University of Southern California Dissertations and Theses

Relationships between physical activity and cognitive ability in older adults: investigating the mediating effects of sleep efficiency and cerebral blood flow

(USC Thesis Other)

Relationships between physical activity and cognitive ability in older adults: investigating the mediating effects of sleep efficiency and cerebral blood flow

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content RELATIONSHIPS BETWEEN PHYSICAL ACTIVITY AND COGNITIVE ABILITY IN
OLDER ADULTS:
INVESTIGATING THE MEDIATING EFFECTS OF SLEEP EFFICIENCY AND CEREBRAL
BLOOD FLOW
by
Laura Fenton
A Thesis Presented to the
FACULTY OF THE USC SCHOOL OF DORNSIFE COLLEGE OF LETTERS, ARTS AND
SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(PSYCHOLOGY )
August 2021
Copyright (2021) Laura Fenton

ii
TABLE OF CONTENTS
List of Tables ................................................................................................................................. iii
List of Figures ................................................................................................................................ iv
Abstract ............................................................................................................................................v
Introduction ......................................................................................................................................1
Physical Activity, Brain Health, and Cognitive Ability ......................................................3
Sleep Efficiency as a Mediator ............................................................................................7
Cerebral Blood Flow as a Mediator ...................................................................................11
Current Study .................................................................................................................................15
Methods..........................................................................................................................................17
Study Population ................................................................................................................17
Measures ............................................................................................................................19
Physical Activity ....................................................................................................19
Sleep Efficiency .....................................................................................................20
Cerebral Blood Flow ..............................................................................................20
Cognitive Ability ...................................................................................................22
Data Analysis .....................................................................................................................24
Results ............................................................................................................................................30
Aim 1 Results .....................................................................................................................30
Aim 2 Results .....................................................................................................................36
Aim 3 Results .....................................................................................................................37

Discussion ......................................................................................................................................43
Study Limitations ...........................................................................................................................47
References ......................................................................................................................................48
Appendix ........................................................................................................................................63
Aim 1 Supplemental Tables ...............................................................................................63
Aim 3 Supplemental Tables ...............................................................................................85

iii
List of Tables
1. Sample Characteristics ...............................................................................................................18
2. Sleep Efficiency as Proposed Mediator .....................................................................................31
3. Global Gray Matter Cerebral Blood Flow as Proposed Mediator .............................................34
4. Aim 3 Model Comparisons ........................................................................................................39
5. Aim 3 Parameter Estimates ........................................................................................................40
6. Physical Activity as Proposed Mediator ....................................................................................63
7. Sleep Efficiency as Proposed Mediator – Light Activity Only .................................................65
8. Global Gray Matter Cerebral Blood Flow as Proposed Mediatory – Light Activity Only ........68
9. Sleep Efficiency as Proposed Mediator – Moderate Activity Only ...........................................70
10. Global Gray Matter Cerebral Blood Flow as Proposed Mediator – Moderate Activity Only .73
11. Hippocampal Cerebral Blood Flow as Proposed Mediator .....................................................75
12. Anterior Cingulate Gyrus Cerebral Blood Flow as Proposed Mediator ..................................78
13. Middle Frontal Gyrus Cerebral Blood Flow as Proposed Mediator ........................................80
14. Superior Frontal Gyrus Cerebral Blood Flow as Proposed Mediator ......................................83
15. Aim 3 Model Comparisons – Light Activity Only ..................................................................85
16. Aim 3 Model Comparisons – Moderate Activity Only ...........................................................86

iv
List of Figures
1. Lifestyle and Intrinsic Factors Related to Alzheimer’s Disease ..................................................2
2. Effects of Aerobic Exercise on Cognitive Domains ....................................................................5
3. Mechanisms Linking Physical Activity to Cognitive Functioning ..............................................6
4. Typical Sleep Architecture ..........................................................................................................7
5. Patterns of Change in Cerebral Blood Flow ..............................................................................12
6. Hypothetical Model of Biomarker Change Across Alzheimer’s Disease .................................13
7. General Procedures for Cerebral Blood Flow Measurement .....................................................21
8. Latent Variable Model ...............................................................................................................24
9. Sleep Efficiency Mediation Model ............................................................................................25
10. Cerebral Blood Flow Mediation Model ...................................................................................25
11. Aim 2 Within- and Between-Person Models ...........................................................................27
12. Aim 3 Within- and Between-Person Models .......................................................................... 29
13. Hypothetical Physical Activity Patterns ..................................................................................38
14. Actual Physical Activity Patterns ............................................................................................39
15. Aim 3 Scatter Plots ..................................................................................................................42

v
Abstract
Due to the lack of disease-modifying treatments for dementia, elucidating ways to
prevent or delay disease onset is a research priority. Physical activity has emerged as a promising
intervention target, but there is still uncertainty regarding the mechanisms through which
physical activity confers benefits on brain health and cognition. The current study investigated
the mediating effects of sleep efficiency and global gray matter cerebral blood flow (CBF)
measured via arterial spin labeling (ASL) on the relationship between physical activity and
cognitive ability in a sample of older adults (N=52, mean age=66.5, 67% Female). Using
mediation models, we tested whether average sleep efficiency and CBF mediated the effect of
physical activity on cognitive ability. To address temporal relationships, random effects models
were used to investigate the effect of daily physical activity on the subsequent night’s sleep
efficiency. Dynamic structural equation models were used to investigate whether day-to-day
fluctuations in physical activity were associated with cognitive ability and CBF. Contrary to our
hypotheses, higher levels of physical activity were not associated with better cognitive ability,
directly or via our proposed mediators. Random effects models revealed no association between
daily physical activity levels and sleep efficiency. Although day-to-day patterns of physical
activity were not associated with cognition or CBF, Bayesian model selection favored models
that included their effect for outcomes of visual memory and CBF. These findings suggest that
free-living levels of physical activity in sedentary older adults may not be associated with sleep,
CBF, or cognitive ability, and that daily patterns of physical activity may provide important
information when looking at brain health and cognitive outcomes.

Keywords: physical activity, sleep, cerebral blood flow, cognition, dementia

1
Introduction
According to the 2020 Alzheimer’s Association Facts and Figures, there are currently 5.8
million people in the United States living with Alzheimer’s Disease (AD) with the number of
new cases of dementia expected to double by 2050 (“Alzheimer’s Disease Facts and Figures,”
2020). Unfortunately, the drugs approved to treat AD by the U.S. Food and Drug Administration
do not stop the progression of the disease. Due to the lack of disease-modifying treatments for
dementia, identifying modifiable risk factors that can prevent or delay the onset of dementia is a
research priority. Physical activity is a promising intervention target as it has not only been
shown to reduce the risk of age-related cognitive decline, but also aids in the management of
cardiovascular risk factors (e.g., diabetes, obesity, hypertension) shown to increase dementia risk
(Baumgart et al., 2015; Edwards et al., 2019). Additionally, physical activity is widely accessible
and cost-effective. Despite the easy accessibility and multitude of positive benefits conferred by
a physically active lifestyle, it is estimated that less than 10% of adults meet the threshold for the
recommended amount of physical activity (150 minutes of moderate aerobic physical activity or
75 minutes of vigorous activity) per week (Tucker et al., 2011; "Physical Activity Guidelines for
Americans, 2008). Meanwhile, it has been estimated that physical inactivity may account for
approximately 21% (over 1.1. million) of cases of AD in the United States. Based on this
estimate, reducing the prevalence of physical inactivity by just 10% could prevent as many as
90,000 cases of AD (Barnes & Yaffe, 2011). The overall purpose of the current study is to
improve our understanding of the mechanisms underlying the association between physical
activity and cognition in order to reduce the incidence of AD.
There are now 12 identified modifiable risk factors for dementia: physical inactivity,
hypertension, obesity, diabetes, late-life depression, social isolation, hearing loss, smoking,

2
traumatic brain injury, alcohol, air pollution, and education (Livingston et al., 2020) (See Figure
1 for a visualization of lifestyle factors related to these risk factors).

Figure 1: Courtesy of Pa Lab, USC INI. Illustrates the general way in which lifestyle and intrinsic factors
interact to confer pathology.

Physical activity is a promising protective factor because it can impact multiple risk
factors simultaneously. Following this logic, the association between physical activity and
cognition may be mediated by the engagement of other risk factors. While there are many
candidate mechanisms, the case for sleep and vascular health are particularly compelling with
mounting evidence implicating their roles early in the AD process (Govindpani et al., 2019;
Mander et al., 2017; Nation et al., 2019; Zlokovic, 2011). Therefore, engagement and
modification of these early markers of disease through physical activity may be a way to delay
and/or avoid dementia onset. Elucidating the ways in which these two factors are related to
physical activity and cognition will not only lend strong support to the importance of exercise
interventions and public health initiatives aimed at increasing engagement in routine physical

3
activity, but also guide clinicians and researchers by providing clinically relevant targets to
engage. For example, evidence for a significant mediating effect of sleep would provide strong
rationale for the inclusion of sleep measures in studies of physical activity and encourage
clinicians to emphasize the importance of both physical activity and sleep when talking to
patients or the general public. Being able to provide this type of directed guidance on the
importance of physical activity and the mechanisms through which it confers cognitive benefits
is of critical importance in order to reduce the prevalence of physical inactivity and avoid
preventable AD cases.
The subsequent sections will provide a review of the literature on physical activity, brain
health and cognitive ability and summarize the evidence to support the roles of sleep efficiency
and cerebral blood flow (CBF) as mediators of the relationship between physical activity and
cognitive ability.

Physical Activity, Brain Health, and Cognitive Ability
Physical activity is broadly defined as the use of skeletal muscles to produce bodily
movement which results in energy expenditure (Caspersen et al., 1985). The notion that physical
activity influences brain health and cognitive ability has been recognized for decades. An early
meta-analytic review of 134 studies found evidence to suggest a small beneficial effect of
exercise on cognitive ability, but issued caution in interpreting this effect, highlighting the need
for additional experimental studies assessing the impact of methodological factors such as
exercise type, participant demographics, and cognitive outcome measures (Etnier et al., 1997).
More recent meta-analyses have added additional support to the idea that exercise interventions
are an effective way to improve cognitive ability in older adults (Colcombe & Kramer, 2003;

4
Hillman et al., 2008; Northey et al., 2018) and that physical activity throughout the lifespan can
lower one’s risk of cognitive impairment later in life (Blondell et al., 2014; Sofi et al., 2011;
Stephen et al., 2017). Despite the general consensus that physical activity is associated with
improved cognitive ability and a lower risk of cognitive decline, recent reviews still emphasize
the fact that there is insufficient evidence to make specific recommendation about the optimal
type, frequency, intensity or duration of physical activity (Kramer & Colcombe, 2018; Stephen et
al., 2017). Additionally, most studies have investigated average levels of physical activity in
older adults, ignoring the importance of day-to-day patterns of activity (Watts et al., 2016).
In an effort to systematically assess the strength of the evidence for physical activity’s
beneficial effects on cognitive outcomes across ages and health conditions and to inform federal
policy related to physical activity guidelines, the 2018 Physical Activity Guidelines Advisory
Committee recently published findings from their review of the literature (Erickson et al., 2019).
The committee found moderately strong evidence for an effect of long-term moderate to
vigorous physical activity on cognition in healthy adults aged 50+ with the largest effect sizes
observed in the cognitive domains of executive function, global cognition and attention. This
strong association between exercise training and executive functioning confirmed findings from
a 2003 meta-analysis looking at domain specific effects of aerobic exercise (Colcombe &
Kramer, 2003) (Figure 2).

5

Figure 2: The results of a meta-analysis examining the effects of aerobic exercise training on various cognitive
domains (Colcombe & Kramer, 2003).

In regard to exercise type, aerobic training had the greatest effect on executive
functioning and global cognition while multi-modal training had the largest effect on episodic
memory (Barha et al., 2017). Looking at evidence from prospective observational studies, the
committee determined that there was strong evidence to suggest higher levels of physical activity
are associated with reduced risk of cognitive decline (Sofi et al., 2011) and dementia (Beckett et
al., 2015). In response to this converging evidence, the field has seen an increase in research on
the underlying mechanisms through which physical activity exerts these effects.
Numerous mechanisms through which physical activity may exert its effect on cognition
have been proposed, ranging in scope from cellular and molecular processes to behavioral
mechanisms (Stillman et al., 2016) (Figure 3). To date, much of the research on causal
relationships between physical activity and cognition has been conducted in animals, focusing on
cellular and molecular mechanisms. Findings from animal models have provided strong support
for the roles of brain-derived neurotropic factor (BDNF) and insulin-like growth factor-1 (IGF-1)
as mediators of the relationship between physical activity and cognitive ability. For example,

6
experimental studies have indicated that blocking BDNF and IGF-1 signaling diminishes the
positive effects of exercise on cognition such as the long-term potentiation of neurons (Cotman
& Berchtold, 2007). At the cellular level, the development of new blood vessels (angiogenesis)
and the growth of new neurons (neurogenesis) have been proposed as potential mediators with
evidence for exercise induced hippocampal neurogenesis in both animals and humans (Erickson
et al., 2011; Van Praag, 2008). Given the strong evidence for these molecular and cellular
changes as a result of physical activity, it is likely that they also give rise to downstream changes
at the brain systems and behavioral level.

Figure 3: Conceptual model of mechanisms linking physical activity to cognitive functioning (Stillman et al.,
2016).

Two more macroscopic candidate mechanisms which may help explain the relationship
between physical activity and cognitive ability in older adults are cerebral blood flow (CBF)
measured via arterial spin labeling (ASL) and sleep efficiency. Multiple reviews have
emphasized the need for research into their roles as mediators of the relationship between
physical activity and cognitive ability (Kennedy et al., 2016; Stillman et al., 2016). Evidence for
these candidate mechanisms is discussed in detail below.

7
Sleep Efficiency as a Mediator
Normal aging is characterized by changes in both sleep quality and quantity with an
estimated 50% of older adults expressing complaints about difficulty in both the initiation and/or
maintenance of sleep (Crowley, 2011). Older adults demonstrate greater sleep-onset latency (i.e.,
time to fall asleep), increased time awake after sleep onset, and decreased time in the deeper
stages of NREM or slow wave sleep (SWS) (Crowley, 2011; Espiritu, 2008; Mander et al., 2017)
(Figure 4).

Figure 4: Typical sleep architecture across a 9-hour sleep window in a younger adult (top) and an older adult
(bottom). Older adults exhibit longer sleep latency (e.g., time to fall asleep), less time in slow wave sleep, more
frequent awakenings and transitions into lighter stages of sleep, and increased time awake after sleep onset.
(Mander et al., 2017).

This combination of factors results in decreased sleep efficiency (i.e., percentage of time
in bed spent asleep). At a more microscopic level, changes also occur in the electrical signaling
observed during the various sleep stages which can be measured with electroencephalography.

