A response duration schedule: effects of training, extinction, and deprivation.

Rats were trained to hold down a lever for at least 40 consecutive seconds. When the lever had been held down for 40 sec, white noise came on. Releasing the bar in the presence of the noise turned off the noise and operated a feeder that delivered a pellet of food. At the end of training, frequency distributions of response durations peaked at 40 to 41 sec. If as in training, holding down the lever produced white noise at the end of 40 sec, and release of the lever terminated the noise and operated the feeder, but no food delivery occurred, duration distributions and several other measures were initially not very different from when food was delivered. However, if during extinction white noise was never produced by lever holding, and feeder operation did not occur upon lever release, most responses were shorter than 1 sec in duration, some were much longer than 41 sec, and duration distributions did not peak at 40 to 41 sec. When reinforcement was reinstated after extinction, performance quickly returned to pre-extinction measures. Further sessions at different levels of deprivation produced only temporary disruptions in performance.