Abstract:
A discrete-trials procedure was superimposed on concurrent-chain schedules with variable-interval initial-link schedules and with fixed-interval and mixed-interval terminal-link schedules. Three different pairs of terminal-link schedules were used and each condition was replicated. The results of this experiment (Experiment 1) suggested the use of a measure of initial-link preference which was modified to take account of the inequalities in obtained terminal-link entries from condition to condition.
The performance under three different types of concurrent-chain schedules was then examined. These three types were discrete trial (with two different trial lengths; Experiment 2), free operant (Experiment 3) and response-contingent blackout (with two lengths of blackout; Experiment 6). In all these experiments the concurrent chains had initial-link schedules averaging 60 sec and fixed-interval terminal-link schedules. One of the terminal-link schedules was always 10-sec long and the other ranged from 10-sec to 60-sec long in the different conditions.
Comparisons of the initial-link performance showed that the length of the discrete trials did not affect preference while the length of the response-contingent blackout did do so. The discrete-trial initial-link preference measures were generally less extreme than the equivalent measures from the free-operant concurrent chains. The initial-link preference measures from the shorter (0.5-sec) contingent blackout were similar to those from the free-operant concurrent chains and were more extreme than those from the longer (1.0-sec) contingent blackout conditions.
The terminal-link performances under the different concurrent-chain schedules, although different in absolute magnitude, all varied in much the same way regardless of the type of procedure in effect.
Performance under the component schedules of concurrent chains was also examined. Single-chain schedules with initial links averaging 60 sec and fixed-interval terminal links similar to those in the concurrent chains were used (Experiment 4). The behaviour under both discrete-trial and free-operant single-chain schedules was examined. In both the initial and terminal links of these schedules the behaviour changed in a similar fashion regardless of the procedure in effect. The terminal-link behaviour in the chains was similar to that in the terminal links of the concurrent chains. The initial-link behaviour in the chains was, however, different, from that in the initial links of the concurrent chains.
The performance under free-operant multiple fixed interval schedules with component schedules similar to the terminal links of the concurrent chains was found to resemble the behaviour in those terminal links (Experiment 5)
Thus the terminal-link behaviour showed remarkable similarity in all the schedules examined regardless of the procedure in effect. The initial-link behaviour, however, differed from procedure to procedure and that in the concurrent-chain initial links could not be predicted from that in the single-chain initial links.