Abstract

Given the existence of alternative physical-geometrical theories that agree on all possible evidence with some standard theory such as general relativity, which empirical equivalent should we project on the available evidence? Epistemological realism, of which Clark Glymour and Michael Friedman are the principal exponents, contends that we should accept the standard theory because it is better tested. ;Our essay explores this view. We show epistemological realism to be problematical in several respects and suggest additions and modifications that appear to us to be essential for philosophical adequacy. To begin, we isolate in Hans Reichenbach's writings, two methods for generating empirical equivalents and show them to be incompatible. Therefore, the epistemological realists' focus on Reichenbach's geometrical method for creating empirical equivalents requires justification. In providing the same, we argue that the geometrical method is a more powerful tool for producing empirical equivalents while on both strategies the standard theory is better confirmed than the non-standard alternatives. ;Next, Friedman and Glymour assert that epistemological realism does not commit one to the analytic/synthetic distinction. Also, that epistemological realism does not entail a theory of meaning that implies non-standard geometries are actually compatible with the standard empirical equivalent. We show that to the contrary, Glymour and Friedman must endorse at least one of these semantic doctrines. That they cannot avoid both. Then, on the assumption that Glymour and Friedman reject the analytic/synthetic distinction, we recast epistemological realism to incorporate the compatibility hypothesis. ;Finally, Friedman and Glymour contend that a space-time theory can be better confirmed than the collection of its observational consequences. Our argument against Glymour includes a defense of Bayesian confirmation theory and establishes Glymour's bootstrap constraint as a sufficient rather than a necessary condition of confirmation. Against Friedman, who denies old evidence can confirm hypotheses and assumes theories, but not collections of observational consequences, evolve by conjunction, we show how these collections can in fact be taken to evolve conjunctively. In modifying epistemological realism to appeal to scientific goals in addition to truth, we believe a more plausible vindication of theories is obtained