Psychosocial Function in Mild Cognitive Impairment: Social Participation is Associated With Cognitive Performance in Multiple Domains

Abstract

Psychosocial function is associated with cognitive performance cross-sectionally and cognitive decline over time. Using data from the COMPASS-ND study, we examined associations between psychosocial and cognitive function in 126 individuals with mild cognitive impairment, an at-risk group for Alzheimer’s disease (AD). Psychosocial function was measured using questionnaires about mental health, social support, and social engagement. Composite scores for five cognitive domains were derived using principal component analysis. Multiple linear regression models were used to test the effects of various psychosocial factors on cognitive performance, controlling for age, sex, education, MoCA scores, and living circumstances. We found that low current participation in one’s social networks, over other psychosocial factors, was associated with worse verbal fluency and processing speed scores than those endorsing normal or high social participation. Our findings provide groundwork for further psychosocial-cognitive analyses in individuals at-risk for AD to better understand the role of poor social engagement in cognitive decline.

What this paper adds

• • This study is one of few to assess the relationship between psychosocial factors and cognitive function in a clinical population with mild cognitive impairment, a critical at-risk group for Alzheimer’s disease.

• • While other studies focus on isolated aspects of social relationships or lack comprehensive analyses of cognitive function, we explored many psychosocial factors and multiple cognitive domains derived from a comprehensive neuropsychological battery.

Applications of study findings

• • Preventative strategies or interventions encouraging social participation and interactions with others may be stimulating and helpful in promoting better cognitive function in those with (or at risk for) Alzheimer’s disease.

Introduction

The number of older adults living with Alzheimer’s disease (AD) is high and expected to increase exponentially over the next decade. Given the profound personal and societal costs of AD, identifying modifiable risk factors is a public health priority. Cognitive function in older adults can be influenced by a broad range of modifiable risk factors (e.g., poor diet, physical inactivity). In a recent systematic review, social isolation has been identified as one modifiable risk factor that hypothetically can prevent or delay up to 45% of dementias at the population level (Livingston et al., 2024). As adults age, social relationships and levels of engagement are modified due to multiple reasons, such as migration of family and friends, reduced social network size, and declining health and cognitive abilities (Cudjoe et al., 2020). Although further research is required to better understand the temporal association between poor social function and cognitive decline, understanding psychosocial functioning in groups with (or at risk for) dementia is an important target for potential intervention. The aim of the current study was to examine associations between psychosocial factors and cognitive performance in individuals with mild cognitive impairment (MCI), who are at risk for developing AD.

In older adults, social relationships play an important role in the protection against depression, coronary heart disease, functional decline, and all-cause mortality (Holt-Lunstad et al., 2010). Similarly, social engagement is thought to protect against cognitive decline and dementia through multiple mechanisms (Berkman et al., 2014; Sommerlad et al., 2023). Early theories suggest that active social engagement offers cognitive stimulation, building cognitive reserve (Stern, 2012). An enriched cognitive reserve then shapes neural pathways that can optimize cognitive performance and compensate for underlying neuropathology through alternative brain networks or cognitive strategies, thereby reducing cognitive decline. It is proposed that there is a dual effect of cognitive reserve, whereby high levels of engagement are associated with positive cognitive function, and consequently, high levels of overall functioning promote an engaging and active lifestyle (Kelly et al., 2017). Finally, social relationships can provide resources that buffer the impact of stress on cognitive health.

Social relationships are divided into structural and functional aspects (Berkman et al., 2014). Structural factors describe characteristics about social networks (e.g., size of social networks, the number of family and friends, living circumstances, and marital status) and social activities (e.g., amount of participation and engagement with others, attending community activities), indicating one’s social connectedness and network ties. Functional measures assess the role and support that are fulfilled by the social network (e.g., perceived social support, loneliness). However, heterogeneity in definitions and measurements of social relationships and a lack of intervention studies have prevented consensus on understanding the role of social function in cognitive decline (Kelly et al., 2017).

