Fig.2: BM response function derived from the FDMCs in Fig.1.
Fig.1: Example of FDMCs measured without cue signal.
Context: It is well known that the basilar membrane (BM) response to sounds is highly nonlinear. Specifically, the function relating BM velocity (= output, in dB) and sound input level (in dB) is nonlinear. At low to mid (i.e. 0--40 dB) and at high sound pressure levels (> 80 dB) the response growth is linear, that is a 10-dB increase in the input will produce a 10-dB increase in the output. At mid to high levels (40--80 dB), however, the response growth is very compressive. In that case a 10-dB increase in the input will produce a less than 10-dB increase in the output (see figure below/aside). A compression ratio of about 5:1 has been usually reported in humans. Moreover, the compression ratio at one place on the BM depends on the relative distance between the place of stimulation and the place of observation, i.e. the position along the BM at which the response is being measured. Because of the frequency-to-place transformation that occurs on the BM (which is also referred to as the tonotopic organization), this place dependence transfers to a frequency dependence. Precisely the compression ratio decreases as the relative frequency separation increases. Lastly, there is recent evidence for a time dependence; the compression ratio at one place may evolve over the course of stimulation. A potential reason for that is the presence of a feedback loop in the auditory system, the medial olivocochlear reflex (MOCR), that controls the cochlear gain via efferent connections to the outer hair cells on the BM. The MOCR has a delay of about 20–25 ms between the onset of the stimulation and the onset of the gain reduction. This gain reduction may be accompanied by a reduction in compression ratio, although behavioral data are inconclusive. The first goal of the present topic is thus to collect additional data in humans in order to verify the hypothesis of MOCR-induced reduction in compression.
In animals, direct measurements of BM responses can be performed using laser velocimetry. In this case the place of observation corresponds to the position on the BM where the laser beam points to. In humans, indirect methods are required. Behavioral techniques have thus been developed, among which the growth of masking (GOM) functions and the temporal masking curves (TMCs) are widely used. As their names indicate, these methods use a masking paradigm to infer the BM response. A forward masking paradigm is typically used (i.e. the masker precedes the target in time). In this case the place of observation corresponds to the frequency of the target. (For more details on these indirect methods and their underlying assumptions see e.g. Oxenham & Bacon, 2003; Ear and Hearing vol.24 pp.352-366 and references therein.)
For some reasons that are not explained here, the GOM and TMC techniques are not adequate to assess the effect of MOCR on the BM response. A new technique has been recently introduced in this regard, namely the fixed-duration masking curves (FDMCs). For a description of the method see Yasin et al., 2013; J. Acoust. Soc. Am. vol.133 pp.4145-4155. The FDMC technique in its "basic" form (i.e. using a total duration of masker + target shorter than or equal to 25 ms) provides an MOCR-free estimation of the BM response. Measuring FDMCs using a "precursor", that is a signal preceding the masker in time and triggering the MOCR, thus provides a means to assess the effect of MOCR on the BM response. Still, a pilot experiment conducted at the Acoustics Research Institute and involving measurements of FDMCs with a precursor revealed large within-listener variability. This variability might be attributable to confusion effects, that is the difficulty of distinguishing the target from the masker in presence of the precursor. The second goal of the present topic is thus to refine the FDMC method in order to obtain stable and reliable estimates of BM responses in humans.
Method: In a first attempt to reduce possible confusion effects in FDMC measurements with a precursor, a contra-lateral broadband cue signal, gated simultaneously with the precursor and masker, was added. In a second attempt, the cue signal was removed and a moderately long training phase (2-5 hours) was given before actual data collection.
Results: In contrast to the literature, the presence of the cue did not reduce within-listener variability. The measurements of FDMCs without cue signal revealed that a rather long training phase (> 5 hours) is needed to improve stability of the FDMC method. The BM responses measured with and without precursor are being analyzed.
My contribution: Principal investigator.
Potential applications: BM compression plays a major role in sound perception. In particular, BM nonlinearity is a crucial component in auditory models to account for simultaneous and temporal masking data. Thus, obtaining a time-dependent estimate of BM compression is of high interest to masking models and auditory models in general. It is also relevant for hearing aids and cochlear implants.
Related contribution:
T. Necciari and B. Laback. Effect of cueing on stability of behavioral measurements of basilar membrane responses with a precursor. Presented at the 169th Meeting of the ASA, May 2015.
Measurements of basilar membrane responses in humans
Illustrations
Post-Doctoral Researcher in Audio Signal Processing & Psychoacoustics
Thibaud Necciari, PhD
Thibaud Necciari's personal Website, 2016-2018
orcid.org/0000-0003-2574-0948 | ResearcherID