Brain Routes for Reading in ASD and Neurotypicals

Background:  Reading utilises two neural pathways: the first (lexical) route visually mapping whole words to their lexical entry to retrieve meaning, the second (nonlexical) mechanically decoding words via general grapheme-phoneme conversion rules of the given language in order to construct its pronunciation. Whilst neurotypical readers typically employ the direct lexical route for reading familiar regular words, a pattern of poor reading comprehension plus precocity at mechanically ‘sounding out’ words (‘hyperlexia’) suggests the same may not be true of readers with autism spectrum disorders (ASD), who may utilise different pathways and who often seem to lack a ‘default’ semantic processing mode.

Objectives:  We attempted to compare visual word processing of short, simple words between participants with an ASD and age- and IQ-matched controls.

Methods:  MEG and EEG recordings were taken as participants passively read single words. Following preprocessing of combined MEG/EEG data, we explored the neuronal generators underlying electrophysiological and neuromagnetic activity with source analysis using high-resolution structural MRI scans from each participant. Results were confirmed in both an anatomically-defined and data-driven regions of interest (ROI) approach.

Results:  The physiological data revealed preferential recruitment of temporal areas associated with the lexical route in control subjects. Though showing less activation than controls in temporal areas, ASD participants showed no preferential use of either route, but additional recruitment of the nonlexical route associated with a dorsal stream of activation in parietal areas and BA 44. Greater activation of this dorsal route in ASD occurred throughout the epoch, from 70ms after word presentation to 450ms. In addition, analysis of semantic differences between experimental words revealed many category-specific semantic differences in the control group that were lacking in the ASC group, particularly in the earliest time-windows.

Conclusions:  In contrast to controls who preferentially employed the lexical route, people with ASC appear to automatically decode even familiar words, showing recruitment of additional pathways and theoretical consistency with hyperlexia, a tendency to mechanically phonologically-decode language whilst, in previous literature. The lack of automatic category-specific differences between words with different meanings in the ASD group supports the suggestion that whilst semantic processing in autism may not be globally impaired, semantic information may not be automatically activated without explicit instruction.