Evolution of strategies in modified sequential assessment games

Abstract

The sequential assessment game is one of the most important game-theoretic models of animal contests. It intends to model contests in which animals gain a progressively more accurate estimate of relative fighting ability by means of repeated bouts of fighting. The model predicts an evolutionarily stable strategy (ESS) that would correspond to an increasing sequence of thresholds for quitting the game. We report on simulated evolution of strategies in modified versions of the game and compare our results with theoretical predictions for the original model. Outcomes of these simulations corroborate some, but not all theoretical predictions for the sequential assessment game. In particular, our results suggest that theoretical analyses of the sequential assessment game with information asymmetry need to take into account factors that have hitherto been ignored in the literature.