8
Two signature oscillations occurring during NREM sleep are slow wave activity (slow waves in
the 0.5-4.5 Hz range) and sleep spindles. Research has demonstrated a reduction in both slow
wave and sleep spindle amplitude and density with age, with the most significant reductions
occurring in the frontal lobe (De Gennaro & Ferrara, 2003, Mander et al., 2013, 2014; Martin et
al., 2013). These microscopic changes are noteworthy, as both slow wave activity and sleep
spindles have been associated with memory consolidation, preparation for hippocampal
encoding, and amyloid and tau deposition (Mander et al., 2013, 2014; Winer et al., 2019),
Additionally, low frequency delta EEG during the first phase of NREM sleep has been
associated with frontally mediated tasks such as executive functioning in older adults (Anderson
& Horne, 2003). In light of these findings, it is not surprising that age related changes in sleep
quality have been associated with changes in cognitive ability.
In healthy individuals, the relationship between both acute and chronic sleep deprivation
and worsened cognitive ability across multiple domains is well established (Goel et al., 2009).
Although total sleep time (TST) is commonly believed to be the most important aspect of sleep
for cognitive ability, in older adults it is more often sleep quality as opposed to quantity that is
associated with cognitive ability. Several studies have found evidence for an association between
sleep efficiency and cognitive ability in healthy older adults (Blackwell et al., 2006; Cavuoto et
al., 2016; Nebes et al., 2009; Wilckens et al., 2016). More continuous and condensed sleep (i.e.,
more efficient sleep) has also been associated with deep, SWS (Wilckens et al., 2016), the type
known to decrease with age and confer important advantages to cognitive performance.
Additional evidence suggests that increased sleep efficiency increases the likelihood of
progressing through the various sleep stages shown to confer cognitive benefits (e.g., slow wave
sleep) (Wilckens et al., 2012). Looking at TST, some cross-sectional and prospective cohort

9
studies of healthy older adults have found evidence that both unusually short and long self-
reported sleep duration increased odds for poor cognitive performance across multiple domains
(Devore et al., 2014; Lo et al., 2016) but other studies have reported no association between total
sleep time and cognitive ability (Blackwell et al., 2006; Nebes et al., 2009). Additional evidence
for the importance of sleep quality comes from research in those at risk for and suffering from
dementia.
While changes in sleep are a common feature of normal aging, there is significant
variation in the severity of sleep disruption experienced by older adults (Vitiello, 2009) with the
most marked changes in those with AD and at highest risk for AD (Mander et al., 2016). For
example, a prospective study of 737 healthy older adults found that individuals with the highest
levels of objectively measured sleep fragmentation were 1.5 times more likely to develop AD
compared to those with low sleep fragmentation (Lim et al., 2013). Amyloid positive individuals
(i.e., individuals in the preclinical phase of AD) have also been shown to exhibit lower sleep
efficiency measured via accelerometer than amyloid negative individuals. (Ju et al., 2013). While
the causal relationship between sleep quality and AD risk is not yet established, there is
increased evidence for a bidirectional relationship in which NREM SWS increases amyloid
accumulation which in turn impairs NREM SWS (Ju et al., 2014). Therefore, interventions to
improve sleep quality in older adults may result in reduced incidence of AD and dementia.
Evidence from epidemiological studies shows that more physically active individuals
report better sleep quality than those who are more sedentary (Youngstedt 2005). Indeed, both
acute and prolonged (e.g., a 12-month exercise intervention) exercise have been associated with
increased time in Stage 2 and SWS and less frequent awakenings after sleep onset (King et al.,
2008; Youngstedt et al., 1997) Findings from randomized controlled trials of physical activity

10
for insomnia in older adults also supports this association, with a Cochrane review finding
evidence to support the use of physical activity as a nonpharmacological method to improve
sleep (Montgomery & Ja, 2009). A subsequent review of the literature echoed this
recommendation, citing evidence for a relationship between moderate amounts of exercise and
improved sleep quality, but emphasizing that more research is needed before a conclusive
statement about the efficacy of physical activity interventions on sleep quality can be made
(Buman & King, 2010).
One limitation in the current literature on physical activity and sleep is the fact that many
studies use measures averaged across several days, which ignores day-to-day variation and limits
the ability to infer the temporal order of the relationship. Evidence from the few studies that have
examined day-to-day variability is equivocal with some studies suggesting it is sleep quality that
predicts physical activity (Baron et al., 2013; Lambiase et al., 2013; Tang & Sanborn, 2014), one
finding evidence to support physical activity levels predicting sleep quality in healthy older
adults (Best et al., 2019), and another finding no day-to-day associations (Mitchell et al., 2017).
Given this conflicting evidence, additional research into the directionality of the relationship is
warranted.
In response to the accumulating evidence for the associations between physical activity,
sleep, and cognitive ability in older adults, two groups have conducted cross-sectional studies
analyzing the relationships between these 3 variables. In their study of cognitively normal older
adults, Wilckens and colleagues found that sleep efficiency, but not total sleep time mediated the
relationship between objectively measured physical activity and multiple cognitive domains
(Wilckens et al., 2018). Conversely, the second study found evidence to support independent
effects of physical activity and sleep efficiency on cognition, suggesting that they may act

11
through separate mechanisms (Falck et al., 2018). These conflicting results may be partially due
to methodological differences. For example, Falck and colleagues collected 14 days of
accelerometer data while Wilckens et al. collected only 7 days of data, with some participants
recording as a little as 4 days. Furthermore, neither group looked at the influence of day-to-day
variability in physical activity on subsequent sleep quality, which limits their ability to make
causal inferences about the directionality of the relationship. Based on this inconsistent pattern of
results and uncertainty regarding the causal nature of the relationships, additional research into
the relationships between physical activity, sleep, and cognition is needed.

Cerebral Blood Flow as a Mediator
Cerebral blood flow measured via arterial spin labeling (ASL) indicates the rate of
delivery of arterial blood to brain tissue and is thought to be an important indicator of
neurovascular health (Zhang et al., 2017). Cerebral perfusion is defined as the delivery of
oxygenated blood to the capillary bed, and is a term often used interchangeably with CBF (Liu &
Brown, 2007). As with sleep, normal aging is characterized by reductions in CBF (Ogoh &
Ainslie, 2009). However, decreased CBF occurs in regions independent of grey-matter atrophy
indicating that is not solely reflective of age-related atrophy, and likely provides unique
information about neurovascular dysfunction and/or underlying pathology (Chen et al., 2011).
Indeed, associations between vascular dysfunction and amyloid pathology are well documented
(Govindpani et al., 2019), and recent work has also shown overlap between CBF and tau PET
binding, suggesting that the regional location of reduced CBF is related to higher tau deposition.
Stronger associations between tau and CBF were also associated with worse global cognition,
suggesting a synergistic effect of tau accumulation and CBF reduction on cognitive ability

12
(Albrecht et al., 2020). In addition to its utility as a proxy measure for neurovascular health,
studies of older adults have also associated CBF with current cognitive ability (Bangen et al.,
2018; Hshieh et al., 2017), future cognitive decline, and conversion from MCI to dementia (Chao
et al., 2010; De Vis et al., 2018). For example, a multisite ADNI study found reduced CBF in
patients with late stage MCI and AD compared to healthy older adults (Wang et al., 2013). Other
studies, however, have found negative associations between CBF and cognitive ability in
cognitively normal, amyloid positive older adults suggesting an interaction between CBF and
amyloid status on cognition (Bangen et al., 2017). This association between higher resting CBF
and worse memory performance may represent a compensatory response to underlying
pathology in which increased levels of CBF are required to maintain cognitive abilities during
the earliest stages of AD (Dai et al., 2009). Further evidence for this compensatory response
comes from research on CBF patterns in individuals with a genetic susceptibility to sporadic AD
(i.e., APOE4 carriers). In APOE4 carriers, there is a marked increase in CBF in middle age,
followed by a steep decrease after middle age (Zhang et al., 2017) (Figure 5).

Figure 5: Patterns of change in cerebral blood flow measured over time in healthy individuals. Expected
effects of vascular risk factors, APOE4 status, aerobic exercise, and cognitive training are shown (Zhang et
al., 2017).

13
This accumulation of evidence for dysregulated CBF in those at risk for dementia has
highlighted the potential utility of CBF as a biomarker and led some researchers to hypothesize
that reduced and/or dysregulated CBF may reflect the early stages of AD and dementia (Sweeney
et al., 2018) (Figure 6).

Figure 6: A hypothetical model of biomarker change across the AD spectrum, implicating changes in brain
vasculature (e.g., CBF) as the first detectable biomarkers (Sweeney et al., 2018).

Indeed, one of the leading hypotheses about the link between vascular health and AD
implicates reduced CBF early in the disease progression of AD. These authors describe a two-hit
vascular hypothesis of AD, in which vascular risk factors leading to impaired blood-brain barrier
dysfunction and reduced CBF constitute the first hit. This is followed by downstream
consequences such as accumulation of toxins, hypoperfusion, reduction of amyloid clearance,
and increased production of amyloid. Increased levels of amyloid in the brain are thought of as
the second hit, which then leads to increased neuronal dysfunction, neurodegeneration, and
dementia (Zlokovic, 2011). Based on this proposed chain of events, improving vascular factors
such as CBF prior to the occurrence of widespread neuronal damage could have tremendous

14
public health benefits. Evidence suggests that this might be possible through the use of physical
activity interventions.
Strong evidence for the relationship between physical activity and CBF comes from
Maass and colleagues. In their 3-month intervention study of cognitively normal older adults,
they found that individuals assigned to the aerobic training group exhibited changes in
hippocampal perfusion which were in turn correlated with improvements in recognition and
spatial memory. The changes in perfusion were also correlated with changes in hippocampal
volume, suggesting a potential for vascular plasticity in the hippocampus as a result of aerobic
exercise (Maass et al., 2015). Cross-sectional research also provides evidence for an association
between CBF and long-term aerobic exercise, showing that endurance trained adults exhibit
levels of cerebral blood flow as much as 17% higher than their sedentary counterparts (Ainslie et
al., 2008). Furthermore, a study that sought to determine a dose-specific relationship between
objectively measured physical activity and CBF in cognitively normal older adults found
evidence for an association between light, moderate to vigorous, and total activity with higher
CBF in lateral and medial frontal regions (Zlatar et al., 2019). This association between physical
activity and CBF is likely a downstream consequence of changes occurring at the molecular and
cellular level as a result of exercise.
Numerous cells, collectively referred to as the neurovascular unit (NVU), work together
to regulate CBF. Under optimal conditions, the NVU responds to neuronal activity through a
process called neurovascular coupling which results in an increase of CBF to activated brain
regions (Sweeney et al., 2018). An important part of this coupling process is the production of
vasoactive factors by the endothelium (Faraci & Heistad, 1998). One of these factors, Nitric
Oxide (NO), is regarded as particularly important due to its effect on cerebral vasodilation and

15
evidence showing that reductions in NO are associated with impaired neurovascular coupling
(Pluta, 2005). Therefore, one mechanism through which physical activity may influence CBF is
through its enhancement of endothelial function. This theory is supported by research on humans
and animal models. A 2015 meta-analysis of randomized controlled trials assessing the effect of
exercise on endothelial function found a positive association between aerobic exercise intensity
and endothelial function (Ashor et al., 2015). In animal models, exercise induced improvements
in endothelial-dependent vasodilation through increased NO production was found (Durrant et
al., 2009). Evidence such as this lends support for further investigations into the relationship
between physical activity and CBF in humans, particularly those at risk for AD.

Current Study
The current study investigated relationships between physical activity, sleep efficiency,
CBF, and cognitive ability in sedentary older adults with normal or subtle cognitive impairment.
Unlike many studies of sleep and physical activity that collect data for 4-14 days (Best et al.,
2019; Falck et al., 2018; Wilckens et al., 2016; Zhu et al., 2017) or rely on self-report measures
(Lo et al., 2016; Nebes et al., 2009; Reid et al., 2010), the study used objectively measured
physical activity and sleep efficiency spanning a period of 30 days. These data, thus, permitted
the study of the real-time effects of physical activity and sleep on CBF and cognitive ability
unbiased by self-report.
The current study contributes to our understanding of the relationship between lifestyle
factors (physical activity and sleep efficiency), brain health (CBF), and cognitive ability in
individuals at risk for dementia. Through the use of robust objective data, we added to the
limited literature on daily relationships between physical activity and sleep efficiency, and the

16
effect of intra-individual fluctuations in physical activity on cognitive outcomes (e.g., CBF and
cognitive performance). Findings from this study are important when considering future
interventions, and underscore the complexity of relationships between physical activity and brain
health in relatively sedentary individuals, something that is of practical importance given the low
percentage of older adults currently meeting the recommended physical activity guidelines.
These goals were pursued in the following three study aims.

Aim 1: Examine whether sleep efficiency and global CBF mediate the relationship between
physical activity and cognitive ability in older adults using mean measures of sleep efficiency
and physical activity collected over a 30-day period. Due to the fact that sleep efficiency may
influence physical activity, a model examining physical activity as a mediator of the relationship
between sleep efficiency and cognitive ability was also tested.
Hypothesis 1: Mean sleep efficiency will mediate the relationship between physical activity and
cognitive ability within the domains of global cognition, executive functioning and episodic
memory.
Hypothesis 2: Global gray matter CBF will mediate the relationship between physical activity
and cognitive ability within the domains of global cognition, executive functioning and episodic
memory.
Aim 2: Examine whether individual differences in effects of physical activity during the day on
sleep efficiency that night account for differences in cognitive ability at baseline.
Hypothesis: Subjects with a stronger positive effect of daily physical activity on the subsequent
night’s sleep efficiency will perform better on cognitive measures of global cognition, executive
functioning and episodic memory.

17
Aim 3: Examine whether individual differences in day-to-day physical activity are associated
with global gray matter CBF and cognitive ability.
Hypothesis: Greater day-to-day variability in physical activity will predicts higher CBF and
cognitive ability.

Methods
Study Population
This project used data from the Lifestyle Enriching Activities for Research in
Neuroscience Intervention Trial (LEARNit), a randomized controlled trial aimed at examining
the effects of exercise on brain health and cognitive functioning in older adults with mild
cognitive impairment. This study is a NIH/NIA funded clinical trial led by Dr. Judy Pa which
began in 2015 and is expected to close enrollment in late 2021. The sample currently consists of
52 older adults (67% female, mean age=66.50). See Table 1 for additional demographics of
current sample. Participants are adults aged 55-80 who are sedentary and expressed mild
concerns about memory and/or attention and/or perform at least 1 standard deviation below
normative levels on tests of attention, executive function, or memory. After an initial screening
visit, participants took part in a 30-day compliance test in which they wear an accelerometer 24
hr/day for 30 days. If they successfully completed the compliance screen, participants were
invited to return for a study visit in which they were administered a cognitive battery, lifestyle
questionnaires, brain MRI scan, PET/CT scan, blood draw and physical assessments. As the trial
is ongoing, all data for the current study was taken from participant’s 30-day compliance test and
cognitive testing and scanning visit immediately following the compliance test.

