Pigeons emitted interresponse times that were reinforced if they fell between an upper and a lower bound (t<IRT<t+t/10). Brief stimuli followed each response; under some experimental conditions the color of these stimuli was correlated with whether the preceding interresponse time was longer or shorter than that specified by the schedule. Preliminary experiments indicated that these "feedback" stimuli acquired no discriminative properties even after prolonged training. A modified procedure, in which t varied cyclically throughout each experimental session, allowed the stimuli to acquire such properties: stimulus control was demonstrated under the training conditions, for two of the pigeons, and under transfer conditions for all three birds. A series of probe conditions, followed by a replication of the simple procedure using a multiple schedule, indicated that the controlling property of the stimuli was not the relation between stimuli, interresponse time, and value of t, but a variable determined by the interaction between the animals' responding and the cyclic procedure. This variable was probably the relative frequency of the less-frequent feedback stimulus.