Abstract
David Marr wrote that Vision is determining "what is where by looking". This privileges the role of objects, their identity and their location. Vision, however, is much more, supporting continuous action, not only with respect to objects but also to other agents. With respect to human beings, this is ubiquitous and infinitely varied. As a possible opportunity for research, consider issues prompted by contact sports. Fencing requires that one block an opponent's sabre. Boxing requires that one block an opponent's punch. In a two-person game, we describe how a blocker counters the pointing actions of an opponent. Contestants face each other, separated by a transparent window. The attacker's job is to hit one of two side-by-side targets on the window as quickly as possible. The blocker's goal is to hit this same target within a short time lag. In our first experiment, reaction times of the blocker were short (measured by the initiation of the attacker's hand movements), approximately, 100-150 milliseconds. This is much shorter than previously reported choice reaction times and suggests that the blocker is sensing some other preparatory action of the attacker. To explore the possible locus of such actions, we replicated our first experiment and added two conditions, allowing the blocker to only see the head and eyes of the attacker or allowing the blocker to only see the torso containing the arms and hand. We find that information is potentially available in each case to support the rapid action, that some subjects use just one of these potential routes whereas others can use both. In either case, the accuracy of these fast blocker responses indicates widespread information about an attacker's intentions, well before hand movements start.
Meeting abstract presented at VSS 2014