Figure 3. Estimated perceived intermittency differs based on trial outcome.

(A) Example photoionization detector (PID) trace (red) and pressure sensor trace (black), and blue lines correspond to inhalation periods. Left: Example estimated perceived odor trace (PID trace sampled during inhalation periods). (B) Average cumulative estimated perceived intermittency (based on estimated perceived odor) across trial time for trials with intermittency values between 0.1 and 0.8 for binary naturalistic (n=1362 trials, left) and square-wave (n=1341 trials, right). (C) Example PID reading (red), sniff trace (black), lick trace (blue) during an example high intermittency trial (top, intermittency = 0.8) and low intermittency trial (bottom, intermittency = 0.3). Gray area indicates 6 s odor stimulus period. Following the stimulus period is the decision period where a water reward is delivered if animals lick for a CS+ (indicated by the water droplet). (D) Time of first lick binned by intermittency using both odor intermittency (blue) and estimated perceived intermittency (black) for binary naturalistic (n=1362 trials, left) and square-wave (n=1341 trials, right). (E) Estimated perceived intermittency vs odor intermittency on hit and miss CS+ trials (n=48 sessions). (F) Right: Square-wave. Accuracy of linear classifier performance in predicting trial identity (CS+ or CS-) for trials of intermittency values between 0.2 and 0.8 (CS+) based on estimated perceived intermittency (gray to black lines). Shuffled control is shown in red. One-sided two-tailed t-test with Bonferroni correction. Left: Binary naturalistic, intermittency values ≥ 0.3, all times are significantly above shuffled control (black bar, p<0001). Intermittency values = 0.2, times ≥4 s are significantly above shuffled control (gray line, p<0001). Right: Square-wave, intermittency values ≥ 0.3, all times are significantly above shuffled control (black bar, p<0001). Intermittency values = 0.2, times ≥3 s are significantly above shuffled control (gray line, p<0001) (n=20 repeats per time bin).

Figure 3—figure supplement 1. Average trial sniff frequency vs odor intermittency on hit and miss trials.

Miss trials are in gray and hit trials are in black. Left: Binary naturalistic. Right: Square-wave (n=48 sessions each, generalized linear model; sniff frequency~trial outcome*intermittency, p>0.05; note: only one session with miss trials for synthetic stimuli at intermittency = 0.5).

Figure 3—figure supplement 2. Pupil dilation and running speed differ between hit and miss trials.

For all graphs miss trials are in gray and hit trials are in black. (A) Heatmap of the % change in pupil diameter from odor onset (time 0 s) for miss and hit trials. (B) Average change in pupil diameter across trial time (odor onset is at time = 0 s and odor offset is at time = 6 s) for hit and miss trials. t-Test at odor offset, p<0.05. (C) Fraction of trials with an increase, decrease, or no change in pupil dilation at the time of odor offset. Miss: increase, 25.7%; decrease, 16.7%; no change, 57.7%. Hit: increase, 34%; decrease, 13.1%; no change, 52.9%. (D) Heatmap of the change in running speed (cm/s) from odor onset (time 0 s) for hit and miss trials. (E) Average change in running speed across trial time (odor onset is at time = 0 s and odor offset is at time = 6 s) for hit and miss trials. t-Test at odor offset, p<0.0001. (F) Histogram of the change in running speed at odor offset for all trials. Two-sample Kolmogorov-Smirnov test, p<0.0001. (G) Cross-correlation of between pupil dilation and instantaneous sniff frequency during odor delivery period. Two-sample t-test at peak, p<0.05. (H) Cross-correlation between running speed and instantaneous sniff frequency during odor delivery period. Two-sample t-test at peak, p<0.05. (I) Cross-correlation between pupil dilation and running speed during odor delivery period. Two-sample t-test at peak, p<0.0001. In all cases a positive lag indicates a delay in the first parameter listed.