Dartmouth Events

Your favorite color makes learning more adaptable and precise

Learning from reward feedback is essential for survival but can become extremely challenging when choice options have multiple features (color, shape, etc.) and feature values (curse of dimensionality). In this talk, I propose a general framework for learning from reward feedback in dynamic, multi-dimensional environments via encoding and updating the average value of individual features. The framework predicts that this feature-based learning occurs not just because it can reduce dimensionality, but more importantly because it can increase adaptability without compromising precision. I present results from a set of experiments in which human subjects change their model of the environment according to this prediction, and moreover, suggest a plausible mechanism for how this could happen in the brain. Finally, I discuss the relevance of these results for studying social behavior.