Neurophysiological Mechanisms of Statistical Learning in Adults with and without Reading Disorders

The artificial grammar learning (AGL) task first introduced by Reber (1967) as well as similar paradigms (e.g., Jost et al., 2015) are thought to elicit implicit statistical learning (SL) of underlying patterns in typical readers. However, previous research has shown that individuals with dyslexia often show difficulty with such incidental learning, on AGL and other SL tasks (Kahta & Schiff., 2016; Singh, Walk and Conway, 2018). Because few studies have investigated this link between statistical learning and reading ability, the current study was designed to examine the neurophysiological and behavioral correlates in adults with and without a reading disorder diagnosis. Sixteen reading impaired and thirty-seven typically reading adults were recruited for the study and completed the AGL, and SL (visual-motor; auditory-motor) tasks, followed by completion of questionnaires eliciting awareness of underlying patterns. During these tasks, behavioral measures such as response times and grammaticality classifications were recorded. Additionally, event-related potentials (ERPs) were also acquired during the computerized tasks. Following this, normed assessments indexing cognitive, reading and spelling ability as well as basic musical ability were administered to participants. Prevalence of attention-deficit symptoms was also accounted for by administration of a checklist. The aim was to assess the underlying mechanisms of implicit-statistical learning such as transition-timing and chunking as well as grammaticality (algebraic patterns and ordinal knowledge) via varied task paradigms (SL and AGL respectively) and non-linguistic stimuli. Although behavioral results were comparable across groups, ERP amplitude differences vary in topology across groups – especially for grammaticality and chunk strength, but not so much for the transition timing paradigms. For atypical readers, correlations were only found between symbol search scores and ERP responses for grammaticality. Thus, overall, the current study highlights the need to assess participants in terms of overall learning capacity before investigating the link between implicit-statistical learning capacity and reading ability. Additionally, findings indicate that participants were not as sensitive to non-linguistic items across learning paradigms as they might have been to purely linguistic items