Forgetting is a Feature, not a Bug: Intentionally Forgetting Some Things Helps Us Remember Others by Freeing up Working Memory Resources

Abstract

We used an item-method directed forgetting paradigm to test whether instructions to forget or to remember one item in a list affects memory for the subsequent item in that list. In two experiments, we found that free and cued recall were higher when a word-pair was preceded during study by a to-be-forgotten (TBF) word pair. This effect was cumulative – performance was higher when more of the preceding items during study were TBF. It also interacted with lag between study items – the effect decreased as the lag between the current and a prior item increased. Experiment 2 used a dual-task paradigm in which we suppressed either verbal rehearsal or attentional refreshing during encoding. We found that neither task removed the effect, thus the advantage from previous TBF items could not be due to rehearsal or attentional borrowing. We propose that storing items in long-term memory depletes a limited pool of resources that recovers over time, and that TBF items deplete fewer resources, leaving more available for storing subsequent items. A computational model implementing the theory provided excellent fits to the data.

Frequency Effects on Memory: A Resource-Limited Theory

“…Popov et al (2018) did not find support for that explanation in a dual task study designed to test that explanation. Participants tried to learn paired-associates in one of four conditions -a control single-task condition, equivalent to Marevic et al (2017), a rehearsal suppression dual task condition, a divided attention dual task condition and a combined rehearsal suppression plus divided attention dual task condition.…”

“…The typical result is that the presence of LF items on the list hurts memory for HF items, and the presence of HF items on the list helps memory for LF items (i.e., performance for HF pure > HF mixed ~ LF mixed > LF pure; see Figure 17). These results cannot be entirely explained by differential rehearsal, because accounting for the number and recency of rehearsals (Ward et al, 2003) or suppressing rehearsal altogether (Popov et al, 2018) does not remove list-composition effects. SAC suggests that fewer resources are depleted while storing HF items, which leaves more resources available for storing LF items in mixed rather than pure LF lists.…”

“…We tested this prediction with a reanalysis of a recent published study on item-based DF (Marevic, Arnold, & Rummel, 2017) and a follow up dual-task study (Popov, Marevic, Rummel & Reder, 2018). The full analysis and results are described in Popov et al (2018), but we will briefly summarize those results.…”

“…The full analysis and results are described in Popov et al (2018), but we will briefly summarize those results. The analysis of preceding study items showed that cued-recall and free-recall paired-associate memory for items presented on study trial k was better when study trial k-1 was TBF, rather than TBR.…”

“…The most straightforward variable to analyze is word frequency, but the effects are likely to occur with any variable that affects how many resources are depleted in processing the preceding study items. Of the nine studies, five studies used word frequency as a factor (Healey & Kahana, 2016;Cox et al, 2018;Reder et al, 2002;Ward et al, 2003), two studies manipulated how many times a studied item had already been presented (Buchler et al, 2008;Aue et al, 2017) and two directed-forgetting studies that manipulated whether each item should be remembered or forgotten (Marevic et al, 2017;Popov et al, 2018).…”