Modeling kinematic forward model adaptation by modular decomposition

The time needed to adapt to a perturbation depends critically on the amount of the available a-priori information: the more we know about the perturbation, the less experience we need to learn how to compensate for it. The drawback of such a model-based approach is the loss of generality, because rigid assumptions do not allow to rapidly adapt to new perturbations. A possible intermediate solution is represented by a modular strategy, in which the generality is gained through new combinations of pre-learned models. Starting from the assumption that modules might represent a way to store a-priori information in the central nervous system, the present paper explores the consequences of such a modular forward model in human motor learning, in the context of reaching movements. In particular, we tested the prediction that in presence of a modular control, perturbations not compatible with the existing modules should be learned with more difficulty than compatible perturbations. To this aim, we confronted human subjects with two different kinematic perturbations of comparable difficulty: one compatible with the natural kinematic modules (or intra-modular) and one incompatible with them (extra-modular). We observed that human subjects adapt faster to intra-modular perturbations, thus providing evidence in favor of the adoption of a modular strategy by the central nervous system. The obtained results have some interesting consequence within the context of modular learning, hereafter discussed.