THREE CONCEPTS

Alan's skin has yellowed, so on Monday he sees the doctor, who examines him and declares that he has jaundice, i.e., the visible expression of an increased concentration of bilirubin in the blood (which I shall abbreviate as an i.c.b.).

i. c.b. but that his yellow skin was produced by a dye with which he was working. On Friday which of the following propositions, if any, should the doctor affirm?

i. Alan's yellow skin was evidence of an i.c.b. and still is.

ii. Alan's yellow skin was but no longer is evidence of an i.c.b.

iii. Alan's yellow skin is not and never was evidence of an i.c.b.

The doctor might be tempted to assert (i) on the ground that Alan's yellow skin is typically the kind of skin associated with an i.c.b. On the other hand, (ii) might be tempting to say since the doctor now has ad­ditional information which makes the original evidence efficacious no longer. Finally, he might be tempted to assert (iii) on the ground that false or misleading evidence is no evidence at all. He might say that Alan's yellow skin is not and never was (real or genuine) evidence of an i.c.b., though on Monday he mistakenly thought it was.

I believe that these three responses represent conflicting tendencies in the way we actually speak about evidence, and that a different but related concept of evidence can be associated with each.

I begin with a notion which I shall call potential evidence. The presence of Alan's yellow skin is potential evidence of an i.c.b. since yellow skin of that sort is generally associated with an i.c.b. That 35 percent of those sam­pled in this district said they would vote for the Democratic candidate is potential evidence that roughly 35 percent of all those voting in the district will vote for him since samples of that size are usually accurate.

First, e can be potential evidence that h even if h is false.¹ Secondly, potential evidence is objective in the sense that whether e is potential evidence that h does not depend upon anyone's beliefs about e or h or their relationship. That Alan has yellow skin is potential evidence that he has an i.c.b. even if no one believes that it is or knows or believes that he has yellow skin or an i.c.b. In these two respects potential evidence is akin to the concept Hempel seeks to define in “Studies in the Logic of Confirmation”^{[1] [2] and to one Carnap calls the classificatory concept of confirmation which he defines using his theory of probability.[3]}

Although Hempel and Carnap in addition allow e as well as h to be false, I am inclined to think that if there is a concept of potential evidence in use it is one that requires e to be true. This, then, is the third feature I attribute to this concept. That Alan has yellow skin is potential evidence that he has an i.c.b. only if he does in fact have yellow skin. The concept that Hempel and Carnap seek to analyze which has no such requirement could be described as “doubly potential” (“e would be potential evidence that h if e were true”). Finally, although both of these authors allow e to entail h (as a “limiting” case), I doubt that there is such a concept of evi­dence in use. The fact that Alan has yellow skin is not evidence that he has skin; it is too good to be evidence.

Can a concept of evidence with these characteristics be defined, and if so will it support proposition (i)? Various definitions of potential evidence will be examined in later sections, after which this question will be addressed.

I turn now to a second concept, veridical evidence, that sanctions prop­osition (iii) above. e is veridical evidence that h only if e is potential evi­dence that h and h is true. However, this is not yet sufficient.

(1) e is veridical evidence that h if and only if e is potential evidence that h, h is true, and there is an explanatory connection between e's being true and h's being true.

At the moment, I shall not try to define the notion of an explanatory connection. I do so in section 8 (excerpted from my The Book of Evidence). Before then, some further comments concerning (1) are in order. The con­cept of evidence characterized here does not require that h's being true correctly explain e's being true; the converse is also possible. That Jones has a severe chest wound can be veridical evidence that he will die, even though the hypothesis that he will die does not explain the fact that he has a severe chest wound. Rather the reverse explanation is correct: he will die because he has a severe chest wound. Alternatively, there may be some common explanation which correctly explains why both e and h are true. The fact that hydrogen and oxygen combine in a simple ratio by volume may be evidence that nitrogen and oxygen do too. (Gay-Lussac indeed took it to be so.) In this case h does not explain e, nor conversely. Still both h and e are explained by appeal to the fact that the pairs of substances involved are gases and gases combine in simple ratios by volume. (Or at a deeper level both h and e are explained by appeal to Avogadro's hypothesis.)

If we can assume—as I shall do here—that whether there is an explan­atory connection between the truth of e and h does not depend on what anyone believes (except, of course, where e and h themselves describe beliefs or intentional actions), then veridical evidence, like potential evi­dence, is an objective concept of evidence. Moreover, it is a concept in accordance with which proposition (iii) should be asserted.

