The removal and restoration of stimulus control.

When a well-learned circle versus ellipse discrimination was made impossibly difficult for the subjects (rhesus monkeys), the controlling stimulus-response topographies were replaced by competing topographies. The identification of two training conditions sufficient to reinstate the original discrimination permitted the following inferences: the original controlling topography had merely decreased in probability of occurrence, whereas the "strength" of the stimulus-response relation remained unchanged; discriminations along the apparently continuous circle-ellipse dimension actually involved several distinct stimulus-control topographies.