Do older adults know when their memory is right?
Older adults remembered less than younger adults but judged the accuracy of their own memories just as well, and both groups' confidence was more informative for remembered word pairs than for single words.
Source
Evidence for Age-Equivalent and Task-Dissociative Metacognition in the Memory Domain
Study at a glance
- Design
- Human experiment — 2 (age group: younger vs older, between) x 2 (test type: item vs associative recognition, within) mixed design with trial-by-trial confidence ratings across four study-test blocks
- N
- N=49 · 28 younger and 29 older adults recruited; 4 younger and 5 older excluded for near-chance discrimination, leaving 24 younger and 25 older adults
- Population
- US undergraduates aged 18-21 and community-dwelling older adults aged 64-85
- Outcome
- Metacognitive efficiency (meta-d'/d'), metacognitive sensitivity (meta-d') and recognition sensitivity (d') for single words and word pairs
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Younger and older adults studied lists of 30 unrelated word pairs, then took two recognition tests: one asking whether single words were old or new, and one asking whether pairs were intact or rearranged. After every answer they rated their confidence from 1 to 6. The researchers used a signal-detection measure, meta-d'/d', which asks how well confidence separates correct from incorrect answers after adjusting for how good memory itself was.
What they found
Older adults had lower recognition sensitivity than younger adults, but metacognitive efficiency did not differ by age, and a Bayes factor of about 4.75 favoured no age difference. Both groups were close to optimal efficiency. Metacognitive efficiency was higher for word pairs than for single words (a small-to-moderate effect), even though memory accuracy did not differ between the two test types. The classic gamma correlation measure showed no age or task effects, so the task difference only emerged with signal-detection measures.
The limits
What it doesn't show
The sample was small and the final groups excluded 4 younger and 5 older adults whose memory was near chance, which may have removed the older adults with the weakest memory and flattered the age comparison. The expected age-by-task interaction in memory (a larger associative deficit for older adults) was not significant, so the study cannot say whether older adults are aware of an associative memory deficit. The explanation that encoding strategies such as linking images boost pair confidence was not measured, and results may not generalise to other kinds of metacognitive judgment (like judgments of learning) or to recall tests. The power analysis was run after data collection, which the authors acknowledge.
Key terms
- Metacognition
- Thinking about one's own thinking, including monitoring how accurate one's memories or answers are.
- d' (d-prime)
- A signal-detection measure of how well someone discriminates old from new items, independent of their bias to say 'yes'.
- meta-d'/d' (metacognitive efficiency)
- How well confidence ratings distinguish correct from incorrect answers relative to actual task performance; 1 means confidence captures all the information available in performance.
- Associative deficit hypothesis
- The idea that older adults have particular trouble binding pieces of information together, so their memory for pairs declines more than memory for single items.
- Bayes factor (BF01)
- A ratio showing how much more likely the data are under the null hypothesis than the alternative; values above 3 are usually taken as moderate evidence for no effect.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What was the main age finding for metacognitive efficiency?
Common questions
Why not just compare average confidence between age groups?
Average confidence reflects bias (older adults here were more confident overall), not accuracy of monitoring; meta-d'/d' asks whether confidence tracks correctness trial by trial while adjusting for memory level.
If older adults remember less, how can their metacognition be equal?
Metacognitive efficiency is measured relative to performance, so someone with weaker memory can still be just as good at knowing which of their answers are right and wrong.
Why might confidence be more accurate for word pairs?
The authors suggest that linking two words during study, for example with a sentence or an image, gives people extra internal cues they can use to judge whether a pair is familiar, though they did not measure strategy use.
More on Metacognition