Fig. 1: Experimental paradigm, block types, and time course of performance.

a Timeline of each block and a single trial of the experiment. At the beginning of each block, two novel stimuli (abstract visual objects) were introduced. On each trial, animals indicated a choice by making a saccade toward one of the two options on the left and right sides of fixation. The selection of each stimulus was rewarded probabilistically based on three reward schedules. Reward contingencies were reversed between better and worse options on a randomly selected trial (between trials 30 and 50). b Different block types. In What blocks (top), reward probabilities were assigned based on stimulus identity, with a particular object having a higher reward probability. In Where blocks (bottom), reward probabilities were assigned based on the location of the stimuli, with a particular side having a higher reward probability regardless of the object appearing on that side. c Outline of the What-only task. Here, only What blocks were used for the entire experiment. Vertical dotted lines, rev, indicate a random reversal point within each block. d Outline of the What/Where task. In this task, What (black) and Where (gray) blocks were randomly interleaved. e–g Time course of performance of the monkeys in each group, measured as the probability of choosing the better option, P(Better), separately for different tasks and block types. Shaded regions indicate error bars. Insets show the averaged performance by reward schedules. Error bars = SEM across subjects (controls: n = 4 in What-only, n = 6 for What/Where; amygdala: n = 4 for both tasks; VS: n = 3 for both tasks). Source data are provided as a Source data file.