Life course engagement in enriching activities: When and how does it matter for cognitive aging?

Abstract

Growing evidence suggests that participation in enriching activities (physical, social, and mental) across the life course is beneficial for cognitive functioning in older age. However, few studies have examined the effects of enrichment across the entire life course within the same participants. Using 2931 participants in the Health and Retirement Study, we linked self-report data from later life and retrospective self-report data from early life and midlife to cognitive performance after age 65. We categorized participants as having either high (top ~25%) or average-to-low (bottom ~75%) level of enrichment during each life period. Thus, eight groups were identified that reflected unique patterns of enrichment during early, mid, and later life (e.g., high-high-high). Using growth curve modeling, we found that life course enrichment patterns predicted both cognitive functioning and the rate of cognitive decline across five time points spanning eight years (Aim 1). Groups with high enrichment during at least one life period had higher performance and slower decline in older age, compared to those who had average to low levels of enrichment throughout all three life periods. We also found that high enrichment during each life period independently predicted better cognitive performance and that high enrichment during early and later life also predicted slower cognitive decline (Aim 2). These findings support the idea that high enrichment is beneficial for cognition in later life and that the effects are long-lasting, even when individuals are inconsistent in enrichment engagement throughout the entire life course.

Since the year 2000, there have been increasing efforts to determine modifiable social and behavioral factors that promote healthy cognitive aging. There is growing evidence that enrichment, that is, participation in physical, social, or intellectually engaging activities, may contribute to the maintenance of cognitive functioning in later life and minimize performance decline (see Hertzog et al., 2008 for a review). To date, the majority of empirical work on enrichment focuses on just one life stage, rather than engagement in cognitively enriching activities across the life course. In part, this is because there are few longitudinal studies that commenced in the mid-20^th century and continue into the 21^st century (for exceptions see Gow, Patty, & Deary, 2017; Schaie & Willis, 2010). Additionally, most studies focus on specific activity domains, rather than more comprehensive multi-domain measures of enrichment.

In the present study, we use current and retrospective self-reports of engagement in enriching activities to examine the predictive value of high enrichment during early, mid, and later-life on cognitive performance and rates of cognitive decline in older age. Much of the existing literature focuses on identifying detriments to cognition, but it is also necessary to identify factors that have a positive influence on cognition (Nyberg & Pudas, 2019). This may inform future interventions designed to improve cognitive health and protect against cognitive decline.

Enriching Activities and the Concept of Cognitive Reserve

During older adulthood, it is typical to see decline across various cognitive domains, with changes more notable after age 65 (Rönnlund et al., 2005; Nilsson et al., 2009). However, there is variability in who is affected by these changes, how much decline there is, and the rate of change over time. This may be due to varying levels of cognitive reserve—a hypothetical construct used to explain individual differences in susceptibility to cognitive impairment (Stern, 2009; 2012). As such, individuals possessing greater amounts of cognitive reserve are theorized to maintain more resources and be better able to cope with higher levels of brain pathology or other neurological insults before reaching the clinical threshold of impairment (Stern, 2009, Stern et al., 2020). Cognitive reserve may reflect an active process, in which the brain actively compensates for neural degradation (Stern, 2002). Alternatively, although not mutually exclusive, it may reflect a more passive process, in which individuals with higher reserve maintain higher functional levels over time (Stern, 2002). Regardless of the specific mechanisms underlying cognitive reserve, increased reserve is hypothesized to lead to better performance and less impairment.

Importantly, implicit in these theories is the possibility that cognitive reserve might be a modifiable protective factor (Livingston, 2017). Cognitive reserve interventions are believed to optimize the protective socio-behavioral factors assumed to contribute to cognitive reserve across the life course and minimize risks associated with vascular disease. These interventions also seek to delay the onset of cognitive decline even among individuals with a genetic disposition to dementia and reduce the population prevalence of Alzheimer’s disease and Alzheimer’s-disease-related-dementias (Langa et al., 2015; 2017). One such intervention that has been hypothesized to increase cognitive reserve is the engagement in enriching activities.

Consistent with the literature, we define enrichment as a broad category of mental, social, and physical activities that may enhance cognitive functioning (Hertzog et al., 2008). Distinct from the concept of engagement, enrichment can be considered as the utility received from doing an activity, while engagement is the act of participation itself. However, these concepts are closely related. Since engagement is considered to be the ongoing investment of personal resources (Stine-Morrow & Manavbasi, 2022), it may take ongoing participation—that is, engagement—in enriching activities in order to reap the benefits. Although the operationalization of enrichment in empirical research varies, there is general consensus about the types of activities that contribute to cognitive reserve. These include education, exercise/physical activity, hobbies, games, social/community engagement, technology use, traveling, relaxation, and developmental/educational activities (Hertzog et al., 2008; Jopp & Herzog, 2010, Nyberg et al., 2019). There is also considerable evidence that enriching activities have a positive impact on cognition in later life. As such, it is hypothesized that engagement in these types of activities increases cognitive reserve (Hertzog et al., 2008), which may be differentially malleable in early, middle, and later life (Fritsch et al., 2007). Below, we review the extant evidence that assesses the relationship between enrichment across the life course and cognitive performance level and the rate of change after age 65.

Early Life Enrichment

Enriching experiences in early life are hypothesized to increase levels of cognitive reserve and lead to better cognitive performance in later life. It has even been argued that cognitive reserve may be most amenable to change during childhood and adolescence (Fritsch et al., 2007). Examples of typical early life enrichment activities include early educational experiences and engagement in complex cognitive activities such as learning a second language, playing sports, or involvement in arts and music (e.g., playing an instrument, singing in a choir, participating in dance). Although the exact time periods of measurement vary, life-course literature typically considers early life as the time from birth through one’s 20s.

