Response Bursts on DRL Schedules

Abstract

<p>The differential reinforcement of low rates schedule typically yields from the subject a distribution of interresponse times (IRTs) that is bimodal. One mode occurs at or near the reinforced value and a second one at IRT values less than 2.0 sec. The IRTs which comprise this latter mode have been termed "response bursts" (Sidman, 1956). Various explanations for the mode at short values have been offered: Blough (1963, 1966), arguing from pigeon data only, suggested that these short IRTs are due to variation in the topography of the pigeon's keypeck response, the double-peck being a variation of the peck. Kramer & Rilling (1970) noted that if the definition of what actually constitutes a response is not clear to the subject, then quick multiple responses might result. Finally, Sidman (1956) has viewed these sequences of rapidly-made responses as evidence of increased response strength at certain times.</p> <p>A comparison of IRT distributions from mice, rats, and pigeons, however, shows such similarities in the patterning of bursts that Blough's suggestion of topographic variation in the pigeon's key peck cannot be a satisfactory explanation for the occurrence of these short IRTs. To test Kramer and Rilling's suggestion that lack of clear response definition may produce multiple responses, mice were trained on a DRL 18 sec schedule and given response feedback in the form of an auditory stimulus. The results of a series of such experiments indicated that while a relatively long-duration auditory stimulus (2.0 sec) would almost eliminate bursting, it did this by acting as an S Δ rather than as response feedback per se. The fact that a short-duration (0.2 sec) auditory stimulus did not suppress bursting showed that bursting was not the result of lack of response definition. Further analysis of the patterning of response bursts lent substance to Sidman's contention that response bursts are a reflection of momentary increase in response strength. Specifically, bursts were found to occur most often as time to reinforcement availability decreased; bursts never occurred immediately after responses which were reinforced (i.e., when time to reinforcement availability was greatest); and the length of a burst run (number of responses) was greater the closer it occurred to the time of reinforcement availability. It was concluded that Sidman's interpretation of the DRL schedule as a temporal discrimination task is consistent with the response burst data from the experiments described here.</p>