18
Sample Characteristics (N = 52)
% Female 67%
Age 66.5 (6.82)
Years Education 16.6 (2.4)
% ApoE+ 27%
MoCA 25.94 (2.68)
Sleep Efficiency (percentile) 73% (6)
Light Physical Activity (minutes/day) 216.43 (76.58)
Moderate Physical Activity (minutes/day) 19.24 (14.7)
Ethnicity
% Caucasian 79%
% African American 10%
% Hispanic 12% (10% Mexican, 2% Cuban)
% Asian American 6%
% Unknown 5%
Medical/Health History
Body Mass Index 27.34 (4.83)
Treated Sleep Apnea 4% (n = 3)
Psychiatric History 32% (n = 17)
Blood Pressure Medication 31% (n = 16)
Anti-Depressants 15% (n = 8)
Sleep Medication 2% (n = 1)
Table 1: Participant characteristics of the LEARNit cohort. Means and standard deviations or counts shown

19
Measures:
Physical Activity
Physical activity was measured using the GENEActiv tri-axis accelerometer which
participants wore for an average of 30.25 days prior to their study visit. Data from the
accelerometer was processed using GGIR, an R-package used to analyze data for physical
activity and sleep research (Migueles et al., 2019). The program performs automatic calibration,
detects abnormally high values or values indicating non-wear, and calculates gravitational
acceleration (i.e., bodily movement). The measure was subcategorized as follows:
Light physical activity Aim 1: Average number of minutes spent in physical activity of between
40-100 gravitational acceleration units across all days; Aims 2 and 3: Minutes spent in physical
activity of between 40-100 gravitational acceleration units each day.
Moderate physical activity Aim 1: Average number of minutes spent in physical activity of
between 100-400 gravitational acceleration units across all days; Aims 2 and 3: Minutes spent in
physical activity of between 100-400 gravitational acceleration units each day.
Total physical activity Aim 1: Summary score of time spent in physical activity between 100-400
gravitational acceleration units across all days (number of minutes multiplied by metabolic
equivalent coefficient (3 for light, 6 for moderate) (Scarmeas et al., 2009); Aims 2 and 3:
Summary score of time spent in physical activity between 100-400 gravitational acceleration
units each day.
Sleep Efficiency
Sleep efficiency is calculated as follows:
𝑠𝑙𝑒𝑒𝑝 𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 =
𝑡𝑖𝑚𝑒 𝑎𝑠𝑙𝑒𝑒𝑝
𝑡𝑖𝑚𝑒 𝑖𝑛 𝑏𝑒𝑑
𝑋 100

20
Sleep efficiency was also measured using the GENEActiv tri-axis accelerometer. Data
from the accelerometer was processed using GGIR, an R-package used to analyze data for
physical activity and sleep research (van Hees et al., 2018).

Cerebral Blood Flow
Cerebral blood flow (CBF) is a measure of the oxygenated blood volume passing through
a specific area of the brain per unit of time, typically quantified as follows:
𝑐𝑒𝑟𝑒𝑏𝑟𝑎𝑙 𝑏𝑙𝑜𝑜𝑑 𝑓𝑙𝑜𝑤 =
(
𝑚𝑖𝑙𝑙𝑖𝑙𝑖𝑡𝑒𝑟𝑠 𝑜𝑓𝑏𝑙𝑜𝑜𝑑
100 𝑔𝑟𝑎𝑚𝑠 𝑜𝑓 𝑡𝑖𝑠𝑠𝑢𝑒
)
(60 𝑠𝑒𝑐𝑜𝑛𝑑𝑠)

CBF was measured using a pseudo-continuous arterial spin labeling (pCASL) MRI scan.
Arterial spin labeling (ASL) is a non-invasive technique that uses magnetically labeled arterial
blood water as an endogenous tracer and provides a highly reliable quantitative measure of CBF.
With pCASL, many short pulses are used to label arterial blood water as it passes through the
labeling plane, towards the region of interest (ROI). The goal is to obtain a tag and control image
in which the static tissue signals are identical, but the magnetization of the inflowing blood is
different. Subtraction of the tag image from the control image provides a quantification of CBF
in the ROI (Liu & Brown, 2007). These general principles are depicted in Figure 7.

21

Figure 7: Depicts general procedures for CBF measurement using arterial spin labeling. The top row
illustrates the sequential order of the labeling and image acquisition. The bottom row illustrates the
quantification procedure of subtracting the label image from the control image (Yousaf et al., 2018).

Structural T1-weighted MPRAGE images were acquired on a Siemens 3T Prisma
scanner. The pCASL sequence consists of an initial M0 image and seven tag-control image
pairs. Scan parameters include a repetition time of 4300ms, an echo time of 36.7ms, a post-label
delay of 2000ms, and a label duration of 1500ms. FOV is 240x240x120mm and resolution is
2.5mm3 isotropic.
Arterial Spin Labeling processing scripts were provided Dr. Dan Albrecht and Dr. Kay
Jann, collaborators at the USC Stevens Neuroimaging and Informatics Institute. To measure
global gray matter CBF, T1-weighted MR images were segmented in SPM12 to produce gray
matter maps. For ROI analyses, FreeSurfer was used to generate subject specific ROIs. A priori
ROIs include the hippocampus, middle frontal gyrus, superior frontal gyrus, and the anterior
cingulate gyrus. These were selected based on previous work demonstrating early patterns of
hypoperfusion in these regions (Chao et al., 2010; De Vis et al., 2018; Montagne et al., 2016)

22
and studies showing their sensitivity to the effects of physical activity (Maass et al., 2015; Zlatar
et al., 2014)

Cognitive Ability
Cognitive ability was measured via a 2-hour neuropsychological battery consisting of
tests designed to measure different cognitive domains. Tests measuring executive functioning,
episodic memory, and global cognition were selected a priori based on prior evidence for their
association with physical activity (Barha et al., 2017; Colcombe & Kramer, 2003; Erickson et al.,
2019). The measures are described below:
Mental Status:
The Montreal Cognitive Assessment (MoCA): A test of multiple cognitive domains
commonly used to screen for mild cognitive impairment and dementia. The test consists
of 7 subdomains: visuospatial/executive; naming; memory; attention; language;
abstraction; and orientation. Participants receive a score out of 30, with 26 or below
indicating some degree of cognitive impairment (Hoops et al., 2009).
Executive Functioning:
Letter-Number Sequencing, Wechsler Adult Intelligence Scale -III (LNS): A measure of
attention and working memory in which individuals are read a series of numbers and
letters and asked to repeat them back to the examiner in sequential and alphabetical order
(Crowe, 2000). The measure results in one overall score with higher scores reflecting
better performance.
Flanker Task: A test of inhibitory control in which participants are asked to respond to an
arrow embedded in the center of 5 arrows pointing left or right. In half of the trials, the

23
arrows point in the same direction as the center arrow (congruent trials). In the other half
of the trials, the arrow points in the opposite direction of the center arrow (incongruent
trials). The difference in reaction time for the congruent and incongruent trials was used
as a measure of inhibitory control (Gothe et al., 2014). Lower scores indicate better
performance.
Episodic Memory:
California Verbal Learning Test, Second-Edition (CVLT): A word list-learning measure
of auditory verbal memory which assesses immediate recall for a word list in addition to
recall after a short and long delay period (Woods et al., 2006). The current used the long
delay free recall score. Higher scores indicate better performance.
Complex Figure Test (CF): A test designed to measure visuospatial constructional ability
and visual memory in which participants are presented with a picture of a complex figure,
and asked to copy it. They are later asked to draw the figure from memory following a
short and long delay period. The Rey-Osterrith and modified Taylor Complex Figure
were counterbalanced across participants (Hubley & Tremblay, 2002). The current study
used the long delay score. Higher scores indicate better performance.
Latent Global Cognition:
All manifest cognitive variables (MoCA, CVLT, Flanker, CF, and LNS) were fit with a
one-factor latent variable model to estimate global cognitive ability. The purpose of estimating a
latent global cognitive ability score was to characterize participants’ general cognitive ability
level. We describe the estimation procedure.
Data Analysis

24
We first present the descriptive results of the key variables in the analysis. As part of pre-
analysis, we estimated a one-factor latent general cognitive ability model. The model is
presented in Figure 9. The latent variable, LV, accounts for the common variance in all of the
cognitive ability variables. The path projecting from the LV to CF was fixed to 1 to scale the
latent variable. The Complex Figure test was selected because it was the strongest indicator of
the latent variable. By fixing this path, the variance of the latent variable could be estimated. The
primary purpose in estimating a latent cognitive ability variable was to include in subsequent
models a general cognitive ability measure that was unbiased by measurement error. All variance
unique to each cognitive ability variable, including measurement error is collected in the residual
variances (labeled e1-e4 in Figure 8).

Figure 8: Latent variable model. MoCA = Montreal cognitive assessment; LNS = Letter-number sequencing;
CVLT = California Verbal Learning Test long delay score; CF = Complex Figure Test long day; Flanker =
Inhibitory control measured via Flanker

MoCA CVLT CF LNS
LV
Flanker
e1 e5 e2 e3 e4
* * * 1 *
*
* * * * *

25
To address aim 1, mediation models were used to examine the indirect effect of physical
activity on cognition by way of sleep efficiency and cerebral blood flow. The models are
depicted in Figures 9 and 10.

Figure 9: Mediation model depicting the proposed mediating effect of sleep efficiency on the relationship
between physical activity and cognitive ability.

Figure 10: Mediation model depicting the proposed mediating effect of cerebral blood flow on the
relationship between physical activity and cognitive ability.

For aim 1, we tested whether mean sleep efficiency across 30 days and global gray matter
CBF mediated the effect of mean levels of total physical activity on measures of mental status,
executive functioning, episodic memory, and latent global cognition. In figures 9 and 10, “a”
indicates the effect of physical activity on the mediator (sleep efficiency or CBF), “b” indicates
the effect of the mediator on cognitive ability, and “c” indicates the effect of physical activity on
Sleep Efficiency
Physical Activity Cognitive Ability
a b
c / c’
Cerebral
Blood Flow
Physical Activity Cognitive Ability
a b
c / c’

26
cognitive ability. C’ is the effect of physical activity on cognitive ability, when adjusting for the
mediator. To test our hypotheses, we computed the indirect effects (a*b in Figures 9 and 10).
Significant indirect effects would provide support for our hypotheses that mean sleep efficiency
and global gray matter CBF partially or fully mediate the relationship between physical activity
and cognitive ability (Hayes, 2015). In addition to the primary analysis with global gray matter
CBF as the mediator, exploratory analyses investigated a priori ROIs (hippocampus, middle
frontal gyrus, superior frontal gyrus, and the anterior cingulate gyrus) to examine region-specific
mediation. We used the bootstrap method to calculate the standard error and associated
confidence interval of our indirect effect (Hayes, 2015). In order to test reverse causality, we also
tested whether physical activity mediates effects of sleep on cognitive ability. Additionally,
because it is possible that sleep efficiency and CBF interact with one another, we tested the
association between these two proposed mediators. Models were adjusted for the effects of age,
sex, and education.
To evaluate aim 2, we used Dynamic Structural Equation Modeling (DSEM) to infer the
temporal effects of physical activity and sleep efficiency on cognitive functioning. This approach
allowed us to examine both within-person relationships in physical activity and sleep efficiency
at the within-person level and individual differences in these parameters at the between-person
level. One of the assumptions of the model is that processes are stable, and therefore the focus is
not on mean change, but rather on variability around a stable mean. Through the model, we
investigated whether deviations around a person’s mean level of sleep efficiency were predicted
by physical activity engagement during the day and vice versa. The general model is depicted in
Figure 11 and can be expressed as the following equation:
𝑆𝐸
!"
= 𝛼
"
+𝛽
"
𝑃𝐴
!"
$
+𝑒
!"
(1)
𝛼
"
= 𝛾
%%
+𝑢
%"
(2)

27
𝛽
"
= 𝛾
&%
+𝑢
&"
(3)

Figure 11: Within-person model estimating the effect of physical activity at day t on sleep efficiency at night t.
Between-person model estimating the effect of individual’s initial score (PAà SE Intercept) and slope scores
(PAà SE Slope) on cognitive ability at baseline.

The within-person model (above the dashed line in Figure 11) was estimated for each
person, p, over as many days as provided (range: 26-32 days). Participant p’s mean
autoregressive effect was used to estimate between-person autoregressive parameters of the
effect of daily physical activity levels on sleep efficiency. As shown in equation 2, the random
intercept, 𝛼
"
, is the distribution of mean sleep efficiency scores across the entire sample and
consists of 𝛾
%%
, the grand mean of sleep efficiency scores and each person’s deviation score
around the mean, 𝑢
%"
. In equation 3, the random slope, 𝛽
"
, is the distribution of the effects of
physical activity on day t on sleep efficiency on night t across the entire sample and consists of
𝛾
&%
, the grand mean of all persons’ regression effects and each person’s deviation score around
the mean, 𝑢
&"
. Between-person random variables (below the dashed line in Figure 11) represent
Physical Activity (t) Sleep Efficiency (t)
Within-Person
Between-Person
α
i
β
i
Cognitive Ability
b2
b1

28
participants’ initial score (𝛼
"
) and slope scores ( 𝛽
"
) for the effect of physical activity levels at
day t on sleep efficiency at night t. Cognitive ability measures were then regressed on the
random intercept and slope variables. The deviance information criterion (DIC) was used to
compare models with and without the regression effect of physical activity on sleep efficiency,
b2 in Figure 11. Models with a lower DIC value are preferred (McNeish & Hamaker, 2020).
Because evidence for the temporal relationship between physical activity and sleep is equivocal,
we also conducted an exploratory analysis to assess the effect of sleep efficiency on physical
activity the following day.
To evaluate aim 3, we used DSEM to investigate the effect of interindividual differences
in day-to-day physical activity on CBF and cognitive ability. Like the bivariate models used in
Aim 2, this model is an autoregressive multilevel lag model in which individual lag-1
autoregressive parameters are estimated for each person at the within-person level and the
regression of cognitive ability and CBF on interindividual differences in the autoregressive
parameters are estimated at the between-person level. The expression for this model is:

𝑃𝐴
!"
= 𝛼
"
+𝜑
"
𝑃𝐴
(!( &)"
$
+𝑒
!"
(4)
𝛼
"
= 𝛾
%%
+𝑢
%"
(5)
𝜑
"
= 𝛾
&%
+𝑢
&"
(6)

At the within-person level, we first estimated effects of physical activity on day t (e.g.,
day 1) on physical activity at day t + 1 (e.g., day 2). Graphically, this is depicted at the within-
person level (above the dashed line) in Figure 12. As shown in equation 5, the random intercept,
𝛼
"
, is the distribution of mean sleep efficiency scores across the entire sample and consists of
𝛾
%%
, the grand mean of physical activity scores and each person’s deviation score around the

29
mean, 𝑢
%"
. In equation 6, the random slope, 𝜑
"
, is the distribution of the effects of physical
activity on day t-1 on physical activity on day t across the entire sample and consists of 𝛾
&%
, the
grand mean of all persons’ autoregressive effects and each person’s deviation score around the
mean, 𝑢
&"
.

Figure 12: Within person model estimating the effect of physical activity at day t on physical activity at day t
+ 1. Between-person model estimating the effect of an individual’s initial score (PAà PA Intercept) and slope
score (PAà PA Slope) on cerebral blood flow and cognitive ability.

