Journals of Gerontology: Psychological Sciences

cite as: J Gerontol B Psychol Sci Soc Sci, 2020, Vol. 75, No. 7, e93–e104
doi:10.1093/geronb/gbz020
Advance Access publication February 19, 2019

Original Research Report

Education and Cognition in Middle Age and Later Life: The

Mediating Role of Physical and Cognitive Activity

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

Yujun Liu, PhD and Margie E. Lachman, PhD
Department of Psychology, Brandeis University, Waltham, Massachusetts.
*Address correspondence to: Yujun Liu, PhD, MS, Psychology Department, Lifespan Initiative on Healthy Aging and Lifespan Lab, 415 South
Street Waltham, MA 02454-9110. E-mail lyujun@brandeis.edu

Received: August 29, 2018; Editorial Decision Date: February 7, 2019

Decision Editor: Nicole Anderson, PhD, CPsych

Abstract
Objectives: Although educational attainment is related to cognitive function in later life, little is known about the mecha-
nisms involved. This study assessed the independent mediating effects of two behavioral variables, physical and cognitive
activity, on the association between educational attainment and cognitive function and change.
Methods: Data were derived from the three waves of the Midlife in the United States (MIDUS) study. Predictors (educa-
tional attainment) were from the 1995 baseline, mediators (physical and cognitive activities) were from the 2004 wave,
and outcomes (cognitive function) were from the 2004 and 2013 waves. Conditional process modeling was applied using
PROCESS in SPSS.
Results: There were both direct and indirect effects of educational attainment on level and change of executive function
(EF) and episodic memory (EM). Physical activity and cognitive activity were both significant mediators for cognitive level.
For mediators of change, however, cognitive activity was significant for EF and physical activity was significant for EM.
Discussion: Physical and cognitive activity are discussed as possible factors for protecting against cognitive decline in later
life. The findings have implications for advancing supportive policies and practices related to maximizing the benefits of
education and physical and cognitive activities for cognition in middle age and later life.
Keywords: Cognition, Education, Health, Mediation analysis
  

There is considerable evidence for wide individual differ- Education and Cognitive Change
ences in the extent of cognitive change in later life (Mella, It has been suggested that educational experiences pro-
Fagot, Renaud, Kliegel, & de Ribaupierre, 2018; Schaie, vide the foundation for continued intellectual stimulation
Willis, & Caskie, 2004). Studies investigating change in across the life course, resulting in improved cognitive func-
multiple cognitive domains reported individual differences tioning in late adulthood (Wilson et al., 2009). However,
in both the rate of change and variations in the patterns there is still considerable controversy about the rela-
of change across a number of cognitive abilities (Mungas tionship between educational attainment and cognitive
et al., 2010; Tucker-Drob, Johnson, & Jones, 2009). Many change. It remains unclear whether higher educational
risk and protective factors for cognitive aging have been attainment slows the rate of cognitive decline over time in
examined, including genetic, health, physical, behavio- middle age and later life (Fritsch et al., 2001; Glymour et
ral, lifestyle, and sociodemographic contributors such as al. 2005). Epidemiologic evidence suggests that individu-
educational attainment (e.g., Alley, Suthers, & Crimmins, als with higher educational or occupational attainment
2007; Hertzog, Kramer, Wilson, & Lindenberger, 2008; have a great degree of cognitive reserve (CR) and show a
Salthouse, 2014).

© The Author(s) 2019. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. e93
For permissions, please e-mail: journals.permissions@oup.com.
e94 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

reduced risk of developing Alzheimer’s disease and other those with lower educational attainment (Wilson et al.,
forms of dementia (Hall et al., 2007; Stern, 2012). CR has 2009).
been suggested to account for the delayed onset of behav- The benefits of physical activity for older adults are
ioral manifestations of dementia among those with brain well established (Kirk-Sanchez & McGough, 2014). There
pathology (Stern, 2012). However, there is mixed evidence is compelling evidence that an active lifestyle has broad
as to whether education is associated with the timing and benefits for cognitive, physical, and psychological health
extent of cognitive declines in normal aging (Stern, 2009, among older adults (Smith et al., 2010). Physical activity
2012). Some research suggests that higher educational can delay or prevent many chronic diseases, including heart
attainment does not protect against cognitive decline (Le disease, type 2 diabetes, some cancers and dementia, which
Carret, Lafont, Mayo, & Fabrigoule, 2003) or even results have been associated with cognitive declines. The cogni-
in a slightly faster rate of cognitive decline among those tive benefits of physical activity are also well documented

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

who develop dementia (Wilson et al., 2009). (Erickson, Hillman, & Kramer, 2015). Zhu and colleagues
(2017) investigated the association of objectively measured
physical activity with incidence of cognitive impairment
Cognitive Benefits of Physical and Cognitive and longitudinal cognition among older adults using data
Activity from 6,452 participants in the United States and indicated
Of interest is whether there are modifiable factors that are that high level of physical activity was associated with
related to educational attainment that can account for dif- lower risk of cognitive impairment and better mainte-
ferences in cognitive aging. Education may cultivate the nance of memory and executive function over time, par-
knowledge, skills, and ability necessary for continued par- ticularly in white adults. Using nationally representative
ticipation in intellectually demanding activities (e.g., read- samples of participants aged 50 years and older from 11
ing, taking courses) or health-promoting behaviors (e.g., European countries (Austria, Germany, Sweden, Denmark,
physical exercise) well into middle and later adulthood. Switzerland, the Netherlands, Belgium, France, Spain, Italy,
Higher educational attainment has been associated with and Greece), Aichberger and colleagues (2010) found cog-
greater participation in various lifestyle activities (Allet nitive benefits from both light intensity activity, such as
et al., 2016; Thrane, 2006), including physical activities leisurely walking, and higher intensity aerobic activity.
and activities that are cognitively demanding (Lachman, Individuals who participated in any type of regular physical
Agrigoroaei, Murphy, & Tun, 2010). activity showed less cognitive decline after 2.5 years, espe-
The beneficial effects of maintaining an engaged lifestyle cially when they engaged in vigorous activities more than
have been demonstrated across several studies, even when once a week (Aichberger et al., 2010). In another study,
activities are introduced later in life (Bherer et al. 2013). Albinet, Boucard, Bouquet, and Audiffren (2010) reported
On the basis of findings from cohort studies and short- that 12 weeks of aerobic training led to enhanced perfor-
term clinical trials, previous studies indicated that indi- mance in executive control and increased heart rate vari-
viduals who continuously engage in high level of physical ability in older men and women aged 65–78 years. These
activity and place significant demands on their intellectual results suggest that aerobic exercise may be an important
resources may maintain or even enhance cognitive poten- brain protective factor as people age. Across cognitive
tial (Fratiglioni, Paillard-Borg, & Winblad, 2004; Lövdén, domains, there is evidence for exercise-related improve-
Bäckman, Lindenberger, Schaefer, & Schmiedek, 2010; ments for both executive functioning (e.g., processing
Park et al., 2014). Compared to other forms of lifestyle speed; Frederiksen et al., 2015) and memory (e.g., spatial/
activity, greater participation in intellectually demanding episodic memory; Erickson et al., 2015). However, there
activities may be especially beneficial for cognitive func- is some evidence to suggest that these functions are dis-
tion (Stine-Morrow et al., 2014). For instance, results of tinctly influenced by physical activity, in that processes that
the Advanced Cognitive Training for Independent and require executive control, in contrast to memory, tend to
Vital Elderly (ACTIVE) study, the first large-scale, ran- exhibit more robust findings (Kramer et al., 1999; Smith
domized trial to test the long-term outcomes of cognitive et al., 2010).
training effects on prevention of decline in daily function,
support the effectiveness of cognitive intervention in main-
taining cognitive health over the long-term and indicate Current Study
modest but detectable far transfer to instrumental activi- Although a number of studies have explored the indepen-
ties of daily living, health-related quality of life, and driv- dent contributions of physical or cognitive activities on
ing outcomes (Tennstedt & Unverzagt, 2013). Conversely, cognition, only a few studies have explored these factors
activities low in cognitive stimulation, such as watching in combination (Sturman et al., 2005). Ghisletta, Bickel,
television, have been related to an increased risk of cogni- & Lövdén (2006) found that activities such as reading a
tive impairment (Wang et al., 2006). Moreover, activities book and playing games were related to changes in percep-
low in cognitive demand may be more prevalent among tual speed, whereas other forms of engagement, for exam-
ple, physical and social activities, were not associated with
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7 e95

