Skip to main content

Research Repository

Advanced Search

A psychoacoustical study of the processing and subsequent detection of frequency-modulation

A psychoacoustical study of the processing and subsequent detection of frequency-modulation Thumbnail


Abstract

Evidence is provided for tuned channels in the human auditory system selectively responsive to frequency-modulation (FM). Thresholds for the detection of small changes in frequency were measured with a 2IFC procedure in two subjects at a number of frequencies. Thresholds were determined before and after exposure to various adapting-stimuli. Large elevations of threshold (often by a factor of three], and in some cases small decreases in threshold (facilitation), were found depending upon the FM parameters of the adapting-stimulus. The degree of elevation (expressed as the adaptation factor equal to the adapted threshold divided by the unadapted threshold) of sinusoidal-FM thresholds was found to be dependent upon the frequency-deviation (±AF) of a sinusoidal- FM adapting-stimulus with the same carrier- (Fc) and modulation-frequency (F ) as the test-stimulus (Chapter 3). A characteristic function was m found consisting of a linear increase in adaptation factor with increasing ±AF up to some value of ±AF (dependent upon Fc) after which a progressive decline in adaptation factor occurred with further increases in ±AF. One possible explanation of the results in terms of the processing of frequency transitions by filters in the carrier-frequency domain was examined. By assuming the existence of rectangular filters with bandwidths equal to the value of ±AF at which peak adaptation occurred, and by introducing a cosinusoidal function to give different weights to different rates-of- change of frequency (df/dt) of the modulating waveform, a good fit to the empirical data was found with adaptation given by the area under the modulating waveform within the filter. A central assumption of the filter model was that the fall in adaptation factor at high values of ±AF was due to the decreasing sweep duration within the filter as ±AF increased beyond the filter bandwidth. This assumption was tested using linear sweeps with a constant df/dt for values of ±AF at and beyond the hypothesised filter bandwidth. The decline in adaptation still occurred even though, with these stimuli, the sweep parameters within the filter remained constant.
One possibility discussed was that the decline in adaptation factor was due to the activation of lateral-inhibitory side-bands by the parts of the frequency transitions outside the filter and a concomitant suppression of filter output. Finally, a good fit to the data was given by Gaussian filters which were regarded as more realistic. The inferred filter bandwidth and their relationship to frequency were similar to critical band- widths (Scharf, 1970] suggesting a common basis for these measures of frequency selectivity. By holding the test Fq of sinusoidal-FM constant and varying the adapting Fc» the selectivity of adaptation effects, tuning- curves, were determined [Chapter 4) and compared to measures of selectivity derived from Chapter 3. Large differences were found, the tuning-curves having markedly narrower bandwidths and a non-Gaussian shape. In Chapter 5 evidence supporting the hypothesis that adaptation is an after-effect of prolonged inhibition between FM-channels tuned to Fq and Fm was found. In the terms of this hypothesis. Fc and Fm tuning-curves represent the range and strength of lateral-inhibitory connections between neighbouring channels. Overall, the data from Chapters 3-5 was consistent with a two-stage model of Fc selectivity in FM processing, the first stage given by the derived Gaussian bandwidths and a higher stage represented by tuning-curves. Both carrier- and modulation-frequency tuning-curves showed evidence of threshold facilitation (Chapters 4 and 5). Thresholds for single linear upward frequency sweeps (up sweeps) were increased by a factor of 2 to 3 following exposure to repetitive (8 Hz) up sweeps but not following exposure to down sweeps or tone bursts) correspondingly, thresholds for down-sweep stimuli were increased only by down sweeps. Sinusoidal FM test stimulus thresholds were elevated by both up-sweeps and down-sweeps and to a lesser extent by tone bursts (Chapter 6). These results suggest the existence in the auditory system of channels specific to upward FM, downward FM, as well as AM or repetition channels. Sawtooth-FM did not reveal directional- specific adaptation effects because of the adapting influence of the rapid frequency transitions between the sweep components of sawtooth waveforms (Chapter 7).
Examination of unadapted thresholds as a function of test-stimulus duration revealed the existence of an integration time of 400 ms for the detection stage of FM processing (Chapter 8).

Publicly Available Date Mar 29, 2024

Files




Downloadable Citations