Evidence for multiple DPOAE components based upon group delay of the 2f(1)-f(2) distortion in the gerbil

The cochlear delay of the 2f(1)-f(2) distortion product otoacoustic emissio n (DPOAE) was measured using the phase gradient method. With a constant f(2 ) and swept f(1), the resulting phase change of 2f(1)-f(2) was used to calc ulate the group delay for f(2) frequencies from 1 to 60 kHz. For f(2) frequ encies between 2 and 60 kHz, the group delays were between 2.2 and 0.11 ms and continuously decreased for increasing f(2) and For increasing primary s timulus levels. For f(2) frequencies below 2 kHz, the group delay decreased to around 1 ms and was largely independent of stimulus level. The ratio cu rves resulting from the f(1) sweeps for high frequencies (f(2) > 16 kHz) di splayed the typical mammalian shape with a peak in the level of 2f(1)-f(2) for a larger primary frequency separation (f(2)/f(1) > 1.15) and decreasing 2f(1) - f(2) level for smaller primary separation. In addition to this typ ical level maximum, for f(2) frequencies from about 1.8 to 16 kHz, the rati o curves displayed a second component in the form of an increase in the lev el of 2f(1)-f(2) for small primary separation at higher primary levels (lev el of f(2) > 30 dB SPL). For f(2) frequencies below 1.8 kHz, only the secon d component and no typical ratio peak as for higher f(2) could be observed and the associated group delay was always close to 0.8 ms. Several possible causes for this behavior are discussed, including different modes of DPOAE generation and modulation as well as changes in the nature of mechanical p rocessing from base to apex in the gerbil cochlea. To evaluate the relative sensitivity of non-linear cochlear mechanics, an iso-distortion threshold curve was constructed from acoustical growth functions of the 2f(1)-f(2) DP OAE at optimum primary separation, by plotting the level of f(2) sufficient to evoke a distortion of -10 dB SPL as a function of f(2). This distortion audiogram resembled the neuronal and behavioral audiogram for frequencies > 2.5 kHz but failed to reflect the sensitivity for lower frequencies. This may be a consequence of more linear frequency processing in the apex. (C) 2000 Elsevier Science B.V. All rights reserved.