Perhaps the most frequently studied source of early life enrichment is education. Several studies have found that more education, as defined as the number of years of schooling completed or the highest degree attained, predicts higher cognitive levels after age 65 (Zahodne et al., 2015; Meng, 2012; Stern; 2009). Others find that education quality, as compared to quantity (i.e., years of education), during early life may be an even stronger predictor of later-life cognitive performance (Card & Krueger, 1992; Crowe, 2012; Sisco, 2015; Walsemann & Ailshire, 2020). There is also evidence that more general mental enrichment may also play a role. Indeed, there is a positive relationship between involvement in early-life cognitive activities (e.g., taking a foreign language, playing sports or instruments) and later life cognition (Chan et al., 2019; Morris et al., 2021; Wilson et al., 2015). Similarly, after-school activities and bilingualism are also associated with better cognitive aging (Bialystok, 2016). Additionally, Lee et al. (2020) found that growing up with grandparents living at home during childhood was beneficial in later life, pointing to an important role for social enrichment during childhood. Furthermore, Dik et al. (2003) found a positive relationship between early life physical activity and later-life processing speed as tested in men, indicating that regular exercise (one to ten hours per week) in early age remains cognitively advantageous many decades later. When tested in women, Middleton et al. (2010) found that those who were physically active (compared to those categorized as physically inactive) throughout the life course had lower odds of cognitive impairment in later life. However, they found the strongest association with physical activity during teenage years, as compared to all other time points throughout the life course. These studies provide strong initial evidence that cognitive, social, and physical enrichment during early life plays an important role in later life cognition.

Although there is ample support for the relationship between education in early life and the level of cognitive functioning later in life, findings are less consistent about the relationship between education and cognitive decline in later life (Lövdén et al., 2020). After classifying individuals as either maintainers (i.e., no change in cognitive performance) or decliners (i.e., decrease in cognitive performance), three studies demonstrated that level of education predicted which cognitive trajectory an individual would follow such that those with higher education were more likely to maintain their cognitive abilities (Habib et al., 2007; Josefsson et al. 2012). However, other longitudinal studies using growth curve modeling and Bayesian analyses, report that despite higher education being strongly related to the average level of functioning over time, it did not influence the rate of age-related cognitive decline (Berggeren, 2018; Cross, 2015; Wilson et al., 2009; Zahodne, 2011). A recent meta-analysis of education and rates of cognitive decline also supports the idea that the association between education and the rate of cognitive change is negligible (Seblova et al., 2020).

Beyond those that focus on education, relatively few studies have examined how early life experiences may relate to the rate of cognitive decline. This is likely due to a lack of longitudinal studies that collect data about childhood experiences. Thus, based on the current evidence, early-life enrichment appears to be positively associated with better later life cognitive functioning but the relationship with the rate of cognitive decline is less clear.

Midlife Enrichment

Midlife is often understudied, but it may uniquely contribute to our understanding of the relationship between enrichment and later-life cognition. Midlife also holds the most flexible age boundaries in life-course literature, ranging anywhere from age 30 to 65, depending on the study. Typical enriching activities during midlife may include the pursuit of additional degrees or professional certificates for some individuals, as well as occupational complexity. Since people in the workforce typically spend a large proportion of their waking hours at their job, occupational complexity provides a good account of the extent of mental stimulation occurring on a day-to-day basis. Other enrichment during this period includes physical activity, as well as social activities including partnership and living arrangements.

On average, most education occurs prior to age 18. But for people who seek college education and higher degrees after high school graduation, the process continues into early adulthood. Lachman et al. (2010) found that higher education was related to better cognition during midlife. These authors also found that more frequent engagement in cognitively complex activities (averaged across a six category response scale ranging from never (0) to daily participation (6)) was related to better memory and executive functioning during adulthood and may moderate the association between education level and cognition. Although this study did not follow participants over time to assess cognition and decline in later years, other work has found that an active cognitive lifestyle in midlife, primarily measured by occupational complexity, is associated with better cognitive performance later in life.

It is well established that during midlife, complexity of work has a positive relationship with cognitive performance (Schooler et al., 1999). A number of studies have also linked occupational complexity to cognition in later life such that higher complexity during working years predicts better cognitive functioning even after retirement (Fisher et al., 2014; Smart et al., 2014; Grzywacz et al., 2016; Fisher et al., 2017; Nexø et al., 2016). This pattern has been observed across several domains including processing speed, as well as both verbal and visuospatial abilities (Finkel et al., 2009). For example, individuals who worked as architects during midlife showed better visuospatial ability in older adulthood. Salthouse et al. (1990) proposed that this was due to the high visuospatial complexity required of their job. Additionally, Karp (2009) found that more mentally stimulating activities at work during midlife was associated with a lower risk of dementia, implicating a possible boost to cognitive reserve.

Moreover, Andel et al. (2016) and Potter et al (2006) found that occupational complexity may even attenuate the rate of cognitive aging over time. However, another longitudinal study found that higher occupational complexity only positively influences cognition, and not the rate of cognitive decline (Lane et al., 2017). Thus, at present, the relationship between occupational complexity and the speed of cognitive change is inconclusive.

Outside of work, Iso-Markku et al. (2016) found that participants who consistently engaged in vigorous physical activity during midlife had a lower chance of impaired cognition compared to persistently inactive participants. Similarly, Kraal et al. (2020) found that level of physical activity during midlife not only predicted higher levels of memory in later life, but more exercise was also associated with slower cognitive decline. In addition, more social engagement during middle age has been found to be associated with less cognitive decline later in life (Marioni et al., 2012; 2014). Similarly, Stieger and Lachman (2021) found that increasing cognitive activity during midlife may slow rates of cognitive decline in older age. Therefore, similar to the findings from early life, it appears that increased enrichment during midlife is associated with better later-life cognition and may potentially attenuate the rate of cognitive change.

