Helping behaviors and cognitive function in later life: The impact of dynamic role transitions and dose changes

Abstract

Despite the burgeoning literature linking prosocial helping behaviors and cognitive function, empirical evidence on whether transitions into and out of helping roles—and how dynamic changes in time commitment—shape cognitive outcomes remain limited. Moreover, most research has focused on formal volunteering, leaving the cognitive outcomes associated with informal helping—assistance provided directly to non-household individuals—largely unexplored. The objective of this study was to investigate the linkages between two forms of helping behaviors—formal volunteering and informal helping—and late-life cognitive function, focusing on dynamic changes in these behaviors over time. Drawing on the life course perspective and two decades of longitudinal data from the U.S. Health and Retirement Study (1998–2020; $N=\mathrm{31,303}$), we employed the asymmetric fixed-effects modeling approach within a multilevel framework to assess how intra-individual changes in helper role status and time commitment shape cognitive function trajectories. Results indicated that transitioning into volunteering and informal helping were both associated with a higher level of cognitive function and a slower cognitive decline, and highlighted how sustained engagement in helping can yield cumulative cognitive benefits that progressively become greater over time. The findings also provide unique evidence on the level of time commitment in helping behaviors needed to achieve cognitive benefits, where moderate levels of helping (approximately 2–4 weekly hours) were consistently linked to robust cognitive benefits for both forms of helping. These findings highlight prosocial helping behaviors as impactful, modifiable lifestyle interventions for promoting cognitive health in aging populations.

Introduction

In light of the growing public health burden posed by Alzheimer’s disease and other related dementias (ADRD), a variety of interventions targeting modifiable risk factors (e.g., hypertension, diabetes, depression, smoking, and air pollution) have been proposed to promote cognitive function and mitigate cognitive decline in later life. A recent review provides growing support for the effectiveness of these interventions in reducing dementia risk, although effect sizes vary and the evidence base remains stronger for some risk factors than others (Livingston et al., 2024). Against this backdrop, there has been increasing interest in modifiable lifestyle factors—such as maintaining a healthy diet, engaging in physical exercise, and participating in socially meaningful activities—that are also associated with better cognitive outcomes (Kivipelto et al., 2020). This growing body of research suggests how individuals spend their time may significantly influence cognitive health, offering insights into strategies for effective interventions.

Among the lifestyle factors receiving increasing attention is social activities aimed at reducing social isolation (Wang et al., 2023). Particularly, increasing prosocial engagement via helping others, such as through formal volunteering and informal helping, has garnered much scientific attention as a candidate for an innovative and cost-effective public health intervention for promoting cognitive function (Anderson et al., 2014). However, empirical evidence directly supporting this proposal is limited in ways that hinder policy efforts for developing effective interventions. First, earlier studies often failed to properly distinguish the act of helping others from the characteristics of the helper, conflating the health effects unique to helping behaviors with the health advantages typically found among helpers. This introduces the selection effects widely discussed in the literature, leading to an overestimation of the effects of helping behaviors on cognitive outcomes (Kail & Carr, 2020). Second, many studies failed to capture the dynamic nature of helping behaviors, which change over time in both status (e.g., volunteering vs. not volunteering) and intensity (i.e., varying levels of time commitment, sometimes referred to as dose). Consequently, little is known about the unique effects of transitioning into and out of a helper role, or how increasing or reducing the time committed to helping others influence cognitive function. Finally, much of the research linking helping to cognitive function has focused on formal volunteering (i.e., helping others under the auspices of formal organizations), while largely overlooking informal helping directly provided to non-household family members (i.e., friends and neighbors), which is far more common in the U.S. compared to formal volunteering. This gap is particularly important given that informal helping is especially prevalent among underserved and disadvantaged communities, for whom mutual aid within informal networks serves as a vital mechanism for meeting daily needs and maintaining well-being (Morrow-Howell & Wang, 2013; Wang et al., 2022).

We address these gaps by investigating the dynamic linkages between two forms of prosocial helping behaviors and cognitive function, using longitudinal data spanning over two decades from the U.S. Health and Retirement Study (HRS), a nationally representative sample of middle-aged and older adults. Building on a growing body of HRS-based research on the topic (Han et al., 2020; Infurna et al., 2016; Kail & Carr, 2020; Proulx et al., 2017; Wang et al., 2022), we extend prior work by drawing from key concepts of the life course perspective to examine how intraindividual changes in volunteering and informal helping—measured in terms of both behavior status and time commitment—affect the level of cognitive function and the rate of cognitive decline across middle and late adulthood. In doing so, we also contribute to the literature by applying a new methodological approach that estimates the impact of dynamic changes in modifiable behaviors on health outcomes, while addressing potential sources of endogeneity (e.g., omitted variable bias, reverse causation) to strengthen causal inference.