Psychosocial Relationships and Cognitive Decline

Multiple meta-analyses provide longitudinal evidence that structural social factors, like reduced social network size, number of social ties, or participation in social activities, are associated with increased dementia risk in late life over and above socioeconomic circumstances, lifestyle behaviors, and physical health (Kelly et al., 2017; Kuiper et al., 2016; Sommerlad et al., 2023). There is substantive evidence that reduced participation in social activities and fewer network ties can predict dementia onset and severity (Dyer et al., 2021; Kotwal et al., 2016; Kuiper et al., 2015, 2016; Piolatto et al., 2022; Wang et al., 2023). In a recent systematic review of 40 cohort studies comparing the influence of multiple social factors on the risk for dementia in cognitively healthy older adults, frequent social contact and engagement with others at baseline were found to be most strongly associated with a decreased risk of dementia (Wang et al., 2023). Moreover, several psychological (e.g., depression, anxiety) and factors of perceived social function (e.g., loneliness, social isolation) predict cognitive decline and dementia (Freire et al., 2017; Shankar et al., 2013; Shen et al., 2022; Sutin et al., 2020). Overall, there is consistent evidence that poor social engagement, small social networks, reduced social participation, and even psychological symptoms are all associated with risk for cognitive decline and dementia incidence. However, due to limitations in longitudinal research (e.g., short follow-up intervals, few repeated assessments), it is proposed that there is likely a bidirectional relationship between psychosocial function and dementia, such that low participation contributes to cognitive decline, which may further reduce levels of social engagement or psychosocial well-being (Sommerlad et al., 2023).

Psychosocial Relationships and Cognitive Function

Many reviews have examined the impact of various social factors on cognitive performance, such as loneliness (Boss et al., 2015), social isolation (Evans et al., 2019), social networks (Yoo, 2022), social activity (Brown et al., 2012; Evans et al., 2019), and social support (Kelly et al., 2017). Each of these reviews presents unequivocal findings regarding positive associations between psychosocial and cognitive functions, such that better psychosocial function at baseline is associated with better cognitive performance at follow-up; however, results vary across studies regarding which specific cognitive domains are associated with social variables. In reviews assessing the role of social networks and social activity on cognitive outcomes in cognitively healthy older adults, frequent participation in social activities was most strongly associated with better global cognitive function, memory, executive function, visuospatial abilities, and processing speed (Evans et al., 2019; Kelly et al., 2017). Similarly, other reviews indicate that quantitative (e.g., network size, complexity) social measures are more strongly associated with changes in episodic memory, working memory, and processing speed compared to qualitative (e.g., relationship satisfaction) measures (Yoo, 2022). This finding is supported by longitudinal studies examining the role of social networks and social engagement on cognitive function, which present that rich social networks and frequent social engagement were associated with better verbal fluency and memory cross-sectionally (Stoykova et al., 2011) and slower memory decline (Barnes et al., 2004; Ertel et al., 2008) and better verbal fluency (Brown et al., 2012) over time in cognitively healthy older adults.

Factors of perceived social function and psychological well-being are also associated with cognitive performance in cognitively healthy older adults. For example, loneliness is linked with decline in global cognition, perceptual speed and visual memory cross-sectionally (O’Luanaigh et al., 2012) and memory, perceptual speed, and visuospatial ability over time (Shankar et al., 2013). Moreover, individuals with depressive and anxiety symptoms have worse memory, language, processing speed, and executive function performance compared to controls (Beaudreau & O’Hara, 2009; Hamilton et al., 2014).

Rationale

Most studies on social-cognitive studies relationships have focused on cognitively healthy older adults or assessed the relationship between psychosocial factors and cognitive impairment over time (e.g., progression to dementia at follow-up). The effects of psychosocial variables on cognitive function have rarely been researched in older adults at risk for AD, particularly in the early stages of cognitive impairment where social engagement might have significant protective effects. Given that 1) psychosocial factors contribute to dementia risk and 2) are associated with cognitive performance in cognitively healthy older adults, it is important to examine the relationship between psychosocial and cognitive function in clinical groups with existing cognitive impairment (e.g., MCI) to understand whether (and which) psychosocial factors are linked to cognitive function. This is particularly critical for groups at risk for AD to elucidate the pathways in which poor psychosocial function may contribute to cognitive decline. Moreover, while many systematic reviews conclude consistent effects of psychosocial factors on cognitive outcomes, results are varied and studies have focused on only single, isolated aspects of social relationships (e.g., focusing on structural or functional social variables) or lack comprehensive analysis of cognitive function (e.g., exploring few cognitive domains, over-reliance on a single measure for assessing cognitive abilities). Therefore, the aim of the current study was to explore associations between various psychosocial factors (ranging from psychological symptoms to structural and functional social measures) and multiple cognitive domains (by using a comprehensive neuropsychological battery to assess memory, executive function, verbal fluency, working memory, and processing speed) in individuals with MCI; and more broadly, for these cross-sectional associations to serve as a starting point in exploring and understanding larger, longitudinal social-cognitive relationships in clinical populations at earlier stages of AD.

