Fig. 1: Measuring behavioral correlates throughout task learning.

a Schematic of an abstract sensory learning task. For home cage task training, animals licked the left port l, or right port (R) for reward for non-match or match stimulus conditions, respectively. For head-fixed task training (2P), non-match stimulus conditions were rewarded (Yes) while match conditions were not (No). During head-fixed task training, animals were primarily trained on directions with fast speeds (95% across T1–T4, 75% for T5) with a smaller fraction of slow speeds trials provided as unexpected stimuli (5% across T1–T4, 25% for T5). b Schematic of two-photon imaging of Prh using chronically implanted microprisms allowed during head-fixed task training. c Learning curves for individual head-fixed animals trained during two-photon imaging. Only imaged animals reaching T5 were analyzed. d High-speed videography was used to measure whisker kinematics during task behavior. e Whisking amplitude during each trial period across training stages. f, g Change in whisker angle (f) and curvature (g) during sample and test stimulus periods across training stages sorted by speed and direction. h Licking rate during each trial period across training stages sorted by choice. Scale bar = 2 mm. Error bars = SEM. n = 90 sessions from 7 animals for (e–h).