The ‘acting as if’ version of the Dispositional Analysis of belief is no doubt the simplest, but it turns out to be too simple. There is a certain narrowness about it. Why should we suppose that a person's overt actions are the only occurrent manifestations of his beliefs? If a belief is a disposition which a person acquires at a certain time and retains for a certain period, it is surely a multiform disposition which manifests itself in many different ways, and not only in the actions which he does.

First, we notice that inaction, as well as action, can be a manifestation of belief. For example, my passenger has the map, but for the past twenty minutes he has not even opened it. He just sits there, doing nothing about it. Thereby he manifests his belief (possibly an incorrect one) that we are on the right road. Again, by sitting quietly in the waiting room, doing nothing, I manifest my belief that the train will not arrive just yet.

This is our first step beyond the too narrow and restrictive Acting-as-if Analysis, and the reader may think it is a very little one. If the acting-as-if philosophers themselves failed to take it, perhaps that was no more than an oversight on their part. For it might well be said that inaction, as well as action, is a form of conduct, at any rate when it is belief-manifesting inaction and not just a bodily quiescence resulting from purely physiological causes.

We take a much bigger step beyond the Acting-as-if Analysis when we notice the close connection there is between belief and some emotional states.

Let us first consider hope and fear. It might be thought that epistemology has no relevance to the study of emotions. But epistemological questions, and even logical ones, do arise about hope and fear. We notice at once that both verbs, ‘to hope’ and ‘to fear’, can govern that-clauses. We also notice that hopes and fears can be reasonable or unreasonable. There are groundless hopes, and even absurd ones (for example, the hope that we shall some day visit Utopia). But there are also reasonable hopes, and it may still have been reasonable to entertain them even though they were eventually disappointed. One may have good grounds for fear too, and one may still have them even though all was well in the end. (The ice was thin, and I feared that it would break, but in fact I got across it safely.) On the other hand, some fears are unreasonable or groundless: which does not of course imply that they are not ‘really’ fears, or that they do not matter to the person who feels them.

Belief and Hope

It is indeed obvious that hope and belief are closely connected.1 If we hope that x will happen, we must at least believe that it is possible that x will happen. Once we are convinced that the thing hoped for is impossible, hope vanishes away. To abolish our hope, it is enough to be convinced that what was hoped for is logically impossible (e.g. the squaring of the circle, which was hoped for long ago). But if our hope is to be retained, we must believe that the thing hoped for is causally possible also. I may be very fond of the sunflower in the garden, but I cannot very well hope that it will say ‘good morning’ to me. It is no doubt logically possible that a sunflower should speak, but I do not believe that it is causally possible.

In very difficult and dangerous situations when ‘we hope against hope’, as we say, it is enough to believe that the event hoped for is at any rate causally possible. If there is just a chance that I shall get over the top of the slippery wall before the wolf catches me, I still hope that I shall manage to do it. But normally (as the phrase ‘against hope’ suggests) we believe more than this. We believe, with some degree of confidence, that the thing hoped for will actually happen. We at least suspect or surmise that it will (these are the traditional words for the lowest degree of belief). We may go farther and have the opinion that it will. We may even be almost sure that it will happen; this is what we call a ‘very confident hope’.

But if we are absolutely sure that the event will happen and have no doubts about it at all, then again we cease to hope: not because we despair, but because we have gone beyond hope into something better. We cease to hope, because we are now more than hopeful. We do not hope for a certainty, or for what we consider to be such. Does anyone hope that the sun will rise next Wednesday? A philosopher might, if he is rather sceptical about induction. He might also say that other people ought, or ‘ought only’, to hope that the sun will rise next Wednesday, because he thinks they have no right to be sure that it will. But they pay no attention to his admonitions. Try as they may, they cannot hope for this event, because they cannot get rid of their complete conviction that it is going to happen.

So far, then, it seems that hope lives in the intermediate region between two opposed convictions. The person who hopes must not be absolutely sure that the event will not happen; but neither must he be absolutely sure that it will. Some degree of incertitude is a characteristic feature of hope.

But there is another belief-factor in hope: the valuational belief that it will be a good thing if x happens (whether good in itself, or good as a means). One does not hope that x will happen if one believes it would be a bad thing for x to happen, or even if one believes it would be neither good nor bad. Here the situation is quite different. Let us consider a politician who is completely sure that it will be a good thing if his party wins the next election. Obviously this does not prevent him from hoping that his party will win. On the contrary, we might be tempted to think that this conviction is a necessary condition for his hope. At any rate, if he only suspects that his party's victory will be a good thing, or thinks so without being anything like sure, we should probably say ‘he half-hopes’ or ‘he has some inclination to hope’, rather than ‘he hopes’. Perhaps if we are to hope that x will happen, it is enough to be nearly sure that it will be a good thing if x happens. But ordinarily, I think, when we hope that x will happen, we are convinced (rightly or wrongly) that it will be a good thing if x happens. So there is an important difference between the two beliefs which we must have if we are to hope for something. There must be some degree of incertitude in our factual belief, the belief that x is going to happen. But in our evaluative belief, the belief that it will be a good thing if x does happen, there need be no incertitude at all; and if there is any, it is likely to be slight.

