Abstract
It has been reported that animals trained in a maze-type apparatus develop maximizing more rapidly than those trained in a Skinner type one. Besides, in the former apparatus a longer inter-trial interval and a smaller number of trials per day are usually used than in the latter one, which may also account for the above difference. The present experiment was designed to investigate rats' performance in a probability learning situation with the above two variables equalized for both apparatus.
Two groups of rats were trained on a spatial 75 : 25 probabilistic schedule for 920 trials using correction method. One group was trained in a T-maze and the other in a 2-lever Skinner box. Inter-trial interval was 13 sec. and the daily number of trials was 40.
Fig. 1 indicates that the T-maze group showed maximizing in only a few hundred trials, while the Skinner-box group remained at a matching level even after 900 trials. Table 1 shows the individual data. The difference between groups was significant (U=21, p<. 05).
These results replicated those of previous studies. Therefore, the difference in performance between apparatus conditions can be attributed to neither inter-trial interval nor the number of trials. The remaining procedural difference is that Ss trained in the T-maze were removed from the apparatus after each trial, while those in the Skinner box were not. This may have produced the present results. This possibility remains to be examined.