Helping behaviors and cognitive function

While health benefits associated with prosocial helping behaviors have been observed across various forms of helping throughout the lifespan, the evidence for middle-aged and older adults largely comes from studies focusing on formal volunteering. Indeed, there are now decades of research demonstrating robust linkages between volunteering and a wide range of physical and mental health outcomes (Anderson et al., 2014; Burr et al., 2021). This, in turn, has given rise to the hypothesis that volunteering may lead to better cognitive outcomes in the context of ADRDs, as well as age-related normal cognitive decline. The conventional theoretical basis for these expected cognitive benefits is that the goal-oriented nature of helping behaviors performed within cognitively stimulating environments may contribute to cognitive reserve by fostering problem-solving skills, adaptive thinking, and sustained mental engagement (Scarmeas & Stern, 2003). This view is increasingly supported by emerging neurobiological models (Burr et al., 2021), which suggest that interrelated neurobiological systems facilitate helping behaviors, which in turn activate the downstream physiological pathways—the hypothalamic–pituitary–adrenal axis (HPA), sympathetic nervous system (SNS), and immune system—for the betterment of the helper’s health, including cognitive function (Brown & Brown, 2015; Inagaki, 2018).

Despite the robust theoretical developments linking helping and cognitive function, the overall evidence, based on a range of research designs ranging from cross-sectional, longitudinal, and randomized controlled trials, remains mixed and inconsistent (for a review, see Keefer et al., 2023). Several recent longitudinal studies showed more promising results, indicating a positive link between volunteering and cognitive outcomes, but these findings are limited by a number of methodological issues that undermine the interpretability and validity of the findings. First, earlier studies often compared cognitive outcomes between volunteers and non-volunteers, raising the question of whether the traits associated with being a volunteer (e.g., genetic factors, health, better socioeconomic resources), rather than the act of volunteering itself, accounted for the better cognitive outcomes observed among volunteers (for an exception, see Kail & Carr, 2020). A related issue that has received insufficient attention is the dynamic nature of volunteering behavior and its effects on cognitive function. Few studies have explicitly conceptualized volunteering as a role that individuals transition into and out of over time, leaving unanswered how role acquisition (e.g., starting to volunteer) and role withdrawal (e.g., ceasing to volunteer) may exert different effects on cognitive outcomes. Prior research often assumed that these effects are equivalent in magnitude but opposite in direction. However, evidence from other domains suggests this assumption may be overly simplistic, as role acquisition and withdrawal may exert differential effects on health and well-being (see Figure 1, scenarios 1 and 2). The life course perspective provides a useful framework for addressing these issues by calling attention to intraindividual changes, where dynamic transitions into and out of helping behaviors are expected to result in unique changes in cognitive trajectories (Ghisletta et al., 2015). That is, transitions may exert asymmetrical effects, with the cognitive benefits of role acquisition through enriched social and cognitive stimulation not necessarily matching the magnitude of the cognitive consequences associated with role withdrawal.

Figure 1.

Three stylized scenarios depicting the potential links between helping transitions and cognitive outcomes.

In addition, earlier studies typically focused on how volunteering influences the level of cognitive function at a given time point without considering how volunteering might alter the rate of cognitive decline, leaving a gap in understanding its influence on long-term cognitive trajectories. If volunteering is linked only to the level of cognitive function and not to the rate of cognitive decline, any initial cognitive benefits from volunteering could be largely offset once one ceases to volunteer, rendering the cognitive benefits of volunteering somewhat limited. However, from a life course perspective, transitions into a volunteer role may represent a pivotal turning point that changes the course of cognitive decline, which could have substantive implications for sustaining helping behavior for an extended duration of time. That is, if volunteering slows the rate of cognitive decline, its benefits could accumulate over time, resulting in progressively greater cognitive advantages as individuals sustain their engagement. (see Figure 1, scenario 3, for an illustration).

Finally, although informal helping shares several features and outcomes with formal volunteering (e.g., social integration, better cardiovascular health) that are theorized to promote cognitive outcomes in later life (Nakamura et al., 2024), this form of helping behavior remains understudied and its cognitive benefits remain largely unknown. Despite being less visible and often overlooked compared to other forms of prosocial behavior (Martinez et al., 2011), informal helping is substantially more prevalent in middle and later adulthood than formal volunteering (Han et al., 2023). Informal helping may be particularly relevant for minoritized populations for whom mutual aid within informal community networks has deep cultural roots in the U.S., especially given the disparities and barriers in access to formal volunteering opportunities that persist across racial and ethnic groups (Morrow-Howell & Wang, 2013).

This study contributes to the scientific literature by addressing key limitations in existing research by explicitly distinguishing the effects of helping behaviors from the characteristics of those who provide the help, examining the impacts of role transitions—including the adoption and withdrawal of helping roles—on cognitive outcomes, and expanding the focus to include informal helping. While most prior research has relied on a status-based approach, comparing cognitive outcomes across individuals, the current study takes a process-based perspective, examining how intraindividual changes in helping behaviors over time shape cognitive trajectories (Luhmann et al., 2011). Additionally, by incorporating both forms of helping status and time commitment across a lengthy observation window, we uncover dose–response parameters and identify the level of engagement needed to produce both short-term and long-term cognitive function benefits.

Dose effects of helping.

