A shape appears on each trial. Predict which outcome (A or B) will follow. Each shape has a hidden probability of producing each outcome — learn the patterns through feedback. Based on probabilistic category learning and the Rescorla–Wagner model (1972).