Abstract
How does a human steer toward a stationary goal while avoiding contact with obstacles in a cluttered environment? Successful steering behavior involves a dynamical interaction between a person's perceived heading and the egocentric locations of the goal and obstacles. Psychophysical data suggest that a goal acts as an attractor of heading, while obstacles act as repellers of heading (Fajen and Warren, 2003, JEP:HPP, 29:343–362). We propose a neural network model that combines neural representations of heading and goal and obstacle positions to generate realistic steering behavior. The model extracts heading from an optic flow field using network layers that model properties of cells in cortical areas MT and MST, and it constructs goal and obstacle representations by combining form and motion cues. The model also contains a circuit that controls smooth pursuit eye movements to maintain fixation on the goal during locomotion. Rotating the eye during locomotion introduces systematic distortion of the optic flow field, and the model corrects for the effects of eye rotation using extra-retinal signals. The model's architecture captures the attractor/repeller dynamics of steering behavior, while clarifying the role of heading perception and eye movements in complex steering tasks. Computer simulations demonstrate model properties on several steering tasks, including approaching goals at different distances and initial viewing angles, and steering in the presence of single and multiple obstacles. Simulation results are compared with the psychophysical data of Fajen and Warren (2003). Supported in part by NSF, ONR, AFOSR, and NGA.