Figure-ground analysis relevant to speech in noise perception in health and disease

I will describe work on generic mechanisms for recognition of auditory targets in noisy backgrounds using stochastic stimuli [1]. The work specifies a general system relevant to speech-in-noise (SIN) perception, deficits in which are a ubiquitous symptom of cochlear damage and central auditory disorder.

Non-invasive human MEG [2] and fMRI [3] work specifies a system involving auditory cortex and intraparietal sulcus. Invasive human local field potential recordings from auditory cortex demonstrate high-frequency oscillatory activity corresponding to figure emergence.
Work in a macaque model demonstrates similar behavioural performance in macaques and man and macaque fMRI demonstrates a network that involves similar areas of high level auditory cortex in the macaque, in rostral parabelt [4]. This definition of network organisation in the macaque provides a basis for neurophysiological work in progress to systematically define mechanisms for generic figure-ground analysis at the level of neurons and neuronal ensembles that is not possible in man.

Data from >100 normal-hearing subjects establishes that the detection of generic figure ground tasks correlates significantly with speech in noise perception using single CVCs, based on a modification of the California Consonants Task to incorporate noise. Data from hearing impaired subjects using conventional amplification or cochlear implants demonstrate significant correlations that are much stronger, accounting for more than a quarter of the variance.

The relationship between the generic figure-ground measure has a likely component related to cochlear processing constituting a common source of variance in the generic figure-ground and SIN task performance, especially in the hearing impaired. But the imaging and recording work on normal subjects demonstrates a central mechanism for the figure-ground analysis that is another factor that contributes to SIN performance. Such central mechanisms represent further determinants of SIN listening success in normal and hearing-impaired listeners.

1. Teki S, Chait M, Kumar S, Shamma S, Griffiths TD. Segregation of complex acoustic scenes based on temporal coherence. Elife 2, e00699 (2013).
2. Teki S, Barascud N, Picard S, Payne C, Griffiths TD, Chait M. Neural Correlates of Auditory Figure-Ground Segregation Based on Temporal Coherence. Cereb Cortex 26, 3669-3680 (2016).
3. Teki S, Chait M, Kumar S, von Kriegstein K, Griffiths TD. Brain bases for auditory stimulus-driven figure-ground segregation. J Neurosci 31, 164-171 (2011).
4. Schneider F, Dheerendra P, Balezeau F, Petkov CI, Thiele A, Griffiths TD. SfN (Washington, DC, 2017).