An evaluation of the abstraction and irrelevant order principles of counting in rats

In two experiments, rats were runway-trained on a variety of reward schedules involving three different reward outcomes: Noyes pellets ®, Kellogg's Corn Pops ®, or a nonrewarded trial (N). In Experiment 1, the training schedule was R'RRN and a single N trial. Following this initial training, half the subjects were tested on the shift schedule RRN and NRRN, while the other half were tested on the schedule RRRN and NRRRN. The results of performance on the shift phase suggest that the rats were not simultaneoously categorizing the reinforcers as similar (food items) and/or different (Noyes pellets vs. Corn Pops). In Experiment 2, rats were trained on one of three schedules, R'RRN, RR'RN, or RRR'N, and then tested on either the shift schedule RRN and >JRRN or RRRN and NRRRN. The results of performance on the shift phase suggest that, contrary to Capaldi's conclusions, the position of the 'different' reinforcer is not irrelevant in terms of affecting the subjects' ability to perform the discrimination.