Beyond existence and aiming outside the laboratory: estimating frequency-dependent and pay-off-biased social learning strategies

Abstract

The existence of social learning has been confirmed in diverse taxa, from apes to guppies. In order to advance our understanding of the consequences of social transmission and evolution of behaviour, however, we require statistical tools that can distinguish among diverse social learning strategies. In this paper, we advance two main ideas. First, social learning is diverse, in the sense that individuals can take advantage of different kinds of information and combine them in different ways. Examining learning strategies for different information conditions illuminates the more detailed design of social learning. We construct and analyse an evolutionary model of diverse social learning heuristics, in order to generate predictions and illustrate the impact of design differences on an organism's fitness. Second, in order to eventually escape the laboratory and apply social learning models to natural behaviour, we require statistical methods that do not depend upon tight experimental control. Therefore, we examine strategic social learning in an experimental setting in which the social information itself is endogenous to the experimental group, as it is in natural settings. We develop statistical models for distinguishing among different strategic uses of social information. The experimental data strongly suggest that most participants employ a hierarchical strategy that uses both average observed pay-offs of options as well as frequency information, the same model predicted by our evolutionary analysis to dominate a wide range of conditions.

Figure 1

(a,c) Instantaneous and (b,d) evolutionary dynamics of frequency-dependent and pay-off-biased social learning.

Figure 2

Sensitivity analysis for the evolutionary model of social learning strategies in the main text. Each row plots the frequencies of five different strategies (individual learning, unbiased social learning, positive frequency dependence, pay-off bias and pay-off conformity) for two-dimensional combinations of parameters. Each individual plot is the frequency of a single strategy after 5000 generations of simulations at all combinations of the two parameters labelled on each axis. (a(i)–(v)) u varied from 0 to 0.5, b varied from 0 to 0.5, while a=0.5+b. (b(i)–(v)) a−b varied from 0 to 0.5, b again from 0 to 0.5. (c(i)–(v)) u again varied from 0 to 0.5, B from 2 to 10. All other parameters not on axes were fixed at B/c=6, u=0.1, a=3/4, b=1/4, w₀=2. The most powerful inference from these simulations is that either pay-off bias (S) or pay-off conformity (SC) dominates the population, unless the environment is very unstable and individual learning is too costly, relative to fitness benefits of optimal behaviour.

Figure 3

(a) Proportion of optimal crop decisions, by round within farm. Vertical lines show 95% profile-likelihood confidence intervals. (b) Proportion of neighbours' crop decisions (circles) and yields (curve) inspected, by round within farm.

Figure 4

The function that determines the reliance on pay-off-biased learning, as a function of the observed difference in means, ${{\displaystyle \overline{\pi }}}_{1,t}-{{\displaystyle \overline{\pi }}}_{2,t}$. See the description of the hierarchical means/conformity strategy in the text. Solid curve, δ=1/10; dashed curve, δ=2.

Figure 5

Maximum-likelihood estimate of strength of positive frequency dependence for the best-fitting model, hierarchical compare means/frequency dependence. Solid curve indicates estimated probability of copying a choice, given its frequency in the group. Dashed line indicates same probability under f=1, unbiased social learning.