Employment Status and Later Life Cognitive Functioning: A Gendered Perspective on the Moderating Role of the Social Network

Abstract

Gender disparities in later life cognition call for a deeper understanding of how social determinants interact to shape cognitive outcomes. This study investigates the gendered moderating role of the social network in the association between employment status (employed, retired, homemaker, unemployed) and cognitive functioning (episodic memory, verbal fluency) among adults aged 50+ in Europe. Using data from the Survey of Health, Ageing and Retirement in Europe (SHARE; waves 4 (2011-2012), 6 (2015-2016), 8 (2019-2020)), we apply linear multilevel models stratified by gender, with episodic memory and verbal fluency as outcomes (N = 145,107). Results indicate that a stronger social network may buffer negative effects of non-employment on episodic memory for women, whereas for men, benefits are primarily indicated at lower social network strength levels. Our findings highlight gender-specific cognitive advantages of social networks, suggesting that gender should be considered a structural factor, not merely a demographic characteristic, in studies of cognitive aging.

Keywords

aging cognition gender differences social network survey of health ageing and retirement in Europe (SHARE)

Introduction

Good cognitive functioning is a requirement for living an independent life (Salthouse, 2012), and it is associated with enhanced well-being in later life (Llewellyn et al., 2008; Pfund et al., 2025). Several studies, however, highlight significant disparities in cognitive functioning among older adults across countries (Arnhold et al., 2025; Asperholm et al., 2019; Skirbekk et al., 2012) as well as within countries (Herlitz et al., 2015; Myrskylä & Margolis, 2014; Saenz et al., 2018; Weber et al., 2014), with gender playing a key role (Asperholm et al., 2019; Weber et al., 2014). To explain these cognitive disparities, several recent studies separately investigate the roles of employment (Vance et al., 2016) and the social network (Grasset et al., 2022; Kelly et al., 2017; Roth, 2022). We examine the two together from a gendered perspective to shed light on their interplay in explaining gendered heterogeneities in later life cognitive functioning.

According to the successful aging model by Rowe and Kahn (1997), participation in productive activities, such as employment, as well as social relations, can be considered as psychosocial enablers of good cognitive functioning in older age. Relatedly, the ‘use it or lose it’ hypothesis postulates that the continued engagement in cognitively stimulating activities in older age promotes positive neuroplasticity, exerting a protective effect on cognitive functioning (Salthouse, 2006).

Prior empirical evidence indicates a compensatory relationship between activity participation (encompassing social, physical, intellectual, and recreational activities) and the social network related to cognitive functioning: for individuals engaging less in such activities, the social network appears to be particularly beneficial for maintaining cognitive functioning (Paiva et al., 2021). Furthermore, the social network-cognition nexus shows substantial gender differences (Zunzunegui et al., 2003), with older men and women navigating their social network differently (Antonucci et al., 2014; Schwartz & Litwin, 2018).

Building on these findings, this study investigates whether the social network moderates the association between employment in older age, measured as employment status (employed, unemployed, retired, homemaker), and cognitive functioning, measured as episodic memory and verbal fluency, and how such a moderating role differs by gender.

As such, this study is the first to systematically integrate a gender-informed perspective on the moderating role of social networks in the employment-cognition relationship, recognizing gender as a structuring force that shapes both social networks and cognitive outcomes. We therefore approach gender not only as a variable for subgroup analysis, but also as a lens through which we interpret the linkages between employment, the social network, and cognitive functioning. Furthermore, while prior studies have predominantly contrasted employment with retirement or treated non-employment as a homogeneous category, we explicitly distinguish between different forms of non-employment (retirement, unemployment, and homemaking).

Literature Review

Employment Status and Cognitive Functioning

Employment represents a central source of cognitive stimulation for older adults, requiring one to perform complex tasks and adapt to new environments while engaging in personal interaction with coworkers or clients (Vance et al., 2016). According to the ‘use it or lose it’ hypothesis, occupational cognitive stimulation results in neurogenesis, benefiting cognitive functioning (Salthouse, 2006). Additionally, employment may reduce psychological stress by providing income security, helping to establish daily routines, and fostering a sense of purpose and self-esteem which, in turn can positively affect cognitive performance (Goodman et al., 2017; Hakanen & Schaufeli, 2012; Stevens-Ratchford, 2011). Evidence predominantly confirms the positive role of employment for cognitive functioning of older adults (Bertogg & Leist, 2023; Greenberg & Burgard, 2021).

However, 62% of 50-year-olds and older in the EU-27 do not work (being either retired, unemployed, or homemakers), the majority being women (Eurostat, 2020). Notably, non-employed older adults are a heterogenous group, with retirees, the unemployed, and homemakers encompassing differing social contexts and implications for cognitive functioning. Retirees, for instance, frequently engage in activities that are similarly structured as employment after their work careers, such as voluntary work, and tend to have more friend-based networks, whereas homemakers often center their lives around existing social networks and neighbors (Ajrouch et al., 2005; Burnett-Wolle & Godbey, 2007). Social networks of unemployed older adults were observed to be family-centered (Haynes et al., 2014).

Research on the relationship between retirement and cognitive functioning predominantly presents non-significant or adverse effects (Alvarez-Bueno et al., 2021; White et al., 2025 for systematic reviews). Unemployment is recognized to negatively affect cognitive functioning (Vélez-Coto et al., 2021), while little is known about the association between homemaking and cognitive functioning (e.g. Bertogg & Leist, 2023; Ice et al., 2020 on the negative association between adult-life homemaking spells and later life cognitive functioning). These mixed findings suggest that the roles of retirement, unemployment and homemaking for cognitive functioning may depend on moderating factors.

Social Network and Cognitive Functioning

For older adults who are retired, unemployed, or homemakers, the social network may serve as an important substitute source for cognitive stimulation, considering that interacting with members of one’s social network involves interpreting body language, organizing and recalling information in conversations, and anticipating the reactions of conversation partners (Vance et al., 2016). Furthermore, regular social contact and emotional as well as instrumental support from network members contribute to fostering positive emotional states, reducing psychological stress, and mitigating the risk of depression (Berkman & Krishna, 2014; Fratiglioni et al., 2004).

Indeed, substantial evidence positively relates the social network to cognitive functioning and cognitive reserve (Fratiglioni et al., 2004; Kelly et al., 2017). Given the complexity of social networks and their strength depending on structural aspects, such as proximity of contacts and frequency of interactions, as well as on functional aspects, such as the roles of network members in providing emotional and instrumental support (Fiori et al., 2007), a diversified social network that fulfils a broader range of functions was often associated with improved cognitive functioning (Ellwardt et al., 2015; Paiva et al., 2021; Pan & Chee, 2020). Relatedly, networks rich in weak or peripheral ties (e.g., acquaintances, neighbors) expose individuals to novel information, diverse social cues, and cognitively demanding interactions (Burt & Burt, 2007; Granovetter, 1973). Research has linked social networks that include a high proportion of weak ties to improved cognitive outcomes (Perry et al., 2022).

Gender Differences in the Role of the Social Network

According to the convoy model, the social network is a group of individuals who accompany a person throughout their life, providing support and adapting alongside them as circumstances change (Kahn & Antonucci, 1980). It views the social network as a dynamic system that includes both stable and evolving ties, shaped by individuals’ personal and situational changes over time (Antonucci et al., 2011). Emphasizing the contextual nature of social ties, the convoy model posits that gender-specific life courses lead to differences in the function of later life social networks between men and women (Antonucci et al., 2014), which may in turn structure the role of social networks for cognitive functioning.

Gender also structures the types of cognitive and social experiences embedded in employment. Men, on average, are more likely to work in positions characterized by greater occupational complexity and more frequent interactions with coworkers or clients, while women’s employment trajectories are more often interrupted or concentrated in more relational but socially homogeneous roles (England, 2010; Fisher et al., 2014). These gendered exposures imply that the cognitive resources lost upon exiting employment differ for men and women, and that the kinds of social networks required to compensate for reduced employment-related stimulation may also diverge across genders.