Methods

Using the COMPASS-ND dataset (Data Release 5), we analyzed data from 126 participants who met the criteria for MCI (M_Age = 71.5 ± 6.4, M_Education = 15.4 ± 3.9). COMPASS-ND is the clinical study of the Canadian Consortium on Neurodegeneration in Aging (CCNA; Chertkow et al., 2019). Participants completed intake interviews and comprehensive evaluations, including clinical and neuropsychological assessment. The study was approved by the Jewish General Research Ethics Board. Data used in this paper are stored on the Longitudinal Online Research Information System (https://www.ccna.loris.ca; Das et al., 2012; Mohaddes et al., 2018). General inclusion and exclusion criteria for the study are listed elsewhere (Chertkow et al., 2019).

MCI Criteria

Participants with MCI were selected based on the following criteria: 1) concern regarding a change in cognition from previous levels based on the participant’s or an informant’s report; 2) impairment in one or more cognitive domains that is greater than what would be expected for the participant’s age and education: WMS-III Logical Memory II score below education-adjusted Alzheimer’s Disease Neuroimaging Initiative (ADNI) cutoffs, a Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) word list recall score less than 6, global Clinical Dementia Rating (CDR) score >0, and a Montreal Cognitive Assessment (MoCA) score between 13–24; 3) assigned a CDR score of ≤ .5 to not be given a diagnosis of dementia; 4) have preservation of independence in functional abilities by having a score greater than 14/23 on the Lawton-Brody Instrumental Activities of Daily Living (IADL) scale; and 5) absence of diffuse subcortical cerebrovascular disease.

Measures

Psychological Symptoms

Geriatric Depression Scale (GDS)

The GDS is a 30-item questionnaire for assessing depressive symptoms in older adults (Yesavage, 1988). It uses a “yes/no” response format to inquire about levels of enjoyment, interest, motivation, and social interactions over the past week. Scores ranges 0–30: 0–9 is normal, 10–19 indicate mild depression, and 20–30 indicate severe depression.

Generalized Anxiety Disorder Scale (GAD-7)

The GAD-7 is a 7-item screening tool for assessing anxiety symptoms in adults (Spitzer et al., 2006). Questions assess how often participants experience different symptoms of anxiety over the past two weeks. Responses range from “not at all” to “nearly every day,” scored from 0 to 3. Total scores ranges 0–21: 0–4 indicate minimal anxiety, 5–9 indicate mild anxiety, 10–14 indicate moderate anxiety, and 15+ indicate severe anxiety.

Social Factors

The COMPASS-ND dataset includes self-reported data on several social factors, including the type and frequency of social engagement in one’s network (e.g., amount of participation in community activities, frequency of interaction with one’s social network) and social function (e.g., feelings of loneliness, social support, and wanting to participate in more social activities). These data were collected prior to the COVID-19 pandemic.

Measures of Social Structure

Living Arrangements

Current living circumstances were measured by a self-reported question: “With whom do you live?”

Social Network

Social network availability and engagement were measured using a collection of questions about levels of perceived social engagement. Participants rated regarding the frequency of current participation with one’s social network, including frequency of telephone and in-person contact. To measure the frequency of current social participation, participants were asked, “Taking only your current situation into consideration, how would you rate your participation/involvement in social activities?” Scores of 0–2 were given to response categories matching “low,” “normal,” and “high.” To measure frequency of telephone contact, participants were asked, “In the past week, how many times did you talk to someone—friends, relatives, or others—on the telephone?” To measure frequency of time spent with others, participants were asked, “In the past week, how many times did you spend time with someone who does not live with you, for instance you went to see them or they came to visit you, or you went out to do things together?” For both questions, responses were scored from 0 to 4, corresponding to “not at all,” “once,” “twice,” “3–4 times,” and “once or more a day.”

