The thesis tested was that a CR which tends to extinguish so long as it plays the secondary or conditioned role, tends to persist and even to become enhanced when it is made the primary or conditioning role in establishing a further second order conditioned response. 24 cats were used. The hind-leg withdrawal response was the reaction tested, while the stimuli were shock, bell, tone, and light. Group I (2 cats) was conditioned to a bell, then the response was extinguished. Group II (4 cats) was conditioned to a bell and then the conditioned response maintained by occasional reinforcements for 30 days. Group III (11 cats) was first conditioned to the bell and then a second order response established to a tone[long dash]no shock was required for this latter. Group TV (1 cat)[long dash]the second order conditioned response was extinguished by omitting the bell. Group V (2 cats from Group III) was brought into third order conditioning, a light stimulus serving as the conditioned stimulus. Third order conditioned responses were not brought to a completion. The results indicate that a stimulus used a.s the primary stimulus in second or third order conditioning is as effective as the primary stimulus at the next lower order, not losing in effectiveness, but apparently gaining when changed from a secondary to a primary role in the conditioning process. There is a retroactive facilitation by the response which is being conditioned upon the conditioning response. Two hypotheses are offered to explain the results (1) The effect is a case of Bahnung; (2) The conditioned symbol (bell) becomes the focus of progressive involvement of the type underlying fear or mild phobia, which thus provides additional drive.