Later Life Enrichment

There is also substantial evidence that engaging in enriching activities during later life affects cognition during the same period. Depending on the study, later life is often considered to begin around age 65 and spans the remainder of the life course, including the oldest-old (commonly defined as 85 years and older). Measures used to assess enrichment during later life typically include engagement in cognitively complex activities through hobbies and leisure, social network ties (e.g., size, frequency of contact), as well as participation in regular exercise.

Like early life and midlife, several studies have found that the frequency and intensity of complex cognitive activity during older age is related to level of cognitive functioning (Wilson et al., 1999; Schooler & Mulatu, 2001; Bielak et al., 2007). This relationship may be the strongest for activities with the highest complexity, such as novel information processing (Bielak et al., 2007). Additionally, intervention studies have found that sustained engagement (averaging 15 hours per week) in complex cognitive activities (e.g., taking a photography class or learning how to quilt) during older adulthood indeed has a positive relationship with cognitive functioning (Park et al., 2014). Another study found that engaging in less cognitive, but still enriching, activities related to arts and culture are also associated with better cognition (Fluharty et al., 2021). Furthermore, ties with a large social network that includes friends as well as family during older adulthood is also positively associated with cognitive performance (Sharifian et al., 2019). More specifically, frequency of contact with friends in later life (averaged across a six category response scale ranging from less than once per year/never (0) to three or more times per week (6)) is positively associated with cognitive performance across several domains including episodic memory, executive function, visuo-construction, processing speed, and language (Meister & Zahodne, 2021). In addition, a 6-month physical activity intervention study yielded a positive relationship between physical enrichment in later life and cognitive functioning (Jonasson et al., 2017).

Additional work has found that enrichment may also help to slow cognitive decline in later life. Hultsch et al. (1999) found that changes in intellectually engaging activities during older age were associated with changes in cognitive functioning such that increased participation was linked to slower decline. Similarly, reductions in cognitive lifestyle activities have been found to be associated with steeper declines in measures of verbal processing speed, as well as episodic and semantic memory (Small et al., 2012). Furthermore, James (2011) found that in older age, enhanced social network structure and function are associated with a slower rate of cognitive decline. Moreover, other research has shown that an increase in exercise during later life is inversely related to cognitive decline (Kramer et al., 2006). For example, Yaffe et al. (2001) found that women with higher levels of physical activity (measured by blocks walked and total kilocalories expended per week) at baseline were less likely to experience cognitive decline. Taken together, it appears that high enrichment in later life is associated with higher cognitive functioning and possibly slower cognitive decline.

Enrichment in Multiple Life Periods

Despite extensive evidence that enrichment is beneficial for cognition in early, mid, and later life, we located comparatively few studies that examine enrichment across more than one life stage within the same individuals. A selection of these studies are reviewed below. They are in accord with theories that consider successful aging to be a lifelong process (Rowe & Kahn, 1997) and that the acquisition and maintenance of cognitive reserve occurs over the life course.

Results from the studies that evaluate the effect of enrichment in multiple life periods support previous findings that cognitive activity during both childhood and middle age are related to later-life cognitive functioning (Wilson et al., 2005; Vermuri et al., 2014). Furthermore, one study that included retrospective self-reports of enrichment during early and middle adulthood found that more leisure activities (based on frequency of participation in 15 different activities) during midlife (40–55), but not early (20–35) or late (60–75) adulthood, was associated with better cognitive functioning at age 79 (Gow et al., 2017). In addition, Cheval and colleagues (2020) found that cognitive resources were positively associated with level of participation in physical activity across mid and later life and that lower cognitive resources were associated with a faster decline of participation in physical activity.

In terms of cognitive decline, Bielak et al. (2014) found that although physical activity throughout the life course (20s, 40s, and 60s) was associated with later-life cognitive functioning, it was unrelated to the rate of cognitive decline. In contrast, Gow et al. (2017) found that more physical activity during later life was related to less cognitive decline but this was not the case for exercise during early or middle adulthood. Thus, there is at least some evidence that is consistent with the previous studies that more enrichment is beneficial to later-life cognitive functioning and possible decline. However, the evidence regarding when in the life course enrichment seems to matter the most (i.e., have the largest effect) is inconclusive. Furthermore, most of this work has focused on one domain at a time—that is, physical, social, or mental enrichment—rather than a composite that incorporates many aspects of enrichment.

The Current Study

We utilized data from the Health and Retirement Study (HRS: Sonnega et al, 2014) to examine the relationship between lifetime enrichment (using current and retrospective survey data) and change in later-life cognitive performance over a period of eight years (five waves). Our first aim was to determine whether having or not having high enrichment across the life course predicts cognitive performance during older age. We hypothesized that individual patterns of enrichment throughout the life course would uniquely predict cognitive performance in later life. For example, individuals who experience high enrichment only during later life may have lower cognitive performance than those who experience high enrichment throughout the entire life course. Due to inconsistencies among previous findings, we hypothesized that high enrichment would not be associated with a change in the rate of cognitive decline. Our second aim was to examine if it matters at which life stage the enrichment occurs. That is, if high enrichment is indeed associated with better cognition as determined by Aim 1, does enrichment at each stage in the life course independently predict cognitive performance, regardless of the other life periods? Based on the previous evidence, we predicted that enrichment during early, mid, and later life would each predict cognitive performance, but perhaps not relate to decline, in older age.