Social Participation

Participants answered questions about the frequency and types of social activities they participated in with others (e.g., volunteer work, community and professional activities, religious activities). Scores of 0–4 correspond to responses of “never,” “at least once a year,” “at least once a month,” “at least once a week,” and “at least once a day.” Scores were categorized as normal (one or more social activities a week) and low social participation (no social activities per week; based on Hämäläinen et al., 2019).

Measures of Social Function

MOS Social Support Survey

This 19-item questionnaire measured the perceived availability of social support (Sherbourne & Stewart, 1991). A composite score (0–100) of emotional/informational support, tangible support, affectionate support, and positive social interactions was used.

Quality of Life

This self-report questionnaire measured subjective perception of the participant’s position in life, in context of their health, comfort, and overall happiness. This scale has 13 items with a total score ranging from 0 to 52.

Loneliness

Loneliness was measured based on a “yes/no” response format for one item in the social support questionnaire: “Do you feel lonely, or do you feel very lonely?”

Cognitive Function

Cognitive function was assessed in five domains using a variety of neuropsychological tests. Composite scores of memory, executive function, processing speed, working memory, and verbal fluency were derived from a principal component analysis of test scores; test administration and selection are described elsewhere (Phillips et al., 2023, submitted). Composite scores were created from specific tests: memory from delayed recall scores on the Rey Auditory Verbal Learning Test, the Benton Visual Memory Test-Revised, and the CCNA-CIMA-Face-Name Association Task, executive function from ratio scores on the Inhibition and Switching conditions of the D-KEFS Color-Word Interference task, processing speed from scores on the WAIS-III Digit Symbol-Coding Test and the CCNA Reaction Time measures, working memory from the WAIS-III Digit Span total score and a ratio score of the Trail Making Test, verbal fluency from scores on phonemic (letter) and semantic (category) fluency on the D-KEFS Verbal Fluency Test. Higher component scores signify better memory, working memory, and verbal fluency but indicate poorer processing speed and executive function.

Statistical Analyses

Data analyses were conducted using R and RStudio (Version 3.6.2). There was less than 6% missing data for psychosocial variables and less than 5% for cognitive variables, which were addressed using mean imputation. We used linear regression models to test the direct effects of psychosocial variables on cognition (i.e., memory, executive function, processing speed, working memory, and language), controlling for age, sex, education, and MoCA scores, and living circumstances (living alone). Into each regression, we entered those control variables, followed by predictor variables of GAD-7 and GDS scores, quality of life score, social support score, loneliness (yes/no), current social participation rating, frequency of telephone participation, frequency of time spent with others, and participation in social activities (low/normal). Each regression was done per cognitive domain, with five regressions calculated overall, to determine which specific psychosocial measures were associated with cognitive performance across multiple domains.

Results

In our MCI sample (N = 126), sex was slightly imbalanced (60% M, 40% F). Most participants (95%) lived with at least another person, and 80% were married or in a common-law partner. Psychological questionnaires indicated mild levels of anxiety (M = 4.3 ± 4.3) and depression (M = 6.8 ± 4.9) on average. Participants reported adequate social support (M = 79.5 ± 19.7) and quality of life (M = 39.8 ± 5.7). Most participants endorsed either “low” (38%) or “normal” (52%) current social participation, compared to high (10%) current participation. Regarding social participation in community activities, 18% had “low” participation versus 82% with “normal” participation. Regarding telephone contact per week, 2% reported none, 6% reported once, 23% reported twice, 42% reported 3–4 time per week, and 28% reported once or more a day. Regarding time spent with others in-person per week, 9% reported none, 16% reported once, 28% reported twice, 40% reported 3–4 time per week, and 7% reported once or more a day. Moreover, 15% reported feelings of loneliness.

Relationships Between Psychosocial Function and Cognitive Performance

Memory

The overall regression was statistically significant (see Table 1). No psychosocial measures were significantly associated with memory scores.

Table 1.

Regression Model Testing Multiple Psychosocial Factors on Memory.

