Humean Testimony

Saul Traiger

Department of Philosophy

Occidental College

Humean Testimony(1)

Epistemology is in the business of formulating norms of acceptable belief. We typically arrive at beliefs through inference. So epistemology is concerned with our inferential practices. Making inferences is something individuals do. If I believe the premises of an argument and you know how to infer something from those premises, it doesn't follow that you will draw the inference, unless you believe the premises. It appears, then, that all the important epistemic work goes on in individual agents. When we build an automobile together, we build it just once, but when we draw the same conclusion, each of us draws it.

This picture of epistemology as individualistic inquiry need not do violence to the obvious point that knowledge is typically acquired through interaction among epistemic agents. I can come to believe that p because you told me that p, but can your telling me that p can constitute a good reason for my believing that p when I don't have your reasons for believing that p? One argument for interpersonal epistemic dependence is the threat of skepticism: Without others, one simply cannot know very much. Of course a positive case for a social epistemology must be made by showing that knowledge depends on consensus, shared expertise, and testimony, that such notions are central to the analysis of knowledge. Epistemologists are hard at work making this positive case.(2)

This social turn in epistemology should be a welcome development to Hume enthusiasts, since Hume long ago identified social aspects of belief transmission and he had much to say about them. Unfortunately, it is not. The problem is not that Hume is ignored but rather that he is accused of clinging to empiricist principles which require the reduction of social to individual evidence. One epistemologist has even introduced a technical term, "Humean Testimony" to refer to testimony which can be reduced to non-social sensory evidence.(3) What are Hume's views on testimony and socially inculcated belief? A close reading of Hume's texts will demonstrate that these widespread characterizations are incorrect. Hume's empiricism does not commit him to epistemic individualism. More significantly, Hume's own theory of belief is decidedly social. If my reading of Hume is correct, proponents of social epistemology will thus lose a target but gain an ally whose insights provide a start at building the positive case.(4)

I. Hume's Empiricism

In A Treatise of Human Nature David Hume appears to reject abstract ideas, substance, the soul and beliefs not "grounded" in antecedent resembling impressions. In the introduction to that work Hume says none of the sciences "can go beyond experience, or establish any principles which are not founded on that authority."(5) The censure of traditional metaphysical notions, such as substance and the soul, reinforces the view that Hume's requirement, that all simple ideas have antecedent resembling impressions, is a necessary condition of justified belief: If you can't get concepts off the ground without antecedent impressions, how can you expect to justify beliefs without them? Hume's distinction, early in the Treatise, between memory and imagination, also suggests that ideas related to precedent resembling impressions enjoy epistemic privilege. (T 8 ff.) Memory preserves the "original form" of impressions. Imagination is epistemically second class; it includes "defective" or "imperfect" memory as well as the mind's deliberate transposition of ideas in "poems and romances." (T 8-9) In a discussion of extension, Hume says that only experience, my impressions and ideas, can inform me of anything concerning the nature of bodies. (T64) A bit later Hume characterizes impressions as "clear and precise" in contrast with perceptions of a "refin'd and spiritual [a] nature" which "can never, but from our fault, contain any thing so dark and intricate." (T72-3) These passages, and others like them, have been seen by many as setting the methodological tone of Hume's entire philosophy.(6) Hume attempts to cut through an epistemological jungle with empiricist principles, systematically doubting most beliefs. The skeptical conclusion of this exercise is the clearing away of most beliefs. The slender empiricist foundation won't support them.

This is admittedly a powerful motivation for reading Hume's remarks on testimony reductionistically. What justifies a belief can depend only on inference from one's private stock of impressions of sense and memory. The justification of any belief must reduce to private impressions. Still, we must separate our expectations about what the empiricism of the beginning of Hume's Treatise suggests from what Hume asserts in the text where testimony is addressed later in Book I. Does Hume appeal to private experience for the justification of belief from testimony? Surprisingly, the non-individualistic themes which many set in opposition to Hume's epistemology are championed in the Treatise and first Enquiry, and the passages in which Hume appears to try to reduce testimony to first person experience rather argue for the irreducibility of testimony as a form of evidence.

