Learning Patterns in Noise: Environmental Statistics explain the Sequential Effect

Abstract

Effects of trial history, or sequential effects, are typically observed in perceptual, motor, and decision making tasks and explained by subjects’ irrational sensitivity to local patterns in stimulus history. We propose that in 2 alternative forced choice reaction time tasks (2AFC), sequential effects are a consequence a rational agent engaging in probability learning but with an inappropriate world model for 2AFC. We manipulate subjects’ world model and show expected changes in sequential effects. Sequential effects are at least in part driven by subjects’ beliefs about their environment.


Back to Table of Contents