Attention to speech: Mapping distributed and selective attention systems

Galit Agmon,Paz Har-Shai Yahav,Michal Ben-Shachar,Elana Zion Golumbic
DOI: https://doi.org/10.1101/2021.02.13.431098
2021-02-13
Abstract:Abstract Daily life is full of situations where many people converse at the same time. Under these noisy circumstances, individuals can employ different listening strategies to deal with the abundance of sounds around them. In this fMRI study we investigated how applying two different listening strategies – Selective vs. Distributed attention – affects the pattern of neural activity. Specifically, in a simulated ‘cocktail party’ paradigm, we compared brain activation patterns when listeners attend selectively to only one speaker and ignore all others, versus when they distribute their attention and attempt to follow two or four speakers at the same time. Results indicate that the two attention types activate a highly overlapping, bilateral fronto-temporal-parietal network of functionally connected regions. This network includes auditory association cortex (bilateral STG/STS) and higher-level regions related to speech processing and attention (bilateral IFG/insula, right MFG, left IPS). Within this network, responses in specific areas were modulated by the type of attention required. Specifically, auditory and speech-processing regions exhibited higher activity during Distributed attention, whereas fronto-parietal regions were activated more strongly during Selective attention. This pattern suggests that a common perceptual-attentional network is engaged when dealing with competing speech-inputs, regardless of the specific task at hand. At the same time, local activity within nodes of this network varies when implementing different listening strategies, reflecting the different cognitive demands they impose. These results nicely demonstrate the system’s flexibility to adapt its internal computations to accommodate different task requirements and listener goals. Significance Statement Hearing many people talk simultaneously poses substantial challenges for the human perceptual and cognitive systems. We compared neural activity when listeners applied two different listening strategy to deal with these competing inputs: attending selectively to one speaker vs. distributing attention among all speakers. A network of functionally connected brain regions, involved in auditory processing, language processing and attentional control was activated when applying both attention types. However, activity within this network was modulated by the type of attention required and the number of competing speakers. These results suggest a common ‘attention to speech’ network, providing the computational infrastructure to deal effectively with multi-speaker input, but with sufficient flexibility to implement different prioritization strategies and to adapt to different listener goals.
What problem does this paper attempt to address?