Abstract

A correct analysis of hierarchical selection processes must specify 1) the objects that succeed differentially as units, and 2) the properties that provide the causal bases for differential success. Here I illustrate how failing to recognize the units/bases distinction creates a contradiction in Elliott Sober's recent account of selection. A revised criterion for units of selection is developed and applied to examples at several biological levels. Criteria for bases of selection are discussed in terms of the degree of context-dependence and directness of a property's effect on the success of units. The significance of previous work by Sober, Wimsatt and Brandon is thereby clarified