A test of the Wagner-Rescorla model’s prediction of no response decrement in a nondifferential following a differential schedule of reinforcement

Abstract

When the pigeons are trained with a multiple variable time (VT) 30-sec VT 60-sec EXT (30-60-EXT). schedule with key colors A, B, and C (respectively), response to A is greater than response to B, and response to B is greater than response to C. When the birds are switched to multiple VT 60-sec VT 60-sec VT 60-sec (60-60-601 with A, B, and C, response to A and B falls to the level previously seen for C. The Wagner-Rescorla model predicts this pattern of results.The model also predicts no decrement in response to B if the switch is to 60-60 training with A and B only (omitting C). The present experiment showed a severe decrement of keypecking in 60-60 with A and B, equivalent to that seen in 60-60-60 with A, B, and C. Other explanations of the performance-failure appear necessary. An alternative description of VT-VT schedules, that they contain no signal for no reward, is considered in terms of the notion that no-reward periods are necessary for the elimination of behaviors that compete with keypecking.

Duration of signals for intertrial reinforcement and nonreinforcement in random control procedures.

“…Second, the current results, along with those of the two-component multiple-schedule work discussed earlier (Brandon, 1981, 1984; Farley, 1980; Gamzu & Schwartz, 1973; Spealman, 1976), show that signaling ITI USs per se in a nondifferential condition is not sufficient for responding to emerge to a target CS. The long-cover (Experiments 1 and 2), multiple-cover (Experiments 2 and 3), and short cover/short CS− (Experiment 2) groups had the extra USs signaled, yet responding did not occur to the target CS.…”

“…The multiple-cover groups in which ITI USs were signaled with no discriminable period of nonreinforcement did not acquire responding; only groups with a true CS-period did. As mentioned earlier, the findings of some three-component multiple-schedule work also suggest that a period of nonreinforcement is necessary (Brandon, 1981(Brandon, ,1984Farley, 1980). However, a consistent CS-period existed within the long-cover conditions of Experiments 1 and 2 and the short-cover conditions of Experiment 3, yet responding to the target was not supported.…”

“…An interesting variant of the multiple-schedule procedure is one in which a third stimulus component is added so that the probability of reinforcement can be the same in the presence of one (target) stimulus as in its absence, but one of the other two (nontarget) stimuli can be correlated with nonreinforcement (Williams, 1976). Of particular interest here, Brandon (1981Brandon ( , 1984 and Farley (1980) found that, in a three-component multiple schedule, responding was maintained to the target stimulus only when one of the stimuli was correlated with nonreinforcement. These results suggest that a discriminable ITI, or period of nonreinforcement, may be necessary for the acquisition of responding to a target cue in a nondifferential procedure.…”

Temporal comparator rules and responding in multiple schedules

“…These findings have come from both the autoshaping and multiple-schedule paradigms (Brandon, 1984;Cooperet al, 1990;Durlach, 1983;Farley, 1980;Goddard & Jenkins, 1987;Marcucella, 1976). Farley (1980, Experiment 1; see also Brandon, 1984) examined maintained keypecking in pigeons trained on a three-cue multiple schedule. Initially, reinforcement was delivered at the same rate in all three components by using a response-independent, variable-time 60-sec (VT 60) schedule.…”