THE A PRIORI ASSUMPTION

I turn now to the second assumption made by many philosophers who try to provide objective accounts of evidence. That assumption is that the evidential relation is a priori: whether e, if true, is evidence that h, and how strong that evidence is, is a matter to be determined completely by a priori calculation, not by empirical investigation.

I will illustrate this idea with brief references to some philosophical theories, the first one being Carnap's. Carnap (1962) embraces both an increase-in-probability and a high probability concept of evidence. But for Carnap the probability relation is entirely a priori. What h's probability is on e is settleable a priori, by reference to the rules of the “linguistic frame­work” (as Carnap calls it). The h-d view of evidence also makes the evi­dential relation a priori, since whether h entails e is a priori. (Even more complex and sophisticated h-d views, which appeal in addition to ideas about simplicity or coherence, are a priori, since whether these additional criteria are satisfied is supposed to be settleable without empirical in­vestigation.) Finally, Hempel's satisfaction and Glymour's bootstrapping criteria again yield concepts in accordance with which one calculates a priori whether e is evidence that h.

What's wrong with this a priori assumption? Let's return to the case of Thomson versus Hertz concerning the electrical character of cathode rays. On the basis of his 1883 experiments in which no deflection of the cathode rays was produced Hertz concluded that his results were evidence, indeed conclusive evidence, that cathode rays are electrically neutral. Thomson in 1897 rejected this claim, not on a priori grounds, but on empirical ones. He assumed, on empirical grounds, that if cathode rays are electrically charged particles, then, since there is gas in the cath­ode tube, when these charged particles pass through the tube they will ionize the gas molecules producing positive and negative charges that will neutralize the charge on the metal plates between which the cathode rays travel.

Similarly, recent scientists offered an empirical reason for rejecting the claim that the burned animal bones in the same layer as stone tools and sediment that looks like wood ash is evidence that the first culinary campfires were built by Peking Man in caves in China between 200,000 and 500,000 years ago. The empirical reason was the new discovery that although the sediment looks like wood ash, it is in fact not this but fine minerals and clay deposited by water.

I am not claiming that all evidential statements are empirical. There are cases where enough information, or at least information of the right sort, is packed into the e-statement to make the claim that e is evidence that h a priori. For example, suppose that e reports that there is a fair lottery consisting of 1,000 tickets, one of which will be selected at random, and that Sam owns 950 of these tickets. Suppose h is the hypothesis that Sam will win. Then, I think, the claim that e is evidence that h is an a priori claim. No empirical fact will render it false.

Let me say that an evidence statement of the form “e is evidence that h” is empirically incomplete if the truth of the statement depends on em­pirical facts in addition to those reported in e. The evidential claims in the two scientific examples at the beginning are empirically incomplete: Hertz's claim that the absence of electrical deflection in his cathode experiments is evidence that cathode rays are electrically neutral, and the claim of the earlier archaeologists concerning evidence for the hypothesis about the first campfires. Both evidential claims were falsified by later empirical discoveries.

A priorists about evidence will surely have a reply. Before considering it, however, let me draw an obvious conclusion, albeit a conditional one. If evidential claims, or many of them, are empirical, not a priori, then it is scientists not philosophers who are in the best position to judge whether e, if true, is evidence that h, and how strong that evidence is. If evidential claims are, by and large, empirical, this, I suggest, is and ought to be an important reason why scientists don't consult philosophical theories of evidence when they try to settle disagreements over evidential claims. Philosophical theories would make such disagreements settleable on a priori grounds. Accordingly, scientists may find philosophical theories of evidence wanting because they give a very mistaken idea of how eviden­tial disputes are usually settled.

4.