Turning to a third concept of evidence, we speak not only of some­thing’s being evidence that h but also of something’s being so-and-so’s evi­dence that h. On Monday the doctor’s evidence that Alan has an i.c.b. was that Alan has yellow skin. I take this to involve at least the claim that on Monday the doctor believed that Alan’s yellow skin is potential evidence of an i.c.b. However, this is not sufficient if on Friday Alan’s yellow skin is potential evidence of an i.c.b.; for the fact that Alan has yellow skin is not on Friday the doctor’s evidence that Alan has an i.c.b., even if on Friday the doctor believes that it is potential evidence. Accordingly, one might be tempted to say that the fact that Alan has yellow skin is the doctor’s evidence that Alan has an i.c.b. only if the doctor believes that this fact is veridical evidence of an i.c.b. More generally,

(2) e is X’s evidence that h only if X believes that e is veridical evidence that h; i.e., X believes that e is potential evidence that h, that h is true, and that there is an explanatory connection between the truth of h and e.^[4]

However, (2) may be too strong in requiring that X believe that h is true and that there is an explanatory connection between h and e. Sup­pose that on Monday the doctor is unsure about whether Alan has an i.c.b. He thinks it probable but he does not know whether to believe it, so he orders tests. Later when the tests reveal no i.c.b. and Alan indignantly asks the doctor “what was your evidence that I have an i.c.b.?” the doctor might reply: “the fact that you have yellow skin.” Even if on Monday the doctor was not sure whether to believe that Alan has an i.c.b. at least he believed that this is probable and that it is probable that this explains his yellow skin. Accordingly, (2) might be weakened as follows:

(23 e is X’s evidence that h only if X believes that e is potential evidence that h, that it is probable that h is true, and that it is probable that there is an explanatory connection between the truth of h and e.

Neither (2) nor (2^Z), however, supplies a sufficient condition. For e to be X's evidence that h it is necessary in addition that

(3) X believes that h is true or probable (and does so) for the reason that e.

The fact that Alan is receiving a certain medical treatment T may be (ve­ridical) evidence that he has an i.c.b. (since treatment T is given only to such people). Even if Alan's doctor knows and therefore believes that the fact that Alan is receiving treatment T is (veridical) evidence that he has an i.c.b., this fact is not the doctors evidence that Alan has an i.c.b. His reason for believing this is not that Alan is receiving treatment T Accordingly, I would add condition (3) to (2) or (2') to obtain sufficient conditions. (2) and (3) can be said to characterize a strong sense of “X's evidence,” (2') and (3) a weak one.

Both (2) and (2') (with (3) added) sanction proposition (ii). When we say that Alan's yellow skin was but no longer is evidence of an i.c.b. we may be understood to be referring to someone’s evidence, in this case the doctor's. We may mean that on Monday the fact that Alan has yellow skin was the doctor's evidence that Alan has an i.c.b., but on Friday it is no longer so. On Friday due to other facts he has learned the doctor no longer believes (it probable) that Alan has an i.c.b. This concept of evidence is thoroughly subjective. Whether e is X's evidence that h depends entirely on what X believes about e, h, and their relationship, and not on whether in fact e is potential or veridical evidence that h.

This subjectivity means that one cannot draw an inference from the fact that e is X's evidence that h to the claim that e is at least some good reason to believe h, or even for X to believe h. It is commonly supposed that evidence bears some relationship to what it is reasonable to believe. Although this may be expressed in a variety of ways perhaps the following simple formulation will suffice for our purposes:

A Principle of Reasonable Belief.

This principle is satisfied by the two objective concepts of evidence. If, in the light of the background information (b) that yellow skin of that type is typically associated with an i.c.b., the fact that Alan has yellow skin is potential (or veridical) evidence that he has an i.c.b., then, given b, the latter fact is a good reason for believing that Alan has an i.c.b. The subjec­tive concept, on the other hand, does not satisfy this principle. The fact that Max has lost ten fights in a row may be his evidence that his luck will change and he will win the eleventh. But this fact is not a good reason at all, even for Max, to believe this hypothesis.

To summarize, then, the three concepts of evidence here characterized provide a way of answering the question of whether the fact that Alan has yellow skin is evidence that he has an i.c.b. It is potential evidence since that kind of skin is typically associated with an i.c.b. It is not verid­ical evidence since the hypothesis is false and his yellow skin is correctly explained not by his having an i.c.b. but by the fact that he was working with a dye. On Monday but not on Friday it was the doctor's evidence that Alan has an i.c.b., since on Monday but not on Friday the doctor be­lieved that Alan has an i.c.b. for the reason that he has yellow skin, which he believed was veridical evidence of an i.c.b.

If potential evidence can be defined, then so can the other two con­cepts via (1), (2), and (3). Of various definitions of potential evidence that appear in the literature two general types will be discussed here because each by itself is not sufficient but if appropriately altered and combined the result may be. The first and most popular type defines evidence in terms of probability, the second in terms of explanation.

2.