Testimony as Evidence

Regarding testimony as evidence fails to predict the sort of epistemic support testimony provides for testimonial belief. As a result, testimony-based belief should not be assimilated into the category of ep istemically inferen tial, evidence-based belief. 1. Reasons, Evidence, and Inference Holmes is trying to determine who committed the murder. His evidence consists of E1, E2, and E3. It occurs to him that he would have these pieces of evidence only if Mustard did it. He concludes that Mustard committed the murder. Here, the basis of Holmes’ conclusion is (a) his belief that his evidence consists of E1, E2, and E3, (b) his belief that he would have these pieces of evidence only if Mustard committed the murder, (c) whatever it is that supports these beliefs, and (d) whatever it is that bears on Holmes’ entitlement to draw the relevant inference from (a) and (b) to the conclusion that Mustard committed the murder. Smith is suffering from some condition that she would like diagnosed. The doctor notes the manifesting symptoms, S1, S2, and S3. On the basis of observing these symptoms, and having good reason to think that a patient manifests S1, S2, and S3 only if she is suffering from disease D, the doctor diagnoses Smith as suffering from D. Here, the basis of the doctor’s diagnosis is (e) his belief that Smith manifests S1, S2, and S3, (f) his belief that one manifests S1, S2, and S3 only if one is suffering from disease D, (g) whatever it is that supports these beliefs, and (h) whatever it is that bears on the doctor’s entitlement to draw the SANFORD C. GOLDBERG 30 relevant inference from (e) and (f) to the conclusion that Smith suffers from D. My description of these two cases is meant to bring out what I regard as an obvious parallel – one that goes to the heart of the epistemic account of inferential, evidence-based belief. In both case, the materials relevant to an epistemic assessment of the conclusion/diagnosis are the same. They consist of (i) the belief(s) that constitute(s) the subject’s evidence, (ii) the belief(s) that constitute(s) the generalization that links this evidence to the target conclusion/diagnosis, (iii) the epistemic support for all of these beliefs, and (iv) the subject’s entitlement to make the transition from these beliefs to the target conclusion/diagnosis itself. If the evidence-involving inference through which the subject acquired the belief is the only source of support for the belief in question, then (i)(iv) are the only materials that are relevant to the epistemic assessment of the belief: these exhaust the belief’s epistemic support. I will call beliefs of this sort Evidence-based beliefs, or E-beliefs for short. E-beliefs are only as epistemically good as the evidence on which they are based, the generalization(s) applied to that evidence, and the subject-drawn inference(s) to the E-belief itself. We can capture the foregoing idea in terms of the notion of supervenience, as follows: (Spv) The epistemic goodness of an E-belief supervenes on considerations (i)-(iv). According to (Spv), there could not be two E-beliefs alike as to (i)-(iv), but different as to the amount of epistemic support they enjoy. Perhaps Gettier considerations will lead one to doubt (Spv). Such a doubt would be based on the idea that there could be two subjects, alike as to (i)-(iv), but differing as to the epistemic goodness enjoyed by their respective E-beliefs, owing to ‘external’ factors regarding which they were non-culpably ignorant. This issue is vexed; a final verdict will depend on the relevant notion of epistemic goodness as well as the nature of Gettier cases. I will return to these topics, briefly, below. Here, however, I propose to circumvent the matter by conditionalizing (Spv), as follows: TESTIMONY AS EVIDENCE 31 1 This is the belief that evidence E1, E2, and E3, would obtain only if Mustard committed the crime. 2 See G oldberg (2007b) where I discuss the epistemology of inferential belief at