Updating the self-appraisal of one’s cognitive performance with seven days of repeated exposure: From test-naïve to experienced

Abstract

Objective:

Self-appraisal of cognitive performance, a potentially useful marker of brain functioning, is typically assessed at a single time point where testees are naïve to what constitutes “good” or “bad” performance. Here, we determine whether familiarizing individuals with self-appraisal with daily memory testing for 7 days provides a more accurate estimate of cognitive functioning and mood.

Method:

225 participants (mean age ± standard deviation: 74.1±8.3 years; 66% female; median education 16.0 years) completed the online Boston Remote Assessment for NeuroCognitive Health (BRANCH), which includes two associative memory tasks, for seven consecutive days. Each day, participants self-appraised their performance. At baseline, they completed various cognitive and mood measures. We computed Pearson’s correlations between task performance and self-appraisal on days 1 and 7 and used linear models to examine the relationship between self-appraisal scores and clinical measures.

Results:

Accuracy (day 1: 0.44±0.12; day 7: 0.81±0.16) and self-appraisal (day 1: 0.36±0.15; day 7: 0.70±0.21) increased, as did the association between accuracy and self-appraisal (day 1: correlation coefficient (r)=0.22, 95% confidence interval (95%CI)=[0.09, 0.34], p=.001; day 7: r=0.69, 95%CI=[0.62, 0.76], p<.001). Self-appraisal scores on day 7, but not day 1, showed significant relationships with in-clinic measures.

Conclusions:

Repeated remote cognitive assessments may help elucidate individuals’ capacities to refine their self-perception of cognitive performance during multi-day learning. The weak association between accuracy and test-naïve self-appraisal warrants caution about using this metric cross-sectionally. Experienced self-appraisal could be especially relevant at the early stages of neurodegenerative diseases when subtle learning difficulties emerge and could improve our capacity to detect early meta-cognitive changes.

Introduction

Self-perceived cognitive functioning represents a person’s general perception of their cognitive abilities. It is common for older adults to self-perceive changes in their cognitive functioning (Rabin et al., 2023), and these changes may or may not be an early sign of underlying neurodegenerative disease (Jessen et al., 2020). Self-perceived cognitive functioning is commonly assessed using questionnaires that probe various aspects of everyday functioning, including a person’s experience of their memory, language, organizational, and everyday functioning abilities (Clare et al., 2011; Rickenbach et al., 2015).

Self-appraisal encompasses a person’s rating of their performance on a test they just completed. Various studies have focused on metacognitive monitoring, which can be prospective or retrospective in nature and may represent an individual’s judgment of how well they learned information they have just learned or their confidence that they may later recall it (Hertzog & Dunlosky, 2011; Siegel & Castel, 2019). On the other hand, retrospective confidence ratings, which ask individuals to judge how well they recalled information, have also been shown to predict future performance (Putnam et al., 2022). Comparing retrospective self-appraisal to objective performance on a test allows for making inferences about a person’s self-awareness of their cognitive performance on a given test. If someone performs well on a test but reports doing poorly, this may reflect a hyperawareness of cognitive issues (Cacciamani et al., 2021). At the same time, a poor performance perceived as good could indicate a lowered awareness of one’s cognitive impairment. While the individual’s cognitive status contributes to their ability to judge their performance, many other factors may also come into play. For example, depressive symptoms have been linked to lower self-rated memory performance (O’Shea et al., 2016).

While it has been documented that older adults can gauge their performance on memory tasks with a certain precision (Zakrzewski et al., 2021), it may be challenging to gauge a good or a poor performance on a test they have never taken. With repeated exposure to the same task and stimuli, a person’s self-appraisal of their performance might change as their understanding of what is expected of them in the task improves. It is well-established that objective performance increases with repeated exposure (Hausknecht et al., 2007). A study by Koriat et al. (2002) demonstrated that repeated learning of a word list not only improved the objective learning curve but also enhanced the subjective learning curve, gradually increasing participants’ accuracy in assessing their memory performance. This suggests that repeated learning plays a crucial role in fine-tuning self-perception of cognitive performance.

