Hume and Testimony
Abstract
Hume’s remarks about testimony in ‘Of Miracles’ appear to commit him to a form of reductionism: the view that our acceptance of testimony depends on non-testimonial evidence. This interpretation, however, ignores the fact that Hume is concerned here with cases in which we are presented with a ‘contrariety of evidence’. The wise man proportions his beliefs to the evidence with which he is provided by testimony; but where experience provides no basis for doubting the testimony, the natural outcome is belief. Hume’s position is also to be distinguished from credulism: the view that our acceptance of testimony is independent of non-testimonial evidence. For while we are initially credulous, this propensity is corrected in light of our experience. Experience also teaches us that some of those from whom we obtain beliefs at second-hand are to be trusted. This enables us gradually to incorporate the experience of others into our own experience and then to acquire beliefs through other sources of testimony. This process of boot-strapping provides us with the enlarged experience on the basis of which testimony may be evaluated where there is a conflict of evidence.
12
Hume and Testimony[1]
Hume’s position on the epistemology of testimony must surely be considered a matter of philosophical interest and importance. Yet the secondary literature contains comparatively little discussion of this aspect of Hume’s philosophy.[2] This omission appears surprising if only because Hume recognises forms of scepticism which would call into question the evidence for any matter of fact which is not immediately available to the senses or memory (EHU 12.22); and this would obviously include evidence in the form of testimony. Furthermore Hume’s awareness that the bulk of our beliefs are acquired at second hand (see, e.g., T 1.3.9.19) lends a special interest to his account of the way in which we arrive at these beliefs. It is natural to approach this account from the perspective of recent writing on testimony which reflects broadly two sorts of philosophical position: reductionism, i.e., that our acceptance of testimony depends on evidence of a non-testimony based kind; and credulism, i.e., that testimony has an intrinsic authority so that our acceptance of it is generally independent of the availability of non-testimonial evidence. Clearly, each of these positions may be refined and qualified in various ways but as formulated they provide a starting-point for considering the nature of Hume's own position.
As a number of writers have pointed out, Hume's remarks about testimony in the context of his discussion ‘Of Miracles’ (EHU § 10) appear unequivocally to commit him to a form of reductionism.[3] Hume begins by observing that ‘there is no species of reasoning more common, more useful, and even necessary to human life, than that which is derived from the testimony of men, and the reports of eye-witnesses and spectators’. He goes on to argue that since the connection of testimony with any event is not a necessary one, our inference from the one to the other must be founded on a species of constant conjunction: in this case, ‘the usual conformity of facts to the reports of witnesses’. In other words, experience provides us with evidence of ‘the veracity of human testimony’ and hence our reliance upon it (10.4). We are unable to perceive any connection between testimony and reality a priori, but to the extent that we find a conformity between them we accept the evidence with which we are thus provided (10.8). Given that the evidence derived from testimony depends on experience, it varies in accordance with that experience, amounting either to a proof or a probability ‘according as the conjunction between any particular kind of report and any kind of object has been found to be constant or variable’ (10.5).
