Seventy-five subjects were trained on a pursuit rotor for 10 trials, with ambient illumination from a strobe light flashing at frequencies of either 2,5,10,15, or 20 per second. A transfer trial followed, with a strobe flashing frequency of 10 per second for all subjects. Results supported hypotheses derived from Adams' closed-loop theory of motor learning that (a) performance would improve during training as a function of amount of visual feedback available, and that (b) if after training visual feedback was reduced, performance would be maintained to the extent that kinesthetic feedback had been learned to be relied upon as an alternate, compensatory, feedback loop. (Author)