Modulation of sound perception by attention, learning and emotion : recent advances on cortical sound coding

Listen to audio

Our interest has focused on the representations of various sound parameters in the A1 primary auditory cortex, and first and foremost on frequency representation (tonotopy). This refers to the spatial organization of neurons in a gradient according to their preferential sensitivity to a given sound frequency. Tonotopy is present in the cochlea and is preserved in the neuronal nuclei of the central auditory pathways. It has been reported in the A1 primary auditory cortex of many mammalian species, including humans, non-human primates and cats. However, in mice, its existence in the A1 area was still under debate in 2010. The absence of tonotopy in mouse area A1 could severely limit the interest of studying cortical sound processing in this species. Between 2010 and 2013, this point was gradually clarified. Work carried out in 2010 (Bandyopadhyay S., Shamma S.A., Kanold P.O., Nat. Neurosci., 2010) using biphotonic calcium imaging in anesthetized mice showed the existence of a coarse tonotopic map in area A1 (a rostro-caudal tonotopy running from high to low frequencies with a gradient of around 3 octaves/mm, and covering the entire A1 cortex). The study was extended to the search for a mapping of neurons based either on the bandwidth of their response, or on their preferential response to a given intensity. For adjacent neurons, a fairly wide heterogeneity of their bandwidths was observed. No gradual spatial organization of neurons evoking "intensity mapping" was observed. On the other hand, a formation of microdomains composed of a few neurons sharing a preferential response to the same intensities was noted. The approach was extended to the search for spatial groupings of neurons, based on the response not to a single parameter but to several. The absence of an authentic cortical tonotopic map in mice, despite a large-scale coarse tonotopic organization, was confirmed by another study (Rothschild G., Nelken I. and Mizrahi A., Nat. Neurosci., 2010). Two studies published in 2011 and 2012, involving electrophysiological analyses in anesthetized mice (Hackett T.A., Barkat T.R., O'Brien B.M., Hensch T.K., Polley D.B., J. Neurosci., 2011; Guo W., Chambers A.R., Darrow K.N., Hancock K.E., Shinn-Cunningham B.G., Polley D.B., Winkowski D.E., Kanold P.O., J. Neurosci., 2012), however, provided evidence in favor of the existence of a tonotopic map in layer L4 of mouse cortical area A1. In 2013, a two-photon calcium imaging study using Fluo-4, strictly focused on the L4 layer of cortical area A1, i.e. the projection layer of thalamo-cortical afferences, (Winkowski D.E., Kanold P.O., J. Neurosci., 2013), provided definitive proof of a tonotopic organization of this cortical layer. Tonotopic organization disappears in layers L2/L3, indicating that a transformation of sensory information takes place during the transition from L4 to L2/L3.

We then turned our attention to recent advances in the neural basis of sound discrimination. Work carried out on the responses of neurons in layers L2/L3 studied by calcium imaging in anesthetized mice (Bathellier B, Ushakova L. and Rumpel S., Neuron, 2012) concluded the existence of a limited repertoire of responses for a given population of neurons. Sound discrimination would lie in the global response of the cortex, which would leverage the combination of distinct modes of response to generate a stimulus-specific global response. This work also revealed the existence of a non-linear mode of operation in the auditory cortex: the responses of neurons in a field that responds to two sounds with distinct modes show that, when responding to a mixture of these two sounds, the response modes are either of one type or another, and are never mixed. All in all, a discriminative representation of sounds emerges in the L2/L3 layer of the auditory cortex only at a global level, in which the functional units are made up of a few hundred neurons. A very small fraction of these functional units account for a large proportion of discrimination for a given pair of sounds.

A notable advance concerning the organization and functioning of neural microcircuits, the fruit of very recently published work, has been discussed (Cossell L., Mrsic-Flogel T.D., Nature, 2015). The strength of synaptic connections between neurons determines how they influence each other's responses. The amplitude of excitatory responses between pairs of cortical neurons varies by two orders of magnitude, but only a small number of synaptic connections are very strong. This heterogeneity of synaptic connection strengths is observed in various cortical areas, but its contribution to information processing in local microcircuits is not known. To address this question, the intensity of the response of neurons in layers L2/L3 of primary visual cortex V1 to various visual stimuli was studied, using two-photon calcium imaging, in anesthetized mice. The imaged region was marked and then, by electrophysiological recordings on cortex slices (patch-clampen whole-cell configuration), the strength of synaptic connections between imaged neurons was measured by the amplitude of their excitatory postsynaptic potentials (EPSPs). Ultra-fast imaging identified pairs of neurons whose responses show a strong temporal correlation. Their distribution reveals the existence of a small population of neuron pairs whose activity is highly correlated. PPSEs with the highest amplitude are associated with these neurons. This result is in line with the hypothesis formulated by Hebb in 1949 (Hebb D.O., 1949), according to which the correlation of neuron activity increases the strength of their synaptic contacts. The results also show that half of the synaptic strength comes from the activity of the 7% of neuron pairs whose response to a stimulus is most strongly correlated. Conversely, weak synaptic connections are those of neurons whose responses are not correlated. Strong, reciprocal synaptic connections often occur between neurons engaged in the same functions. This simplifies the apparent complexity of neural networks. The establishment of strong synaptic connections could amplify the response to specific stimuli in cortical microcircuits. The authors propose that weak synaptic connections, on the other hand, ensure the response plasticity of neural networks. This is an incentive to rethink the way we explore brain microcircuitry.