Examining the role of aging perceptions in Subjective Cognitive Decline

Abstract

OBJECTIVE:

While Subjective Cognitive Decline (SCD) is gaining ground as a “pre-clinical” risk state for AD, its utility depends on our understanding of the factors linked to SCD. Rarely examined sociocultural factors including perceptions of aging may relate to the subjective experience of cognitive aging. Identifying such associations will help to refine the utility while setting the stage for addressing modifiable factors contributing to SCD.

METHODS:

The study consisted of N=136 participants (68% female; 73% White; 22% Black, Age_mean=74.72; Education_mean=16.01). Questionnaires assessed SCD, depressive symptoms, and age perceptions (essentialist aging beliefs, subjective age, age group identification, and explicit/implicit age stereotypes). Cognitive functioning was measured with a semantic interference and learning task.

RESULTS:

SCD was correlated with essentialist aging beliefs, age identification and depressive symptoms (r_range=.18-.22, p_range=.009-.02, CI_range=.00-.39). Essentialist aging beliefs were correlated with subjective age and age group identification (r_range=.22-.42, p_range<.001-.003, CI_range=.08-.57). Both age group identification and essentialism were correlated with depressive symptoms (r_range=.22, p_range=.009-.01, CI_range=.04-.39). In the adjusted regression model including, depressive symptoms, age perceptions and SCD, only SCD was associated with cognition (b=−.31, p<.001).

CONCLUSION:

Although correlated with SCD, perceptions of aging do not explain the relationship between SCD and performance on a sensitive cognitive test among older adults

1. Introduction

Subjective cognitive decline (SCD), the experience of cognitive decline in the absence of clinically significant objective cognitive impairment¹, has received attention as a potential indicator of preclinical Alzheimer’s disease (AD) for decades². A growing number of studies linking SCD to AD biomarkers such as amyloid burden, hippocampal volume, and cerebrospinal fluid (CSF) tau levels have reinforced the value of SCD for identifying and understanding the earliest stages of AD^3–8, an increasingly important endeavor amidst recurring failed clinical intervention trials in symptomatic patients⁹. While screening for SCD would be an inexpensive, quick, and broadly deployable manner of aiding preclinical diagnosis and improving the efficacy of clinical trials, its subjective nature complicates its straight-forward interpretation as a risk state for disease.

Fundamentally, SCD is based on self-reflection, a complex and imperfect process influenced by a variety of biological (neurodegenerative disease), psychological (mood and personality), and sociocultural forces^10–12. In order to increase the specificity of SCD as a marker for preclinical AD, careful consideration of how SCD is measured (task factors), and the characteristics of the person in whom it is measured (person factors) is needed. Recent work on task factors has shown that asking individuals to compare their cognition to others their age improves the association between SCD and performance on objectively measured cognitive tasks sensitive to very early AD¹³. Regarding person factors, it is well known that subjective perceptions and/or expressions of cognitive decline are linked to psychological states and traits, most notably depressive symptoms¹⁴. Although mood is routinely considered in understanding the context in which SCD occurs, more stable belief systems relating to aging are not formally considered or evaluated as part of clinical neurological or neuropsychological assessments.

Lifespan psychologists have demonstrated substantial heterogeneity in how individuals view and react to the changing ratio of biopsychosocial gains to losses that accompany aging ^15,16. Moreover, individuals’ perceptions about aging in general (i.e., age stereotypes, essentialist aging beliefs) as well as their own aging (i.e., subjective age, age group identification) have important impacts on age-related outcomes including cognitive functioning. For example, negative age stereotypes (i.e., generalized representations of older adults as inactive, impaired, or incompetent which can operate both within and outside of conscious control¹⁷) have been shown to have widespread negative effects on self-concept, mood, health, longevity and cognitive performance in older adults^17–20. These detrimental effects are hypothesized to manifest when individuals internalize negative stereotypes into their cognitive schemas in a manner similar to gender and racial stereotypes²¹. In addition, holding strong essentialist beliefs about aging (i.e., that aging is a fixed and immutable process that cannot be influenced) has also been observed to influence cognition in older adults, in addition to exacerbating the effect of negative age stereotypes on cognitive function^22,23. Finally, beliefs about one’s own aging including subjective age (i.e., the age a person feels) and age group identification (in this case, the extent to which one identifies with older adults) have also been identified as playing an important role in overall subjective health and well-being^24,25 as well as objective cognitive functioning in older age^26,27.

