Overview
- A peer-reviewed PLOS Biology paper published July 16 found that when listeners shift focus between speakers the brain represents both speech streams for about one to two seconds.
- The team identified a reproducible EEG neural signature that appears during switches and a drop in alpha-band power consistent with a brief rise in cognitive effort.
- Analysis showed engagement with a new speaker begins and completes before disengagement from the previous speaker, producing an asymmetric handoff rather than an instant switch.
- The experiment used EEG on 24 young, normal-hearing adults in a simulated cocktail-party setup, with behavioral checks showing roughly 86 percent accuracy in following the cued speaker.
- Authors report tentative evidence from large-language-model tests that word-by-word predictive context may reset after a switch, but they stress the result is model-dependent and that broader replication and tests with older or hearing-impaired listeners are needed before engineering translation.