Finally, we may notice that the incertitude of our factual belief makes it possible for us to have hopes about the unobserved present, and even about the past. Usually we are unsure about the unobserved present, and often we are unsure about the past. For example, I can hope that my remarks yesterday evening did not cause offence to anyone, because I am not sure what affects they had on my audience. John has been ill, but I hope he is perfectly well by now. I can have this hope so long as I am not absolutely sure what his present state of health is. Did we win the boat-race this afternoon? So long as no news has reached me about the result of the race, I can hope that we did win.

Such hopes may seem paradoxical. Ordinarily our hopes are concerned with the future. It might be suggested, then, that when I say ‘I hope that John is perfectly well by now’ I am really hoping that evidence will be forthcoming which will show that he was indeed perfectly well at the time when I made my remark; and that in the same way, I hope to receive evidence later that what I said yesterday evening caused no offence.

On the contrary, it does not feel like this at all, even though in some of these examples I do also hope that future evidence will be forthcoming. (For instance, I hope that the evening paper will inform me that we did win the boat-race.) When I hope that John is now well, it feels like a hope about what now is. When I hope that my remarks yesterday caused no offence, it feels like a hope about what formerly was—the state of mind of my audience yesterday evening.

Moreover, so far as the hoping person can tell, there may be very little likelihood that such future evidence will be forthcoming. He may even be quite sure (correctly or not) that he is never going to get any; and as we have seen, this makes it impossible for him to hope that he will get it. There may be some chance of receiving information which will show that my remarks yesterday did not in fact cause offence. But I might have a similar hope about something I said at a meeting thirty years ago; and it is much too late to hope that evidence will be forthcoming which will show that I caused no offence on that occasion.

Belief and Fear

We may now turn to the relation between fear and belief. We tend to think of hope and fear as a symmetrical pair, as if fear were the same attitude as hope but with a ‘change of sign’ (like the change from + to −); or as if the only difference between hope and fear was the difference between pro and con. There is indeed a difference of this kind between them. When we hope that x will happen, we believe that it will be a good thing if x happens. When we fear that x will happen, we believe that it will be a bad thing if x happens. If you and I have both promised to play cricket this afternoon, I hope that it will rain this afternoon, whereas you fear that it will.

We can also detect some important resemblances between the two emotions so long as we confine our attention to fear that… Some degree of incertitude is a necessary constituent of both. Unless one has doubts about the reliability of induction, one does not fear that the sun will fail to rise next Wednesday, any more than one hopes that it will rise. More paradoxically, perhaps, if one is quite sure that some very unpleasant event will happen, one can no longer fear that it will happen, any more than one can hope that some very pleasant thing will happen when one is quite sure that it will.

For the past twelve hours the conscience-stricken schoolboy has feared that the Headmaster will summon him for an interview to-morrow. But now the matter is settled. The summons has come, and this fear ‘that’ is abolished. It is not that the boy's fear has been abolished altogether. Far from it. But it assumes a different (and worse) form. Fear ‘that’ the interview will happen is replaced by fear ‘of’ the interview itself. He contemplates this future event with fear and trembling. Nevertheless, he does not, and even cannot, continue to fear that it will occur. He may now have other fears ‘that’, for example the fear that he will be beaten or that he will be expelled from the school or that he will disgrace himself by shedding tears, since he cannot be sure that these things will happen in the course of the interview, but only that there is an appreciable probability that they will. But the particular fear ‘that’ which he previously had—the fear that the interview would take place—really has been abolished, just as hope is abolished when we come to be sure that the event hoped for is actually going to happen.

Again, fear ‘that’ can be extended to the unobserved present, and even to the past, as hope can. As I can hope that John is quite well by now, I can fear that William is still pretty ill. As I can hope that my remarks yesterday evening caused no offence, I can fear that those I made a week ago did cause great offence. And here again, there has to be some degree of incertitude. If I am quite sure that William is still pretty ill, I can be very sorry that it is so, but I cannot fear that it is so. Nor can I fear that my remarks caused offence if I have conclusive evidence that they did. I can only regret very much that I made them.

But now we encounter a complication. One might say to another person (not to oneself) ‘I fear that p’ when one is quite sure that p is true. For example the newsagent at the railway station, when asked for a copy of to-day's Times, might say ‘I am afraid that we have no copies left; we sold the last one an hour ago’. Or again, when I am asked ‘Who did it?’ I may reply ‘I am afraid that I did’. But there is some incertitude here: not about the fact itself, but about the way one's hearer will take it. The newsagent fears that you will be distressed to hear that there is no copy left. I fear that you will be shocked to hear that I am the person who did it. If the speaker were quite sure that his information would distress or shock his hearer, he would put his remark differently. He would say ‘I am very sorry to have to tell you that…’ instead of ‘I am afraid that…’.

Fear ‘Of’ and Fear ‘That’

So far, then, there are important parallels between hope and fear. But it would be a great mistake to suppose that the only difference between them is the difference between pro and con. For one thing, fear is a more passive attitude than hope. We can be frightened by something, made or forced to fear. But there is nothing which is related to hope as being frightened is related to fear. We can hope but we cannot be ‘be-hoped’. We may no doubt be encouraged or persuaded to hope; but we can resist such encouragements and persuasions, and we often do. Hope is more autonomous than fear. It is not just something which we suffer. Consequently, fear is even more liable to be unreasonable than hope is. The mere thought of something can frighten us, though there is no evidence at all that it is going to happen.