In general, earlier studies on the health and well-being benefits associated with time committed to volunteering suggest a non-linear relationship: health benefits generally increase with hours spent volunteering, but there is typically an upper limit beyond which these benefits level off or diminish (Burr et al., 2011; Windsor et al., 2008). While there is no consensus on the threshold, existing evidence indicates that volunteering is beneficial to health unless it exceeds what is considered very high levels of commitment (e.g., over 15 or 20 hours per week; Windsor et al., 2008). Acknowledging this non-linear pattern, studies focusing on cognitive function often treat volunteering hours as a categorical variable and generally find that the magnitude of cognitive benefits increases progressively with time commitment, from low (less than 2 hours weekly), to moderate (2–4 hours), and high (more than 4 hours), although with some diminishing returns (Han et al., 2020; Kail & Carr, 2020; Proulx et al., 2017).

However, prior studies have largely overlooked how dynamic changes in time commitment (dose) to helping others may influence cognitive function, often assuming that the dose effect of volunteering is constant regardless of an individual’s history of prior volunteering (i.e., researchers assume no interaction between past and current volunteering hours), although there are reasons to question the validity of this assumption. In the broader literature on role transitions in later life, abrupt changes in time committed to major roles are found to be more detrimental to health and well-being than gradual changes (Moustafa et al., 2020). For instance, sudden transitions into an intense caregiving role have been shown to be more stressful and harmful to health outcomes compared to more gradual transitions (Burton et al., 2003). Similarly, abrupt retirements are generally considered to lead to worse health and adjustment outcomes than phased retirements (Dingemans & Henkens, 2014; van Solinge, 2012). A parallel can also be drawn from the literature on physical activity programs and interventions that recommend beginning with low intensity activities and slowly increasing the exercise intensity (Cress et al., 2005), as studies often show that moderate levels of exercise lead to better health benefits than vigorous exercise for individuals with low to no baseline activity levels (Scott et al., 2022).

This implies that gradual changes in time commitment to helping behaviors, both in the process of role acquisition and withdrawal, lead to more favorable cognitive outcomes compared to abrupt changes. However, whether and how changes in the intensity of formal volunteering and informal helping influence cognitive function remain an open empirical question, one that is crucial for providing evidence that helps in the design of effective interventions to support cognitive health in later life.

Study Objectives

Building on the growing body of theoretical and empirical work linking helping behaviors to better health, this study examines the complex ways in which cognitive function is associated with formal volunteering and informal helping—two common ways individuals help others and contribute to their communities in later life. Specifically, we employ a methodological approach that allows us to assess whether and how dynamic intraindividual changes in helping behaviors, measured by role status and time commitment, influence cognitive function levels and the rate of cognitive decline over time. We hypothesize that transitioning into a helping role would lead to higher cognitive function and a slower rate of cognitive decline, while withdrawing from the helping role would result in lower cognitive function and a faster rate of decline. Additionally, we anticipate that a gradual transition (e.g., moving from a non-volunteer to a moderate dose of volunteering or gradually moving from a high to moderate/low dose) would yield better cognitive outcomes than an abrupt change (e.g., transitioning to a high-dose of volunteering from a non-volunteer status or abruptly ceasing high-dose volunteering). Finally, although research on informal helping is limited, we expect a similar pattern of results with informal helping behaviors.

Methods

Data Source and Study Sample

This study utilized longitudinal data from the Health and Retirement Study (HRS), a nationally representative biennial panel survey of U.S. adults aged 51 and older, as well as their spouses (Health and Retirement Study, 2024). While the HRS began collecting data in 1992, the current study focuses on 12 waves of data collected from 1998 to 2020, as some study variables were inconsistently measured before 1998. Given that the HRS regularly replenishes its sample with samples of younger cohorts to maintain its representativeness, “baseline” in this study is defined as the first wave in which a participant entered the HRS. Data for this study were primarily taken from the RAND constructed longitudinal data file—an easy-to-use dataset based on HRS core data that accounted for missing information and inconsistencies across waves (RAND Center for the Study of Aging, 2023). An exception was made for measures on helping behaviors, which were not included in the longitudinal RAND dataset and were instead taken directly from the HRS core data available from the Survey Research Center at the University of Michigan.

We analyzed 12 waves of biennial data collected between 1998 and 2020. Among over 36,000 respondents interviewed in the HRS during this period, we first included 35,082 individuals who were age-eligible and completed at least one non-proxy interview. Missing data on key study variables were minimal (approximately 0.57% of person-wave observations) and these participants were excluded from the analysis. This resulted in 198,818 person-wave observations collected from 34,924 participants. Following recommendations in the literature (Breuer & deHaan, 2024), we excluded singleton observations ($n=\mathrm{3,621}$) from the analyses. The final sample included 195,197 person-wave observations collected from 31,303 participants averaging approximately 6.2 waves of observations per respondent.

Measures

Cognitive function.

Respondents’ cognitive function was assessed at each wave using a modified version of the Telephone Interview for Cognitive Status (m-TICS) widely used in large population surveys (Crimmins et al., 2011). The measure included an immediate and delayed 10-noun free recall test of memory (range: 0–20), a serial 7s subtraction test of working memory (range: 0–5), and a backward counting test of mental processing (range: 0–2). A composite score was calculated by summing across these tests, with higher scores indicating better cognitive function (range: 0–27).

