Competing speech streams are simultaneously represented in the human cortex during attention switching
Summary
An EEG study demonstrates that when switching attention between two competing speech streams, the brain transiently encodes both streams, with engagement to the new one beginning before disengagement from the old. The study uses Temporal Response Functions to map neural tracking and compares four context models derived from a large language model to probe semantic prediction during switches. The findings provide insight into neural attention switching and illustrate how AI language models can inform understanding of human cognition.