Figure 3.

Mice maintain natural sampling behavior in stop-signal task. a, Example raw lick data during sucrose training and correct no-stop trials in a well trained animal. Three randomly chosen trials for each condition are overlaid and aligned by the first lick to demonstrate the low variability in licking in both sucrose training and no-stop trial types. Definitions of LD and ILI are shown. b, c, Distributions of LDs and ILIs during sucrose training (gray) and correct no-stop trials (black). Largely overlapping distributions indicate that sampling behavior was left largely unaffected by the stop-signal task. Mean ± SD for LD and ILI for each condition are shown. Jitter, defined as the SD, generally decreased for LDs and ILIs after training in the stop-signal task, indicating that licking became slightly more precise with training. d, Raw lick data for a correct-stop trial demonstrating reaction-time measurement. Based on the distribution of ILIs from correct no-stop trials in c, the next absent lick on a correct-stop trial can be accurately estimated at an average of 62 ms from the end of the last recording lick (62 ms mean, solid gray line; dashed gray box indicates 95% confidence interval). This time interval was added to all reaction-time measurements on correct-stop trials. e, Example of a stop-signal task session in a well trained animal demonstrating how change in the animal's motivation is controlled for in reaction-time measurements. Only trials that occur when the animal is motivated to drink, defined by maintaining ≥80% correct no-stop trial performance in rolling 10 trial windows (outlined in black box), are used for data analysis. This ensures that the animal's performance during a correct-stop trial represents a decision and not simply unmotivated licking.