Popov, V., Marevic, I., Rummel, J., & Reder, L. M. (2019). Forgetting Is a Feature, Not a Bug: Intentionally Forgetting Some Things Helps Us Remember Others by Freeing Up Working Memory Resources. Psychological Science, 30(9), 1303–1317. https://doi.org/10.1177/0956797619859531
Popov et al tells a story about how item-method directed forgetting has an _aftereffect_ in addition to the primary effect. The aftereffect is such that to-be-remembered (TBR) items are remembered better when then were preceded with to-be-forgotten (TBF) items. Moreover, this effect is cumulative, such that memory is even better for items that are preceded by multiple TBF items. The interpretation is that TBR items consume cognitive “resources” to remember them, whereas TBF items to not. The lack of engagement of these resources for TBF items benefits following items by allowing those items to “consume” those additional resources, thus improving their encoding and retention. This is consistent with the resource-depletion-and-recovery theory of Lynne Reder (senior author).
They report 2 experiments, 1 is a reanalysis of existing data, the other is a new experiment. The new experwiment introduced controls in the deisgn to rule out alternative hypotheses: mainly rehersal or attentional borrowing (on TBF trials, one will selectiely rehearse previous TBR items). These used articulatory suppression for this. The results hold, suggesting that the aftereffect is indeed real, due to “richer encoding following TBF trials” rather than due to covert rehearsal of previous items.
Their analyses focus on Bayesian logistic regression, and they report Bayes factors (BF) for all of their analyhses. This allows them to report the amount of evidence in favor of their research hypothesis and the null hypothesis.
They included a computational model of the resource-depletion-and-recovery theory and used it to fit the empirical data. It did so … um … really well. R-sq values of 0.99. Seems way too good to be true. This was not a very convincing addition, as it seems to me that their way overfit their model parameters to get (practically) perfect model fits.
In the end, I’m convinced that the aftereffect is real. They had well-controlled experiments, and results that replicated across two experiments. However, I am not convinced by the resource-depletion-and-recovery account of the results. Everything here about “consuming resources” hinges on an indirect interpretation from better vs. worse subsequent memory performance. There is no actual measure, either behaviorally or neurally, in the moment that suggests resources are being SPARED on TBF trials.
Decent read, but I don’t leave convinced of their theoretical account.