The Evidence of Testimony
Epistemologists do not seem to have paid much attention to the evidence of testimony. But according to our ordinary way of thinking, testimony is one of our most important sources of knowledge. Everyone claims to know a very large number of geographical and historical truths, for which he has only the evidence of testimony. All of us here would claim to know that China is a very large and populous country, though none of us, perhaps, has been within three thousand miles of it. Every English schoolboy knows that Charles II was restored to the throne in 1660, and that Britain was once part of the Roman Empire, though these ‘known facts’ are facts about the remote past. The same applies to facts about the very recent past as well. If an important debate takes place in Parliament, millions of people claim to know about it next day, just by reading newspapers or listening to the wireless.
To take an even more striking example, each of us would claim to know how old he is, that is, how many years have elapsed since he was born. But he has only the evidence of testimony to assure him that he was born in such and such a year; and equally he has only the evidence of testimony to assure him that this present year is 1967. How do I know, or what grounds have I for believing, that to-day is January 23, 1967? If it is a case of ‘being sure and having the right to be sure’, I have acquired this right by reading what is written on a calendar, or at the top of the front page of the newspaper which was delivered at my home this morning. I am often uncertain what day of the week it is. Is it Wednesday or Thursday? But in my ordinary unphilosophical moments, I assume that this question can be conclusively settled by consulting the appropriate written sources, such as to-day's newspaper.
Indeed, each of us depends on testimony for almost all that he claims to know about anything which is beyond the range of his own first-hand observation and memory; and one of the most important functions of memory itself is the remembering of what we have learned from other people by means of speech and writing.
This reliance on testimony plays a fundamental part not only in our cognitive lives, but in our practical lives as well. In very many of our practical undertakings we depend in one way or another on the spoken or written information we receive from others. To catch a train, I must rely on the information I receive from the timetable. To catch a bus, I must rely on the written words which I see on the front of it, e.g. the word ‘Paddington’, which is an abbreviation for ‘This object goes to Paddington’. I find my way about the country by relying on the written testimony of sign-posts and milestones, and if I use a map instead, I am relying on testimony too. It could even be said that we are relying on testimony whenever we use a ruler or a tape-measure to measure the length of something, e.g. the testimony of the manufacturer that this stretch on the ruler's edge is 8½ inches long.
But now let us imagine a detached and very reasonable observer of the human scene, brought up on Locke's principle that one must never believe any proposition more firmly than the evidence warrants.1 Would he not think that our ordinary attitude to testimony is absurdly credulous? Of course, no one believes everything that he is told, nor everything that he reads; still less does he always believe it with complete conviction. But in nine cases out of ten we do give at least some credence to what we are told or what we read. There is of course the tenth case. The answer to some very important question, practical or theoretical, may depend on the correctness of so-and-so's testimony. Then we shall be more cautious. Or the event which he describes to us may be very improbable in the light of all the other relevant evidence which we have, for example, if he tells us that a Flying Saucer landed in the University Parks half an hour ago. Or perhaps it may be very much in the testifier's own interest that we should believe him; for if we do, we shall give him the money for which he asks. Or again, we may have found on many previous occasions that he himself or others like him (beggars for example) made statements which turned out later to be false. Even so, we are usually prepared to give some weight to the testimony which is offered to us, and it only happens very seldom that we just reject it ‘out of hand’. Even habitual liars and romancers tell the truth sometimes. Improbable events do happen. Even when it is to the speaker's own interest that his statement should be true, it does sometimes turn out to be true all the same.
The principle which we follow in the great majority of cases seems to be something like this: What there is said to be (or to have been) there Is (or was) more often than not. And that is why our disinterested Lockean observer might think us absurdly credulous. I do not think we extend this principle to the future (‘What there is said to be going to be, there will be, more often than not’). We are not quite as credulous as that. We all know how unreliable even our own first-hand predictions can be. But even though we limit our principle to testimony about the present and the past, what justification can we have for accepting it? Must we not admit that our ordinary attitude to testimony is indeed unreasonably credulous?
