56 THE JOURNAL OF PHILOSOPHY rewards which accords all credit, and so all the personal benefits tha go along with credit, to the first research program to discover a particular fact or procedure, and none to other programs pursuin the same goal. As a consequence of the priority system, workers in competing research programs are involved in a winner-takes-all rac for personal rewards. Call this the rewards race. I will argue that the strong parallel between the payoff structure of the benefits race and that of the rewards race, that is, between the scheme by which science benefits society and the scheme by which science rewards its practitioners, is not accidental: the fact that scientific rewards are distributed according to the winner-takes-all priority system, rather than some other scheme, is explained by the winnerconfers-all manner in which science benefits society. The structure of the benefits race, then, explains society's implementation of the rewards race.

The form of the explanation is as follows. A scientific reward scheme such as the priority rule acts as a system of incentives, encouraging researchers to devote their time and energy to some research programs in preference to others. Different reward schemes, then, may result in different allocations of resources among competing research programs. Society has an interest in adopting a reward scheme that promotes an allocation with a relatively high expected payoff. I will show that the priority system promotes an especially efficient allocation of resources in" winner-confers-all" situations, that is, in situations where almost all benefit is extracted from a goal the very first time