Conformity biased transmission in social networksby Andrew Whalen, Kevin Laland

Journal of Theoretical Biology


Jury Selection as a Biased Social Process

Hayward R. Alker Jr., Carl Hosticka, Michael Mitchell

Transmission through a biased graphene bilayer barrier

Johan Nilsson, A. H. Castro Neto, F. Guinea, N. M. R. Peres

Are the Networks Biased? ``Calling'' States in the 2000 Presidential Election

J. Wilson Mixon, Jr., Amit Sen, E. Frank Stephenson

Brittleness in Electric Transmission Network

Yan Limei, Jin Hongzhang, Rong Panxiang


Conformity biased transmission in social networks

Andrew Whalen n, Kevin Laland

School of Biology, University of St Andrews, Harold Mitchel Building, St Andrews, Fife, KY16 9TH, UK

H I G H L I G H T S  We modeled the invasion of novel behaviors through social networks.  Network structure, conformity bias, and learning biases were varied.  When learners used a conformity bias novel behaviors were less likely to spread.  Gross network structure had limited impact on the likelihood a novel behavior spread.  However, high degree nodes were disproportionately the source of novel behaviors. a r t i c l e i n f o

Article history:

Received 27 February 2015

Received in revised form 28 May 2015

Accepted 17 June 2015

Available online 30 June 2015


Social networks

Social learning


Cultural transmission a b s t r a c t

In this paper we explore how the structure of a population can differentially influence the spread of novel behaviors, depending on the learning strategy of each individual. We use a series of simulations to analyze how frequency dependent learning rules might affect how easily novel behaviors can spread through a population on four artificial social networks, and three real social networks. We measured the likelihood that a novel behavior could spread through the population, and the likelihood that there were multiple behavioral variants in the population, a measure of cultural diversity. Surprisingly, we find few differences between networks on either measure. However, we do find that where a behavior originated on a network can have a substantial impact on the likelihood that it spreads, and that this location effect depends on the learning strategy of an individual. These results suggest that for first-order analysis of how behaviors spread through a population, social network structure can be ignored, but that the social network structure may be useful for more fine-tuned analyses and predictions. & 2015 Elsevier Ltd. All rights reserved. 1. Introduction

Many animals are able to learn behaviors socially, by attending to the behavior of other individuals (Laland and Galef, 2009). This allows individuals to learn novel behaviors more quickly and cheaply than through asocial trial-and-error. One important consequence of social learning lies in its ability to facilitate the transmission of behavioral variants through a population. However social learning is not intrinsically adaptive (Rogers, 1988;

Rendell et al., 2010). Theoretical models suggest that in order for social learning to increase the fitness of animals in a population, individuals must be selective in who, when, and how they use social information (Boyd and Richerson, 1985; Rogers, 1988;

Henrich and McElreath, 2003; Laland, 2004; Rendell et al., 2011).

This implies that animals will rarely copy others randomly, a theoretical finding that is borne out by extensive research into animals and human social learning (Hoppitt and Laland, 2013).

Heuristics specifying the circumstances under which individuals copy others are often termed ‘social learning strategies' (Laland, 2004), although ‘transmission' biases (Boyd and Richerson, 1985;

Henrich and McElreath, 2003) and ‘trust' (Corriveau and Harris, 2010) are related concepts. Previous research has employed theoretical and computational tools to examine these questions (see Rendell et al., 2010; Hoppitt and Laland, 2013). However much of this research has treated who and how questions independently. In this paper we examine together how the choice of whom an individual learns from, and how they learn, collectively affect how novel behaviors spread through a social network.

We build on a long line of modeling research looking into the how question by analyzing what types of strategies are likely to evolve in unstructured populations. In early work on this topic,

Boyd and Richerson (1985) explored the evolutionary outcomes of a conformity biased transmission strategy in a spatially variable environment, and concluded that human social learning should commonly evolve to exhibit a conformity bias. They defined conformity biased transmission (henceforth ‘conformist transmission') as transmission where the likelihood of adopting the majority behavior was greater than the observed frequency of

Contents lists available at ScienceDirect journal homepage:

Journal of Theoretical Biology 0022-5193/& 2015 Elsevier Ltd. All rights reserved. n Corresponding author. Tel.: þ44 01334 463009.

E-mail address: (A. Whalen).

Journal of Theoretical Biology 380 (2015) 542–549 the behavior in the population. Subsequently, Henrich and Boyd (1998) found that in a spatially and temporally changing environment, conformist transmission would evolve whenever social learning would evolve. Wakano and Aoki (2007) extended and upheld these results, but concluded that a very strong conformity bias might not be adaptive, because it may present novel beneficial behaviors from spreading through the population, thus limiting the advantage of social learning. Both Nakahashi et al. (2012) and

Kandler and Laland (2013) reached similar conclusions; the first using an island model, where individuals were spread on a series of environmentally varying islands, and the second using reactiondiffusion models. These results suggest that conformity biased transmission is likely to be adaptive under a wide variety of situations.

While the above studies do explore both who individuals learn from (e.g. learning from island neighbors) and how they do (applying a conformist transmission rule), their representation of social structure is at best, relatively crude. Many of these models examine social structures at a comparatively large scale that driven by differences in the underlying environmental structure.

In reality social structure will arise at several different scales, including within demes. Directed social learning may arise through more fine-grained spatial structure, where individuals disproportionately learn from others that are physically proximate, or more salient in some other respect, compared to individuals within the same deme but more distant in space, or less attractive as models. Social structure may play an important role in the value of social learning since the value of social information depends not only on how you use it, but also from whom it came. If the structure of the population inhibits the spread of novel behaviors, it may decrease the usefulness of social learning. Conversely, if the structure allows only beneficial behaviors to spread, it may increase the usefulness of social learning. Thus, it is likely that the usefulness of a given learning strategy, like conformity biased copying, may depend on the relationships of who learns from whom. We call this set of relationships the social network of the population.