In human movement and sports science, manipulations of perception and action are common and often comprise the control of events, such as opening or closing liquid crystal goggles. Most of these events are externally controlled, independent of the actions of the participants. Less common, although sometimes desirable, are event manipulations that are dependent on the unconstrained movements of participants. As an example, we describe a method we used previously to manipulate vision of basketball jump shooters on the basis of on-line registration of their own movements. The shooters wore liquid crystal goggles that opened or shut as a function of specific kinematic features of these movements. The novel aspect of this method is that the criteria for detecting movement patterns and performing the appropriate manipulations are adjustable to the specific sport context and the complexity and variations of the unconstrained movements. The method was implemented as a finite state machine: a computer system that can be used for pattern recognition. We discuss this method, how it works and the potential it has for studying perceptual-motor skills in sport. Furthermore, the results of the basketball experiment are briefly summarized and complemented with new analyses.