It is tempting to read into these remarks a general claim about the source of our justification for beliefs which are based on testimony – namely, that this depends on the more fundamental evidence provided by the individual’s own observation and experience. We are able to rely on testimony only to the extent that we have been able to establish a correlation between the reports of witnesses and the facts themselves. This has been characterised as the Reductionist Thesis [RT]: a reduction of testimony as a form of evidence to a species of inductive inference.[4] When we infer from the reports of witnesses that certain facts obtain, we are engaged in a form of inductive inference whose justification ultimately depends upon our experience of a constant conjunction between the kind of report in question and the kind of fact to which it testifies. Now there appear to be obvious difficulties with RT understood as the claim that we are able to trust testimony only because experience has shown it to be reliably correlated with the facts. For what is meant here by ‘experience’? It seems plain that it cannot mean individual experience or observation, because it is simply false that any of us individually has been able to establish the reliability of testimony in this way. On the other hand, if it means common experience then it is surely question-begging (for our justification in relying upon the observations of others is just what is at issue).[5] In fact, when Hume goes on to refer to cases in which testimony involves a contest of ‘opposite experiences’ – i.e., the experience in which reliance upon testimony is founded, on the one hand, and the experience which is in conflict with the fact testified to on the other - it seems clear that he can only mean in each case the common experience of mankind. Hence the charge that in so far as he is committed to RT, Hume’s account of belief based upon testimony is viciously circular.[6]
There seems, however, good reason to question the attribution to Hume of a general thesis about the epistemology of testimony such as RT. For one thing, it assumes that Hume is in general concerned with the justification of evidence derived from testimony. This, however, is surely at odds with Hume’s naturalistic approach to the kinds of belief with which epistemology is concerned: as, e.g., in the case of his treatment of belief in the existence of body, where the question at issue concerns the causes of the belief rather than its truth (T 1.4.2.1). If there are no ‘arguments of philosophy’ by which the truth of the principle of the existence of body might be established, then we might expect Hume to question the possibility of establishing the general veracity of testimony by any such argument. More controversially, we might also take Hume’s denial of the existence of any argument on which the principle of the uniformity of nature is founded (T 1.3.6.4-11) to provide the basis for an alternative naturalistic account of our propensity to engage in inductive inference. Thus, in the case of testimony we should also expect that Hume would be concerned with the explanation of beliefs which are acquired in this way rather than their justification.
This last point is arguably borne out by Hume’s remarks about historical testimony in T 1.3.4. In this section Hume illustrates his claim that inferences of effects from causes depend on experience in the form of impressions of sense or memory by referring to our acceptance of historical testimony concerning Caesar’s death on the Ides of March.[7] According to Hume, the basis for our belief about Caesar’s death is the causal relation of the history we are reading to a chain of testimony originating in the experiences and reports of eye-witnesses and spectators. The chain of causes and effects runs from the characters or letters present to our memory or senses back to the original event; it thus illustrates the general point that causal inference is founded in impressions of the senses or memory (‘all reasonings concerning causes and effects are originally deriv’d from some impression’ – T 1.3.4.3). Now it is possible to understand Hume as saying here that the historical belief to which he is referring is warranted or justified by the existence of a chain of causes and effects initiated by the death of Caesar and terminating in the characters or letters presently perceived or remembered. Yet it seems obvious that our belief that Caesar was killed on the Ides of March is more secure than any beliefs we might have about the supporting chain of testimony.[8] This point has force, however, only if we take Hume to be concerned here with the justification of historical beliefs.[9] The alternative is to see Hume as attempting to explain how such beliefs arise, i.e., by relating them to impressions of sense or memory construed as the effects of ideas in the minds of historians whose testimony contributes links to a chain of such testimony. (Hume subsequently attempts to explain how the ideas received from reading historical accounts of Caesar’s death may be enlivened by their association with the impressions of eye-witnesses to that event so that they become ideas of belief – T 1.3.13.6).[10] What Hume has in mind in T 1.3.4 might appropriately be described as the ‘natural propagation of belief’[11] – an attempt to show how beliefs about the death of Julius Caesar are an outcome of the causal relation between the original events and the impressions received from historical accounts of those events.
