Psychophysics and neurophysiology of temporal factors in hearing by the goldfish: amplitude modulation detection.

Abstract

Goldfish [Carassius auratus] and human psychophysical observers were trained to detect the onset of sinusoidal amplitude modulation impressed on continuous noise or tonal carrier signals (fc) and the just-detectable depth of modulation (m) was measured as a function of modulation frequency (fm) between 5 and 400 Hz. The noise modulation-detection function for goldfish is nearly flat out to 400 Hz at a value of m corresponding to a peak-to-trough intensity difference of about 4 dB. The human function has a low-pass characteristic with a 3-dB down point at about 50 Hz. Temporal integrating mechanisms of the human auditory system do not apparently operate in the same way in goldfish. The tonal modulation function for goldfish has a band pass characteristic with increasing sensitivity up to 200 Hz at about 3 dB/octave, and a steeper decline to 400 Hz. Fish are able to detect a 0.4% modulation (.DELTA.I [intensity difference] = 0.07 dB) at 200 Hz. The human function generally shows less sensitivity than the goldfish and appears to be determined by 2 processes: the detection of amplitude fluctuations at lower values of fm, and the detection of spectral components at higher values of fm. The greater modulation sensitivity of the goldfish may be due to its lack of temporal integration, which is concomitant with an inability to process the acoustic spectrum. The activity of single neurons of the goldfish saccular nerve was analyzed in response to the same amplitude-modulated signals used in the psychophysical studies. The synchronization of spikes to the sinusoidal envelope was measured, and the value of m necessary to produce a given degree of synchronization plotted as a function of fm produced neural modulation-detection functions for comparison with the psychophysical functions. For the tonal carriers, these functions show band-pass characteristics with a best modulation frequency that varies between 20 and 300 Hz in different neurons. The high sensitivity of neurons to amplitude modulated (AM) tones is an example of temporal contour enhancement, the origin of which is not known. Neural modulation functions for noise carriers are flatter and show less sensitivity than for tones. Generally, the best sensitivity of the ensemble of neurons corresponds well with the psychophysical functions, suggesting that behavioral detection is based on a response to particular temporal patterns of neural activity. The correspondence between the behavioral and neural data in goldfish indicates that no major neural transformations intervene between input to the brain and the behavioral response. This is probably not true for humans. For a given neuron, the bandwidth of response to pure-tone signals is from 2 to 7 times the best modulation frequency. The phase angle between the peak of the stimulus envelope and the peak of the modulation period histogram varies with fm in a way that depends on the neuron''s best modulation frequency and pure-tone bandwidth. Pure tone and modulation ''tuning'' are apparently determined by the same mechanisms. Whether modulation tuning is determined mechanically in the periphery or is a cellular property remains to be determined.