such changes. Using data from a large biracial community 48.3% of the sample. Nine years later, the second wave
of older adults, Sturman and colleagues (2005) found that (MIDUS 2) included data from about 75% (N= 4,963) of
the beneficial effects of physical activity on the rate of cog- the respondents who participated in the follow-up study.
nitive decline over 6 years were reduced and no longer sta- As is typically found, those who participated at the second
tistically significant when cognitive activity was adjusted wave showed some differences on MIDUS 1 variables com-
and in analyses that eliminated persons with the lowest pared with those who dropped out of the study (Radler
cognitive performance at baseline. Cognitive function was & Ryff, 2010). Compared to the dropouts, longitudinal
measured by the East Boston Tests of Immediate Memory participants were more highly educated, t(6,757) = 12.48,
and Delayed Recall, the Mini-Mental State Examination, p < .001, mean years of education 14.06 versus 13.21;
and the Symbol Digit Modalities Test. They argued that were more likely to be women, 53.8% versus 48.3%,
physical activity alone does not protect against cognitive χ2(1) = 17.49, p < .001; and had higher self-rated health,

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

decline among older adults (Sturman et al., 2005). With t(6,759) = 10.42, p < .001, 3.61 versus 3.33 on a 5-point
data from 250 participants aged between 18 and 44 years scale where 1 = poor, 5 = excellent. Dropouts did not differ
old, the results from the INSIGHT study, a comprehen- from longitudinal participants in terms of age at MIDUS 1,
sive, multidisciplinary brain training system, indicated t(6,711) = .70, p = .48, 46.14 versus 46.39 years old. The
that physical activity was more important than cogni- average age of the longitudinal participants (the sample
tive activity for adaptive reasoning and problem solving we included in this study) was 58.69 (SD = 11.37), with
(Daugherty et al., 2018). 53% women. A majority of the participants (93%) were
Given the varied findings about the relative benefits of white, and more than 70% of the participants were mar-
physical and cognitive activity, the current study exam- ried or cohabiting. The average education level of the par-
ined both forms of activity together to consider whether ticipants at MIDUS 2 (M2) was 14.32 years (SD = 2.62).
they have independent contributions to cognitive level and MIDUS 3 (M3) was conducted 9.12 years later, on average
changes more than 9 years. In addition, although some (SD = 0.53). Of the sample from M2, 76.9% of those el-
research has examined mechanisms for exercise interven- igible (N = 3,294) were retested (Hughes, Agrigoroaei,
tions, little attention has been paid to possible mediators in Jeong, Bruzzese, & Lachman, 2018). The descriptive sta-
prospective longitudinal studies (Hertzog et al., 2008). Our tistics for the M2 sample and the participants who had
study also considered whether physical and cognitive activ- longitudinal data for cognitive variables are shown in
ity are mediators of the relationship between educational Supplementary Table A.
attainment and cognitive level and change. Those with At M3, participants ranged in age from 42 to 92 years
higher education are expected to engage more frequently in (M = 64.30, SD = 11.2) and had a mean education level
both cognitive and physical activity. Specifically, we tested of 14.6 years (SD = 2.6). Women made up 55.3% of the
the following hypotheses: (a) education was expected to be sample. The Brief Test of Adult Cognition by Telephone
positively related to both level and change in cognition, such (BTACT; the psychometric properties of the BTACT are
that individuals with higher levels of educational attain- reported in Lachman, Agrigoroaei, Tun, & Weaver, 2014)
ment would demonstrate better performance on cognitive was administered for the first time at M2, in a separate
function and show less cognitive decline; (b) participants telephone interview, with a completion rate of 86%
with higher levels of educational attainment were expected (N = 4,206) of eligible participants. As with M2, at M3
to report being more physically and cognitively active; (c) the BTACT was administered in a separate telephone inter-
more frequent cognitive and physical activity was expected view, with a completion rate of 82% (N = 2,693) of eligible
to be related to better cognitive performance, independent participants. The cognitive tests at M2 and M3 were con-
of education; and (d) finally, we predicted that the associa- ducted on average 9.32 years apart (SD = 0.45). About 85%
tion between educational attainment and cognition would of the survey sample at M2 (4,206 of 4,963 participants)
be mediated by both physical and cognitive activities. and about 82% (2,693 of 3,294 participants) of the sur-
vey sample at M3 completed the cognitive phone interview.
There were no significant demographic differences, includ-
Method ing age, gender, education, race, marital status, and health,
between participants who completed the cognitive phone
Participants interview and those who did not. Those who participated
This study includes the three waves of the Midlife in the at the third wave showed some differences on M2 vari-
United States (MIDUS) national database. The first wave ables compared with those who dropped out of the study.
(MIDUS 1) was collected between 1995 and 1996 with Compared to the dropouts at M3, longitudinal participants
7,108 noninstitutionalized participants in the 48 contig- were more highly educated, t(4,198) = 10.53, p < .001;
uous states selected via random digit phone dialing (Brim, M = 14.69 versus 13.83 years of education; were younger,
Ryff, & Kessler, 2004). The original participants ranged in t(4,204) = 5.11, p < .01, M = 55.20 versus 57.18 years
age from 24 to 75 years (M = 46.40, SD = 13.00), had a old; and had higher self-rated health, t(4,204) = 11.09,
mean education level of 13.21 years, and women made up p < .001, M = 3.68 versus 3.34 on a 5-point scale where
e96 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