There is another difference, and it is more relevant to our discussion. Fear is not always fear that.… One can also fear an entity—a wolf, or a precipice, or a headmaster. One can fear an event, either an event which is actually happening, such as the thunderstorm which is now going on, or a future event (e.g. the interview with the headmaster to-morrow) if one is absolutely sure that it is going to happen. And here the passive character of fear is most obvious. These entities or events frighten us. They make us afraid.

To put it in another way: hope is always a propositional attitude. It is always hope that… Hopes ‘of’ and hopes ‘for’ are quite easily reducible to hopes ‘that’. The hope of victory is just the hope that our side will win. Belinda hoped for a pony (you have given her a donkey instead). But her hope for a pony was the hope that she would have a pony, or that she would be presented with one. Fear can be a propositional attitude too (fear that…) but it need not be. There is also fear of…, where the ‘object’ of our fear is not a proposition but an entity or an event. We fear the wolf. But we cannot hope the Siamese cat; or rather, it makes no sense to say we can, though we can very well hope that this agreeable creature will visit us again some day.

Consequently, fear ‘of’ is not abolished by certitude, as fear ‘that’ is. I do not fear the wolf any less because I am sure that it is now there in front of me. The schoolboy does not fear to-morrow's interview any less when he is quite sure that it is going to happen. This certitude does indeed deliver him from his previous fear that the interview would happen. But, as we have seen, his fear ‘that’ is replaced by something worse—fear ‘of’. One does not hope for a certainty. But it might be said that unpleasant certainties are what we fear most of all. Indeed, fear ‘of’ may be so intense that we are beyond the use of that-clauses altogether. Our intellectual powers are for the time being paralysed. This is what happens in the extreme fear which we call ‘terror’.

There are other questions which would have to be considered in any complete account of the epistemology of hope and fear. For instance, can one both hope that x will happen and fear that it will happen? Presumably one can, if one believes that the happening of x would be a good thing in some ways but a bad thing in others. In the middle years of the fourth century a patriotic Christian might both hope and fear that Julian would be the next Emperor. Again, what should be said about the odd but morally respectable phenomenon of hope on behalf of another person? (‘I hope for his sake that he will win the prize.’) We may even have such hopes—and fears too—when the other person is fictitious and known by us to be so, for instance when we are reading a novel or watching a play. Indeed, we may go farther and ‘identify ourselves’ with the other person, as if his hopes and fears were our own.

It would not be relevant to discuss these questions here. Enough has been said to show that hope and fear are ways in which a person's beliefs are manifested. Hopes and fears are quite important ‘belief-indicators’. It is not true, then, that action is the only belief-indicator we can find, nor even that action and inaction are the only ones.

We can also notice something else which the Acting-as-if Analysis omits. Hope and fear, like other emotional states, are something more than modes of bodily behaviour. Fear is, no doubt, more ‘bodily’ than hope. At any rate, in fear ‘of’ there are very noticeable bodily symptoms, though they are less easily noticeable in fear ‘that’. But neither fear nor hope is reducible to a conjunction or series of outward and publicly-observable bodily states. Both have an inner aspect as well. We must feel them or ‘live through them’ at first hand if we wish to know what they are like. Moreover, if we wish to discover what a person is afraid of, or what his fears ‘that’ are concerned with, it is appropriate to go and ask him. His introspective report may not be decisive (there are what we call ‘unconscious fears’) but at any rate it is relevant evidence. And certainly it is relevant evidence if we wish to learn what he hopes for. Indeed, it is almost the only evidence we can get, since hope—unlike fear—does not show itself by any obvious bodily symptoms.

So if a person's hopes and fears throw some light, at any rate, on his beliefs, A's own introspective evidence is not, after all, quite irrelevant to the question ‘Does A believe that p?’ If we wish to answer that question, we may have to consider what goes on in A's own inner life. Here it is worth noticing that even action itself has an ‘inner aspect’ too. Even if belief were purely and simply a disposition to act as if a proposition were true, inward and private events would still have some relevance to the analysis of belief. Bodily movements do not count as actions unless the agent has the appropriate kinaesthetic sensations. Otherwise, his bodily movements would just be physical happenings, like the movements of a clockwork mouse across the table, and we could not say that he made them. But these kinaesthetic sensations are private experiences of his own. If he were to report to us that he did not feel any such sensations while his bodily movements were taking place, we could not say that he was acting as if such and such a proposition were true; he was not performing an action at all. Moreover, to know just what it is that he is doing we must know what his purpose is. We can often guess for ourselves what it is; but sometimes we guess wrong, and then we draw mistaken inferences about his beliefs (as in the example of the man who struck a match in the cellar, pp. 259–60, above). So there is some point in asking him what his purpose was. Here again his introspective evidence is not decisive (there are what we call ‘unconscious purposes’) but it is at any rate relevant to the question ‘what is he doing?’—or more vulgarly, ‘what on earth is he up to?’

