Seven rats were exposed to a procedure where responses on an observing lever (observing responses) changed the schedule on a food lever from a mixed schedule to a schedule with a discriminative or uncorrelated stimulus for a fixed duration; Observing responses produced a stimulus associated with a variable-ratio 50 (S+) or a stimulus associated with extinction (S-) in the S+/S- condition, an S- in the S- only condition, and an S+ in the S+ only condition, while they produced two uncorrelated stimuli in the uncorrelated condition. Experiment 1 employed two stimulus durations (30 or 6s) and two schedules of stimulus presentation (a variable-interval 30s or a fixed ratio 1), and examined observing response rates in the S+/S-, S- only, S+ only, and uncorrelated conditions. Experiment 2 employed two component durations (80 or 300s), and examined observing response rates in the S+/S- and S- only conditions. In both experiments, rates of observing responses were maintained in both S+/S- and S+ only conditions, but decreased in the S- only and uncorrelated conditions. These results indicate that an S- is not a conditioned reinforcer irrespective of manipulated variables.