It is thus possible that SCD may relate to aging perceptions. Moreover, given the links between aging perceptions and objective cognitive performance, it is possible that such perceptions explain the association between SCD and objective cognition. Finally, as depressive symptoms have been shown to relate to all constructs of interest (SCD, cognition, and age perceptions), it is important to consider such symptoms when examining SCD in relation to age perceptions and cognitive functioning among older adults.

The first aim of this study was to examine the extent to which perceptions of aging including negative age stereotypes, essentialist aging beliefs, subjective age and age group identification are linked to SCD. The second aim is to determine if these attitudes explain the previously demonstrated association between SCD and objective cognitive functioning. We hypothesized that feeling older than one’s age, having more fixed beliefs about aging, identifying more closely with older adults, and holding more negative age stereotypes would each be associated with higher SCD. Second, we hypothesized that such perceptions may contribute to and explain the association between SCD and a sensitive measure of cognitive function.

2. Methods

2.1. Participants

Participants for this study were selected from the parent study on Subjective Cognitive Decline which currently comprises of 157 participants recruited from the Columbia University Medical Center Aging, and participants from the Dementia Neurology Clinic[1] (n=12), two of the clinical cases were referred to the neurology clinic through a memory-concern screener administered in the Columbia University Department of Obstetrics & Gynecology. The remaining participants were referred from ongoing aging studies at Taub Institute at Columbia University (n=145). Referral aging studies included the Alzheimer’s Disease Research Center (n=73), Washington Heights Inwood Columbia Aging Project (n=35), Testing Olfaction in Primary care to detect Alzheimer’s disease and other Dementias (n=11), and Cognitive Reserve and Reference Ability Neural Network studies (n=22), Imaging inflammation in elders with different clinical and biomarker profiles of Alzheimer’s disease (n=2), Concerns About Memory Problems (n=2). To be included in the parent study, participants were required to have performed within normal limits on standard neuropsychological testing (demographically adjusted z-scores above −1.5) within the last 12 months (see Supplementary Table S1 for neuropsychological screening measures) and have no history of neurological conditions or ongoing psychiatric diagnoses. All participants provided written consent and procedures were approved by the Columbia University IRB. A total of 136 participants had available measures selected for study (67.6 % females; n=92) with a mean age of 73.69 (SD=6.79) years and mean education of 16.22 (SD=2.42). Participants self-identified as White (73 %, n=99), Black (22%, n=30), Other (4 %, n=5) and Asian (1%, n =2)¹. A total of six individuals (6.9 %) identified as Hispanic.

2.2. Measures

2.2.1. Subjective Cognitive Decline Questionnaire

Participants completed an age-anchored, 20-item SCD questionnaire (i.e., “Do you have difficulty with […] compared to others of your age?). Ratings in each item ranged from (0-no problem to 6-major problem) summing to a total score that ranged from 0–120 with higher scores indicating increased subjective complaints (see Chapman and colleagues 2021 for scale¹³). Previously reported Cronbach’s alpha for this scale was .93 in line with this sample’s Cronbach alpha which was .94¹³.

2.2.2. Perceptions of age and aging

Age Stereotypes:

Age stereotypes were measured both explicitly and implicitly. Explicit age stereotypes were measured through a questionnaire consisting of 10 items that examined the extent to which participants agreed with positive or negative attributes of older adults. Each item contained opposing ends of one attribute (i.e., dependent-independent) in a scale from −3 to 3, with 0 being neutral¹⁶. Total explicit age stereotypes scale ranged from −30 to 30 with higher scores representing endorsement of more negative explicit aging stereotypes. Cronbach’s alpha for this scale was of .87 in this sample.