1 = poor, 5 = excellent. Dropouts did not differ from lon- 1 = never, 2 = less than once a month, 3 = once a month,
gitudinal participants in sex, 55.3% versus 52.5% female, 4 = several times a month, 5 = once a week, and 6 = sev-
χ2(1) = .14, p =.71. Compared to dropouts, longitudinal eral times a week. Higher score indicates more frequent
participants performed significantly better on all cognitive physical activity. We computed the mean score across sum-
tests and factors at M2 (Hughes et al., 2018). mer and winter in all three settings for both moderate and
vigorous intensity. We selected the activity intensity and
setting with the maximum value to represent the highest
Dependent Variables frequency of physical activity across all intensity levels and
Episodic memory (EM) was measured by immediate and domains (Cotter & Lachman, 2010).
delayed free recall. Following exploratory and confirma-
tory factor analysis (Lachman et al., 2010, 2014), an EM

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

composite factor score was computed as a standardized Independent Variable and Covariates
mean of the z-scored measures loading on the factor. Education, the independent variable from M1, was opera-
Executive function (EF) was measured by working tionalized as the total number of years of formal schooling.
memory (measured by backward digit span), verbal fluency Covariates, which were from M2, included age (coded in
(measured by category fluency), reasoning (measured by years), gender (men coded as 1, women coded as 2), marital
number series completion), executive functioning (meas- status (married coded as 1, separated, divorced, widowed,
ured by task-switching [Stop and Go Switch Task]) and and never married all coded as 0), race (Caucasian coded as
speed of processing (measured by 30 seconds and counting 1, African American, and others coded as 0), and self-rated
task, or 30-SACT). An EF composite score was computed physical health, which was reported by participants on a
following exploratory and confirmatory factor analysis 5-point scale ranging from 1 (poor) to 5 (excellent).
(Lachman et al., 2010). The EF factor score was computed
as a standardized mean of the z-scored measures loading
on the factor. Statistical Analysis
Both factor scores at M3 were standardized using the Descriptive information and correlations were computed
means and standard deviation from M2 to allow for exam- for all study variables. Before conducting main analyses,
ination of change. Change in EM and EF was analyzed by continuous predictor variables were mean-centered for
using M3 scores as the dependent variable and including moderation analyses so that the intercepts could be inter-
M2 scores as a predictor. preted as the average scores. Conditional process modeling
was applied using PROCESS in SPSS. Three criteria need
to be satisfied to indicate a mediation relationship (Baron
Mediating Variables & Kenny, 1986): (a) the predictor variable needs to signifi-
Frequency of engaging in cognitive activities cantly predict the outcome variable, (b) the predictor vari-
The cognitive activity variable, from M2, was created by able must significantly predict the mediator variable(s), and
averaging the self-reported frequencies on a 6-point scale (c) the mediator variable(s) must significantly predict the
(1 = never, 2 = once a month, 3 = several times a month, outcome variable while controlling for the predictor vari-
4 = once a week, 5 = several times a week, and 6 = daily) of able. If both direct and indirect effects remain significant,
engaging in four cognitive activities: reading books, mag- the association is said to be partially mediated (Hayes &
azines, or newspapers; doing word games such as cross- Preacher, 2014).
word puzzles or Scrabble; attending educational lectures or Multiple mediation analyses were based on 1,000 boot-
courses; and writing (e.g., letters, journal entries, or stories; strapped samples using Hayes’ PROCESS Macro v2.15
Lachman et al., 2010). (Hayes & Preacher, 2014), allowing for formal tests of the
total, direct, and indirect effects of educational attainment
on cognitive function at M2 and cognitive change from M2
Frequency of physical activity to M3. The predictor variable was educational attainment
Physical activity, from M2, was created by twelve ques- at M1, the two mediator variables were physical activity
tions assessing the participants’ frequency of vigorous (e.g., and cognitive activity at M2, and covariates were from M2;
competitive sports such as running, vigorous swimming, outcomes were cognitive function at M2 and the change of
or high intensity aerobics; digging in the garden or lifting cognitive function from M2 to M3.
heavy objects) and moderate intensity (e.g., leisurely sports
such as light tennis, slow or light swimming, low-impact
aerobics, or golfing without a power cart; brisk walking Results
or mowing the lawn with a walking lawnmower). These
questions referred to frequency of physical activities sepa- Findings of Univariate and Bivariate Analyses
rately for the summer and winter months, in three different The descriptive statistics for the sample at M2 and longi-
settings (i.e., home, work, and leisure), with ratings from tudinal sample are displayed in Supplementary Table A.
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7 e97

On average, participants’ cognitive function level at M3 mediation effects on the relationship between educational
(EM: M = −0.04, SD = 0.98; EF: M = −0.15, SD = 0.74) attainment and level of EF.
was lower than their cognitive function level at M2 (EM: As shown in Table 2, the results also supported the
M = 0.02, SD = 0.99; EF: M = 0.06, SD = 0.97). partial mediation effects of physical activity and cognitive
Means and correlations between all variables are shown activity on the association between educational attain-
in Table 1. Participants who were older, non-white, had ment and EM at M2. The direct effect between educational
lower income and education, poorer physical health, and attainment and EM was significant (direct effect: 0.060,
lower physical and cognitive activity were more likely to 95% CI: 0.047, 0.071). The indirect effects between edu-
show lower levels of EF and EM. Women had lower EF and cational attainment and EM through both physical activity
men had lower EM. (indirect effect: 0.003, 95% CI: 0.001, 0.005, κ2 = 0.008)
and cognitive activity (indirect effect: 0.007, 95% CI:

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

0.004, 0.009, κ2 = 0.033) were also significant, indicating
Findings From Mediation Models that both physical activity and cognitive activity had small
Level of cognition but significant partial mediation effects on the relationship
For the mediation models, coefficients and 95% confidence between educational attainment and level of EM.
intervals (CIs) are provided (see Table 2). Model 1 tested
whether educational attainment was related to EF at M2 Change in cognition
and whether this relationship was mediated by physical Model 3 and Model 4 tested whether physical activity
activity and cognitive activity. For the mediational effect, and cognitive activity mediated the relationship of edu-
kappa squared (κ2) is provided as a measure of effect size, cational attainment and 9-year change in cognitive func-
as recommended by Preacher and Kelley (2011). With tion (EM and EF; see Tables 3). To accomplish this, we
the guidelines of Cohen (1988), small, medium, and large added M2 cognition as a predictor variable, in addition
effect sizes are stated as 0.01, 0.09, and 0.25, respectively, to age, sex, race, marital status, and self-reported health,
for mediation analysis. The total effects model that does and used M3 cognition as dependent variables. The di-
not consider the effect of the mediator demonstrated that rect effects model found a significant relationship between
educational attainment was significantly related to EF educational attainment and EF change (0.012, 95% CI:
(total effect: 0.124, 95% CI: 0.114, 0.134). As shown in 0.004, 0.020) and a significant relationship between cog-
Table 2, while controlling for age, sex, race, self-reported nitive activity and EF change (0.033, 95% CI: 0.003,
health, and household income, educational attainment 0.062). Participants with higher education showed less
was positively related to physical activity (0.072, 95% decline in EF and those with more frequent cognitive ac-
CI: 0.056, 0.087) and cognitive activity (0.065, 95% CI: tivity showed less decline in EF. The indirect effects for EF
0.055, 0.075). Both physical activity and cognitive activ- change through cognitive activity (indirect effect: 0.001,
ity were also positively related to EF, and the direct path 95% CI: 0.001, 0.002, κ2 = 0.006) was also significant,
between educational attainment and EF was significant indicating that cognitive activity had a small but signifi-
(direct effect: 0.108, 95% CI: 0.097, 0.118). The media- cant mediation effects on the relationship between educa-
tion analysis demonstrated that both physical activity tional attainment and change in EF.
(indirect effect: 0.004, 95% CI: 0.002, 0.006, κ2 = 0.012) The direct effect of educational attainment on EM
and cognitive activity (indirect effect: 0.013, 95% CI: change was significant as well (0.023, 95% CI: 0.009,
0.009, 0.016, κ2 = 0.035) had small but significant partial 0.036), indicating that participants with higher education

