Fig 1. Rats reliably learn a “cloud of dots” visual decision-making task.

(a) Task design, with example stimulus frames for upper hemifield (top) and lower hemifield (bottom) trials (left: easy, right: difficult). Stimulus duration is 0.5 s; all other task epochs have variable duration. (b) Virtual head fixation algorithm, condition is active for portion of trial marked by green line in (a). (c) Proportion of trials completed increased as animals were trained on head fixation. Across animals trained on head fixation after learning the visual rule, the mean proportion of completed trials on day 1 of training was 0.483; this increased to 0.635 by day 9. Black trajectories denote animals whose neural recordings were included in this dataset, gray trajectories denote animals who were trained but no recordings were performed. (d) Animals typically reached stable performance above 90% accuracy on completed easy trials in fewer than 30 sessions (median = 16 sessions, +/−10 std). Color scheme as in (c). (e) Psychometric performance on single sessions after reaching performance criterion on easy stimuli, prior to neural recordings, for each animal included in neural dataset. Error bars indicate standard error of the mean. The underlying data for this figure are available for download from 10.17632/5ms7gcb67j.1.