This thesis examines the integration of visual and somatosensory information in the planning and execution of goal-directed hand action by having subjects point, with or without vision, to targets that can be defined either by visual or by proprioceptive cues. The first experimental chapter examines the performance of healthy control subjects while subsequent
chapters study patients with impairments to either the visual or somatosensory systems. Other chapters use prismatic displacement to distort the visual input of healthy normal subjects.
The study using patients with neglect reveals that the curvature of their reaches is abnormal only when visual cues are available and the conclusion is drawn that in these patients the basis of their neglect must be visual rather than spatial. In the cases of patients with somatosensory impairments performance was almost always improved by having visual
cues available, despite the fact that these cues could not signal the location of the proprioceptively-defined target directly. The final chapter, a case study of a single patient with unilateral somatosensory impairment, reveals that being able to see the area of workspace immediately adjacent to the proprioceptively-defined target significantly improves proprioceptive
localization of that target.
In all cases, and in particular after prism adaptation, performance varies according to whether visual and/or proprioceptive cues are available and a case is argued for a change in the frame of reference that is used to plan reaches under visual or proprioceptive conditions. In addition it is suggested that when planning a reach, information from each system is
weighted according to the goodness of its source.