The seminar is sponsored by NYU-ECNU Institute of Brain and Cognitive Science at NYU Shanghai.
Abstract:
Decision-making tasks such as navigation, negotiation, military strategy, writing a paper, and planning a career involve long sequences of decisions with multiple options at each step, leading to a combinatorial explosion of the decision tree. To make such decisions tractable, people must prune the tree. To investigate how people do this, we introduce a set of experimental paradigms based on a variant of tic-tac-toe. People play against each other, against computers of different strengths, choose between two moves, or evaluate positions. We analyze moves made, evaluations, reaction times, and eye movements. Using AI-inspired models, we break human skill in this task down into three components: the size of the decision tree, the noise level, and the quality of the evaluation function. Our results that people think remarkably like computers in tasks with large decision trees.
Biography:
Wei Ji Ma (native Dutch spelling: Whee Ky Ma) is Associate Professor of Neural Science and Psychology. He received his Ph.D. in physics from the University of Groningen and studied computational neuroscience as a postdoc with Christof Koch and Alex Pouget. Wei Ji spends his free time on RCEF, enjoying interactive theater and dance, playing board games and badminton, and exploring NYC's gastronomy.