Helping behaviors.

The two forms of helping behaviors were assessed at each wave as status and time commitment. Formal volunteering status was assessed with the item “have you spent any time in the past 12 months doing volunteer work for religious, educational, health-related or other charitable organizations?”, where the response was coded dichotomously (1 = yes; 0 = no). Informal helping status was similarly assessed with the item, “Did you spend any time in the past 12 months helping friends, neighbors, or relatives who did not live with you and did not pay you for the help?” Participants were further asked a set of anchoring questions that capture the annual time committed to helping (e.g., “Altogether, would you say the time amounted to less than 100 hours, more than 100 hours, or what?” followed by “Would it be less than 200 hours, more than 200 hours or what?”). A multi-category measure for the annual time committed to each form of helping was created (e.g., 0 = did not volunteer; 1 = 1–99 hours; 2 = 100–199 hours; 3 = 200+ hours), which was employed in the analyses as a categorical variable to capture possible threshold or non-linear effects.

Control variables.

We controlled for a set of time-varying covariates (TVC) measured at each wave to control for demographic, social and health characteristics that could confound the key estimates. These variables included age in years (range = 51–109), marital status (1 = married or partnered; 0 = not), labor force status (1 = not working; 2 = part–time work; 3 = full–time work), household wealth (assets minus debts in $1,000, inverse hyperbolic sine-transformed), activity of daily living (ADL) limitations (range: 0–5), and depressive symptoms, evaluated with the eight-item version of the Center for Epidemiologic Studies Depression Scale (CES-D; range: 0–8). These TVCs were lagged by one wave in the analyses to adjust for participants’ characteristics prior to the helping transitions, thereby accounting for potential confounding by prior levels of these covariates that may influence both the likelihood of engaging in helping transitions (i.e., treatment/exposure) and subsequent cognitive outcomes. We also included a set of time-invariant covariates (TIC) in the analyses, which included gender, race/ethnicity, birth year, and education.

Analytic Strategy

The key research questions addressing the linkages between helping behaviors and cognitive function were examined using multilevel models, with observations (level 1) nested within persons (level 2). Several analytical techniques were incorporated into the basic multilevel modeling framework. First, helping transitions were examined using an asymmetrical fixed effects modeling approach (Allison, 2019), which relaxes the common assumption that transitions in both directions—i.e., role adoption (positive) and withdrawal (negative)—have effects of equal magnitude. Instead, this approach allows for the possibility that entering and exiting a helping role may have distinct cognitive effects. Similarly, changes in the time committed to helping were also allowed to yield asymmetric effects—for example, increasing the time commitment from 1–99 hours to 100–199 hours would lead to changes in cognitive function that is different from the magnitude of change induced by reduction in time commitment from 100–199 hours to 1–99 hours (for equations, see Supplementary Materials).

The asymmetric measures were incorporated into a series of within-between random effects (WBRE) models, which decompose all time-varying independent variables, including the helping transitions, into between-person (BP; person-mean of time-varying variables) and within-person (WP; deviation from the person-mean at each wave) components. This approach yields WP estimates that are at least as unbiased as those from fixed effects models, which are independent of selection effects attributed to all stable inter-individual differences, both observed and unobserved (Bell & Jones, 2015). We chose the WBRE approach over fixed effects models primarily for its ability to more accurately estimate cognitive function trajectories by accounting for substantial inter-individual variability in the rate of cognitive function decline through specifying random effects for age (Tucker-Drob, 2019). Cognitive function trajectories were estimated using an approach that primarily tracks longitudinal changes in cognitive function (i.e., individual change, assessed with linear and quadratic age measures at each wave), while accounting for cross-sectional age-differences in cognitive function (assessed with birth year), which has been shown to be crucial when the study sample includes a wide range of birth cohorts (Morrell et al., 2009). We further added an interaction term between age and birth year to control for potential cohort-differences in the rate of cognitive decline (Morrell et al., 2009). Together, these strategies ensure that the key model estimates (e.g., ${\beta }_{10}$ and ${\beta }_{20}$ below) reflect average intra-individual changes in cognitive function in response to changes in helping behaviors that are unconfounded by person-level characteristics.

Given our interest in cognitive health outcomes unique to each form of helping behavior, formal volunteering and informal helping were examined in separate models, with the other form of helping included as a lagged covariate in each model. For each helping behavior, we estimated two sets of models based on the two asymmetric measures of helping: helping status (yes or no) and time commitment (hours). We first estimated a model with asymmetric measures of helping transitions (and dose changes) along with all study variables. The level-1 WP estimates from this model allowed us to investigate whether participants showed changes in the level of cognitive function as a function of helping transitions. For example, the abbreviated level-1 equation predicting person i’s cognitive function at time $t$ as a function of helping status transitions was as follows:

