As we age, almost all of us will complain about increased difficulty to communicate in noisy environments. An important issue is why some individuals, even without measurable loss in audibility, experience more difficulties than others.  At present, it is estimated that over 10% of individuals showing significant difficulty for understanding speech-in-noise (SPiN) have actually clinically-normal audiograms. In particular, synaptopathy - the loss of synapses connecting the cochlea to the auditory nerve, caused by aging or noise exposure, is thought to be an important factor contributing to this problem. Yet, recent attempts to assess synaptopathy in humans have produced mixed results. These works were mostly focused on the impairment caused on temporal envelope coding. 

Although previous works showed the pivotal role of TFS for SPiN understanding, it is currently unclear how synaptopathy impairs the neural mechanisms coding for the spectral shape of sounds. The INSPECTSYN project will address this fundamental question by following an integrated multidisciplinary approach combining computational modeling, psychophysics and electrophysiology. In particular, we will examine how FFRs reflect the amount of synaptopathy in an impaired ear and use this proxy to provide an evaluation, through a novel perspective, of the contribution of synaptopathy to SPiN deficits.

Human language acquisition is based on the development of neural processes for selectively extracting and processing acoustic parameters of speech signals, in particular amplitude and frequency modulations (AM-FM). Previous studies show that auditory processing of these acoustic cues is already in place at birth, but remains inefficient until adulthood. In particular, young children show more difficulty than adults in perceiving speech in a noisy environment. Yet, the developmental stages of this speech processing efficiency are still scarce. 

In this new interdisciplinary research project at the interface between biological and computational sciences conducted in collaboration between the INCC and STMS laboratories, we are exploring a new combination of electrophysiological measurements (EEG) and computational models of the auditory periphery and the midbrain to translate the empirical changes observed into physiologically plausible, latent parameters. These results should provide further information regarding the origins of auditory development and associated perceptual changes observed from birth to adulthood. 

Beyond words, speech carries a lot of information about a speaker through its prosodic structure. Humans have developed a remarkable ability to infer others’ states and attitudes from the temporal dynamics of the different dimensions of speech prosody (i.e. pitch, intensity, timbre, rhythm). However, we still do not have a computational understanding of how high-level social or emotional impressions are built from these low-level dimensions. We recently developed a data-driven approach combining voice-processing techniques (using a specifically-designed audio software) and psychophysical reverse-correlation to expose the mental representations or ‘prototypes’ that underlie such inferences in speech. 

 In particular, we are currently running experiments to characterize prosody processing deficits in patients after stroke. One motivating perspective to the deployment of such approach in clinical contexts is to provide further characterization of impairments, hopefully leading in the long term to the development of novel individualized rehabilitation strategies. Yet, further theoretical work on the approach is also needed. For instance, the reverse-correlation approach as considered here relies on the assumption that the algorithm the brain uses to make these inferences from speech prosody could be approximated by template-matching, i.e. a linear comparison of the input stimulus with one single prototype. This is a strong assumption that needs further work to understand in which conditions it can reasonably be applied.