Signal Detection, Noise, and the Evolution of Communication

Abstract

Signal detection theory has had limited application in studies of animal communication. Yet by specifying constraints placed by noise on a receiver’s performance, it provides a way to investigate optimal performance and thus the evolution of communication. Noise in this case is anything influencing a receiver’s receptors other than a signal of interest. The essential features of signal detection theory are (1) a distinction between the detectability of a signal in noise and the criterion or threshold for a receiver’s response and (2) a realization that any decision by a receiver to respond has four possible outcomes, not all of which are independent. Although presented here in terms of a receiver’s threshold for response to one kind of signal, signal detection theory applies also to more complex criteria for response as well as complex discriminations among multiple signals. A receiver’s optimal performance always depends on the payoffs of the four possible outcomes of a decision to respond and on the detectability of a signal. By incorporating detectability, signal detection theory can provide a complete explanation for the evolution of exaggerated signals. An alternative explanation, based only on sexual selection and necessary costs of signals, does not do so. In particular, signal detection theory shows that exaggeration of signals should evolve so as to improve the detectability of signals by receivers. By shifting the emphasis from a receiver’s preferences and to its performance, this theory also clarifies the co-evolution of signalers and receivers. The result is a signal-detection balance, in which signals reach optimal but not ideal detectability and receivers reach optimal but not ideal performance. The crucial importance of the detectability of signals by receivers means that noise in natural situations, just as much as costs and benefits for the participants, determines the features of communication.