Jonathan Nelson, Max Planck Institute for Human Development, Germany
The Mathematics, Psychology, and Neuroscience of Information
Abstract: Beginning in the 1950s, several ideas were proposed in statistics for quantifying the expected usefulness of an experiment. These ideas have inspired research in diverse fields, ranging from psychology to medical decision making and machine learning. I will overview two current research projects. One project brings together entropy and information models from mathematics and physics, providing a unified explanation of diverse empirical findings in psychology. Another project explores how the human brain represents expectation of information for probabilistic classification. We show that expectation of information overlaps with some previously identified reward areas, but has different spatial and temporal signatures.