BEER Lab Meeting - Maria Pykälä

Understanding information diffusion within social networks is fundamental to research across the social sciences from economics to cultural evolution. Traditionally, work on diffusion has either focused on structural features of social networks, or on evolved individual learning biases with no reference to the social network. The complete two-step social learning strategy, consisting of both selection of models from a network, and learning from the selected models, is poorly understood. The objective of this experiment is to empirically investigate subjects’ tie formation choices and learning strategies using a controlled laboratory experiment. Subjects are separated into two groups, demonstrators and social learners, and undertake a two-armed bandit task. Demonstrators are placed in different exogenous network structures and can receive information on which is the higher payoff choice through individual trial-and-error. In addition, demonstrators access choices and payoffs of their social ties. Social learners cannot learn through trial-and-error. Instead, combining experimental traditions from cultural evolutionary social learning experiments, and economic network experiments, social learners observe demonstrator network structures and purchase social ties with demonstrators to access social information regarding the higher payoff choice to make. Given social learners’ tie choices and access to information, using a partial strategy method enables us to have complete information regarding the learning strategies used in each possible informational situation. The proposed design hopes to simultaneously answer the following two questions: 1. Given a social network structure in which information is diffused, which nodes, and how many nodes, are preferred as ties to acquire information from? 2. Given social ties are in place, how is information used from them?