Figure 3.

Active Sampling Strategies Emerge across Task Learning

(A) Diagram to show extraction of inhalation duration from example nasal flow trace.

(B) Example nasal flow traces from one mouse showing emergence of rapid sniffing between early and late trials.

(C) MID for example in (B) calculated for each trial (first 500 ms of stimulus) in purple dots. Blue crosses show corresponding sniff frequency for each trial.

(D) Plot showing how MID changes between early and late trials, for learning (n = 38) and passive (n = 42) mice. Thick red lines show significant reductions in MID (faster sniffing); thick blue lines show significant increases in MID (slower sniffing; p < 0.01, unpaired t tests).

(E) Cumulative histograms of MID change (late-early) for learning and passive mice. Black arrowhead shows point of significant difference (STAR Methods).

(F) Left: example flow traces showing the first inhalation after odor onset for early and late trials. Dotted gray line indicates where flow = 0 for each sniff cycle shown. Right: heatmap showing change in inhalation duration as a function of sniff number since odor onset, sorted by ΔMID.

(G) Example nasal flow traces during CS⁺ presentations for example “high motivation” (left) and “low motivation” mice (right), for an early (top) and late (bottom) trial. “Motivation” here refers to the number of licks during the odor stimulus (“anticipatory” licks). Licks are shown as blue ticks. Droplet indicates when mouse would receive reward. Note that sniff changes only occur for the “high motivation” mouse.

(H) ΔMID (averaged for each cell [n = 18] across CS⁺ and CS⁻ stimuli) as a function of the mean number of anticipatory licks in late trials (calculated from CS⁺ trials only).

(I) ΔMID across learning (averaged for each cell across CS⁺ and CS⁻ stimuli, n = 18) as a function of the reaction time calculated from divergent lick patterns (Figure S2A).