You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

Poster C71, Thursday, November 9, 10:00 – 11:15 am, Harborview and Loch Raven Ballrooms

Neural Correlates of Atypical Categorical Perception in Dyslexia

Sara Beach1,2, Tracy M. Centanni2, Ola Ozernov-Palchik2,3, Sidney C. May2, Dimitrios Pantazis2, Tyler K. Perrachione4, John D. E. Gabrieli2;1Harvard University, 2Massachusetts Institute of Technology, 3Tufts University, 4Boston University

Many studies find that individuals with dyslexia perceive speech sounds less categorically than typically-developing individuals do, which likely impedes mapping speech sounds to print during reading acquisition. This deficit manifests in less consistent labeling of individual sounds, a shallower slope of the identification function over a sound continuum, weaker discrimination of two sounds that lie across a phonemic boundary, and better discrimination of sounds that lie within a phonemic category. The reason for these behavioral phenomena has not been conclusively identified. At least a subset of dyslexics seems to have auditory processing deficits, particularly for rapid spectrotemporal features that cue consonant identity, and there is evidence that neural hyperexcitability contributes to more variable sensory encoding. In parallel, dyslexics fail to benefit from regularities in acoustic signals, leading to a deficit in stimulus-specific adaptation, which may diminish the ability to abstract both short- and long-term representations. In this study, we explored how these brain-based theories account for categorical perception deficits in dyslexia. Our first aim was to determine whether neural mismatch responses pattern categorically, such that between-category mismatches are greater than within-category mismatches, or whether mismatch responses are graded according to the acoustic distance between the deviant and the standard. This would reveal the degree to which subphonemic information is preserved in automatic speech processing, and whether this differs in dyslexia. Our second aim was to spatially and temporally characterize neurophysiological phoneme adaptation by testing whether classifier error is reduced for subsequent stimulus repetitions, and whether this effect differs in dyslexia – exploring for the first time how neural adaptation affects the coherence of neural representations. Adults with and without dyslexia completed two tasks while undergoing magnetoencephalography (MEG). First, participants labeled 40 instances each of ten continuum tokens, in random order, as either ‘ba’ or ‘da’. From their responses, we derived each individual’s identification function and selected from it five equidistant tokens representing two endpoints, two within-category tokens, and an ambiguous midpoint. In the second task, participants were exposed to these tokens in a passive roving-oddball paradigm. Trains of four to six token repetitions were presented with a 250-ms interstimulus interval, for a total of 3000 items, while participants watched a silent movie. We conducted univariate analyses of the magnitudes of the neural mismatch response to the deviant syllable and of neural adaptation within a syllable train, as the deviant (the first token) gradually became a standard. We also trained a machine learning classifier to distinguish token identity and tested it on held-out portions of the MEG data. Using exemplars selected to best represent each individual’s perceptual space, preliminary data show how categorical representation of speech is encoded in the cortical response patterns of adults with and without dyslexia, and include specific examination of the M100 component, at whose latency the persistence of subphonemic information has been unclear. MEG data reveal neurophysiological adaptation to these repeated exemplars and varied decodability of each exemplar with respect to its position on the continuum and its repetition history.

Topic Area: Perception: Speech Perception and Audiovisual Integration

Back to Poster Schedule