Figure 1.

Task. Shown are the sequences of trial events and blocks. Each session consisted of four blocks with different action-outcome associations. Actions designated as 1 and 2 were counterbalanced across days between the leftward and rightward movements from the odor port to the left or right reward delivery well. Outcomes were short delay, long delay, big size, and small size, counterbalanced across the two sides (and actions) within each session. In the figure, drops of fluid indicate times of reward delivery, and those depicted with dotted lines indicate the time of omission of a reward delivered in the previous block or session. Epochs during the reward delivery period are color-coded to indicate positive and negative prediction error epochs in which reward delivery or omission was unexpected, and control epochs in which reward delivery or omission was fully expected at the beginning of the block. Comparisons were made during these epochs between firing rates at the beginning of blocks (first five trials) versus the end of blocks (last 15 trials). Odors at the beginning of trials instructed whether reward would be available only on the left (forced choice), only on the right (forced choice), or on either side (free choice). Thus both forced choice odors became predictive of the associated outcome toward the end of blocks when rats had learned the action-outcome associations for that block.