character—the degree to which he is capable of ‘keeping his head’ in a situation of this particular sort, or rather in a situation which he himself believes to be of this sort.

Acting As If p Were True When One does not Believe It

We have seen that a person may fail to act as if a proposition were true, even though he does believe it. He may also act as if a proposition were true even though he does not believe it. Here we must distinguish between ‘not believing that p’ and ‘believing that not-p’. though in colloquial English the two phrases are often used as if they were equivalent.

Let us first consider an example in which there is no belief either way, the neutral state of suspense of judgement. An Oxford teacher of mine once betted on a horse because its name was ‘Aristotle’. He was a man eminent for his good sense, who subsequently became the Head of his College. He was well aware that a horse's name, however distinguished, has no relevance, either favourable or unfavourable, to the creature's chances of winning a race. He did not believe the proposition ‘Aristotle will win’ (though he did not disbelieve it either) and then he proceeded to act as if the proposition were true.

Perhaps it will be said that this action was ‘not serious’ and therefore does not count. But if what a man does is an infallible sign of what he believes, there could not be any un-serious actions at all. It is notorious, however, that un-serious actions do sometimes occur, even in our serious-minded age; and there was a time when they occurred quite frequently. Moreover, if the word ‘serious’ is introduced into the acting-as-if analysis of belief, we get into a logical difficulty. Suppose we said that believing a proposition consists in being disposed to act as if it were true when (and only when) one is acting seriously. What is meant by ‘acting seriously’? Surely it can only mean ‘acting in accordance with one's beliefs’; and that would make the Acting-as-if Analysis circular.

But it is also possible to act as if p were true when one disbelieves the proposition p (or believes its contradictory, not-p, if the reader prefers to put it that way). This may be illustrated by a well-known story about King Canute. Canute's courtiers asserted that he ruled the sea, from which it would logically follow that the sea would obey any command that he might give it. Canute did not believe this. Not only so; he disbelieved it. And he then proceeded to act as if it were true. He sat on this throne in Chichester Harbour when the tide was coming in, and commanded the sea to go back; and the sea took no notice whatever of his command. He acted as if a proposition were true when he disbelieved it, or rather, because he disbelieved it and wanted, if possible, to disprove it.

A somewhat similar procedure is an important part of scientific method. One of the best ways of disproving a hypothesis is to act in accordance with it, and show that the predicted consequences of it do not in fact occur. This illustrious Anglo-Scandinavian monarch may perhaps be regarded as one of the founders of Scientific Empiricism.

Perhaps it will be said that when Canute did this he was acting as if the proposition ‘Canute rules the sea’ were false. In a way, he was. Certainly he was acting as a highly intelligent man would act who wished to prove that the proposition was false. But if we take this line about the example of Canute, it only shows what an ambiguous criterion this acting-as-if criterion of belief is. For here the very same action can be described in two mutually incompatible ways: either as an example of acting as if a proposition p were true, or as an example of acting as if p were false. A man disposed to act as if the courtiers' hypothesis were true might be expected to do what Canute did; and a man disposed to act as if the hypothesis were false (provided he was intelligent enough) might also be expected to do what Canute did.

Purely Theoretical Beliefs

The Acting-as-if Analysis also gets into difficulties about beliefs which we hold in a purely theoretical way, without ever having occasion to act on them. (It must be remembered that speaking and writing do not count here as actions. That was the point of the remark ‘acts speak louder than words’.) Those beliefs which never manifest themselves in our actions at all may be called ‘purely theoretical’ for brevity's sake, though this adjective applies to the believing rather than the propositions believed, which may quite well be propositions about matters of fact. It is obvious that such beliefs exist. Everyone has some beliefs which are purely theoretical in this sense, and some people have a great many.

For example, I myself believe, on the authority of geologists, that in a remote geological epoch this country was not yet an island, and there was a continuous land-bridge between it and the Continent where the Straits of Dover now are. It is no doubt conceivable that circumstances might arise in which one would act as if this proposition were true. If one were a palaeontologist, one might investigate the sea-bottom outside Dover Harbour in the hope of finding fossils of extinct land-animals there. But in point of fact, no such circumstances have ever arisen in my own case, nor in the case of many thousands of other people who hold this belief. Neither they nor I have ever in fact acted as if this proposition about the remote past were true; yet there is no doubt that we do believe it.

It is true that if we are greatly interested in some proposition about the past, our belief may well affect our actions. For example, we believe that Alexander the Great invaded India in the fourth century BC, and that there were Greek-speaking people in India and in Bactria for a century or two afterwards. I myself happen to be interested in these events and in their consequences. So if there were an exhibition of Greco-Buddhist sculpture in London, I might quite well travel to London in order to have a look at it. On the other hand, I should do no such thing if there were an exhibition of fifteenth century printed books. I do believe firmly that printing began in the fifteenth century, but am not particularly interested in this proposition. In any case, it is quite impossible that we should ‘do something about’ all the propositions we believe even when opportunities arise for doing it. Non omnia possumus omnes.