II. Conviction and the Testimony of Others

What role does the appeal to one's own impressions have in Hume's theory of the formation and justification of belief from testimony? This is difficult to answer because Hume does not sharply distinguish between descriptive and normative matters. A case in point is section iv of Book I, Part III, "Of the component parts of our reasoning concerning cause and effect". (T82-4) Hume argues that even where it seems that there are no impressions on which to base beliefs, the mind "must never lose sight of them entirely, nor reason merely on its own ideas, without some mixture of impressions, or at least of ideas of the memory, which are equivalent to impressions." (T82) The beliefs which concern Hume are those in which the mind "carries its view beyond those objects, which it sees or remembers." (T82) Hume focuses on historical belief and his example is the belief that Caesar was killed in the senate house on the Ides of March.(7)

Should we read Hume's thesis, that cause and effect reasoning must involve impressions, as a prescription - don't hold beliefs about history (or anything else) without founding them on impressions of sense or memory; or is Hume offering a psychological law of belief formation - all beliefs are products of reasoning from impressions of sense and memory, even when the contents of such beliefs suggest the contrary? Elizabeth Anscombe selects the first reading and there appears to be much in the text to recommend it.(8) Hume says that the belief about Caesar's demise is founded on the unanimous testimony of historians, that such testimony comes to us through impressions of sense or memory. We receive "characters and letters" which we remember to stand for ideas, and we can link those ideas in the minds of eye-witnesses to Caesar's death, where they began, to the person from whom we learned the fact, the testifier. Without such a chain, anchored in impressions, that is, "without the authority either of the memory or senses our whole reasoning would be chimerical without foundation." (T83)

Though it is difficult, as Anscombe admits, to untangle the chain of evidential support Hume has in mind here, it does appear that Hume is concerned with the evidence for historical belief, that he wants to distinguish well founded from chimerical belief. If we reasoned merely from our ideas, without impressions, we might hold beliefs not grounded in historical fact. Caesar is not a chimera because there is a chain of evidence terminating in impressions of him. Without this chain we have no response to the skeptic.

How does a chain of testimony justify a belief? Anscombe sheds some light on this complex passage. Hume says "it is impossible for us to carry on our inferences in infinitum;" we must stop at impressions of sense or memory. Anscombe locates two chains which must stop. First my belief that Caesar was killed is traced to my impressions of characters or letters. We can represent this chain as follows:

(1) impressions of language ---> belief

The impressions of language are the cause of my belief, but not the justification for it. If it were, I might accept anything I'm told. The warrant for this belief, on Anscombe's reading, comes from what a chain which shows the connection of the characters and letters to the fact it represents:

(2) death of Caesar ---> eye-witness impression ---> idea --> characters/letters

Just as the impressions of characters and letters are the cause of belief, the justification for the belief depends on taking those characters and letters as the effect of a causal chain from Caesar's death to the witnessing of the death by spectators and the subsequent passing on of belief through testimony. Anscombe criticizes this theory of testimony. On her interpretation, Hume seeks the justification of historical belief in belief about a chain of historical record which traces, by inference from effect to cause, back to the event which the belief is about. Hume establishes the epistemic credentials of the impressions which induce belief in these cases by appeal to the chain of testimony which induced it. Anscombe's Wittgensteinian point is that the belief about the historical record depends on the belief about the event which it is supposed to support. The belief that Caesar was killed on the Ides of March is more secure than any beliefs about chains of testimony to support it.(9) She says: "For let us ask: why do we believe that there were eye-witnesses at that killing? Certainly for no other reason than that we believe it happened."(10) Anscombe diagnoses Hume's mistake as motivated by Cartesian skepticism. Only hyperbolic doubt can raise the issue of the warrant for a belief in the existence of Caesar. The very same doubt would block a solution which traces testimony back to eyewitness impressions.

Is Hume guilty of answering skepticism about the existence of Caesar by appeal to an historical chain of perceptions which has lower epistemic credentials than then belief in Caesar itself? Although Hume would allow the possibility of doubting Caesar's existence, such doubt would not be Cartesian in nature.(11) The reason for this is that Hume's purpose in this section is not to account for the justification of historical beliefs, but to explain an important feature of their causal origin.

