Reinforcement pattern learning: Do rats remember all prior events?

Abstract

Three experiments in a runway investigated how rats remember events occurring in temporal order. In Experiment 1, the reinforced trial was always the last (fifth) trial of the day for Group 4NR, the first trial for Group R4N, and the third trial for Group 2NR2N. For Group Nlrg, one ofTrials 2-5 was randomly reinforced. The running speeds for Group 2NR2N, an inverted V with its peak on Trial 3, were of special interest. An inverted V was also obtained for Group S-ITI (Experiment 2), which received the 2NR3N schedule at a 30-sec ITI, and for Group L-ITI (Experiment 3), which had a 30-min IT!. However, slow running on Trial 4 and not slow running on Trial 6 were demonstrated for Groups LINT (Experiment 2) and S-INT (Experiment 3), which received the same schedule as Groups S-ITI and L-ITI, except for a long or short interval inserted between Trials 3 and 4. These results suggest that rats remember all prior events occurring at an equal ITI whether the ITI is short or long, whereas they reset their memory when a longer or shorter interval elapses after the reinforced trial.

Counting in rats: Its functional significance and the independent cognitive processes that constitute it.

“…How Group RRN of Experiment 3 could have learned this strategy in initial training and thus how this strategy can explain the behavior of Group RRN in shift when it received NRRN and RRN training is far from clear. We are also extremely skeptical that rats use this “slow-if-identical, fast-if-otherwise” strategy on the basis of prior findings that indicate that rats can and do attend to, remember, and use many more than the previous two reinforced events of a series (e.g., Capaldi & Verry, 1981; Yazawa & Fujita, 1984).…”

The sequential view: From rapidly fading stimulus traces to the organization of memory and the abstract concept of number

“…In more recent times, the ability of rats to remember multiple reward events has been demonstrated not only by extinction data, but by acquisition data from serial leaning and counting tasks. These recent studies have shown that rats can remember multiple rewarded events, mixtures of rewarded and nonrewarded events, and multiple nonrewarded events (see, e.g., Yazawa & Fujita, 1984). The seriallearning and counting studies will be mentioned in greater detaillater.…”

“…If the animal remembered merely the prior trial of the series, it could not have responded correctly on the current trial because, owing to the use of both series, reward was followed by both reward and nonreward and nonreward was followed by both reward and nonreward. Thus, to respond correctly on Trial 3, the animal had to remember two prior reward events and their order-if RN, then nonreward, and ifNR, then reward (see also Yazawa & Fujita, 1984).…”

Learning about trial sequences disrupts the partial reinforcement extinction effect in classical conditioning.

“…To obtain such evidence, animals would need to be trained with a fixed sequence of N and R trials, and show that they respond more to the R trial in the sequence. Such evidence was provided by Yazawa and Fujita (1984) who described a runway experiment in which three groups of rats were given five trials, composed of one R trial and four N trials, every day. The first group was always reinforced on the first trial of the day (RNNNN), the second group was reinforced on the last trial (NNNNR), and the third group was reinforced on the third trial (NNRNN).…”