Understanding Frequency-dependent Causation

Causal effects often vary with the relative frequency of the causal property in the population of interest. (For example, "being a female candidate for high office causes a stir" only because such events are rare.) Such causal effects are called frequency-dependent, and they pose interesting problems for both philosophy of causality and philosophy of biology. 1 Elliot Sober 2 argues that certain cases of frequency-dependence "constitute counterinstances" of the causal model proposed by Ronald Giere. 3 To solve the problem Sober raises, Giere's model must provide a satisfactory solution to the question: How can a property causally act on all members o f a population when the property is effective only when it is rare? In response to Sober, Giere 4 has recently proposed a special extension to his model to accommodate frequency-dependent claims. John Collier's s proposed defense of Giere's original model restricts the domain of frequencydependent claims. In this paper I argue that neither Giere's nor Collier's proposals adequately answers the question posed by Sober's problem. More importantly, I shall argue that neither proposed revision is needed;that Giere's original model adequately answers the above question provided frequencydependent causation is correctly understood. While my primary aim is to arrive at this solution, I shall also show how it sheds light on the appropriate methods for learning about frequency-dependent causal effects in science.