Individual autoregressive effects were then used to estimate differences in the parameter
between-persons. We then tested whether these individual differences in the intercept (a1 and a2)
and slope (b1 and b2) were associated with levels of CBF and measures of cognitive ability
(below dashed line in Figure 12). The DIC was used to compare models with and without the
autoregressive effect of physical activity, b1 and b2 in Figure 12. Statistical significance and
meaningfulness of effects were evaluated using the 95% credible interval and model fit
comparisons.

Physical Activity (t)
Physical Activity
(t+1)
Within-Person
Between-Person
α
i
φ
i
Cerebral Blood Flow
Cognitive Ability
a1
b1
b2
a2

30
Results
Aim 1 Results
Mediation models (Tables 2 & 3) did not reveal any significant indirect (a*b pathways in
Figures 9 and 10) effects, indicating no statistically significant evidence for the mediating role of
sleep efficiency or CBF on the relationship between physical activity and cognitive ability.
Additionally, no significant associations between physical activity and sleep efficiency or CBF
(paths denoted as a in Figures 9 and 10) were identified. There were, however, significant
associations between the proposed mediators and cognitive ability (paths denoted as b in Figures
9 and 10). Sleep efficiency was associated with better performance on the MoCA (b=0.12,
SE=0.05, p=0.024), Complex Figure (b=0.20, SE=0.09, p=0.025), and Flanker (b=-4.04,
SE=1.64, p=0.013). CBF was associated with better performance on the MoCA (b=0.11,
SE=0.04, p=0.009) and Letter-Number Sequencing (b=0.10, SE=0.05, p=0.035). Results from
exploratory models investigating the mediating role of CBF in a priori ROIs can be found in the
appendix (Tables 6 – 9). Although no significant indirect effects were identified, CBF in several
ROIs were significantly associated with cognitive outcome measures. The directionality of these
relationships differed, with increased CBF associated with both better and worse performance.
None of the associations reported above survived correction for multiple comparisons.
Models revealed significant direct effects (c and c’ pathways in Figures 9 and 10) of
physical activity on cognitive ability. However, these effects were in the opposite direction of
our hypotheses. Higher mean levels of physical activity were associated with lower inhibitory
control measured via the Flanker (b=1.00, SE =0.38, p=0.009), and worse performance on
Letter-Number Sequencing when adjusted for the mediator (b=-0.03, SE=0.01, p=0.038). These
associations did not survive correction for multiple comparisons.

31
Post-hoc analyses investigated whether there were differential effects when looking at
only light or only moderate physical activity, as opposed to the total physical activity measure
(Appendix tables 2 - 5). Results did not differ, aside from the direct negative effect of physical
activity on cognitive ability disappearing when looking at moderate physical activity alone.
Mediation models to assess for reverse causality (Appendix table 1) did not reveal any
indirect effects, indicating no statistically significant evidence for the mediating role of physical
activity on the relationship between sleep efficiency and cognitive ability.
Exploratory analyses investigating the relationship between our proposed mediators
revealed a significant association between sleep efficiency and global gray matter CBF, in which
better sleep efficiency was associated with increased CBF throughout the brain (b=0.50,
SE=0.22, p=0.026). The association remained when adjusting for age, sex and education.
Table 2: Sleep Efficiency as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.02 0.03 [-0.04,
0.07]
0.545 0.02 0.03 [-0.05,
0.08]
0.642
b. SE à Latent Cog 0.18 0.11 [-0.05,
0.40]
0.122 0.00 0.04 [-0.07,
0.07]
0.899
c’. PA à Latent Cog 0.00 0.03 [-0.05,
0.05]
0.993 0.00 0.01 [-0.01,
0.02]
0.616
a*b. PA à SE à Latent
Cog
0.00 0.01 [-0.01,
0.01]
0.599 0.00 0.00 [0.00,
0.00]
0.900
c. PA à Latent Cog 0.00 0.03 [-0.05,
0.06]
0.906 0.00 0.01 [-0.01,
0.02]
0.639

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p

32
a. PA à SE 0.02 0.02
[-0.03,
0.07]
0.430 0.02 0.03
[-0.04,
0.07]
0.576
b. SE à MoCA 0.11 0.05
[0.003,
0.21]
0.044* 0.12 0.05
[0.02,
0.22]
0.024*
c’. PA à MoCA -0.002 0.01
[-0.03,
0.02]
0.860 0.00 0.01
[-0.02,
0.02]
0.985
a*b. PA à SE à MoCA 0.002 0.003
[-0.003,
0.01]
0.469 0.002 0.003
[-0.01,
0.01]
0.596
c. PA à MoCA 0.00 0.01
[-0.02,
0.02]
0.992 0.002 0.01
[-0.02,
0.03]
0.877

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.02 0.02 [-0.03,
0.07]
0.419 0.02 0.03 [-0.03,
0.07]
0.528
b. SE à CVLT 0.08 0.09 [-0.09,
0.24]
0.359 0.08 0.08 [-0.08,
0.24]
0.319
c’. PA à CVLT 0.004 0.01 [-0.02,
0.03]
0.744 0.003 0.01 [-0.02,
0.03]
0.833
a*b. PA à SE à CVLT 0.001 0.003 [-0.004,
0.01]
0.569 0.001 0.003 [-0.004,
0.01]
0.609
c. PA à CVLT 0.01 0.01 [-0.02,
0.03]
0.668 0.004 0.01 [-0.02,
0.03]
0.763

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.02 0.02 [-0.03,
0.06]
0.498 0.01 0.03 [-0.04,
0.06]
0.683
b. SE à CF 0.18 0.10 [-0.002,
0.67]
0.053 0.20 0.09 [0.03,
0.37]
0.025*
c’. PA à CF -0.004 0.02 [-0.04,
0.03]
0.821 -
0.001
0.02 [-0.03,
0.03]
0.934
a*b. PA à SE à CF 0.003 0.01 [-0.01,
0.01]
0.548 0.002 0.01 [-0.03,
0.04]
0.694

33
c. PA à CF -0.001 0.02 [-0.04,
0.04]
0.964 0.001 0.02 [-0.03,
0.04]
0.965

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.03 0.02 [-0.02,
0.07]
0.269 0.02 0.03 [-0.03,
0.07]
0.489
b. SE à Flanker -3.98 1.62 [-7.16, -
0.81]
0.014* -4.04 1.64 [-7.25, -
0.84]
0.013*
c’. PA à Flanker 1.01 0.35 [0.33,
1.69]
0.003* 1.00 0.38 [0.25,
1.74]
0.009*
a*b. PA à SE à
Flanker
-0.10 0.11 [-0.32,
0.11]
0.347 -0.08 0.12 [-0.30,
0.15]
0.517
c. PA à Flanker 0.91 0.34 [0.24,
1.58]
0.008* 0.92 0.37 [0.19,
1.65]
0.014*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.01 0.03 [-0.04,
0.06]
0.572 0.01 0.03 [-0.05,
0.07]
0.670
b. SE à LNS 0.07 0.07 [-0.06,
0.20]
0.270 0.07 0.06 [-0.05,
0.20]
0.231
c’. PA à LNS -0.03 0.02 [-0.06,
0.002]
0.063 -0.03 0.01 [-0.06, -
0.002]
0.038*
a*b. PA à SE à LNS 0.001 0.002 [-0.003,
0.01]
0.621 0.001 0.003 [-0.004,
0.01]
0.700
c. PA à LNS -0.03 0.02 [-0.06,
0.003]
0.079 -0.03 0.01 [-0.06,
0.00]
0.052

PA = All light and moderate physical activity; SE = Sleep efficiency; MoCA = Montreal cognitive assessment;
CVLT = California Verbal Learning Test long delay score; CF = Complex Figure Test long day; Flanker =
Inhibitory control measured via Flanker; LNS = Letter-number sequencing; Bold rows indicate a significant
relationship (p<.05); Covariates = Age, Sex, & Education.

34
Table 3: Global Gray Matter Cerebral Blood Flow as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.07 0.06 [-0.04,
0.18]
0.195 0.04 0.05 [-0.07,
0.15]
0.452
b. CBFà Latent Cog 0.07 0.07 [-0.07,
0.20]
0.333 -0.01 0.01 [-0.04,
0.02]
0.567
c’. PA à Latent Cog 0.01 0.03 [-0.05,
0.07]
0.794 0.01 0.01 [-0.01,
0.02]
0.526
a*b. PA à CBFà
Latent Cog
0.01 0.01 [-0.01,
0.02]
0.428 0.00 0.00 [0.00,
0.00]
0.660
c. PA à Latent Cog 0.01 0.03 [-0.05,
0.07]
0.669 0.01 0.01 [-0.01,
0.02]
0.538

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.09 0.05 [-0.01,
0.19]
0.062 0.07 0.05 [-0.02,
0.17]
0.122
b. CBFà MoCA 0.08 0.03 [0.01,
0.14]
0.018* 0.11 0.04 [0.03,
0.18]
0.009*
c’. PA à MoCA -0.01 0.01 [-0.03,
0.02]
0.532 -0.01 0.01 [-0.03,
0.02]
0.597
a*b. PA à CBF à
MoCA
0.01 0.01 [-0.002,
0.02]
0.121 0.01 0.01 [-0.003,
0.02]
0.161
c. PA à MoCA 0.00 0.01 [-0.03,
0.02]
0.969 0.001 0.01 [-0.02,
0.03]
0.914

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.09 0.05 [-0.01,
0.19]
0.062 0.07 0.05 [-0.02,
0.17]
0.122
b. CBFà CVLT 0.05 0.05 [-0.05,
0.16]
0.328 0.04 0.06 [-0.07,
0.16]
0.442

35
c’. PA à CVLT 0.003 0.01 [-0.02,
0.03]
0.825 0.003 0.013 [-0.02,
0.03]
0. 834
a*b. PA à CBF à
CVLT
0.01 0.01 [-0.01,
0.02]
0.394 0.003 0.004 [-0.01,
0.01]
0.460
c. PA à CVLT 0.01 0.01 [-0.02,
0.04]
0.585 0.01 0.01 [-0.02,
0.03]
0.672

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.07 0.05 [-0.02,
0.16]
0.140 0.05 0.04 [-0.03,
0.14]
0.231
b. CBFà CF 0.07 0.08 [-0.08,
0.22]
0.365 0.09 0.08 [-0.07,
0.25]
0.282
c’. PA à CF -0.01 0.02 [-0.04,
0.03]
0.786 -
0.004
0.02 [-0.03,
0.03]
0.792
a*b. PA à CBF à CF 0.01 0.01 [-0.04,
0.02]
0.452 0.01 0.01 [-0.01,
0.02]
0.404
c. PA à CF 0.00 0.02 [-0.04,
0.04]
0.993 0.001 0.02 [-0.04,
0.04]
0.976

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.10 0.05 [-0.002,
0.20]
0.055 0.08 0.05 [-0.02,
0.17]
0.138
b. CBFà Flanker 1.22 1.24 [-1.21,
3.64]
0.325 1.32 1.38 [-1.39,
4.03]
0.339
c’. PA à Flanker 0.88 0.36 [0.18,
1.58]
0.013* 0.89 0.37 [0.16,
1.62]
0.016*
a*b. PA à CBF à
Flanker
0.12 0.14 [-0.16,
0.39]
0.397 0.10 0.13 [-0.16,
0.36]
0.458
c. PA à Flanker 1.00 0.337 [0.34,
1.66]
0.003* 0.99 0.37 [0.26,
1.72]
0.008*

Unadjusted for covariates Adjusted for covariates

36
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.09 0.06 [-0.02,
0.20]
0.118 0.06 0.05 [-0.04,
0.15]
0.279
b. CBFà LNS 0.09 0.05 [-0.01,
0.19]
0.064 0.10 0.05 [0.01,
0.20]
0.035*
c’. PA à LNS -0.04 0.02 [-0.07, -
0.01]
0.020* -0.03 0.01 [-0.06,
0.01]
0.010*
a*b. All PA à CBF à
LNS
0.01 0.01 [-0.01,
0.02]
0.231 0.01 0.01 [-0.01,
0.02]
0.326
c. All PA à LNS -0.03 0.02 [-0.06,
0.01]
0.101 -0.03 0.02 [-0.06,
0.002]
0.072

PA = All light and moderate physical activity; CBF = Global gray matter cerebral blood flow; MoCA =
Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex
Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number
sequencing;Bold rows indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Aim 2 Results
Model results suggest that physical activity levels during the day did not have a
statistically significant effect on sleep efficiency the same night. The average beta effect across
all people was 0.02 with a between-person variance of .001. Using these values, we constructed
the 95% confidence band around the mean effect, which was: -0.04 and 0.06. In other words,
about half of the distribution was slightly above zero whereas the other half was slightly below
zero. The DIC value for the model with the autoregressive effect of physical activity on sleep
was 4,181.29 whereas the DIC value for the model without the effect was 4,178.99. This small
difference in DIC values suggested no strong preference for either model. As a result of the null
correlation between physical activity and sleep efficiency, further associations between the effect
of physical activity on sleep and cognition were not investigated.
We tested for endogeneity by testing the effect of sleep efficiency at day t on physical
activity at day t+1 (i.e., the next day). The random effect of sleep efficiency on physical activity

37
was nonsignificant, as the average effect across individuals was -0.01 with a between person
variance of 0.16 [.95 CI: -0.79 – 0.77]. The DIC value for the model with the regression effect of
sleep on physical activity was 4136.107 whereas the DIC value for the model without the effect
was 4137.338. Due to this negligible difference, the model comparisons were inconclusive.

Aim 3 Results
Within individuals, autoregressive multilevel lag models revealed a significant effect of
physical activity at time t (e.g., day 1) on physical activity at time t+1 (e.g., day 2) such that a
one-unit increase in physical activity at time t is expected to decrease physical activity at time
t+1 by .46 units [.95 CI: -0.70 – -0.22] E.g., When one exercises on Saturday, they may be tired
or sore, and less likely to exercise on Sunday. Like the effect of physical activity on sleep
efficiency, the autoregressive effect is assumed to be constant over the entire study window, that
is, it captures the correlation between any two randomly selected sequential days. The negative
autoregressive coefficient describes the nonlinear cyclical change that occurs in physical activity
over the study window. Therefore, this result does not indicate that individuals are steadily
decreasing. Rather, the negative autoregressive effect means that if someone engages in physical
activity levels above their norm on one day, the probability of their activity dipping below their
norm on the next day increases and vice-versa. Greater negative autoregressive parameters also
imply that people deviate further from their equilibrium on days they are more active or less
active. Between individuals, the variance suggested that 95% of participants’ autoregressive
effects fell between -0.70 and -0.22. Given the wide 95% confidence interval, there is variability
in the extent to which participants’ activity one day predicted activity the next day. Figure 13
depicts three average physical activity trajectories: one for a participant who engages in small

38
day-to-day changes in physical activity (the dashed red line); one for a participant who engages
in mean (moderate) day-to-day changes in physical activity (the solid black line); and one for a
participant who engages large day-to-day changes in physical activity (the dotted blue line). The
participant with the larger negative autoregressive effect would have greater and steeper
increases in physical activity from one day to the next (blue line) while persons with smaller
negative autoregressive effect would have smaller and more gradual change in physical activity
from one day to the next (red line). The latter person – the participant with the least negative
autoregressive effect – is predicted to remain closer to his or her mean level of physical activity
from day to day. Figure 14 depicts three actual physical activity trajectories of participants in the
study.

