The control of rapid limb movement in the cat

In cats trained to track a moving display by making rapid, isometric force adjustments, responses are characterized by extremely short reaction times (60-70 ms) and a stereotyped temporal configuration. The animal uses early derivatives of display movement to scale force responses to target stimuli of different sizes according to a learned relationship between initial display motion and required force (Ghez and Vicario 1978a, 1978b). In the present study we altered that relationship by using double stimulation and delayed feedback to assess the animals' ability to update their responses. In experiments where a second target stimulus followed the first after a controlled interval (15-120 ms) on random trials, the animal modified its response in the appropriate direction with little or no increase in reaction time. When the second stimulus called for a return to baseline, the animal aborted the ongoing response. When the second stimulus called for a doubling of force, the animal increased its phasic force output; however, this increase was not sufficient to reach the new target level and late responses were emitted. The control response which followed each experimental double stimulation trial showed consistent differences from other controls in the amplitude of both peak force and peak dF/dt. Control responses following trials calling for a return were reduced in size; those following stimuli requiring response doubling were increased. We concluded that the experimental trials not only elicited modification of ongoing responses but also caused the animal to alter its internalized gain function relating initial display derivatives to required force. In experiments where feedback was delayed after giving a first target stimulus such that the compensatory display failed to reflect the animal's initial response, the animal emitted a new updated response 70-80 ms after the first. The display trajectory which caused the cat to update its response on delayed feedback trial was identical to that of control trials with long reaction times. In this case, however, the information eliciting response updating had to be derived as a difference between the actual display trajectory and that expected by the animal, based on its experience with the tracking task. This suggests that the animal develops an internal model of display properties which is used to determine when a new response is required.