Abstract
Four rats had continuous access to activity wheels first, then access for 1 hr per day, and, subsequently, continuous access. Limiting S's access to the wheel substantially increased the total frequency of running. A distributional analysis of response duration, burst duration, and interburst interval showed that the increased frequency arose almost entirely from a shortening of the interval between successive bursts. In contrast, speed of the individual response and number of responses per burst changed only negligibly. If S were running, the probability that it would either stop or continue did not differ appreciably for the conditions of continuous or limited access to the wheel. But if S were not running, the probability that it would start running was appreciably greater for limited than for continuous access.