Search

International audience; Cochlear synaptopathy (or the loss of primary auditory synapses) remains a subclinical condition of uncertain prevalence. Here, we investigate whether it affects humans and whether it contributes to deficits in suprathreshold speech-in-noise intelligibility or auditory temporal processing ability. For 94 normal-hearing listeners (aged 12-68 years; 64 women), we measured click-evoked auditory brainstem responses (ABRs), self-reported lifetime noise exposure, and speech reception thresholds (SRTs) for sentences and words in stationary and fluctuating maskers. Temporal processing ability was assessed by measuring the slope of threshold-duration functions (TDSs), gap detection thresholds (GDTs) and frequency modulation detection thresholds (FMDTs). TDSs were inferred from absolute detection thresholds for brief (2-20 ms) pure tones and were corrected for the effects of long tone (500 ms) thresholds. GDTs were measured as the shortest detectable silent gap between two marker tones (2 kHz, 80 dB SPL). FMDTs were measured for a carrier tone at 1.5 kHz modulated at 2 Hz. Based on animal research, we assumed that the shallower the growth of ABR wave-I amplitude-level function (in units of μV/dB), the higher the risk of suffering from synaptopathy. We found shallower wave-I growth with increasing age but not with increasing noise exposure. SRTs in noise, TDSs and FMDT were not correlated with wave-I growth rates nor were GDTs when adjusted for the effect of absolute threshold. TDSs, GDTs and FMTDs were not correlated with noise exposure. Altogether, the data are consistent with age-related but not noise-induced synaptopathy. In addition, the data dispute the notion that synaptopathy impairs the intelligibility of supra-threshold speech in noise or the detection of gaps, frequency modulation and brief sounds. [Work supported by the Oticon Foundation (ref. 15-3571), Junta de Castilla y Leon (SAP023P17), MINECO (BFU2015-65376-P), and the European Regional Development Fund].

International audience; Cochlear implants (CI) aim to restore hearing to patients suffering from profound to severe hearing-loss by directly stimulating the auditory nerve with electrical pulses. Due to hardware, surgical, or physiological reasons, the encoding and the transmission of acoustic cues are not optimal, leading to important behavioral gaps between CI and normal hearing listeners (especially for speech in noise recognition). To improve CI-user performance, efforts made in developing new coding strategies can be assisted by computational models which could assess how much information is transmitted to the auditory nerve by a given coding strategy, instead of testing CI users on psychophysical tasks which can be time-demanding and expensive. In the present study, we propose a computational model to objectively determine the amount of information transmitted by cochlear implants. The model is a combination of an electrode-nerve interface model with a phenomenological model of the auditory nerve in response to electrical stimulation. It accounts for the effects of electrical pulse shapes, spread of excitation, neural survival, and neural response properties (such as refractoriness, facilitation, or latency) on information transmission. From the acoustic signal, the model computes the electrical pulses delivered to each electrode, the current received by each auditory nerve fiber and finally, the timings of the evoked spikes. Based on stimulus reconstruction and information theory, an objective metric is proposed to quantify the amount of information remaining in the resulting spike trains across all the fibers along the cochlea. This framework would allow to better identify the sources of information loss (e.g. is more information lost due to deafferentation or by current spread?), and to guide the development of new coding strategies. Also, the model parameters could be individualized in order to make listener-specific predictions and thus facilitate choosing the coding strategy for each patient that maximizes information transmission.