The collateralbehaviorof pigeonsunder differential-reinforcement-of-low-rate (DRL) schedules was evaluated for its role in controlling DRL performance. Two of three pigeons engaged in high rates of collateral keypecking under schedules up to DRL 28 sec. Rate of collateral pecking was positively correlated with DRL efficiency. Topographical features of the collateral behavior were inconsistent with the notion that collateral behavior mediates DRL performance through response·produced stimuli. Rather, the collateral behavior appeared to interfere with the operant response, delaying it long enough to meet the DRL requirement.
328When human and nonhuman subjects are exposed to schedules of reinforcement which require temporal spacing of responses, they frequently develop nonrandom patterns of collateral behavior between occurrences of the operant response. Though these collateral behaviors may merely be induced and functionally insignificant, the topographical and temporal stereotypy of these patterns has lead to speculation that they are functional in determining the temporal spacing of the operants. Such speculation is encouraged by studies which demonstrate that decrements in performance occur under differential-reinforcement-of-Iow-rate (DRL) schedules when subjects are prevented from engaging in collateral behavior (Laties, Weiss, & Weiss, 1969;Richardson & Loughead, 1974;Schwartz & Williams, 1971). A parsimonious interpretation of these observations is that collateral behaviors facilitate DRL performance by interfering with and thereby delaying the operant response (Schwartz & Williams, 1971). Laties et al. (1969), however, argued for a stronger interpretation, suggesting that collateral behaviors could be described as a discriminated chain of responding in which each response serves as a discriminative stimulus for the following response. In this way, the subject's own behaviors serve as a "clock," indicating the time for the operant.Although neither of these interpretations of collateral behavior has been formalized, several means of discriminating between them seem reasonable. Specifically, we shall argue that support for mediation will be demonstrated when: (1) the precision of timing under DRL schedules is functionally related to properties of the collateral behavior; (2) the collateral behavior is highly stereotyped and deviations from the stereotyped pattern lead to nonreinforcement; and (3) the sequences of collateral behavior are heterogeneous, thereby providing a set of discriminable stimuli to be used for mediation.If mediation is not supported in these ways, then we shall argue that the functional significance of collateral behavior may be related to its ability to interfere with the operant response. One prediction of this interference hypothesis is that the degree of incompatibility between the collateral and operant behaviors should be positively related to DRL efficiency.The preceding arguments were tested in an experiment in which the location and sequential patterns of collateral and operant ...