Concurrent performance as bouts of behavior

Log-survivor analyses of interresponse times suggest that the behavior of rats responding under single variable-interval schedules is organized into bouts (i.e., periods of engagement and disengagement). Attempts to generalize this analysis to the key pecking in pigeons, however, have failed to produce the characteristic broken-stick appearance typically obtained with rats. This failure may be due to a relatively low rate of reinforcement for engaging in alternative behavior experienced by pigeons. The present study tested this hypothesis by exposing four pigeons to concurrent schedules of reinforcement for key pecking, first without a changeover delay (COD) and then with a COD. In this arrangement, one of the concurrent options was treated as the target response and the rate of reinforcement for that option was manipulated across conditions. The other option provided explicit reinforcement for engaging in an alternative response (i.e., explicit reinforcement for disengaging from the target response). In the absence of a COD, log-survivor plots for three of the pigeons were approximately linear, thus providing no evidence that responding was organized into bouts. When a COD was present, plots were broken stick in appearance, indicating a bout structure had been generated in the pigeons' behavior. Both bout length and the rate of bout initiations were a function of differences in rate of reinforcement. These data suggest that behavior may become organized into bouts when contingencies create sufficiently long visits to both the target behavior and the extraneous behavior. Fits of a double-exponential model deviated systematically from the actual plots due to the presence of a plateau between the two limbs. An alternative, double-gamma, model was explored, and it provided a considerably better fit than did the double-exponential.