Social learning strategies modify the effect of network structure on group performance

Abstract

The structure of communication networks is an important determinant of the capacity of teams, organizations and societies to solve policy, business and science problems. Yet, previous studies reached contradictory results about the relationship between network structure and performance, finding support for the superiority of both well-connected efficient and poorly connected inefficient network structures. Here we argue that understanding how communication networks affect group performance requires taking into consideration the social learning strategies of individual team members. We show that efficient networks outperform inefficient networks when individuals rely on conformity by copying the most frequent solution among their contacts. However, inefficient networks are superior when individuals follow the best member by copying the group member with the highest payoff. In addition, groups relying on conformity based on a small sample of others excel at complex tasks, while groups following the best member achieve greatest performance for simple tasks. Our findings reconcile contradictory results in the literature and have broad implications for the study of social learning across disciplines.

Figure 1. Performance over time for different strategies in a fully connected network.

N denotes the number of components of the system and K represents the number of interdependent components (see Methods); s stands for sample size. (a) Simple environment with a global optimum and (b) complex environment with multiple local optima. Red dotted lines: best member (s=3); red dashed lines: best member (s=9); turquoise thin lines: conformity (s=3); turquoise thick lines: conformity (s=9); black dotted lines: individual learning; grey dashed lines: random copying. Based on n=100 agents and 1,000 repetitions.

Figure 2. Number of unique solutions in the population over time.

N denotes the number of components of the system and K represents the number of interdependent components; s stands for sample size. (a) Simple environment with a single optimum and (b) complex environment with multiple local optima. Red dotted lines: best member (s=3); red dashed lines: best member (s=9); turquoise thin lines: conformity (s=3); turquoise thick lines: conformity (s=9); black dotted lines: individual learning; grey dashed lines: random copying. Based on n=100 agents and 1,000 repetitions.

Figure 3. Performance of different networks as a function of social learning strategy.

(a) Group performance averaged across efficient (red solid lines) and inefficient (turquoise dashed lines) networks. Shadings around the lines show the region between the best and worst performing network in each category. (b) Average performance at the final time step (t=200) for each network. Error bars show +−2 s.e. of the mean. Left panels: individuals rely on the best member (s=3) strategy; right panels: individuals rely on the conformity (s=3) strategy. Inefficient networks outperform efficient networks when individuals rely on the best member strategy. Efficient networks outperform inefficient networks when individuals rely on the conformity strategy. Results in the left panel replicate the findings of Lazer and Friedman while results in the right panel are in line with Mason and Watts. Based on n=100 agents and 1,000 repetitions.

Figure 4. Relationship between network diameter and average performance.

(a) Individuals rely on the best member (s=3) strategy; (b) individuals rely on the conformity (s=3) strategy. Red circles indicate efficient networks and turquoise triangles indicate inefficient networks. Straight lines show best linear fits. The relationship between diameter and performance is positive when individuals rely on the best member strategy (Pearson correlation coefficient r=0.83) and is negative when they rely on the conformity strategy (r=−0.77).