Name:Kevin Yang

School/Affiliation:McMaster University

Co-Authors:Ellia Baines, Shu Sakamoto, Laurel Trainor

Virtual or In-person:In-person

Short Bio:

Kevin is a 4th year undergraduate thesis student at McMaster University working in the Auditory Development Lab.

Abstract:

Natural environments contain mixtures of sounds from multiple sources. The auditory system must organize these inputs into perceptually distinct streams so that individual sources can be represented and tracked. Although attention is known to affect stream formation and selection, differences between music and speech remain unclear.

In this study, participants were assigned to one of two groups. The speech group listened to two concurrent talkers. The music group listened to two concurrent piano parts from duet recordings. Stimuli within a group were matched for duration and overall level. On each trial, a cue instructed either integration (attend to both streams) or segregation (attend to one designated stream). Conditions were presented in counterbalanced blocks. To maintain engagement, short behavioral probes asked about the currently attended stream. electroencephalogram (EEG) was recorded continuously during listening.

We used temporal response function decoding to compare two simple models of stream formation: a Separate model that assumes early, distinct stream representations, and a Combined model that assumes encoding of the acoustic mixture. Time-resolved decoding examined early processing (50–100 ms) and later processing (200–500 ms). We asked whether music shows earlier distinct stream formation than speech, and how attention (integration vs. segregation) changes the balance between Independent and Whole-Scene representations.

The poster will report a mixed effect model of group and model comparisons across these latency ranges. The goal is to clarify how stimulus type, selective attention, and timing relate to the neural mechanisms that support auditory stream formation in complex, overlapping scenes.