Coincidence detection in the cochlear nucleus : implications for the coding of pitch

Abstract

The spatio-temporal pattern in the auditory nerve (AN), i.e. the temporal pattern of AN fiber activity across the tonotopic axis, provides cues to important features in sounds such as pitch, loudness, and spatial location. These spatio-temporal cues may be extracted by central neurons in the cochlear nucleus (CN) that receive inputs from AN fibers innervating different cochlear regions and are sensitive to their relative timing. One possible mechanism for this extraction is cross-frequency coincidence detection (CD), in which a central neuron converts the degree of cross-frequency coincidence in the AN into a rate response by preferentially firing when its AN inputs across the tonotopic axis discharge in synchrony. We implemented a CD model receiving AN inputs from varying extents of the tonotopic axis, and compared responses of model CD cells with those of single units recorded in the CN of the anesthetized cat. We used Huffman stimuli, which have flat magnitude spectra and a single phase transition, to systematically manipulate the relative timing across AN fibers and to evaluate the sensitivity of model CD cells and CN units to the spatiotemporal pattern of AN discharges. Using a maximum likelihood approach, we found that certain unit types (primary-like-with-notch and some phase lockers) had responses consistent with cross-frequency CD cell. Some of these CN units provide input to neurons in a binaural circuit that process cues for sound localization and are sensitive to interaural level differences. A possible functional role of a cross-frequency CD mechanism in the CN is to increase the dynamic range of these binaural neurons. However, many other CN units had responses more consistent with AN fibers than with CD cells. We hypothesized that CN units resembling cross-frequency CD cells (as determined by their responses to Huffman stimuli) would convert spatio-temporal cues to pitch in the AN into rate cues that are robust with level. We found that, in response to harmonic complex tones, cross-frequency CD cells and some CN units (primary-like-with-notch and choppers) maintained robust rate cues at high levels compared to AN fibers, suggesting that at least some CN neurons extend the dynamic range of rate representations of pitch beyond that found in AN fibers. However, there was no obvious correlation between robust rate cues in individual CN units and similarity to cross-frequency CD cells as determined by responses to Huffman stimuli. It is likely that a model including more realistic inputs, membrane channels, and spiking mechanism, or other mechanisms such as lateral inhibition or spatial and temporal summation over spatially distributed inputs would provide insight into the neural mechanisms that give rise to the robust rate cues observed in some CN units.