Why behaviour patterns that animals learn socially are locally adaptive

Recent models of the social transmission of behaviour by animals have repeatedly led their authors to the counterintuitive (and counterfactual) conclusion that traditional behaviour patterns in animals are often not locally adaptive. This deduction results from the assumption in such models that frequency of expression of socially learned behaviour patterns is not affected by rewards or punishments contingent upon their expression. An alternative approach to analysis of social learning processes, based on Staddon–Simmelhag’s conditioning model, is proposed here. It is assumed that social interactions affect the probability of introduction of novel behaviour patterns into a naive individual’s repertoire and that consequences of engaging in a socially learned behaviour determine whether that behaviour continues to be expressed. Review of several recently analysed instances of animal social learning suggests that distinguishing processes that introduce behaviour patterns into the repertoires of individuals from processes that select among behavioural alternatives aids in understanding observed differences in the longevity of various traditional behaviour patterns studied in both laboratory and field. Finally, implications of the present approach for understanding the role of social learning in evolutionary process are discussed. During their lifetimes, individual animals can acquire behaviour in one of two ways: by individual learning or by social learning. Individual learning refers to behaviour acquired by an animal as the result of its own experience of the rewards and punishments contingent upon engaging in various acts, while social learning refers to those instances in which the acquisition of behaviour is influenced by observation of or interaction with another animal or its products (Heyes 1994). Classic examples of socially learned behaviour include the song dialects of white-crowned sparrows, Zonotrichia leucophrys (Marler 1970), sweet-potato washing and wheat placer mining exhibited by macaques on Koshima Island in Japan (Kawamura 1959) and the termite fishing engaged in by chimpanzees in Gombe National Park, Tanzania (Goodall 1973). The numerous behavioural processes supporting social learning in animals have been discussed at length elsewhere (Galef 1988; Whiten & Ham 1992; Heyes 1994) and will not be considered further here. Both individual and social learning are forms of phenotypic plasticity enabling animals to acquire behaviour that is adaptive in local habitat (Boyd & Richerson 1988). Theoreticians have argued that individual learning and social learning have different patterns of costs and benefits that make one or the other superior in any given environment. For example, Rogers (1988) discussed a hypothetical species, the ‘snerdwump’, whose members inhabit a variable environment and learn what foods to eat either by individual learning (sampling among available foods and discovering which is the most nutritionally valuable) or by copying the food choices of a member of the previous generation. In environments that are relatively constant across generations, snerdwumps that avoided exposure to poisons by copying the food choices of their elders would prosper, while snerdwumps that lived in environments that fluctuated significantly between generations and learned what to eat by copying members of the previous generation would be unable to discover superior foods that their brethren might identify while learning what to eat by trial and error. Psychologists studying behavioural traditions in animals have generally assumed that no learned behaviour, whether acquired socially or individually, will be long maintained in an individual’s repertoire unless that behaviour is at least as likely to produce rewards as are available alternatives (Galef 1976; Heyes 1993). Furthermore, most students of social learning have assumed implicitly 0003–3472/95/051325+10 $08.00/0 ? 1995 The Association for the Study of Animal Behaviour