We performed experiments to investigate how humans acquire an internal representation of virtual objects through the execution of reaching movements across the object surface. Subjects were instructed to make reaching movements between points lying on the boundary of a planar virtual surface of varying stiffness. Results suggest two types of internal representation: force perturbations and object boundaries. In the first case, a rectilinear hand movement is enforced by opposing the interaction forces. In the second case, the trajectory is conformed to the object boundary so as to reduce interaction forces. While this dichotomy is evident for very rigid and very soft objects, the likelihood of identifying a surface boundary depended, in a smooth and monotonic way, on the average force experienced in the first movements. This continuum of interactive behaviors suggests that the nervous system uses a weighted combination of two control strategies, one generating a compensatory response, the other a compliant motion.