Belief and Surprise

It is however conceivable that there might be intelligent creatures who neither hope nor fear. Some human creatures have wished, or professed to wish, to be ‘free’ both from hope and from fear; and if this wish were fulfilled, they would still be capable of holding beliefs. But there are two other feeling-states which are more closely connected with belief. They are surprise and doubt. Indeed, they are so closely connected with belief that we find it very difficult to conceive of a believing being who is incapable of feeling them. If there were such a creature, belief, for him, would be something very different from what it is for us.

We might describe surprise and doubt as ‘cognitive feelings’. Traditionally there is a third member of this peculiar class of feelings: the feeling of confidence. Its status is much more controversial, and its existence has sometimes been denied. But let us first consider surprise. No one will deny that we do sometimes experience surprise.

Perhaps it would be a little strange to say that surprise is an emotion, as hope and fear are. But it is certainly something that we feel. Sometimes it has organic repercussions too. At any rate, intense surprise can have them; we may gasp with surprise. And of all feeling-states, surprise is perhaps the one which is most closely connected with belief. It is the feeling which we have when some proposition which we believe turns out to be false. We also feel some surprise, though not so much, when we come across evidence against some propostition which we believe, even though this adverse evidence is by no means conclusive. If I believe that a colleague has gone away for the week-end, I am a little surprised to see a light in his rooms when I walk across the quadrangle on Saturday evening.

The connection between belief and surprise is so close that we sometimes express our belief that p by saying ‘I should be surprised if not-p’. We can also indicate the degree of our belief by specifying the degree of surprise we should feel if p were to be falsified. The lowest degree of belief, traditionally called ‘suspecting that…’, is indicated by saying ‘I should be just a little surprised if not-p’. The middle degrees (‘thinking that…’ or ‘having the opinion that…’) are indicated by saying ‘I should be a good deal surprised if not-p’. When we are very nearly sure but not quite, we say ‘I should be greatly surprised if not-p’. The highest degree of belief (conviction, being absolutely sure) is not usually indicated at all. When I am quite sure that we shall arrive by 7 o'clock, I usually just say ‘we shall arrive by 7 o'clock’, if I say anything. But perhaps you ask me whether I am absolutely sure about it; and then I reply ‘I shall be very surprised indeed if we don't’.

Moreover, if a person is surprised when a proposition p is falsified, this is about the strongest evidence we can have that he did, until then, believe the proposition for some period of time; and the degree of his surprise is about the strongest evidence we can have concerning the degree of his belief.

But are there perhaps two different kinds of surprise, not only the surprise of falsified belief, but also the surprise of sheer novelty? In the latter case, it might be said, what clashes with, or is removed by, the newly-ascertained fact is not a previously held belief, but sheer ignorance. The first European who observed a duck-billed platypus was no doubt much surprised to find a furry four-legged oviparous creature with a beak like a duck. (It is called ornithorhynchus paradoxus to this day.) Did he believe beforehand that no such creature existed, and was his surprise due to the empirical falsification of this negative proposition which he had, until then, believed? Surely it is more likely that he had never even conceived of such a creature at all? In that case he could not have believed beforehand that no such creature existed, and the surprise which he felt could not have anything to do with the empirical falsification of a proposition previously believed.

We have to admit, I think, that any belief of his which was refuted by these surprising observations was an ‘implicit’ or ‘tacit’ belief. The propositions ‘no furry four-footed creature is oviparous’ and ‘no furry four-footed creature has a beak like a duck’ were taken for granted by him, or assumed without question, without being explicitly assented to. But still, it was not true that he considered this strange creature with a completely open mind, as the phrase ‘sheer ignorance’ might suggest. His previous observations of various members of the animal kingdom, and what he had been taught about them or read about them, determined the way he ‘approached’ the creature which he saw. A child of two years old would not have felt surprise when he saw his first duckbilled platypus. For him, it would have been no odder than anything else. When everything is new, nothing is surprising.

It is, however, possible to resist or suppress one's surprise. That is what the old lady did who exclaimed ‘There's no such animal!’ when she was shown a giraffe at the zoo. If I interpret the story rightly, she embraced the hypothesis that she was having a visual hallucination. No doubt she was confronted by a choice of evils. But it was too much to have to give up her belief that all the species of quadrupeds in the animal kingdom were represented in pictures of the animals going into Noah's Ark (no dappled quadruped with a very long neck and two small horns on its head was among them). It was better and more comfortable to suppose that she was now having an hallucination. This is an excellent way of protecting one's beliefs against empirical refutation. The lady was strong-minded enough to make use of it there and then. We cannot always manage that. But a day or two afterwards, when the memory of some belief-refuting experience has faded a little, we can often persuade ourselves that we must have been dreaming when we had it.

