Quantifying diffusion in social networks: a Bayesian approach

Abstract

Social learning is often broadly defined as ‘learning that is influenced by observation of, or interaction with, another animal (typically a conspecific) or its products’ (Heyes 1994). However, the phrase ‘influenced by’ is unacceptably vague, and we prefer to characterize social learning as learning that is facilitated by observation of, or interaction with, another individual, or its products (Hoppitt and Laland 2013). In addition, we prefer the more specific term ‘social transmission’ to refer to the process by which behavioural traits spread through groups. We define social transmission as occurring when ‘the prior acquisition of a behavioural trait T by one individual, A, when expressed either directly in the performance of T or in some other behaviour associated with T, exerts a lasting positive causal influence on the rate at which another individual, B, acquires and/or performs the behavioural trait’ (Hoppitt and Laland 2013). The study of social learning was initially motivated by an interest in the cognitive or psychological mechanisms underpinning social learning (e.g. Galef 1988; Heyes 1994), leading to research conducted in controlled laboratory conditions (e.g. Zentall et al. 1996). In recent years, the focus has shifted to animal traditions and culture, driven by the discovery of group-specific behaviour in a number of taxa, including primates (e.g. Whiten et al 1999; S. Perry et al. 2003), cetaceans (e.g. Rendell and Whitehead 2001) and birds (e.g. Madden 2008). Such group-specific behaviour patterns often appear to be the result of different behavioural innovations spreading through groups by social transmission. Many researchers are now studying the conditions under which novel behavioural traits spread and form traditions in the field or in a captive group context, in which the subjects are free to interact with one another (e.g. meerkats (Suricata suricatta) (Thornton and Malapert 2009); vervet monkeys (Chlorocebus pygerythrus) (Van de Waal et al. 2010); and humpback whales (Megaptera novaeangliae) (Allen et al. 2013)). Such research has motivated the development of novel methods for studying social learning in freely interacting groups (Laland and Galef 2009; Kendal, Galef et al. 2010; Hoppitt and Laland 2013). Laboratory experiments provide valuable insights into learning mechanisms, and other aspects of social interaction, but it remains challenging in a natural or field context to ascertain whether social learning has occurred and quantify its impact without the use of sophisticated statistical methods. In our use of the terminology, ‘social transmission’ can be distinguished from ‘diffusion’, as the latter term refers to the observed spread of a trait through a group, irrespective of the cause of the spread. Therefore, a trait might be said to have diffused through a group without any evidence that this occurred by social transmission. For instance, the diffusion may result from independent asocial learning by each individual, or there may be an unlearned social influence on behaviour, as occurs, for instance, when animals influence each other’s movements, as is reported in sticklebacks (Atton et al. 2012).