Some further observations on serial enumeration and categorical flexibility

Rats were runway-trained in a variety of serial discriminations that involved three different reward outcomes: Noyes pellets (R), a Kellogg’s Corn Pop (R’), and 30 sec of confinement in an unbaited goal area (N). Each of three groups experienced a series that consisted of a single non-rewarded trip down the runway, irregularly mixed with a longer series that for one group (n = 3) consisted of three trips rewarded with the pellets, followed by a terminal nonrewarded trip (RRRN/N). A second group (n = 6) had the single N series mixed with a series that included an initial trip rewarded with a Pop, and three subsequent trips rewarded with pellets, pellets, and nothing, respectively (R’RRN/N). The remaining group was trained with an RRN/N series pair. The interval between trips within a series was about 20 sec, and the interval between series was about 10 min. All rats developed strong, accurate anticipation of the terminal nonreward event of the longer series. In a 4-day transfer shift phase, a matched one-half of the rats originally trained R’RRN/N were shifted to a new series pair, RRRN/NRRRN, as were all the animals trained RRN/N. The remaining one-half of the matched rats were shifted to RRN/NRRN, as were the animals originally trained on the RRRN/N series pair. The transfer to the new series was plainly more powerful in animals initially trained with the R’RRN series than it was in either of the control groups. This shift result is taken to mean that rats classified the differing motivational events of the R’RRN series into at least three numerical categories: (1) one R’, (2) two Rs, and (3) three rewards.