But suppose we adopted the incredulous or non-credulous attitude which our disinterested Lockean observer would presumably recommend. Let us try to imagine a society in which no one would ever accept another person's testimony about anything, until he had completely satisfied himself about the bona fides of that person, his powers of accurate observation and capacity for recalling accurately what he had observed. Such a society would hardly be a society at all It would be something like the State of Nature described by Hobbes, in which the life of every man is solitary, poor, nasty, brutish and short. What our Lockean observer calls credulity is a necessary condition for social cooperation. What he calls credulity is not only in the long-term interest of each of us. It has a moral aspect too. If some people make a virtue of accepting testimony so readily, they are not wholly mistaken. Am I treating my neighbour as an end in himself, in the way I wish him to treat me, if I very carefully examine his credentials before believing anything he says to me? Surely every person, just because he is a person, has at least a prima facie claim to be believed when he makes a statement? This claim is not of course indefeasible. But it might well be argued that we have a duty to trust him unless or until we find pretty convincing reasons for mistrust, and even to give him ‘the benefit of the doubt’ if we have some reasons for mistrusting him, though not conclusive ones.
A Conflict between Charity and the ‘Ethics of Belief’
Perhaps some religious moralists may think that these conflicts are illusory. Does charity really require that one should shut one's eyes to the facts or the empirically-supported probabilities? If it is an ‘unconditional pro-attitude’ towards other people, the knowledge or reasonable belief that your neighbour's statements are false will do nothing to weaken it—not even if you think that he is trying his best to deceive you or mislead you. ‘He is rather a liar, of course, and I can't believe a word of it, but he is a good fellow all the same.’ If you are a charitable person, you are supposed to have a pro-attitude towards him as he is, faults and all (‘What he is’ of course includes his capacities for becoming better). Surely it is at least logically possible to be at once charitable and clear sighted about the defects, intellectual or moral, of one's neighbours?
This combination of qualities is indeed logically possible, and is even occasionally achieved in some very admirable persons. It may well be true that the conflicts we have mentioned arise only for those who are trying to be charitable without yet being so, or are trying to be less uncharitable than they have hitherto been. But unfortunately this is the position in which many of us are most of the time; and then, for us, there can quite well be a conflict between the precepts of charity and the precepts of the Ethics of Belief.
As has been pointed out already, the study of belief is one of the regions where epistemology and moral philosophy overlap. But epistemological questions are our main concern at present.
‘Accept What You Are Told, Unless You See Reason To Doubt It’
Whatever degree of charity we have, be it great or little, our ordinary practice is to accept what we are told unless or until we see reason to doubt it. We do seem to follow the principle ‘What there is said to be (or have been) there is (or was) more often than not’. There are of course certain special occasions when the principle is temporarily switched off as it were, or put into cold storage for a while. When someone says ‘I am now going to tell you a story’ he is warning us that what he is about to say is not to be taken as testimony. The principle ‘What there is said to have been there was, more often than not’ is to be ignored for the time being. A similar switching-off occurs when we begin to read a book which we have borrowed from the section marked ‘Novels’ in the library. The sentences uttered by characters in a play are not to be taken as testimony either, though many of them have the form of statements about empirical matters of fact, for instance, ‘The wind bites shrewdly, it is very cold’. We are not to take this as a weather-report. But the important point to notice about these occasions is that they are special and exceptional ones. Special conventions and devices have to be used to convey to us that for the time being the principle we ordinarily follow is not to be applied. And that principle is something like ‘What there is said to be, or to have have been, there is, or was, more often than not’.
To follow this principle may be socially expedient or even socially indispensable. It may be charitable too. But considered in a cool hour, in the way our detached Lockean observer would consider it, it is a very curious principle indeed. It is not even easy to decide what kind of a principle it is. We are at first inclined to suppose that it is itself a proposition which we believe, something which is itself either true or false. In that case, we cannot dispense ourselves from the task of asking what grounds we have for claiming that it is true, or at any rate more probable than not. Certainly it is neither self-evident nor demonstrable. It looks like an inductive generalisation. If that is what it is, no one of us is entitled to believe it unless he himself, by his own personal observation, has been able to verify at least some of the testimony which he has received from others.
Such first-hand verification is in a way a wasteful procedure. No one would wish to verify all the statements he hears or reads, even if he could. The whole point of testimony is that it is a substitute for first-hand experience, or an extension of first-hand experience, whereby each of us can make use of the experiences which other persons have had. When a piece of testimony is verified by our own first-hand experience, we no longer need it. I want you to tell me what happened on the other side of the hill because I was not able to go there and see for myself. I want Tacitus to tell me what happened in the reign of the Emperor Claudius because I was not alive at the time. The testimony which I cannot verify for myself is the testimony which I need to have. I might as well not have had it, if I can find out for myself that what you tell me is true. On the other hand, unless I can find out for myself that at least some of the things I am told are true, the testimony I receive is equally useless to me, because I have no ground for believing that any of them are true.
