Systematic errors in indirect estimates of basilar membrane travel times

There exist in the literature three attempts to derive basilar membrane travel times from the phase versus frequency characteristics of responses to tones in the auditory nerve [Anderson et al., J. Acoust. Soc. Am. 4., 1131-1139 (1971)], cochlear nucleus [Gibson et al., in Psychophysics and Physiology of Hearing, edited by Evans and Wilson (Academic, New York, 1977), pp. 57-68], and basilar membrane [Robles et al., J. Acoust. Soc. Am. 59, 926-939 (1976)]. It is argued in this paper that these derivations probably have overestimated the actual mechanical travel times. Travel time was originally defined by von Békésy as the latency between the onset of a click stimulus and the onset of basilar membrane vibration. For a linear bandpass system, the frequency-domain equivalent of this latency is the high-frequency asymptotoic slope of the phase lag versus frequency characteristic, which is not generally a linear function. In the neural studies (auditory nerve and cochlear nucleus) it was assumed that the phase versus frequency characteristic was a straight line. Slopes derived under a linear assumption are probably closer to the weighted average group delay (i.e., the center of gravity of the click response) than they are to travel time. In the Mössbauer study of basilar membrane mechanics the latency of the response to clicks was compared with the low-frequency slope of the phase characteristic. The comparison should have been made with the high-frequency slope.