Table 1. Correlations for All Variables at Midlife in the United States (MIDUS) 2, N = 4,206

Variables M(SD) or % 1 2 3 4 5 6 7 8 9 10

1. Age 55.43 (12.45) —

2. Sex (% female) 55.3 .004 —
3. Physical activity 4.40 (1.31) −.287** −.055** —
4. Cognitive activity 3.94 (0.85) .027 .028 .118** —
5. Health 3.54 (1.01) −.184** −.024 .209** .160** —
6. Education 14.28 (2.62) −.144** −.103** .215** .240** .260** —
7. Race/ethnicity 91.9 .095** .000 .077** .065** .065** .040** —
(% white)
8. Episodic memory 0.02 (0.99) −.339** .222** .187** .135** .183** .213** .058** .157** —
9. Executive function 0.06 (0.97) −.431** −.113** .300** .238** .298** .412** .128** .317** .433** —

Note: **p < .05; SD = standard deviation.

 e98

Table 2. Coefficients, Standard Errors, and 95% Confidence Intervals for the Mediation Model for Executive Function and Episodic Memory

Direct effects Path coefficients

EF PA CA Indirect effect for EF

Variables b(SE) CI b(SE) CI b(SE) CI b(SE) CI

Age −0.029***(0.001) −0.032,−0.027 −0.027***(0.002) −0.030, −0.024 0.005***(0.001) 0.003, 0.006

Sex (female) −0.147***(0.026) −0.198,−0.095 −0.111***(0.041) −0.190, −0.032 0.076***(0.024) 0.028, 0.124
Race/ethnicity 0.391***(0.045) 0.304,0.479 0.323***(0.069) 0.189, 0.461 0.070(0.042) −0.025, 0.157
(white)
Education 0.108***(0.006) 0.097,0.118 0.072***(0.008) 0.056, 0.087 0.065***(0.005) 0.055, 0.075
Health 0.116***(0.014) 0.089,0.142 0.169*** (0.022) 0.126, 0.207 0.067***(0.013) 0.044, 0.093
PA 0.059***(0.011) 0.038,0.079 0.004***(0.001) 0.002, 0.006
CA 0.194***(0.019) 0.159,0.2228 0.013***(0.001) 0.009, 0.016

EM PA CA Indirect effect for EM

b(SE) CI b(SE) CI b(SE) CI b(SE) CI

Age −0.024***(0.001) −0.026,−0.021 −0.027***(0.002) −0.030, −0.024 0.005***(0.001) 0.003, 0.006

Sex 0.484***(0.027) 0.425,0.542 −0.109***(0.041) −0.188, −0.029 0.076***(0.024) 0.028, 0.124
(female)
Race/ethnicity 0.223***(0.048) 0.122,0.324 0.323***(0.069) 0.195, 0.468 0.070(0.042) −0.025, 0.157
(white)
Education 0.060***(0.006) 0.047,0.071 0.072***(0.008) 0.056, 0.087 0.065***(0.005) 0.055, 0.075
Health 0.057**(0.014) 0.081,0.137 0.168*** (0.022) 0.126, 0.207 0.067***(0.013) 0.044, 0.093
PA 0.044***(0.013) 0.020,0.068 0.003**(0.001) 0.001, 0.005
CA 0.107***(0.021) 0.067,0.146 0.007***(0.001) 0.004, 0.009

Notes: CA = cognitive activity; EF = executive function; EM = episodic memory; PA = physical activity.

*p < .1, **p < .05, ***p < .01.
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

Table 3. Coefficients, Standard Errors, and 95% Confidence Intervals for the Mediation Model for EF and EM Change

Direct effects Path coefficients

EF PA CA Indirect effect for EF

Variables b(SE) CI b(SE) CI b(SE) CI b(SE) CI

Age −0.015***(0.004) −0.017,−0.013 −0.016***(0.002) −0.021, −0.011 0.007***(0.001) 0.005, 0.010

Sex −0.028(0.020) −0.065,0.009 −0.092*(0.049) −0.187, 0.004 0.057**(0.024) 0.006, 0.108
(female)
Race/ethnicity 0.094***(0.038) 0.024,0.164 0.231**(0.092) 0.051, 0.411 0.115**(0.042) 0.019, 0.212
(white)
Education 0.012**(0.004) 0.004,0.020 0.065***(0.011) 0.043, 0.084 0.037***(0.005) 0.026, 0.048
Health 0.033***(0.011) 0.012,0.053 0.156*** (0.022) 0.103, 0.208 0.014(0.013) −0.014, 0.042
PA 0.013*(0.008) −0.002,0.029 0.001(0.001) −0.001, 0.002
CA 0.033**(0.015) 0.003,0.062 0.001***(0.001) 0.001, 0.002
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

EM PA CA Indirect effect for EM

b(SE) CI b(SE) CI b(SE) CI b(SE) CI

Age −0.014***(0.001) −0.029,−0.008 −0.017***(0.002) −0.018, −0.009 0.007***(0.001) 0.005, 0.010

Sex 0.316***(0.042) 0.183,0.124 −0.153*(0.049) −0.157, 0.0 0.057**(0.024) 0.006, 0.108
(female)
Race/ethnicity 0.056***(0.075) 0.288,0.120 0.264**(0.092) 0.051, 0.411 0.115**(0.042) 0.019, 0.212
(white)
Education 0.023***(0.009) 0.009,0.036 0.052***(0.011) 0.043, 0.084 0.050***(0.005) 0.026, 0.048
Health 0.056***(0.019) 0.021,0.092 0.168*** (0.022) 0.103, 0.208 0.025(0.013) −0.003, 0.053
PA 0.035**(0.011) 0.034,0.078 0.003***(0.001) 0.001, 0.005
CA 0.045*(0.019) −0.159,0.233 0.002(0.001) −0.001, 0.004

Notes. CA = cognitive activity; EF = executive function; EM = episodic memory; PA = physical activity.