Raising the question of the epistemic status of historical beliefs at this early juncture in the Treatise, before Hume has given an account of belief would have been inappropriate. The Caesar passage occurs right after Hume has "sunk" the question of how experience gives rise to the principle that every effect must have a cause in the question of why we conclude that "particular causes necessarily have such particular effects." (T82) Section iv, "Of the component parts of our reasoning concerning cause and effect" is the first part of Hume's epistemologizing transformation of the discussion of causation to a discussion of causal inference. That account will depend on Hume's thesis that a belief is a lively idea related to a present impression. Hume's purpose in Section iv is to show that causal reasoning, the reasoning which will turn out to bring about belief, always involves impressions of sense or memory, never simply ideas alone. His example is the historical belief about Caesar. Appearances to the contrary, this has nothing directly to do with the epistemic warrant of such beliefs. Hume's discussion of justification only occurs after the account of belief has been provided.

Hume's aim is to show that reasoning from cause and effect involves impressions of sense and memory. For historical beliefs, presumably a tough case, it would seem that Hume need only remind us that such beliefs get started by the reception of testimony, and he has already explained that one receives testimony from impressions of language. What does the second chain, from the event via the eye-witness to testimony, have to do with belief formation? If Hume is not concerned with the justification of belief here, why does he even mention this second chain?

Though Hume's language suggests that the second chain provides a stopping place in a regress of reasons, it really plays a more modest role. As Hume points out later, (T113) for impressions of characters and letters to induce belief, one must take those characters and letters as reports or beliefs of the testifier, and not as stories, fables, dreams, or mere utterances. Though it is possible for an apparent testifier to misrepresent her words as signalling belief when they do not, usually what brings about the testifier's belief is the experience of letters and characters which she took as belief. She was told. When Hume says "Tis obvious all this chain of argument ... is first founded on those characters or letters which are seen or remember'd" (T83) he means that each testifier (except the eye-witness) in the second chain must have such impressions and must take them as indicating belief in order for the chain to continue; otherwise the chain stops. This is just an account of one aspect of the natural propagation of belief and it shouldn't be taken as an account of how beliefs are justified.(12) Reading it as an account of justification depends on misreading what Hume says is founded on what. It is not that my belief in Caesar is founded on (justified by) the eye-witness impressions of the murder. Rather my belief (and the belief of the testifiers other than the eye-witness) is founded on (caused by) impressions of testimony which I take as testimony. The existence of the second chain, whether I consider it consciously or not, causes me to take the impressions as testimony.(13)

The difference, then, between my reading of this section and Anscombe's is that she sees the first chain as inducing belief and the second as inducing justified belief while I see both chains as necessary for inducing belief. Merely having impressions of language is not enough to bring about belief. Impressions of testimony have a rich casual history which Hume only begins to explain here.

The non-normative, naturalistic reading of this section is further supported by what Hume says at the close of the section:

I need not observe, that 'tis no just objection to the present doctrine, that we can reason upon our past conclusions or principles, without having recourse to those impressions, from which they first arose. For even supposing these impressions shou'd be entirely effac'd from the memory, the conviction they produc'd may still remain; and 'tis equally true, that all reasonings concerning causes and effects are originally deriv'd from some impression; in the same manner, as the assurance of a demonstration proceeds always from a comparison of ideas, tho' it may continue after the comparison is forgot. (T84)

Conviction does not require that we reason explicitly, that is, that we remember the impressions which gave rise to belief. Our conviction is not a matter of being justified, of legitimizing our inference. The greater part of humankind, I think Hume would say, rarely reflects on the existence of chains of testimony which induce their beliefs. But induce them they do.

Although she doesn't cite it, there is a parallel passage in the Enquiry Concerning Human Understanding which appears to support Anscombe's reading against mine. In Part I of Section V of the Enquiry Hume again reminds us that although beliefs are the effect of custom, the "great guide of human life," there must always be "some fact ... present to the senses or memory, from which we may first proceed in drawing these conclusions." (E45) Hume's conclusion looks like a classic statement of the foundationalism Anscombe attributes to him:

If I ask why you believe any particular matter of fact, which you relate, you must tell me some reason; and this reason will be some other fact, connected with it. But as you cannot proceed after this manner, in infinitum, you must at last terminate in some fact, which is present to your memory or senses; or must allow that your belief is entirely without foundation. (E 46)

The admittedly most natural reading of this is as a regress argument; impressions are the basic "facts" which halt the regress.