Method

Transparency and Openness

We report inclusion criteria and details about the variable sources and creation process. This study was not preregistered. The R package “lme4” was used to create multilevel growth models (Bates et al., 2015), and results were confirmed in Stata 17. All variables and instructions for use are publicly available on the Health and Retirement Study (HRS) website (https://hrs.isr.umich.edu/data-products) for registered users. The HRS is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. General information about recruitment in the HRS nationally representative longitudinal panel, the study design, and protocol content are provided in Sonnega et al. (2014). Because we conducted our analyses on publicly available, deidentified data, our study on “life course engagement in enriching activities and cognitive aging” did not meet the definition of “human subjects” research and therefore, did not require review from the University of Michigan Institutional Review Board. Our analysis code is available to researchers upon request.

Participants and Procedure

We identified a subsample of HRS non-proxy participants who were aged 65 and over in the 2010 biennial interview (n = 4470) and provided complete data for cognition variables from 2010 to 2018 (final analytic sample: n = 2931). Inclusion in this analytic sample required that: a) participants had complete cognitive performance data available for the years 2010, 2012, 2014, 2016, and 2018 with no imputations; and b) information about early-life, midlife, and later-life engagement in social, physical, and mental activities collected by HRS during this time period. We also conducted attrition analyses to examine whether demographic variables predicted missing data and imputations¹. Table 1 provides an overview of the source of the variables and Tables 2 and 3 include descriptive information relevant for analysis steps linked to the two study aims.

Table 1.

Variables used to score mental, physical, and social enrichment in early life, midlife, and later life

Life Stage	Enrichment Variable	Source	Points	Response Categories	Proportion of Sample
Early	Bilingualism	LHMS	0 1	English only More than English	.89 .11
Early	Pre-school	LHMS	0 1	Did not attend preschool Attended preschool	.89 .11
Early	High school (HS)	LHMS	−1 0 1	Did not attend HS Attended HS—normal curriculum Attended HS—vocational or college prep courses	.06 .53 .41
Early	Foreign Languagea	LHMS	0 1	Did not learn a foreign language in HS Learned a foreign language in HS	.50 .42
Early	Complex Cognitive Activitiesa	LHMS	−1 0	Did not participate in ≥ 1 activity Participated in ≥ 1 activity	.31 .60
Early	Household Books	LHMS	0 1	≤ 1 bookshelf in household > 1 bookshelf in household	.72 .28
Early	Clubsa	LHMS	0 1	≤ 1 club > 1 club	.48 .39
Early	Sportsa	LHMS	0 1	No sports ≥ 1 sport	.50 .42
Early	Household (Age 10)	LHMS	−1 0	Live in orphanage, foster care, or with ≥ 6 people Live with ≤ 5 people	.45 .55
Mid	Job Enrichmentb	LHMS	−1 0	Unenriching job Enriching job	.27 .45
Mid	Physical Activity (Ages 30–39)	LHMS	0 1	< 1x per week ≥ 1x per week	.72 .28
Mid	Physical Activity (Ages 40–49)	LHMS	0 1	< 1x per week ≥ 1x per week	.66 .27
Mid	Household (Age 40)	LHMS	−1 0	Live alone or with ≥ 6 people Live with ≤ 5 people	.27 .74
Late	Community/Social Engagement	PLQ	0 1	< 1x per week ≥ 1x per week	.52 .48
Late	Hobbies	PLQ	−1 0	< 1x per week ≥ 1x per week	.34 .66
Late	Physical Activity	PLQ	−1 0	< 1x per week ≥ 1x per week	.36 .64
Late	Live with Partner	PLQ	−1 0	Do not live with partner Live with partner	.33 .62

Note: LHMS = Life History Mail Survey, PLQ = Psychosocial and Lifestyle Questionnaire.

^aParticipants who did not attend high school were not given scores for these items.

^bParticipants who did not work outside the home were not given a score for this item.

Table 2.

Descriptive statistics for the eight life course enrichment pattern groups [Aim 1]

	Group
	HHH 1	HHL 2	HLL 3	HLH 4	LHH 5	LLL 6	LLH 7	LHL 8
n	121	223	366	104	103	1480	212	322
Gender (% women)	46.28	62.33	73.50	66.35	39.80	62.64	53.77	46.27
Age at baseline (mean; range 65–94)	71.36	72.60	72.91	72.45	71.50	73.61	72.54	73.41
Race
% White	88.43	95.07	90.16	92.31	92.23	81.14	89.15	85.71
% Black	08.26	04.04	08.47	05.77	04.85	14.86	07.54	11.49
% Other	03.31	00.90	01.37	01.92	02.91	03.99	03.30	02.80
Wealtha (mean)	3.01	3.05	2.99	2.81	2.96	2.97	3.02	3.07
Cognition at baseline (mean; range 0–35)	25.61	24.80	24.54	24.54	24.13	22.21	23.17	23.39

Note: H refers to high enrichment, L refers to average to low enrichment

^aMean household wealth is coded in quintiles with 5 as the top 20%

Table 3.

Descriptive statistics for early, mid, and later life enrichment groups [Aim 2]

	Early Life Enrichment		Midlife Enrichment		Later Life Enrichment
	Higha	Average/Low	Highb	Average/Low	Highc	Average/Low
	1	0	1	0	1	0
n	814	2117	769	2162	540	2391
Gender (% women)	65.48	58.15	50.07	63.78	51.85	62.07
Age at baseline (mean; range 65–94)	72.84	73.87	72.50	73.96	71.98	73.92
Race
% White	91.52	83.18	89.73	83.00	90.19	84.44
% Black	06.83	12.79	07.53	12.43	06.42	12.18
% Other	01.67	03.70	02.71	03.32	02.52	03.30
Wealthd (mean)	3.00	2.99	3.03	2.99	2.98	3.00
Cognition at baseline (mean; range 0–35)	24.77	22.58	24.25	22.81	24.16	22.97

Note:

^aHigh early life enrichment included participants from Aim 1 enrichment groups 1–4.

^bHigh midlife enrichment included participants from Aim 1 enrichment groups 1, 2, 5, and 8.

