Implicit, Explicit, and Predictive Perceptual Processing in Dyslexia

Abstract

During reading acquisition, neural circuits for auditory and visual perception undergo specialization and are incorporated into an efficient network. However, in many children, dyslexia impairs the development of accurate and fluent reading. The core neurobiological differences that cause this specific reading disability remain unelucidated. Some prominent theories attribute dyslexia to atypical speech-sound representations, while others implicate slow incorporation of otherwise intact representations into explicit processes. Other, more mechanistic theories trace deficits to poor learning of the stimulus regularities that support predictive, efficient perception. We evaluated these hypotheses in four experiments conducted in adults with and without dyslexia, aiming to determine (a) whether implicit processing of faces, print, and speech is altered in dyslexia, (b) whether explicit stimulus-identification processes are compromised in dyslexia, and (c) whether adaptive and predictive perceptual mechanisms are dysfunctional in dyslexia. In Experiment 1, neural decoding of magnetoencephalography (MEG) data revealed how, in typical readers, information about the perceptual category and perceptual ambiguity of speech syllables emerges during passive exposure and active categorization. Experiment 2 identified, in dyslexia, neural speech representations of comparable quality; however, additional processing stages were observed on the way to slower explicit behavioral identification of the syllables. Experiments 1 and 2 suggest that implicit speech processing is preserved in dyslexia, although explicit access to sublexical speech representations requires more neural resources – consistent with the hypothesis of impaired access to otherwise intact representations. Experiments 3 and 4 embedded stimuli in contexts to measure neural responses to predictable and unpredictable stimulation. In Experiment 3, control and dyslexia groups showed equivalent sensitivity to syllable repetition, but subsequent repetitions had a cumulative effect only in controls, suggesting that the implicit, internal model of short-term stimulus consistency is insufficiently plastic in dyslexia. Experiment 4 used electroencephalography (EEG) to contrast implicit repetition effects with top-down effects of violated expectations, finding that perceptual predictions about faces and print were poorly integrated into intact feedforward processing in dyslexia. Experiments 3 and 4 suggest a deficient mechanism for prediction-mediated perceptual learning in dyslexia. Taken together, this dissertation highlights the value of computational, neuroscientific approaches to evaluating causal theories of dyslexia.