$$\begin{gathered} {\mathrm{C}\mathrm{o}\mathrm{g}}_{ti}={\beta }_{00}+{\beta }_{10}(H_{ti}^{+}-\overline{H_{ı}^{+}})+{\beta }_{20}(H_{ti}^{-}-\overline{H_{ı}^{-}})+{\beta }_{30}({\mathrm{A}\mathrm{g}\mathrm{e}}_{ti}-{\overline{Age}}_i)+ \\ {\beta }_{40}({\mathrm{T}\mathrm{V}\mathrm{C}}_{t-1i}-{\overline{TVC}}_i)+{\beta }_{50}({\mathrm{C}\mathrm{o}\mathrm{g}}_{t-1i}-\overline{{Cog}_i})+e_{ti}, \end{gathered}$$

where ${\beta }_{10},{\beta }_{20}$, and ${\beta }_{30}$ denote coefficients for the two asymmetric effects of helping transitions (i.e., positive and negative transitions) and age. The effects of TVCs were captured with ${\beta }_{40}$; all time-varying confounders were lagged by one wave, ensuring that confounders (measured at $t-1$) preceded helping transitions (measuring change between $t-1$ and $t$) and the cognitive outcomes measured at $t$, maintaining the correct temporal order to support improved causal inference. In incorporating lagged covariates, we adopted the approach proposed by McNeish and Matta (2020) to retain baseline observations that do not have prior observations in the analyses. In a similar manner, we further considered effects of the lagged cognitive outcome, which has been shown to contribute to yielding a conservative estimate of the treatment effect through accounting for feedback effects (i.e., past levels of cognitive function affecting helping transitions; Demetrescu et al., 2023). To assess the robustness of our findings, we estimated additional models that included a set of additional health controls at $t-1$ (i.e., number of chronic conditions and self-rated fair/poor health), along with models excluding the lagged cognitive function outcome. As reported in the Supplementary Materials, the results of these sensitivity analyses were consistent with the main findings, providing additional confidence in the robustness of our results.

The question of whether helping transitions at each wave altered the rate of cognitive decline was addressed in the subsequent model with the inclusion of an interaction term between linear age and the asymmetric measures of helping, where the interaction terms were also decomposed into WP and BP components (Schunck, 2013). All model estimates were adjusted by HRS survey weights at both levels 1 and 2; robust standard errors are reported, which accounts for the clustering of observations by sampling units (Heeringa et al., 2017). All analyses were performed using the MIXED procedure in Stata (Version 18).

Results

Study Sample Characteristics

Descriptive characteristics of the study sample are shown in Table 1. On average, participants scored 15.9 out of 27 on the cognitive function scale during the study period. Volunteering and informal helping were recorded in 35.9% ($n=\mathrm{66,727}$) and 56.8% ($n=\mathrm{102,762}$) of all person-wave observations, respectively. Participants exhibited substantial changes in helping behaviors over time (not shown in Table 1). Specifically, there were 12,455 positive transitions (role adoption) and 15,625 negative transitions (role withdrawal) in volunteering; for informal helping, there were 19,022 positive transitions (role adoption) and 25,999 negative transitions (role withdrawal). Similarly, there were significant changes in the intensity of helping (dose changes) during the study period (see Supplementary Materials for more details).

Table 1.

Weighted Descriptive Characteristics of the Study Sample

	Mean	(SD)	%
Time-varying characteristicsa
Cognitive function (0–27)	15.86	(4.42)
Volunteering
No			64.1%
Yes			35.9%
Time commitment
1–99 hours			20.8%
100–199 hours			8.2%
200+ hours			7.0%
Informal helping
No			43.2%
Yes			56.8%
Time commitment
1–99 hours			41.1%
100–199 hours			9.4%
200+ hours			6.2%
Age (51–109)	65.89	(9.88)
Married or partnered			64.7%
Labor force status
Not working			56.0%
Part-time work			13.9%
Full-time work			30.1%
Household wealth (IHS-transformed; −8.6–14.7)	5.11	(2.74)
Median value in $1,000	168.01
ADL limitations (0–5)	0.26	(0.77)
Depressive symptoms (0–8)	1.42	(1.96)
Background characteristicsb
Female			53.8%
Race-ethnicity
White, non-Hispanic			75.0%
Black, non-Hispanic			11.0%
Other, non-Hispanic			4.5%
Hispanic			9.5%
Birth year (1892–1965)	1945.80	(13.60)
Education (in years; 0–17)	12.96	(3.11)

Notes.

^aperson-wave observations N = 195,197.

^bPerson N = 31,303.

IHS = inverse hyperbolic sine. ADL = activities of daily living.

Multilevel Model Results

Volunteering and cognitive outcomes.

The model results linking volunteering status transitions and cognitive outcomes are presented in Panel A, Table 2. In Model A1, positive transitions (role adoption) were associated with an increase in cognitive function ($b=0.175,p<.001$), while negative transitions (role withdrawal) were linked to a reduction in cognitive function ($b=-0.156,p<.001$). In Model A2, we examined whether these transitions were linked to the rate of cognitive decline. Consistent with our hypothesis, positive transitions were associated with a slower rate of cognitive decline ($b=0.010,p=.002$), indicating that transitioning into and maintaining a volunteer role may lead to accumulated cognitive benefits. In contrast, negative transitions were associated with a faster rate of cognitive decline ($b=-0.014,p<.001$).

Table 2.

