Fig. 7: Combinatorial coding of odors in the PN activity.

a Fractions of PNs (top) and LNs (bottom) activated by each odor. The number above each bar indicates the number of cells tested with that odor. b The fractions of PNs and LNs activated by odors were correlated (Pearson Correlation R = 0.69, P = 0.002, n = 17 common odors). c Trajectories of PN responses to six odors projected on a 3-dimensional principal components (PC) space. The trajectories remained near origin before odor delivery (2 s timepoint) and began to diverge between 2 and 3 s timepoint, and gradually returned to origin (numbers on the plot indicate some timepoints). d Accuracy of odor classification based on PN population responses as a function of the response duration used. Each dot represents the average accuracy for test odors in a given trial. The accuracy for all durations exceeded the chance level (dashed line at 1/6). Black lines: means, error bars: s.e.m. e The classification accuracy improves with an increase in the number of PNs taken for the analysis. Each dot represents a different random sampling of the given number of PNs. Black lines: means, error bars: s.e.m. f Correlation between mean classification accuracy (calculated in (e)) and number of PNs taken (Pearson Correlation R = 0.93, P = 1.5 × 10⁻⁰⁵, n = 12). Source data for (a, b, d–f) are provided as a Source Data file.