Belief and Doubt

Surprise is something which we feel, and it is so closely connected with belief that the rather paradoxical phrase ‘cognitive feeling’ is quite naturally applied to it. Much the same could be said about doubt. This too is closely connected with belief; and since doubting, like surprise, is an inward experience which we ‘live through’, we are quite willing to say that we feel doubt about something, not merely that we have it. If we believe a proposition p, we feel surprised when it turns out to be false. But even when it has, as yet, been neither verified nor falsified, we feel doubt of other propositions which would be improbable if p be true. We do, of course, have to notice that they would be improbable if p be true, and conceivably we might fail to notice this. But sometimes it is very obvious. For instance, I have lost my umbrella, and I believe that I left it behind in the bus yesterday afternoon. Then I doubt the proposition ‘I shall get it back again’. I do not go so far as to disbelieve this proposition, not yet at any rate, since articles left behind in buses are very occasionally recovered by their owners. Doubting is not the same as disbelieving, though sometimes, for courtesy's sake, we may say to another person ‘I doubt it’ when in fact we firmly disbelieve it, as we may also say ‘I think so’ when in fact we are perfectly sure that it is so.

Still, there is a connection between doubting and disbelieving, and a feeling of doubt might be described as a felt inclination to disbelieve. But it is not a merely wishful inclination, or at any rate it need not be. There are indeed persons who enjoy doubting for doubting's sake, as there are others who find doubting almost unbearable. But if and so long as we do believe that p, it is reasonable to doubt another proposition q which would obviously be rendered unlikely if p be true. Doubt of q is something we are rationally committed to, so long as we do believe that p. It is, however, open to us to doubt p itself instead. For example, I am sure that it will continue to rain all day. Then at 1 p.m. it is reported to me that the wind has already veered a point or two. Either I must doubt this report (does my informant really know the difference between ‘veering’ and ‘backing’?) or else I must doubt whether the rain is going to continue all day; or at any rate, if I accept the report, I must ‘reduce’ my belief from a conviction to an opinion, and hold it with less confidence than I did.

We see from this example that there is a direct connection between doubt and surprise, as well as a connection between each of them and belief. What I doubt is also what would surprise me, if it were true. For our present purpose, however, the most important point is that the doubts which a person feels throw some light on his beliefs, as his surprise does also, and moreover that doubt, like surprise, is something inward and introspectible. It may sometimes manifest itself by hesitant action. But if we wish to know whether a person doubts a proposition, we usually have to go and ask him. This is true of the milder degrees of surprise too. Intense surprise (‘being astonished that…’) is likely to show itself by publicly-observable bodily symptoms; but even then we may have to go and ask the person what it is that he is astonished at.

The Feeling of Confidence

No one would deny that surprise and doubt are relevant to the analysis of belief; and it is quite natural to describe them as feelings. But what is to be said about the feeling of confidence? It has been thought that the feeling of confidence varies in degree, and that the different degrees of it may be used to define the different degrees of belief, from barely suspecting that… at the bottom end of the scale to complete conviction at the top, with the various degrees of opinion in between. On this view, the maximum degree of confidence is expressed by saying ‘I feel sure that…’. If we accept the Dispositional Analysis of belief, we cannot take over this traditional doctrine just as it stands. But the feeling of confidence might still be relevant to the analysis of belief, even though that analysis takes a dispositional form. It might still be true that anyone who believes a proposition has a feeling of confidence when he actually entertains that proposition; and moreover, that anyone who has a higher degree of belief with regard to p than he has with regard to q, feels more confidence when he actually entertains the proposition p than he feels when he actually entertains the proposition q. To put it in another way, the feeling of confidence which a person has when he actually entertains a proposition might still be an important ‘belief-symptom’, so important that if it were absent we should conclude that he did not really believe the proposition. (Cf. the absence of surprise when the proposition is falsified.)

But is there such a feeling at all? The view that there is has been severely criticized by some contemporary philosophers, for instance Professor William Kneale (Probability and Induction, ch, 2). We shall consider his criticisms later.2 But before we do so, it is worth while to mention one or two linguistic points.

On the traditional view, the highest degree of confidence is expressed by saying ‘I feel sure that…’. But we notice that in everyday English there is also a rather different use of ‘feel sure’. It is quite usual to say ‘I feel much more sure of this than of that’. Here there seem to be degrees within the feeling of sureness itself. Yet we might add ‘All the same, I do not feel absolutely sure of either’. And now feeling sure does not admit of degrees; instead, there are degrees of approximation to it.

This double usage of ‘feel sure’ might trouble a philosopher. But we can find analogues for it. ‘This line is straighter than that: all the same, neither of them is quite straight.’ ‘Your cup is fuller than mine, but yours is not quite full.’ ‘You were more punctual than I was, but neither of us was absolutely punctual.’ ‘This duster is much cleaner than that one, though neither of them is quite clean.’ We also notice that confidence itself is sometimes spoken of as if it did not admit of degrees. ‘How confident are you that it will go on raining all day?’ ‘I do not feel completely confident about it, but I think it will.’ For ordinary everyday purposes, we often do not find it necessary to distinguish very sharply between degrees of approximation to x and degrees of x itself. But for philosophical purposes we do need to distinguish between them; and it seems better (dare one say, more accurate?) to treat ‘confidence’ as something which does admit of degrees, and sureness as something which does not. The more confident we are, the nearer we are to sureness: sureness is the maximum degree of confidence. At any rate this is the usage I propose to adopt.