^cHigh later life enrichment included participants in Aim 1 enrichment groups 1, 4, 5, and 7.

^dMean household wealth is coded in quintiles with 5 as the top 20%.

Measures

Cognition

Following criteria set by Crimmins et al. (2011), cognition was operationalized using a global cognition score calculated using immediate and delayed word recall tests, serial 7’s, and backward counting. Because participants were 65 or older, additional orientation and identification tests including the week, day, president, vice president, and object naming were added to the global cognition score in accordance with previous research (Walsemann & Ailshire, 2020). Cognitive performance scores ranged from zero to 35.

We included four waves of composite global cognition scores found in the RAND-HRS public data (2010, 2012, 2014, and 2016) together with the scores for the 2018 wave that we calculated using the same HRS method detailed in Ofstedal et al. (2005). Only participants who had complete cognitive data (i.e., five waves of global cognition scores) were included in the analytic sample. Imputed values were not used in any of our analyses.

Enrichment Variables and Scores

Part of the HRS core biennial battery includes a Psychosocial and Lifestyle Questionnaire (PLQ) that collects information on participant wellbeing, social relationships, and participation in work and social activities (Smith et al, 2017). Random alternating 50% subsets of panel participants assigned to a face-to-face interview complete this self-administered paper questionnaire each core wave. An additional HRS questionnaire, the Life History Mail Survey (LHMS: Smith et al., 2022), was mailed to subsets of panel participants in 2015 and 2017. The LHMS collects participants’ retrospective reports about their educational, residential, familial, and health history from early, mid, and later life. Aggregated information from 2015 and 2017 is available in the cross-wave LHMS public data file available on the HRS website: https://hrsdata.isr.umich.edu/data-products/lhms-cross-wave.

We used the psychosocial and LHMS data to calculate enrichment scores, selecting items (i.e., social, physical, and mental activities) deemed by previous literature to be enriching (Hertzog et al., 2008). Table 1 provides a list of included enrichment variables, how they were scored, and the proportion of the analytic sample assigned each score based on whether or not they reported engaging or failing to engage in each enriching activity. The sample proportions for each item score reported in Table 1 are for the total analytic sample. Participants did not receive a score for items that were unanswered. For each life period, items describing activities in which the majority of the analytic sample experienced received a score of 0 and for other, non-majority responses, points were either added to or subtracted from the total enrichment score. Specific cutoffs were determined for each item separately, based on the statistical distributions and hypothesized direction of each relationship. The evidence motivating the inclusion of each of our variables and the direction of their effects can be found in Supplemental Table 1.

We also note that heterogeneity in our enrichment variables occurs as a function of life period. For example, the variables used to determine enrichment in early life were not necessarily the same as those used for middle and later life. While testing the same items during all three life periods would allow for maximum control, we were limited by variables available in the HRS database. Furthermore, even if the HRS had indeed used the same items at each time point, enrichment often will look different depending on the life period (i.e., physical activity in early life may be seen through participation in sports but during later life, it may include less strenuous physical activities, such as walking). Because of this, regardless of how each aspect of enrichment was presented for each time point, we attempted to incorporate variables that would test for aspects of social, physical and cognitive enrichment within each life stage. The following sections describe how we calculated total enrichment scores for each life period.

Early Life.

In early life, participants received points toward a total enrichment score if they were bilingual, attended preschool, attended high school, took extra vocational or prep courses during high school, had more than one bookshelf’s worth of books in the home, or were involved in at least one sport. We subtracted points from the total enrichment score for this life period if participants a) did not attend high school; b) did not engage in any of ten complex cognitive activities in high school (e.g., play a musical instrument, do woodwork, take ballet classes, or perform in school theatre); or c) grew up in an orphanage, foster home, or household with six or more individuals. All items were from the LHMS data.

Midlife.

In midlife, participants received points toward a total enrichment score if they participated in vigorous or moderate physical activity at least once a week on average (measured separately for ages 30–39 and 40–49). We subtracted points from participants’ total enrichment score if they reported living alone or with more than six people at age 40, and if their primary job during midlife was not enriching. The job enrichment score was a composite of responses to six items about their most important job between age 30 and 40. Participants rated the degree (from strongly disagree to strongly agreed) to which their job was interesting and enjoyable, their skills matched the job, the skills learned on the job were valuable, they felt they had some personal control of their work, they could rely on their coworkers, and how physically demanding the job was. All items were from the LHMS data.

Later life.

In later life, participants received a point toward a total enrichment score if they participated in social activities at least once a week. We subtracted points from the total enrichment score if they did not engage in at least two mentally stimulating activities (listed in Table 1) at least once per week, did not exercise or walk at least once a week, or did not live with a spouse or partner. These items are included in the PLQ.

Enrichment Groups.

We calculated overall enrichment scores by summing participants’ points for all items, separately for each life period (early, middle, later life). To test our hypotheses that focused on high enrichment, as opposed to enrichment more broadly, with an appropriately selective sample, we decided a priori to select enrichment scores in the highest quartile (25%) or quintile (20%) to identify individuals who experienced ‘high enrichment.’ In a post hoc assessment of the enrichment score distribution for each life period, we observed a natural cutoff that formed around the top 25% of scores (actual percentages varied; see Results for details). As a result, we categorized scores from each participant in each life period as either ‘high enrichment’ if they were in the top approximately 25% (dummy coded as 1), or ‘average-to-low enrichment’ if they were in the remaining 75% or so (dummy coded as 0).

Based on these scores, participants were divided into eight groups that reflected their life course enrichment pattern— i.e., where each participants’ enrichment scores fell (high or average-low) for each life stage (i.e., early, mid, later). As shown in Table 2, individuals in Group 1 had high enrichment scores in early, mid, and later life, whereas those in Group 6 had average-low enrichment scores across the entire life course. This categorization allowed us to examine if enrichment group membership would predict differences in the level of later-life cognitive performance and decline over time (Aim 1).

