A neural circuit for coordinating reaching with grasping: autocompensating variable initial apertures, perturbations to target size, and perturbations to target orientation

A neural network model is presented, that extends principles of the VITE (vector integration to end-point) model (Bullock and Grossberg, 1988, Bullock et al., 1998, 1999) of primate reaching to the more complex case of reach-grasp coordination. The main new planning problem addressed by the model is how to simulate human data on temporal coordination between reaching and grasping, while at the same time remaining stable and compensating for altered initial apertures and perturbations of object size and object location/orientation. Simulations of the model replicate key features of four different experimental protocols with a single set of parameters. The proposed circuit computes reaching to grasp trajectories in real-time, by continuously updating vector positioning commands, and with no precomputation of total or component movement times. The model consists of three generator channels: transport, which generates a reaching trajectory; aperture, which controls distance between thumb and index finger; and orientation, which controls hand orientation vis-a-vis target's orientation.