Abstract

Evolutionary biology, indeed any science that attempts to reconstruct prehistory, faces practical limitations on available data. These limitations create the problem of contrast failure: specific observations may fail to discriminate between rival evolutionary hypotheses. Assessing the risk of contrast failure provides a way to evaluate testing protocols in evolutionary science. Here I will argue that part of the methodological critique in the Spandrels paper involves diagnosing contrast failure problems. I then distinguish the problem of contrast failure from the more familiar philosophical problem of underdetermination, and demonstrate how contrast failure arises in scientific practice with an investigation into Lewontin and White’s estimation of an adaptive landscape.