Regression Model
Predictor	B	SE	T-Value
Age
(Years)	−.04	.01	−3.12**
Sex
Male	−.28	.18	−1.55
Education
(Years)	.02	.02	.43
MoCA score
(Total score)	.15	.03	5.54***
Living circumstances
Living with spouse, common-law partner, family member, or friend	Ref.
Living alone	.06	.28	.22
Geriatric anxiety disorder scale (GAD-7)
(Total score)	−.04	.02	−1.53
Geriatric depression scale (GDS)
(Total score)	.03	.02	1.36
Quality of life
(Total rating)	.04	.02	1.95
Social support
Total score	−.01	.00	−2.06*
Loneliness
No	Ref.
Yes	−.22	.28	−.77
Current social participation
Low	Ref.
Normal	.11	.18	.61
High	−.07	.31	−.23
Telephone frequency
None	Ref.
Once per week	−1.38	.82	−1.67
Twice per week	−1.46	.79	−1.86
Three-four times per week	−1.45	.78	−1.86
Once or more per day	−1.23	.81	−1.52
Time spent with others frequency
None	Ref.
Once per week	.57	.38	1.47
Twice per week	.49	.38	1.3
Three-four times per week	.40	.38	1.04
Once or more per day	.45	.48	.92
Desire for social participation
No	Ref.
Yes	.08	.16	.51
Amount of social activities
Normal	Ref.
Low	.28	.21	1.32

Note. ***p < .001, **p < .01, *p < .05; B, unstandardized beta coefficient; SE, standard error; Ref., reference condition.

Executive Function

The overall regression model was not statistically significant (see Table 2). No psychosocial measures were significantly associated with executive function scores.

Table 2.

Regression Model Testing Multiple Psychosocial Factors on Executive Function.

Regression Model
Predictor	B	SE	T-Value
Age
(Years)	.03	.02	2.19*
Sex
Male	−.45	.21	−2.21*
Education
(Years)	−.04	.02	−1.51
MoCA score
(Total score)	−.07	.03	−2.26*
Living circumstances
Living with spouse, common-law partner, family member, or friend	Ref.
Living alone	.09	.32	.28
Geriatric anxiety disorder scale (GAD-7)
(Total score)	−.02	.03	−.74
Geriatric depression scale (GDS)
(Total score)	.02	.03	.60
Quality of life
(Total rating)	.00	.02	.02
Social support
Total score	.01	.01	1.10
Loneliness
No	Ref.
Yes	.03	.33	.11
Current social participation
Low	Ref.
Normal	.07	.21	.35
High	.25	.36	.70
Telephone frequency
None	Ref.
Once per week	.37	.95	.39
Twice per week	.06	.91	.07
Three-four times per week	.43	.91	.48
Once or more per day	−.17	.94	−.18
Time spent with others frequency
None	Ref.
Once per week	−.02	.45	−.04
Twice per week	−.09	.44	−.22
Three-four times per week	.09	.44	.20
Once or more per day	−.10	.56	−.18
Desire for social participation
No	Ref.
Yes	−.10	.19	−.51
Amount of social activities
Normal	Ref.
Low	−.33	.24	−1.36

Note. ***p < .001, **p < .01, *p < .05; B, unstandardized beta coefficient; SE, standard error; Ref., reference condition.

Verbal Fluency

The overall regression was statistically significant (see Table 3). Current social participation was significant associated with verbal fluency performance. Specifically, those endorsing “low” current social participation (M = −.14 ± 1.03) had worse verbal fluency scores than those endorsing “normal” current social participation (M = −.05 ± .91, β = .47, p < .05) and “high” current social participation (M = .81 ± 1.03; β = 1.08, p < .001). We found no statistically significant differences between men and women on verbal fluency.

Table 3.

Regression Model Testing Multiple Psychosocial Factors on Verbal Fluency.

Regression Model
Predictor	B	SE	T-Value
Age
(Years)	.01	.01	.42
Sex
Male	−.06	.18	−.33
Education
(Years)	.03	.02	1.27
MoCA score
(Total score)	.13	.03	4.41***
Living circumstances
Living with spouse, common-law partner, family member, or friend	Ref.
Living alone	−.10	.28	−.35
Geriatric anxiety disorder scale (GAD-7)
(Total score)	−.02	.02	−.88
Geriatric depression scale (GDS)
(Total score)	.02	.02	1.07
Quality of life
(Total rating)	.01	.02	.59
Social support
Total score	−.01	.01	−1.88
Loneliness
No	Ref.
Yes	.06	.29	.21
Current social participation
Low	Ref.
Normal	.47	.18	2.53*
High	1.08	.32	3.42***
Telephone frequency
None	Ref.
Once per week	1.04	.83	1.25
Twice per week	1.20	.80	1.51
Three-four times per week	1.16	.79	1.48
Once or more per day	.89	.82	1.09
Time spent with others frequency
None	Ref.
Once per week	−.76	.39	−1.94
Twice per week	−.14	.38	−.36
Three-four times per week	−.49	.39	−1.36
Once or more per day	−.53	.49	−1.08
Desire for social participation
No	Ref.
Yes	−.13	.16	−.80
Amount of social activities
Normal	Ref.
Low	.60	.21	2.78**