This interpretation looses its plausibility when the passage is understood in context. At the beginning of the section Hume says that philosophy can seductively draw us away from "the bustle of the world, and the drudgery of business," allowing us to indulge our private reason. Once the role of custom in belief formation is appreciated, however, there is no chance that we will remain isolated in our own private impressions of sense and memory. Custom takes us beyond them. The passage quoted above comes right after Hume argues that a man who sees "the remains of pompous buildings", could not infer the prior existence of a civilization unless he had "perused volumes" of ancient history or had received testimony after some other fashion. (E45) Without the belief that there is a historical causal chain from witnesses, through testimony, to the existence of the civilization whose remains one is now viewing, one would not take the perceptual evidence as evidence of a historical civilization. The textual support for this interpretation is somewhat cleaner in the Enquiry than the Treatise; Section V of the Enquiry has two parts. Part I, which we've been considering, discusses the formation of belief through custom. Part II raises and seeks to answer normative question: "Wherein, therefore, consists the difference between such fiction and belief." (E47)

It is a mistake, then, to see Hume questioning our historical beliefs in these passages. Hume's topic is conviction, and conviction does not require reflective argument and reasoning of the sort Anscombe is critical. Hume's account of conviction from testimony emphasizes the social scope of belief. Though we can't have beliefs without some admixture of impressions, it is significant that the kind of impressions on which Hume chooses to focus are "rich"; they make reference to a community of thinkers like ourselves. The impressions which induce belief are not private patterns of colors, sounds, and the like. Hume thus provides a non-individualistic account of belief formation rather than a brand of verificationism.

III. Justification and Belief from Testimony

Though his concern is not with justification in the passages just considered, Hume does not ignore the question of the evidence for historical belief. Hume's remarks on justification build on his view about the natural formation of belief just discussed. In subsequent sections of the Treatise Hume considers how chains of testimony can vary in ways which support or detract from the believability of testimony. When belief occurs by causal chains like the Caesar chain, there's no problem about justification. The epistemic status of the belief is not at issue, though not for logical or criteriological reasons, as Wittgenstein might have held. Once one appreciates what I have called the "richness" of belief formation, there are many ways transmission of belief can go astray, and Hume is interested in these.

Paradigm cases of belief formation for Hume involve our taking other persons as truth tellers.(14) Having explained how we come to take other people as truth tellers in standard cases of historical belief, Hume is painfully aware that we "have a remarkable propensity to believe whatever is reported, even concerning apparitions, enchantments, and prodigies..." (T113) How can we curb this propensity? One temptation is to say that we should ultimately never depend on the testimony of others alone; we must always check the alleged facts against our own experience. Hume appears to be saying something like this when he asserts that experience is the "true standard" of "all judgments", including judgments otherwise reached by testimony. (T113) In the Enquiry, Hume also emphasizes that beliefs from testimony must be "founded on" experience:

It will be sufficient to observe, that our assurance in any argument of this kind is derived from no other principle than our observation of the veracity of human testimony, and of the usual conformity of facts to the reports of witnesses. It being a general maxim, that no objects have any discoverable connexion together, and that all the inferences, which we can draw from one to another, are founded merely on our experience of their constant and regular conjunction; it is evident, that we ought not to make an exception to this maxim in favour of human testimony, whose connexion with any event seems, in itself, as little necessary as any other. (E111)

How are we to understand Hume's reference to "our experience" here?

In an influential paper and more recently in his book, Testimony, C.A.J. Coady suggests two ways to understand the phrase "our experience" in this passage. On Coady's view, both are problematic.(15) One reading is that by "our experience" Hume means the experience of each of us. Each person must found beliefs based on testimony on their own first person impressions of the facts reported. This "reductive thesis," as Coady calls it, is too restrictive. The other interpretation is that by "our experience" Hume means the collective experience of humankind as reported in testimony. Coady's objection to this second view is that it is circular. It appeals to the veracity of testimony to support the veracity of testimony.

Let's consider the two readings in turn, and Coady's objections. The requirement that one check testimony against one's own testimony-free experience unacceptably restricts the class of justified beliefs. It rules out most historical beliefs, most scientific claims, perhaps most common knowledge. Our epistemic interdependence is a fact which is incompatible with a thoroughgoingly individualistic approach to evidence.(16) So if Hume requires that testimony reduce to testimony free evidence, so much the worse for Hume's epistemology.