Implicit age stereotypes were assessed using an adapted Implicit Association Test (IAT)^16,28.This computerized test assesses if participants have automatic positive or negative associations with older adults by measuring their reaction times when forced to make such associations. The total score of the IAT, ‘d’, is derived by subtracting log transformed “old-negative” reaction time latencies from “old-positive” reaction time latencies. A faster average response time in the “old-negative” block compared to the “old-positive” block was interpreted as an implicit preference towards negative age stereotypes. Higher d values represent increased endorsement of negative implicit stereotypes of aging.

Essentialist beliefs about aging:

Essentialist beliefs were assessed via a 4 item scale²⁹ which reflects a continuum of self-reported beliefs on how fixed or malleable the aging process is. Example items include “To a large extent, a person’s age biologically determines his or her abilities,” and “No matter at what point in life, you can always influence your own aging”. Each item is rated on a scale from 0 (do not agree) to 6 (absolutely agree). Two items were reversed scored, resulting in a range of 0–24 with higher scores indicating increased essentialist beliefs (i.e., age as an irreversible and fixed process). Cronbach’s alpha for this scale has been previously reported between 0.55–0.72²⁹ in line with the samples Cronbach’s alpha which was .56.

Subjective age:

Felt age was measured with one item that asked participants to indicate how old they felt. Chronological age is then subtracted from felt age to generate a discrepancy score with positive scores indicating feeling older than one’s chronological age³⁰.

Age group identification:

Age group identification was measured with two items that assessed the extent to which participants identify with “older people”. Items included “I identify with older people” and “I have a lot in common with older people”³¹. Each item is rated on a scale of 1 (do not agree) to 7 (absolutely agree) with higher scores indicating greater age group identification. Total score ranged from 2–14. Cronbach’s alpha for this scale was of .88 in this sample.

2.2.3. Cognitive functioning

Cognitive functioning was measured with The Loewenstein-Acevedo Scales of Semantic Interference and Learning (LASSI-L)³² a recently developed list-learning test that measures susceptibility to proactive semantic inference and the ability to recover from proactive semantic interference. As part of this measure participants are asked to read aloud a list of 15 words, List A, from three semantic categories: fruits, musical instruments, and articles of clothing. Participants are then asked to recall the items (A1) learned with the three semantic categories as cues (e.g., “Can you tell me all the words on the list that were fruits[3] ?”). List A is then read again, followed by another cued recall (A2). Participants are then presented with a second list (new set of 15 words, List B) from the same semantic categories (fruits, musical instruments, and articles of clothing), followed by recall (B1, susceptibility to proactive semantic interference). Following list B participants are presented with List B again, and recall (B2, recovery from semantic interference). Immediately following B2, participants are asked to recall all of the words from List A (A3, susceptibility to retroactive semantic interference). For the purposes of this study susceptibility to proactive interference (B1) was selected given its previous associations with biomarkers of AD such as amyloid load and AD signature volumetric loss and its previous shown association with SCD^32,33.

2.2.4. Self-reported depressive symptoms

Self-reported depressive symptoms were measured via the Geriatric Depression Scale (GDS)³⁴. This scale consists of 15 items assessing various depressive symptoms such as hopelessness, fear of the worst happening, etc. A total score for this study was computed excluding one item that asks individuals about their memory (as this would be expected to overlap with SCD). The total score ranged from 0 – 14 with higher scores indicating higher endorsement of depressive symptoms. Cronbach’s alpha of the GDS was .75 within this sample.

2.3. Statistical analyses

Means and standard deviations are reported on participants’ demographics (age, gender and education), questionnaires of age perceptions and SCD, and cognitive function in Table 1. Two tailed correlations were conducted between demographics (age and education) and SCD, perceptions of aging and depressive symptoms. Independent sample t-test were conducted to examine differences with regards to gender group. Given our hypothesis, bivariate analyses were conducted with one-tailed Spearman correlations to examine associations between SCD and perceptions of aging. Two regression models were then conducted to examine SCD’s association with cognitive performance when first adjusting for depressive symptoms and demographics. The second regression model adjusted for depressive symptoms, demographics and perceptions of aging. In all comparisons, statistical significance was defined at the α = 0.05 level (i.e., p < 0.05).

