When artificial grammar acquisition in Parkinson's disease is impaired: the case of learning via trial-by-trial feedback.

Although there is strong evidence that human category learning is mediated by qualitatively distinct systems, the neural substrates of procedural category learning remain largely unclear. The present research sought to investigate the role of the basal ganglia in the acquisition of categorical knowledge via an examination of the ability of people with Parkinson's disease (PD) to learn an artificial grammar (AG) system in a format akin to habit learning tasks such as probabilistic classification learning. Eighteen nondemented patients with PD were compared with 22 matched controls on a task modified so that participants' grammar learning depended on making use of feedback provided on each trial. Results showed that patients with PD exhibited abnormal AG learning, with the deficit more pronounced early in the task. Impaired categorisation performance in patients was not related to declarative knowledge obtained during the test and was independent of frontal functioning. The findings presented here indicate that prior inconsistencies across category learning tasks in PD performance may be explained, at least in part, by whether the task necessitates the formation of associations between stimulus cues and categorical responses, and support the hypothesis that the basal ganglia play a specific role in procedural learning in complex feedback-based categorisation tasks.