A Visionary Approach to Listening: Determining The Role Of Vision In Auditory Scene Analysis

To recognize and understand the auditory environment, the listener must first separate sounds that arise from different sources and capture each event. This process is known as auditory scene analysis. The aim of this thesis is to investigate whether and how visual information can influence auditory scene analysis. The thesis consists of four chapters. Firstly, I reviewed the literature to give a clear framework about the impact of visual information on the analysis of complex acoustic environments. In chapter II, I examined psychophysically whether temporal coherence between auditory and visual stimuli was sufficient to promote auditory stream segregation in a mixture. I have found that listeners were better able to report brief deviants in an amplitude modulated target stream when a visual stimulus changed in size in a temporally coherent manner than when the visual stream was coherent with the non-target auditory stream. This work demonstrates that temporal coherence between auditory and visual features can influence the way people analyse an auditory scene. In chapter III, the integration of auditory and visual features in auditory cortex was examined by recording neuronal responses in awake and anaesthetised ferret auditory cortex in response to the modified stimuli used in Chapter II. I demonstrated that temporal coherence between auditory and visual stimuli enhances the neural representation of a sound and influences which sound a neuron represents in a sound mixture. Visual stimuli elicited reliable changes in the phase of the local field potential which provides mechanistic insight into this finding. Together these findings provide evidence that early cross modal integration underlies the behavioural effects in chapter II. Finally, in chapter IV, I investigated whether training can influence the ability of listeners to utilize visual cues for auditory stream analysis and showed that this ability improved by training listeners to detect auditory-visual temporal coherence.