Figure 13: Hypothetical physical activity patterns for individuals with more negative autoregressive effects
(blue line), less negative autoregressive effects (red line), and autoregressive effects falling in the middle
(black line)

39

Figure 14: Actual physical activity patterns for participants with more negative autoregressive effects (blue
line), less negative autoregressive effects (red line), and autoregressive effects falling in the middle (black line)

Although point estimates of the effect of the autoregressive effect on CBF and cognitive
ability variables included credible intervals that encompassed zero (equivalent to a statistically
nonsignificant effect), DIC values favored including the autoregressive effect of physical activity
when looking at performance on Complex Figure and levels of global gray matter CBF (Table
4). That is, between-person differences in the autoregressive parameter appear to have utility for
predicting CBF and visual memory despite the 95% credible intervals encompassing zero. DIC
differences of less than 5 units were considered inconclusive (Burnham & Anderson, 2004).
Table 4: Aim 3 Model Comparisons

With AR effect Without AR effect
Superior model with
covariates
Dependent DIC R
2
DIC R
2
DIC
difference
DIC R
2

40

MoCA = Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF =
Complex Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number
sequencing; CBF = Global gray matter cerebral blood flow; DIC = Deviance information criterion (models
with values at least 5 units smaller are preferable); Covariates = Age, Sex, & Education

The phi coefficients in Table 5 quantify the effect of the autoregressive parameter on
CBF and Complex Figure. The 95% credible intervals were wide and encompassed zero, but the
autoregressive effects account for large unique portions of variability in cognitive outcomes (8 %
for Complex Figure and 15% for CBF). More negative autoregressive effects were associated
with better Complex Figure Test recall scores, and higher levels of global gray matter CBF.
(Figure 15).

Table 5: Aim 3 Parameter Estimates

No Covariates Covariates
Effect Posterior
Mean
.95 Credible
Interval
Posterior
Mean
.95 Credible
Interval
Intercept (alpha) -0.12 [-1.03, 0.92] 0.05 [-1.04, 0.73]
Intercept (phi) -0.46 [-0.52, -0.41] -0.46 [-0.52, -0.40]
Var (alpha) 0.09 [0.001, 1.44] 0.18 [0.004, 1.60]
Latent Cog 17792.49 0.27 17795.93 0.16 -3.44 17789.39 0.30
MoCA 16840.91 0.31 16808.17 0.32 32.74 16820.13 0.30
CVLT 16789.21 0.26 16777.80 0.22 11.40 16803.81 0.23
CF 16819.40 0.33 16826.85 0.25 -7.46 16835.24 0.33
Flanker 16891.60 0.30 16894.00 0.22 -2.40 16897.27 0.32
LNS 16120.10 0.30 16123.38 0.27 -3.28 16073.02 0.51
CBF 16569.50 0.37 16629.64 0.22 -60.14 16633.73 0.37

41
Var (phi) 0.01 [0.003, 0.03] 0.01 [0.003, 0.03]
Res. Var. (PA) 730.50 [677.76, 789.64] 730.45 [678.18, 788.57]
Regression Effects
PA --> CF -1.21 [-22.52, 17.43] -0.10 [-9.35, 10.65]
phi --> CF -3.31 [-15.15, 6.30] -3.54 [-12.91, 8.44]
R-squared

CF 0.33 [0.01, 0.99] 0.33 [0.045, 0.92]

No Covariates Covariates
Effect Posterior
Mean
.95 Credible
Interval
Posterior
Mean
.95 Credible
Interval
Intercept (alpha) -0.18 [-1.33, 0.60] 0.06 [-0.99, 1.02]
Intercept (phi) -0.46 [-0.52, -0.40] -0.47 [-0.52, -0.40]
Var (alpha) 0.08 [0.001, 1.47] 0.27 [0.01, 2.04]
Var (phi) 0.01 [0.004, 0.03] 0.01 [0.01, 0.03]
Res. Var. (PA) 729.34 [677.03, 785.68] 731.71 [680.85, 788.03]
Regression Effects
PA --> CBF 2.64 [-43.18, 47.20] 0.76 [-15.21, 18.13]
phi --> CBF -10.14 [-32.72, 8.11] -5.96 [-26.79, 12.77]
R-squared

CBF 0.37 [0.02, 0.99] 0.37 [0.08, 0.89]

PA = All light and moderate physical activity; CF = Complex Figure Test long day; CBF = Global gray
matter cerebral blood flow; Alpha = Mean level; Phi = autoregressive effect; Covariates = Age, Sex, &
Education

42

Figure 15: Scatter plots of associations between autoregressive effects of daily physical activity with Complex
Figure performance and global gray matter CBF. All light and moderate physical activity.

Post-hoc analyses investigated whether there were differential effects when looking at
only light or only moderate physical activity, as opposed to the total physical activity measure
(Appendix tables 10 and 11). Model comparisons did not differ when looking at the

43
autoregressive effect of only light physical activity. When looking at the effect of moderate
activity, DIC values favored including the autoregressive effect of physical activity for outcomes
of latent cognition and CVLT, in addition to Complex Figure and global gray matter CBF.

Discussion
Despite a plethora of prior evidence for protective effects of physical activity, few older
adults engage in the recommended amount of physical activity thought to benefit cognition and
protect against cognitive decline. Therefore, it is important to understand whether physical
activity, even at low levels, is associated with brain health. The current study demonstrates the
complexity around understanding whether individual differences in physical activity in an
otherwise sedentary population offers benefits to people’s cognitive ability. There are several
important takeaways from the current study. First, differences in day-to-day physical activity
may predict individual differences in visuospatial memory and cerebral blood flow. To our
knowledge this is the first study to show that day-to-day variability in physical activity may
influence brain health and cognition. Second, results suggest that day-to-day physical activity
and sleep efficiency in sedentary, free-living older adults are not significantly associated with
one another. Finally, our proposed mediators, sleep efficiency and CBF, were significantly
associated not only with cognitive ability outcomes, but also with each other. Implications for
these findings are discussed below.
Day-to-day patterns of physical activity appears to influence both visual memory
performance and CBF throughout the brain. The directionality of the relationship suggests that,
even in individuals with low levels of activity overall, alternating between “high exertion” and
“rest days” may prove beneficial. Given the relatively sedentary nature of our sample, increased

44
day-to-day variability may be reflective of individuals who are engaging in more intentional
activity on certain days (e.g., mowing the lawn, walking with a friend, intensive cleaning), as
opposed to lower activity across all days (e.g., light housework). This could help to explain the
association between day-to-day variability and global gray matter CBF. The association between
day-to-day variability and visual memory is more difficult to interpret, particularly given he null
effect of day-to-day variability in physical activity on other measures of cognitive ability. Future
experimental research in which exercise variability is deliberately manipulated between groups is
warranted in order to confirm or refute these findings.
The lack of a significant association between physical activity and sleep efficiency adds
to the limited literature on temporal associations between these processes. Models with the
effects of physical activity on sleep efficiency and sleep efficiency on physical activity were
inconclusive, as DIC values were nearly equivalent. Several prior studies of daily associations
between objectively measured physical activity and sleep efficiency in older adults have also
failed to find a significant relationship (Best et al., 2019; Kishida & Elavsky, 2016; Mitchell et
al., 2017). One prior study did, however, find evidence for a temporal association between sleep
efficiency and physical activity the following day (Maya J Lambiase, Kelley Pettee Gabriel,
Lewis H Kuller, 2013). There are several possible explanations for these results. First,
quantification of sleep and physical activity often differs across studies. For example, although
the work by Lambiase et al. (2013) and Mitchell et al. (2017) used the same accelerometer,
different cut points were used to define moderate to vigorous physical activity. Furthermore,
most studies have focused on the total time spent in physical activity, but neglected to consider
other characteristics such as type and/or timing. Associations between sleep and physical activity
also vary based on these factors. For example, longer durations of intense activity (e.g., at least

45
30 minutes of moderate to vigorous aerobic activity) performed in the first half of the day may
be necessary to observe an association with sleep efficiency. Additionally, the setting in which
the activity is performed (e.g., outside in sunlight) may play an important role. Finally, proximal
relationships between physical activity and sleep efficiency may simply be absent when looking
at older adults free-living patterns. In other words, some sort of intervention (e.g., aerobic
exercise program or sleep hygiene education) may be necessary to elicit associations between
day-to-day activity and sleep patterns.
Despite the lack of support for our proposed mediation pathways, both of our proposed
mediators were significantly associated with cognitive outcomes. Specifically, greater sleep
efficiency was related to higher MoCA scores, better inhibitory control, and better visual
memory. Higher levels of global gray matter CBF were related to higher MoCA scores and better
working memory. This lends support to the idea that, if physical activity were able to engage
these mediators, the proposed pathways may exist. This is the very purpose of LEARNit, as the
goal is to determine whether physical activity intervention improves brain health and cognitive
functioning. Thus, the current null findings may be consistent with the preintervention goal of
identifying a sedentary sample of older adults in which the beneficial effects of physical activity
are not yet present. To be enrolled in the current stud, participants must have reported engaging
in less than 60 minutes of structured aerobic activity per week. Therefore, the absence of
significant associations between physical activity and cognition (directly or via mediators) may
be a result of participants’ relatively sedentary activity patterns. Indeed, despite several studies
reporting associations between light levels of physical activity and cognitive performance in
older adults (Johnson et al., 2016; Wilbur et al., 2012), other work suggests that physical activity
levels in the moderate to vigorous range may be necessary (Fanning et al., 2017; Kerr et al.,

46
2013). In light of these findings, further investigation into whether our proposed mediation
pathways are present in more physically active individuals, and at what specific dose (e.g.,
intensity and length) is warranted.
Finally, exploratory analyses revealed an unexpected significant association between
sleep efficiency and global gray matter CBF, suggesting that better sleep efficiency is associated
with higher CBF throughout the brain. To date, few studies have investigated associations
between sleep quality and CBF in older adults, and none have assessed sleep efficiency
specifically (André et al., 2021). Due to the close temporal association between collection of
sleep efficiency and CBF data in the current study, the directionality of the relationship cannot be
inferred. However, based on literature suggesting the importance of the coupling of brain waves
and cerebral blood flow for amyloid clearance, the association between the two processes in the
context of aging warrants further study (Fultz et al., 2019).
A note is in order regarding the use of Bayesian approaches compared to more traditional
null hypothesis significance tests (i.e., comparing p-values to an alpha cut-off value). Bayesian
estimators generate distributions of parameter estimates, which make it impossible to generate a
single test statistic for each parameter. Statistical decision making, thus, must rely on the .95
credible interval (i.e., intervals that encompass zero can be interpreted as “not statistically
significant”), the Deviance Information Criteria, and scientific judgement about whether the
parameter adds to our understanding. While the modeling approaches used did not lead to
conclusions based on traditional p-values, the Bayesian approach employed in the current study
suggests that the data supported several of our hypotheses more than the null hypothesis (i.e.,
these models had lower DIC values). Although the study findings currently provide partial

47
support for some of our hypotheses (namely, those in Aim 3), these models will be updated as
new LEARNit data are collected, cleaned, and archived for these analyses.
Study Limitations
There are several limitations to the proposed study. First, the sample size was small, as
data collection is currently underway. Despite the study consisting of a small number of
participants, there was some support for the third aim that greater physical activity followed by
good rest predicts better visual memory and increased global gray matter CBF.
Second, the sampling strategy was one of convenience rather than a random sample. We,
therefore, cannot rule out selection bias. Participants were older adults who were willing to
participate in an exercise intervention program, implying some level of motivation to engage in
health-promoting behaviors. Furthermore, the majority of the sample consisted of white, highly
educated individuals, limiting our ability to generalize findings.
Third, participants in the current study did not fill out sleep diaries in addition to wearing
the accelerometer. The lack of a sleep diary prevents cross-validation of accelerometer derived
sleep measures, weakening the validity of the sleep measures obtained. However, past studies
have validated sleep data obtained via accelerometer against measures of self-report sleep and
polysomnography, providing evidence to support the use of accelerometer derived sleep data in
the absence of sleep diaries (van Hees et al., 2015). Physical activity measures did not include
the type or timing of activity, limiting our ability to investigate the potential influence of these
factors.

48
References
Ainslie, P. N., Cotter, J. D., George, K. P., Lucas, S., Murrell, C., Shave, R., Thomas, K. N.,
Williams, M. J. A., & Atkinson, G. (2008). Elevation in cerebral blood flow velocity with
aerobic fitness throughout healthy human ageing. Journal of Physiology, 586(16), 4005–
4010. https://doi.org/10.1113/jphysiol.2008.158279
Albrecht, D., Lisette Isenberg, A., Stradford, J., Monreal, T., Sagare, A., Pachicano, M.,
Sweeney, M., Toga, A., Zlokovic, B., Chui, H., Joe, E., Schneider, L., Conti, P., Jann, K., &
Pa, J. (2020). Associations between vascular function and Tau PET are associated with
global cognition and amyloid. Journal of Neuroscience, 40(44), 8573–8586.
https://doi.org/10.1523/JNEUROSCI.1230-20.2020
Anderson, C., & Horne, J. A. (2003). Prefrontal cortex: Links between low frequency delta EEG
in sleep and neuropsychological performance in healthy, older people. Psychophysiology,
40(3), 349–357. https://doi.org/10.1111/1469-8986.00038
André, C., Laniepce, A., Chételat, G., & Rauchs, G. (2021). Brain changes associated with sleep
disruption in cognitively unimpaired older adults: A short review of neuroimaging studies.
In Ageing Research Reviews (Vol. 66, p. 101252). Elsevier Ireland Ltd.
https://doi.org/10.1016/j.arr.2020.101252
Ashor, A. W., Lara, J., Siervo, M., Celis-Morales, C., Oggioni, C., Jakovljevic, D. G., &
Mathers, J. C. (2015). Exercise Modalities and Endothelial Function: A Systematic Review
and Dose–Response Meta-Analysis of Randomized Controlled Trials. Sports Medicine,
45(2), 279–296. https://doi.org/10.1007/s40279-014-0272-9
Bangen, K. J., Clark, A. L., Edmonds, E. C., Evangelista, N. D., Werhane, M. L., Thomas, K. R.,
Locano, L. E., Tran, M., Zlatar, Z. Z., Nation, D. A., Bondi, M. W., & Delano-Wood, L.

