Abstract: Reward can improve working memory performance. However, there has been controversy on whether reward can regulate the retrieval of working memory. Some studies have presented reward signals before the retrieval stage, with the results showing that reward can affect the retrieval phase during working memory. Klink et al. (2017) indicated that reward cues presented in the retrieval phase cannot affect working memory. However, this finding lacks support in the literature and needs to be further tested. The present study aims to explore the mechanism and effect of reward on the precision of working memory during retrieval.
A total of 24 participants (experiment 1) and 60 participants (experiment 2) were recruited for two experiments. Experiment 1 included two stages: association learning and memory test. The participants first established the connection between color and value through association learning. Thereafter, a working memory test was conducted. Reward cues were presented with the memory probe during the working memory test. Experiment 2 included four stages: pretest, association learning, posttest, and confidence assessment. Pre- and post-test tasks were the same, in which the participants performed two memory tests in sequence during the retrieval phase. Colored cues were not related to rewards in the pre-test but associated with rewards in the post-test. Confidence judgment required the participants to evaluate their confidence that they can infer the color of the second cue based on the color of the first cue. This undertaking aimed to test whether the participants have expectations for the order of cues.
Results of experiment 1 showed that the effect of reward was significant in the memory test stage. In particular, memory performance with high and low reward cues was better than that with no reward cues. Results of the post-test of experiment 2 showed a significant main effect of test order. Memory performance of the first test item was particularly better than that of the second test item. Moreover, there was a significant test order by serial position of the high-reward cue interaction effect. Memory performance of the first test item was better when the high-reward cue appeared before the low-reward cue. Memory performance of the second test item was better when the low reward cue appeared before the high-reward cue. The participants’ working memory capacity was differentiated according to their performance during the pre-test to investigate the individual differences of the reward effects. Results likewise indicated that reward did not affect the memory performance of the high-capacity group. For the low-capacity group, memory performance was affected by the test order, serial position of the high-reward cue, and their interaction. Thus, the effect of reward was more evident with the low-capacity group.
The present study was the first to observe that rewards can directly regulate working memory retrieval without affecting encoding and retention. In addition, the regulation mechanism of reward on working memory retrieval involved the redistribution of working memory resources. Such a reward regulation was affected by the test order and working memory capacity. Moreover, findings indicated that individuals can adjust cognitive strategies before or during learning according to the item value to promote working memory processing. They can also adjust retrieval precision after learning according to the value of information set by the test. This flexible working memory regulation mechanism plays an important role in promoting human cognitive activities and social adaptation.