Acoustic distortion as a measure of frequency selectivity: relation to psychophysical equivalent rectangular bandwidth.

The magnitude of cubic intermodulation distortion generated when two tones are progressively separated in frequency reaches a broad maximum when the distortion frequency falls just over half an octave below the high-frequency stimulus (f2), when this distortion is measured with a microphone in the ear canal. For the component 2f1-f2, this peak occurs at an f2/f1 ratio of approximately 1.2. The tuning, magnitude, and mean group delay of this distortion peak was measured for a fixed f2 of 4 kHz at 40 dB SPL and a varied f1 at 55 dB SPL in eight human subjects with normal hearing. The distortion peak measures were compared with the frequency selectivity at 4 kHz of the same eight subjects derived using a forward-masking notched-noise paradigm. In the six subjects from whom good, repeatable levels of distortion were measured, a significant negative correlation was found between the tuning of the distortion peak and the psychophysical bandwidth at f2. It is concluded that the tuning of the distortion peak may provide an objective measure of frequency selectivity in the human cochlea.