49
(2017). Cerebral blood flow and amyloid-β interact to affect memory performance in
cognitively normal older adults. Frontiers in Aging Neuroscience, 9(JUN), 1–14.
https://doi.org/10.3389/fnagi.2017.00181
Bangen, K. J., Werhane, M. L., Weigand, A. J., Edmonds, E. C., Delano-Wood, L., Thomas, K.
R., Nation, D. A., Evangelista, N. D., Clark, A. L., Liu, T. T., & Bondi, M. W. (2018).
Reduced regional cerebral blood flow relates to poorer cognition in older adults with type 2
diabetes. Frontiers in Aging Neuroscience, 10(SEP), 1–12.
https://doi.org/10.3389/fnagi.2018.00270
Barha, C. K., Davis, J. C., Falck, R. S., Nagamatsu, L. S., & Liu-Ambrose, T. (2017). Sex
differences in exercise efficacy to improve cognition: A systematic review and meta-
analysis of randomized controlled trials in older humans. Frontiers in Neuroendocrinology,
46(March), 71–85. https://doi.org/10.1016/j.yfrne.2017.04.002
Barnes, D. E., & Yaffe, K. (2011). The projected effect of risk factor reduction on Alzheimer’s
disease prevalence. The Lancet Neurology, 10(9), 819–828. https://doi.org/10.1016/S1474-
4422(11)70072-2
Baron, Jk., Reid, K., & Zee, P. (2013). Exercise to Improve Sleep in Insomnia: Exploration of
the Bidirectional Effects. Journal of Clinical Sleep Medicine, 9(8), 2239–2248.
Baumgart, M., Snyder, H. M., Carrillo, M. C., Fazio, S., Kim, H., & Johns, H. (2015). Summary
of the evidence on modifiable risk factors for cognitive decline and dementia: A population-
based perspective. Alzheimer’s and Dementia, 11(6), 718–726.
https://doi.org/10.1016/j.jalz.2015.05.016
Beckett, M. W., Ardern, C. I., & Rotondi, M. A. (2015). A meta-analysis of prospective studies
on the role of physical activity and the prevention of Alzheimer’s disease in older adults.

50
BMC Geriatrics, 15(1), 1–7. https://doi.org/10.1186/s12877-015-0007-2
Best, J. R., Falck, R. S., Landry, G. J., & Liu-Ambrose, T. (2019). Analysis of dynamic,
bidirectional associations in older adult physical activity and sleep quality. Journal of Sleep
Research, 28(4). https://doi.org/10.1111/jsr.12769
Blackwell, T., Yaffe, K., Ancoli-Israel, S., Schneider, J. L., Cauley, J. A., Hillier, T. A., Fink, H.
A., & Stone, K. L. (2006). Poor sleep is associated with impaired cognitive function in older
women: The study of osteoporotic fractures. Journals of Gerontology - Series A Biological
Sciences and Medical Sciences, 61(4), 405–410. https://doi.org/10.1093/gerona/61.4.405
Blondell, S. J., Hammersley-Mather, R., & Veerman, J. L. (2014). Does physical activity prevent
cognitive decline and dementia?: A systematic review and meta-analysis of longitudinal
studies. BMC Public Health, 14(1), 1–12. https://doi.org/10.1186/1471-2458-14-510
Buman, M. P., & King, A. C. (2010). Exercise as a Treatment to Enhance Sleep. American
Journal of Lifestyle Medicine, 4(6), 500–514. https://doi.org/10.1177/1559827610375532
Burnham, K. P., & Anderson, D. R. (2004). Multimodel Inference. Sociological Methods &
Research, 33(2), 261–304. https://doi.org/10.1177/0049124104268644
Caspersen, C., Powell, K., & Christenson, G. (1985). Physical Activity, Exercise, and Physical
Fitness: Definitions and Distinctions for Health-Related Research. SAGE Open, 100(2),
126–131. https://doi.org/10.1177/2158244017712769
Cavuoto, M. G., Ong, B., Pike, K. E., Nicholas, C. L., Bei, B., & Kinsella, G. J. (2016).
Objective but not subjective sleep predicts memory in community-dwelling older adults.
Journal of Sleep Research, 25(4), 475–485. https://doi.org/10.1111/jsr.12391
Chao, L. L., Buckley, S. T., Kornak, J., Schuff, N., Madison, C., Yaffe, K., Miller, B. L.,
Kramer, J. H., & Weiner, M. W. (2010). ASL perfusion MRI predicts cognitive decline and

51
conversion from MCI to dementia. Alzheimer Disease and Associated Disorders, 24(1), 19–
27. https://doi.org/10.1097/WAD.0b013e3181b4f736
Chen, J. J., Rosas, H. D., & Salat, D. H. (2011). Age-associated reductions in cerebral blood flow
are independent from regional atrophy. NeuroImage, 55(2), 468–478.
https://doi.org/10.1016/j.neuroimage.2010.12.032
Colcombe, S., & Kramer, A. F. (2003). Fitness Effects on the Cognitive Function of Older
Adults: A Meta-Analytic Study. Perspectives on Psychological Science, 14(2), 213–217.
https://doi.org/10.1177/1745691617707316
Cotman, C. W., & Berchtold, N. C. (2007). Physical activity and the maintenance of cognition:
Learning from animal models. Alzheimer’s and Dementia, 3(2 SUPPL.), 30–37.
https://doi.org/10.1016/j.jalz.2007.01.013
Crowley, K. (2011). Sleep and sleep disorders in older adults. Neuropsychology Review, 21(1),
41–53. https://doi.org/10.1007/s11065-010-9154-6
Dai, W., Lopez, O. L., Carmichael, O. T., Becker, J. T., Kuller, L. H., & Gach, H. M. (2009).
Mild cognitive impairment and alzheimer disease: Patterns of altered cerebral blood flow at
MR imaging. Radiology, 250(3), 856–866. https://doi.org/10.1148/radiol.2503080751
De Vis, J. B., Peng, S. L., Chen, X., Li, Y., Liu, P., Sur, S., Rodrigue, K. M., Park, D. C., & Lu,
H. (2018). Arterial-spin-labeling (ASL) perfusion MRI predicts cognitive function in
elderly individuals: A 4-year longitudinal study. Journal of Magnetic Resonance Imaging,
48(2), 449–458. https://doi.org/10.1002/jmri.25938
Devore, E. E., Grodstein, F., Duffy, J. F., Stampfer, M. J., Czeisler, C. A., & Schernhammer, E.
S. (2014). Sleep duration in midlife and later life in relation to cognition. Journal of the
American Geriatrics Society, 62(6), 1073–1081. https://doi.org/10.1111/jgs.12790

52
Durrant, J. R., Seals, D. R., Connell, M. L., Russell, M. J., Lawson, B. R., Folian, B. J., Donato,
A. J., & Lesniewski, L. A. (2009). Voluntary wheel running restores endothelial function in
conduit arteries of old mice: Direct evidence for reduced oxidative stress, increased
superoxide dismutase activity and down-regulation of NADPH oxidase. Journal of
Physiology, 587(13), 3271–3285. https://doi.org/10.1113/jphysiol.2009.169771
Edwards, G. A., Gamez, N., Escobedo, G., Calderon, O., & Moreno-Gonzalez, I. (2019).
Modifiable risk factors for Alzheimer’s disease. Frontiers in Aging Neuroscience, 11(JUN),
1–18. https://doi.org/10.3389/fnagi.2019.00146
Erickson, K. I., Hillman, C., Stillman, C. M., Ballard, R. M., Bloodgood, B., Conroy, D. E.,
Macko, R., Marquez, D. X., Petruzzello, S. J., & Powell, K. E. (2019). Physical Activity,
Cognition, and Brain Outcomes: A Review of the 2018 Physical Activity Guidelines.
Medicine and Science in Sports and Exercise, 51(6), 1242–1251.
https://doi.org/10.1249/MSS.0000000000001936
Erickson, K. I., Voss, M. W., Prakash, R. S., Basak, C., Szabo, A., Chaddock, L., Kim, J. S.,
Heo, S., Alves, H., White, S. M., Wojcicki, T. R., Mailey, E., Vieira, V. J., Martin, S. A.,
Pence, B. D., Woods, J. A., McAuley, E., & Kramer, A. F. (2011). Exercise training
increases size of hippocampus and improves memory. Proceedings of the National
Academy of Sciences of the United States of America, 108(7), 3017–3022.
https://doi.org/10.1073/pnas.1015950108
Espiritu, J. R. D. (2008). Aging-Related Sleep Changes. Clinics in Geriatric Medicine, 24(1), 1–
14. https://doi.org/10.1016/j.cger.2007.08.007
Falck, R. S., Best, J. R., Davis, J. C., & Liu-Ambrose, T. (2018). The Independent Associations
of Physical Activity and Sleep with Cognitive Function in Older Adults. Journal of

53
Alzheimer’s Disease, 63(4), 1469–1484. https://doi.org/10.3233/JAD-170936
Fanning, J., Porter, • G, Awick, E. A., Ehlers, • D K, Roberts, • S A, Cooke, • G, Burzynska, • A
Z, Voss, • M W, Kramer, • A F, & Mcauley, • E. (2017). Replacing sedentary time with
sleep, light, or moderate-to-vigorous physical activity: effects on self-regulation and
executive functioning. Journal of Behavioral Medicine, 40, 332–342.
https://doi.org/10.1007/s10865-016-9788-9
Faraci, F. M., & Heistad, D. D. (1998). Regulation of the cerebral circulation: Role of
endothelium and potassium channels. Physiological Reviews, 78(1), 53–97.
https://doi.org/10.1152/physrev.1998.78.1.53
Fultz, N. E., Bonmassar, G., Setsompop, K., Stickgold, R. A., Rosen, B. R., Polimeni, J. R., &
Lewis, L. D. (2019). Coupled electrophysiological, hemodynamic, and cerebrospinal fluid
oscillations in human sleep. Science, 366(6465), 628–631.
https://doi.org/10.1126/science.aax5440
Gothe, N. P., Fanning, J., Awick, E., Chung, D., Wójcicki, T. R., Olson, E. A., Mullen, S. P.,
Voss, M., Erickson, K. I., Kramer, A. F., & McAuley, E. (2014). Executive function
processes predict mobility outcomes in older adults. Journal of the American Geriatrics
Society, 62(2), 285–290. https://doi.org/10.1111/jgs.12654
Govindpani, K., McNamara, L. G., Smith, N. R., Vinnakota, C., Waldvogel, H. J., Faull, R. L., &
Kwakowsky, A. (2019). Vascular Dysfunction in Alzheimer’s Disease: A Prelude to the
Pathological Process or a Consequence of It? Journal of Clinical Medicine, 8(5), 651.
https://doi.org/10.3390/jcm8050651
Hayes, A. F. (2015). An Index and Test of Linear Moderated Mediation. Multivariate Behavioral
Research, 50(1), 1–22. https://doi.org/10.1080/00273171.2014.962683

54
Hillman, C. H., Erickson, K. I., & Kramer, A. F. (2008). Be smart, exercise your heart: exercise
effects on brain and cognition. Nature, 9.
Hoops, S., Nazem, S., Siderowf, A. D., Duda, J. E., Xie, S. X., Stern, M. B., & Weintraub, D.
(2009). Validity of the MoCA and MMSE in the detection of MCI and dementia in
Parkinson disease. Neurology, 73(21), 1738–1745.
https://doi.org/10.1212/WNL.0b013e3181c34b47
Hshieh, T. T., Dai, W., Cavallari, M., Guttmann, C. R., Meier, D. S., Schmitt, E. M., Dickerson,
B. C., Press, D. Z., Marcantonio, E. R., Jones, R. N., Gou, Y. R., Travison, T. G., Fong, T.
G., Ngo, L., Inouye, S. K., & Alsop, D. C. (2017). Cerebral blood flow MRI in the
nondemented elderly is not predictive of post-operative delirium but is correlated with
cognitive performance. Journal of Cerebral Blood Flow and Metabolism, 37(4), 1386–
1397. https://doi.org/10.1177/0271678X16656014
Hubley, A. M., & Tremblay, D. (2002). Comparability of total score performance on the Rey-
Osterrieth Complex Figure and a modified Taylor Complex Figure. Journal of Clinical and
Experimental Neuropsychology, 24(3), 370–382. https://doi.org/10.1076/jcen.24.3.370.984
Johnson, L. G., Butson, M. L., Polman, R. C., Raj, I. S., Borkoles, E., Scott, D., Aitken, D., &
Jones, G. (2016). Light physical activity is positively associated with cognitive performance
in older community dwelling adults. Journal of Science and Medicine in Sport, 19(11),
877–882. https://doi.org/10.1016/j.jsams.2016.02.002
Ju, Y. E. S., Lucey, B. P., & Holtzman, D. M. (2014). Sleep and Alzheimer disease pathology-a
bidirectional relationship. Nature Reviews Neurology, 10(2), 115–119.
https://doi.org/10.1038/nrneurol.2013.269
Ju, Y. E. S., McLeland, J. S., Toedebusch, C. D., Xiong, C., Fagan, A. M., Duntley, S. P.,

55
Morris, J. C., & Holtzman, D. M. (2013). Sleep quality and preclinical Alzheimer disease.
JAMA Neurology, 70(5), 587–593. https://doi.org/10.1001/jamaneurol.2013.2334
Kennedy, G., Hardman, R. J., MacPherson, H., Scholey, A. B., & Pipingas, A. (2016). How
Does Exercise Reduce the Rate of Age-Associated Cognitive Decline? A Review of
Potential Mechanisms. Journal of Alzheimer’s Disease, 55(1), 1–18.
https://doi.org/10.3233/JAD-160665
Kerr, J., Marshall, S. J., Patterson, R. E., Marinac, C. R., Natarajan, L., Rosenberg, D.,
Wasilenko, K., & Crist, K. (2013). Objectively Measured Physical Activity Is Related to
Cognitive Function in Older Adults. Journal of the American Geriatrics Society, 61(11),
1927–1931. https://doi.org/10.1111/jgs.12524
King, A. C., Pruitt, L. A., Woo, S., Castro, C. M., Ahn, D. K., Vitiello, M. V., Woodward, S. H.,
& Bliwise, D. L. (2008). Effects of moderate-intensity exercise on polysomnographic and
subjective sleep quality in older adults with mild to moderate sleep complaints. Journals of
Gerontology - Series A Biological Sciences and Medical Sciences, 63(9), 997–1004.
https://doi.org/10.1093/gerona/63.9.997
Kishida, M., & Elavsky, S. (2016). An intensive longitudinal examination of daily physical
activity and sleep in midlife women. Sleep Health, 2(1), 42–48.
https://doi.org/10.1016/j.sleh.2015.12.001
Kramer, A. F., & Colcombe, S. (2018). Fitness Effects on the Cognitive Function of Older
Adults: A Meta-Analytic Study—Revisited. Perspectives on Psychological Science, 13(2),
213–217. https://doi.org/10.1177/1745691617707316
Lim, A. S. P., Kowgier, M., Yu, L., Buchman, A. S., & Bennett, D. A. (2013). Sleep
Fragmentation and the Risk of Incident Alzheimer’s Disease and Cognitive Decline in

56
Older Persons. Sleep, 36(7), 1027–1032. https://doi.org/10.5665/sleep.2802
Liu, T. T., & Brown, G. G. (2007). Measurement of cerebral perfusion with arterial spin
labeling: Part 1. Methods. Journal of the International Neuropsychological Society, 13(3),
517–525. https://doi.org/10.1017/S1355617707070646
Livingston, G., Huntley, J., Sommerlad, A., Ames, D., Ballard, C., Banerjee, S., Brayne, C.,
Burns, A., Cohen-Mansfield, J., Cooper, C., Costafreda, S. G., Dias, A., Fox, N., Gitlin, L.
N., Howard, R., Kales, H. C., Kivimäki, M., Larson, E. B., Ogunniyi, A., … Mukadam, N.
(2020). Dementia prevention, intervention, and care: 2020 report of the Lancet
Commission. In The Lancet (Vol. 396, Issue 10248, pp. 413–446). Lancet Publishing
Group. https://doi.org/10.1016/S0140-6736(20)30367-6
Lo, J. C., Groeger, J. A., Cheng, G. H., Dijk, D. J., & Chee, M. W. L. (2016). Self-reported sleep
duration and cognitive performance in older adults: A systematic review and meta-analysis.
Sleep Medicine, 17, 87–98. https://doi.org/10.1016/j.sleep.2015.08.021
Maass, A., Düzel, S., Goerke, M., Becke, A., Sobieray, U., Neumann, K., Lövden, M.,
Lindenberger, U., Bäckman, L., Braun-Dullaeus, R., Ahrens, D., Heinze, H. J., Müller, N.
G., & Düzel, E. (2015). Vascular hippocampal plasticity after aerobic exercise in older
adults. Molecular Psychiatry, 20(5), 585–593. https://doi.org/10.1038/mp.2014.114
Mander, B. A., Rao, V., Lu, B., Saletin, J. M., Ancoli-Israel, S., Jagust, W. J., & Walker, M. P.
(2014). Impaired prefrontal sleep spindle regulation of hippocampal-dependent learning in
older adults. Cerebral Cortex, 24(12), 3301–3309. https://doi.org/10.1093/cercor/bht188
Mander, B. A., Rao, V., Lu, B., Saletin, J. M., Lindquist, J. R., Ancoli-Israel, S., Jagust, W., &
Walker, M. P. (2013). Prefrontal atrophy, disrupted NREM slow waves and impaired
hippocampal-dependent memory in aging. Nature Neuroscience, 16(3), 357–364.

