A computational approach to information-sharing and social generalization in collective risky decision-making

The aim of the project A computational approach to information-sharing and social generalization in collective risky decision-making by Wataru Toyokawa was twofold:
(1) To develop theoretical models and conduct interactive online behavioural experiments to identify under which conditions social learning and information-sharing strategies yield adaptive and maladaptive group behaviour,
2) To estimate cognitive processes underlying integrations of socially shared information and individual experiences in search problems using computational modelling.

Focusing on computationally tractable gambles, the project’s goal was to establish a research paradigm that will be scaled to more complex questions tackled in larger projects. The project’s results will be published soon. Toyokawa says: “I believe that the reinforcement learning models used and developed here should be general enough to apply to non-human animals too.” This will soon be proofed because Toyokawa will collaborate with animal researchers in the cluster to synthesize understanding of different systems within a unified computational learning model.