Vervet monkeys received food reinforcement contingent on autogrooming. Experiment 1 reinforced grooming on a schedule of increasing intermittency and grooming increased in frequency and duration; with only pauses reinforced, grooming decreased in frequency and duration. Experiment 2 demonstrated differentiation of operant autogrooming; in each session a different single form of grooming was reinforced (for example, grooming the tail only), and that form increased in frequency while other forms became less frequent. In Experiment 3 scratching was succesfully conditioned with a method that selectively reinforced variety in behavior; reinforcement was contingent on a shift in scratching form. In Experiment 4, with no contingencies on grooming, a prefood stimulus did not increase autogrooming whether or not grooming had previously resulted in contingent reinforcement. The form of conditioned autogrooming resembled the form of unconditioned autogrooming. The discussion suggests how reinforcement principles can account for changes in the topography of operant behavior.Key words: operant autogrooming, operant scratching, response differentiation, response topography, response variety, resurgence, variable-duration schedule, fixed-duration schedule, vervet monkeysIn selecting examplars of instrumental or operant behavior, both Thorndike (1911) and Skinner (1938) originally chose biologically neutral responses, with no unconditional relation to the reinforcer (i.e., responses that were not elicited by the reinforcer). In the words of Thorndike (191 1), responses should be chosen so that conditioning would proceed without the "helping hand of instinct" (p. 30). After having conditioned arbitrarily selected responses such as latch pulling in cats with food, Thorndike extended the paradigm to more natural patterns of behavior, also with no connection to the reinforcer; for example, in cats autogrooming responses could be conditioned with food. Lorge (1936) further demonstrated that face washing, standing up, begging, and scratching in rats could be conditioned with food. The view then was that reinforcement works regardless of the biological relevance of the reinforcer for the response and regardless of the specification of the response (Lorge, 1936).Yet, not all attempts have been successful in conditioning autogrooming and otherWe thank the Institute of Neurophysiology, Panum Institute, University of Copenhagen for providing us with space, equipment, and animals. We thank