To Overwrite or to Recall? Individual Differences in Motor Adaptation
The central nervous system predicts the consequences of motor commands by leaning multiple internal models of external perturbations and of the body. It is not well understood, however, how new internal models are created. Here, we propose a novel computational model of motor adaptation in which a stochastic Bayesian decision-making process determines whether i) a previously updated expert perturbation model is recalled and updated, ii) a novice model is selected and is updated into a new expert, or iii) the \"body\" model is updated. Results from computer simulations provide insights into various and contradictory experimental data on savings and error-clamp, and predicts qualitative individual differences in adaptation. We verified these predictions in a visuomotor adaptation experiment in which we varied the perturbation amplitudes as well as the amount of noise added to perturbation, and added \"trigger\" trials in the error-clamp condition. Single trigger trials led to largely qualitatively different behavior and can therefore be used to probe individual differences in memory updates between \"one-model\" and \"two-model\" learners. \"One-model\" learners continuously update the body model, showing no savings during re-adaptation to the perturbation, and gradual decay during error clamp. In contrast, \"two-model\" learners switch between an updated expert model and the body model, showing large savings during re-adaptation and stochastic lags during error clamp. Our results thus support the view that motor adaptation belongs to the general class of human learning according to which new memories are created when no existing memories can account for discontinuities in sensory data.\n\nShort summary/significanceWhen movements are followed by unexpected outcomes, such as following the introduction of a visuomotor or force field perturbation, or the sudden removal of such perturbations, it is unclear whether the central nervous system updates existing memories or creates new memories. Here, we propose a novel model of adaptation, and investigate, via simulation and behavioral experiments, how the amplitude and schedule of the perturbation, as well as the characteristics of the learner, lead to updates of existing memories or creation of new memories. Our results provide insights into a number of puzzling and contradictory experimental data on savings and error-clamp, as well as large qualitative individual differences in adaptation.