In the multiday Boston Remote Assessment for Neurocognitive Health (BRANCH), older adults complete the exact same remote cognitive tasks every day for a week. Diminished practice effects (i.e., reduced performance improvements on repeated testing), particularly when retesting the same stimuli, represent a decrement in memory consolidation that may be a hallmark of Alzheimer’s disease (Papp et al., 2024; Samaroo et al., 2020). Previous studies have shown that performance on BRANCH improves over time (Jutten et al., 2023; Papp et al., 2024; Weizenbaum et al., 2024) and that differences in learning curves are associated with cognitive performance as measured using in-clinic pencil-and-paper neuropsychological tests (Weizenbaum et al., 2024). BRANCH also includes measures of self-appraisal of performance. With multiday BRANCH, we can investigate whether we can observe changes in test performance and self-appraisal over repeated exposure.

We asked participants to rate their performance on BRANCH following each daily assessment. We aimed to investigate how self-appraisal of cognitive performance changes after repeated exposure to the same stimuli within the multiday BRANCH. We hypothesized that self-appraisal after the first exposure would be less strongly associated with actual performance than self-appraisal after a week of daily repeated exposure because individuals learn to understand what a good performance is from taking the same tests multiple times. Further, we hypothesized that higher self-appraisal would be associated with better cognitive functioning, better self-perceived cognitive functioning, and fewer depressive symptoms. We also expected self-appraisal after repeated exposure to be more strongly related to the other measures than on the first day, as participants had the benefit of multiple opportunities to assess their performance, allowing them to refine their self-perception based on their multiple past performances.

Methods

Participants

Participants were recruited through the Harvard Aging Brain Study (P01AG036694), the Instrumental Activities of Daily Living Study (R01AG053184 and R01AG067021), and the Subjective Cognitive Decline Study (R01AG058825), all of which focused on aging and preclinical Alzheimer’s disease. In addition to the three parent studies’ inclusion criteria, participants were selected for the present study if they were cognitively normal (Clinical Dementia Rating [CDR] global score = 0, Mini-Mental State Examination [MMSE] score >25, and Logical Memory Delayed Recall [LMDR] score above education-adjusted cutoffs). Participants who had a recent history of alcohol or drug abuse, severe head trauma, or a current unstable medical condition or primary psychiatric diagnosis were excluded. A few participants (n = 11, 5%) received a CDR global score of 0.5 but were considered cognitively normal in a multidisciplinary consensus meeting based on their overall clinical presentation and performance on neuropsychological testing. These participants were also included in the study. All participants gave written informed consent before any study procedures. The institutional review board of Massachusetts General Brigham approved this study.

Materials

Boston Remote Assessment of Cognitive Health (BRANCH)

The Boston Remote Assessment of Cognitive Health (BRANCH) comprises two associative memory tasks and a digit–symbol substitution task. The two associative memory tasks are (1) the ‘Face–Name’ (FNAME) test, an adaptation of the Face–Name Associative Memory Exam (Rentz et al., 2011), in which twenty face-name pairs are shown with participants asked to identify the learned faces following a delay; and (2) the ‘Groceries Prices’ test, where participants needed to associate visually represented grocery items with prices ranging from $1.09 to $12.99 (Castel, 2005). BRANCH is described in more detail elsewhere (Papp et al., 2024; Weizenbaum et al., 2024). As described in Weizenbaum et al. (2024), participants attested to completing tasks independently without recording stimuli or responses at the beginning of each session. If participants failed to respond in the allotted time, a pop-up message would appear asking participants to confirm that they were ready to proceed with the test. Study coordinators were available to support participants via phone and email as needed.

In the present study, we focus exclusively on the composite performance accuracy of the two associative memory tasks, as they assess the same cognitive process. Performance accuracy was captured by dividing the number of correct responses by the total number of responses. The resulting score fell on a scale from 0 to 1, with higher scores representing better performance. After completion of each task, participants were prompted, but not required, to “please rate [their] performance on the test today,” using a visual analog scale ranging from 0 to 100 (with higher scores indicating better self-perceived performance). We will refer to this as their ‘self-appraisal.’ The self-appraisal ratings were rescaled to the same scale as performance accuracy, ranging from 0 to 1, with higher scores indicating better performance and self-appraisal.