Note. ***p < .001, **p < .01, *p < .05; B, unstandardized beta coefficient; SE, standard error; Ref., reference condition.

Processing Speed

The overall regression was statistically significant (see Table 4). Although not reaching statistical significance, there was a trend of social participation associated with processing speed scores. Specifically, those endorsing “low” current social participation (M = .11 ± 1.09) had slower processing speed than those endorsing “normal” current social participation (M = .00 ± .92, β = −.40, p = .058) and “high” social participation (M = −.46 ± .95, β = −.70, p = .056). We found that men performed better on processing speed tests compared to women; however, this difference was not statistically significant (β = −.39, t = −1.89, p = .07).

Table 4.

Regression Model Testing Multiple Psychosocial Factors on Processing Speed.

Regression Model
Predictor	B	SE	T-Value
Age
(Years)	.01	.01	.94
Sex
Male	−.39	.21	−1.89
Education
(Years)	−.03	.02	−1.15
MoCA score
(Total score)	−.08	.03	−2.44*
Living circumstances
Living with spouse, common-law partner, family member, or friend	Ref.
Living alone	−.06	.32	−.17
Geriatric anxiety disorder scale (GAD-7)
(Total score)	−.01	.03	−.43
Geriatric depression scale (GDS)
(Total score)	−.03	.03	−1.25
Quality of life
(Total rating)	−.03	.02	−1.33
Social support
Total score	.01	.01	1.34
Loneliness
No	Ref.
Yes	.11	.33	.34
Current social participation
Low	Ref.
Normal	−.40	.21	−1.91
High	−.69	.36	−1.93
Telephone frequency
None	Ref.
Once per week	−.88	.95	−.93
Twice per week	−.44	.91	−.49
Three-four times per week	−.95	.90	−1.05
Once or more per day	−.95	.93	−1.02
Time spent with others frequency
None	Ref.
Once per week	.31	.44	.71
Twice per week	.09	.44	.21
Three-four times per week	.29	.44	.67
Once or more per day	.57	.59	1.03
Desire for social participation
No	Ref.
Yes	.15	.19	.79
Amount of social activities
Normal	Ref.
Low	−.16	.24	−.66

Note. ***p < .001, **p < .01, *p < .05; B, unstandardized beta coefficient; SE, standard error; Ref., reference condition.

Working Memory

The overall regression model was statistically significant (see Table 5). No psychosocial measures were significantly associated with working memory scores.

Table 5.

Regression Model Testing Multiple Psychosocial Factors on Working Memory.

Regression Model
Predictor	B	SE	T-Value
Age
(Years)	−.01	.02	−.50
Sex
Male	−.44	.21	−2.08*
Education
(Years)	−.04	.03	−1.71
MoCA score
(Total score)	−.11	.03	−3.45***
Living circumstances
Living with spouse, common-law partner, family member, or friend	Ref.
Living alone	.06	.33	.17
Geriatric anxiety disorder scale (GAD-7)
(Total score)	.02	.03	.53
Geriatric depression scale (GDS)
(Total score)	−.00	.03	−.09
Quality of life
(Total rating)	.01	.02	.43
Social support
Total score	.00	.01	.37
Loneliness
No	Ref.
Yes	.02	.34	.06
Current social participation
Low	Ref.
Normal	−.16	.22	−.75
High	−.27	.37	−.72
Telephone frequency
None	Ref.
Once per week	−.20	.98	−.20
Twice per week	.54	.93	.58
Three-four times per week	.41	.93	.44
Once or more per day	.20	.96	.21
Time spent with others frequency
None	Ref.
Once per week	−.05	.46	−.10
Twice per week	−.16	.45	−.36
Three-four times per week	−.20	.45	−.43
Once or more per day	−.20	.58	−.34
Desire for social participation
No	Ref.
Yes	.01	19	.04
Amount of social activities
Normal	Ref.
Low	−.08	.25	−.33

Note. ***p < .001, **p < .01, *p < .05; B, unstandardized beta coefficient; SE, standard error, Ref., reference condition.

