Abstract

The biological and social sciences both deal with populations that are heterogeneous with regard to important causes of interest, in the sense that the same cause often exerts very different effects upon distinct members of the population. For instance, welfare-to-work programs are likely to have different effects on the economic prospects of trainees depending on such variables as education, prior work experience, and so forth. Moreover, it is rarely the case in biology or social science that all such complicating variables are known and can be measured. In such circumstances, generalizations about the effect of a factor in a given population average over these differences, and hence take on a probabilistic character. Consequently, a causal generalization that holds with respect to a heterogeneous population as a whole may not hold for a given sub-population, a fact which raises a variety of difficulties for explanation and prediction. ;The overarching theme of the dissertation is that knowing how a cause produces its effect is the key to knowing when a particular causal relationship holds and when it does not. More specifically, the proposal is the following. Suppose that X is a cause of Y in the population P. Then there is a mechanism, or mechanisms, present among at least some of the members of P through which X influences Y. So if we know the mechanism and the kinds of things that can interfere with it, then we are in a much better position to say when the causal generalization will hold and when it will not. This intuitive idea has been endorsed by several philosophers; however, what has been lacking is a systematic exploration of the proposal and its consequences. That is what I aim to provide. The approach to the heterogeneity problem is developed in the context of an example drawn from biomedical science, namely, research into the causal mechanisms by which HIV attacks the human immune system. Moreover, I argue that my approach to the problem of heterogeneity sheds new light on some familiar philosophical issues that are relevant to the biological and social sciences, namely, ceteris paribus laws and methodological holism versus methodological individualism