Reward produced memories regulate memory-discrimination learning, extinction, and other forms of discrimination learning.

Abstract

In memory-discrimination learning, reward-produced memories are differentially rewarded such that they are the only stimuli available to support discriminative responding. Memory-discrimination learning was used in this study as follows: Reward-produced memories that were assumed to regulate instrumental performance in previously reported extinction and discrimination learning investigations were isolated and explicitly differentially reinforced (prior to a shift to extinction) in each of 4 runway investigations with rats. Results obtained here in the explicit discrimination learning stage and in the subsequent extinction stage were consistent with the prediction of the memory view and with prior discrimination learning and extinction findings. The memory interpretation was applied to memory-discrimination learning, to extinction, and to 2 other types of discrimination learning. It appears that a theory must use reward-produced memories to explain all 4 types of discrimination learning.

Reconciling reinforcement learning models with behavioral extinction and renewal: Implications for addiction, relapse, and problem gambling.

“…Thus, when the agent does not receive reward during the extinction condition, it takes longer to determine that the extinction condition is different and will require a different situation representation (a “splitting” of the state). This is very compatible with Capaldi and colleagues’ suggestion that extinction arises from a discrimination between acquisition and extinction conditions (Capaldi, 1957, 1958; Capaldi & Birmingham, 1998; Capaldi & Lynch, 1968).…”

“…Reinstatement by provision of unsignaled rewards (Pavlov, 1927;Rescorla, 2004;Robbins, 1990) can also be envisioned as a cued recall of the original state over other states. The presence of recent/available rewards is one of the most salient differences between states (Capaldi, 1957(Capaldi, , 1958Capaldi & Birmingham, 1998;Capaldi & Lynch, 1968).…”

Stimulus control of anticipatory responding in instrumental learning as revealed in serial learning tasks

“…Under the XNY and ZNN series, Position 3 cues were associated with both Y (XNY series) and N (ZNN series), so position cues cannot explain either interevent anticipation on Trial 2 or discriminative responding on Trial 3. But item learning can by assuming that, on any given trial, animals may remember each of the reward events that occurred on one or more prior trials (see, e.g., Capaldi, 1992Capaldi, , 1994Capaldi & Birmingham, 1998). Specifically, in the XNY series, the rats remembered X and X+N on Trials 2 and 3, respectively, whereas, in the ZNN series, the rats remembered Z and Z + N, respectively on Trials 2 and 3.…”

The role of us signal value in contingency, drug conditioning, and learned helplessness

“…For example, in , the number of food pellets delivered in a one-way runway apparatus on Trial I indicated whether reward or nonreward would occur on Trial 2. Results showed that discriminative responding on Trial 2 was much better the larger the difference in reward magnitude between the two trials .There is nowevidencethat reward-produced memories may serve an important signaling function in discrimination learning and extinction (Capaldi & Birmingham, 1998).…”

“…The signaling properties offood have also been studied in appetitive conditioning by Capaldi and his associates (Capaldi 1994;Capaldi, Alptekin, Miller, & Birmingham, 1997;Capaldi & Birmingham, 1998). For example, in , the number of food pellets delivered in a one-way runway apparatus on Trial I indicated whether reward or nonreward would occur on Trial 2.…”