Post-Hoc Analyses on Current Social Participation

Further analyses of current social participation in our sample ¹ demonstrated that participants with low (N = 49), normal (N = 65), and high (N = 12) current participation had similar age, education, and living circumstances (see Table S1). There were significant differences between levels of social participation on depression and quality of life scores, with those reporting low social participation endorsing higher depressive symptoms and lower quality of life than those with normal and high social participation. Of the women, 16% reported high participation, 49% reported normal participation, and 35% reported low participation. Of the men, 5% reported high participation, 53% reported normal participation, and 42% reported low participation. These differences in social participation between men and women were not statistically significant. Moreover, adding an interaction between sex and current social participation into the regression models for verbal fluency and processing speed did not account for significant additional variance.

Discussion

The goal of the present study was to examine associations between psychosocial function and cognitive performance in multiple domains in individuals with MCI. We found that current social participation within one’s social network was uniquely associated with cognitive performance on neuropsychological tests of verbal fluency and processing speed, after accounting for demographic factors, MoCA scores, living circumstances, and all other psychosocial variables. There were no associations between psychosocial factors and cognitive performance on memory, executive function, and working memory.

Through our post-hoc analyses, we found limited evidence of sex differences across social participation categories and no interaction between sex and current participation on verbal fluency or processing speed performance. The high participation group had few participants and were mostly female; however, our findings were largely driven by cognitive performance in the larger, more balanced low participation group. In addition, we controlled for significant differences in depression and quality of life by entering these predictors before current social participation in our analyses. We also considered lifetime social participation versus current social participation, which indicated that 53% of people with normal lifetime participation and 10% of people with high lifetime participation now reported current low social participation. This means that many participants currently reporting low participation had normal or high lifetime participation, indicating a decline in perceived participation over time. These changes in levels of perceived engagement can be associated with individual factors (e.g., changes in health, increasing frailty, chronic disease, disability, fewer opportunities for physical and cognitive stimulation) and changes in social networks related to aging (e.g., reduced network size due to death or migration of family and friends); as such, these variables may be implicated in our low participation group. While other studies have linked social participation with cognitive function in various domains (e.g., memory, working memory, and executive function) in cognitively healthy older adults (see Kelly et al., 2017 for a systematic review), we found relationships between low social participation and poor processing speed and verbal fluency scores in those with MCI. This suggests that social participation may influence cognitive performance differently in MCI compared to cognitively healthy older adults.

We found that low current social participation was associated with poorer verbal fluency scores in individuals with MCI. Given that social interactions typically involve word generation and verbal fluency (i.e., accessing one’s lexical and semantic networks and retrieving words from vocabulary), one’s level of social participation may indicate how often these skills are practiced through interactions with others. Previous evidence demonstrates that high social engagement and participation are associated with better performance on verbal fluency in cognitively healthy older adults (Bourassa et al., 2017; Brown et al., 2012). While these studies have measured social participation using the number of social activities per week or the number of people interacted with regularly, we measured social participation using both objective (participation in number of activities) and subjective (rating of current participation) methods and found that only subjective perceptions of social participation were associated with verbal fluency. While our findings are not entirely consistent with previous evidence on social activities and verbal fluency in cognitively healthy individuals, our results corroborate a more general link between social participation and verbal cognitive performance and show a particular association between perceived current participation and verbal fluency in individuals with MCI. This suggests that frequent communication and engagement with others may provide opportunities to exercise and facilitate verbal skills (e.g., lexical access, semantic retrieval), or alternatively, participants with higher verbal ability may be able to better maintain relationships and pursue interactions; although longitudinal analyses are required to confirm these hypotheses.

