a, Schematic depicting the change in task from Go/NoGo (T5) to left/right choice (L/R). The motor response changed from Go to GoRight, and from NoGo to GoLeft. The category rule remained the same. b, Possible changes in neuronal responses between T5 and left/right choice. Top, choice/reward-selective neurons. Bottom, uniquely category-selective neurons. c, d′ throughout category learning and the change in task, aligned to criterion (>66% correct, n = 9 mice). d, Example HLS maps before (T5) and after (L/R) the task change. Scale bars, 30 μm. Hue: preferred category; lightness: response amplitude; saturation: category selectivity. e, CTI of all recorded neurons, calculated using either the relevant or the irrelevant rule, before (T5) and after (L/R) the task change. T5: P = 1.3 × 10−161, L/R: P = 1.3 × 10−24, two-tailed WMPSR test (n = 2,389). Grey lines denote CTI = 0.1. f, Top, inferred spike rate for stimuli ordered along the relevant dimension (black), or the irrelevant dimension (blue) across all Go category-selective neurons (defined at T5, CTI > 0.1). PT5 rel = 6.5 × 10−28, PT5 irrel = 1.6 × 10−5, PL/R rel = 1.6 × 10−19, PL/R irrel = 0.38, two-tailed WMPSR test (n = 407). Data are mean ± s.e.m. Bottom, mean activity per stimulus. g, As in f, for NoGo category-selective neurons. PT5 rel = 1.9 × 10−4, PT5 irrel = 0.45, PL/R rel = 7.7 × 10−4, PL/R irrel = 0.21, two-tailed WMPSR (n = 48). h, Predictor weights and response amplitudes of significantly modulated neurons. P < 0.05, at least one predictor, F-statistic (1,904 neurons). Scale bars, 50 neurons. Top row, normalized predictor weights for each neuron. Left, neurons with a negative category weight (category 1, NoGo/GoLeft). Middle, neurons with a positive category weight (category 2, Go/GoRight). Right, no significant category modulation. Middle row, average (normalized) activity in correct, incorrect and missed trials of categories 1 and 2 separately. Bottom row, per group, the mean normalized response to all presented stimuli at T5 (left) and L/R (right).
Source data