Frederick Schmitt, who coined the term "Humean Testimony," looks more closely than Coady at the first of the two readings of "our experience".(17) Schmitt interprets Hume as a reliabilist about of justification, where the reliability of testimony is based on the receiver of testimony's determination that the testifier is an expert on the matter to which she attests. The problem is establishing that the testifier is indeed an expert. The receiver of testimony can establish the credentials of the testifier only if he already has mastery of the testifier's field of expertise. The testimony of others would be of no use. To judge it requires knowing what the testifier knows.

We already have strong grounds for rejecting this interpretation of Hume.

By saying that the veracity of testimony is a matter of our observation, Hume means simply that what people report is usually true. That's an observation we, as students of human nature, can make and Hume does make. Hume's remarks don't commit him to a particular account of how we make this observation. He certainly does not say that the only way to make it is for each person to check all testimonial claims against their own first person experience. Are there any passages which could support such a first person reading of "our experience" as "each person's impressions?" We've already seen that the sense in which beliefs based on testimony trace back to one's own impressions is not a matter of the justification of such beliefs, but one of belief formation. So for any belief, as indeed for any idea, there will be antecedent impressions of sense and/or memory as causal antecedents. But those causal antecedents need not have anything to do with the justification of those beliefs. Schmitt is certainly right that the evidential value of testimony is often established by determining the expertise of testifiers. There are, however, many ways to determine the expertise of testifiers other than by being an expert oneself.

Coady's alternative reading is no more sympathetic. If we read "our experience" non-individualistically, then Hume appears to be saying that we have common, shared experience that testimony in general is reliable. But what sort of experience could count as such evidence, once we have ruled out individualistic evidence? The only evidence left is testimonial evidence, the very type of evidence which was in question in the first place. On this reading, Hume's normative account of testimony is viciously circular.

What, exactly, does Hume say that "our experience" is supposed to establish? The circularity objection depends on our experience establishing the veracity of testimony in general. Hume discusses the issue of the veracity of testimony in general. But he doesn't try to establish that testimony is, in general, true. He says that the connection between any particular bit of testimony and any particular event reported is "as little necessary as any other." Testimony is an effect. While testimony is often the effect of the event it reports, the fact that it is testimony shouldn't elevate its status, its connection to a cause, above that of any other effect. Hume observes that we have a tendency to treat testimony as a special effect, that we overrate testimony. The crucial point is that Hume does not argue for elevating the status of testimony. On the contrary, his point is that there is nothing special about testimony which warrants a generalization we wouldn't make in other cases. So Hume is not saying that our common experience ought to establish a single, over-arching generalization about testimony. Rather he is criticizing that inference.

What makes these passages interesting and controversial is that they involve, in important ways, several key aspects of Hume's philosophy - his skepticism, his naturalism, and his critique of religious dogmatism. The individualistic reading common to Anscombe, Coady, Schmitt, and Webb, overemphasizes Hume's skepticism. It takes Hume's empiricism about the formation of ideas and overextends it into a theory of evidence. Part of my task has been to show how Hume keeps these things separate.

There is a skeptical element in Hume's account of testimony, one which fits nicely with other skeptical themes. Causation is a relation which does not depend on ideas. Causal connections are not conceptual connections. This point is part of Hume's famous skeptical argument. It's also part of his remarks on the connection between testimony and the event reported by testimony. When Hume says that this connection is overrated, he's saying that the connection is a causal one, and hence not a conceptual one. The falsehood of a piece of testimony is never impossible. The skepticism here, however, is mitigated. Hume thinks that experience is our guide in causal inference, and his point is that experience informs our inferences about testimony just as it does our testimony-free inferences. They're on the same footing.

This is the naturalistic dimension of Hume's theory of testimony. The phenomenon of human testimony is part of the natural, causal order. There are regularities in the complex behavior of belief transmission through testimony, as human agents ply their beliefs on other human agents. Formulating psychological laws is part of Hume's science of human nature, a critical part, since those laws will help us evaluate the claims of testifiers and hence to appreciate other regularities and formulate other laws.

