GATSBY COMPUTATIONAL NEUROSCIENCE UNIT

The Game of Life: Frugal Sampling Makes It Simpler

In many of our decisions we cannot consult explicit statistics telling us about the relative risks involved in our actions. In lieu of explicit statistics, we can arrive at proxy statistics by sampling from the payoff distributions that embody the alternatives they face. How competitive these proxies and our ultimate choices are depends on the size of the sample we take. Recent studies of decisions from experience observed that people tend to rely on genuinely small samples from payoff distributions, and on functionally small samples (via recency). In this paper, we formally investigate possible reasons behind people’s reliance on small samples. Three key results emerged: First, small samples amplify the difference between the average earnings associated with the payoff distributions, thus rendering choices simpler. Second, people who use a simple choice heuristic or choose in accordance with prospect theory benefit from this amplification effect, while Bayesian updaters may not. Third, though not providing a strictly veridical portrayal of the world, small samples can give rise to surprisingly competitive choices. We discuss these findings in the contexts of other debated benefits of small samples and the influential life-is-a-gamble metaphor.