57
https://doi.org/10.1038/nn.3324
Mander, B. A., Winer, J. R., Jagust, W. J., & Walker, M. P. (2016). Sleep: A Novel Mechanistic
Pathway, Biomarker, and Treatment Target in the Pathology of Alzheimer’s Disease?
Trends in Neurosciences, 39(8), 552–566. https://doi.org/10.1016/j.tins.2016.05.002
Mander, B. A., Winer, J. R., & Walker, M. P. (2017). Sleep and Human Aging. Neuron, 94(1),
19–36. https://doi.org/10.1016/j.neuron.2017.02.004
Martin, N., Lafortune, M., Godbout, J., Barakat, M., Robillard, R., Poirier, G., Bastien, C., &
Carrier, J. (2013). Topography of age-related changes in sleep spindles. Neurobiology of
Aging, 34(2), 468–476. https://doi.org/10.1016/j.neurobiolaging.2012.05.020
Maya J Lambiase, Kelley Pettee Gabriel, Lewis H Kuller, and K. A. M. (2013). Temporal
relationships between physical activity and sleep in older women. Med Sci Sports Exc,
23(1), 1–7. https://doi.org/10.1038/jid.2014.371
Migueles, J. H., Rowlands, A. V., Huber, F., Sabia, S., & van Hees, V. T. (2019). GGIR: A
Research Community–Driven Open Source R Package for Generating Physical Activity and
Sleep Outcomes From Multi-Day Raw Accelerometer Data. Journal for the Measurement of
Physical Behaviour, 2(3), 188–196. https://doi.org/10.1123/jmpb.2018-0063
Mitchell, J. A., Godbole, S., Moran, K., Murray, K., James, P., Laden, F., Hipp, J. A., Kerr, J., &
Glanz, K. (2017). and Physical Activity. 48(10), 1950–1956.
https://doi.org/10.1249/MSS.0000000000001000.No
Montagne, A., Nation, D. A., Pa, J., Sweeney, M. D., Toga, A. W., & Zlokovic, B. V. (2016).
Brain imaging of neurovascular dysfunction in Alzheimer’s disease. Acta
Neuropathologica, 131(5), 687–707. https://doi.org/10.1007/s00401-016-1570-0
Montgomery, P., & Ja, D. (2009). Physical exercise for sleep problems in adults aged 60 + (

58
Review ). Cochrane Database of Systematic Reviews, 4, 13.
https://doi.org/10.1002/14651858.CD003404.www.cochranelibrary.com
Nation, D. A., Sweeney, M. D., Montagne, A., Sagare, A. P., D’Orazio, L. M., Pachicano, M.,
Sepehrband, F., Nelson, A. R., Buennagel, D. P., Harrington, M. G., Benzinger, T. L. S.,
Fagan, A. M., Ringman, J. M., Schneider, L. S., Morris, J. C., Chui, H. C., Law, M., Toga,
A. W., & Zlokovic, B. V. (2019). Blood–brain barrier breakdown is an early biomarker of
human cognitive dysfunction. Nature Medicine, 25(2), 270–276.
https://doi.org/10.1038/s41591-018-0297-y
Nebes, R. D., Buysse, D. J., Halligan, E. M., Houck, P. R., & Monk, T. H. (2009). Self-reported
sleep quality predicts poor cognitive performance in healthy older adults. Journals of
Gerontology - Series B Psychological Sciences and Social Sciences, 64(2), 180–187.
https://doi.org/10.1093/geronb/gbn037
Northey, J. M., Cherbuin, N., Pumpa, K. L., Smee, D. J., & Rattray, B. (2018). Exercise
interventions for cognitive function in adults older than 50: A systematic review with meta-
Analysis. British Journal of Sports Medicine, 52(3), 154–160.
https://doi.org/10.1136/bjsports-2016-096587
Ogoh, S., & Ainslie, P. N. (2009). Cerebral blood flow during exercise: Mechanisms of
regulation. Journal of Applied Physiology, 107(5), 1370–1380.
https://doi.org/10.1152/japplphysiol.00573.2009
Pluta, R. M. (2005). Delayed cerebral vasospasm and nitric oxide: Review, new hypothesis, and
proposed treatment. Pharmacology and Therapeutics, 105(1), 23–56.
https://doi.org/10.1016/j.pharmthera.2004.10.002
Reid, K. J., Baron, K. G., Lu, B., Naylor, E., Wolfe, L., & Zee, P. C. (2010). Aerobic exercise

59
improves self-reported sleep and quality of life in older adults with insomnia. Sleep
Medicine, 11(9), 934–940. https://doi.org/10.1016/j.sleep.2010.04.014
Scarmeas, N., Luchsinger, J. A., Schupf, N., Brickman, A. M., Cosentino, S., Tang, M. X., &
Stern, Y. (2009). Physical activity, diet, and risk of Alzheimer disease. JAMA - Journal of
the American Medical Association, 302(6), 627–637.
https://doi.org/10.1001/jama.2009.1144
Sofi, F., Valecchi, D., Bacci, D., Abbate, R., Gensini, G. F., Casini, A., & Macchi, C. (2011).
Physical activity and risk of cognitive decline: A meta-analysis of prospective studies.
Journal of Internal Medicine, 269(1), 107–117. https://doi.org/10.1111/j.1365-
2796.2010.02281.x
Stephen, R., Hongisto, K., Solomon, A., & Lönnroos, E. (2017). Physical Activity and
Alzheimer’s Disease: A Systematic Review. The Journals of Gerontology. Series A,
Biological Sciences and Medical Sciences, 72(6), 733–739.
https://doi.org/10.1093/gerona/glw251
Stillman, C. M., Cohen, J., Lehman, M. E., & Erickson, K. I. (2016). Mediators of physical
activity on neurocognitive function: A review at multiple levels of analysis. Frontiers in
Human Neuroscience, 10(DEC2016), 1–17. https://doi.org/10.3389/fnhum.2016.00626
Tang, N. K. Y., & Sanborn, A. N. (2014). Better quality sleep promotes daytime physical activity
in patients with chronic pain? A multilevel analysis of the within-person relationship. PLoS
ONE, 9(3). https://doi.org/10.1371/journal.pone.0092158
Tucker, J. M., Welk, G. J., & Beyler, N. K. (2011). Physical activity in U.S. adults: Compliance
with the physical activity guidelines for Americans. American Journal of Preventive
Medicine, 40(4), 454–461. https://doi.org/10.1016/j.amepre.2010.12.016

60
van Hees, V. T., Sabia, S., Jones, S. E., Wood, A. R., Anderson, K. N., Kivimäki, M., Frayling,
T. M., Pack, A. I., Bucan, M., Trenell, M. I., Mazzotti, D. R., Gehrman, P. R., Singh-
Manoux, B. A., & Weedon, M. N. (2018). Estimating sleep parameters using an
accelerometer without sleep diary. Scientific Reports, 8(1), 1–11.
https://doi.org/10.1038/s41598-018-31266-z
Van Praag, H. (2008). Neurogenesis and exercise: Past and future directions. NeuroMolecular
Medicine, 10(2), 128–140. https://doi.org/10.1007/s12017-008-8028-z
Vitiello, M. V. (2009). Recent advances in understanding sleep and sleep disturbances in older
adults: Growing older does not mean sleeping poorly. Current Directions in Psychological
Science, 18(6), 316–320. https://doi.org/10.1111/j.1467-8721.2009.01659.x
Wang, Z., Das, S. R., Xie, S. X., Arnold, S. E., Detre, J. A., & Wolk, D. A. (2013). Arterial spin
labeled MRI in prodromal Alzheimer’s disease: A multi-site study. NeuroImage: Clinical,
2(1), 630–636. https://doi.org/10.1016/j.nicl.2013.04.014
Watts, A., Walters, R. W., Hoffman, L., & Templin, J. (2016). Intra-individual variability of
physical activity in older adults with and without mild Alzheimer’s disease. PLoS ONE,
11(4), 1–15. https://doi.org/10.1371/journal.pone.0153898
Wilbur, J., Marquez, D. X., Fogg, L., Wilson, R. S., Staffileno, B. A., Hoyem, R. L., Morris, M.
C., Bustamante, E. E., & Manning, A. F. (2012). The Relationship Between Physical
Activity and Cognition in Older Latinos. The Journals of Gerontology Series B:
Psychological Sciences and Social Sciences, 67(5), 525–534.
https://doi.org/10.1093/geronb/gbr137
Wilckens, K. A., Erickson, K. I., & Wheeler, M. E. (2012). Age-related decline in controlled
retrieval: The role of the PFC and sleep. Neural Plasticity, 2012.

61
https://doi.org/10.1155/2012/624795
Wilckens, K. A., Erickson, K. I., & Wheeler, M. E. (2018). Physical Activity and Cognition: A
Mediating Role of Efficient Sleep. Behavioral Sleep Medicine, 16(6), 569–586.
https://doi.org/10.1080/15402002.2016.1253013
Wilckens, K. A., Hall, M. H., Nebes, R. D., Monk, T. H., & Buysse, D. J. (2016). Changes in
Cognitive Performance Are Associated with Changes in Sleep in Older Adults With
Insomnia. Behavioral Sleep Medicine, 14(3), 295–310.
https://doi.org/10.1080/15402002.2014.1002034
Winer, J. R., Mander, B. A., Helfrich, R. F., Maass, A., Harrison, T. M., Baker, S. L., Knight, R.
T., Jagust, W. J., & Walker, M. P. (2019). Sleep as a Potential Biomarker of Tau and β-
Amyloid Burden in the Human Brain. The Journal of Neuroscience : The Official Journal of
the Society for Neuroscience, 39(32), 6315–6324.
https://doi.org/10.1523/JNEUROSCI.0503-19.2019
Woods, S. P., Delis, D. C., Scott, J. C., Kramer, J. H., & Holdnack, J. A. (2006). The California
Verbal Learning Test - second edition: Test-retest reliability, practice effects, and reliable
change indices for the standard and alternate forms. Archives of Clinical Neuropsychology,
21(5), 413–420. https://doi.org/10.1016/j.acn.2006.06.002
Youngstedt, S. D., O’Connor, P. J., & Dishman, R. K. (1997). The effects of acute exercise on
sleep: A quantitative synthesis. Sleep, 20(3), 203–214.
https://doi.org/10.1093/sleep/20.3.203
Yousaf, T., Dervenoulas, G., & Politis, M. (2018). Advances in MRI Methodology. In
International Review of Neurobiology (1st ed., Vol. 141). Elsevier Inc.
https://doi.org/10.1016/bs.irn.2018.08.008

62
Zhang, N., Gordon, M. L., & Goldberg, T. E. (2017). Cerebral blood flow measured by arterial
spin labeling MRI at resting state in normal aging and Alzheimer’s disease. Neuroscience
and Biobehavioral Reviews, 72, 168–175. https://doi.org/10.1016/j.neubiorev.2016.11.023
Zhu, W., Wadley, V. G., Howard, V. J., Hutto, B., Blair, S. N., & Hooker, S. P. (2017).
Objectively Measured Physical Activity and Cognitive Function in Older Adults. Medicine
and Science in Sports and Exercise, 49(1), 47–53.
https://doi.org/10.1249/MSS.0000000000001079
Zlatar, Z. Z., Hays, C. C., Mestre, Z., Campbell, L. M., Meloy, M. J., Bangen, K. J., Liu, T. T.,
Kerr, J., & Wierenga, C. E. (2019). Dose-dependent association of accelerometer-measured
physical activity and sedentary time with brain perfusion in aging. Experimental
Gerontology, 125(July). https://doi.org/10.1016/j.exger.2019.110679
Zlatar, Z. Z., Wierenga, C. E., Bangen, K. J., Liu, T. T., & Jak, A. J. (2014). Increased
hippocampal blood flow in sedentary older adults at genetic risk for Alzheimer’s disease.
Journal of Alzheimer’s Disease : JAD, 41(3), 809–817. https://doi.org/10.3233/JAD-132252
Zlokovic, B. V. (2011). Neurovascular pathways to neurodegeneration in Alzheimer’s disease
and other disorders. Nature Reviews Neuroscience, 12(12), 723–738.
https://doi.org/10.1038/nrn3114

63
Appendix
Aim 1 Supplemental Tables
Appendix Table 1: Physical Activity as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. SE à PA 0.47 0.55 [-0.61,
1.54]
0.393 0.36 0.61 [-0.84,
1.55]
0.560
b. PA à MoCA 0.00 0.01 [-0.03,
0.02]
0.860 0.00 0.01 [-0.02,
0.02]
0.985
c’. SE à MoCA 0.11 0.05 [0.00,
0.21]
0.044* 0.12 0.05 [0.02,
0.22]
0.024*
a*b. SE à PA à MoCA 0.00 0.01 [-0.01,
0.01]
0.865 0.00 0.00 [-0.01,
0.01]
0.985
c. SE à MoCA 0.11 0.05 [0.00,
0.21]
0.043* 0.12 0.05 [0.02,
0.22]
0.023*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. SE à PA 0.51 0.57 [-0.62,
1.63]
0.377 0.41 0.62 [-0.80,
1.63]
0.506
b. PA à CVLT 0.00 0.01 [-0.02,
0.03]
0.744 0.00 0.01 [-0.02,
0.03]
0.833
c’. SE à CVLT 0.08 0.09 [-0.09,
0.24]
0.359 0.08 0.08 [-0.08,
0.24]
0.319
a*b. SE à PA à CVLT 0.00 0.01 [-0.01,
0.02]
0.740 0.00 0.01 [-0.01,
0.01]
0.834
c. SE à CVLT 0.08 0.09 [-0.09,
0.25]
0.356 0.08 0.08 [-0.08,
0.25]
0.320