Now certainly it does not follow from the Acting-as-if Analysis that anyone who believes a proposition does in fact act as if the proposition were true, but only that he would act thus if circumstances were to arise to which the proposition were (for that person) practically relevant, and would have acted thus if such circumstances had ever arisen since the time when he first acquired his belief. Beliefs which are never in fact acted upon might perfectly well exist, if the Acting-as-if Analysis is correct. The difficulty is that there would be no way of discovering that they exist. Neither the believer himself, nor others, could possibly find out that he does believe some propositions (very many, perhaps) in this purely theoretical way. It is ail very well for him to say to others, or even privately to himself, ‘I believe that p’. But this, according to the Acting-as-if Analysis, is very weak evidence for the hypothesis that he does believe it. We must have deeds, not words; and so must he, if he is to find out what his own beliefs are. Yet in point of fact, each of us is quite sure that he does believe many propositions in this purely theoretical way, and that many propositions (not necessarily the same ones) are believed in this purely theoretical way by others.

Moreover, beliefs held in this purely theoretical way need not be beliefs about the past. There are many other examples: beliefs about the remote future, such as the belief that some millions of years from now the earth will be too cold to be habitable; beliefs about objects which are remote in space, for instance the belief that there are albatrosses in the South Atlantic, or that there are mountains at the South Pole; beliefs about very general causal laws, for instance the belief that when a particle moves with a velocity approaching the speed of light, its mass increases.

It is true that any one of these beliefs might very well manifest itself in the actions of the appropriate scientific specialists. Such beliefs might determine what apparatus these men install in their laboratories, what experiments they undertake, what expeditions they made to remote parts of the earth's surface or (some day) to remote parts of the solar system. Nevertheless, these beliefs can be held, and are held, by laymen like ourselves, who never have any opportunity of acting on them; and since we never do in fact act as if these propositions were true, neither we ourselves nor others who observe our actions would be entitled to affirm that we do believe them, if the Acting-as-if Analysis is correct. But actually we know very well that we believe them. How do we know this? Is it possible that we just know it by introspection, as the traditional theories of belief maintained? How very upsetting that would be!

Finally, the Acting-as-if Analysis may be criticized for taking an unduly narrow view of human nature, and of the motives which make us believing beings. One of these motives undoubtedly is that we wish our actions to be successful. We often have to act without knowledge of some of the relevant circumstances; and we very seldom, or never, know (in any strict sense of the word ‘know’) what the results of our action are going to be. We often have to use a proposition as a premiss in our practical reasoning without knowing it to be true. When someone gets into a train at Oxford in order to attend a committee meeting in London, he cannot know that he will arrive in time or even that he will arrive at all. He has to be content with believing, on the authority of the railway timetable (which is by no means infallible), that the train will reach London not much later than the time-table said it would.

But this practical motive, this desire that our actions should be successful, is not the only motive we have for acquiring beliefs. There is also the desire for knowledge for its own sake. Knowledge is something which we value as an end in itself. It is good to know how things are. But here again, we often find that knowledge in any strict sense of the term is not available. All the same, the desire for knowledge is still there, and belief is a partial satisfaction of that desire; a second-best, but much better than nothing. That is why there are purely theoretical beliefs, which make no difference to our actions.

It is therefore not so very surprising that theory which defines belief as a disposition to act as if a proposition were true is not altogether satisfactory. It is indeed true, and important too, that many of the propositions which we believe are used as premisses in our practical reasoning. One of the functions of belief is to provide us with guidance in our actions and our practical decisions. But this is not its only function. Another is to give a partial and second-best satisfaction to our desire to know how things are.

• 1.

  See below, pp. 256–9.

• 2.

  Ramsey Himself does not actually use these words, so far as I have been able to discover. But he does speak of ‘adding a proposition hypothetically to [one's] stock of knowledge’ (p. 247 n.) and he also says ‘it belongs to the essence of any belief that we deduce from it and act on it in a certain way’ (p. 251 ad. fin.).

• 3.

  Something more will have to be said later about this concept of ‘not really believing’ when we discuss Half-Belief. See Lecture 4, below.

• 4.

  See G. Ryle The Concept of Mind, pp. 181–5.

• 5.

  Cf. In vino Veritas, though Veritas here primarily means sincerity about one's wishes and emotional attitudes, rather than sincerity about one's beliefs.

• 6.
  Proceedings of the Aristotelian Society, 1932–33, pp. 134–5.
• 7.

  Cf. Aristotle, Nicomachean Ethics, Book VII, ch 3.