An empirical test of 2-dimensional signal detection theory applied to Batesian mimicry

Signal detection theory (SDT) has been invoked to help explain why imperfect mimics of particularly unprofitable or abundant models might experience no further selection to improve their mimicry. However, most tests of SDT have focused on single dimensions of mimetic phenotypes, or used multivariate techniques to compress many dimensions of phenotype into a single scale. Here, we explicitly tested SDT in both one and two dimensions by asking human subjects to discriminate computer-generated mimics and models that varied continuously in both size and/or color. We arrived at two major conclusions. First, although subjects can use prey size or color to help discriminate profitable and unprofitable prey that vary in only one dimension, responses of subjects to prey that vary in two dimensions are poorly represented by multidimensional SDT. Second, because different individuals within groups may use different strategies, the behavior of groups is often better fit by more complex models. In general, humans give more weight to color when making discriminations than is optimal. This bias may indicate that they believe that color has higher relative validity than size. More studies on the behavior of natural predators when foraging on multidimensional prey are urgently needed.