Concerning gender differences in later life cognitive functioning, research indicates that women experience substantial cognitive benefits from a stronger social network (Tomioka et al., 2018; Zunzunegui et al., 2003). Moreover, engagement with weaker ties, such as friends, has been associated with decelerated cognitive decline in women only (Zunzunegui et al., 2003). A potential underlying mechanism is that women receive higher levels of emotional support from friends than men (Liao & Scholes, 2017; Liebler & Sandefur, 2002), with women’s friend-based social networks being connected with improved mental health outcomes. Conversely, in men, the receival of high social support from very close contacts, such as their spouse or partner, is associated with decelerated cognitive decline (Liao & Scholes, 2017).

Taken together, these strands of research suggest an integrated framework in which employment and social networks represent partially overlapping sources of cognitively enriching activity. Employment typically provides structured, socially diverse, and cognitively demanding engagement, whereas social networks can supplement or compensate for this stimulation (Berkman et al., 2000; Fratiglioni et al., 2004). In this view, social networks may operate as moderators in the employment–cognition relationship, altering the extent to which reduced employment-related stimulation translates into cognitive consequences. Gender, as a key organizing principle of life-course roles, shapes both employment experiences and network configurations (Risman, 2004), and therefore may condition the strength and direction of this moderating process.

The few existing studies on the association between employment status and the social network in relation to cognitive functioning predominantly focus on the transition from employment to retirement, highlighting the importance of the social network for cognitive functioning following retirement (Börsch-Supan & Schuth, 2013; Grotz et al., 2018). However, scientific evidence for other forms of later life non-employment is missing.

Moreover, empirical evidence has pointed to gender differences in the interlinkage between activity participation and the social network in relation to cognitive functioning (Tomioka et al., 2018; Zunzunegui et al., 2003), showing a more pronounced effect of social engagement on cognitive functioning for women. However, gender differences in the link between (non-)employment and the social network in relation to cognitive functioning have not been considered so far.

This study contributes to the literature by examining simultaneously how four employment statuses (employed, retired, unemployed, and homemaker) are associated with cognitive functioning and by investigating whether social network strength moderates these relationships in gender-specific ways using cross-national survey data.

Based on the literature discussed above, we hypothesize that cognitive functioning differs across employment statuses in later life, with employed individuals exhibiting higher cognitive functioning than those who are retired, unemployed, or engaged in homemaking and that the strength of the social network moderates the relationship between employment status and cognitive functioning. We expect that a strong social network positively moderates associations for retired, unemployed, and homemaking individuals by providing alternative sources of cognitive stimulation that partially compensate for the reduced exposure to cognitively demanding interactions that employment typically affords. In light of prior research indicating gender differences in the roles of the social network and activity participation in relation to cognitive decline, we expect the social network as a moderator to be particularly crucial for women’s later life cognitive functioning.

Research Design

Data Source

We analyze data from the Survey of Health, Ageing and Retirement in Europe (SHARE), a panel survey on health, social, and economic conditions, as well as social engagement, of non-institutionalized people aged 50 and above in Europe and Israel (Börsch-Supan et al., 2013). In this study, we use data from waves 4, 6, and 8 (conducted in 2011-2012, 2015-2016, and 2019-2020, covering 27 countries), which include a social network module providing detailed information on the respondents’ social surroundings (Bergman et al., 2019; SHARE-ERIC, 2024a; SHARE-ERIC, 2024b; SHARE-ERIC, 2024c).

We restrict the sample to observations of participants aged at least 50 years living in Europe with information on all variables used in our models. We further excluded observations of participants with current employment status Don’t know (24 observations), Other (2127 observations), Permanently sick or disabled (4326 observations), or refused to answer (18 observations). Our final sample comprises 82,090 observations (28,231 in Wave 4; 29,874 in Wave 6; 23,985 in Wave 8) for women (N = 49,465) and 63,017 observations (21,914 in Wave 4; 23,300 in Wave 6; 17,803 in Wave 8) for men (N = 38,974). The average number of waves per respondent is 1.64 (SD = 0.74).

Outcome: Cognitive Functioning

The outcome variables cover episodic memory and verbal fluency, two dimensions of cognitive functioning that are sensitive to aging-related cognitive decline (Gonzalez-Burgos et al., 2019; Machado et al., 2018). The variable episodic memory represents the combined immediate and delayed recall test score, ranging from 0 to 20 total words recalled (mean = 9.16). In the immediate recall test, the interviewer reads a list of ten words aloud, asking the respondent to recall as many of these words as possible within 1 minute. The delayed recall test takes place a few minutes later, with the respondents being asked to recall as many words as possible from the previously read-out words without rehearing them. In the verbal fluency test, the participants are asked to name as many animals as possible within 60 seconds. Scores in our sample range from zero to 97 (mean = 20.09).

Independent Variables

The independent variables are social network strength and employment status. Social network strength is operationalized using a composite indicator. The social network scale (sn scale) is treated as an ordinal variable rather than continuous, as it was constructed from a composite of multiple discrete components and categorized into five meaningful levels. The highest category, sn scale = 4, indicates a strong social network, whereas the lowest category, sn scale = 0, is assigned only to individuals who reported having no social network at all. The sn scale was first introduced and validated by Litwin and Stoeckel (2016) and was composed using information on individuals’ network size, proximity to contacts (number of contacts living within 25 km), their frequency of contact (number of individuals with weekly or more contact), received support (number of contacts with very close emotional ties), and diversity of contacts (number of different types of contacts). Weighing each of these traits equally, a raw score ranging from 0 to 20 was converted to the sn scale (0 = 0, 1 = 1–5, 2 = 6–10, 3 = 11–15, and 4 = 16–20). In our sample, the sn scale follows a bell-shaped distribution, centered on level 2, with 2.82% of total respondents in our samples at level 0, 25.91% at level 1, 46.44% at level 2, 21.34% at level 3, and 3.49% at level 4. For the categorical variable employment status, we use the participants’ self-reported current employment status with the categories employed (including self-employed), retired (from own work only), homemaker, and unemployed.

We control for sociodemographic and economic factors that are potential confounders. The sociodemographic variables include the continuous variable age at the interview (mean ages are 67.64 years (women) and 68.01 years (men)) and age squared, accounting for non-linear age-related cognitive decline. Highest educational attainment is divided into the three categories primary (International Standard Classification of Education [ISCED] levels 0-1), secondary (ISCED levels 2-4), and post-secondary education (ISCED levels 5-6). Marital status is divided into the categories married (including registered partnerships), divorced, widowed, and never married. Considering that parenthood is both related to employment status and the social network, we include a binary control variable children, indicating whether the respondent is childless (Fanelli & Profeta, 2021; Muller et al., 2020). We consider the economic situation of the respondent by including the financial wealth quintile – the relative wealth position within a country, based on self-reported real and financial assets per household, and adjusted for household size. We further include survey wave as a categorical variable and learning, a binary variable indicating first participation in the cognitive test, accounting for learning effects in repeated cognitive testing (Collie et al., 2003). Table 1 presents summary statistics for the sn scale and the control variables, categorized by employment status and gender.

Table 1.