*p < .1, **p < .05, ***p < .01.
e99

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

e100 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

showed less decline in EM. As shown in Table 3, there was Discussion

a significant relationship between physical activity and EM
This study demonstrated the role of both physical and
change (0.035, 95% CI: 0.034, 0.078), indicating that those
cognitive activity in the relationship between educational
with more frequent physical activity showed less decline
attainment and individual differences in cognitive function
in EM. The indirect effect for EM change through physi-
and change therein at middle age and later life in a large
cal activity (indirect effect: 0.003, 95% CI: 0.001, 0.005,
cohort from across the United States. The results were con-
κ2 = 0.037) was also significant, indicating that physical
sistent with previous findings that individuals with higher
activity had small but significant mediation effects on the
educational attainment had better cognitive functioning in
relationship between educational attainment and change in
later adulthood (Wilson et al., 2009). Given the importance
EM. The results of the models for EF and EM level and
of understanding the disparate findings in previous litera-
change are presented in Figures 1 and 2, respectively. Note
ture on the association between educational attainment

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

that for all four models we tested whether age was a mod-
and cognitive change in later life, our study added to this
erator, and found no significant moderating effects for age.
discussion by showing both direct and indirect effects for
the association between educational attainment and cogni-
tive change.
Previous studies have shown wide variability in aging-
related changes in cognition (Mella et al., 2018; Schaie
et al., 2004). The present study examined two modifiable
behavioral factors, both physical and cognitive activity, to-
gether as possible mediators that could mitigate cognitive
decline in old age and clarify mechanisms linking educa-
tion and cognition (Lachman et al., 2010). In line with our
hypotheses and previous work (Hall et al., 2007; Stern,
2012), participants with higher levels of educational attain-
ment reported being more active both physically and cog-
nitively. The results provided support for the independent
effects of physical activity and cognitive activity on level
Figure 1. Mediation models for executive function (Model 1) and ex- of both EF and EM. Consistent with previous findings that
ecutive function change (Model 2). Model 1: the relationship of educa- higher educational attainment was associated with greater
tion and level of executive function, mediated by physical activity and
participation in various lifestyle activities (Allet et al.,
cognitive activity. *p < .05. Model 2: the relationship of education and
change in executive function, mediated by physical activity and cog-
2016; Hess, 2014; Lachman et al., 2010; Thrane, 2006),
nitive activity. Model 2 parameters are presented in square brackets. our study also found that higher educational attainment
*p < .05. Indirect effect of physical activity: 0.004, 95% CI: 0.002, 0.006, was associated with greater levels of physical and cognitive
κ2 = 0.012 [0.001, 95% CI: −0.001, 0.002, κ2 = 0.001]; indirect effect of activities. Education can provide advantages to older adults
cognitive activity: 0.013, 95% CI: 0.009, 0.016, κ2 = 0.035 [0.001, 95% CI: by increasing access to resources and opportunities for en-
0.001, 0.002, κ2 = 0.006].
gaging in various physical activities and continued intellec-
tual stimulation across the life course, thereby resulting in
improved cognitive functioning in late adulthood (Wilson
et al., 2009).
Although the effect sizes were small, and we cannot
make direct causal inferences from the data, the findings
that cognitive activity was significant as a predictor and
mediator of EF change and that physical activity was a sig-
nificant predictor and mediator for EM change adds to this
literature, supporting the notion that physical and cognitive
activity may play a role in protecting against age-related
declines in cognition. We found no moderating effects of
Figure 2. Mediation models for episodic memory (Model 3) and epi- age on the mediating effects of physical activity and cogni-
sodic memory change (Model 4). Model 3: the relationship of education tive activity, suggesting that, contrary to predictions, the
and level of episodic memory, mediated by physical activity and cogni-
effects were of equal magnitude across the adult life span.
tive activity. *p < .05. Model 4: the relationship of education and change
in episodic memory, mediated by physical activity and cognitive activ-
Physical activity has been found in other studies to be
ity. Model 4 parameters are presented in square brackets. *p < .05. related to both cognitive function and cognitive change
Indirect effect of physical activity: 0.003, 95% CI: 0.001, 0.005, κ2 = 0.008 (Barnes, Yaffe, Satariano, & Tager, 2003; Renaud, Bherer, &
[0.003, 95% CI: 0.001, 0.005, κ2 = 0.037]; indirect effect of cognitive activ- Maquestiaux, 2010; Robinson & Lachman, 2018). Within
ity: 0.007, 95% CI: 0.004, 0.009, κ2 = 0.033 [0.002, 95% CI: −0.001, 0.004, the cognitive domain, there is evidence for exercise-related
κ2 = 0.001].
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7 e101

improvements for both executive functioning (e.g., process- and cognitive activities. There is also the possibility that
ing speed; Frederiksen et al., 2015) and memory (e.g., spa- errors in reporting these activities may be correlated, so
tial/episodic memory; Erickson et al., 2015). Older adults that any confounding between the self-report measures
who have completed a physical activity program that pro- may be increased by relying on self-report. In future stud-
duces significant increases in cardiorespiratory fitness often ies, mediation analysis with latent variables could include
show enhanced EM (Barnes et al., 2003; Kramer et al., a correlated error term between cognitive and physical
1999; Smith et al., 2010). Although previous evidence activities. Future work could also use objective assessments
shows EF is affected by physical activity (Kramer et al., of activity and more items at each wave of assessment to
1999; Smith et al., 2010), our findings indicated that physi- help reduce measurement error (Barnett, van der Pols, &
cal activity mediated the relationship between educational Dobson, 2005).
attainment and change in EM but not EF. Another limitation is that the participants in the cog-

