ABSTRACT
This chapter presents some underlying neuro-computational mechanisms that can serve for such multimodal integration in infants in order to reach and grasp objects. Touching objects permits to discover the physical limits of the body and to construct its spatial representation with the use of vision and proprioception. This spatial representation is however non-linear, the same tactile stimulus can correspond to different visuo-motor pairs. The neural mechanisms involved in the construction of reach cells, visuo-motor neurons have been found to encode the preferential visual movement Direction in extrinsic coordinates in 3D space. The chapter explores how reaching a place or an object can be seen as learning sensory-motor attractors. It discusses several neural mechanisms found important for multimodal integration, reaching and grasping, as well as the importance of sighted touch for anticipatory movements. The grasping "procedure" can be seen as the building of an attractor basin around the desired object.