There is another linguistic point which is worth considering. We often speak of confidence as if it were a pro-attitude. It certainly is when we have (or feel) confidence ‘in’ another person. We may also have (or feel) confidence ‘in’ an instrument or a device. I have not much confidence in my watch—it is often five minutes slow—whereas I have a good deal of confidence in the College clock. Confidence ‘in’ is closely connected with belief ‘in’, a topic which we shall have to discuss later.3 For the present, however, we are only concerned with belief ‘that’. Are we to say that the confidence which enters into it is likewise a pro-attitude? And would it follow from this that whenever we believe a proposition, we wish it to be true? Is all believing, or all believing ‘that’, a kind of wishful thinking? Obviously it is not. However addicted we are to wishful thinking, we are often nearly sure, or even quite sure, that a proposition is true when we would much prefer it to be false.

But still it does seem a little odd to say ‘I feel very confident that we shall lose the game’ or ‘I feel pretty confident that the Judge will send me to prison for five years’. Do they hope to lose the game? Does the accused man want to be sent to prison for five years? The answer to both these questions might conceivably be ‘Yes’; and if it is, we shall no longer be puzzled. The other side have lost all their matches so far. It would be nice if they could win for once, and we are almost sure they will. There is something to be said for going to prison; it gives one an opportunity for meditation, and it is a way of atoning for one's offence. But such wishes are unusual, and what are we to say when they are absent—when there is no wish at all to lose the match or to be sent to prison?

Let us consider another example, an extreme example which makes the difficulty still more obvious. When the little boy is terrified of the bogey-man in the dark cupboard, should we say he feels confident or very confident that there is a bogey-man inside? He does feel nearly sure that there is, or even quite sure. But does he feel any confidence at all? We might be inclined to say that confidence is the last thing he feels.

There is indeed some connection between confidence and courage, or at any rate between confidence and boldness, the readiness to take risks or to ‘have a go’, cost what it may. Confidence is a manifestation of vitality (perhaps the little boy is rather deficient in that valuable quality) and a confident person is likely to be energetic and enterprising, for good or ill. But this is what we call self-confidence. It is a kind of trust in oneself or in one's own capacities, somewhat like the trust one may have in another person. It is related to belief ‘in’ and not to belief ‘that’. The highest degree of it is what we call ‘being sure of oneself’.

The little boy is far from being sure of himself. In this alarming situation, he has no trust at all in his own capacities. If we say of him ‘confidence is the last thing he feels’, it is self-confidence that we speak of, as we quite often do when the word ‘confidence’ is used. And though he has little self-confidence, or none at all, he may still feel very confident about the proposition ‘there is a bogey-man in the cupboard’. It is this propositional or ‘epistemic’ confidence which concerns us now, and not self-confidence; and it was this ‘epistemic’ confidence which the footballers expressed when they said ‘we feel confident that we shall lose the game’ and the accused man expressed when he said ‘I feel confident that the Judge will send me to prison’.

Kneale's Criticisms

But whatever complexities there may be in our ordinary use of the words ‘confident’ and ‘sure’, we are perfectly willing to speak of feeling confident (not only of being confident) and of feeling sure (not only of being so). We are also perfectly willing to speak of feeling more confident about one proposition than we feel about another. The traditional doctrine about a feeling of confidence which varies in degree is supported by the evidence of ordinary language, for whatever that support may be worth.

Nevertheless, it has been held that there is no such feeling, even though we do ordinarily speak as if there were, Let us now consider Professor William Kneale's discussion of this question (Probability and Induction,4 Part I, Section 4, ‘The Nature of Opinion’). He has made some formidable objections to the traditional doctrine. It is true that his main aim in this chapter is to criticize another traditional doctrine, the view that probability is to be defined in terms of degrees of confidence, whereas we are concerned with the doctrine that degrees of belief are to be defined in terms of them. Again, the distinction which he has in mind is the distinction between opinion and knowledge, whereas we are concerned with the distinction between opinion and conviction (being sure that…). There is still an important difference between opinion and conviction even when the conviction does not amount to knowledge, as it often does not. When I am, or feel, sure that p, p may in fact be false; and even if it be true, I may not have conclusive reasons for believing it. We are also concerned with the distinction between opinion and suspecting that…, which Kneale does not discuss, (Perhaps he would regard suspecting that… as the weakest variety of opinion.)

But although the doctrine I am discussing is not the main target of Kneale's criticism, he does make some shots at it, and it may well be thought that he hits the bull's eye. Confidence is supposed to be something which we feel. But he himself can discover no such feeling when he is rationally opining (p. 14). Perhaps we need not discuss the rather unplausible view which he then proceeds to criticize. It is concerned with the measurement of probability. Probabilities are measured by fractions; but can we really suppose that for every degree of probability between 0 and 1 there is a corresponding degree of felt confidence? (p. 15). Degrees of belief might still be definable in terms of degrees of confidence even though degrees of probability are not. And if we do wish to define degrees of belief in this way, we are not compelled to suppose that the series of degrees is continuous. The degree of felt confidence, and the degree of belief likewise, might rise by finite jerks or jumps as we pass along the series, from suspecting that… at the bottom to conviction at the top.