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

The results from our study that cognitive activity was nitive sample were not screened for cognitive impairment
an important factor for EF and change therein, beyond or dementia; all participants were included in the anal-
educational attainment and physical activity, supported ysis. Although only a small percentage of participants had
and extended previous findings that environmental condi- stroke, heart disease, or other factors that might affect
tions—such as cognitive stimulation in the home—are most cognitive function, a goal for future waves of MIDUS is to
robustly associated with aspects of cognition such as lan- screen for cognitive impairment. In addition, there was a
guage, attention, and other EMs (Greenfield & Moorman, lack of racial diversity in the MIDUS sample; the vast ma-
2018; Noble et al., 2015; Peyre et al., 2016). jority of participants were non-Hispanic whites. Thus, ad-
The present results suggest the importance of taking a ditional research is needed to examine the generalizability
developmental and longitudinal approach to investigate of the findings in more diverse samples.
individual differences in antecedent variables that lead to
greater decrements for some persons and maintenance of
high levels of functioning for others. Educational attain- Implications for Policy and Practice
ment is related to better cognitive functioning and physical As Hertzog and colleagues (2008) indicated, previous re-
and cognitive activity were supported as possible mecha- search and practice focus has been on short-term gains in
nisms. Because we were able to examine both the direct older adults, rather than proactive intervention at younger
and indirect effects, we can conclude that educational at- ages to produce long-term effects. In general, more at-
tainment has not only a direct effect but also indirect effects tention should be given to how interventions in mid-
through physical and cognitive activity. Furthermore, the life could be structured to promote and enhance health
results provide evidence that both cognitive and physical and well-being, productivity, and cognitive development
activity have independent effects for level and change. In (Hertzog et al., 2008). Our findings regarding the mediat-
previous research these two forms of activity have not ing effect of physical activity and cognitive activity on the
typically been studied together. Moreover, this study was association between educational attainment and cognitive
conducted with a large longitudinal sample to test whether function over time have important implications for under-
physical and cognitive activity are unique contributors to standing the factors predicting cognitive functioning and
cognitive function and change in middle age and later life. cognitive change associated with aging. The study results
The study also includes a test battery that covers multiple suggest that engaging in physical and cognitive activity in
key aspects of cognition that are associated with cognitive midlife may be one explanation for how educational attain-
aging and sensitive to change across the adult life span ment affects level of cognitive functioning at old age. This
(Lachman & Tun, 2008). study highlights the importance of engaging in physical and
cognitive activity in midlife and older adulthood and can
inform future policy work and intervention development
Limitations and Future Research aimed at enhancing physical and cognitive health in an
There are some limitations that should be considered as aging population. However, a recent report suggests that
future studies continue and expand on this work. One limi- only about 20% of adults meet the recommended guide-
tation is a lack of objective measurements of both physical lines for physical activity (Clarke, Ward, Norris, & Schiller,
and cognitive activity. Although self-report measurements 2017). Policies and programs are needed to promote reg-
are useful to help gain insight into one’s level of physi- ular physical activity among older adults as a means to
cal and cognitive activity, they possess several limitations maintaining cognitive health.
in terms of reliability and validity, such as a capacity to This study also provided support for the benefits of
over- or underestimate true physical and cognitive activ- cognitive activities for older adults including formal
ity and potential issues of recall and response bias (e.g., cognitive training and informal cognitive interventions.
social desirability, inaccurate memory; Prince et al., 2008; Park and colleagues (2014) found that older adults who
Shephard, 2003). In addition, participants’ level of EF and learned quilting or digital photography had more memory
EM may differentially influence the reporting of physical improvement than those who only socialized or did less
e102 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

cognitively demanding activities. Formal cognitive train- References

ings such as the ACTIVE should be supported and pro- Aichberger, M. C., Busch, M. A., Reischies, F. M., Ströhle, A.,
moted among older adults (Brodziak, Wolińska, Kołat, & Heinz, A., & Rapp, M. A. (2010). Effect of physical inactivity on
Różyk-Myrta, 2015). In the ACTIVE trials, healthy adults cognitive performance after 2.5 years of follow-up. Longitudinal
65 years and older participated in 10 sessions of memory results from the Survey of Health, Ageing and Retirement
training, reasoning training, or processing-speed training. (SHARE). GeroPsych, 23, 7–15. doi:10.1024/1662-9647/
The sessions improved participants’ cognitive skills in the a000003
area in which they were trained. Most of these improve- Albinet, C. T., Boucard, G., Bouquet, C. A., & Audiffren, M. (2010).
ments persisted 10 years after the training was completed Increased heart rate variability and executive performance after
(Rebok et al., 2014). aerobic training in the elderly. European Journal of Applied
Many physicians view mild cognitive impairment, age- Physiology, 109, 617–624. doi:10.1007/s00421-010-1393-y

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

related decline in memory, and other cognitive processes Allet, L., Giet, O., Barral, J., Junod, N., Durrer, D., Amati, F., …
as a transitional phase between normal cognitive aging Puder, J. J. (2016). Educational level is related to physical fitness
and dementia, although data regarding the actual rate of in patients with type 2 diabetes—a cross-sectional study. PloS
crossover to dementia are inconclusive (Blieszner, Roberto, One, 11, e0164176. doi:10.1371/journal.pone.0164176
Wilcox, Barham, & Winston, 2007). Understanding the fac- Alley, D., Suthers, K., & Crimmins, E. (2007). Education and cogni-
tors that relate to cognitive function and cognitive change tive decline in older Americans: Results from the AHEAD sample.
in midlife may be important for improving early interven- Research on Aging, 29, 73–94. doi:10.1177/0164027506294245
Baron, R. M., & Kenny, D. A. (1986). The moderator–media-
tions for dementia and related disorders. Consistent with
tor variable distinction in social psychological research:
previous findings that physical activity reduces the rate of
Conceptual, strategic, and statistical considerations. Journal of
cognitive decline in word list delayed recall and clinical
Personality and Social Psychology, 51, 1173–1182. doi:10.1037
dementia rating among older adults at risk (Lautenschlager
%2F0022-3514.51.6.1173
et al., 2008), our findings indicated that physical activ-
Barnes, D. E., Yaffe, K., Satariano, W. A., & Tager, I. B. (2003). A lon-
ity reduced the rate of decline in EM. This is particularly
gitudinal study of cardiorespiratory fitness and cognitive func-
important as the transition from midlife to older adulthood tion in healthy older adults. Journal of the American Geriatrics
typically coincides with the transition from working to Society, 51, 459–465. doi:10.1046/j.1532-5415.2003.51153.x
retirement, where regular activity from a steadier schedule Barnett, A. G., van der Pols, J. C., & Dobson, A. J. (2005). Regression
can be disrupted and there is a need to find strategies to to the mean: What it is and how to deal with it. International
maintain consistent activity (Robinson & Lachman, 2018). Journal of Epidemiology, 34, 215–220. doi:10.1093/ije/dyh299
In summary, this study was a strong test of the inde- Bherer, L., Erickson, K. I., & Liu-Ambrose, T. (2013). A review of
pendent mediation effects of physical activity and cognitive the effects of physical activity and exercise on cognitive and
activity on the relationship between educational attainment brain functions in older adults. Journal of Aging Research, 2013,
and cognitive function. This study adds to the literature on 657508. doi:10.1155/2013/657508
the importance of both physical and cognitive activity in Blieszner, R., Roberto, K. A., Wilcox, K. L., Barham, E. J., &
the cognitive function and cognitive change in middle age Winston, B. L. (2007). Dimensions of ambiguous loss in couples
and later life. The longitudinal findings suggest that educa- coping with mild cognitive impairment. Family Relations, 56,
tional attainment, physical activity, and cognitive activity 196–209. doi:10.1111/j.1741-3729.2007.00452.x
may be particularly important precursors to cognitive func- Brim, O. G., Ryff, C. D., & Kessler, R. C. (Eds.). (2004). How healthy
tion that may have the potential for mitigating cognitive are we?: A national study of well-being at midlife. Chicago, IL:
declines in middle age and later life. University of Chicago Press.
Brodziak, A., Wolińska, A., Kołat, E., & Różyk-Myrta, A. (2015).
Guidelines for prevention and treatment of cognitive impair-
Supplementary Material ment in the elderly. Medical Science Monitor, 21, 585–597.
Supplementary data are available at The Journals of doi:10.12659/MSM.892542
Gerontology, Series B: Psychological Sciences and Social Clarke, T. C., Ward, B. W., Norris, T., & Schiller, J. S. (2017). Early
release of selected estimates based on data from the National
Sciences online.
Health Interview Survey, January–September 2016: Lack of
health insurance coverage and type of coverage. Hyattsville, ML:
Funding Center for Disease Control and Prevention.
Cohen, J. (1988). Statistical power analysis for the behavioral sci-
This study was funded by the National Institute on Aging
ences. (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
(NIA) Grants no. P01-AG020166 and U19-AG051426.
Cotter, K. A., & Lachman, M. E. (2010). No strain, no gain:
Psychosocial predictors of physical activity across the adult
Conflict of Interest lifespan. Journal of Physical Activity and Health, 7, 584–594.
Retrieved from https://www.ncbi.nlm.nih.gov/pmc/articles/
None reported.
PMC2972196/
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7 e103