Descriptive Statistics of the Mean or Percentages for the Control Variables by Employment Status for Women and Men (Women/Men)

Variables	Employment situation (women/Men)				Total % or mean (women/Men)
Variables	Employed	Retired	Unemployed	Homemaker	Total % or mean (women/Men)
Outcomes
Episodic memory (Mean (SD))	11.44 (3.20)/10.45 (3.14)	8.85 (3.72)/8.23 (3.47)	9.95 (3.36)/9.08 (3.43)	8.24 (3.77)/8.70 (3.53)	9.42 (3.77)/8.83 (3.53)
Verbal fluency (Mean (SD))	23.64 (7.64)/22.73 (7.62)	19.39 (7.56)/18.91 (7.40)	20.56 (7.74)/19.72 (7.76)	16.30 (7.22)/19.40 (7.71)	20.06 (7.86)/20.13 (7.67)
Covariates
sn scale = 0 (%)	14.99/22.32	69.88/71.77	2.47/5.32	12.66/0.59	100.00/100.00
sn scale = 1 (%)	18.41/25.80	62.10/70.20	2.60/3.64	16.88/0.37	100.00/100.00
sn scale = 2 (%)	21.12/25.66	62.09/71.08	2.53/2.94	14.26/0.32	100.00/100.00
sn scale = 3 (%)	26.82/27.21	58.33/70.19	2.32/2.36	12.53/0.25	100.00/100.00
sn scale = 4 (%)	29.79/28.73	53.73/68.68	2.55/2.29	13.93/0.30	100.00/100.00
Age (in years) (mean)	57.40/58.43	71.78/72.01	57.03/57.79	67.65/64.04	67.64/68.01
Primary education (%)	6.65/10.60	62.93/85.70	2.13/3.32	28.29/0.37	100.00/100.00
Secondary education (%)	21.69/27.00	62.45/68.89	2.95/3.72	12.92/0.39	100.00/100.00
Post-secondary education (%)	37.00/33.74	56.41/64.17	1.88/1.89	4.71/0.20	100.00/100.00
Married (%)	25.20/26.11	54.48/70.92	2.73/2.67	17.59/0.30	100.00/100.00
Widowed (%)	6.27/7.02	80.11/91.84	0.73/0.85	12.88/0.30	100.00/100.00
Divorced (%)	32.47/36.32	60.94/56.95	4.46/6.43	2.13/0.30	100.00/100.00
Never married (%)	34.03/34.68	59.61/56.12	3.59/8.39	2.77/0.81	100.00/100.00
Children: Yes (%)	21.99/25.79	60.70/71.05	2.48/2.87	14.83/0.30	100.00/100.00
Children: No (%)	23.36/27.59	64.79/66.26	2.64/5.53	9.20/0.63	100.00/100.00
Financial wealth: Q1 (%)	17.42/23.19	63.01/68.86	4.34/7.44	15.23/0.51	100.00/100.00
Financial wealth: Q2 (%)	22.81/26.67	58.85/69.87	2.63/3.14	15.71/0.32	100.00/100.00
Financial wealth: Q3 (%)	23.47/25.92	59.13/71.62	2.24/2.18	15.16/0.28	100.00/100.00
Financial wealth: Q4 (%)	23.29/25.21	61.64/72.80	1.60/1.77	13.47/0.23	100.00/100.00
Financial wealth: Q5 (%)	23.98/28.47	62.37/69.66	1.49/1.55	12.16/0.32	100.00/100.00
Survey wave 4 (%)	22.59/30.65	59.10/65.01	2.56/4.03	15.74/0.31	100.00/100.00
Survey wave 6 (%)	17.78/25.99	68.18/70.61	1.77/3.15	12.26/0.24	100.00/100.00
Survey wave 8 (%)	25.26/20.13	57.03/77.46	3.03/1.94	14.68/0.47	100.00/100.00
Learning: 1st participation (%)	27.28/34.36	54.06/60.04	4.08/5.26	14.58/0.33	100.00/100.00
Learning: Re-participation (%)	20.34/23.01	63.42/74.30	1.95/2.36	14.28/0.33	100.00/100.00
N (total)	18,146/16,356	50,111/44,490	2,047/1,964	11,786/207	82,090/63,017
% (total)	22.10/25.95	61.04/70.60	2.49/3.12	14.36/0.33	100.00/100.00

Analytic Strategy

First, we provide a descriptive overview of the two cognitive functioning dimensions, episodic memory and verbal fluency, by employment status and social network strength, separately for men and women. Subsequently, we apply linear multilevel models with random intercepts, with observations on level 1, individuals on level 2, and countries on level 3, separately for men and women (supported by likelihood ratio tests) (Raudenbush & Bryk, 2002). We use this method to consider the hierarchical panel data structure of repeated observations of individuals and individuals being nested in countries of residence; country differences explain substantial shares of the total variations in episodic memory (10.6% for women and 6.3% for men) and verbal fluency (18.9% for women and 15.8% for men).

For each dependent variable (episodic memory and verbal fluency), we run two model specifications, separately for each gender. Models 1a (episodic memory) and 1b (verbal fluency) include employment status and sn scale, whereas models 2a-b additionally include an interaction term between employment status and sn scale. Models 1a-b evaluate the associations between employment status and cognitive functioning, as well as social network strength and cognitive functioning. Models 2a-b address the moderating role of social network strength in the relationship between employment status and cognitive functioning. In all models, we use sn scale = 2 as the reference category, as it represents the most common and median level in our sample, providing a stable baseline from which we can compare lower and higher levels more meaningfully.

Robustness and Sensitivity Checks

We conduct several robustness and sensitivity checks to test the stability of our findings. These are presented in the supplementary materials (Tables S.3-10; Figures S.1-2).

We address potential confounding and reverse causality by performing the main analysis (models 2a-b) excluding (i) all observations from wave 8 as parts of the data collection for wave 8 were conducted during COVID-19 lockdowns using computer-assisted telephone interviews (CATI), while lockdown measures may have temporarily affected components of our social network measure, such as the frequency and diversity of contacts (Table S.3); (ii) observations of individuals with no social network (sn scale = 0) (Table S.4); and (iii) reporting cognitive impairment (diagnosed with dementia or any other serious memory impairment) (Table S.5), in the light of research linking cognitive impairment to social network decline and social isolation (Dyer et al., 2021; Hajek & König, 2025) and resulting reverse causality concerns. (iv) We test gender differences in interaction effects by estimating a pooled model with three-way interactions (employment status × sn scale × gender) (Table S.6) and (v) test robustness to differences in age composition or potential survivor bias by restricting the sample to participants aged 50 to 69 years (Table S.7). We test sensitivity to the social network operationalization using (a) the continuous raw score underlying the sn scale, ranging from 0 to 20 (Table S.8); and (b) two separate components of the sn scale: (i) very close contacts (0 = 0, 1 = 1, 2 = 2–3, 3 = 4–5, and 4 = 6–7 persons) and (ii) diversity of contacts (presence of spouse; other family; friend; and other) (Tables S.9-10; Figures S.1-2). Both indicators range from 0 to 4 (reference: 2), probing the extent to which each facet drives the gendered moderation of the composite sn scale. Prior research suggests weak ties may be particularly important for women’s cognitive stimulation (Perry et al., 2022; Tomioka et al., 2018; Zunzunegui et al., 2003), while close supportive ties may be more central for men (Liao & Scholes, 2017).

Results

Descriptive Statistics

Women. Among women, employed respondents exhibit the highest cognitive scores (mean = 11.44, SD = 3.20 in episodic memory; mean = 23.64, SD = 7.64 in verbal fluency), while homemakers show the lowest score (mean = 8.24, SD = 3.77; mean = 16.30, SD = 7.22) – with the unemployed (mean = 9.95, SD = 3.36; mean = 20.56, SD = 7.74) and retirees (mean = 8.85, SD = 3.72; mean = 19.39, SD = 7.56) ranging in between (Figure 1, left-hand side). A positive social network gradient is evident in episodic memory scores among employed, retired, and homemaking respondents. In verbal fluency, we observe a positive gradient for retirees and homemakers. In both cognitive domains, the gradient is the steepest for retirees and homemakers.

Figure 1.

Mean Episodic Memory and Verbal Fluency by Employment Status and sn scale for Women and Men. Note: The dashed red line denotes the mean cognitive outcome of the related employment status. Own calculations on SHARE waves 4, 6, and 8

Men. Similarly, in men (Figure 1, right-hand side), employed respondents score the highest (mean = 10.45, SD = 3.14 in episodic memory; mean = 22.73, SD = 7.62 in verbal fluency). Lower scores are observed in the unemployed (mean = 9.08, SD = 3.43; mean = 19.72, SD = 7.76) and retirees (mean = 8.23, SD = 3.47; mean = 18.91, SD = 7.40). In men, we observe a positive social network gradient in episodic memory scores among employed and retired respondents. In verbal fluency, there is a positive gradient among the employed, the retirees, and the unemployed. In both cognitive domains, the gradient is the steepest for unemployed men. Our sample includes only 124 observations of homemaking men; therefore, we refrain from further interpreting related results.