Helping Status Transitions and Cognitive Function

	Panel A: Volunteering				Panel B: Informal helping
	Model A1		Model A2		Model B1		Model B2
	β	(SE)	β	(SE)	β	(SE)	β	(SE)
Time-varying characteristics: WP
Role acquisition	0.175***	(0.029)	0.176***	(0.032)	0.141***	(0.029)	0.134***	(0.034)
Role withdrawal	−0.156***	(0.031)	−0.113**	(0.036)	−0.179***	(0.024)	−0.143***	(0.025)
Age	−0.094***	(0.003)	−0.094***	(0.003)	−0.091***	(0.003)	−0.090***	(0.003)
× Role acquisition			0.010**	(0.003)			0.009***	(0.003)
× Role withdrawal			−0.014***	(0.003)			−0.011***	(0.003)

Notes. Person-wave observations N = 195,197; person N = 31,303. Estimates based on weighted data. Standard errors (in parentheses) are adjusted for survey weights, clustering, and stratification. WP: Within-person. Full model results presented in Supplementary Table S1.

^**p < .01.

^***p < .001.

Model results from Models A3 and A4, demonstrating the dose effects, are shown in Figure 2 (estimates are presented in Table S2 in the Supplementary Materials). The results from Model A3 indicated that findings based on status transitions described above masked substantial heterogeneity observed in dose changes (see top panel). Specifically, transitioning from non-volunteering (i.e., 0 hours) into 1–99 hours ($b=0.141,p<.001$) and 100–199 hours ($b=0.309,p<.001$) of volunteering were both associated with higher levels of cognitive function. Transitioning into 200+ hours also showed a similar positive effect, but the magnitude of the effect was smaller compared to that of transitioning into 100–199 hours of volunteering ($b=0.230,p=.048$). In contrast, withdrawing from any level of volunteering to non-volunteering resulted in lower cognitive function, with the largest effect observed when withdrawing from 200+ hours ($b=-0.366,p<.001$). Furthermore, increasing time commitment from any non-zero volunteering hours category to a higher dose was associated with better cognitive function (e.g., 100–199 hours to 200+ hours; $b=0.235,p<.001$), whereas reducing the time commitment to volunteering led to worse cognitive function only when the change in dose was relatively dramatic (i.e., 200+ hours to 1–99 hours; $b=-0.162,p=.039$).

Figure 2.

Formal Volunteering Dose Changes and Cognitive Function. Estimates used to create these figures are presented in Table S2 (see Supplementary Materials).

Finally, results from Model A4, shown on the bottom panel of Figure 2, demonstrate the links between changes in volunteering dose and the rate of cognitive decline. Transitions from a non-volunteer to 100–199 hours ($b=0.015,p=.028$) and 200+ hours ($b=0.025,p=.019$) of volunteering were associated with a slower rate of cognitive decline, but not for 1–99 hours of volunteering ($b=0.004,p=.239$). However, increasing time commitment from any non-zero volunteering status did not result in changes to the rate of cognitive decline. In contrast, the only instance where a reduction in volunteering led to faster cognitive decline occurred when withdrawing from 1–99 hours of volunteering ($b=-0.013,p<.001$).

Informal helping and cognitive outcomes.

The results for linkages between informal helping status transitions and cognitive function levels are presented in Panel B, Table 2. Similar to the findings based on volunteering, positive transitions led to improved cognitive function, while negative transitions resulted in worse cognitive function (see Model B1). Similarly, in Model B2, positive transitions led to a slower rate of cognitive decline, while negative transitions led to a faster rate of cognitive decline.

Dose effects of informal helping are shown in Figure 3 (see also Table S2 in the Supplementary Materials). In Model B3 (top panel), transitioning into any level of informal helping and increases in the dose of informal helping generally resulted in better cognitive function, with the exception of an increase from 100–199 hours to 200+ hours, which was statistically unrelated to changes in cognitive function. In contrast, withdrawing from any level of informal helping to no informal helping led to lower cognitive function, while reductions in informal helping to a non-zero level did not result in changes in cognitive function. Finally, in Model B4 (bottom panel, Figure 3), transitioning from a non-helper to 1–99 hours and 100–199 hours of informal helping led to a slower rate of cognitive decline, whereas withdrawing from each level of time commitment to non-helping also led to a faster rate of cognitive decline. Other dose changes in informal helping were unrelated to changes in the rate of cognitive decline.

Figure 3.

Informal Helping Dose Changes and Cognitive Function. Estimates used to create these figures are presented in Table S2 (see Supplementary Materials).

Discussion

The primary aim of the study was to investigate the cognitive function benefits associated with two forms of helping behaviors, focusing on dynamic changes in these behaviors over time. Guided by the life course perspective and leveraging two decades of rich longitudinal data from a large, national data source, Health and Retirement Study, we examined how role transitions and changes in time commitment shape cognitive trajectories in later life, both in terms of the level of cognitive function and the rate of cognitive decline. Findings from this study offer novel evidence based on the asymmetric effects of transitions and dose changes, providing new insights into how sustained engagement in prosocial activities may yield cumulative cognitive benefits that grow over time. These results highlight the potential of helping behaviors as an accessible and impactful lifestyle intervention for many older persons to potentially help preserve and promote cognitive health in later life.

