THE DESIRABILITY OF EVIDENCE
I shall now briefly turn to the question of why evidence is desirable. When one has a hypothesis h why should one seek evidence that h?
The answer, which is straightforward, is that one wants one's hypothesis to be true or at least probable, and one wants a good reason for believing it.
To have evidence is to satisfy both desires, at least on the theory of evidence in section 4. If e is veridical evidence that h, then h is true, and if e is (only) potential evidence that h then h is probable given e. Moreover, given background information b, if e is potential or veridical evidence that h, then, following the principle of reasonable belief, given b, e is a good reason for believing h. According to the theory of section 4 evidence that h provides a good reason for believing h only if (a) h is probable given e, and (b) there is probably an explanatory connection between h and e, given h and e. The Wheaties example provides a case satisfying (a) but not (b), while the case of the stalled car satisfies (b) but not (a). And in neither case does e provide a good reason for believing h. The fact that this man eats Wheaties is not a good reason for believing that he will not become pregnant. And the fact that my car won't start this morning is not a good reason for believing the very specific hypothesis h of section 3.On the other hand, the alternative definitions of evidence that I rejected do not jointly satisfy the twin desires of truth/probability and good reasons for believing. The “increase in probability” definition spawns the lottery case in which the hypothesis that Bill will win is not probable (though its probability is increased over its prior probability), nor in this case does the information cited provide a good reason for believing that Bill will win. The “high probability” definition, although it satisfies the desire for probability, generates the Wheaties case in which, as we have just seen, the “evidence” provides no reason for believing the hypothesis. Finally, the explanation definition satisfies neither the truth/probability desire nor that for a good reason for believing, as is shown by the case of the stalled car.[17]
8.