Standard clinical and cognitive measures

At baseline, participants completed the Preclinical Alzheimer’s Cognitive Composite–5 (PACC5; Donohue et al., 2014; Papp et al., 2017). PACC5 performance is rated as a Z-score, calculated as the mean score of five measures (the Mini-Mental State Examination (Folstein et al., 1975), Wechsler Memory Scale–Revised Logical Memory Delayed Recall (Wechsler & Stone, 1987), Digit-Symbol Coding Test (Wechsler, 1981), the free plus total recall of the Free and Cued Selective Reminding Test (Grober et al., 2009), and category fluency). The PACC5 Z-score has a mean of 0 and a standard deviation of 1 in a cognitively unimpaired sample. Higher scores represent better global cognitive performance. The PACC5 and its components have shown favorable reliability and validity (Donohue et al., 2014; Papp et al., 2017).

The Clinical Dementia Rating (CDR), a semi-structured interview to determine cognitive and functional impairment (Hughes et al., 1982), was administered to participants and their study partners. Study partners were the participants’ spouses, children, friends, or family members. Based on information from both sources, scores between 0 (no impairment) and 3 (severe impairment) are given in various areas (“boxes,” including memory, judgment and problem-solving, home and hobbies, and personal care). Scores from the separate boxes are summed to a “sum of boxes” score ranging from 0–18). Further, a global CDR score ranging from 0 to 3 is calculated using a previously established scoring logic (Morris, 1993); the global score scale corresponds to the individual box scores.

Participants and their study partners also completed the Cognitive Function Index (CFI), a 15-item questionnaire about self-perceived cognitive functioning (Amariglio et al., 2015). Items are scored between 0, 0.5, or 1 and are summed for a total score (range 0–15). Higher scores indicate greater subjective cognitive complaints. The CFI has been shown to have a strong internal reliability, sensitivity to change over time, and discriminative abilities (Amariglio et al., 2015).

Finally, as a measure of depressive symptoms, we administered the 30-item Geriatric Depression Scale (GDS; Yesavage et al., 1982). Each item is rated 0 if the response does not indicate depressive symptoms and 1 if it does. Total scores thus potentially range from 0–30, with higher scores representing more depressive symptoms. The GDS has high internal consistency, strong test-retest reliability, and good criterion and construct validity (Yesavage et al., 1982).

Procedures

Participants completed the clinical measures at an initial in-clinic baseline visit and subsequently completed the BRANCH remotely and unsupervised on their own device (personal computer or mobile device) once daily for seven consecutive days. Each BRANCH session included the same stimuli and required approximately 10 to 15 minutes to complete.

Statistical analysis

All analyses were run in R version 4.3.2. We created composites of the objective performance (i.e., accuracy) and the self-appraisal metrics for the FNAME and the Groceries Prices tasks by averaging both tasks’ metrics. These composites serve as the primary outcomes of this study.

Pearson’s correlations were calculated for the composite accuracy and self-appraisal on day 1 and compared to the correlations on day 7. Linear models were used to examine further the relationship between the day 1 (“test-naïve”) and day 7 (“experienced”) composite self-appraisal scores and the PACC5, CDR, CFI, and GDS. To account for the potentially modifying effects of baseline demographic characteristics, these models were adjusted for age, sex, education, as well as objective performance on the task. We repeated these analyses for the FNAME and Groceries Prices tasks separately; these results are shown in Supplemental Table 1. As self-appraisal ratings were not a forced response, we excluded day 7 self-appraisals of precisely 50 (the default on the visual analog scale; occurred in n = 9 participants [4%]) in sensitivity analyses, reported in Supplemental Table 2. There were no outliers. Analyses were run with the available data.

Last, we calculated the difference between task accuracy and self-appraisal on days 1 and 7 by subtracting accuracy from self-appraisal and subsequently classified participants’ relative estimations over the seven-day testing period into the following groups: (a) consistent underestimation (i.e., participants self-appraised their performance as worse than their actual accuracy on day 1 and on day 7); (b) consistent overestimation (i.e., participants self-appraised their performance as better than their actual accuracy on day 1 and on day 7); (c) variable estimation (i.e., participants either self-appraised their performance as better or as worse than their actual accuracy on day 1 and vice-versa on day 7). We report descriptive statistics for these groups. Group differences were computed using t-tests or chi-square tests, as appropriate.

Transparency and Openness

The data and analyses used for this publication can be made available upon reasonable request from the corresponding author. This study was not preregistered.

Results

We included 225 cognitively unimpaired participants (74.1 ± 8.3 years old; 66% female). Baseline characteristics are displayed in Table 1.