Volunteering and cognitive outcomes

Consistent with previous research, we observed significant cognitive benefits associated with formal volunteering. The findings based on asymmetric measures indicated that transitioning into a volunteer role led to a higher level of cognitive function, whereas withdrawing from the role resulted in lower levels of cognitive function. However, the more intriguing evidence emerged from models examining rates of cognitive decline: we found that sustained engagement in volunteering led to a slower rate of cognitive decline, which demonstrated how the cognitive benefits from volunteering may accumulate over time.

In the context of the life course perspective, this finding suggested that adopting a volunteering role may serve as a critical turning point that alters the trajectory of cognitive aging. This perspective emphasizes that transitions into new roles, such as volunteering, may provide opportunities for individuals to engage in enriching social and cognitive activities, potentially delaying age-related cognitive decline through improved cognitive reserve (Scarmeas & Stern, 2003). Conversely, withdrawing from a volunteering role may represent a negative turning point, disrupting these enriching experiences and accelerating cognitive decline. These findings highlight volunteering as a dynamic and modifiable lifestyle behavior that can shape cognitive trajectories in later adulthood.

The analyses based on the dose of volunteering (i.e., time commitment) provided additional insights into how helping behaviors through the auspices of formal organization may shape cognitive aging. In general, increases in time commitment to volunteering were associated with higher levels of cognitive function, although the transition from no volunteering to 200+ hours yielded cognitive benefits were not as pronounced in magnitude and statistical significance (see Figure 2, bottom panel). However, this should not be interpreted as a relative lack of value in high-intensity volunteering: analyses focusing on the rate of cognitive decline suggested that transitioning to and sustaining high-dose volunteering (i.e., 200+ hours) leads to progressively greater cognitive benefits over time. This may be because the immediate benefits of high-dose volunteering could be offset by the challenges of an abrupt transition, which are known to be disruptive and create role conflict. Over time, sustained engagement at this intensity may facilitate role adjustment, at which point the higher levels of social and cognitive activity may translate into substantial cognitive benefits. This view aligns with findings from research reported from the Baltimore Experience Corps, a randomized controlled trial (RCT) designed around high-intensity volunteering (median of 400+ annual hours), which demonstrated that cognitive benefits emerged only among participants engaged for an extended period of over 24 months (Brydges et al., 2020).

Our findings also revealed distinct patterns in how reductions in volunteering time commitment can affect cognitive function. Abrupt reductions in time commitment were associated with lower levels of cognitive function, while gradual reductions or withdrawals led to less detrimental outcomes, which was consistent with findings from the broader literature on role transitions in later life (Moustafa et al., 2020). However, as modeled here, any pathway to a complete role withdrawal (e.g., from any level of time commitment to not volunteering) resulted in significantly worse cognitive function, suggesting the importance of sustaining volunteering at a lower dose for maintaining cognitive function in later life.

Informal helping and cognitive outcomes

Our findings shed new light on cognitive outcomes associated with informal helping, an under-appreciated form of prosocial engagement that has received relatively less attention compared to formal volunteering (Martinez et al., 2011). Similar to volunteering, informal helping was associated with significant cognitive benefits, where transitions into informal helping roles led to higher levels of cognitive function, while withdrawing from these roles resulted in lower levels of cognitive function. Moreover, sustained engagement in informal helping at low (1–99 hours) and moderate (100–199 hours) doses led to slower rates of cognitive decline, suggesting that, like volunteering, the cognitive benefits of helping accumulate over time. These findings underscored the importance of considering informal helping alongside volunteering, when designing programs to help modify lifestyle strategies aimed at promoting cognitive health through prosocial helping behaviors.

However, our findings regarding informal helping and cognitive function were somewhat different from our findings on volunteering in ways that merit further consideration. That is, unlike volunteering, high levels of informal helping were associated with immediate cognitive benefits, as indicated by higher levels of cognitive function, but sustained engagement in high-dose helping was unrelated to accumulated benefits over time. This distinction may reflect the nature of informal helping, which often involves more reactive and repetitive tasks rather than the structured, goal-oriented activities involving the experiences typically associated with formal volunteering (Taniguchi, 2012). Moreover, informal helping often lacks the formal role assignment or public recognition that comes with formal volunteering, which may diminish its potential to foster a strong sense of purpose or role identity. Over the long run, high-dose informal helping may become part of the routine demands of daily life, blending into daily hassles with relative low reward or stimulation rather than serving as a distinct and cognitively enriching activity. This may explain why sustained high-dose informal helping, unlike formal volunteering, was not associated with accumulated cognitive benefits over time. More research is needed here.

Implications

This study contributed to a growing recognition of lifestyle factors as critical components of healthy cognitive aging by shedding light on the role of formal volunteering and informal helping. The findings should be understood in the context of recent trends in the prosocial engagement among older adults, which suggest that participation in these helping behaviors has stalled or even declined among recent cohorts of older adults (Han et al., 2023). At the same time, there have been dramatic increases in sedentary behavior and “screen time” over the recent years, both of which are associated with worse cognitive and physical health outcomes (Livingston, 2019; Saunders et al., 2020). These trends underscore the need for considering helping behaviors as a public health intervention strategy to mitigate the negative effects of unhealthy lifestyles. Helping behaviors provide an accessible and meaningful way for older adults to stay socially connected and cognitively engaged, addressing the public health concerns stemming from social isolation and physical inactivity.