The reasonable plan might seem to be that each person should test for himself a not very large part (one tenth perhaps?) of all testimony he receives, thereby rendering that part useless as testimony; and then he would be able to estimate what degree of confidence, great or little, he is entitled to have concerning the remaining nine-tenths. He would sacrifice a little of it in order to be able to use the rest to supplement and extend his own very limited first-hand experience. At least, this would seem to be the reasonable plan if the principle we are discussing (‘What there is said to be (or to have been) there is (or was) more often than not’) is indeed an inductive generalization.
Now certainly each of us is sometimes able to ‘check’ the testimony which he receives from others. In a strange town I have often had to ask a passer-by where the nearest post office is, or where the railway station is; and usually (though by no means always) I have found for myself that the information given to me was correct.
But when we consider the enormous number and variety of the beliefs which each of us holds on the evidence of testimony alone, it is obvious that the amount of first-hand confirmation he has is tiny indeed in comparison. It is nothing like large enough to justify the generalisation ‘what there is said to be, or have been, there is, or was, more often than not’. In a very simple and primitive society, where no one can read or write or listen to the radio, the situation would be easier. Then, if someone tells me that there was a wolf sitting beside the village well at midnight, I can go to the well myself this morning and see what look like the footmarks of a wolf in the mud beside it. But in a civilised and highly-educated community it is another matter. We have only to consider the vast mass of historical propositions which every educated person believes. I can personally remember a few of the events that happened in the reign of King Edward VII. But I certainly cannot remember anything that happened in the reign of King Edward the Confessor, to say nothing of the reign of Hadrian or Septimius Severus. Yet I do very firmly believe that both these Emperors visited Britain, and that Septimius Severus died at York. I hold this belief on nothing but the evidence of testimony. The best I have managed to do by way of ‘checking’ the testimony for myself is to read the Latin text of what I am told is a copy of the Historia Augusta; and this hardly amounts to first-hand verification.
Suppose however that we re-stated our principle in a much weaker form: ‘What there is said to be (or to have been) there is (or was) in at least one case out of every five.’ This seems a very modest principle, even a rather sceptical one. But if it is supposed to be an inductive generalisation, the evidence which any one person has for believing it would still be quite insufficient. In a civilized and literate society, the amount of testimony which each of us has been able to test and verify for himself is far too small to justify any inductive estimate of the ‘overall’ reliability of testimony in general: too small, that is, in relation to the enormous number and variety of all the beliefs he has, which are supported partly or wholly by testimony spoken or written, or conveyed in other ways (for example, by means of maps). Whatever estimate any one person tried to make of its reliability, whether favourable or unfavourable, he would not have nearly enough first-hand evidence to justify it. Indeed, the habit of accepting testimony is so deep-rooted in all of us that we fail to realize how very limited the range of each person's first-hand observation and memory is.
‘First-Hand’ and ‘Second-Hand’
Before we consider whether there is some other way of interpreting our principle, it may be worth while to say something about the contrast between ‘first-hand’ and ‘second-hand’. We contrast the first-hand knowledge which each of us acquires by means of his own observation, introspection and memory with the second-hand knowledge (or beliefs) which he acquires by means of testimony. Yet it is important to notice that there is, after all, something first-hand about the acceptance of testimony itself. Testimony has to be conveyed to us by means of perceptible events or entities—audible or visible words, or sometimes visible signs or gestures; or occasionally by tangible means, as when we converse with a deaf and blind person by ‘tapping out’ words on his hand. And the person who accepts the testimony must perceive these perceptible events or entities for himself. Unless he has this first hand experience, he cannot accept the testimony, nor even reject it.
It is true of course that I may believe what John said although I did not hear him say it. But then some third person, Bill for instance, must tell me that John said so-and-so; and I must still hear for myself what Bill tells me, or see it for myself if he tells me in writing. How ever many hands or mouths John's story has passed through before it reaches me, it will not reach me at all unless there is some first hand perceptual experience of mine at the end.
