Fig. 1: Multisensory task contingencies delay reaction time.

a Schematic of task setup. b Example trial structure with reward availability for each cohort. Three cohorts of mice were presented with the same sensory stimuli: continuous drifting gratings that occasionally changed orientation and direction (visual trial) and a continuous tone that changed frequency content (auditory trial, Supplementary Fig. 1). Cohorts differed in reward structure. Noncontingently exposed (NE) mice were not rewarded contingently to the stimuli. Unisensory trained mice (UST) were rewarded for licks to the left spout after visual trials only, i.e., trained on vision only (cyan blocks denote the reward windows). Multisensory trained mice (MST) were rewarded and trained to lick (for instance) left to report visual changes and right to report auditory changes, i.e., discriminate modality. For NE mice, reward windows were temporally decorrelated from the sensory stimuli, and randomly occurred outside the stimulation period (these windows are denoted as cyan blocks with an asterisk). The trial windows indicate the time window used post-hoc to compare stimulus-related lick rates across cohorts; colors of these windows correspond to the different trial types (blue: visual; red: auditory; gray: catch). For NE mice and auditory trials in UST mice, licks to the visual spout and auditory spout that happened to fall in these windows were defined as surrogate “hits” and “errors” (see Methods). ITI: inter-trial interval. c The upper panels show behavioral response rates (dots) and model fits (lines: solid lines for responses to the correct—rewarded—side, dashed lines for responses to wrong—unrewarded—side) for an example session of a noncontingently exposed (NE) mouse. The bottom panels show the average psychometric fits for each mouse obtained by averaging parameters over sessions. Each session was fit with a two-alternative signal detection model (black lines in upper panels, colored in lower panels). d Same as c, but for UST animals. Note how visual hit rates increase as a function of the amount of visual change, but not auditory change. The relatively high lick rate to the visual spout upon auditory changes arises because only that spout was associated with reward in this task. e Same as c, but for MST animals. Hit rates increased as a function of both visual and auditory change. f D-prime across cohorts. Visual d-prime was comparable for UST and MST (ANOVA, n = 151 sessions, F(1,29) = 1.60, p = 0.22,), and lower than auditory d-prime (ANOVA, n = 139, F(1,261) = 36.26, p = 5.84 × 10⁻⁹). Each dot is the average over sessions for each animal. Error bars denote the median and interquartile ranges. g The detection threshold for visual orientation changes was comparable for UST and MST (ANOVA, n = 151 sessions, F(1,31) = 0.45, p = 0.51). h Reaction time for the same subjectively salient visual stimuli (see Methods) was significantly shorter for UST compared to MST (ANOVA, n = 3917 trials, F(1,3865) = 60.1, p = 1.11 × 10⁻¹⁴). Saliency levels: sub = subthreshold, thr = threshold, sup = suprathreshold, max = maximal change. Boxplot: dot, median; box limits, 25th and 75th quartiles; whiskers, 1 × interquartile range. **p < 0.01, ***p < 0.001.