This analysis demonstrated that, despite the large number of theoretical response modes that groups of several tens of neurons could generate, local auditory cortex populations generate only a small repertoire of functionally distinct response modes. Interestingly, a similar result was obtained when two second long sounds were presented ( Figure S4). We then sought to determine the spatial organization of the neurons that underlie distinct response modes. We calculated the mean firing rate of neurons in response to the groups of sounds

associated to the different modes, which were identified in the above analysis. Interestingly, pairs of response modes observed in a given population www.selleckchem.com/products/Docetaxel(Taxotere).html corresponded to the firing of partially overlapping subgroups of neurons (Figures

4A and 4D). To assess the similarity of tuning of neurons associated to the same or different subgroups, we computed their signal correlations. We found that members of the same subgroup had significantly higher signal correlations than neuron pairs across groups (same mode: 0.76 ± 0.07, n = 37; different modes: 0.53 ± 0.11, n = 23 modes, Wilcoxon test p = 2 × 10−4). Furthermore, the centroids of the 3-MA neuronal subgroups corresponding to two distinct response modes were significantly more distant to each other than when the neurons of the local population are spatially randomized (Figure 4E). This indicated an organization of the modes into different spatial domains, which was also visually evident in many examples (Figures 4A and 3C). This observation is consistent with previous estimations of the spatial layout of neurons suggesting a patchy organization of neuronal subgroups in the cortex (Rothschild et al., 2010). The low number of observed response modes suggests that local activity patterns form discrete representations of sounds. A prediction from this scenario would be that for a continuous transition between two stimuli exciting two modes an abrupt change in response patterns would be observed because the population could generate no intermediate response pattern. Alternatively, the low number of response modes could merely

reflect biases or gaps in the set of out tested sounds. To determine if abrupt changes in response patterns could be observed, we first identified local populations in anaesthetized mice showing at least two response modes using a broad set of different sounds (Figure 5A). We selected two basis sounds that were falling in either response mode and constructed linear mixtures from them. Next, we retested the same population with the new set of stimuli to map the transition across modes with higher resolution. When the mixture ratio was varied continuously, we observed abrupt transitions in the population activity patterns that are visible in both the raw activity plots and the similarity matrices (Figures 5B, 5C, and S5).