On the same page, however, Kneale makes a remark about knowledge which is very relevant to the question we are discussing. ‘It is obvious that knowledge itself is not accompanied by confidence. When we realise that 2 + 2 = 4, we do not sweat with any feeling of supreme intensity’ (p. 15, ad fin.). Here again, what Kneale has in mind is the contrast between knowledge and rational opinion. But what he says about knowledge could be applied to any conviction, whether it amounts to knowledge or not. Whatever proposition we are sure of, whether true or false, a priori or empirical, whether our sureness is reasonable or unreasonable, do we find ourselves ‘sweating with a feeling of supreme intensity’ when we entertain that proposition? Conceivably we might, when we have just come to be sure about something, like Archimedes when he rushed out of the bath shouting ‘Eureka!’ But then it is the coming to be sure, perhaps after a long period of doubt and unsuccessful enquiry, which has this strong feeling-tone, accompanied very likely by organic sensations, as the phrase ‘sweating with’ suggests. The sureness, once we have it, and have got used to having it, is no longer manifested in this exciting way.

Why then do we speak of ‘feeling confident’? Kneale's answer is this: ‘When we speak, as we admittedly do, of feeling confident, we are referring, I think, to the absence of serious doubt or questioning from our minds, much as when we speak of feeling tranquil we are referring to the absence of uneasiness’5 (p. 16). And he agrees that doubt itself can properly be described as something which we feel, since there is at any rate a feeling-element in it, ‘a feeling of frustration and restlessness’ (ibid.)

As we have seen, Kneale is mainly concerned here with degrees of probability. But what he has said suggests a way of reformulating the traditional doctrine about degrees of belief: or rather, perhaps, a way of turning it upside down. Where the traditional doctrine speaks of degrees of confidence, should we speak of degrees of doubt instead? A person who merely suspects that p feels a good deal of doubt about the proposition p, when he actually entertains the proposition. In the middle degrees of belief (traditionally called ‘opinion’) he feels some doubt about the proposition p, but not so much. Within opinion itself there are degrees, and he might express these by saying how much doubt he feels: some doubt, a little doubt, and just a very little doubt when he is nearly sure but not quite. Finally, the highest degree of belief (conviction, being quite sure) would be expressed by saying ‘I do not feel any doubt at all that p’. Conviction, then, so far from being a feeling of supreme intensity, would be the absence of a feeling: the complete absence of any feeling of doubt. In this way, we might hope to dispel the mystery which has somehow gathered round the traditional doctrine of ‘a feeling of confidence which varies in degree’. Feeling more confidence is turned into feeling less doubt, and feeling less confidence into feeling more doubt. And the supposed supreme degree of confidence, conviction, just consists in feeling no doubt at all.

Confidence and Doubt

This is a very attractive proposal. Doubt is something which is perfectly familiar to everyone. Doubt is something which we feel, or at any rate it is something which we experience or ‘live through’. We all know what it is like to doubt, and some of us enjoy it. How happy we should be if everything we say about confidence could be said, and said more clearly, by talking about doubt instead!

But can it? Let us consider conviction. Believing a proposition p with conviction (being completely sure that p) cannot just consist in not having any feelings of doubt about p. There are many propositions which a person feels no doubt about, because he has never considered them at all. If you have never heard of Great Snoring, you do not feel any doubt about the proposition ‘Great Snoring is a village in Norfolk’.

Indeed, there are propositions about which one cannot feel any doubt, because one cannot even entertain them. (It is worth while to mention this, because a state of conviction or complete sureness is sometimes expressed by saying ‘I cannot doubt that…’.) For instance, there are many propositions in Higher Mathematics which I myself cannot doubt, because I cannot at all understand the technical symbols in which they are formulated. The most I can manage to be convinced of is that some necessary truths or other are formulated in this (to me) unintelligible manner.

We encounter other difficulties when we consider the lower degrees of confidence, those which we are supposed to feel in suspecting and in opinion. Ordinarily we are quite content to say ‘I feel more confident about p than about q’. For instance, I feel more confident about the proposition ‘it will rain some time today’ than I feel about the proposition ‘it will rain within the next hour’. Are we to say, instead, that I feel less doubt about the first proposition than I feel about the second? It is true that I do. But if we stop at that point in our analysis of ‘feeling more confident… than…’ we have left something out. For here again, I might quite well feel less doubt about p than I feel about q without believing either of them. I greatly doubt the proposition ‘some human being will visit Sirius some day’. I also doubt the proposition ‘some human being will visit the planet Neptune some day’ and I doubt it a good deal less. But I do not believe either of them. It may be that it would be reasonable for me to believe the second one, at any rate with the lowest degree of belief (‘suspecting that…’), if I were better informed about the technology of space-travel. But as it is, I doubt the first more than the second without believing either of them. In the same way, a sceptical person might doubt the existence of telepathy and he might doubt the existence of telekinesis still more, without believing that either of them exists.

It seems, then, that degrees of belief cannot be defined in terms of degrees of doubt. It is indeed true, and relevant, that when we are more nearly sure of p than of q, we feel less doubt about p than we feel about q. But ‘being more nearly sure of…’ cannot be equivalent to ‘being disposed to feel less doubt about…’; and ‘being quite sure of…’ cannot be equivalent to ‘being disposed not to feel any doubt about…’. Something relevant is being said when these analyses are offered, but something else which is equally relevant is being omitted.