What can ground a belief from testimony, according to Hume, is the believer's experience of human nature, that is, his or her beliefs about the testifier. When we receive testimony we make use of information about the testifier. Hume looks at that information to see what it justifies and what it doesn't. In the parallel passage in the Treatise, Hume says it is only "our experience of the governing principles of human nature, which can give us any assurance of the veracity of men." (T113) Beliefs based on testimony must be judged by utilizing knowledge of human nature in order to determine the quality of the testimony. That means that in particular cases one must figure out whether someone lies, exaggerates, frightens or pleases when they offer testimony. Here too "our experience" must mean each person's causal inferences about particular testifiers.

Judgements about the reliability of testimony will make use both of specific facts about the testifier and psychological laws, which Hume refers to as "governing principles." What are these governing principles? Hume's remarks are not systematic here, but it should come as no surprise that some involve the passions, since for Hume, the effect of belief "is to raise up a simple idea to an equality with our impressions, and bestow on it a like influence on the passions." (T119) Hume gives us several examples of the interplay of passion and belief: a coward will easily assent to accounts of danger; a sorrowful person accepts sorrowful accounts. (T120) Hume's extensive discussions of such factors shows that to justify belief from testimony requires that we understand a great deal of the psychology of human beings.

Now how do we establish these general psychological principles? Is it circular to allow that some of these principles are transmitted through testimony, if those same principles are to be used in the evaluation of testimony? Suppose I want to evaluate Smith's claim that it is snowing in Singapore. I notice that Smith loves attention. I learn, through testimony, i.e. by reading a psychology text or a novel, that people who crave attention may say things for their shock value. This psychological principle may be invoked when I evaluate Smith's claim. Here testimony is used to evaluate other testimony, but not in a way which is circular. The invocation of the psychologist's testimony may or may not be appropriate in a particular occasion, but that's a different matter. Just as I can use or abuse sensory evidence to check a sensory claim, I can use or abuse testimony to check a claim presented through testimony. The threat of circularity is no greater in the first case than in the second. I need not be an expert about the climate of Singapore to evaluate Smith's claimed expertise, though I need some expertise in folk psychology.

While a full account of Hume's positive theory of testimony is beyond the scope of this paper, a brief overview can be given.(18) We've seen that Hume has a two part theory of belief; one part devoted to the formation of belief, the other to the justification of belief. In both parts testimony is at center stage. Hume does not give testimony-free accounts of belief formation and justification and then apply them to testimony. On the contrary, his first example of causal inference is the inference to the belief that Caesar died in the Senate House on the Ides of March. Testimony is central because it illustrates the main feature of causal inference, that it is the only species of philosophical relation which takes us beyond our senses. Inferences from testimony take us beyond our senses by utilizing the senses of other epistemic agents.

Beliefs from testimony may be formed only when we take the reports of others as testimony. Hume has a rudimentary account of the difference between words and utterances which we interpret as fiction, poetry, and the like, and those which we take as testimony. (T97-8) Our interpretation of words and utterances as testimony or fiction depends on our beliefs about the causal origin of those words or utterances.

Those very beliefs about the causal origin of the words and utterances of others form the basis of our overly zealous credulity and its regulation, as outlined above. Once "our experience" is read nonindividualistically, it becomes clear that our experience of testifiers, i.e. our beliefs about human nature, inform our evaluation of their testimony. Hume looks to contiguity and resemblance to help account for the credulity, both particular cases of people believing things they shouldn't, and the more general phenomenon of widespread credulity, which Hume characterizes as the most universal and conspicuous weakness in human nature. (T112) In a particular case, for example, the purported contiguity of the Red Sea with the events testified to by Moses, will often enliven the latter to the point of inculcating belief in those who "have seen Mecca or the Holy Land." (T110) Eloquence, pride in "exciting the admiration of others" (E117), surprise and wonder, are all responsible for our "too easy faith in the testimony of others." (T112) The "intimate connexion" between words and utterances, the ideas in the minds of the testifier, and the facts they purport to represent, cause us to chronically overestimate of the credibility of testimony.

