Hidden preferences and the evolution of signals

Abstract

Mechanisms concerned with the recognition of biological signals act as important agents of selection on the appearance or ‘form’ of signals. Recognition of a given signal form can be achieved by many equally efficient alternative mechanisms. These alternative mechanisms will be selectively neutral and subject to change by genetic drift, thus preventing the fixation of a signal form that is optimal in releasing the receiver’s behaviour. Because signal form is a multidimensional trait with an almost infinite potential to vary, it is expected that some novel forms of signals always exist that elicit responses more readily than any of those signals that the receiver has experienced during its evolutionary history. The existence of such ‘hidden preferences’ is illustrated in simple models of recognition mechanisms based on artificial neural networks. The fundamental evolutionary instability of recognition mechanisms perhaps explains why biological signals are so variable in form, and why, in experiments, animals sometimes show greater responses to novel forms of signals than they do to familiar forms.