The Path Space Conception of Components of Fitness

Marshall Abrams
Duke University, Center for Philosophy of Biology

     Full text: Not available
     Last modified: June 14, 2005
     Presentation date: 07/16/2005 11:00 AM in MACK 236
     (View Schedule)

Abstract
I describe what I call the "path space" conception of an organism and its environment. I argue that it provides a useful framework for thinking about the relationship between models in evolutionary theory and real-world biological processes, and suggest that the path space conception can provide the core of a unified conception of evolutionary "forces" and their interactions. There has been controversy in recent years over whether such forces are causal elements in the world or instead are merely artifacts of models; the path space conception provides a way of understanding evolutionary forces as real. The path space perspective can be viewed as way of extending the propensity interpretation of fitness, though it does not depend on the propensity interpretation of fitness. In this talk I focus on relationships between components of fitness rather than on relationships between natural selection and other forces such as migration and drift.

The path space conception treats an organism type (genotype or phenotype) together with the organism's environment as composing an abstract mechanism constituted by conditions which are relatively stable. Objective probabilities of interactions involving the organism and its environment then determine objective probabilities over sets of paths, i.e. sequences of states of the organism and the environment consistent with the stable conditions. The relevant probabilities can be viewed either as propensities or as probabilities of some other kind such as the author's "mechanistic probabilities". Various types of interactions between elements in the environment and the organism then have determinate probabilities. In particular, the probability of an occurrence such as the performance of a function is the probability of the set of paths in which such an interaction occurs. Conditional probabilities involving sets of paths are particularly useful for understanding interactions between the organism and the environment.

I will use the path space conception to help clarify three related issues concerning components of fitness:

Traits need not combine linearly; a trait may confer a greater advantage when combined with some traits than when combined with alternative traits. One can postulate and even measure such relationships between traits, especially in particular cases. However, we have had no clear general conception of the nature of the probabilistic and causal relationships between different traits.

Traits need not have their effects simultaneously. For example, a trait might produce functional effects only early in life, or only at night, or only when a predator approaches in a certain manner. Another trait might produce functional effects at other times. How do these different effects combine to produce an overall level of fitness?

It is often said that a particular trait under selection is both beneficial and detrimental. Such claims are puzzling. It seems that what's claimed is that trait A both increases fitness relative to alternative trait a and also decreases fitness relative to a, but it is not easy to provide a general analysis of this idea. Counterfactuals about non-existent combinations of traits might help, but such counterfactuals seem particularly problematic.