Table 1.

Participant Characteristics.

Variable	Whole sample
N	225
Age in years	74.1 ± 8.3
Female sex, N (%)	148 (65.8)
Education in years, M (IQR)	16.0 (15–18)
Race, N (%)
Black	23 (10)
Asian	7 (3)
Native American/Alaskan Native	1 (1)
White	194 (86)
Ethnicity, N (%)
Hispanic	3 (1)
Non-Hispanic	222 (99)
PACC5	0.14 ± 0.72
CDR
Global, M (IQR)	0 (0–0)
Sum of Boxes	0.14 ± 0.33
CFI
Participant	2.46 ± 2.45
Study partner	1.49 ± 2.32
GDS	4.56 ± 4.13
BRANCH objective performance
Day 1	0.44 ± 0.12
Day 7	0.81 ± 0.16
BRANCH self-appraisal
Day 1	0.36 ± 0.15
Day 7	0.70 ± 0.21

Note: All data displayed as mean ± standard deviation, except otherwise noted.

Abbreviations: BRANCH, Boston Remote Assessment for NeuroCognitive Health; CDR, Clinical Dementia Rating; CFI, Cognitive Function Index; GDS, Geriatric Depression Scale; IQR, interquartile range; M, median; PACC5, Preclinical Alzheimer’s Cognitive Composite–5.

Day 1: Test-naïve self-appraisal

On the first day of testing, the performance accuracy of the FNAME and Groceries Prices tests’ composite was (mean =) 0.44 ± (standard deviation =) 0.12, while the participants’ self-appraisal was 0.36 ± 0.15. Performance accuracy and self-appraisal had a weak association (correlation coefficient (r) = 0.22, 95% confidence interval (95%CI) = [0.09, 0.34], p = .001). Linear models that were adjusted for baseline demographics and initial task accuracy revealed that the self-appraisal scores on day 1 were associated only with the GDS (estimate (B) = −0.005, 95%CI = [−0.010, −0.0003], p = .039) so that a lower GDS was associated with higher self-appraisal. None of the other clinical measures were associated with test-naïve self-appraisal. All associations are displayed in Table 2.

Table 2.

Linear Models Showing the Relationship Between Self-Appraisal and Various Clinical Measures.

Measure	Day 1		Day 7
	B [95%CI]	P-value	B [95%CI]	P-value
PACC5	−0.022 [−0.053, 0.010]	.175	0.058 [0.020, 0.096]	.003
CDR-SB	−0.011 [−0.071, 0.048]	.703	−0.079 [−0.151, −0.007]	.032
CFI  Participant  Study partner1	−0.003 [−0.011, 0.005] −0.002 [−0.012, 0.006]	.478 .492	−0.012 [−0.021, −0.002] −0.007 [−0.018, 0.004]	.017 .195
GDS	−0.005 [−0.010, −0.0003]	.039	−0.008 [−0.014, −0.002]	.008

Note: All models are adjusted for baseline age, sex, years of education, and objective performance on day 1. Effects with p < .05 are displayed in bold.

¹Data missing for n = 5 (2%).

Abbreviations: B, estimate; CDR-SB, Clinical Dementia Rating Sum of Boxes; CFI, Cognitive Fit Index; GDS, Geriatric Depression Scale; PACC, Preclinical Alzheimer’s Cognitive Composite; 95%CI, 95% confidence interval.

Day 7: Experienced self-appraisal after repeated exposure

Accuracy scores improved quickly over the first three days of testing, after which the improvement tapered off until reaching 0.81 ± 0.16 on day 7. Self-appraisal also increased over repeated exposures up to 0.70 ± 0.21 on day 7. Figure 1 displays the change in objective performance and self-appraisal over the seven days of BRANCH. By day 7, objective performance and self-appraisal showed a substantially stronger association (r = 0.69, 95%CI = [0.62, 0.76], p < .001) than on day 1, as seen in Figure 2.

Figure 1.

Change in objective performance and self-appraisal over seven days of testing. Objective performance is shown in orange, and participant self-appraisal is shown in purple. For both metrics, we averaged the scores across FNAME and Groceries.

Figure 2.

Relationship between objective performance and self-appraisal on the first (left) and last day of testing (right).