It is not just that there is less doubt in some cases than in others, and that in complete conviction (being absolutely sure that…) there is no doubt at all. There is something else which counteracts the doubt. In conviction this anti-doubting factor is completely victorious. Doubt is squeezed out altogether. In opinion, it is not squeezed out altogether. Some doubt is still felt, even in the higher degrees of opinion, when the ‘opined’ proposition is actually being entertained; but the anti-doubting factor has the upper hand. It still has the upper hand in suspecting that…, but only just. Its victory is ‘a very near thing’. The victory may be short-lived too. Suspecting is a less stable state of mind than opinion. We easily slip out of it into the neutral state of suspense of judgement. Then we just wonder whether p is true or false. Sometimes we go farther, and being to suspect that not-p, whereas we had previously suspected that p.

What is this anti-doubting factor? It is exactly what we call ‘confidence’, and it has to be mentioned, under some name or other, in any tolerable analysis of the degrees of belief. If, or to the extent that, our believing is reasonable, the degree of confidence varies with the strength of the evidence we ourselves have for the proposition believed, while the degree of doubt varies with the strength of the evidence we ourselves have against the proposition. (As has been pointed out before, it is the evidence we ourselves have which is relevant here, not merely the evidence which ‘there is’, nor the evidence which other people have.)

But does it follow from this that confidence is something which we ‘feel’, as doubt is? Let us confine ourselves, for the moment, to the lower degrees of belief, from suspecting that… at the bottom to being nearly, but not quite, sure at the top (that is, to beliefs which fall short of conviction). There is a sense in which all of them are conflict-situations. At any rate, this is true of coming to suspect that… and coming to have the opinion that… And this conflict is something which we feel; it is even agonizing sometimes. If we do eventually arrive at a more or less stable state of suspecting that… or of having the opinion that…, the conflict is terminated. But a certain state of tension remains between the two opposing factors, doubt and confidence. One side has gained ground, the other has lost it. But both still retain their hostile posture, because neither has achieved a decisive victory. Each is still exerting pressure on the other. It is this state of tension which we ‘feel’ when we suspect that p or have the opinion that p; and we still feel it a little in the highest degree of opinion (when we are very nearly sure, but not quite). We can describe our condition either by saying how much confidence we feel or by saying how much doubt we feel. But if we are to be explicit we ought to mention both.

This is the conclusion we reach if we use the word ‘feel’ in an emotion-like sense. Now let us consider conviction. In this sense of the word ‘feel’, it is true that in the state of conviction (being completely sure that…) there is no feeling at all. There may well have been some while we were coming to be sure. But once we are sure, and so long as we remain so, there is no feeling at all, in this sense of the word ‘feel’. Being sure is not a conflict-situation, and there is no felt tension in it.

But the word ‘feel’ also has a wider sense than this. It is applied to any mental state or mental process which is introspectible. What is felt, in this sense, is what can be (though it need not actually be) an object of introspection; what we feel is just something which we ‘live through’. We may recall Samuel Alexander's distinction between ‘enjoyment’ and ‘contemplation’. ‘Feeling’, in this sense, is equivalent to what he called ‘enjoying’. Emotions themselves are often lived through, though not always, if there are what we call unconscious emotions (e.g. unconscious fears). But they are certainly not the only mental states or happenings which are lived through.

Now is being sure something which can be felt in this sense of the word ‘feel’ (the ‘living through’ sense) though in the emotion-like sense it cannot? I suggest that it is. As we have seen, in ordinary everyday speech we are perfectly content to speak of ‘feeling sure’ as well as ‘being sure’. If you ask me whether I feel sure that it is going to rain, and I reply that I do, are we bound to suppose that my reply is false? It would have to be, if the truth is that I have no feeling at all about the proposition ‘it is going to rain’ in any sense of the word ‘feel’.

Sureness is indeed the absence of a feeling, if the word ‘feel’ is used in its emotion-like sense. Sureness is a state of mind in which conflict or tension is absent. But it might still be a state of mind which is lived through or ‘enjoyed’ by the person who has it; and that, I suggest, is the sense which the word ‘feel’ has when a person says ‘I feel sure’. (Cf. ‘Did you feel very frightened?’ ‘No, oddly enough I felt perfectly calm at the time’.) Feeling sure is rather like feeling calm, or feeling tranquil as Kneale puts it.7 But he is mistaken in supposing that feeling tranquil is just the absence of uneasiness. It is a positive state which we sometimes ‘live through’. Should we wish to say that when a man feels well, the truth is only that he does not feel at all ill?

• 1.

  Cf. J. Harrison Christian Virtues, Aristotelian Society Supplementary Volume, 1963, pp. 80–1, and J. M. O. Wheatley, ‘Wishing and Hoping’, Analysis, June 1958.

• 2.

  Pp. 382–5, below.

• 3.

  See Lecture 9, below.

• 4.

  Oxford University Press, 1949.

• 5.

  Kneale himself does not wish to define opinion in terms of doubt, though he agrees that ‘anyone who opines feels some doubt about what he opines’ (op. cit., p. 16). I think this is because he is concerned only with the distinction between rational opinion, on the one hand, and knowledge on the other; and not with the distinction between opinion, whether rational or not, and conviction whether rational or not.

• 6.

  On ‘unconscious’ beliefs, see below, Lecture 3, pp. 290–301.

• 7.
  Probability and Induction, p. 16.