Stochastic characterization of motor learning, retention and transfer

In the present research, the notion of a generalized motor program underlying movement coordination, and instantiated through learning, was interpreted as a coordination dynamics. Three experiments were directed at the potentially beneficial (positive transfer) and potentially adverse (interference) effects of practicing specific aspects of the dynamics on the dynamics as a whole. Measures of stochastic nonlinear dynamics and recurrence quantification analysis were applied to evaluate these potential effects. ^ The motor task was a bimanual coordination in which the oscillations of manipulanda in the left and right hands were synchronized and transcribed circles. The task allows four possible patterns when decomposed into inphase (synchronous) and antiphase (anti-synchronous) movements in two directions. A single potential function V of the bimanual relative phase incorporates all four relative-phase patterns and the Fokker-Planck equation relates the probability distribution P of relative phase to V. ^ In three experiments, participants practiced either a single pattern or all four patterns. Practice was preceded by and followed by a test of all four patterns. The results of the experiments suggest that the intervening practice, viewed as either facilitating (transfer) or impeding (interference), exerted a general, non-specific effect on the coordination dynamics. Indifferent to the type of pattern that was practiced, the harder coordination patterns (by an immediate post-test) and all patterns (by a delayed post-test) became less variable (but not more precise) and the component motions increased in dynamical complexity (the recurrent structure expressing the determinism of the fine time-scale correlations of the four patterns). ^ The major lesson of the experiments and their results, however, was methodological. In respect to spatially and temporally complex motor skills, the successful application of more refined quantifiers of learning, transfer and interference processes requires a greater understanding of individual differences in performance and learning trends, as well as a more abstract geometric description of the motions. The steps to be taken in this respect constituted the larger part of the discussion. Hopefully, they will inform the next round of analyses directed at the acquisition and retention of movement patterns conducted with the modern tools of statistical physics and nonlinear dynamics.