Self-appraisal scores on day 7 showed significant relationships with all in-clinic measures except for the study partner-reported CFI (Table 2). Of note, the direction of the relationship between self-appraisal and PACC5 flipped from a (non-significant) negative association on day 1 to a positive association on day 7. When separating the two associative learning tasks, we noted that the results for test-naïve self-appraisal were identical, while the association between GDS and experienced Groceries self-appraisal was marginally stronger than between GDS and experienced FNAME self-appraisal (see Supplemental Table 1). The overall results remained the same when removing participants who may not have rated their performance on the last day of testing (see Supplementary Table 2).

Self-appraisal relative to objective performance

Out of 217 participants who completed all days of testing, most consistently underestimated their performance (i.e., self-rated their performance as lower relative to their objective performance) on day 1 and day 7 (n = 129, 59.4%; Figure 3). A smaller proportion consistently overestimated their performance (n = 22, 10.1%). Others varied in their estimation (n = 66, 30.4%), most initially overestimating their performance and underestimating it by the end of testing (n = 47, 21.7%).

Figure 3.

Discrepancy between participants’ objective performance and self-appraisal on days 1 (Y-axis) and 7 (X-axis). The top right quadrant (in green) shows consistent overestimation (n = 22, triangles), while the bottom left quadrant (in purple) shows consistent underestimation (n = 128, circles). The other quadrants (in yellow) show variable estimation (n = 67, squares).

Objective performance was lower on day 1 and day 7 among participants who consistently overestimated their performance compared to participants in the other two groups. Conversely, those who consistently underestimated their performance had the highest accuracies of all participants on both day 1 and day 7. Participants who consistently overestimated their performance had lower baseline PACC5 scores than participants who consistently underestimated their performance (Table 3). The groups did not differ in any other demographic aspect. All baseline descriptive statistics by group are displayed in Table 3.

Table 3.

Descriptive Statistics of the Three Under and Overestimation Groups.

Variable	Consistently underestimated a	Consistently overestimated b	Variably under/overestimated c	Group differences
N (%) 1	129 (59.0)	22 (10.1)	67 (30.9)	—
Age in years	74.1 ± 8.6	74.5 ± 8.3	73.9 ± 7.8	n.d.
Female sex, N (%)	83 (64.8)	16 (72.7)	45 (67.2)	n.d.
Education in years, M (IQR)	18 (15–18)	17 (14–18)	16 (14–18)	n.d.
PACC5	0.24 ± 0.71	−0.22 ± 0.76	0.14 ± 0.69	a > b
CDR sum of boxes	0.16 ± 0.34	0.05 ± 0.15	0.14 ± 0.36	n.d.
CFI  Participant  Study Partner	2.70 ± 2.73 1.67 ± 2.49	2.55 ± 2.09 1.34 ± 1.83	2.01 ± 1.97 1.28 ± 2.23	n.d. n.d.
GDS	5.11 ± 4.37	4.36 ± 4.66	3.63 ± 3.36	n.d.
BRANCH objective performance  Day 1  Day 7	0.47 ± 0.13 0.84 ± 0.15	0.39 ± 0.10 0.72 ± 0.20	0.41 ± 0.09 0.81 ± 0.14	a > b, c a, c > b
BRANCH self-appraisal  Day 1  Day 7	0.30 ± 0.13 0.66 ± 0.21	0.53 ± 0.09 0.81 ± 0.19	0.43 ± 0.13 0.73 ± 0.18	b > c > a b > a

Note: All data displayed as mean ± standard deviation, except as noted otherwise.

¹data missing for n = 8 (3.6%).

Abbreviations: BRANCH, Boston Remote Assessment for NeuroCognitive Health; CDR, Clinical Dementia Rating; CFI, Cognitive Function Index; GDS, Geriatric Depression Scale; IQR, interquartile range; M, median; n.d., no difference; PACC5, Preclinical Alzheimer’s Cognitive Composite–5.

Discussion

In this study, we investigated participants’ self-appraisal of memory performance on a remote cognitive battery over seven consecutive days of testing. Specifically, we were interested in how self-appraisal relates to objective performance before and after repeated assessments with the same test material. We noted that the participants’ objective performance and subjective experience improved with repeated exposures over seven days. Initial, test-naïve self-appraisal was only weakly related to actual performance, but repeated exposure to the same stimuli fortified this relationship by the last day. Experienced but not test-naïve self-appraisal related to several clinical measures, including self-reported cognitive complaints and mood.