Table 1.

Descriptives of SCD, depressive symptoms, age perceptions and association with demographics age, education and gender (n = 136).

	Descriptives		Associations with demographics
			Age			Education			Gender
	M (SD) / n (%)	Sample Range	r	p	CI	r	p	CI	M Difference^	p	CI
Age	73.69 (6.79)	56–92	-	-	-	-	-	-	-	-	-
Education	16.22 (2.42)	10–20	-	-	-	-	-	-	-	-	-
Gender - Female	92 (68%)	-	-	-	-	-	-	-	-	-	-
SCD Age anchored (0 – 120)	18.82 (15.79)	0 – 60	.14	.10	−.04, .31	−.10	.24	−.27, .07	0.10	.97	−5.64, 5.85
Explicit stereotypes (−30 – 30)	−7.01 (9.06)	−30 – 17	−.06	.48	−.22, .11	.25	.004	.09, .39	4.46	.007	1.25, 7.66
Implicit stereotypes (RTold-positive – RTold-ngeative )	0.04 (0.14)	−0.35 – 0.43	−.05	.55	−.22, .11	.06	.49	−.23, .25	0.01	.80	−0.04, 0.06
Essentialist beliefs (0 – 24)	8.89 (4.97)	0 – 24	.14	.10	−.02, .31	.04	.64	−.11, .20	1.06	.25	−0.73, 2.85
Subjective age (felt age -chronological age)*	−14.21 (11.66)	−49 – 5	.05	.56	−.11, .22	.17	.06	−.00, .08	1.16	.59	−3.10, 5.42
Age identity	6.84 (3.20)	0 – 12	.20	.02	.01, .38	−.11	.21	−.28, .08	0.91	.12	−0.24, 2.06
Depressive symptoms (0 – 14)	1.25 (1.80)	0 – 10	−.03	.71	−.21, .13	−.11	.19	−.27, .06	0.54	.11	−0.11, 1.89
B1 Proactive interference (0 – 16)	8.82 (2.86)	1 – 15	−.26	.002	−.42, −.08	.27	.002	.10, .41	−0.37	.49	−1.40, 0.67

Note. M= Mean, SD = Standard deviation, r = Spearman correlation coefficient, p = significance p-value, CI = confidence intervals. Confidence intervals for correlations were calculated from 1000 bootstrap correlations. Bolded values reflect statistical significance.

^{^}Mean difference between males and females. Positive values reflect larger means in men than women; negative values reflect larger mean values in women compared to men.

^*n=133 with one outlier removed.

3. Results

3.1. Demographics, SCD, Perceptions of aging and depressive symptoms

Table 1 displays the summary scores and bivariate associations for self-reported SCD, perceptions of aging, and depressive symptoms. One participant reported their felt age as 300 years which was treated as an outlier and removed from analyses. SCD was not correlated with any demographic variables. Regarding perceptions of aging, males endorsed more negative explicit age stereotypes than women (p <.05). Finally, more negative explicit stereotypes and older subjective age and cognitive functioning were positively correlated with level of education (see Table 1).

As reported in Table 2, higher SCD was correlated with stronger essentialist beliefs, greater identification with older people and more depressive symptoms. Similarly, depressive symptoms were correlated with stronger essentialist beliefs and greater identification with older people. Significant correlations were also observed between these perceptions of aging such that stronger essentialist beliefs were associated with greater identification with older people. Older subjective age was correlated with stronger essentialist beliefs about aging and identification with older people. A regression model including all age attitudes, however, showed no unique associations between age attitudes and SCD (see Supplementary Table S2). Finally, only SCD was significantly correlated with cognitive function.

Table 2.

Bivariate associations between SCD, perceptions of aging, and depressive symptoms.