Coady thinks that the only way Hume can make sense of causal laws involving testimony is to establish correlations between facts verified through testimony-free experience and testimony. He then has us imagine a world of Martians in which such correlations never hold.(19) The absurdity of such a world should force us, Coady argues, to reject Hume's reductive approach to testimony. On my interpretation, however, individual sensory evidence is not isolated from testimony in the way required to generate the Martian world. The possibility that beliefs from testimony are all false degenerates into the possibility that all our beliefs are false, since, for Hume, beliefs from testimony are often as central to the support of our full systems beliefs as any others. For Hume, the inferences which generate beliefs from testimony are causal inferences, the same used to generate any other belief. A full scale indictment of testimony would lead to a rejection of all our causal inferences.

Hume's Treatise presents no comprehensive theory of justified belief, but much is said about the avoidance of unjustified belief. Although Hume occasionally contrasts unjustified belief to belief based on "truth and reality," (T121) it would be a mistake to take such remarks as suggesting a program other than the one in which he is here engaged, namely the program of tagging problematic beliefs and letting others alone. Hume's theory of belief formation, a theory which stresses the public causal history of impressions of testimony, shows that all beliefs are in principle capable of being called into question, including our most cherished beliefs about Caesar. But we'd need other, potentially defeating knowledge before we could reasonably entertain doubts about such established beliefs. Humean testimony neither question-beggingly depends on the veracity of testimony nor reduces to first-person knowledge of the facts in question. Rather, it appeals to independent facts about the testifier, psychological facts which can be established without our knowledge of the subject matter of the testifier.

Acceptable beliefs from testimony are generally justified by default; they involve background conjunctions of conditions like: "if the testifier is not providing a sorrowful account, and you are not sorrowful; and ... " This is a rather different account of justification than one which traces all beliefs to supporting impressions. Yet there is still a role for one's own impressions in this picture and it would be a mistake to claim that there is no individualistic strain in Hume's epistemology. In a letter to the Rev. Hugh Blair, Hume writes:

No man can have any other experience but his own. The experience of others becomes his only by the credit which he gives to their testimony; which proceeds from his own experience of human nature.(20)

There are two points to be made. The first is that one only "has" one's own perceptions. One doesn't feel the pain of other persons. The second is that in spite of this, one can still learn about and from other persons through testimony. The second point does not entail that beliefs must be justified by stripping away all reference to other persons. On the contrary, Hume asserts that we give credit to testimony, that is we take testimony to be justified, by utilizing what we know about human nature. That knowledge is usually itself the result of testimony and education. What is disallowed is only that our insight is the result of the direct experience of other persons' perceptions.

Why has so much enthusiasm has been generated for the reductionist interpretation of Hume's theory of testimony? As I noted at the start, the tendency results in part from the running together of Hume's empiricism about the formation of ideas and his account of justified belief. Another is the failure to appreciate the historical context in which Hume's remarks on testimony are formulated. The primary concept of evidence in the seventeenth century was testimonial and authoritative evidence, not what Locke called "the evidence of things."(21) While some modern philosophers, notably Descartes and Locke, were reviving ancient skepticism about historical belief, Hume was stressing the indispensability of testimony as a form of evidence.(22) While Hume backed the tradition in his emphasis on testimony, he bucked it by his stand on the proper application of testimonial evidence. Hume used his analysis of testimony to attack religious belief which sought support in the claimed occurrence of miracles. He achieves this and more by a naturalistic account of probability and belief which covers all beliefs, including both beliefs from testimony and beliefs from "pure" sensory evidence.

The legacy of misinterpretations of Hume on testimony has obscured the fact that Hume's account suggests that both the formation and the justification of our beliefs we are introduced to the rich public realm of language use, human passion and action. Only by looking there can we begin to understand how we become "acquainted with such existences, as by their removal in time and place, lie beyond the reach of the senses and memory." (T108)

Saul Traiger

Occidental College


1. An earlier version of the first half of this paper was delivered at the Fourteenth Hume Conference, Marburg, Germany, August 15, 1988. I'd like to thank Annette Baier, Geoffrey Sayre McCord, and Eric Steinberg for comments and suggestions. A version of the full paper was delivered at Nanzan University, Nagoya, Japan. Professor Hiroyuki Hattori translated the paper into Japanese and served as interpreter, enabling a discussion with Nagoya area philosophers from which this paper has benefitted enormously.