In summary, the descriptive results hint at a positive association of both employment and social network strength with cognitive functioning. However, the social network gradient varies between employment categories, being most pronounced among retirees and homemakers for the sub-sample of women and among the unemployed for the sub-sample of men. Table S.1 in the supplementary materials displays the distribution of sn scale scores within each employment status.

The Role of Employment and the Social Network

To analyze how the employment status and the social network relate to cognitive functioning, we perform multilevel models. Tables 2 and 3 show the results of models 1a-b for episodic memory and verbal fluency, respectively, with full regression tables presented in the supplementary material (Table S.2).

Table 2.

Women: Results of the Random Intercept Models With the Outcome Variables Episodic Memory and Verbal Fluency

	Episodic memory		Verbal fluency
	Model 1a b (SE)	Model 2a b (SE)	Model 1b b (SE)	Model 2b b (SE)
Fixed effects
Employment status (ref.: Employed or self-employed)
Retired	−0.163*** (0.038)	−0.180*** (0.046)	−0.613*** (0.077)	−0.629*** (0.095)
Unemployed	−0.410*** (0.070)	−0.532*** (0.098)	−0.745*** (0.144)	−0.901*** (0.202)
Homemaker	−0.405*** (0.045)	−0.421*** (0.058)	−1.135*** (0.093)	−1.091*** (0.120)
sn scale (ref.: sn scale = 2)
sn scale = 0	−0.509*** (0.066)	−0.191 (0.167)	−0.583*** (0.136)	−0.149 (0.344)
sn scale = 1	−0.200*** (0.027)	−0.069 (0.060)	−0.508*** (0.055)	−0.445*** (0.124)
sn scale = 3	0.276*** (0.026)	0.142*** (0.051)	0.670*** (0.053)	0.563*** (0.105)
sn scale = 4	0.396*** (0.052)	0.263*** (0.097)	1.247*** (0.108)	1.308*** (0.199)
Interaction (Ref.: Employed or self-employed x sn scale = 2)
sn scale = 0
Retired × sn scale = 0		−0.370** (0.184)		−0.439 (0.379)
Unemployed × sn scale = 0		0.179 (0.455)		0.905 (0.938)
Homemaker × sn scale = 0		−0.522** (0.250)		−1.175** (0.515)
sn scale = 1
Retired × sn scale = 1		−0.163** (0.068)		−0.024 (0.141)
Unemployed × sn scale = 1		−0.194 (0.172)		0.053 (0.355)
Homemaker × sn scale = 1		−0.151* (0.089)		−0.310* (0.184)
sn scale = 3
Retired × sn scale = 3		0.172*** (0.060)		0.109 (0.124)
Unemployed × sn scale = 3		0.582*** (0.169)		0.550 (0.347)
Homemaker × sn scale = 3		0.138* (0.087)		0.238 (0.179)
sn scale = 4
Retired × sn scale = 4		0.134 (0.118)		−0.087 (0.244)
Unemployed × sn scale = 4		0.386 (0.336)		−0.213 (0.691)
Homemaker × sn scale = 4		0.344** (0.168)		−0.074 (0.346)
Intercept	4.760*** (0.581)	4.804*** (0.581)	10.189*** (1.267)	10.205*** (1.268)
Random effects: Var.
country	0.838	0.839	8.02	8.014
country-individuals	3.845	3.843	17.486	17.487
Residual	5.526	5.525	22.852	22.849
Number of observations	82,090	82,090	82,090	82,090

Note. Models 1a (Women) and 1b (Men) are without interactions. Models 2a (Women) and 2b (Men) include an interaction term between Employment status and the sn scale. Significance levels are *p < 0.1, **p < 0.05, ***p < 0.01.

Table 3.

Men: Results of the Random Intercept Models With the Outcome Variables Episodic Memory and Verbal Fluency

	Episodic memory	Episodic memory	Verbal fluency	Verbal fluency
	Model 1a b (SE)	Model 2a b (SE)	Model 1b b (SE)	Model 2b b (SE)
Fixed effects
Employment status (ref.: Employed or self-employed)
Retired	−0.273*** (0.040)	−0.211*** (0.050)	−0.654*** (0.088)	−0.629*** (0.107)
Unemployed	−0.453*** (0.072)	−0.279*** (0.106)	−1.043*** (0.155)	−0.947*** (0.231)
Homemaker	−0.332 (0.203)	−0.277 (0.305)	−0.784 (0.441)	−0.433 (0.662)
sn scale (ref.: sn scale = 2)
sn scale = 0	−0.618*** (0.067)	−0.338** (0.140)	−1.039*** (0.146)	−0.794*** (0.303)
sn scale = 1	−0.156*** (0.027)	−0.056 (0.053)	−0.486*** (0.059)	−0.413*** (0.114)
sn scale = 3	0.157*** (0.032)	0.182*** (0.061)	0.504*** (0.070)	0.488*** (0.133)
sn scale = 4	0.296*** (0.072)	0.501*** (0.135)	0.823*** (0.157)	0.634** (0.292)
Interaction (Ref.: Employed or self-employed × sn scale = 2)
sn scale = 0
Retired × sn scale = 0		−0.326** (0.159)		−0.254 (0.345)
Unemployed × sn scale = 0		−0.936*** (0.328)		−1.121 (0.713)
Homemaker × sn scale = 0		−0.232 (0.885)		−1.677 (1.920)
sn scale = 1
Retired × sn scale = 1		−0.128** (0.061)		−0.088 (0.132)
Unemployed × sn scale = 1		−0.291* (0.153)		−0.259 (0.332)
Homemaker × sn scale = 1		−0.241 (0.452)		−0.862 (0.980)
sn scale = 3
Retired × sn scale = 3		−0.028 (0.072)		0.015 (0.156)
Unemployed × sn scale = 3		−0.137 (0.208)		0.243 (0.451)
Homemaker × sn scale = 3		0.058 (0.619)		0.027 (1.343)
sn scale = 4
Retired × sn scale = 4		−0.283 (0.159)		0.249 (0.345)
Unemployed × sn scale = 4		−0.453 (0.489)		0.613 (1.059)
Homemaker × sn scale = 4		1.098 (1.328)		1.477 (2.882)
Intercept	5.948*** (0.698)	5.875*** (0.699)	10.884*** (1.569)	10.858*** (1.571)
Random effects: Var.
country	0.461	0.461	6.46	6.463
country-individuals	3.665	3.662	18.231	18.229
Residual	5.414	5.415	24.957	24.962
Number of observations	63,017	63,017	63,017	63,017

Women. Episodic memory is significantly negatively associated with being retired (β = −0.163, p < 0.01), unemployed (β = −0.410, p < 0.01) and being a homemaker (β = −0.405, p < 0.01), in reference to being employed (Table 2, model 1a). Model 1a further reveals a positive association between episodic memory and the sn scale (reference: sn scale = 2).

In verbal fluency among women (Table 2, model 1b), being retired (β = −0.613, p < 0.01), unemployed (β = −0.745, p < 0.01), and being a homemaker (β = −1.135, p < 0.01) show negative associations. Furthermore, model 1b positively relates verbal fluency to the sn scale.

Men. Among men (Table 3, model 1a-b), unemployment shows a pronounced negative association with episodic memory (β = −0.453, p < 0.01) in comparison to employment. We also find a significant negative association for being retired (β = −0.273, p < 0.01). In addition, we observe a positive association between episodic memory and the sn scale. Notably, a complete lack of social connections (sn scale = 0) exhibits a particularly pronounced negative relationship with episodic memory (β = −0.618, p < 0.01) in men.

Verbal fluency of men is negatively associated with being retired (β = −0.654, p < 0.01) and being unemployed (β = −1.043, p < 0.01), and positively associated with the sn scale (Table 3, model 1b). As in episodic memory, the association between the lack of a social network (sn scale = 0) and verbal fluency in men is substantial in size (β = −1.039, p < 0.01).

