Evidence for neuronal periodicity detection in the auditory system of the Guinea fowl: Implications for pitch analysis in the time domain

SummaryEvidence for periodicity analysis was obtained by recording from 420 single units in the auditory midbrain nucleus (MLD) of awake Guinea fowls (Numida meleagris). The results were compatible with a neuronal correlation model consisting of three main components: an oscillator, an interval multiplier and a coincidence unit. The model makes use of a neuronal time constant in order to measure the periodicities of auditory signals.For 180 units the sequence of spike intervals in response to tone bursts and amplitude modulations (AM) was studied with 10 μs resolution. In 69 of these units (38%) amplitude fluctuations like stimulus onset or the modulation cycles produced periodic spike trains resembling damped oscillations. The periods of these oscillations did not correspond to either the best frequency (BF) of these units or the periodicities of the stimuli. They were interpreted as multiples of a neuronal time constant, τ1=0.4 ms, probably a minimal synaptic delay. These units were tuned to AM-signals with particular combinations of the modulation frequency, fm, and the carrier frequency, fc. The corresponding periods τm and τc were related to the intrinsic oscillation by a periodicity equation: m·τm+n·τc = 1·τ1, where a few small integers for m, n and 1 were adequate to describe all observed properties of a unit.Variation of fm or fc shifted the phase delays of the coupled spike activities proportional to m · τm or n · τc, respectively. These effects were explained by coincidence of neuronal activity phase coupled to fc, with intrinsic oscillations triggered by the fm-cycles. The coincidence condition at the level of the recorded units was given by the periodicity equation. Psychophysical experiments using AM-signals indicated that the described mechanisms, together with the same neuronal time constant, τ1 are adequate to explain pitch perception in humans.