The shapes and levels of psychophysical detection threshold versus frequency functions for sinusoidal electrical stimulation of the deafened cochlea vary from subject to subject. These variations have been shown previously to be correlated with nerve-survival patterns. The shapes of these functions are critical in the design and calibration of analog processors for auditory prostheses. This paper examines two stimulus features that may contribute to the shapes of these functions: phase duration and stimulus duration. Psychophysical experiments were performed with unilaterally deafened and implanted macaque monkeys. Effects of phase duration were studied by measuring thresholds for single symmetric biphasic pulses presented at a rate of 1 per trial. Psychophysical detection thresholds for these pulses decreased, in the region of maximum slope, at a rate averaging -6.2 dB per doubling of phase duration, which is steeper than the rate of decrease in thresholds reported for single auditory-nerve fibers. Slope was weakly correlated with threshold level. Thresholds for long-duration sinusoids were consistently lower and the slopes of the threshold versus phase-duration functions were consistently steeper than those for single pulses. Thresholds for sinusoids and pulse trains decreased as a function of stimulus duration for durations up to at least 300 ms. The rate of decrease as a function of stimulus duration depended on the phase duration of the stimulus, and for long phase-duration signals, it depended on frequency. The rate of decrease was correlated with the absolute-detection-threshold level. We conclude that phase duration contributes significantly to, but does not completely account for, the steep slopes of the threshold versus frequency functions for long-duration sinusoids. Temporal integration is greater for longer phase-duration signals, giving rise to steeper slopes of threshold functions for long-duration sinusoids as compared to those for single pulses in the 1- to 5-ms/phase range. Current neural data and models do not account for the steep slopes of psychophysical threshold functions.

No abstract