In summary, models 1a-b hint at a positive relationship between employment and cognitive functioning, as well as a positive relationship between social network strength and cognitive functioning. While women show substantial positive associations between higher sn scale levels and cognitive functioning, men exhibit a pronounced negative relationship between the lack of a social network and cognitive functioning.

The Social Network as a Moderator

To examine the moderating role of social network strength on the relationship between employment status and cognitive functioning, we conducted multilevel models, adding interaction terms between employment status (reference: employed or self-employed) and the sn scale levels (reference: sn scale = 2) (Tables 2 and 3, models 2a-b). Figure 2 visualizes the estimated marginal means with 95% confidence intervals.

Figure 2.

Episodic Memory and Verbal Fluency by Gender and Employment Status: Estimated Marginal Means With 95% Confidence Intervals for sn scale = 0, sn scale = 1, sn scale = 2, sn scale = 3, and sn scale = 4. Note: Own calculations on SHARE waves 4, 6, and 8 (82,095 observations of women and 63,019 observations of men). Discrete predictors are held constant at their proportions

Women. Table 2 (Model 2a) shows that for women, the association between being retired and episodic memory is mostly positively moderated by the sn scale (sn scale = 0: β = −0.370, p < 0.05; sn scale = 1: β = −0.163, p < 0.05; sn scale = 3: β = 0.172, p < 0.01), but no significant interaction is observed for sn scale = 4. For unemployed women, we find a pronounced positive interaction for sn scale = 3 (β = 0.582, p < 0.01), but no significant interactions for the other sn scale levels, where the number of observations is relatively low. For homemaking women, our results reveal significant positive interactions (sn scale = 0: β = −0.522, p < 0.05; sn scale = 1: β = −0.151, p < 0.1; sn scale = 3: β = 0.138, p < 0.1; sn scale = 4: β = 0.344, p < 0.05).

For women’s verbal fluency (Table 2, model 2b), we find positive interactions between homemaking and social network strength for below-reference levels (sn scale = 0: β = −1.175, p < 0.05; sn scale = 1: β = −0.310, p < 0.1).

Men. For episodic memory in men (Table 3, model 2a), retirement shows significant positive interactions with the social network for levels up to the reference level (sn scale = 0: β = −0.326, p < 0.05; sn scale = 1: β = −0.128, p < 0.05), while we observe no significant interactions for higher sn scale levels. Similarly, we find positive interactions between unemployment in men and the sn scale up to the reference level: a reference level score (sn scale = 2) as opposed to sn scale = 0 exhibits a pronounced positive interaction, whereas the positive interaction is considerably lower in magnitude and only marginally significant for sn scale = 1 (sn scale = 0: β = −0.936, p < 0.01; sn scale = 1: β = −0.291, p < 0.1). For verbal fluency in men, we find no significant interactions between employment status and the sn scale.

For both genders, models 2a-b show that most statistically significant coefficients correspond to modest adjusted differences between employment statuses (<0.15 SD). Substantively larger gaps are concentrated among respondents with low sn scale scores, such as in unemployed men with sn scale = 0 (−0.34 SD in episodic memory) and homemaking women with sn scale = 0 (−0.25 SD in episodic memory).

Robustness and Sensitivity Checks

To test the stability of our findings, we conducted several robustness and sensitivity checks (Tables S.3-10; Figures S.1-2).

Excluding SHARE wave 8 observations, results remain consistent with our main models in terms of statistical significance and direction of effects (Table S.3). When excluding respondents with no social network (sn scale = 0), the positive moderating role of the social network persists for both genders, with women showing significant interactions for retirement and homemaking across higher sn scale levels, and men continuing to show pronounced interactions for retirement and unemployment at sn scale = 1 (Table S.4). The results are robust to the exclusion of cognitively impaired respondents (Table S.5). The three-way interactions (employment status × sn scale × gender) (Table S.6) reveal that the moderating role of the sn scale differs significantly by gender for unemployment and retirement in selected sn scale scores, consistently favoring women. Restricting the sample to respondents aged 50-69 yields results mostly consistent with the main models (Table S.7). The moderating patterns remain largely intact: for women, stronger social networks continue to buffer the negative associations of non-employed statuses for episodic memory. For men, the protective role of the social network at lower levels (sn scale = 0 and 1) persists for retirement and unemployment in episodic memory. Using the continuous raw score underlying the sn scale (ranging from 0 to 20) instead of the categorical measure (Table S.8) confirms positive interactions between non-employment and the social network for both genders.

We re-estimated in a sensitivity check the gender-stratified multilevel models 2a-b, replacing the sn scale with two of its components (i) diversity of contacts and (ii) very close contacts (Tables S.9-10; Figure S.1-2 of the supplementary material). Among women, a greater diversity of contacts is associated with higher predicted episodic memory in retirees and homemakers compared to employed respondents. Episodic memory patterns for very close contacts are less uniform. For verbal fluency, diversity shows selective positive interactions, while they are mostly non-significant for very close contacts (Table S.9). Among men, the diversity of contacts shows comparatively limited interactions with non-employment concerning episodic memory, yet lower levels of very close contacts are linked to lower episodic memory scores for retirees and the unemployed, compared to the employed. Verbal fluency interactions remain mostly non-significant (Table S.10).

Discussion

This study examined how employment status (employed, retired, unemployed, homemaker) relates to cognitive functioning and whether social network strength moderates these associations differently for women and men. Overall, the findings suggest a positive moderating role of the social network for non-employed statuses, revealing gender-specific patterns. For women’s episodic memory, our results indicate a buffering role of social ties when employment-related cognitive stimulation is absent. For men, moderation effects are concentrated among individuals with weak or absent social ties.

In aging societies, it becomes increasingly important to identify contributors to maintaining good cognitive functioning into advanced age and reduce inequalities in cognitive functioning. Previous studies examining the interplay between older people’s activity participation and their social network and its association with cognitive functioning have predominantly focused on activities considered as social engagement, such as voluntary work or participation in clubs or community organizations (Litwin & Stoeckel, 2016; Paiva et al., 2021). These studies found more pronounced associations between social network strength and cognitive functioning for less socially engaged individuals. Unlike prior studies, which mainly focus on social engagement activities, our study analyzes employment status from a gender-informed perspective and tests whether social networks moderate its association with cognitive functioning, thus addressing an understudied mechanism for cognitive resilience in later life.

The hypothesis of a positive interaction between non-employment and social network strength is mostly confirmed for women’s episodic memory, underscoring a potential additional importance of social connections for older women who are retired or homemakers. For men, the findings are more mixed. A weak social network appears to exacerbate the adverse effects of unemployment and retirement on episodic memory, with particularly large negative associations for unemployed men lacking any social ties. For verbal fluency, less support is found for our hypothesis. Negative associations with a weak social network are observed primarily among homemaking women and unemployed men. The weaker results for verbal fluency align with existing evidence indicating that aging has a more adverse impact on episodic memory compared to semantic memory, which is assessed in the verbal fluency test (Levine et al., 2002). We refrain from interpreting estimates for homemaking men or drawing gender comparisons for homemakers, considering their low number of observations.

Motivated by research showing gender differences in the roles of weak ties (cognitive stimulation) and very close ties (support), we examined the moderating roles of the diversity of contacts and very close contacts on the relationship between employment status and cognitive functioning. Following prior literature, we conceptualize the diversity of contacts as bridging ties that provide cognitively stimulating exposure to novel social interactions, whereas very close contacts represent bonding ties that supply emotional support (Liao & Scholes, 2017; Tomioka et al., 2018).

The results suggest that the observed moderation patterns may be driven by different social network dimensions for women and men. For women, higher diversity of contacts is consistently linked to higher predicted episodic memory in retirees and homemakers, suggesting that diverse ties may partly substitute for employment-related cognitive stimulation. Among men, very close (bonding) ties appear to be more relevant moderators for retirement and unemployment, suggesting that emotionally close ties may be particularly important when employment-related social interaction is absent. These findings align with evidence linking women’s diversified, friend-rich networks to improved cognition (Tomioka et al., 2018; Zunzunegui et al., 2003) and men’s particular reliance on close, often spouse-centered ties (bonding ties), for socioemotional resources, which may in turn affect cognitive functioning (Liao & Scholes, 2017).

