Our task consisted of sequences of 120 two-alternative forced choice trials. On each trial two stimuli were drawn from a set of four stimuli and shown to the participant, who had to choose one. Following the choice a numerical reward between 1 and 99 was displayed under the chosen stimulus. Then, the next trial began. Participants were instructed to try to maximize the total number of reward points throughout the experiment. The reward on each trial depended on the participant’s choice: each stimulus was associated with a specific reward distribution from which rewards were sampled. The four reward distributions associated with the four stimuli were approximately Gaussian and followed a two-by-two design: the mean of the Gaussian could be either high or low (60 or 40), and the standard deviation could be either large or small (20 or 5), resulting in four reward distributions in total ((60, 20), (60, 5), (40, 20) and (40,5)). We recorded choices and pupil dilation traces from 27 human subjects, which completed 4 di!erent sequences (blocks) of 120 trials each. The data they generated is formatted as a MATLAB table named "dt", with each row corresponding to a single trial, and each column (accessible by name) corresponding to a measured variable.

Columns:

• ID: unique participant identifier.
• Block: Block number
• trial: Trial number
• stim1: ID of stimulus on the left side of the screen.
• stim2: ID of stimulus on the right side of the screen.
• stim_chosen: ID of the chosen stimulus
• rt: reaction time in seconds.
• reward: reward in points.
• pupil_stimuli: pupil dilation time series, sampled at 10 Hz, % change relative to baseline, first sample time-locked to stimulus onset.
• pupil_reward: pupil dilation time series, sampled at 10 Hz, % change relative to baseline, first sample time-locked to reward onset

Stimulus ID corresponds to associated reward distribution (1: (60, 20), 2: (60, 5), 3: (40, 20), 4: (40, 5)). For a more detailed description of the task and pupil data
processing, see bioarxiv pre-print.