Error Correction Process

As a first step toward analyzing error correction in these patients, it is necessary to establish that a first-order linear error correction model provides an adequate account of the patients' performance. Given their neurological impairments and hypotheses concerning the role of the basal ganglia and cerebellum in online error correction (Flament and Ebner, 1996; Lawrence, 2000; Smith et al., 2000), we considered it possible that a qualitatively different strategy might characterize the performance. A second-order error correction model has been shown to provide a better fit under certain circumstances. For example, these higher-order models are more appropriate when expert musicians tap at a fast pace (Pressing and Jolley-Rogers, 1997). In our study, the proportion of runs that were adequately accounted for by a first-order error correction model was roughly equivalent across the groups. Thus, we infer that, qualitatively, both patient groups used a strategy similar to that of the control participants in how they used asynchrony information to adjust their performance. However, the results indicate a quantitative deficit in error correction for the PD patients. The estimates for the first-order parameter a tended to be lower for this group, although the result was only marginally significant. If this finding were replicated, it would suggest that a dopamine-related deficit in the striatum reduces the gain at which the error signal influences the next outgoing motor command (see also Malpani and Rakitin, this volume).

0 0

Post a comment