Speech can be learned and perceived on the basis of vision, as demonstrated by prelingually deaf adults who rely on seeing spoken language. Visual speech perception (lipreading/speechreading) is not limited to deaf individuals, as almost everyone demonstrates some visual spoken word recognition;but ability varies widely from individual to individual. A mechanistic account at the neural level is needed to help explain lipreading ability, its individual variation, its potential for plasticity, and its role in the spoen language processing system. Based on extensive perceptual research and initial neural evidence, this development project will test a novel hypothesis that visual spoken word representations are stored in the high-level vision cortical pathway. Neuroimaging, computational, and behavioral methods will be used to test this hypothesis in individuals with normal hearing and vision. To test this hypothesis requires overcoming deficiencies in the resolution of conventional functional magnetic resonance imaging (fMRI) data and achieving control over visible spoken stimuli. We will apply advanced neuroimaging techniques (rapid adaptation fMRI and connectivity analyses) and computational modeling of speech dissimilarity to localize in the cerebral cortex activity in response to visible spoken nonsense syllable and word stimuli. Localizers will be used to identify in individuals an area in high-level visual corte that has previously been shown to be selective for visible speech, an area selective for non-speech face motion, areas selective for visual orthographic word forms and for semantic processing, the lateral occipital complex, the fusiform face area, and the human visual motion area. A longterm goal of this project is to determine whether the visual speech pathway is organized the same way as the auditory speech pathway, with multiple levels of representation from speech features to words. The visual speech pathway might be hierarchically shallow, with visual features that are not specific to speech projecting directly to speech representations in high-level vision cortical areas. This project will test specific hypotheses about the selectivity f localized brain areas for visible spoken words and syllables. It will also investigate adaptation, connectivity, and psychophysiological effects in whole brain analyses focused on relationships among brain areas that might contribute to visual perception of speech. The project involves a highly innovative collaboration between speech science researchers and vision scientists using a multidisciplinary approach that can benefit both areas of research. Detailed understanding of how the complex and dynamic visual speech stimulus is processed in the visual pathway will contribute to vision science. Clinical relevance. If evidence is obtained for high-level visual system involvement in speech recognition, innovative clinical methods for training lipreading in those with hearing loss become available from vision science. Individuals with hearing loss frequently depend on visible speech in face-to-face communication, and enhanced ability to use visual information could improve the quality of their lives. Explanations for individual difference in lipreading and methods to improve lipreading have been sought for over a century. In future studies, we can use the proposed methods to isolate causes for lipreading differences. Studies can test the hypothesis that poor lipreading corresponds to poorly tuned neuronal spoken visual word form representations with excessive adaptation for dissimilar and poorly discriminated stimuli, and good lipreading corresponds to minimal neuronal adaptation beyond a certain level of stimulus dissimilarity.
Individuals with hearing loss must rely to varying extent on visual speech perception, but individuals vary greatly in their ability to derive speech informatio from visual input. This project seeks to understand the perceptual and neural basis for visual speech processing and is a novel collaboration between speech and visual neuroscientists who will apply recently developed neuroimaging methodology for localizing visual speech activations in the cerebral cortex. Using outcomes of the project, new diagnostic and training methods for speech perception (visual and multisensory) based on vision science and cognitive neuroscience will be developed.