Daugherty, A. M., Zwilling, C., Paul, E. J., Sherepa, N., Allen, C., (MIDUS) study. Journal of the International Neuropsychological
Kramer, A. F., … Barbey, A. K. (2018). Multimodal fitness and Society, 24, 805–820. doi:10.1017/S1355617718000425
cognitive training to enhance fluid intelligence. Intelligence, 66, Kirk-Sanchez, N. J., & McGough, E. L. (2014). Physical exercise
32–43. doi:10.1016/j.intell.2017.11.001 and cognitive performance in the elderly: Current perspectives.
Erickson, K. I., Hillman, C. H., & Kramer, A. F. (2015). Physical Clinical Interventions in Aging, 9, 51–62. doi:10.2147/CIA.
activity, brain, and cognition. Current Opinion in Behavioral S39506
Sciences, 4, 27–32. doi:10.1016/j.cobeha.2015.01.005 Kramer, A. F., Hahn, S., Cohen, N. J., Banich, M. T., McAuley, E.,
Fratiglioni, L., Paillard-Borg, S., & Winblad, B. (2004). An active Harrison, C. R., … Colcombe, A. (1999). Ageing, fitness and neu-
and socially integrated lifestyle in late life might protect against rocognitive function. Nature, 400, 418–419. doi:10.1038/22682
dementia. Lancet. Neurology, 3, 343–353. doi:10.1016/ Lachman, M. E., Agrigoroaei, S., Murphy, C., & Tun, P. A. (2010).
S1474-4422(04)00767-7 Frequent cognitive activity compensates for education dif-

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

Frederiksen, K. S., Verdelho, A., Madureira, S., Bäzner, H., ferences in episodic memory. American Journal of Geriatric
O’Brien, J. T., Fazekas, F., … Waldemar, G.; LADIS Study. Psychiatry, 18, 4–10. doi:10.1097/JGP.0b013e3181ab8b62
(2015). Physical activity in the elderly is associated with Lachman, M. E., Agrigoroaei, S., Tun, P. A., & Weaver, S. L. (2014).
improved executive function and processing speed: The LADIS Monitoring cognitive functioning: psychometric properties of
Study. International Journal of Geriatric Psychiatry, 30, 744– the Brief Test of Adult Cognition by Telephone. Assessment, 21,
750. doi:10.1002/gps.4220 404–417. doi:10.1177/1073191113508807
Fritsch, T., McClendon, M. J., Smyth, K. A., Lerner, A. J., Lachman, M. E., & Tun, P. A. (2008). Cognitive testing in large-scale
Chen, C. H., Petot, G. J., … Friedland, R. P. (2001). Effects of surveys: Assessment by telephone. In Hofer, S. M. & Alwin, D. F.,
educational attainment on the clinical expression of Alzheimer’s (Eds.), Handbook of cognitive aging: Interdisciplinary perspec-
disease: Results from a research registry. American Journal tives (pp. 506–523). Thousand Oaks, CA: Sage.
of Alzheimer’s Disease and Other Dementias, 16, 369–376. Lautenschlager, N. T., Cox, K. L., Flicker, L., Foster, J. K.,
doi:10.1177/153331750101600606 van Bockxmeer, F. M., Xiao, J., … Almeida, O. P. (2008). Effect
Ghisletta, P., Bickel, J. F., & Lövdén, M. (2006). Does activ- of physical activity on cognitive function in older adults at risk
ity engagement protect against cognitive decline in old age? for Alzheimer disease: A randomized trial. JAMA, 300, 1027–
Methodological and analytical considerations. The Journals 1037. doi:10.1001/jama.300.9.1027
of Gerontology, Series B: Psychological Sciences and Social Le Carret, N., Lafont, S., Mayo, W., & Fabrigoule, C. (2003).
Sciences, 61, 253–261. doi:10.1093/geronb/61.5.P253 The effect of education on cognitive performances and its
Glymour, M. M., Weuve, J., Berkman, L. F., Kawachi, I., & implication for the constitution of the cognitive reserve.
Robins, J. M. (2005). When is baseline adjustment useful in Developmental Neuropsychology, 23, 317–337. doi:10.1207/
analyses of change? An example with education and cognitive S15326942DN2303_1
change. American Journal of Epidemiology, 162, 267–278. Lövdén, M., Bäckman, L., Lindenberger, U., Schaefer, S., &
doi:10.1093/aje/kwi187 Schmiedek, F. (2010). A theoretical framework for the study of
Greenfield, E. A., & Moorman, S. M. (2018). Childhood socio- adult cognitive plasticity. Psychological Bulletin, 136, 659–676.
economic status and later life cognition: Evidence from the doi:10.1037/a0020080
Wisconsin longitudinal study. Journal of Aging and Health. Mella, N., Fagot, D., Renaud, O., Kliegel, M., & de Ribaupierre, A.
doi:10.1177/0898264318783489 (2018). Individual differences in developmental change:
Hall, C. B., Derby, C., LeValley, A., Katz, M. J., Verghese, J., & Quantifying the amplitude and heterogeneity in cognitive change
Lipton, R. B. (2007). Education delays accelerated decline on a across old age. Journal of Intelligence, 6, 10. doi:10.3390/
memory test in persons who develop dementia. Neurology, 69, jintelligence6010010
1657–1664. doi:10.1212/01.wnl.0000278163.82636.30 Mungas, D., Beckett, L., Harvey, D., Farias, S. T., Reed, B.,
Hayes, A. F., & Preacher, K. J. (2014). Statistical mediation analy- Carmichael, O., … DeCarli, C. (2010). Heterogeneity of cogni-
sis with a multicategorical independent variable. British Journal tive trajectories in diverse older persons. Psychology and Aging,
of Mathematical and Statistical Psychology, 67, 451–470. 25, 606–619. doi:10.1037/a0019502
doi:10.1111/bmsp.12028 Noble, K. G., Engelhardt, L. E., Brito, N. H., Mack, L. J., Nail, E. J.,
Hertzog, C., Kramer, A. F., Wilson, R. S., & Lindenberger, U. Angal, J., … Elliott, A. J.; PASS Network. (2015). Socioeconomic
(2008). Enrichment Effects on adult cognitive development: disparities in neurocognitive development in the first two years of
Can the functional capacity of older adults be preserved and life. Developmental Psychobiology, 57, 535–551. doi:10.1002/
enhanced? Psychological Science in the Public Interest, 9, 1–65. dev.21303
doi:10.1111/j.1539-6053.2009.01034.x Park, D. C., Lodi-Smith, J., Drew, L., Haber, S., Hebrank, A.,
Hess, T. M. (2014). Selective engagement of cognitive resources: Bischof, G. N., & Aamodt, W. (2014). The impact of sus-
Motivational influences on older adults’ cognitive func- tained engagement on cognitive function in older adults:
tioning. Perspectives on Psychological Science, 9, 388–407. The Synapse Project. Psychological Science, 25, 103–112.
doi:10.1177/1745691614527465 doi:10.1177/0956797613499592
Hughes, M. L., Agrigoroaei, S., Jeon, M., Bruzzese, M., & Peyre, H., Bernard, J. Y., Hoertel, N., Forhan, A., Charles, M.-A.,
Lachman, M. E. (2018). Change in cognitive performance from De Agostini, M., & Ramus, F. (2016). Differential effects of fac-
midlife into old age: Findings from the Midlife in the United States tors influencing cognitive development at the age of 5-to-6 years.
e104 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 7

