A statistically based measure of the acoustic reflex and its relation to stimulus loudness.

Abstract

To develop an objective statistic for quantifying acoustic-reflex magnitude, a computer sampled baseline and perstimulatory conductance and susceptance. For each component, the difference between baseline and perstimulatory means was divided by the pooled standard deviation of the samples yielding a statistic similar to d′. Component d′’s were combined by taking the square root of the sum of the squared d′s. The resultant d′ was used to represent acoustic-reflex magnitude in an objective, statistically based format. Stimulus level required to elicit reflexes of specific magnitudes was estimated using the PEST adaptive procedure. Magnitude-intensity functions were determined for a 1-kHz tone, 50- and 300-Hz wide noise bands, and white noise. Stimuli at estimated levels were then used as references in loudness balances among all stimulus pairs, yielding four equal-loudness contours. Equal-loudness and equal-reflex contours had similar shapes, suggesting a relationship between loudness and the acoustic reflex. The magnitude-intensity functions were quite steep, indicating that small intensity changes produced large changes in reflex magnitude. We concluded that the data do not permit rejection of the hypothesis that stimuli eliciting equal reflexes are equally loud. However, the quantification technique has value in that the notion of reflex ’’threshold’’ can be replaced with a mathematically rigorous measure of reflex magnitude. Experiments to refine this technique are in progress.