Key-pecking intermittently produced a set of brief exteroceptive stimulus changes under two-component multiple schedules of conditioned reinforcement. Throughout the study, free access to grain was concurrently provided on an intermittent basis via a variable-interval tape. Free food presentations scheduled by the tape were delivered if no peck had been emitted for 6 sec, and the brief stimulus changes produced by responding under the multiple schedules were those which accompanied food presentation. The second component of each multiple schedule was always associated with a 1-min, variable-interval schedule of conditioned reinforcement. The schedule associated with the first component was systematically varied and conditioned reinforcement was either absent (extinction) or programmed on a 1-, 3-, 6-, or 12-min variable-interval schedule. Under these conditions, rate of responding in the manipulated component decreased monotonically with a decrease in the frequency of conditioned reinforcement. In addition, contrast effects were often obtained in the constant, second component. These results are similar to those obtained with similar multiple schedules of primary reinforcement.In their review of positive conditioned reinforcement, Kelleher and Gollub (1962) suggested and gave many illustrations of an experimental approach involving the study of conditioned reinforcement under conditions in which primary reinforcement is not removed from the experimental situation. This approach involves the application of chained schedules of reinforcement and related procedures. In a chained schedule, responding in the presence of a given exteroceptive stimulus is reinforced on a given schedule with a subsequent exteroceptive stimulus. A chained schedule can involve any number of exteroceptive stimuli, each one of which is correlated with a given member or component. Since primary reinforcement terminates the schedule, and since each of the other exteroceptive stimuli in the chain can potentially serve as a conditioned reinforcer for behavior in the respective preceding member, chained schedules can be employed to assess the con-