Heuristics, hacks, and habits: Boundedly optimal approaches to learning, reasoning and decision making

AbstractHumans regularly perform tasks that require combining information across several sources of information to learn, reason, and make decisions. Bayesian models provide a computational framework, and a normative account, for how humans carry out these tasks. However, exact inference is intractable in most real-world situations, and extensive empirical work shows that human behavior often deviates significantly from the Bayesian optimum. A promising possibility is that people instead approximate rational solutions using bounded available resources. In this workshop, we bring together leading researchers from cognitive science, neuroscience and machine learning to build a better understanding of bounded optimality in how humans learn, reason and make decisions.

Return to previous page