Auditory cortical onset responses revisited. II. Response strength.

Most neurons of the auditory pathway discharge spikes locked to the onset of an acoustic stimulus, but it is largely unknown in which way the acoustic parameters of sound onsets shape the neuronal responses. In this paper is analyzed the number of spikes discharged by single neurons in primary auditory cortex of barbiturate-anesthetized cats to the onsets of tones of characteristic frequency. The time course of the peak pressure (i.e., the envelope) was altered by parametrically varying sound pressure level (SPL), rise time, and rise function (linear or cosine-squared). For both rise functions, rise time had manifold, and in some cases dramatic, effects on conventional spike count-level functions. In general, threshold SPL, dynamic range, and the lowest SPL at which monotonic spike count functions saturated increased with prolongation of the rise time. In neurons with mostly nonmonotonic spike count-level functions, "best SPL" increased and the descending high-SPL arms flattened, so that functions obtained with long rise times were often monotonic whereas those obtained with shorter rise times were highly nonmonotonic. Consequently, the "tuning" to SPL was less sharp for longer rise time tones, and spike count versus rise time functions changed from "short-pass" to "long-pass" with an increase in SPL. Systematic effects of rise time persisted when spike counts were plotted against the rate of change of peak pressure or against the maximum acceleration of peak pressure. However, when spike counts were plotted as a function of the instantaneous peak pressure at the time of response initiation, the functions obtained with different rise times, and even with different rise functions, were in close register. This suggests that the stimulus-dependent component of first-spike latency can be viewed as an integration window, during which rate of change of peak pressure is integrated. The window commences with tone onset and its duration is inversely related to the maximum acceleration (or, for linear rise functions, the rate of change) of peak pressure and the neuron's transient sensitivity. The present findings seriously question, for onset responses, the usefulness of the spike count-level function and measures derived from it, such as threshold SPL, dynamic range, best SPL, or degree of nonmonotonicity. They further cast doubt onto the validity of current concepts of intensity coding at cortical levels, because most neurons' onset responses are not indicative of a signal's steady-state SPL. However, they suggest a mechanism by which a neuronal population will sample a given transient in an orderly, sensitivity-dependent, temporal sequence. The sampling rate is automatically adjusted to, and adjusted by, the rapidity of the signal's change. And the instantaneous properties of the transient could be represented by the ratios and spatial distribution of responses across the simultaneously active subpopulation. Such a mechanism could provide the basis for the demonstrated capability of discrimination of rapid transients.