Abstract

Is a scientist who accepts a hypothesis in part due to moral, or prudential, or aesthetic considerations thereby open to criticism as unscientific? May such considerations play a legitimate role, even when considering the acceptability of a hypothesis from a purely epistemic vantage point? I believe they may. Such considerations cannot be entirely purged from science, but nor does their role render science an arational enterprise. ;Some hold that the acceptability of a hypothesis should be determined entirely by considering how probable it is compared to its rivals. This view assumes that truth is our only epistemic goal. But were this the case, there would be no rationale for accepting less-than-certain hypotheses. That rationale is to be found in the fact that our epistemic goal is two-fold. We seek to accept what is true, but we also seek new information. It is this pursuit of new information that justifies the risk of accepting what is false. ;Thus the epistemic evaluation of rival hypotheses must be in light of both aspects of our epistemic goal simultaneously. Such evaluations can be made by means of a decision- theoretic epistemology such as that developed by Isaac Levi and others. Such theories define the epistemic utility of a hypothesis as a function of its probability and its content. The hypothesis with the greatest epistemic utility---the "epistemic winner"---best meets both aspects of our epistemic goal jointly, and should thus be accepted. ;So far there seems no need to appeal to non-epistemic considerations like those referred to at the outset. However, the two aspects of our epistemic goal compete with one another and might be weighted in a variety of ways. In some cases, emphasizing truth over informativeness yields one epistemic winner, while the opposite emphasis yields a rival epistemic winner. Here is where moral, prudential and aesthetic considerations play a legitimate role. These non-epistemic considerations determine how our epistemic objectives are to be weighted. The general rule is: choose the weighting of epistemic objectives whose epistemic winner has the greatest non-epistemic utility among all epistemic winners