The acceptance of testimony is first hand in another way as well. Testimony has to be understood by the person who receives it, and he must understand it for himself. No one else can understand it for him. If it is offered to him in a foreign language which he does not know, or in a technical terminology which he cannot follow, he may ask someone else to translate it or interpret it for him. But he still has to understand for himself what the translator or interpreter tells him.
Sometimes this second condition is fulfilled, but the first—the perceptual one—is not. This happens quite frequently in dreams. In our dream we may have copious and complicated mental imagery, either visual or auditory; quite a complicated story may be presented to our minds, and often we understand it perfectly. We do not just dream that we understand it. Dreaming that one understands what one does not in fact understand can also happen. But I wish to consider the case where we really do understand the words which present themselves to us in our dream—and surely it is not at all an uncommon case. Very often these words are combined into sentences which preport to describe some empirical matter of fact. Sometimes it is as if we just heard them without even seeming to see anyone who utters them.
These sentences, however, do not count as testimony, not even if they turn out subsequently to be true, as they may if the dream is a telepathic of clairvoyant one. They do not count as testimony, because the words composing them are not physical sounds or physical marks which we perceived. The first hand experience we had, when they presented themselves to us, was not an experience of perceiving, but of imaging, though we were not aware of this at the time.
Much the same could be said of waking hallucinations. Suppose you are a motorist trying to find your way to a village in a remote part of Norfolk. You have a visual hallucination of a signpost with the words ‘Great Snoring 2½ miles’ written on it. What is written on signposts is a form of testimony. But they have to be real physical signposts with real physical letters written on them. Hallucinatory words on an hallucinatory signpost do not count as testimony, though here again it is conceivable that these hallucinatory words do describe what is actually the case. Perhaps you do turn down the little lane which appears to be indicated by the signpost there appears to you to be, and perhaps you do arrive at Great Snoring after 2½ miles. Students of paranormal cognition might then suggest that the visual hallucination was a paranormal experience, the manifestation in consciousness of an unconscious ‘extra-sensory perception’ of the whereabouts of the village. But whatever we think of this explanation, we can hardly say that you accepted (and acted upon) a piece of testimony, as you would have been doing if the signpost had been a real one, perceived in the normal manner.
The conclusion we must draw is that the testimony received by a particular person can never be more reliable than the first hand experience by means of which he receives it. To be sure, when I am delirious the doctor may tell me that the spoken or written sentences which I claim to hear or to see are not to be taken as testimony, and that I am not to believe them or be at all worried or frightened by them. But then he himself is giving me testimony about the non-testimonious character of these hallucinatory words. Perhaps I believe what he tells me. But I am not entitled to believe it, unless I am first entitled to believe that the sounds he appears to be uttering really are what they appear to be—events in the public and physical world. If I suspect that the words I seem to hear him utter are hallucinatory too, I must also suspect that they are not to be relied on. It comes to this: before I am entitled to believe what someone is saying, I must make sure that it is really being said; and before I am entitled to believe what I read I must make sure that the written words really are there on the page or the signpost or the noticeboard.
Furthermore, one is only entitled to rely on testimony if one understands it correctly. When an English traveller in Italy sees the word ‘calda’ written on a water-tap he may think it means ‘cold’ and act accordingly, with unfortunate results. There is nothing wrong with his perceptual capacities. He is not having a visual hallucination or illusion, and he is not dreaming. The word ‘calda’ is really there on the tap just as it appears to be. But he misunderstands the testimony which is offered to him, and is therefore mistaken when he relies on it. Moreover, the understanding which is required may have to be something more than the mere ‘dictionary-knowledge’ which would have been quite sufficient in this example. One may need to be able to ‘interpret’ what is said or written. If the caller is told at the door that Mrs So-and-so is not at home, he may misinterpret what is said to him, and believe mistakenly that the lady is not in the house, though he understands the dictionary-meaning of the words perfectly well. The testimony which we accept can never be more reliable than our own capacity for understanding the words (or signs or gestures) by means of which it is conveyed to us.
If we prefer to put it so, our lack of understanding may prevent us from discovering what the testimony conveyed to us actually is, and what we believe will then be something different from what the speaker or writer was telling us. Viewed in this way, our lack of understanding has the same kind of results as an illusion or hallucination might have. If an Italian were to have an hallucination of the word ‘fredda’ when he looked at the water-tap, the results for him would be much the same as they were for our English traveller, who thought that ‘calda’ meant cold.