Theoretically, our findings underscore the notion that social networks operate as role-contingent resources that shape cognitive resilience in later life. Furthermore, our study highlights gendered patterns in the relationship between employment status, social network strength, and cognitive functioning, pointing to structural differences in how older adults access and benefit from social and cognitive resources. The substantial estimated benefits of a strong social network for women’s episodic memory align with prior research, showing that women benefit from broad social networks that are more emotionally supportive and functionally diverse than men’s (Liebler & Sandefur, 2002; Zunzunegui et al., 2003). These networks may act as substitute forms of cognitive stimulation when employment is absent, supporting the ‘use it or lose it’ hypothesis within gendered social realities. In contrast, men often rely more heavily on their spouses for emotional support, which may lead to the cognitive functioning – non-employment relationship being more pronounced in the absence of those core connections (Liao & Scholes, 2017; Schwartz & Litwin, 2018). Decomposing the sn scale into diversity of contacts and very close contacts supports this interpretation, suggesting that men’s episodic memory may be more contingent on the presence of core, very close contacts, while women’s outcomes may be more responsive to cognitively stimulating, diverse networks. From a gender-relational perspective (Risman, 2004), these findings reflect how socially constructed roles and expectations differentially position men and women to accumulate or deplete forms of social capital that support cognitive resilience in later life.

Limitations

The study has some important limitations. First, our models rely on current employment status and do not capture heterogeneity in cognitive demands of the work respondents perform or have performed in the past. This matters, first, because the theoretical argument linking employment to cognition rests on differences in cognitive stimulation that may vary between professions and, second, in the light of the cognitive reserve theory, arguing that lifelong cognitively stimulating experiences (e.g. employment) may protect against age-related decline (Stern et al., 2020). However, SHARE provides only broad measures of the current occupation group for employed respondents (International Standard Classification of Occupations 2008 major group), constraining interpretations concerning occupational cognitive demand.

Second, sample composition and selection may affect generalizability. SHARE surveys community-dwelling older adults, which may result in underrepresentation of individuals with severe health limitations. In addition, the employment status groups differ markedly in age, raising the possibility of age-related selection (e.g., survivor bias among retirees). We address this concern by flexibly controlling for age and by conducting robustness checks restricting the sample to ages 50-69; nevertheless, selection processes may still influence the estimated associations. Furthermore, although robustness checks yielded consistent results, selective attrition cannot be fully ruled out in longitudinal survey data.

Third, our study design estimates adjusted associations rather than causal effects. For instance, reverse mechanisms remain plausible. Cognitive decline can affect employment status (e.g., ability to work) and may also lead to social withdrawal (Dyer et al., 2021; Hajek & König, 2025). Furthermore, although the multilevel models account for the nested structure of the data, they cannot rule out confounding through unobserved characteristics, such as early-life cognitive abilities, that may jointly shape employment, social networks, and later-life cognition (Hofer & Clouston, 2014). Gender-specific moderation patterns should be interpreted cautiously, as they may partly reflect underlying differences in baseline distributions and sample composition, including selection into employment and social network categories, rather than distinct underlying mechanisms. We mitigate these concerns with robustness checks, excluding respondents reporting diagnosed cognitive impairment and respondents without social ties; however, the observational design still does not rule out reverse causality or unobserved confounding.

Conclusion

The present study provides important insights, using the sn scale as a proxy for social network strength, which considers functional and structural aspects, allowing us to draw a more comprehensive picture of participants’ social relationships. The results demonstrate that social networks may compensate for the cognitive stimulation lost from non-employment, with women benefiting from diverse networks and men relying on close ties. These findings have practical implications for promoting cognitive health among non-employed older adults and highlight the importance of considering gender as a structural factor shaping access to cognitively protective resources.

Future research and data collection on later life cognitive functioning may consider work histories and social networks to identify causal processes. Furthermore, it should unpack how institutionalized gender roles and their intersection with class and family structures shape gendered trajectories of cognitive aging and access to social capital in older adulthood.

Supplemental Material

Supplemental material - Employment Status and Later Life Cognitive Functioning: A Gendered Perspective on the Moderating Role of the Social Network

Supplemental material for Employment Status and Later Life Cognitive Functioning: A Gendered Perspective on the Moderating Role of the Social Network by Thomas Arnhold, Daniela Weber and Valeria Bordone in Research on Aging

Footnotes

Acknowledgments

This paper uses data from SHARE Wave 4 (doi: https://doi.org/10.6103/SHARE.w4.900), Wave 6 (doi: https://doi.org/10.6103/SHARE.w6.900), and Wave 8 (doi: https://doi.org/10.6103/SHARE.w8.900) see Börsch-Supan et al. (2013) for methodological details. The SHARE data collection has been funded by the European Commission, DG RTD through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, DASISH: GA N°283646) and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, SHARE-COVID19: GA N°101015924) and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, VS 2020/0313 and SHARE-EUCOV: GA N°101052589 and EUCOVII: GA N°101102412. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the U.S. National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, BSR12-04, R01_AG052527-02, HHSN271201300071 C, RAG052527 A) and from various national funding sources is gratefully acknowledged (see ).

Thomas Arnhold

Daniela Weber

Valeria Bordone

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Vienna Science and Technology Fund [grant number LS22-008] and APART-GSK of the Austrian Academy of Sciences [grant number APART-GSK/11844].

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental material for this article is available online.

Author Biographies

Thomas Arnhold is a researcher at the International Institute for Applied Systems Analysis and a PhD candidate at the Vienna University of Economics and Business. His research focuses on employment and social relationships across the life course and their connection to healthy aging.

Daniela Weber is a Senior Research Scholar at the International Institute for Applied Systems Analysis and an Assistant Professor at the Vienna University of Economics and Business. Her interdisciplinary research focuses on the influence of environmental factors on healthy aging, adopting a comprehensive life-course perspective.

Valeria Bordone is an Associate Professor for Life course, Generations, Ageing at the Department of Sociology of the University of Vienna. Her research mainly focuses on social inequalities across the life span, family dynamics, and sustainable ageing.

References

Ajrouch

K. J.

Blandon

A. Y.

Antonucci

T. C.

(2005). Social networks among men and women: The effects of age and socioeconomic status. The Journals of Gerontology: Series B, 60(6), S311–S317. https://doi.org/10.1093/geronb/60.6.S311

Alvarez-Bueno

Cavero-Redondo

Jimenez-Lopez

Visier-Alfonso

M. E.

Sequi-Dominguez

Martinez-Vizcaino

(2021). Effect of retirement on cognitive function: A systematic review and meta-analysis. Occupational and Environmental Medicine, 78(10), 761–768. https://doi.org/10.1136/oemed-2020-106892

Antonucci

T. C.

Ajrouch

K. J.

Birditt

K. S.

(2014). The convoy model: Explaining social relations from a multidisciplinary perspective. The Gerontologist, 54(1), 82–92. https://doi.org/10.1093/geront/gnt118

Antonucci

T. C.

Birditt

K. S.

Sherman

C. W.

Trinh

(2011). Stability and change in the intergenerational family: A convoy approach. Ageing and Society, 31(7), 1084–1106. https://doi.org/10.1017/S0144686X1000098X

Arnhold

Szenkurök

Weber

(2025). Mapping Inequalities in the Health of Older Adults Around the World: Heterogeneities in cognitive and physical functioning (pp. 515–536). Vienna Yearbook of Population Research. https://doi.org/10.1553/p-mcm9-5b3b

Asperholm

Högman

Rafi

Herlitz

(2019). What did you do yesterday? A meta-analysis of sex differences in episodic memory. Psychological Bulletin, 145(8), 785–821. https://doi.org/10.1037/bul0000197

Bergman

Kneip

Scherpenzeel

(2019). Survey participation in the survey of health, ageing and retirement in Europe (SHARE), wave 1-7. Based on release 7.0.0. SHARE working paper series 41-2019. Munich. MEA, Max Planck Institute for Social Law and Social Policy.