For a second analysis, we created two groups for each life period (see Table 3). Participants with high enrichment scores (top ~25%) were compared to those with average to low scores separately for early, mid, and later life. This allowed us to test whether high enrichment in each life period predicted later life cognition, regardless of enrichment status during the other periods (Aim 2).

Covariates

Adjusted models that included covariates controlled for sociodemographic factors including race, wealth, age at baseline (2010), and gender. Variables created by RAND (RAND HRS Longitudinal File 2018) derived from HRS data were used for race and total household wealth (assets minus debts). Race was represented by two variables, the first dummy coded as one for white with all other races as the reference group, the second coded one for black with all other races as the reference group. Due to the uneven distribution of wealth in participants, wealth coded into quintiles (5 = top 20%; 1 = bottom 20%) for the analyses. Gender was dummy coded, with women as the reference group. Covariate selection was based on prior research using HRS data and enrichment literature (Gow et al., 2017; Walsemann et al., 2020; Zahodne et al., 2015) indicating variables that may have established relationships with cognition in older adulthood.

Statistical Analyses

We conducted unadjusted and adjusted growth curve models (fitted in a multilevel modeling framework) to assess the predictive value of enrichment on repeated observations of cognitive performance over eight years. We plotted the mean cognitive performance scores over time by enrichment group and determined a linear specification was the best fit. We first tested whether life course patterns of enrichment predicted later life cognitive performance and decline (Aim 1). We then examined whether high enrichment during any life period (included as separate predictors within the same model) independently predicted later-life cognition and cognitive decline (Aim 2). Linear mixed models were fit using Restricted Maximum Likelihood (REML) and tested via the Satterthwaite method for t-tests. Adjusted regression models included the covariates (race, wealth, age at baseline, and gender).

Results

Descriptive Statistics

High enrichment refers to the scores in the top 27.8% in early life, 26.2% in midlife, and 18.4% in later life. Therefore, participants with ‘average to low’ enrichment constituted the remaining 72.2%, 73.8%, and 81.6%, respectively. Demographic information and the breakdown of membership in the enrichment groups examined for Aims 1 and 2 can be found in Tables 2 and 3, respectively.

Aim 1: Does life course enrichment group predict cognitive performance in later life?

Estimates from linear mixed models are reported in Table 4. In the unadjusted model for Aim 1, cognitive performance was the outcome variable whereas time, enrichment group, and a time by enrichment group interaction were the predictors. The adjusted model for Aim 1 included the same variables plus covariates. Patterns of significance among the variables of interest were similar in the two models, thus the adjusted model is described here. As expected, the level of cognitive performance decreased over time, and enrichment group predicted cognitive performance in later life. A group-by-time interaction was also present, suggesting that enrichment group membership also predicted the rate of cognitive decline in older age.

Table 4.

Aim 1 multilevel growth curve model results assessing cognitive change over five waves

	Aim 1 Unadjusted Model		Aim 1 Adjusted Model
Fixed Effects	Estimate (b)	Standard Error	Estimate (b)	Standard Error
Intercept	25.73	0.22	33.35	0.87
Time	−0.23***	0.05	−0.23***	0.04
Enrichment Group	−0.38***	0.04	−0.3***	0.05
Time x Enrichment Group	−0.03**	0.01	−0.03**	0.01
Gender: Men			0.43***	0.12
Age			−0.13***	0.01
Race: White			1.83***	0.34
Race: Black			−1.33***	0.37
Wealth			0.00	0.04
Random Effects
Variance of random intercept	8.47	0.31	7.12	0.27
Variance of random slope	0.24	0.018	0.22	0.02
Residual Variance	6.42	0.09	6.48	0.02

^*p < .05.

^**p < .01.

^***p < .001

To compare cognitive performance between individuals who experienced high enrichment during at least one period during their lifetime (groups 1, 2, 3, 4, 5, 7, 8) to those who did not have high enrichment at any point (group 6), we conducted follow-up contrasts with effect coding. Overall, cognitive performance was significantly worse for group 6 than for groups 1, 2, 3, and 7 (all ps < .033; see Figure 1 for trajectories by group). The difference between group 6 and groups 4, 5, and 8 did not reach significance (ps > .05), likely as a result of small group sizes. Additionally, group 6 had a significantly steeper rate of cognitive decline as compared to group 1 (p = .025). Furthermore, although there was no difference in overall cognitive performance between groups 6 and 8, there was an interaction such that the rate of decline differed between groups (p = .009). No other interactions were significant (ps > .05).

Figure 1. The relationship between life course enrichment patterns and cognition in later life.

Note. Mean cognitive performance scores of each group are plotted at each time point (2010–2018), with bars representing standard error. H refers to high enrichment, L refers to average to low enrichment.

Aim 2: Does enrichment in each life period independently predict cognitive performance?

Linear mixed model results revealed enrichment in early, middle, and later life significantly predicted cognitive performance in older age (Table 5) such that high enrichment was associated with better cognitive performance (see Figure 2). Additionally, early life and later life enrichment significantly predicted the rate of cognitive decline². However, the association between the rate of decline and midlife enrichment was not significant.

Table 5.