Although not statistically significant, we observed a trend between current social participation and processing speed, such that low social participation was associated with slower processing speed in individuals with MCI. Importantly, processing speed is a sensitive marker of cognitive aging and is less resistant to age-related cognitive decline, compared to other cognitive abilities (Lindenberger et al., 1993). In fact, changes in processing speed are thought to underpin cognitive decline in more complex cognitive abilities (Lindenberger et al., 1993). As such, we consider processing speed to be a fundamental, lower-level cognitive process that influences age-related decline in other complex cognitive tasks. Previous research demonstrates a link between processing speed and social participation in cognitively healthy older adults (Ghisletta et al., 2006; Lövdén et al., 2005). While our findings did not reach a conventional criterion for statistical significance, they extend a previous link between social participation and processing speed in individuals with MCI.

The Role of Social Participation in Cognitive Function

We investigated the role of multiple psychosocial variables on cognitive performance and found that only perceived current social participation was linked with cognitive function, even after controlling for living circumstances, depression, and loneliness. Current social participation was not highly correlated with other psychosocial variables—correlations between current social participation and other psychosocial measures entered in the regression models ranged from r = .1–.3, except for quality of life (r = .39). This lowers that the possibility that our results are explained by other psychosocial variables (e.g., one’s current social participation being driven by depression or anxiety about their levels of participation). Additionally, the difficulty of determining the directionality of social-cognitive relationships and the possibility of reverse causality (i.e., poor cognitive function is not a consequence but rather a cause of reduced social participation) is a well-documented concern in the literature (Sommerlad et al., 2023). Although it is difficult to rule out this possibility, we attempt to address this by using MoCA scores as covariates; that is, the relationship between social participation and cognitive performance in processing speed and verbal fluency is reliable regardless of the participant’s cognitive status. As such, these results should be considered as a useful starting point to test future hypotheses on social-cognitive relationships.

To explain the pathways between social participation and cognitive performance, it is hypothesized that social participation and engagement with one’s social network may mitigate the impact of negative physiological consequences of stressors, or act as a resource to enhance well-being and adaptive lifestyle behaviors (Berkman et al., 2014). Another possible mechanism is via the role of social engagement in offering cognitive stimulation, which can preserve cognitive and neural networks (Berkman et al., 2014; Stern, 2012). Compared to memory and executive function domains, it is possible that lower-order abilities in our processing speed and fluency tasks are implicated in a complex set of cognitive processes that occur more frequently during social participation (e.g., processing speech, searching through language networks, sequencing words and sentences, monitoring verbal cues and speech). Particularly in our sample with MCI, we hypothesize that social participation may offer cognitively stimulating opportunities or have a protective effect against decline in lower-level cognitive processes, which requires further testing using longitudinal data.

Limitations and Future Directions

Although the COMPASS-ND dataset includes diverse psychosocial information, the psychometric nature of tests available may have affected our lack of findings for other psychosocial variables. For example, some measurements of psychosocial variables were not standardized or were limited to a singular question or item within a questionnaire (e.g., loneliness). Regarding study design, we used mean imputation for missing data and included a high number of variables in our analyses with a limited sample size, which could have impacted power and reliability of findings. Although this study was novel in examining cross-sectional associations between psychosocial and cognitive function in MCI, we could not determine the directionality, temporal nature, and mechanisms of these results without longitudinal data. Furthermore, future research comparing cognitively healthy controls and clinical groups (e.g., MCI, AD) with greater variability in psychosocial well-being and cognitive performance will clarify how psychosocial function impacts complex cognitive processes, particularly as cognitive function changes due to increasing cognitive impairment or conversion to AD.

In our study, we found that social participation has a positive relationship with cognitive performance in multiple domains in individuals with MCI. Specifically, we found that low current social participation was associated significantly with weaker verbal fluency and near-significantly with slower processing speed. To our knowledge, this is the first study to explore associations between a range of psychosocial factors and cognitive performance in various domains in an at-risk diagnostic group. Our findings extend the social-cognitive associations found in previous studies with cognitively healthy older adults to individuals with MCI, and suggest that interventions encouraging social participation and interactions with others may be stimulating and helpful in promoting better cognitive function in those at-risk for developing AD.

Ethical Statement

Ethical Approval

The study was approved by the Jewish General Research Ethics Board. Data used in this paper are stored on the Longitudinal Online Research Information System (https://www.ccna.loris.ca).

ORCID iD

Sana Rehan https://orcid.org/0000-0002-9750-6870