Ss in the experimental group were asked to imagine an idiosyncratic noxious scene the cessation of which would be contingent upon imagining a neutral stimulus (a ringing bell). This was repeated 30 times. A second group was asked to imagine noxious and neutral scenes in an unpaired manner. A third group received no imagery training. In the experimental test phase, all Ss were asked to estimate the size of circles to establish a baseline. With Groups 1 and 2, E said the word "bell" when the Ss either overestimated (during the over-estimation condition) or underestimated (during the under-estimation condition) the size of the circles. Deviation of the estimates of circle size was influenced by use of the word "bell" in Group 1 only supporting the covert negative reinforcement hypothesis. © 1972.