Berkman

L. F.

Glass

Brissette

Seeman

T. E.

(2000). From social integration to health: Durkheim in the new millennium. Social Science & Medicine, 51(6), 843–857. https://doi.org/10.1016/S0277-9536(00)00065-4

Berkman

L. F.

Krishna

(2014). Social network epidemiology. In: SocialEpidemiology (2nd ed., pp. 234–289). Oxford University Press.

10.

Bertogg

Leist

A. K.

(2023). Gendered life courses and cognitive functioning in later life: The role of context-specific gender norms and lifetime employment. European Journal of Ageing, 20(1), 7. https://doi.org/10.1007/s10433-023-00751-4

11.

Börsch-Supan

Brandt

Hunkler

Kneip

Korbmacher

Malter

Schaan

Stuck

Zuber

on behalf of the SHARE Central Coordination Team . (2013). Data resource profile: The survey of health, ageing and retirement in Europe (SHARE). International Journal of Epidemiology, 42(4), 992–1001. https://doi.org/10.1093/ije/dyt088

12.

Börsch-Supan

Schuth

(2013). Early retirement, mental health and social networks. In: Börsch-Supan

Brandt

Litwin

Weber

(Eds.), Active ageing and solidarity between generations in Europe (pp. 337–348). De Gruyter. https://doi.org/10.1515/9783110295467.337

13.

Burnett-Wolle

Godbey

(2007). Refining research on older adults’ leisure: Implications of selection, optimization, and compensation and socioemotional selectivity theories. Journal of Leisure Research, 39(3), 498–513. https://doi.org/10.1080/00222216.2007.11950119

14.

Burt

R. S.

Author (2007). Brokerage and closure: An introduction to social capital. Oxford University Press.

15.

Collie

Maruff

Darby

D. G.

McStephen

(2003). The effects of practice on the cognitive test performance of neurologically normal individuals assessed at brief test–retest intervals. Journal of the International Neuropsychological Society, 9(3), 419–428. https://doi.org/10.1017/S1355617703930074

16.

Dyer

A. H.

Murphy

Lawlor

Kennelly

S. P.

Study Group, for the NILVAD . (2021). Social networks in mild-to-moderate alzheimer disease: Longitudinal relationships with dementia severity, cognitive function, and adverse events. Aging & Mental Health, 25(10), 1923–1929. https://doi.org/10.1080/13607863.2020.1745146

17.

Ellwardt

Van Tilburg

T. G.

Aartsen

M. J.

(2015). The mix matters: Complex personal networks relate to higher cognitive functioning in old age. Social Science & Medicine, Special Issue: Social Networks, Health and Mental Health, 125(1), 107–115. https://doi.org/10.1016/j.socscimed.2014.05.007

18.

England

(2010). The gender revolution: Uneven and stalled. Gender & Society, 24(2), 149–166. https://doi.org/10.1177/0891243210361475

19.

Eurostat . (2020). Ageing Europe: Looking at the lives of older people in the EU: 2020 edition. https://doi.org/10.2785/628105

20.

Fanelli

Profeta

(2021). Fathers’ involvement in the family, fertility, and maternal employment: Evidence from central and eastern Europe. Demography, 58(5), 1931–1954. https://doi.org/10.1215/00703370-9411306

21.

Fiori

K. L.

Smith

Antonucci

T. C.

(2007). Social network types among older adults: A multidimensional approach. The Journals of Gerontology: Series B, 62(6), P322–P330. https://doi.org/10.1093/geronb/62.6.P322

22.

Fisher

G. G.

Stachowski

Infurna

F. J.

Faul

J. D.

Grosch

Tetrick

L. E.

(2014). Mental work demands, retirement, and longitudinal trajectories of cognitive functioning. Journal of Occupational Health Psychology, 19(2), 231–242. https://doi.org/10.1037/a0035724

23.

Fratiglioni

Paillard-Borg

Winblad

(2004). An active and socially integrated lifestyle in late life might protect against dementia. The Lancet Neurology, 3(6), 343–353. https://doi.org/10.1016/S1474-4422(04)00767-7

24.

Gonzalez-Burgos

Hernández-Cabrera

J. A.

Westman

Barroso

Ferreira

(2019). Cognitive compensatory mechanisms in normal aging: A study on verbal fluency and the contribution of other cognitive functions. Aging (Albany NY), 11(12), 4090–4106. https://doi.org/10.18632/aging.102040

25.

Goodman

W. K.

Geiger

A. M.

Wolf

J. M.

(2017). Leisure activities are linked to mental health benefits by providing time structure: Comparing employed, unemployed and homemakers. Journal of Epidemiology & Community Health, 71(1), 4–11. https://doi.org/10.1136/jech-2016-207260

26.

Granovetter

M. S.

(1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380. https://doi.org/10.1086/225469

27.

Grasset

Proust-Lima

Mangin

J.-F.

Habert

M.-O.

Dubois

Paquet

Hanon

Gabelle

Ceccaldi

Annweiler

David

Jonveaux

Belin

Julian

Rouch-Leroyer

Pariente

Locatelli

Chupin

Chêne

on behalf of the Memento Cohort Study group . (2022). Explaining the association between social and lifestyle factors and cognitive functions: A pathway analysis in the Memento cohort. Alzheimer's Research & Therapy, 14(1), 68. https://doi.org/10.1186/s13195-022-01013-8

28.

Greenberg

Burgard

(2021). Cumulative employment intensity and complexity across the life course and cognitive function in later life among European women and men. Annals of Epidemiology, 58(1), 83–93. https://doi.org/10.1016/j.annepidem.2021.01.006

29.

Grotz

Meillon

Amieva

Andel

Dartigues

J.-F.

Adam

Letenneur

(2018). Occupational social and mental stimulation and cognitive decline with advancing age. Age and Ageing, 47(1), 101–106. https://doi.org/10.1093/ageing/afx101

30.

Hajek

König

H.-H.

(2025). Prevalence of loneliness and social isolation among individuals with mild cognitive impairment or dementia: Systematic review and meta-analysis. BJPsych Open, 11(2), Article e44. https://doi.org/10.1192/bjo.2024.865

31.

Hakanen

J. J.

Schaufeli

W. B.

(2012). Do burnout and work engagement predict depressive symptoms and life satisfaction? A three-wave seven-year prospective study. Journal of Affective Disorders, 141(2), 415–424. https://doi.org/10.1016/j.jad.2012.02.043

32.

Haynes

Banks

Hill

(2014). The relationship between employment and social networks in the older population: A comparative European study. International Journal of Social Economics, 41(4), 321–335. https://doi.org/10.1108/IJSE-10-2012-0201

33.

Herlitz

Dekhtyar

Asperholm

Weber

(2015). Gender differences in memory and cognition. In Pachana

N. A.

(Ed.), Encyclopedia of geropsychology (pp. 1–7). Springer. https://doi.org/10.1007/978-981-287-080-3_225-1

34.

Hofer

S. M.

Clouston

(2014). On the importance of early-life cognitive abilities in shaping later-life outcomes. Research in Human Development, 11(3), 241–246. https://doi.org/10.1080/15427609.2014.936173

35.

Ice

Ang

Greenberg

Burgard

(2020). Women’s work-family histories and cognitive performance in later life. American Journal of Epidemiology, 189(9), 922–930. https://doi.org/10.1093/aje/kwaa042

36.

Kahn

R. L.

Antonucci

T. C.

(1980). Convoys over the life course: Attachment, roles, and social support. In: Life-span development and behavior (pp. 253–286). Academic Press.

37.

Kelly

M. E.

Duff

Kelly

McHugh Power

J. E.

Brennan

Lawlor

B. A.

Loughrey

D. G.

(2017). The impact of social activities, social networks, social support and social relationships on the cognitive functioning of healthy older adults: A systematic review. Systematic Reviews, 6(1), 259. https://doi.org/10.1186/s13643-017-0632-2

38.

