Neural coding of speech sound in the telencephalic auditory area of the mynah bird

Summary The responses of neurons in field L in the auditory neostriatum of the mynah bird, Gracula religiosa, were recorded during presentation of intact or manipulated mimic voices. A typical mimic voice konnichiwa elicited responses in most of the neurons. Neurons in the input layer (L2) of field L showed many peaks on peristimulus time histograms while those in other layers (L1 and L3) exhibited only one or two peaks. Several neurons in L1 and L3 responded only to the affricative consonant /t/ in the intact mimic voices. They did not respond to the affricative consonant in the isolated segment or to the one in the playbacked voice in reverse. Forty-five percent of the neurons (33/ 73) decreased in firing rates at the affricative consonant in the isolated segment compared with in the intact voice. Some of these neurons, in which neither the affricative consonant in the isolated segment nor bursts of noise alone elicited responses, exhibited clear phasic responses to /t/ in the case when bursts of noise with particular central frequencies preceded the affricative consonant. The responsiveness of these neurons appears to receive temporal facilitation. These results suggest that these neurons code the temporal relationship of speech sound.Abbreviations HVc hyperstriatum ventrale, pars caudale - TFN temporally facilitated neuron - TSN temporally suppressed neuron