The role of metacognitive uncertainty in the delayed retrieval shift of older adults

ABSTRACT

Strategic choice behavior of older adults in many skill acquisition tasks can be characterized as a delayed and/or incomplete shift to a more efficient retrieval-based strategy, even when older adults possess the requisite knowledge to use it successfully. The noun-pair lookup task (NPLT) requires verification of whether a centrally presented word pair matches one of a set of pairs displayed at the top of the screen. Because the pairings do not change, verification can be made from memory as the associations are learned. This study examines the role of metacognitive uncertainty in explaining older adults’ delayed retrieval shift in the NPLT. Older and younger adults participated in a NPLT with previously learned items and new items. For each trial, the look-up table was shown only if the scanning strategy was selected. Some participants were given a precue informing whether the item had been previously learned. Retrieval strategy choice was low for older adults but precueing increased its frequency. Older adults’ retrieval choices had minimal costs on NPLT accuracy, suggesting that the delayed retrieval shift can be exacerbated by metacognitive uncertainty that was reduced by precueing. The role of metacognitive uncertainty in older adults’ retrieval avoidance was supported by a robust item-level regression effect of retrospective confidence judgments during prelearning tests and later NPLT retrieval strategy choices for older adults.

KEYWORDS:

• Skill acquisition
• memory retrieval
• strategy choice
• associative learning
• metacognition
• aging

Disclosure statement

No potential conflict of interest was reported by the author(s).

Open practices statement

All of the data are available at https://osf.io/nt7cm/.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/13825585.2022.2112142

Notes

1. We determined in advance that a sample size of 160 participants was required in order to detect an Age X Condition interaction of f = 0.2 with 0.71 power.

2. As reported in the Methods section, the older groups with or without precueing differed somewhat in trials to criterion during prelearning despite random assignment. An anonymous reviewer commented that this difference could account for the observed NPLT precueing effects. We therefore conducted an additional analysis which includes cycles-to-criterion as a covariate while specifying interactions of the covariate with all experimental factors. Cycles to criterion was a significant predictor of retrieval behavior, and it also significantly interacted with Match, Block, Status X Block, Match X Block, and Status X Match X Block. However, controlling on cycles to criterion did not substantively change the inferences about precueing effects. The only changes were that the Condition X Age interaction became significant, and the Age X Match interaction was no longer significant. We concluded that this variable could not have produced the precueing benefits found for older adults.

3. Additional details are provided in Supplemental Materials.

4. See the Supplementary Materials for a GLM model conducted on the accuracy data including all conditions and blocks.

5. These confidence judgments should not be affected by precueing because they were collected prior to the precueing manipulation.

6. Although we chose to model cumulative retrieval over the first three blocks to capture early confidence-retrieval choice relationships, this criterion is a somewhat arbitrary. Therefore, we computed this model using different criteria: blocks 1 to 2, and blocks 1 to 4. The results did not substantively change with the exception that confidence was only a marginally significant predictor when using the expanded range of blocks 1 through 4.

7. We repeated this model for blocks 1 to 2, and blocks 1 to 4. The results did not substantively change with the exception that the three-way interaction term was marginally significant when using blocks 1 to 2.

Additional information

Funding

This publication was made possible with a Ruth L. Kirschstein National Research Service Award (NRSA) Institutional Research Training Grant from the National Institutes of Health (National Institute on Aging) Grant # 5T32AG000175-30.