Two kinds of drift?

Philosophers of biology have recently been debating about whether random genetic drift is a distinct process from that of natural selection. One camp argues that drift is a process of “indiscriminate sampling” that is logically and ontologically distinct from the “discriminate sampling” process that is natural selection. The other camp argues that, rather than being two autonomous processes, natural selection and drift are just two aspects or facets of a single process. I argue that the two positions involved in this dispute are not mutually exclusive. Indeed, by explicitly taking into account its sources, I argue that two kinds of drift are theoretically possible. The first kind – that I call circumstantial-drift – involves a process of its own, separate from the process of natural selection, or, more specifically, it requires that differences in a circumstantial trait cause differences in reproductive output of biological entities. The second kind – that I label probabilistic-drift – requires no other process than a probabilistic causal relation between a heritable physical trait and reproductive output, i.e. requires only the causal relation on which the work of natural selection could also depend. If both of these kinds of drift are theoretically possible, then neither defining, nor empirically identifying drift need to be zero-sum games. [ABSTRACT FROM AUTHOR]