The investigators collected behavioral data from two macaque monkeys (M1 and M2) who were presented with a direction discrimination task with post-decision wagering. These monkeys received prior training to report the net direction of motion of a set of randomly moving dots. In each trial, the subject fixed their gaze on an initial fixation point, then red dots would appear on the periphery of the screen to indicate direction targets. After a delay, the subject would be presented with a random dot stimulus for 100–900 ms; motion direction and strength (proportion of dots which moved in a coherent direction out of the group of randomly moving dots) varied in each trial. In a random half of the trials, a third target (sure target) would appear on the screen following stimulus presentation. After a delay, the fixation point disappeared to signal the monkey to saccade to the appropriate direction target (right target for rightward motion and left target for leftward motion). The monkeys received a large reward for choosing the correct direction target; choosing the incorrect direction target resulted in a timeout. Monkeys could opt out of direction discrimination by choosing the third target (sure target) in trials where it was presented for a small reward.

A total of 86,622 and 60,733 trials were completed by monkeys M1 and M2, respectively. The researchers have shared data showing the frequency at which monkeys selected the correct direction target in relation to stimulus motion duration and strength. This catalog record also links to related software and code referenced by the associated publication to support the modeling of partially observable Markov decision processes (POMDPs).