Learning behaviorally stable solutions to producer--scrounger games

Animal decision making is influenced by past experience in many biological contexts such as mating, avoiding predators, and foraging. In behavioral games, what constitutes a good or bad decision about which alternative to use depends on the behavior of other individuals. Solutions to games can take the form of a stable equilibrium frequency (SEF) of alternative tactics. In this study, we ask whether individuals within flocks of ground-feeding passerines (Lonchura punctulata) engaged in a producer--scrounger game adjust their behavior and converge on the SEF by using a fixed rule of thumb or by learning to estimate payoffs by the process of responding to contingencies in reinforcements obtained from each alternative tactic. After being trained either in a high-scrounging (HS) or a low-scrounging (LS) food condition, flocks of birds were provided with identical foraging conditions over 2 successive test phases in which we expected the SEF of scrounging to decrease and then to increase. Birds trained in the HS condition scrounged more than those trained in the LS condition and continued to do so even when subsequently tested in the same conditions. This effect of past experience is inconsistent with the use of a fixed rule alone. An improvement in the efficiency of scrounging behavior within experimental phases provided further evidence of learning in this game-theoretic context. This experiment provides the first empirical evidence that group-level adjustments in scrounger use to different environmental conditions are mediated by learning the payoffs associated with each tactic. Copyright 2010, Oxford University Press.