Optimal use of simplified social information in sequential decision-making

AO_SCPLOWBSTRACTC_SCPLOWSocial animals can improve their decisions by attending to the choices made by others. The rewards gained by attending to this social information must be balanced against the costs of obtaining and processing it. Previous work has investigated the behaviour of rational agents that respond optimally to a full sequence of prior decisions. However, such full sequences are potentially difficult to perceive and costly to process. As such, real animals are likely to rely on simpler forms of information when making decisions, which in turn will affect the social behaviour they exhibit. In this paper I derive the optimal policy for rational agents responding to specific simplified forms of social information. I show how the behaviour of agents attending to the total aggregate number of previous choices differs from those attending to more dynamic information provided by the most recent prior decision, and I propose a hybrid strategy that incorporates both information sources to give a highly accurate approximation to the optimal policy with the full sequence. Finally I analyse the evolutionary stability of each strategy depending on the cost of cognition and perception, showing that a hybrid strategy dominates when this cost is low but non-zero, while attending to the most recent decision is dominant when costs are high. These results show that agents can employ highly effective social decision-making rules without requiring unrealistic cognitive capacities, and point to likely ecological variation in the social information different animals attend to.