Cerebellar degeneration reduces memory resilience after extended training

Abstract

Cerebellar patients are impaired in motor adaptation. Further, motor adaptation can be divided into a fast component and a slow component, and the cerebellum is known to be especially crucial for the slow component. We tested whether the cerebellar deficit in motor adaptation can be ameliorated by training paradigms targeting slow learning using four visuomotor tasks: standard, gradual, overlearning and long intertrial intervals. We measured slow learning in patients and age-matched controls using a standard paradigm developed for reaching movements by Smith in 2006. The paradigm quantifies slow learning as the magnitude of spontaneous recovery of a previously learned and washed-out adaptation. Cerebellar patients had slower learning and reached a lower level of final adaptation, as seen in previous studies. Nevertheless, both groups had robust spontaneous recovery. Moreover, spontaneous recovery was increased in both groups in the overlearning paradigm with no significant difference between them. Computational modeling suggested that increased spontaneous recovery in the Control group reflects a changed slow adaptation system. That is, in the overlearning in controls, the slow system forgot more slowly and had a slower response to errors. In contrast, the same modeling suggests that no such change in the slow system occurred in the Cerebellar group. Rather, increased spontaneous recovery in this group seems to be the result of greater accumulated slow adaptation in the overlearning. We used our modeling results to predict that overlearning should have slower adaptation in the counterperturbation. Modeling further suggested we would not see these difference in the cerebellar patients. Follow-up analysis confirmed these model predictions. Taken together, our results imply that residual slow learning in cerebellar patients is expressed during increased training trials, but the primary cerebellar deficit is not improved.