Rhythm modulates perception and neural tracking of speech in a speech-in-noise task

Humans are able to tune into a single speaker even when multiple people are speaking, a phenomenon known as the “cocktail-party effect” (Cherry, 1953). Neural entrainment to the quasi-rhythmic quality of the speech envelope has been proposed as a potential underlying mechanism (Selective Entrainment Hypothesis). McAuley and his colleagues (2020) have previously tested this hypothesis using a speech-in-noise perception task where sentences from a target and a background speaker were presented simultaneously. Their results showed that rhythm distortion in the target speech stream significantly decreased task performance whereas rhythm distortion in the background speech stream improved it. The current study aims to 1) replicate the previous behavioral results with a more naturalistic set of stimuli and, 2) to examine a more direct neural entrainment measure using Electroencephalography (EEG). Twenty participants listened to 100 stimuli in each of the 3 conditions (Target Rhythm Distorted (TRD), Background Rhythm Distorted (BRD), and No Rhythm Distortion (NRD)). They were instructed to attend to the target sentences and answer questions about them. Behavioral results replicated previous findings that response accuracy was significantly higher in BRD condition than NRD and TRD conditions. Phase-locking value (PLV), a metric indexing neural tracking of speech envelopes, was extracted for both target and background streams. PLVs were significantly higher for target than background streams across all conditions. Further, the difference in PLV between target and background streams was modulated by conditions in a subset of EEG channels, providing neural entrainment evidence that underlie the behavior patterns.