Physiological and psychoacoustical estimation of human cochlear input/output curves

Abstract

[EN] Human cochlear input/output (I/O) curves are not completely understood because they can be obtained using indirect methods only. The temporal masking curve (TMC) method is a favoured psychophysical method to infer human cochlear I/O curves, but is inconvenient for clinical applications. Distortion Product Otoacoustic Emissions (DPOAE) I/O curves share many of the characteristics of cochlear I/O curves and could be a useful alternative to the TMC method, but its generation mechanisms are not completely known, and it is uncertain how DPOAE I/O curves relates to the cochlear I/O curve. The aim of the present work is to test if the two methods can be used indistinctly to infer human cochlear I/O curves in normal hearing listeners. The approach is to compare individual estimates of cochlear I/O curves inferred with the two methods. If the results are consistent, this would provide support for the assumptions of both methods. The results showed reasonably good correspondence between I/O curves of the two methods for frequencies above ~2 kHz but not for lower frequencies. At lower frequencies, the DPOAE I/O curves frequently presented plateaus and notches, which were not present in the I/O curves inferred from TMCs. The DPOAE I/Os were measured using the group average stimulation level parameters (primary level rule) of Kummer et al. (1998). Simulations are presented that aimed at testing if individual differences from the DPOAE group average primary level rule could explain the plateaus and notches in the DPOAE I/O curves at lower frequencies. The results suggest that the primary level rule that maximizes the DPOAE level is also a good primary level rule to estimate the I/O curve of the underlying non-linearity and that even a small deviation from this rule may lead to notches in the DP I/O curve. It is a common hypothesis that maximum DPOAE levels occur when the basilar membrane excitation by the two primary tones is equal at the cochlear site tuned to the higher of the two primary frequencies. A novel TMC-based approach is also presented here designed to test this hypothesis. The results support this hypothesis as the levels required for equal excitation inferred from the TMC method coincide with the empirically found DPOAE primary levels that produce maximum response. Cochlear I/O curves as inferred from TMCs were finally compared to estimates obtained from DPOAEs I/O using two additional primary level rules. The first rule consisted in individualized primary levels inferred from TMCs that obtained equal cochlear excitation and the second used empirically found primary levels optimized individually for maximum DPOAE level. The results showed that the correspondence between I/O curves inferred from TMCs and DPOAEs remained high at frequencies above ~2 kHz but did not improve at lower frequencies, independently of the primary level rule used. Reasons for the poorer correspondence between I/O curves inferred from TMCs and DPOAEs at lower frequencies and particularly for the plateaus and notches in the DPOAE I/Os are discussed. Directions of future research are suggested.