Cognitive Development, 40, 152–162. doi:10.1016/j. Medicine, 72, 239–252. doi:10.1097/PSY.0b013e3181

cogdev.2016.10.001 d14633
Preacher, K. J., & Kelley, K. (2011). Effect size measures for media- Stern, Y. (2009). Cognitive reserve. Neuropsychologia, 47, 2015–
tion models: Quantitative strategies for communicating indi- 2028. doi:10.1016/j.neuropsychologia.2009.03.004
rect effects. Psychological Methods, 16, 93–115. doi:10.1037/ Stern, Y. (2012). Cognitive reserve in ageing and Alzheimer’s
a0022658 disease. Lancet. Neurology, 11, 1006–1012. doi:10.1016/
Prince, S. A., Adamo, K. B., Hamel, M. E., Hardt, J., Connor Gorber, S1474-4422(12)70191-6
S., & Tremblay, M. (2008). A comparison of direct versus self- Stine-Morrow, E. A. L., Payne, B. R., Roberts, B. W., Kramer, A. F.,
report measures for assessing physical activity in adults: A sys- Morrow, D. G., Payne, L., … Parisi, J. M. (2014). Training ver-
tematic review. The International Journal of Behavioral Nutrition sus engagement as paths to cognitive enrichment with aging.
and Physical Activity, 5, 56. doi:10.1186/1479-5868-5-56 Psychology and Aging, 29, 891–906. doi:10.1037/a0038244

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/7/e93/5336661 by guest on 06 August 2024

Radler, B. T., & Ryff, C. D. (2010). Who participates? Accounting Sturman, M. T., Morris, M. C., Mendes de Leon, C. F., Bienias, J. L.,
for longitudinal retention in the MIDUS national study of health Wilson, R. S., & Evans, D. A. (2005). Physical activity, cogni-
and well-being. Journal of Aging and Health, 22, 307–331. tive activity, and cognitive decline in a biracial community popu-
doi:10.1177/0898264309358617 lation. Archives of Neurology, 62, 1750–1754. doi:10.1001/
Rebok, G. W., Ball, K., Guey, L. T., Jones, R. N., Kim, H.-Y., archneur.62.11.1750
King, J. W., … & Willis, S. L. (2014). Ten-year effects of the Tennstedt, S. L., & Unverzagt, F. W. (2013). The ACTIVE study:
ACTIVE cognitive training trial on cognition and everyday Study overview and major findings. Journal of Aging and Health,
functioning in older adults. Journal of the American Geriatrics 25, 3S–20S. doi:10.1177/0898264313518133
Society, 62, 16–24. doi:10.1111/jgs.12607 Thrane, C. (2006). Explaining educational-related inequalities in health:
Renaud, M., Bherer, L., & Maquestiaux, F. (2010). A high level Mediation and moderator models. Social Science and Medicine
of physical fitness is associated with more efficient response (1982), 62, 467–478. doi:10.1016/j.socscimed.2005.06.010
preparation in older adults. The Journals of Gerontology, Series Tucker-Drob, E. M., Johnson, K. E., & Jones, R. N. (2009). The cog-
B: Psychological Sciences and Social Sciences, 65, 317–322. nitive reserve hypothesis: A longitudinal examination of age-asso-
doi:10.1093/geronb/gbq004 ciated declines in reasoning and processing speed. Developmental
Robinson, S. A., & Lachman, M. E. (2018). Perceived control and Psychology, 45, 431–446. doi:10.1037/a0014012
cognition in adulthood: The mediating role of physical activity. Wang, J. Y., Zhou, D. H., Li, J., Zhang, M., Deng, J., Tang, M., …
Psychology and Aging, 33, 769–781. doi:10.1037/pag0000273 Chen, M. (2006). Leisure activity and risk of cognitive impair-
Salthouse, T. A. (2014). Correlates of cognitive change. Journal ment: The Chongqing aging study. Neurology, 66, 911–913.
of Experimental Psychology. General, 143, 1026–1048. doi:10.1212/01.wnl.0000192165.99963.2a
doi:10.1037/a0034847 Wilson, R. S., Beckett, L. A., Barnes, L. L., Schneider, J. A., Bach, J.,
Schaie, K. W., Willis, S. L., & Caskie, G. I. (2004). The Seattle lon- Evans, D. A., & Bennett, D. A. (2002). Individual differences in
gitudinal study: Relationship between personality and cogni- rates of change in cognitive abilities of older persons. Psychology
tion. Neuropsychology, Development, and Cognition. Section and Aging, 17, 179–193. doi:10.1037/0882-7974.17.2.179
B, Aging, Neuropsychology and Cognition, 11, 304–324. Wilson, R. S., Hebert, L. E., Scherr, P. A., Barnes, L. L.,
doi:10.1080/13825580490511134 Mendes de Leon, C. F., & Evans, D. A. (2009). Educational
Shephard, R. J. (2003). Limits to the measurement of habitual physi- attainment and cognitive decline in old age. Neurology, 72,
cal activity by questionnaires. British Journal of Sports Medicine, 460–465. doi:10.1212/01.wnl.0000341782.71418.6c
37, 197–206. doi:10.1136/bjsm.37.3.197 Zhu, W., Wadley, V. G., Howard, V. J., Hutto, B., Blair, S. N., &
Smith, P. J., Blumenthal, J. A., Hoffman, B. M., Cooper, H., Hooker, S. P. (2017). Objectively measured physical activ-
Strauman, T. A., Welsh-Bohmer, K., … Sherwood, A. (2010). ity and cognitive function in older adults. Medicine and
Aerobic exercise and neurocognitive performance: A meta- Science in Sports and Exercise, 49, 47–53. doi:10.1249/
analytic review of randomized controlled trials. Psychosomatic MSS.0000000000001079