Aim 2 multilevel growth curve model results assessing cognitive change over five waves

	Aim 2 Unadjusted Model		Aim 2 Adjusted Model
Fixed Effects	Estimate (b)	Standard Error	Estimate (b)	Standard Error
Intercept	22.9	0.09	29.08
Time	−.46***	0.03	−0.45***	0.02
Early Life Enrichment	1.72***	0.17	1.39***	0.16
Midlife Enrichment	0.79***	0.17	0.72***	0.16
Late Life Enrichment	0.61**	0.19	0.43**	0.18
Time x Early Life Enrichment	0.09*	0.04	0.09*	0.04
Time x Midlife Enrichment	0.03	0.04	0.03	0.04
Time x Late Life Enrichment	0.15**	0.05	0.14**	0.05
Gender: Men			0.55***	0.12
Age			−0.12***	0.01
Race: White			1.65***	0.33
Race: Black			−1.32***	0.37
Wealth			0.00	0.04
Random Effects
Variance of random intercept	7.94	0.29	6.7	0.28
Variance of random slope	0.23	0.02	0.21	0.02
Residual Variance	6.44	0.09	6.48	0.09

^*p < .05.

^**p < .01.

^***p < .001

Figure 2. Life stage as an independent predictor of cognitive performance and decline.

Note. Mean cognitive functioning scores of each group are plotted at each of the 5 time points (2010–2018), separately for early, mid-, and later life, with bars representing standard error.

To compare the role of enrichment during each life period on cognitive performance in later life, we conducted a post hoc analysis in which we calculated pseudo-R² values³ using backward elimination for each predictor in our model. This analysis yields an index of effect size for each predictor that allows us to compare the magnitude of the effects. We find that early life enrichment had the largest effect (adjusted model pseudo R² = .028), whereas middle and later life explained the same amount of variance in cognitive performance (adjusted model pseudo-R²= .006).

Discussion

In this study, we used current and retrospective self-report data to categorize individuals’ level of enrichment (operationalized as engagement in physical, social, and cognitively complex activities) across the life course during their early, middle, and later years. We then examined the relationship between high enrichment and cognitive performance and decline in older age (≥ 65 years) over a span of eight years. We found that a person’s life course pattern of enrichment predicted both the level of cognitive performance, as well as the rate of cognitive decline. Indeed, those who did not have high enrichment during any life stage were the worst performers. Additionally, a high level of enrichment during early and later life, independent of enrichment during other stages, predicted better cognitive performance and slower decline, but high enrichment during midlife only predicted level of cognitive performance. This supports the idea that enrichment may increase cognitive reserve and therefore allow higher functioning for longer.

For Aim 1, we sought to determine whether having high enrichment across the life course predicts cognitive performance during older age. We found that having high enrichment during at least one life period was predictive of a higher level of cognitive performance over eight years in this sample aged 65 to 94 years at baseline. These results are consistent with past findings that have investigated enrichment effects at individual life periods, but rarely from a life course perspective covering early, middle, and later life. Although we cannot conclude a causal relationship from these results alone, we may speculate as to why this association might exist. Wilson et al. (2005), for example, hypothesized that this relationship may be explained by a link between enrichment during early and midlife and enrichment during later life, suggesting that enriched individuals are more likely to remain in or continue to seek out enriching environments. However, the findings from Aim 2 indicate that those with high enrichment during early and/or midlife, but not later, life, still outperformed those who had no high enrichment at all. This suggests that the relationship between high enrichment in early and midlife and later-life cognition cannot be completely explained by engagement in enriching activities in later life.

Alternatively, in line with the Scaffolding Theory of Aging and Cognition (STAC), enriching activities throughout the life course could have a positive effect on cognitive outcomes in later life by altering neural network pathways used in ways that create compensatory scaffolding (Reuter-Lorenz & Park, 2014). Cognitive performance in individuals with high enrichment may also be enhanced in later life compared to individuals with less enrichment due to enhanced neural networks that attenuate the adverse effects of brain aging. In addition, because our findings show that participation in enriching activities even in later life is associated with cognitive performance, it is plausible that the STAC model could also explain the relationship between enrichment and rate of cognitive decline (Reuter-Lorenz & Park, 2014). In essence, if positive plasticity is possible in older adulthood, continued participation in enriching activities could stave off more rapid cognitive decline.

For Aim 2, we tested whether enrichment at each stage in the life course independently predicted cognition. In accordance with our hypothesis, we found that enrichment during early, middle, and later life, all predicted cognitive performance, regardless of enrichment status during other periods. When examining the magnitude of each effect (using pseudo-R² values) it appears as though high enrichment during early life may be most strongly associated with cognitive ability in later life (2.8% variance accounted for). However, both mid-life and later-life enrichment still explained a significant amount of variance in cognitive performance (0.6% variance accounted for), suggesting that the benefits of enrichment during these life periods should not be ignored.

Contrary to our hypothesis, high enrichment during early and later, but not middle, life also predicted a slower rate of cognitive decline. These findings may support the cognitive reserve theory of differential preservation, which states that advantages gained by enrichment are increased over time by individuals not only starting at a higher baseline cognitive level, but also having a more gradual rate of cognitive decline (Salthouse, 2006; Tucker-Drob et al., 2009). This is corroborated by the finding that participants with average to low enrichment in early and later life not only start at a lower baseline of cognitive performance but also experience steeper rates of cognitive decline. Thus, cognitive reserve does not appear to be maintained to the same degree over time. This would suggest that engaging in enrichment only in early or later life may still have potential benefits on an individuals’ rate of cognitive change in later years. This can be contrasted to the idea of preserved-differentiation, in which we would have expected to see different baseline levels of cognition based on enrichment group but similar trajectories of cognitive decline across groups (Tucker-Drob et al., 2009). However, it is important to note that relatively few studies in the past have found that increased participation in enriching experiences across the lifespan are related to a slower rate of cognitive decline in later life (for a recent study see Stieger & Lachman, 2021). Thus, we should be cautious in interpreting these findings that do not appear to be highly robust across the literature. It may be the case that while this effect may exist, it may be relatively small, and therefore, not found in previous work that uses smaller samples. As such, more work is needed in this area to further elucidate the relationship with cognitive decline and what factors may influence this association.

