Does cheating pay? Re-examining the evolution of deception in a conventional signalling game

The study of reliability, or ‘honesty’, in communication between individuals with conflicting interests has been a major focus of game theoretical modelling in evolutionary biology. It has been proposed that mixed populations of honest and deceptive signallers can be evolutionarily stable in a model of conventional, or ‘minimal cost’, signals of competitive ability, and evolutionary simulations have been presented to support this hypothesis. However, we find that these results are questionable on both theoretical and methodological grounds. Here, we examine the theoretical issues raised by this model and examine the proposed ‘cheating’ strategy through the use of a genetic algorithm. Our evolutionary simulations do not support the hypothesis that deception can be evolutionarily stable in this game. Intuition and common sense have it that animals communicate using ambiguous threat displays that have an underlying probabilistic mixed strategy type of mechanism, but there remains no working game theoretical model of such a communication system.