	SCD			Explicit age stereotypes			Implicit age stereotypes			Essentialist aging beliefs			Subjective age			Age identity
	r	p	CI	r	p	CI	r	p	CI	r	p	CI	r	p	CI	r	p	CI
Explicit	−.03	.73	−.20, .15
Implicit	−.06	.47	−.24, .11	.13	.14	−.04, .29
Essentialist beliefs	.18	.04	.01, .34	.06	.48	−.12, .23	−.17	.06	−.31, .01
Subjective age	.08	.38	−.11, .26	.10	.24	−.10, .26	−.05	.57	−.22, .13	.42	<.001	.26, .57
Age identity	.21	.02	.00, .37	−.15	.07	−.32, .04	−.07	.42	−.25, .12	.25	.003	.08, .42	.26	.002	.10, .42
Depressive symptoms	.22	.009	.07, .39	.15	.08	−.03, .32	−.02	.81	−.19, .15	.30	<.001	.15, .45	.14	.11	−.03, .29	.22	.01	.04, .37
Proactive Semantic Interference	−.34	<.001	−.50, −.19	.09	.15	−.10, .27	.02	.42	−.15, .19	−.07	.23	−.25, .11	.10	.12	−.08, .27	−.13	.06	−.30, .02

Note.

^*Confidence intervals (CI) calculated from 1000 bootstrapping samples.

Regression analyses with cognitive functioning as outcome

The overall model was significant (F(5,132) =8.37, p <.001) and explained 25% of the cognitive outcome. Higher SCD, older chronological age and fewer years of education were independently associated with lower cognitive scores. There was no significant association between cognitive functioning and depressive symptoms. Table 4 summarizes results of the regression model adjusting for all perceptions of aging, demographics and depression. The overall model was significant (F(10,122) = 4,27, p <.001) and explained 26% of the cognitive outcome (semantic proactive interference). SCD, age and education were associated with semantic interference. Neither age attitudes nor depressive symptoms were associated with cognitive functioning.

Table 4.

SCD, age perceptions, depressive symptoms & demographics predicting cognitive functioning

	B (SE)	Beta	p
SCD	−0.055 (0.015)	−.307	<.001
Subjective age	0.024 (0.022)	.096	.276
Age group identification	0.014 (0.079)	.016	.860
Essentialist aging beliefs	−0.016 (0.051)	−.028	.754
Explicit age stereotypes	0.010 (0.028)	.032	.717
Implicit age stereotypes	−0.919 (1.652)	−.044	.579
Depressive symptoms	−0.087 (0.139)	−.054	.532
Age	−0.099 (0.035)	−.236	.005
Gender (0=Men, 1=Women)	0.724 (0.510)	.120	.158
Level of Education	0.286 (0.103)	.239	.007

4. Discussion

SCD is a risk factor for future cognitive decline in older adults^1,2,14. In fact, we would argue that many if not all individuals who develop Alzheimer’s disease dementia likely move through a state of SCD at some point early in their disease course³⁵. Increasingly, studies are revealing the utility of SCD for detecting important imaging and cognitive markers of pre-clinical AD^3–8,13. However, subjective perceptions of cognitive decline are not always indicative of pathologic cognitive aging³⁶, and it is important to identify the factors that may contribute to SCD more broadly³⁷. This is important because it informs the circumstances under which SCD is most likely to be a marker of emerging AD, and because it allows for identification of potentially modifiable factors that can improve perceptions of cognitive decline. Historically, while much emphasis has been placed on mood as a primary contributor to SCD^38–41, it is increasingly appreciated that both SCD and mood may together herald emerging AD⁴². In this study, we examine SCD in relation to aging perceptions including subjective age, age stereotypes, essentialist aging beliefs, and age group identification in an effort to determine whether such perceptions may contribute to the experience of SCD or even detract from its association with a sensitive marker of cognitive functioning.

In a series of bivariate analyses, we first examined correlations among SCD, age perceptions, depressive symptoms and cognition. There were a number of important findings. First, results confirmed the well-established link between SCD and depressive symptoms^36,43. Second, results supported the idea that the different age perceptions measured in this study represented unique constructs, as several perceptions were not or only weakly correlated with the other constructs (e.g., age stereotypes), and none of the correlations between perceptions exceeded a medium effect size (r = .40). Consistent with previous work, aging essentialism, age group identification, and subjective age were associated with one another⁴⁴.

