Empirical tests of reciprocity theory: Problems in assessment

Observations and experiments designed to test Trivers' (1971) theory of reciprocal altruism face two difficulties. First, in many cases the costs and benefits of behaviors being exchanged cannot be expressed directly in terms of effects on the actors' fitness. This is particularly true when an exchange of cooperative acts involves different types of behavior, such as grooming and alliance formation in non-human primates. Second, in many social groups individuals differ widely in their ability to confer benefits on others. High-ranking animals can, for example, offer greater assistance than low-ranking animals and adults can help offspring more than offspring can help adults. As a result, what appears not to be a reciprocal interaction from the observer's point of view may in fact support Trivers' theory when costs and benefits are calculated from the animals' own perspectives. We discuss these issues with special reference to our own experiments on reciprocal exchanges of grooming and alliance formation in free-ranging vervet monkeys.