Episodic and semantic feeling-of-knowing in aging: a systematic review and meta-analysis

Abstract

A complex pattern of preservation and deterioration in metacognition in aging is found, especially regarding predicting future memory retrieval (i.e., feeling-of-knowing, FOK). While semantic FOK (sFOK) is preserved with age, studies on episodic tasks (eFOK) produce equivocal findings. We present a meta-analysis of 20 studies on eFOK and sFOK, analyzing the difference in metacognitive sensitivity between 922 younger and 966 older adults, taking into account the difference in memory performance. The sFOK studies yielded no overall age effect (8 effects, g = -0.10 [-0.29, 0.10]). However, we found a reliable age-group difference on eFOK (22 effects, g = 0.53 [0.28, 0.78]), which was moderated when considering recognition performance. Moreover, using aggregated data of 134 young and 235 older adults from published and unpublished studies from our lab, we investigated memory performance as an explanation of the eFOK deficit. We show that older adults are less metacognitively sensitive than younger adults for eFOKs which is, at least partly, due to the age-related memory decline. We highlight two non-exclusive explanations: a recollection deficit at play in the first and second order tasks, and a confound between first order performance and the measure used to assess metacognitive sensitivity.

Figure 1

Basic episodic (top) and semantic (bottom) FOK paradigms. FOKs are made after a cued-recall attempt in both cases and are made on a “yes/no” or a Likert-scale. The recognition can be an “old/new” task or a multiple-alternative forced-choice. First-order performance is usually the proportion of correct recognition. Metacognitive sensitivity is assessed by comparing FOKs with memory performance in the recognition phase.

Figure 2

Flow chart of the steps for identification and selection of records included in the systematic review and meta-analysis.

Figure 3

Forrest plot of the effect of eFOK deficit in OA. Confidence interval of the overall estimated effect does not overlap with 0.

Figure 4

Forrest plot showing an absence of sFOK deficit in OA. Confidence interval of the overall estimated effect does overlap with 0.

Figure 5

Funnel plot centered on the overall effect size (vertical line) for eFOK model (A) and sFOK model (B). The white areas are the 95% confidence intervals of the overall effect sizes. Points in the gray areas are outliers.

Figure 6

(A) Forrest plot of the effect of eFOK deficit in OA. Confidence interval of the overall estimated effect does not overlap with 0. (B) Forrest plot of the effect of recall deficit in OA. Confidence interval of the overall estimated effect does not overlap with 0. (C) Forrest plot of the effect of eFOK deficit in OA for half of OA with the best recall performance and half of YA with the worst recall performance. Confidence interval of the overall estimated effect does not overlap with 0. (D) Forrest plot of the effect of eFOK deficit in OA for half of OA with the worst recall performance and half of YA with the best recall performance. Confidence interval of the overall estimated effect does not overlap with 0.

Figure 7

(A) Forrest plot of the effect of eFOK deficit in OA using the Hamann coefficient. Confidence interval of the overall estimated effect does not overlap with 0. (B) Forrest plot of the effect of eFOK deficit in OA using type-II d’. Confidence interval of the overall estimated effect does not overlap with 0.

Figure 8

(A) Forrest plot of the effect of eFOK deficit in OA for half of OA with the best recognition performance and half of YA with the worst recognition performance. Confidence interval of the overall estimated effect overlaps with 0. (B) Forrest plot of the effect of eFOK deficit in OA for half of OA with the worst recognition performance and half of YA with the best recognition performance. The confidence interval of the overall estimated effect does not overlap with 0.