Thèse soutenue

FR  |  
EN
Auteur / Autrice : Thomas Hoellinger
Direction : Agnès Roby-Brami
Type : Thèse de doctorat
Discipline(s) : Neurosciences
Date : Soutenance en 2009
Etablissement(s) : Paris 6

Résumé

FR  |  
EN

Since antiquity, humans have attempted to reproduce their own movements as demonstrated by multitudes of artistic expressions. The recognition of action has been the object of many studies since the discovery of mirror neurons both in humans and monkeys, suggesting a resonance of action observation in the central nervous system. However these mechanisms are controversial and few studies have attempted to define the movement characteristics that are important for action recognition. This work was based on the analysis of human kinematic characteristics. The first aim was to define important variables for action and the second was to compare them to variables implicated in action recognition by an observer. In the methodological part, we developed a method of 3D reconstruction from electromagnetic recordings carried out with a restricted number of sensors. This method allows the study of 3D hand displacements as well as the configuration of the whole arm. Moreover, it allows the creation of a display made up of points or a stick diagram. The experimental part was based upon a protocol in which participants were asked to reach, grasp and lift cylindrical objects onto a target placed 18cm above the support surface. The visual appearance of the different objects was identical but the weight varied. In the first experiment, we studied the kinematic characteristics of the endpoint. When the weight of the object could be anticipated, the kinematics of the majority of the variables were invariant across the different weights. When the objects were presented randomly and the weight could therefore not be anticipated, these invariant characteristics disappeared. We found that when the object was heavier, the duration of the grasp increased and the acceleration during the lift decreased. This experiment was also carried with a deafferented patient. This revealed the importance of proprioceptive information in the regulation of the kinematics of action. A computational model of the effect of the weight on the movement characteristics in the different conditions of anticipation and proprioceptive feedback was developed in order to better understand the mechanisms behind the variations observed. In the second study we analysed the mechanisms which allow the weight of an object to be judged during observation of its lifting (observers saw either a recording of their own movements or those of another person). In the two visual conditions (working point or stick diagram), we observed that the response of the participants was correlated with the weight of the objects. Acceleration during the lift phase had the highest correlation with the subjects’ responses. In other words, the higher the acceleration, the lower the perceived weight of the object. The variability of the responses suggested that observers do not use other variables which also vary with the weight, neither do they base judgement on maximal height, which may be misleading. This might suggest that the response is not due to a mirror resonance. Acceleration during lifting seems to be specific to the weight during action even when anticipation is not possible. Moreover, it is particularly important for the visual judgement of the intrinsic properties of objects. Overall, these results show that kinematic analysis, when coupled with models, is a useful tool to increase understanding of the mechanisms of human motor control and action recognition. The methodology used (electromagnetic sensors and a kinematic model) could also allow a richer graphic environment to be created (an avatar, for example) so that experimental conditions can be altered, particularly for the study of action recognition (changing the view point, for example). The perspectives seem to be numerous in the field of rehabilitation as well as for virtual reality (or augmented reality) or for fun or educational purposes