If we accept a proposition p on the evidence of testimony, this evidence can never be stronger than our evidence for believing (1) that certain words have in fact been spoken or written (2) that p is in fact the proposition which these words convey. And our evidence for these two beliefs has to be first-hand. Each of us must hear or see for himself, and no-one else can do it for him; and each of us must understand for himself if he understands at all. It is true that someone else may have to explain to me what a particular word or sentence means. But I still have to understand this explanation for myself.
No doubt these considerations are perfectly obvious. It is surely undeniable that there is something first hand about the acquisition of any belief whatever, whether it is based on testimony or not. But we tend to forget this, if we lay great emphasis on the concepts of public verifiability and public knowledge. We speak as though there were some formidable entity called ‘the public’ (or perhaps we call it ‘Science’ with a capital S). But there is no such entity, and if there were it could not know anything or verify anything. There are only human beings who co-operate with one another. One very important way in which they do it is by giving each other testimony. I tell other persons that I have verified a proposition p, and they tell me that they have verified it too. But each of us has to do his verifying for himself at first hand, however important it is for each of us to learn that others have done it.
What has now been pointed out is that there is also something first hand about this ‘learning that others have done it’, since each of us must hear or see for himself what others say or write, and each must understand for himself what they are telling him.
Another Interpretation of the Principle
We may now return to the principle ‘What there is said to be (or to have been) there is (or was) more often than not’. Hitherto we have assumed that this principle is itself something which we believe; and if we interpret it in that way, we shall have to suppose that we believe it very firmly, perhaps even with complete conviction. At any rate, each one of us seems to be guided by it all the time. Of course, we do not believe that what there is said to be or to have been, there always is or was; still less do we believe this with complete conviction. The qualification ‘more often than not’ is an essential part of the principle itself. But even though this qualification is put in and borne in mind, it is difficult to see how such a belief could be reasonable. This is because the degree of our belief (if it is indeed belief) would still be far greater than our evidence justifies.
When we consider the vast amount of testimony which each of us accepts, especially if he is a civilized and educated person, we find that each of us is only able to verify a very small part of it by means of his own first-hand observation and memory. Perhaps there might be enough first-hand verification to justify him in suspecting or surmising that what there is said to be or have been there is, or was, more often than not. (‘Suspecting’ and ‘surmising’ are traditional names for the lowest possible degree of belief.) But any higher degree of belief than this would surely be too high, and could not be justified by the amount of first-hand confirmation each of us can get for the vast mass of statements, on all manner of subjects, which he hears or reads. Even if someone were to say ‘I think, without being at all sure, that what there is said to be (or to have been) there is (or was) more often than not’—expressing a not very confident opinion—he would still be unreasonably credulous. Credulity may be socially expedient, even socially indispensable, and it may be charitable. But it is credulity still.
There is however another way of interpreting the principle we are discussing. Perhaps it is not itself a proposition which we believe, still less a proposition believed with complete conviction. Instead, it may be more like a maxim or a methodological rule. In that case, it is better formulated in the imperative than the indicative mood. We might put it this way: ‘Believe what you are told by others unless or until you have reasons for doubting it.’ Or we might say ‘Conduct your thoughts and your actions as if what there is said to be (or to have been) there is (or was) more often than not’. If this is what our principle is, we no longer have to ask what evidence there is for believing it, because it is not itself something believed. It does of course concern believing, and could be described as a policy for forming beliefs. But a policy is not itself believable, since it is not itself either true or false. We could perhaps say that we believe ‘in’ it, in the sense in which some people believe in Classical Education and others believe in taking a cold bath every morning before breakfast. But believing in a policy or procedure is very different from believing that something is the case.
All the same, the adoption of a policy does have to be justified. We cannot ask what evidence there is for it (since it is not something which is either true or false). But we can ask what reasons there are for adopting it; or if it never was consciously adopted, we can ask what reasons there are for retaining it, once we have reflected on it and have noticed what sort of a policy it is. And there are in fact pretty cogent reasons for adopting, or retaining, this particular policy. They are economic reasons (in rather a broad sense of the term) because they are concerned with the intelligent use of scarce resources.
