Fig. 7: Antagonism through complex mixtures in single neurons.

a Experimental setup for imaging OSN somata in the epithelium. b Example image of OSN somata and selected ROIs from one of three mice used in this dataset. Scale bar in image is 20 µm. c Stimulus design for complex mixtures. OSN responses to each of 16 components are measured. Odor mixtures are made from subsets of 2, 4, 8, or 12 of the 16 individual odor components. d Example mixture responses from the OSNs outlined in (b). Each point is the average of three trials. The data from each OSN is fit with a sigmoid that reflects the maximum response from that cell. Odor tuning profile for each OSN is shown on the right. Scale bar units are ΔF/F. e Deviation from each mixture response to the sigmoidal fit the data. Positive values reflect antagonism and negative values reflect synergy. For each OSN: n = 24 2-part mixtures, n = 20 4-part mixtures, n = 20 8-part mixtures, and n = 20 12-part mixtures. Mean of all mixtures of a given complexity is the horizontal black bar and error is SEM. f Cumulative distribution of all deviations from the sigmoidal fit for each mixture complexity. Data collected from 129 OSNs in three mice. Expanded traces are inset. Right, histogram of bootstrapped P value distributions for each mixture complexity, testing the significance of the observed deviations from the sigmoidal fit against a null distribution. 500 random deviations for each mixture complexity were selected and compared with distributions with a vanishing mean and the same standard deviation using a two-sample Kolmogorov–Smirnov test. This process was repeated 100,000 times. Red horizonal dashed line corresponds to P values = 0.05. g Data were normalized for each OSN by dividing each mixture response by the asymptote of the sigmoidal fit to the data. Antagonism is not overrepresented in a small number of strongly responding cells.