Practice effects are a well-described phenomenon in cognitive testing (Duff et al., 2024; Goldberg et al., 2015; Jutten et al., 2020; Young et al., 2023) and have even been demonstrated over relatively extended follow-up durations (e.g., six months). With multiday BRANCH, participants were exposed to the same memory test with the same stimuli every day for a week. In line with the practice effects literature, participants performed exponentially better over the first two or three days of testing, with the increase in performance subsequently slowing down somewhat at the group level over the remaining three to four days. Previous studies with BRANCH have shown how these brief, daily assessments may provide further insight into learning in the context of healthy aging, as well as in Alzheimer’s disease (Jutten et al., 2023; Papp et al., 2021; Samaroo et al., 2020; Weizenbaum et al., 2024). Still, it is not as well-understood how self-appraisal changes over the course of such a repeated testing paradigm.

Several studies have asked people about their perceived performance on specific cognitive and functional tests (Eslinger et al., 2005; Huber & Fulton, 2024; Massimo et al., 2013; Mora & Suchy, 2023; Scherling et al., 2016). For example, poorer self-appraisal of performance on an executive functioning task was predictive of future decline in individuals with mild cognitive impairment (Scherling et al., 2016). These studies were cross-sectional and performed in individuals with a degree of cognitive impairment. It is known that self-awareness of cognitive ability is diminished in the presence of impairment, a state termed anosognosia. However, the findings reported in those studies likely do not translate to the general population of cognitively unimpaired older adults.

Unfamiliarity with the task might cause a person to be unaware of what constitutes a good or a poor performance on that test, making it harder for them to estimate their ability. Even though someone may have a general idea of how they did (e.g., “I remembered about half the faces”), it is the opportunity to compare the self-perceived success on one assessment with that of a previous assessment that helps contextualize one’s performance (e.g., “Today I was able to remember three more faces than yesterday”). With repeated exposure to the same test, one can learn better what is expected of them and revise their self-perceived performance. In our study, test-naïve self-appraisal was somewhat lower than objective performance, but over time, participants rated their performance as increasingly better. The increase in self-appraisal appeared to follow a linear trajectory, i.e., self-appraisal went up gradually with approximately the same amount each day. Thus, an individual’s ability to update their perception of their performance appears to lag behind the increase in their actual performance: the distance between self-appraisal and accuracy grew over the first few days of the week and gradually shrunk by the end of the seven days of testing. Consequently, we postulate that self-appraisal provides essential additional information, mainly when probed after familiarization with the task. The question remains: What does experienced self-appraisal of cognitive performance reflect, and how does it differ from test-naïve self-appraisal?

Test-naïve self-appraisal appeared random: it was not associated with objective performance or most clinical measures taken at baseline, such as study partner- and self-perceived cognitive functioning. We only observed a subtle relationship between test-naïve self-appraisal and self-reported mood, where those who reported more depressive symptoms self-appraised their performance as being worse. This is in line with the well-established finding that individuals with more depressive symptoms will report lower self-perceived cognitive functioning (O’Shea et al., 2016). At the same time, depressive symptoms may also impact objective performance, although participants included in this study did not endorse high levels of depressive symptoms. The impact of mood complaints on objective performance is thus likely minimal.

By the end of the week, participants had completed the same test seven times, giving them an experienced self-appraisal of their performance. We observed that on the last day of testing, self-appraisal was more strongly associated with objective performance and that it was also related to cognitive and functional status, as well as self-reported cognitive functioning and mood. The relationships with cognitive and functional measures corroborate that the self-appraisal reflects cognition. Those who performed better on the in-clinic cognitive composite thought they performed better on the BRANCH tasks by the seventh day. Similarly, those who consider their global self-perceived cognitive functioning to be worse self-rated their performance on the test as worse after seven days. Interestingly, these associations were not apparent on the first day of testing. Perhaps the test-naïve self-appraisal that individuals made after the first exposure to the test without a frame of reference for a good performance is more strongly related to factors we did not investigate here, such as personality or neuropsychiatric symptoms (Martyr et al., 2019). Put together; these findings suggest that self-appraisal, particularly after repeated exposure to the same test, is more than simply a proxy for objective cognitive functioning; it represents a person’s ability to update their understanding of their cognitive functioning with repeated exposure, as well as their mood and potentially other factors not investigated here, such as implicit theories the participant may have (e.g., “I am getting better because I have done this test before”), familiarity with being tested, or the expectation of being asked to self-reflect on their performance.