From one point of view, the scarcity from which each of us suffers is a scarcity of first-hand observations, or more generally of first-hand experiences. One of our misfortunes is that no human being is ubiquitous. If I am in Oxford at 10 a.m. this morning, I cannot directly observe what is going on in Newcastle at that time. Indeed, as I sit here looking out of the window, I cannot directly observe what is going on behind the thick screen of thorn-bushes ten yards away. Still less are we ‘ubiquitous in time’, if such a phrase is allowable. Each of us has some first-hand access to the past by means of his own memories. But the span of time which they cover is very limited. I certainly cannot remember the Norman Conquest, nor even what happened in the year 1890. Moreover, the past which each of us remembers is only his own past, what he himself perceived or felt or did on various past occasions, and what he then came to believe or came to know. He does not even remember the whole of his past, in the sense of being able to recall any part of it he pleases, though it is possible that in some subconscious or unconscious way he ‘retains’ a good deal more of it than he can now recall. And in the past which he is able to recall, he suffered from the same scarcity of first-hand observations as he does now. He cannot recollect even one occasion when he was able to see through a brick wall, or touch something from a distance of two hundred yards.
But ‘scarce’ is of course a relative term. After all, each one of us has had a good many first-hand experiences, and goes on having them all the time so long as he is alive and awake. If we call them ‘scarce’, we must have some good or end in mind, and we are pointing out that no one individual has enough of them to enable him to achieve it. What is this good or end? It is knowledge, which is something we desire for its own sake, and also as a means for achieving other goods which we desire.
To put it in another way, there are many questions which each of us desires to settle or to answer; and no one person has anything like enough first-hand evidence to enable him to settle more than a very few of them. Each of us would like to know what happened before he was born, and what is happening now on the other side of the wall. His own first-hand observations and his own first-hand memories will not enable him to answer these questions. If he cannot know the answers to them, he would still like to be able to hold the most reasonable beliefs that he can, on the best evidence he can get. And very often indeed the only evidence he can get is the evidence of testimony. He must either accept what others tell him, for what it may be worth; or else he must remain in a state of suspended judgement, unable to find any answer at all to many of the questions which he desires to answer.
The economic aspect of this situation comes out in another way when we notice that there is an exchange of testimony between different individuals. I have some information which you need but do not at present possess, and you have some which I need but do not at present possess. I tell you what I have observed and you have not; and you tell me what you have observed and I have not. This exchange is advantageous to both of us. Though both of us suffer from a deficiency of first-hand knowledge, I can do something to remedy your deficiency and you can do something to remedy mine. But in a way this is something better than an exchange. For I do not lose the information which I give you, as I should if I gave you my hat; and you do not lose the information which you give me in return, as you would if you gave me your umbrella. Both of us gain and neither of us loses.
As has been mentioned already, this exchange of testimony has a moral aspect too. The information you give me will be useless to me, or worse than useless, unless you give it honestly. And if I trust you to give it honestly, you too must be honest in giving your information to me. One way of formulating the policy we are discussing is ‘Accept what you are told by others unless or until you have specific reasons for doubting it’; and this is closely related to the moral rule ‘Trust your neighbour unless or until you have specific reasons for distrusting him’. Prudence and charity go hand in hand here. Or at least they go hand in hand some of the way, though a seeker after knowledge would probably stop giving his neighbour ‘the benefit of the doubt’ rather sooner than charity would recommend.
But at present it is the prudential or economic aspect of this exchange of testimony which mainly concerns us. The moral aspect of it is only relevant in so far as the policy we are considering will not in fact succeed, unless there is at least a modicum of honesty and mutual trust among those who practise it. In a community of incorrigible liars or incurable romancers, the exchange of testimony would not help very much to solve the problem which arises from the scarcity of each person's first-hand experiences. And the testimony of incurable theorisers, who cannot report an observed fact without putting their own interpretation on it, would not be much better. Our policy will work best in a community of honest and hard-headed empiricists who have a respect for facts and for one another. It must also be assumed, I think, that the majority of the persons from whom one receives testimony are sane or in their right minds, and are usually capable of distinguishing between hallucinations and normal perceptions. Finally, most testimony (though not all) is given ‘after the event’, perhaps a long time after. The recipient is then at the mercy of any defects there may be, whether of omission or commission,3 in the memory of the testifier. So each of us has to assume that the memories of most other persons are not very defective in either of these ways.