Bivariate analyses also revealed novel, selective correlations between SCD and aging perceptions. In particular, higher SCD was correlated with stronger beliefs that aging is a fixed and inevitable process (i.e., aging essentialism) and a stronger sense of identification with older adults. These results suggest that both general beliefs about the aging process as well as perceptions about one’s own aging relate to SCD. Somewhat unexpectedly, neither subjective age nor age stereotypes (explicit and implicit) were correlated with SCD. Interestingly, depressive symptoms showed the same pattern of correlations with age perceptions. That is, depressive symptoms were most strongly correlated with aging essentialism and age group identification. The fact that SCD and depressive symptoms show the same pattern of correlations raises the possibility that depression mediates the association between age perceptions (essentialism / age group identification) and SCD, or that such age perceptions mediate the association between SCD and depressive symptoms. This cross-sectional work may thus serve as preliminary evidence for longitudinal models designed to map the temporal pathways between these constructs. Specifying the pathways through which these constructs affect SCD, or vice versa, will likely have key repercussions as the field moves towards early nonpharmacological interventions.⁴⁵

Results of the first regression model bolster previous findings in our lab¹³ and in a growing number of studies that when measured against sensitive cognitive outcomes, SCD relates to objective cognitive performance. In a recent study, we showed that age-anchored SCD (perceptions that memory is worse than others the same age) mapped onto three challenging memory tasks. The current study extends that finding to a fourth measure and also shows that SCD is associated with cognitive performance when adjusting for depressive symptoms. Whether or not depressive symptoms should be considered as a covariate remains unclear, as it is possible that new onset depressive symptoms may confer increased risk for AD⁴⁶.

Finally, despite the associations between certain aging perceptions and SCD, there were no bivariate associations between any aging perceptions and objective cognitive performance. Moreover, adding perceptions of aging into the regression model did not change or detract from the extent to which SCD captures cognitive performance on a task that is sensitive to pre-clinical AD. There are several implications of these findings. First, it appears that there is non-overlapping variance between SCD and aging perceptions versus SCD and objective cognition. Stated differently, the subjective experience of cognitive decline may reflect multiple, independent pathways of information arising from both belief systems and cognitive functioning.

The current findings raise interesting questions about the directionality of the observed associations. Ongoing longitudinal work is dedicated to clarifying the temporal dynamics and mechanisms underlying the association between SCD and age perceptions. While the current analyses were based on the premise that perceptions of aging are stable beliefs that are longstanding and may influence SCD, it’s certainly possible that higher SCD leads people to identify more with older adults and develop stronger essentialist views on aging. One might speculate that the lack of an association between SCD and subjective age might be due to the reduced number of individuals that actually endorsed feeling older than their age. The average score for subjective age was negative with only one case reported feeling older than their age with the majority feeling their actual age or younger. This could have skewed the impact that this feeling may have on SCD. Similarly, with regards to age stereotypes, explicit stereotypes had a negative mean indicating that on average this group did not endorse strong negative stereotypes towards older adults and thus might not be related to this construct.

The current study had several limitations including the fact that the sample was comprised largely of non-Hispanic whites; inclusion of a more diverse set of participants who might hold different cultural beliefs of aging could reveal different results and should be examined in future studies. A second limitation was a lack of a distinction between new onset versus longstanding depressive symptoms, each of which might relate differently to SCD and age perceptions and bring more clarity to the question at hand.

To conclude, SCD relates not only to depressive symptoms, but to specific perceptions of aging including essentialist aging beliefs and age group identification. Importantly, these associations do not detract from the ability of SCD to detect variability on a sensitive measure of cognitive functioning. Routine examination of SCD as part of wellness visits in older adults will create opportunities to identify early signs of cognitive dysfunction and also to address potentially modifiable attitudes about the aging process.

Table 3.

SCD, depressive symptoms and demographics predicting cognitive functioning.

	B (SE)	Beta	P
SCD	−0.053 (0.014)	−.297	<.001
Depressive symptoms	0.057 (0.130)	−.036	.660
Age	−0.100 (0.033)	−.238	.003
Gender (0=Men, 1=Women)	0.704 (0.484)	.116	.148
Level of Education	0.312 (0.096)	.261	.001