There is another reason for doubting whether Hume’s remarks about testimony in ‘Of Miracles’ commit him to anything like RT. This has to do with the philosophical context in which these remarks occur. In brief, while experience is our only guide in reasoning about matters of fact, we must recognise that ‘All effects follow not with like certainty from their supposed causes’ (10.3). Thus, ‘A wise man … proportions his beliefs to the evidence’: i.e., evidence ranging from ‘proof’ (in the case of events which have been found to be constantly conjoined) to ‘probability’ (where the events in question are more variably related). ‘All probability … supposes an opposition of experiments and observations’, where the smaller must be deducted from the greater in order to establish the force of the superior evidence (10.4). Now all this applies to the case of testimony given that various factors may result in a ‘contrariety of evidence’, such as conflict among the witnesses themselves as well as doubts about their character (10.7). But there is also the nature of the fact testified to: the more unusual it is, the weaker the testimony-based evidence for its occurrence (10.8).[12]
Hume’s concern in ‘Of Miracles’ is, then, with cases in which the evidence provided by testimony involves this kind of conflict within experience. It is precisely in such cases that some assessment must be made of the credibility of the witnesses and/or the likelihood of what they relate. But whatever the basis for this kind of assessment, there seems no reason to suppose that Hume would be committed to the view that it is normally employed in arriving at beliefs on the basis of testimony. In fact, there will be many cases in which there is no ‘contrariety of evidence’: i.e., where the testimony involved is of a uniform kind, and provided by apparently reputable witnesses; and where the facts testified to appear to be consistent with our experience both of the natural world and also of human nature itself. In these cases our acceptance of the testimony will not require us to measure the credibility of the witnesses against the prior probability of what they report as determined in each case by our own experience (or experience acquired vicariously through the many reports from apparently reputable sources of the various aspects of human nature). This, indeed, is reflected in what Hume says in T 1.3.13.9 about the belief that Caesar was killed on the Ides of March: the uniformity of the testimony connecting this event with the history we are presently reading enables the vivacity of the impressions of the original witnesses to be conveyed to our ideas so that they become beliefs about the death of Caesar.[13]
Education and testimony.
Before going any further we should consider what Hume himself would be prepared to count as examples of testimony. Hume appears to understand by ‘testimony’ something like the words of others, ones connected with ideas which represent certain facts, which are offered to us as reports of those facts. The relation of these reports to the corresponding facts is not only, according to Hume, a causal one: the ideas involved are also supposed to resemble these facts, and to this extent testimony ‘is to be consider’d as an image as well as an effect’ (T 1.3.9.12).[14] Hume is sometimes prepared to use ‘testimony’ as being more or less equivalent to ‘evidence’, as in conventional references to the testimony of the senses (as well as to the testimony of the understanding, of memory, and of history itself). But when he is concerned with testimony as a source of belief he generally has in mind historical narratives which purport to provide knowledge of past events – ones which have been witnessed either by the writer or by other eye-witnesses whose testimony provides the basis for the narrative. Hume has no doubt of the indispensability of evidence which is provided in this form, as we have seen already. Bearing in mind also that education is responsible for ‘more than one half of those opinions, that prevail among mankind’ (T 1.3.9.19), it is clear that on Hume’s own account most of what we believe must come to us at second hand.
Hume recognises that 'testimony' – in some sense of that term – covers a great deal more than the reports of historians. As we have just seen, there is also the source of belief provided by education (T 1.3.9.16-19).[15] While education reflects the influence of custom, this is not the customary association of ideas which arises from the regular conjunction of different objects, as in the case of our causal beliefs. Rather, it is a matter of the repetition of a certain idea which enables it to take hold on the mind and so give rise to belief.[16] It seems clear that Hume is not thinking here of systematic instruction but something more like the passing on of the experience of others as part of our upbringing. This effect of custom or habit may even outweigh the effects of our experience of the constant union of cause and effect. Thus, our acceptance of testimony in this form is not in general the product of inductive inference. What is more, most of our opinions are acquired in this way rather than through either abstract or experiential reasoning, and the ideas involved may operate on the mind in the same way as those of the senses, memory and reason. In this sense, education is an artificial cause of belief, in contrast to these other natural causes. Of course, the fact that education tends to result in this way in the uncritical acceptance of certain ideas means that it can easily be a source of prejudice (EHU 12.4), as well as providing maxims which may be contrary to reason. It thus seems clear that education, so understood, is not for Hume a source of evidence in the same way as those forms of testimony which may be evaluated according to our experience both of the kind of witness in question and also the correlation between what he or she reports and what we know about the world itself.