Rational Compression In Choice Prediction

To successfully navigate its social environment, an agent must construct and maintain representations of the other agents that it encounters. Such representations are useful for many tasks, but they are not without cost. As a result, agents must make decisions regarding how much information they choose to store about the other agents in their environment. Using choice prediction as an example task, we illustrate the problem of finding agent representations that optimally trade off between downstream utility and information cost, before presenting the results of two behavioural experiments designed to examine this tradeoff in human social cognition. We find that people are sensitive to the balance between representation cost and downstream value, while still deviating from optimality.