Levine

Svoboda

Hay

J. F.

Winocur

Moscovitch

(2002). Aging and autobiographical memory: Dissociating episodic from semantic retrieval. Psychology and Aging, 17(4), 677–689. https://doi.org/10.1037/0882-7974.17.4.677

39.

Liao

Scholes

(2017). Association of social support and cognitive aging modified by sex and relationship type: A prospective investigation in the English longitudinal study of ageing. American Journal of Epidemiology, 186(7), 787–795. https://doi.org/10.1093/aje/kwx142

40.

Liebler

C. A.

Sandefur

G. D.

(2002). Gender differences in the exchange of social support with friends, neighbors, and co-workers at midlife. Social Science Research, 31(3), 364–391. https://doi.org/10.1016/S0049-089X(02)00006-6

41.

Litwin

Stoeckel

K. J.

(2016). Social network, activity participation, and cognition: A complex relationship. Research on Aging, 38(1), 76–97. https://doi.org/10.1177/0164027515581422

42.

Llewellyn

D. J.

Lang

I. A.

Langa

K. M.

Huppert

F. A.

(2008). Cognitive function and psychological well-being: Findings from a population-based cohort. Age and Ageing, 37(6), 685–689. https://doi.org/10.1093/ageing/afn194

43.

Machado

Barroso

Molina

Nieto

Díaz-Flores

Westman

Ferreira

(2018). Proposal for a hierarchical, multidimensional, and multivariate approach to investigate cognitive aging. Neurobiology of Aging, 71(1), 179–188. https://doi.org/10.1016/j.neurobiolaging.2018.07.017

44.

Muller

J. S.

Hiekel

Liefbroer

A. C.

(2020). The long-term costs of family trajectories: Women’s later-life employment and earnings across Europe. Demography, 57(3), 1007–1034. https://doi.org/10.1007/s13524-020-00874-8

45.

Myrskylä

Margolis

(2014). Happiness: Before and after the kids. Demography, 51(5), 1843–1866. https://doi.org/10.1007/s13524-014-0321-x

46.

Paiva

A. F.

Cunha

Voss

Matos

A. D.

(2021). The interrelationship between social connectedness and social engagement and its relation with cognition: A study using SHARE data. Ageing and Society, 43(8), 1735–1753. https://doi.org/10.1017/S0144686X2100129X

47.

Pan

Chee

K. H.

(2020). The power of weak ties in preserving cognitive function: A longitudinal study of older Chinese adults. Aging & Mental Health, 24(7), 1046–1053. https://doi.org/10.1080/13607863.2019.1597015

48.

Perry

B. L.

McConnell

W. R.

Peng

Roth

A. R.

Coleman

Manchella

Roessler

Francis

Sheean

Apostolova

L. A.

(2022). Social networks and cognitive function: An evaluation of social bridging and bonding mechanisms. The Gerontologist, 62(6), 865–875. https://doi.org/10.1093/geront/gnab112

49.

Pfund

G. N.

James

B. D.

Willroth

E. C.

(2025). Bidirectional relationships between well-being and cognitive function. Psychological Science, 36(5), 350–366. https://doi.org/10.1177/09567976251335578

50.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Appl. Data anal. Methods (second). Sage Publications, Inc.

51.

Risman

B. J.

(2004). Gender as a social structure: Theory wrestling with activism. Gender & Society, 18(4), 429–450. https://doi.org/10.1177/0891243204265349

52.

Roth

A. R.

(2022). Social connectedness and cognitive decline. The Lancet. Healthy Longevity, 3(11), e723–e724. https://doi.org/10.1016/S2666-7568(22)00217-3

53.

Rowe

J. W.

Kahn

R. L.

(1997). Successful aging. The Gerontologist, 37(4), 433–440. https://doi.org/10.1093/geront/37.4.433

54.

Saenz

J. L.

Downer

Garcia

M. A.

Wong

(2018). Cognition and context: Rural-urban differences in cognitive aging among older Mexican adults. Journal of Aging and Health, 30(6), 965–986. https://doi.org/10.1177/0898264317703560

55.

Salthouse

(2006). Mental exercise and mental aging: Evaluating the validity of the “Use It or Lose It” Hypothesis. Perspectives on Psychological Science, 1(1), 68–87. https://doi.org/10.1111/j.1745-6916.2006.00005.x

56.

Salthouse

(2012). Consequences of age-related cognitive declines. Annual Review of Psychology, 63(1), 201–226. https://doi.org/10.1146/annurev-psych-120710-100328

57.

Schwartz

Litwin

(2018). Social network changes among older Europeans: The role of gender. European Journal of Ageing, 15(4), 359–367. https://doi.org/10.1007/s10433-017-0454-z

58.

SHARE-ERIC . (2024a). Survey of health, ageing and retirement in Europe (SHARE) wave 4. Release version: 9.0.0. Author. [Data set]. https://doi.org/10.6103/SHARE.w4.900

59.

SHARE-ERIC . (2024b). Survey of health, ageing and retirement in Europe (SHARE) wave 6. Release version: 9.0.0. Author. [Data set]. https://doi.org/10.6103/SHARE.w6.900

60.

SHARE-ERIC . (2024c). Survey of health, ageing and retirement in Europe (SHARE) wave 8. Release version: 9.0.0. Author. [Data set]. https://doi.org/10.6103/SHARE.8.900

61.

Skirbekk

Loichinger

Weber

(2012). Variation in cognitive functioning as a refined approach to comparing aging across countries. Proceedings of the National Academy of Sciences, 109(3), 770–774. https://doi.org/10.1073/pnas.1112173109

62.

Stern

Arenaza-Urquijo

E. M.

Bartrés-Faz

Belleville

Cantilon

Chetelat

Ewers

Franzmeier

Kempermann

Kremen

W. S.

Okonkwo

Scarmeas

Soldan

Udeh-Momoh

Valenzuela

Vemuri

Vuoksimaa

the Reserve, Resilience and Protective Factors PIA Empirical Definitions and Conceptual Frameworks Workgroup . (2020). Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimer's and Dementia: The Journal of the Alzheimer’s Association, 16(9), 1305–1311. https://doi.org/10.1016/j.jalz.2018.07.219

63.

Stevens-Ratchford

R. G.

(2011). Longstanding occupation: The relation of the continuity and meaning of productive occupation to life satisfaction and successful aging. Activities Adaptation & Aging, 35(2), 131–150. https://doi.org/10.1080/01924788.2011.574255

64.

Tomioka

Kurumatani

Hosoi

(2018). Social participation and cognitive decline among community-dwelling older adults: A community-based longitudinal study. The Journals of Gerontology: Series B, 73(5), 799–806. https://doi.org/10.1093/geronb/gbw059

65.

Vance

D. E.

Bail

Enah

C. C.

Palmer

J. J.

Hoenig

A. K.

(2016). The impact of employment on cognition and cognitive reserve: Implications across diseases and aging. Nursing: Research and Reviews, 6(1), 61–71. https://doi.org/10.2147/NRR.S115625

66.

Vélez-Coto

Rute-Pérez

Pérez-García

Caracuel

(2021). Unemployment and general cognitive ability: A review and meta-analysis. Journal of Economic Psychology, 87(1), Article 102430. https://doi.org/10.1016/j.joep.2021.102430

67.

Weber

Skirbekk

Freund

Herlitz

(2014). The changing face of cognitive gender differences in Europe. Proceedings of the National Academy of Sciences of the United States of America, 111(32), 11673–11678. https://doi.org/10.1073/pnas.1319538111

68.

White

Arp

Coe

(2025). Retirement and cognitive function in later life. Social Science Research Network. (SSRN Scholarly Paper No. 5334947). https://doi.org/10.2139/ssrn.5334947

69.

Zunzunegui

M.-V.

Alvarado

B. E.

Del Ser

Otero

(2003). Social networks, social integration, and social engagement determine cognitive decline in community-dwelling Spanish older adults. The Journals of Gerontology: Series B, 58(2), S93–S100. https://doi.org/10.1093/geronb/58.2.S93

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.83 MB

0.00 MB