Kreitewolf, J. | bioRxiv Science

bioRxiv2018

Perceptual grouping in the cocktail party: contributions of voice-feature continuity

Cocktail parties pose a difficult yet solvable problem for the auditory system. Previous work has shown that the cocktail-party problem is considerably easier when all sounds in the target stream are spoken by the same talker (the voice-continuity benefit). The present study investigated the contributions of two of the most salient voice features -- glottal-pulse rate (GPR) and vocal-tract length (VTL) -- to the voice-continuity benefit. Twenty young, normal-hearing listeners participated in two experiments. On each trial, listeners heard concurrent sequences of spoken digits from three different spatial locations and reported the digits coming from a target location. Critically, across conditions, GPR and VTL either remained constant or varied across target digits. Additionally, across experiments, the target location either remained constant (Experiment 1) or varied (Experiment 2) within a trial. In Experiment 1, listeners benefited from continuity in either voice feature, but VTL continuity was more helpful than GPR continuity. In Experiment 2, spatial discontinuity greatly hindered listeners abilities to exploit continuity in GPR and VTL. The present results suggest that selective attention benefits from continuity in target voice features, and that VTL and GPR play different roles for perceptual grouping and stream segregation in the cocktail party.

neuroscience↗

bioRxiv2018

Irrelevant speech disrupts memory more when spoken by a familiar talker

When listening, familiarity with an attended talkers voice improves speech comprehension. Here, we instead investigated the effect of familiarity with a distracting talker. In an irrelevant-speech task, we assessed listeners working memory for the serial order of spoken digits when a task-irrelevant, distracting sentence was produced by either a familiar or an unfamiliar talker (with rare omissions of the task-irrelevant sentence). We tested two groups of listeners using the same experimental procedure. The first group were undergraduate psychology students (N=66) who had attended an introductory statistics course. Critically, each student had been taught by one of two course instructors, whose voices served as familiar and unfamiliar task-irrelevant talkers. The second group of listeners were family members and friends (N=20) who had known either one of the two talkers for more than ten years. Students, but not family members and friends, made more errors when the task-irrelevant talker was familiar versus unfamiliar. Interestingly, the effect of talker familiarity was not modulated by the presence of task-irrelevant speech: students experienced stronger working-memory disruption by a familiar talker irrespective of whether they heard a task-irrelevant sentence during memory retention or merely expected it. While previous work has shown that familiarity with an attended talker benefits speech comprehension, our findings indicate that familiarity with an ignored talker deteriorates working memory for target speech. The absence of this effect in family members and friends suggests that the degree of familiarity modulates memory disruption.

neuroscience↗