Procedural or declarative deficit in adults with developmental dyslexia? A study of Artificial grammar learning

Artificial grammar learning (AGL) has frequently been employed to investigate the procedural hypothesis of dyslexia. However, most studies did not distinguish whether performance depended upon the acquisition of grammatical rules (procedural memory), distributional knowledge (statistical learning) or reference to individual memory traces (instances). An AGL paradigm was administered to 32 young adults with dyslexia and 60 controls. During the learning phase, adults with dyslexia learned artificial grammar and then utilized procedural memory in the grammaticality judgment and recognition tasks to a similar extent as controls. An overall group difference was observed only on the grammaticality judgment task; however, the performance of both groups was similarly modulated by the grammaticality of the stimuli. Adults with dyslexia acquired and used individual memories (instances) in both grammaticality judgment and recognition tasks. However, they exhibited some difficulty in forming individual memories, as indicated by their greater tendency to mistakenly identify non-grammatical control items as previously seen during the training phase (recognition task). On untrained items, adults with dyslexia also showed an advantage with high compared to low-bigram frequency stimuli, indicating reliance on sublexical statistical cues to foster their performance. In conclusion, adults with dyslexia acquired procedural rules and showed sensitivity to the distributional properties of the stimuli, providing little support for the procedural hypothesis of dyslexia. This study highlighted both spared and compensatory mechanisms in adults with dyslexia. Previous conflicting results may depend on not having isolated which components (procedural vs declarative memory vs statistical learning) are involved in the AGL paradigm.