The significant associations between cognitive performance in later life and enrichment during every life period demonstrates the value in using a life course model. Because participation in enriching activities during each point throughout the life course was independently related to cognitive outcomes in later life, this suggests that enrichment, regardless of when it is experienced, is a valuable use of an individuals’ time and resources. Furthermore, since each period throughout the life course is complex and unique to each person, it is encouraging that there is not just one life period that meaningfully contributes to cognitive performance in later life. This provides individuals the flexibility to participate in highly enriching activities whenever it is possible for them. Due to the time-consuming nature of education during early life, raising children and/or building a career during midlife, or health limitations that may be experienced in later life, many individuals are unable to consistently experience high enrichment across the entire lifespan. Our findings show that participation in enriching activities during any life period may be beneficial for later life cognition, although it does seem to be best when enrichment is continued and maintained in all life periods.

Limitations and Future Directions

We also note some limitations of the current study and possible future directions. Level of enrichment during midlife, unlike early and later life, was not significantly associated with rate of cognitive decline. Although it may be the case that middle age is unique such that cognitive reserve may be less malleable during this time, it is also possible that the available measures for midlife were not sufficiently sensitive to categorize levels of enrichment. Similar to previous work, occupation-related details were used to determine level of enrichment during middle age. However, because many women in the cohorts in the sample (born between 1916 and 1945) did not work outside the home during midlife, the enrichment scores for this life period relied more on physical and social, rather than cognitive, measures of enrichment. The lack of a score for cognitive enrichment during midlife for 28% of our participants could explain why this relationship did not reach significance. In addition, racial and geographical disparities in opportunities for enrichment likely play a role (Staben et al., 2022).

Our primary analyses used a binary score (high/low) to summarize the overall level of engagement in enriching activities for each life period. This dichotomy was determined by the distribution of scores assigned for reported engagement in a selected set of cognitive, physical, and social activities (see Table 1 and Supplemental Table S1). It facilitated comparing engagement in early, midlife and later life given the different number of activities included in the HRS dataset. We acknowledge the limitations associated with applying an arbitrary cut point to define high versus low subgroups. Overall, findings from a supplemental analysis of the continuous scores (Supplemental Tables S2 and S3) were similar to those using the dichotomy, with the exception that the time by early-life enrichment interaction was no longer significant. Future research is needed to address inconsistencies in the literature regarding the definition and role of early-life enrichment on the rate of later-life cognitive decline.

More generally, while the use of Health and Retirement Study (HRS) data provides enormous benefits (e.g., a large sample, a considerable number of batteries/variables, longitudinal data), it also comes with inherent limitations. For example, the self-report data collected about early and mid-life are completed retrospectively, that is, several decades after they occurred. There may be significant error in reporting on events from long ago, which may be worse for individuals with lower cognitive abilities (e.g., memory). Research shows, however, that reports of salient facts about personal life events and activities (i.e., autobiographical memory; see Thomsen, 2015 for a review) are generally reliable (Belli, 1998; Smith et al., 2021). Until more longitudinal studies of lifespan development are conducted beginning in early life that continue into old age, we are limited to the use of retrospective responses.

Additionally, the HRS dataset uses a global measure of cognition, which yields a general cognitive functioning score for each participant in each wave of data collection. This type of measure was selected primarily due to time constraints in order to reduce the burden for participants completing all the batteries of the HRS. However, the broad nature of this assessment does not allow us to identify which aspects of cognitive functioning relate most to enrichment. It may be the case that enrichment is differentially related to various cognitive domains. Thus, future work should prioritize using more sensitive cognitive functioning measures for longitudinal studies.

Another limitation includes the racial breakdown of the sample. The HRS seeks to include a racially heterogeneous sample of the United States population over age 50. However, because our sample was largely composed of white participants, our findings may not be generalizable to people over age 65 in different racial/ethnic population subgroups. Further exploration of enrichment should be pursued among more racially balanced samples and younger birth cohorts to determine whether these results are robust and replicable.

More broadly, we should consider the gaps that remain in the field that can be studied to inform future intervention work. Although we did not seek to uncover which types of enrichment primarily drove the relationship between enrichment and cognition, it would be worthwhile to examine the differential effects of social, physical, and mental enrichment on cognition and cognitive decline in later life. It may be the case that mentally enriching activities are more or less important, as compared to social and physical activities, for later life cognition. The degree of impact from each type of activity may also depend on when in the life course it occurs. Furthermore, it is important to explore individual differences in the experience of enrichment. In the present study, we were unable to determine the degree of enrichment gained from each activity, which likely varies from person to person. Factors such as novelty, which is known to increase attentional engagement (Stine-Morrow & Manavbasi, 2022), may enhance enrichment such that a novice piano player might yield more benefits from playing the piano than a life-long pianist of the same age. There may also be individual differences in the strength of the association between enrichment and later life cognition. Perhaps those who start off with lower cognitive performance, or have particularly unenriching early lives, may gain the most from engaging in enriching activities. It would be valuable to identify factors that make someone particularly susceptible to the benefits of enrichment.

Conclusion

In conclusion, we find that having high enrichment during at least one life period is associated with better cognitive performance. Additionally, high enrichment during early, middle, and later life, regardless of enrichment status during other periods, all predict cognitive performance in later life. Thus, it appears that high enrichment may be contributing to cognitive reserve across the lifespan, although this relationship may be strongest during early life. As such, enrichment does indeed matter for later life cognition, and matters across the entire life course. These findings suggest that it may never be too late to benefit from interventions designed to increase engagement in enriching physical, social and cognitive activities.

Significance:

The present study suggests that enrichment does indeed matter for cognitive aging, and is beneficial across early, middle, and later life. This has implications for the creation of interventions designed to increase engagement in enriching physical, social and cognitive activities across the entire life course.