Effects of Parkinson disease on two putative nondeclarative learning tasks: probabilistic classification and gambling.

OBJECTIVE To assess performance on two nondeclarative (implicit) memory tasks of Parkinson disease (PD) patients without dementia in the earlier or later stages of the disease (Hoehn and Yahr Scale scores of 1-2.5 or 3-4, respectively). BACKGROUND Different subtypes of nondeclarative memory appear to depend on different components of frontostriatal circuitry. Performance on a probabilistic classification learning (PCL) task was impaired by striatal damage (eg, in PD or Huntington disease) but not by circumscribed frontal lobe damage. On the other hand, performance on the Iowa Gambling Task (IGT) was impaired by damage to the prefrontal cortex. METHOD AND RESULTS On the PCL, the learning of the control (age- and education-matched) group (n = 19) and the early PD group (n = 16) was comparable with each other, and both groups showed better performance than the later PD group (n = 16). On the IGT, the control group learned better than both of the PD groups. The control and early PD groups were similar on measures from the Wisconsin Card Sorting Test, Stroop Test, Mini-Mental State Examination, and Beck Depression Inventory II. CONCLUSIONS The PCL and IGT tasks appear to rely on different parts of the frontostriatal circuitry in patients with early PD. The current finding that IGT performance was impaired in early PD implies ventromedial prefrontal cortical dysfunction early in the disease.