When we compared self-appraisal and objective performance, we noted that most participants consistently reported their performance to be worse than it was. Others have previously shown that older individuals with or without cognitive impairment often underestimate their cognitive performance, as compared with their study partners (Amariglio et al., 2021; Cacciamani et al., 2017) or with objective performance (Scherling et al., 2016; Vannini et al., 2017). In line with this previous work, we found that those who consistently underestimated their performance had better cognitive functioning as measured using an in-clinic cognitive composite, as well as better BRANCH performance, than those who overestimated their performance or those who estimated their performance variably. Underestimators did not differ from the other participants in other aspects, such as age, sex, education, mood, or functional status. According to the cue-utilization theory, individuals judge their performance by relying on intrinsic, extrinsic, and mnemonic cues rather than objective accuracy (Koriat, 1997). Factors such as the perceived difficulty of the task (intrinsic), repetition frequency (extrinsic), and ease of retrieval of the material (mnemonic) may have contributed to the self-appraisals found here. Future work will have to elucidate further the value of informed self-appraisal in biomarker-confirmed neurodegenerative diseases such as Alzheimer’s disease, including in individuals with cognitive impairments, to determine whether it may be used to identify individuals at risk for cognitive decline. Future studies might also include a control group assessing repeated self-appraisal without repeated testing to investigate how self-appraisal changes on its own.

Constraints on generality

Most participants were highly educated and predominantly White and female, thus limiting the generalizability of these findings to other populations. As multiday BRANCH is rolled out in other cohorts, we aim to include minoritized individuals to verify whether these findings hold in the broader population. Using a slider scale to capture self-appraisal somewhat limited our qualitative interpretation of the self-appraisal data, as a given number on the sliding scale can represent different things to different people. We also could not determine with total confidence that participants responded to the self-appraisal question, as the slider’s default was set to 50 (the middle of the scale), and a response was not forced. We tried to mitigate these limitations by avoiding the interpretation of given self-appraisals as “bad,” “average,” or “good” and by running sensitivity analyses excluding participants who rated their performance as exactly 50/100 on the last day. Our reasoning was that on the first day, a rating of 50 might indicate uncertainty about the objective performance, whereas a rating of 50 on the last day more likely reflects non-response. The results presented here did not change when excluding participants who likely did not respond to the self-appraisal questions on the last day of testing. We investigated relative estimations of cognitive performance by considering any difference greater than zero between objective performance and self-appraisal to indicate over or underestimation, but it should be noted that there are other ways to investigate the accuracy of self-appraisals of cognitive performance. Further, BRANCH is administered digitally, precluding those without Internet access from participating in this research or benefiting from this study’s findings. As the digital divide continues to narrow, this point may become somewhat less concerning. Future work will need to further elucidate the value of repeated self-appraisal as a diagnostic or prognostic tool in neurodegenerative disease, for instance by uncovering the factors that moderate the relationship between objective cognitive performance and the self-appraisal of such performance.

Conclusion

In conclusion, repeated assessments of individuals’ self-perceived performance on a cognitive test may help elucidate their capacity to update their self-perception during multiday learning. Our finding that test-naïve self-appraisal and accuracy were only weakly correlated warrants caution about using self-appraisal as a proxy for metamemory accuracy cross-sectionally. Repeated objective and self-perceived performance may be especially helpful in distinguishing healthy aging from the earliest stages of neurodegenerative diseases such as Alzheimer’s disease.

Key points.

Question:

How does self-appraisal of cognitive test performance change among cognitively normal older adults after short-term repeated exposure?

Findings:

Self-appraisal on the first day of testing was only weakly correlated with objective performance and other clinical measures, yet after seven days of taking the same test, self-appraisal was related more strongly to objective performance and associated with various measures of cognition and mood.

Importance:

Asking someone about their self-perceived performance on a test after repeated exposure provides an improved estimation of their objective cognitive performance than after a single assessment.

Next steps:

To further distinguish the worried well from those who harbor underlying disease, repeated self-appraisal may be combined with daily momentary assessments of an individual’s mood, activities, health, and state of mind.