Moreover, our findings emphasize the importance of tailoring interventions to older adults with varying needs and capacities, particularly for those who may face barriers to initial engagement or risk disengagement due to lack of resources and health problems (Gonzales et al., 2015). Given the potential importance of sustained engagement in helping behaviors to maximize cognitive benefits as identified in this study, practitioners should consider identifying and implementing best practices for promoting long-term prosocial engagement for those older persons who are able or who want to participate in such activities. For volunteering, strategies could include providing stipends to offset costs, offering flexible scheduling, and fostering social interaction and support among peer volunteers and staff (Sellon, 2014). For older adults unable to maintain high-dose volunteering, accommodating continued engagement at lower doses represents an ideal strategy, which may help to sustain the cognitive benefits of volunteering while minimizing the detrimental effects of abrupt disengagement. These efforts are especially critical for older adults in sub-optimal health, as increasing evidence indicates that the health benefits of volunteering are often greater for those in poorer health (de Wit et al., 2022; Han & Park, 2024). By creating opportunities that align with diverse abilities and circumstances of older adults, public health initiatives can leverage the full potential of helping behaviors to enhance cognitive health, foster social connections, and counteract the adverse effects of unhealthy lifestyles in aging populations.

Finally, while more research is needed to understand the antecedents of and barriers to engagement and withdrawal from informal helping, a first step would be to assess the needs and motivations of potential helpers within communities. Community leaders and organizers may want to devise innovative approaches to encourage informal helping, such as fostering a culture of mutual aid, promoting intergenerational exchanges, and providing accessible platforms to match older adults with informal helping opportunities within communities. These efforts can help maximize the reach and impact of informal helping, ensuring its benefits for both individual cognitive health and community well-being.

Limitations and Contributions

This study has limitations that should be acknowledged. First, the lack of detailed information on the specific helping activities in which respondents engaged for both volunteering and informal helping limits our ability to understand the mechanisms linking helping behaviors to cognitive function. More detailed measures would provide greater insight into which types of activities contribute to cognitive outcomes and help clarify the observed differences in the patterns of results between formal volunteering and informal helping. Second, although the measurement approach used to assess time committed to helping others in this study is known to minimize item non-response and help overcome several forms of systematic bias (Hauser & Willis, 2004), the reliance on self-reported measures of helping behaviors raises the possibility of recall bias. Similarly, the instrument employed in this study to evaluate cognitive function (i.e., m-TICS) used, while well-validated and widely used, were not clinical assessments (Crimmins et al., 2011). As such, we are not able to draw clinical implications from the study findings. Finally, although our modeling strategy incorporated several techniques to support causal inference, the estimates may still be biased by factors we could not fully address, such as residual time-varying confounding and measurement error, and therefore, the results reported here should not be interpreted as strictly causal estimates.

Despite these limitations, this study makes several important contributions to the substantive and methodological literature on the topic. To the best of our knowledge, this is the first study to track role transitions and dose changes as they unfold over a long period of time, investigating how the dynamic changes in helping behavior shape cognitive trajectories in middle and late adulthood. By extending the recently introduced asymmetric fixed effects modeling approach—previously limited to continuous and binary predictor variables only (Allison, 2019)—to ordinal measures of helping behaviors, we also made a methodological contribution. This approach allowed us to uncover new evidence demonstrating how transitioning into and sustaining a helper role may lead to accumulated cognitive benefits in later life, while also providing unique insights into the level of time commitment that may be needed to achieve cognitive benefits. Importantly, because we incorporate helping behavior status, time commitment to helping others, and cognitive function across a lengthy observation window, we are able to specify the parameters of two prosocial engagements needed to produce both short-term and long-term cognitive benefits, which helps to bridge two strands of prior research: studies linking dose changes to immediate cognitive level changes (Kail & Carr, 2020) and studies linking helping status to long-term outcomes, including cognitive impairment (Infurna et al., 2016; Wang et al., 2022). In doing so, the current study clarifies the data-generating mechanism linking short-term cognitive benefits to longer-term gains, offering a more nuanced understanding of how helping behaviors may slow down cognitive aging and protect against cognitive impairment. Finally, the analytic approach used in this study, including strategies to address multiple forms of endogeneity and temporal ordering of key variables, helped to provide additional confidence in the potential causal relationships underlying our estimated effects.

In conclusion, our findings underscored the potential of prosocial helping behaviors as modifiable lifestyle factors for promoting cognitive health in aging populations. Future research should build on these results by incorporating more detailed measures of helping activities to better understand the types of tasks and contexts that yield the greatest cognitive benefits. Additionally, exploring the contextual factors that shape and prolong prosocial engagements in later life, such as cultural norms, social support, and accessibility of opportunities, could help identify strategies to promote sustained participation, where feasible. Such efforts would provide valuable insights into optimizing the design of interventions and ensuring that the benefits of helping behaviors are accessible to diverse populations.