How are events and event properties recognized? We argue that qualitative properties of time extended trajectories in a 3-D optical phase space provide the information for visual event identification. Using simulations of a rigid body event, a ball rolling on a U-shaped surface, we show that the qualitative properties of trajectories in event phase space map into optical phase space. We review psychophysical and physiological evidence suggesting that ordinally scaled properties of phase space trajectories can be detected directly without detection of metrically scaled accelerations. A critical review of extant analyses of optical flow reveals that the common assumption of rigid motion and smoothness of optical flow have been required because analysis has been based on extremely brief samples of optical flow. Recent analysis has shown that approaches based an such brief samples are bound to be unstable. In contrast, the amount of structure in time extended trajectories overcomes the problem of instability due to measurement noise and makes assumption of rigidity unnecessary. However, due to the large number of optical texture elements projected from surfaces in an event and to the effects of ocdusion, a large number of trajectories appears in the optics. The degrees of freedom problem is to collapse this large number down to a single form capturing the type of event. We show how symmetries apparent across trajectories in optical phase space might be use to achieve reduction and coherence. Finally, we briefly discuss the problems introduced by motion of the observer of an event.