(A) Peak heights in trial-averaged ΔF/F traces are greater for cells from trace conditioned mice than controls. ΔF/F traces from tone onset to air-puff onset were averaged across all trials after learning and peak heights in these traces were determined and plotted for trace (blue bar), pseudo-conditioned (green bar) and spontaneous activity data (red bar) respectively (* indicates p<0.01). (B) Distribution of reliability scores for cells pooled from all mice that learned the trace conditioning task. (C) Reliability score depends on the number of trials averaged or the bin-width. In Figure 3D,F of the main text, reliability scores were calculated by averaging different numbers of trials (trial bin widths of 24 [on average] and 5 respectively). In this figure, we plotted the cell-averaged reliability score calculated using trial a range of trial bin-widths. The point highlighted with a blue circle indicates the bin-width used for Figure 3D and the red-highlighting indicates the bin-width used for Figure 3E. In Figure 3D,E of the main text, reliability scores were calculated by averaging different numbers of trials (trial bin widths of 24 [on average] and 5 respectively). The size of the highest peak in the averaged trace from randomly time-shuffled trials depends on the number of time-shuffled trials averaged. Hence, the reliability score depends on the trial-bin width used to calculate it. As seen in this panel, the highest cell-averaged reliability scores for trial bin-widths of 5 and 25 closely match the reliability scores seen in Figure 3D,F respectively. (D) Reliability scores for neuronal responsiveness at locally determined peak times do not change with learning. This plot shows the mean (solid line) and standard error (shaded areas) of the reliability score for neuron activity at a peak-timing determined locally, within each five-trial window. The blue, green and red curves are for neurons from trace learners, pseudo-conditioned mice and spontaneous activity respectively. (E) Activity peak timings progressively change towards the final peak timings. We computed the local activity peak timing within 10-trial windows over the entire session. Plotted here is the mean, absolute difference between these local peak timings and the final activity peak timings for cells from trace learners (blue), pseudo-conditioned mice (green), spontaneous activity (red) and non-learners (cyan). The solid curves depict mean values and the shaded areas, SEM. Final activity peak timings were computed using trials after the learning trial. (F) Activity peak timings early in the session are different from final, learning-related peak timings. Activity peak timings were calculated for the early half of the session by averaging activity traces for trials 1 to 25. The final peak timing for each cell was determined by averaging every alternate trial from trials 26 to 50. These two sets of peak-timings were then compared with the final, post-learning peak timings determined by averaging a non-overlapping set of trials from the 26–50 trial window. The absolute differences in peak timing have been plotted for trace learners (blue), pseudo-conditioned mice (green), spontaneous activity (red) and non-learners (cyan). For trace learners, the mean timing-difference is significantly higher for peak timings determined early in the session (solid colored bars) than for those determined in the latter half (bars with black hatch pattern). Peak timing difference does not decrease for all controls. The red line indicates the average frame time across datasets (* indicates p<0.01). (G) Cell-groups sharing peak activity timing are randomly distributed in the field of view. Neuron masks from an example field of view in a trace learner mouse, color-coded as per the timing (frame number) of peak calcium fluorescence within the tone to puff period. Scale bar represents 50 microns. (H) Average, pair-wise distances between neurons sharing the same peak timing (left) and random neuron pairs (right) are plotted, with error-bars marking SEM.