We have been discussing a policy for forming beliefs. It is designed to remedy a certain sort of scarcity from which each individual person suffers, a scarcity of first-hand experiences. This policy may be formulated in two ways: (1) ‘Accept what you are told, unless or until you have specific reasons for doubting it’ (2) ‘Conduct your thoughts and actions as if it were true that what there is said to be (or to have been), there is (or was) more often than not’.
Whichever formulation we prefer, we now see that a number of conditions must be fulfilled if this policy is to succeed. It is logically possible that none of them ever are fulfilled, and it is pretty certain that not all of them are always fulfilled in fact. There are incorrigible liars, romancers and theorisers. There are careless observations, defective memories, and hallucinations or dreams mistaken for normal perceptions.
It may well seem that if ‘safety first’ is one's motto, the wisest course would be not to accept testimony at all. The policy of accepting it, unless or until one has specific reasons for doubting it, is likely to yield a pretty mixed bag of beliefs, in which there will be many incorrect ones. It is likely that there will also be a good many others which will be inaccurate, correct in some respects but incorrect in others. You tell me that an airship is coming over. I believe you, and rush out into the garden to enjoy this unusual spectacle. There is indeed a lighter-than-air aircraft in the sky, but it is only a kite-balloon which has come adrift. Moreover there may in fact be specific reasons for doubting someone's testimony (for example, he has very poor eyesight, or he has a habit of telling others what they want to hear, whether it is true or not). But we may not know that he has these defects, nor have any evidence for believing that he has them. The policy we are considering does not say ‘Accept what others say unless or until there are specific reasons for doubting it’, and would be useless to us if it did. Instead, it says ‘unless or until you have specific reasons for doubting’ or ‘unless you and them’. We may very well fail to find them, although they do exist.
But ‘safety first’ is not a good motto, however tempting it may be to some philosophers. The end we seek to achieve is to acquire as many correct beliefs as possible on as many subjects as possible. No one of us is likely to achieve this end if he resolves to reject the evidence of testimony, and contents himself with what he can know, or have reason to believe, on the evidence of his own first-hand experience alone. It cannot be denied that if someone follows the policy of accepting the testimony of others unless or until he has specific reason for doubting it, the results will not be all that he might wish. Some of the beliefs which he will thereby acquire will be totally incorrect, and others partly incorrect. In this sense, the policy is certainly a risky one. If we prefer the other formulation ‘Conduct your thoughts and actions as if it were true that what there is said to be or to have been, there is or was, more often than not’ we still have to admit that the policy is a risky one, and it would still be risky if we substituted ‘in one case out of every three’ for ‘more often than not’.
But it is reasonable to take this risk, and unreasonable not to take it. If we refuse to take it, we have no prospect of getting answers, not even the most tentative ones, for many of the questions which interest us (for example ‘What is the population of London?’ ‘Did London exist 300 years ago?’ ‘If it did, what was its population then?’).
It must be admitted, of course, that many of these questions could not even be asked unless some testimony had already been accepted. The stay-at-home inhabitant of Little Puddlecombe has only the evidence of testimony for believing that there is such a place as London at all. And if he is a safety-first philosopher, whose policy it is to reject the evidence of testimony, or to ignore it on the ground that it is so unreliable, he does not even know, or has no ground for thinking, that there is the question ‘what is the population of London?’ and therefore cannot even wish to know the answer to it. Similarly, I could not even wish to know more about the character of the Emperor Severus Alexander unless I already believed that there was such a person; and for this belief I have only the evidence of written testimony. If I were to reject that evidence, I could not even ask what kind of a person he was.
So if anyone refused to follow the policy we are recommending, and preferred the contrary policy of rejecting all testimony unless and until he had conclusive reasons for accepting it, this would certainly save him a great deal of trouble. There would be very many questions about which he would not have to worry himself, because he would not be able to consider them at all. On the other hand, we might be inclined to think that he was rather less than human. He cannot value knowledge very highly, if he does not even attempt to get it when there is a risk that his attempt will fail. He rejects the policy we are recommending because he does not really care very much for the end which it is designed to achieve. He prefers to cultivate his own garden, and a very, very small garden it will be.
In our cognitive enterprises, as in some of our practical ones, ‘nothing venture, nothing have’ is a better motto than ‘safety first’. But it is only better as a means. If you do not want to have, there is no reason why you should venture.