Frequency tuning or spectral filtering has served as a basis for much of our understanding of complex sound processing throughout the auditory pathway. Neural processing along other stimulus dimensions, namely intensity and time, also plays important roles in forming representations of complex acoustic signals. We recently re-examined some of these issues in the primary and lateral fields of the auditory cortex in the awake marmoset, a highly vocal primate. Our data showed that a large majority of neurons (~70-80%) exhibited highly non-monotonic discharge rate versus stimulus intensity functions. For a given neuron, this nonlinear response property applies to both tones, narrowband and broadband stimuli (e.g., bandpass noises, vocalizations). Many neurons we studied only responded within a narrow range of sound intensities (as narrow as 10-20 dB). The optimal sound levels that these neurons were tuned to were approximately evenly distributed across the entire range of sound intensities tested. The suppression of discharges at sound levels higher than the tuned intensity appeared to be caused by neural inhibitions. An important implication of this intensity tuning is that it further divides the task of processing a complex sound among cortical neurons with similar frequency tuning characteristics.
When tested by temporally modulated signals, most cortical neurons in awake marmosets exhibited bandpass modulation transfer functions based on discharge rate. The best modulation frequency, at which a neuron gave maximum discharge rate, was largely distributed in the range of 8-64 Hz in our samples (centered at 25-30 Hz). Only a small percentage of neuron had best modulation frequencies higher than 64 Hz. Phase-locked discharges were low-pass in nature (regarding modulation frequency), limited to modulation frequencies below ~30 Hz and were generally not as clear an indicator of changing modulation frequency as measures of discharge rate. Interestingly, many neurons had nearly identical best modulation frequencies regardless a stimulus was modulated in amplitude or frequency, or regardless modulations were applied to tones, narrowband or broadband carriers. These findings suggest an intrinsic temporal integration mechanism for an auditory cortical neuron that is applied to all of its time-varying inputs, and that the temporal integration window is in the order of 30-40 msec in the cortical fieldsA we studied.
(This work has been supported by NIH-NIDCD Grant R01-DC03180 and Whitaker Foundation Grant RG-96-0268).