A test for S-S associations in a conditional counting task

Conditional numerical discriminations were established in runway-trained rats by presenting daily a pair of reward series that differed in the number of rewarded trials preceding a terminal nonrewarded trial. One kind of series contained three rewarded trials (RRRN or NRRRN), which always occurred in either a smooth, black runway or a rough, white runway (counterbalanced). The other series contained four rewarded trials (RRRRN or NRRRRN), which occurred in a runway of the other brightness and texture. The reward on each rewarded trial was 5.045-g Noyes pellets until Day 22. For Days 22-32, the series containing four Rs was altered so that the number of pellets on each respective trial was 14, 7, 3, and 1. To test whether S-S associations controlled the serial discriminations after Day 32, we rearranged the order of elements in the 14-7-3-1-0 series to 20-7-3-14-0. Rapid approach to the terminal series element on the test day would have implied the operation of S-S associations (14 pellets would have been a cue for upcoming 7, not 0), but the approach to that element remained slow, suggesting that the animals continued their counting of rewarded trials during the training with the S-S-relevant 14-7-3-1-0 series, irrespective of the order of the series elements.