Social learning and cooperative behaviour:
Evaluating the influence of social learning
in social dilemmas

Abstract

Humans are characterised by a strong proclivity towards two traits: cooperation and social learning.
Our ability to cooperate in large unrelated groups and selectively copy and improve upon cultural
traits through cumulative cultural evolution are fundamental drivers for our widespread success as a
species. However, we know comparatively little regarding the extent that cooperative behaviour
spreads through social learning. This thesis focused on three widely studied social learning strategies
(payoff bias, conformity and prestige bias) and investigated their influence on cooperative behaviour
across three studies. I used online experiments (studies 1 and 2) and an agent-based model (study 1) to
assess cooperation (and spite) in economic games and a questionnaire and semi-structured interviews
to investigate social learning and cooperation within a real-life context, environmental sustainable
behaviour (study 3).

In study 1, model comparison indicated that the changes in cooperation observed across rounds in a
cooperative game were best explained by payoff biased social learning than by prestige bias or
conformity, though a follow up agent-based model showed that the effect of social learning was small
compared to variation within participant’s inclination to cooperation. There was additionally some
evidence of higher cooperation under a snowdrift payoff structure than under a prisoner’s dilemma.
Study 2 found marked preferences towards altruism over spiteful behaviour: participants were not
influenced by the source (success weighted or frequency weighted) of social information but were
slightly more altruistic after viewing altruistic social information. Study 3 found indirect support for
payoff bias and conformity within real-life sustainable behaviour decisions, but little support for
prestige bias. Individuals reported perceiving strong environmental injunctive norms in their friends
and family (conformity) and that the biggest barrier to their sustainable behaviour was the costs of
such behaviour (payoff). Sources of social information such as familiar or knowledgeable individuals
were generally preferred to conformity or high status weighted social information. Like study 1, this
effect was small compared to the influence of individual’s personal norms towards sustainability. In
addition, experimentally manipulated vignettes framed with payoff, conformity or prestige information did not encourage individuals to choose a more sustainable over less sustainable participant reward.

Overall, the findings in this thesis indicate a clear hierarchy in social learning strategy use: payoff bias
was the most influential, then conformity and then prestige. This generally supported previous
research identifying the influence of payoff bias but contrasted with strong theoretical predictions
concerning prestige. However, the impact of social learning throughout this thesis was small
compared to other influential factors. The thesis is concluded by discussing some methodological and
theoretical limitations of the studies presented alongside ways that these general findings may be built
upon by further applied and theoretical research.