2. Kornblith, Hilary, "Some Social Features of Cognition", Synthese 73 (1987) pp. 27-41; John Hardwig "Epistemic Dependence", Journal of Philosophy 82, 7 (1985) pp. 335-349; K. Lehrer and C. Wagner, Rational Consensus in Science and Society (Dordrecht: D. Reidel, 1981); Alvin Goldman, "Epistemic Paternalism: Communication Control in Law and Society", The Journal of Philosophy LXXXVIII, 3 (March, 1991) pp. 113-131 and "Foundations of Social Epistemics", Synthese 73 (1987) pp. 109-144.

3. Cf. Frederick Schmitt; "Justification, Sociality, and Autonomy", Synthese 73 (1987) pp. 43-85.

4. I have taken care not to call the non-individualist a proponent of "social epistemology." This latter phrase has been appropriated, not just as a journal title, but as an ideology with much broader implications than the epistemological view to which I'm referring. Cf. Steve Fuller, Social Epistemology (Bloomington: Indiana University Press, 1988).

5. David Hume, A Treatise of Human Nature L.A. Selby-Bigge, ed., Oxford, 1888, p. xviii, hereafter "T".

6. See John Passmore, Hume's Intentions Third Edition, London, Duckworth, 1980, 65-104, for an excellent discussion of the basis for phenomenalist and positivist readings of Hume's epistemology. Passmore adduces evidence for these and other "intentions" in the text.

7. I'm taking the liberty of talking about belief here even though Hume doesn't officially introduce the topic of belief until somewhat later in the Treatise. Hume's official subject here is cause and effect reasoning. By no later than section vi it is clear that the product of such reasoning is belief.

8. Elizabeth Anscombe, "Hume and Julius Caesar" in From Parmenides to Wittgenstein: Collected Philosophical Papers Volume 1 Oxford, Basil Blackwell, 1981, 86-9; reprinted from Analysis 34 1 (1973).

9. Anscombe, op. cit. p. 88

10. Ludwig Wittgenstein, On Certainty, New York, Harper and Row, 1969, p. 26 ff.

11. Anscombe is not alone in attributing Cartesian motives to Hume's epistemology. Cf. Thomas Reid, Works, I, Edinburgh, 1863 p. 91, and Passmore, op. cit., pp. 7-17.

12. Hume has many other things to say about the formation of belief; testimony is only one of a number of causal factors. Cf. David Hume, Enquiry Concerning Human Understanding, L.A. Selby-Bigge and P.H. Nidditch, eds., Oxford, Oxford University Press, 1975, hereafter "E". "Of Miracles", pp. 109-130.

13. Anscombe, op. cit. p.90

14. This is borne out at T115, where Hume contrasts the non-credible belief in immortality with the "true and established judgment; such as is derived from the testimony of travellers and historians."

15. Coady, C.A.J., Testimony (Oxford: Oxford University Press, 1992) pp. 80 ff., "Testimony and Observation"; American Philosophical Quarterly 10, 2 (April, 1973). See also Mark Owen Webb, "Why I Know About As Much As You: A Reply to Hardwig" Journal of Philosophy XC 5, May 1993 pp. 260-270; Michael Welbourne; "The Community of Knowledge", Philosophical Quarterly 31, 1981 pp. 302-14.

16. Cf. Hartwig, Schmitt, Webb, op. cit., but also H.H. Price, Belief (London: George Allen & Unwin, 1969) pp. 112-129.

17. Schmitt, op. cit. pp. 43-53.

18. Further work on this topic, and on the place of Hume's testimony-based epistemology and its place in Book I of Hume's Treatise is the subject of my "Beyond Our Senses: Recasting Book I, Part III of Hume's Treatise", manuscript.

19. Op. cit. p. 85 ff.

20. The Letters of David Hume, ed. J.Y.T.Greig, Oxford, Oxford University Press, 1932, Vol. I p.349

21. John Locke, An Essay Concerning Human Understanding (Oxford: Oxford University Press, 1975) pp. 657-668. See also Ian Hacking, The Emergence of Probability (Cambridge: Cambridge University Press, 1975) pp. 31-38.

22. Cf. Peter Jones, Hume's Sentiments (Edinburgh: University of Edinburgh Press, 1982) pp. 44 ff.