Visually guided grasping in depth is systematically inaccurate
33.319, Sunday, 18-May, 8:30 am - 12:30 pm, Jacaranda Hall
Claire Walker1, Carlo Campagnoli1, Fulvio Domini1,2; 1Department of Cognitive, Linguistic & Psychological Sciences, Brown University, 2Center for Neuroscience and Cognitive Systems@UniTn, Italian Institute of Technology
Current theories postulate that the visual system is divided into two parallel pathways: a dorsal stream that guides motor actions and a ventral stream for conscious perception and recognition of objects. It has been hypothesized that the dorsal stream has access to accurate 3D information. However, recent empirical findings cast serious doubts on this hypothesis, since we found that reach-to-grasp actions are highly inaccurate when visual feedback of the hand and haptic feedback of the objects are absent (Campagnoli et al., 2012; Foster et al., 2010). Here we investigated to what extent the visual feedback of the limb may improve the accuracy of a reach-to-grasp action. Participants were asked to grasp along the depth axis a virtual, disparity-defined vertical cylinder. In different trials, the simulated cylinder had different diameters (30 mm and 50 mm) and appeared at two different distances (280mm and 400mm). The cylinder was visible throughout every trial, whereas the visual feedback of the tips of the grasping fingers (e.g. index and thumb) was present from the start of the grasping action until the fingers reached a specified distance from the cylinder. In six consecutive blocks, this distance was 70mm, 30mm, 20mm, 15mm, 10mm, and 5 mm. In two control blocks, the visual feedback was either absent altogether (first block) or always present (last block). In all conditions, the Final Grip Aperture (FGA) was highly inaccurate, showing that the visual feedback of the fingers is not sufficient for an accurate reach-to-grasp action. Interestingly, we observed a dramatic improvement in FGA accuracy from the 70mm block to the 30mm block. However, no further improvement was observed in subsequent blocks (30 – 5 mm). These results are in contrast with the hypothesis that the dorsal stream performs a veridical metric analysis of objects for motor action.