Variabilité sensorimotrice et adaptation – VARAD
Variability is an inherent feature of motor output both between and within subjects in all categories of tasks. Yet traditionally this feature of motor behavior has been dismissed as insignificant in many theories of motor control and operationally treated as a standard deviation in distributional statistics. In theories about the variability of motor output, the most common position has been that variability reflects noise. Recently some optimal control theories have been proposed stating that motor system minimizes variability both in sensory information and in motor output. This principle of optimality suggests that variability is a nuisance and should be controlled in order to minimize 'costs' functions. In other words, variability is regarded as a dysfunctional aspect of a sensorimotor system, resulting from some random fluctuations that compromise the deterministic relation between a 'sensori- input and a 'motor- output. This view has been challenged by other theoretical approaches that regard variability as a central factor for adapting behaviors. On the one hand the dynamic systems theory places greater emphasis on the space-time characteristics of coordination patterns in tasks requiring multiple degrees of freedom. Within this framework variability reflects the capacity for flexible and adaptive sensorimotor systems to adapt to specific environmental and tasks demands. On the other hand, the selectionist approach of operant conditioning postulates that new behaviors are shaped based on variations of other responses: variability is necessary to adapt to changes in reinforcement contingencies. Thus variations of behavior leading to the final response are selectively reinforced driving gradual changes in the topography of the response. It follows that variability is useful ' if not necessary ' for learning new behaviors. This hypothesis have rarely been tested, except for one study in which experimental manipulations of the level of variability revealed that only rats trained to produce variable responses were able to acquire a difficult-to-learn sequence. To our knowledge, the possibility to control behavioral variability has never been tested for sensorimotor control in humans, except for one of our recent study. In this set of experiments we showed that one could increase or decrease variability of latencies distributions in saccades and manual responses, independently from the medians, by manipulating reinforcement contingencies. Our interpretation is that some of the observed variability reflects environmental constraints controlling the response instead of depending solely on internal noise. In other words, we propose that variability in behavior also results from loose environmental constraints. A fundamental question involving variability has been the understanding of the emergence of new responses using the concept of shaping. In shaping a response-reinforcer contingency is gradually changed to select environment-behavior relations that progressively approach some criterion response topography. In contrast, the classical engineer-like approach of sensorimotor control usually regards a change in the system as a re-calibration of some parameter. For instance in adaptation phenomenon in which the outcome of the system is modified by applying some constant perturbation, the system progressively adapt in order to compensate for this change in its dynamics. Within the oculomotor system, this has been extensively studied by shifting a target during a saccade so that the latter becomes dysmetric. The conventional view has been to view saccade adaptation as a servo-mechanism, in which the presence of an 'error signal' changes the parameters of the saccadic system, thereby reducing the error over many trials. According to this conceptualization, saccade adaptation is learning only in the most restricted sense of the word, in that a change in behavior occurs as a consequence of an adjustment of some unspecified internal gain. We propose to test whether variations of motor behaviors are under the control of reinforcement contingencies. Moreover we intend to test for a relationship between variability and the ability to adapt to changes in the environment. These hypothesis will be addressed in the saccadic system and the arm movement system and involve behavioral studies. We anticipate that this project will provide a solid framework for studying other forms of sensorimotor plasticity. If the results corroborate our hypothesis this will constitute a breakthrough in the study of motor learning, with fundamental implications for the understanding of plasticity in other systems.
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Help of the ANR 0 euros
Beginning and duration of the scientific project: - 0 Months