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p

64
a. SE à PA 0.36 0.50 [-0.62,
1.35]
0.469 0.24 0.57 [-0.88,
1.36]
0.674
b. PA à CF 0.00 0.02 [-0.04,
0.03]
0.821 0.00 0.02 [-0.03,
0.03]
0.934
c’. SE à CF 0.18 0.10 [0.00,
0.37]
0.053 0.20 0.09 [0.03,
0.37]
0.025*
a*b. SE à PA à CF 0.00 0.01 [-0.01,
0.01]
0.811 0.00 0.00 [-0.01,
0.01]
0.934
c. SE à CF 0.18 0.10 [-0.01,
0.37]
0.056 0.20 0.09 [0.02,
0.38]
0.026*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. SE à PA 0.62 0.51 [-0.37,
1.61]
0.221 0.43 0.58 [-0.71,
1.56]
0.459
b. PA à Flanker 0.10 0.04 [0.03,
0.17]
0.003* 0.10 0.04 [0.03,
0.17]
0.009*
c’. SE à Flanker -0.40 0.16 [-0.72, -
0.08]
0.014* -0.40 0.16 [-0.73, -
0.08]
0.013*
a*b. SE à PA à
Flanker
0.06 0.06 [-0.05,
0.18]
0.282 0.04 0.06 [-0.08,
0.16]
0.493
c. SE à Flanker -0.34 0.16 [-0.64, -
0.03]
0.031* -0.36 0.16 [-0.68, -
0.04]
0.027*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. SE à PA 0.35 0.57 [-0.76,
1.46]
0.538 0.30 0.67 [-1.01,
1.60]
0.658
b. PA à LNS -0.03 0.02 [-0.06,
0.00]
0.063 -0.03 0.01 [-0.06,
0.00]
0.038*
c’. SE à LNS 0.07 0.07 [-0.06,
0.20]
0.270 0.07 0.06 [-0.05,
0.20]
0.231
a*b. SE à PA à LNS -0.01 0.02 [-0.04,
0.02]
0.533 -0.01 0.02 [-0.05,
0.03]
0.651

65
c. SE à LNS 0.06 0.07 [-0.07,
0.19]
0.341 0.07 0.06 [-0.05,
0.18]
0.279

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. SE à PA 0.38 0.56 [-0.72,
1.48]
0.501 0.33 0.66 [-0.97,
1.63]
0.619
b. PA à Latent Cog 0.00 0.03 [-0.05,
0.05]
0.993 0.00 0.01 [-0.01,
0.02]
0.616
c’. SE à Latent Cog 0.18 0.11 [-0.05,
0.40]
0.122 0.00 0.04 [-0.07,
0.07]
0.899
a*b. SE à PA à Latent
Cog
0.00 0.01 [-0.02,
0.02]
0.993 0.00 0.00 [-0.01,
0.01]
0.719
c. SE à Latent Cog 0.18 0.11 [-0.04,
0.39]
0.113 0.00 0.04 [-0.08,
0.07]
0.934

PA = All light and moderate physical activity; SE = Sleep efficiency; MoCA = Montreal cognitive assessment;
CVLT = California Verbal Learning Test long delay score; CF = Complex Figure Test long day; Flanker =
Inhibitory control measured via Flanker; LNS = Letter-number sequencing; Bold rows indicate a significant
relationship (p<.05); Covariates = Age, Sex, & Education.

Appendix Table 2: Sleep Efficiency as Proposed Mediator – Light Activity Only
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.10 0.09 [-0.07,
0.27]
0.251 0.08 0.10 [-0.11,
0.26]
0.428
b. SE à MoCA 0.11 0.05 [0.002,
0.21]
0.046* 0.05 0.05 [0.01,
0.22]
0.026*
c’. PA à MoCA -0.01 0.05 [-0.09,
0.08]
0.919 0.01 0.05 [-0.08,
0.10]
0.886
a*b. PA à SE à MoCA 0.01 0.01 [-0.01,
0.03]
0.330 0.01 0.01 [-0.02,
0.03]
0.470
c. PA à MoCA 0.01 0.05 [-0.08,
0.10]
0.893 0.02 0.05 [-0.08,
0.12]
0.743

66
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.10 0.09 [-0.07,
0.26]
0.271 0.08 0.10 [-0.12,
0.27]
0.400
b. SE à CVLT 0.08 0.09 [-0.09,
.25]
0.351 0.08 0.08 [-0.08,
0.24]
0.311
c’. PA à CVLT 0.004 0.05 [-0.09,
0.10]
0.931 -
0.003
0.05 [-0.10,
0.10]
0.955
a*b. PA à SE à CVLT 0.01 0.01 [-0.01,
0.03]
0.497 0.01 0.01 [-0.01,
0.03]
0.529
c. PA à CVLT 0.012 0.05 [-0.09,
0.11]
0.818 0.004 0.05 [-0.10,
0.10]
0.941

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.09 0.09 [-0.08,
0.26]
0.294 0.06 .0.10 [-0.13,
0.25]
0.535
b. SE à CF 0.19 0.10 [0.002,
0.37]
0.047* 0.20 0.09 [0.03,
0.38]
0.024*
c’. PA à CF -0.04 0.06 [-0.15,
0.08]
0.556 -0.02 0.06 [-0.13,
0.10]
0.739
a*b. PA à SE à CF 0.02 0.02 [-0.02,
0.06]
0.381 0.01 0.02 [-0.03,
0.05]
0.556
c. PA à CF -0.02 0.07 [-0.15,
0.12]
0.793 -0.01 0.07 [-0.14,
0.12]
0.909

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.12 0.09 [-0.05,
0.29]
0.151 0.09 0.10 [-0.11,
0.28]
0.370
b. SE à Flanker -4.21 1.67 [-7.49, -
0.94]
0.012* -4.20 1.66 [-7.46, -
0.95]
0.011*

67
c’. PA à Flanker 3.90 1.32 [1.31,
6.50]
0.003* 3.90 1.38 [1.20,
6.59]
0.005*
a*b. PA à SE à
Flanker
-0.52 0.45 [-1.41,
0.37]
0.250 -0.37 0.56 [-1.27,
0.53]
0.417
c. PA à Flanker 3.38 1.31 [0.81,
5.96]
0.010* 3.53 1.36 [0.87,
6.18]
0.009*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.08 0.09 [-0.10,
0.26]
0.377 0.07 0.11 [-0.15,
0.29]
0.524
b. SE à LNS 0.08 0.07 [-0.05,
0.21]
0.226 0.08 0.06 [-0.04,
0.20]
0.202
c’. PA à LNS -0.11 0.06 [-0.23,
0.01]
0.064 -0.11 0.05 [-0.21, -
0.01]
0.040*
a*b. PA à SE à LNS 0.01 0.01 [-0.01,
0.02]
0.466 0.01 0.01 [-0.01,
0.03]
0.580
c. PA à LNS -0.11 0.06 [-0.23,
0.02]
0.087 -0.10 0.06 [-0.21,
0.01]
0.061

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE 0.09 0.10 [-0.11,
0.29]
0.374 0.08 0.13 [-0.16,
0.33]
0.504
b. SE à Latent Cog 0.18 0.11 [-0.04,
0.40]
0.110 -0.01 0.03 [-0.07,
0.06]
0.887
c’. PA à Latent Cog -0.02 0.10 [-0.21,
0.17]
0.847 0.01 0.03 [-0.04,
0.07]
0.635
a*b. PA à SE à Latent
Cog
0.02 0.02 [-0.03,
0.06]
0.469 0.00 0.00 [-0.01,
0.01]
0.888
c. PA à Latent Cog 0.00 0.10 [-0.19,
0.19]
0.984 0.01 0.03 [-0.05,
0.07]
0.668

68
PA = All light and moderate physical activity; SE = Sleep efficiency; MoCA = Montreal cognitive assessment;
CVLT = California Verbal Learning Test long delay score; CF = Complex Figure Test long day; Flanker =
Inhibitory control measured via Flanker; LNS = Letter-number sequencing; Bold rows indicate a significant
relationship (p<.05); Covariates = Age, Sex, & Education.

Appendix Table 3: Global Gray Matter CBF as Proposed Mediator – Light Activity Only
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.39 0.19 [0.03,
0.75]
0.036* 0.29 0.17 [-0.05,
0.64]
0.091
b. CBFà MoCA 0.08 0.03 [0.01,
0.14]
0.022* 0.10 0.04 [0.02,
0.18]
0.011*
c’. PA à MoCA -0.03 0.05 [-0.12,
0.07]
0.589 -0.02 0.05 [-0.11,
0.07]
0.713
a*b. PA à CBF à
MoCA
0.03 0.02 [-0.01,
0.07]
0.101 0.03 0.02 [-0.01,
0.07]
0.128
c. PA à MoCA 0.01 0.05 [-0.09,
0.10]
0.918 0.01 0.05 [-0.08,
0.11]
0.779

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.39 0.19 [0.03,
0.75]
0.036* 0.29 0.17 [-0.05,
0.64]
0.091
b. CBFà CVLT 0.06 0.05 [-0.05,
0.16]
0.302 0.05 0.06 -0.06,
0.16]
0.407
c’. PA à CVLT -0.002 0.05 [-0.10,
0.09]
0.974 -
0.003
0.05 [-0.10,
0.09]
0.943
a*b. PA à CBF à
CVLT
0.02 0.02 [-0.03,
0.07]
0.362 0.01 0.02 [-0.02,
0.05]
0.431
c. PA à CVLT 0.02 0.05 [-0.08,
0.12]
0.700 0.01 0.05 [-0.09,
0.11]
0.837

Unadjusted for covariates Adjusted for covariates

69
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.30 0.18 [-0.05,
0.65]
0.090 0.21 0.16 [-0.11,
0.53]
0.191
b. CBFà CF 0.08 0.08 [-0.08,
0.23]
0.332 0.09 0.08 [-0.07,
0.25]
0.263
c’. PA à CF -0.03 0.06 [-0.16,
0.09]
0.584 -0.03 0.06 [-0.14,
0.09]
0.621
a*b. PA à CBF à CF 0.02 0.03 [-0.03,
0.08]
0.408 0.02 0.02 [-0.02,
0.06]
0.380
c. PA à CF -0.01 0.07 [-0.15,
0.13]
0.874 -0.01 0.07 [-0.14,
0.12]
0.891

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.40 0.19 [0.03,
0.76]
0.035* 0.30 0.19 [-0.07,
0.66]
0.110
b. CBFà Flanker 1.04 1.26 [-1.43,
3.51]
0.408 1.17 1.39 [-1.55,
3.90]
0.399
c’. PA à Flanker 3.43 1.37 [0.73,
6.12]
0.013* 3.55 1.34 [0.92,
6.19]
0.008*
a*b. PA à CBF à
Flanker
0.41 0.54 [-0.65,
1.47]
0.448 0.35 0.49 [-0.61,
1.31]
0.478
c. PA à Flanker 3.84 1.28 [1.34,
6.34]
0.003* 3.90 1.34 [1.28,
6.53]
0.004*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.37 0.21 [-0.05,
0.77]
0.081 0.23 0.19 [-0.14,
0.59]
0.219
b. CBFà LNS 0.10 0.05 [0.00,
0.20]
0.049* 0.11 0.05 [0.01,
0.21]
0.031*
c’. PA à LNS -0.14 0.06 [-0.25, -
0.02]
0.019* -0.12 0.05 [-0.21, -
0.03]
0.012*

70
a*b. PA à CBF à LNS 0.04 0.03 [-0.02,
0.09]
0.183 0.02 0.02 [-0.02,
0.07]
0.278
c. PA à LNS -0.10 0.07 [-0.23,
0.03]
0.114 -0.10 0.06 [-0.20,
0.01]
0.087

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.30 0.22 [-0.12,
0.72]
0.164 0.16 0.20 [-0.24,
0.56]
0.425
b. CBFà Latent Cog 0.08 0.07 [-0.06,
0.22]
0.262 0.10 0.09 [-0.07,
0.26]
0.266
c’. PA à Latent Cog -0.05 0.07 [-0.18,
0.09]
0.499 -0.04 0.08 [-0.18,
0.11]
0.623
a*b. PA à CBF à
Latent Cog
0.02 0.03 [-0.03,
0.08]
0.418 0.02 0.02 [-0.03,
0.06]
0.483
c. PA à Latent Cog -0.02 0.08 [-0.19,
0.14]
0.797 -0.02 0.09 [-0.19,
0.15]
0.804

PA = All light and moderate physical activity; CBF = Global gray matter cerebral blood flow; MoCA =
Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex
Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number sequencing;
Bold rows indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Appendix Table 4: Sleep Efficiency as Proposed Mediator – Moderate Activity Only
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE
-0.03 0.05 [-0.13,
0.07]
0.554 -0.02 0.06 [-0.13,
0.09]
0.723
b. SE à MoCA
0.12 0.05 [0.002,
0.21]
0.045* 0.12 0.05 [0.02,
0.22]
0.024*
c’. PA à MoCA
-0.01 0.02 [-0.05,
0.03]
0.677 -0.01 0.02 [-0.05,
0.03]
0.602
a*b. PA à SE à MoCA
-0.003 0.01 [-0.01,
0.01]
0.570 -0.002 0.01 [-0.02,
0.01]
0.725

71
c. PA à MoCA
-0.01 0.02 [-0.05,
0.03]
0.578 -0.01 0.02 [-0.05,
0.03]
0.527

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE -0.02 0.05 [-0.12,
0.08]
0.717 -0.01 0.06 [-0.13,
0.10]
0.814
b. SE à CVLT 0.08 0.09 [-0.09,
0.25]
0.717 0.09 0.08 [-0.07,
0.24]
0.296
c’. PA à CVLT 0.03 0.02 [-0.02,
0.07]
0.219 0.03 0.03 [-0.02,
0.08]
0.220
a*b. PA à SE à CVLT -0.002 0.004 [-0.01,
0.01]
0.722 -0.001 0.01 [-0.01,
0.01
0.811
c. PA à CVLT -0.03 0.02 [-0.02,
0.07]
0.228 0.03 0.03 [-0.02,
0.08]
0.250

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE -0.04 0.05 [-0.13,
0.06]
0.466 -0.02 0.06 [-0.13,
0.09]
0.697
b. SE à CF 0.19 0.09 [0.01,
0.37]
0.042* 0.20 0.09 [0.03,
0.38]
0.023*
c’. PA à CF 0.03 0.03 [-0.03,
0.09]
0.305 0.02 0.03 [-0.04,
0.09]
0.453
a*b. PA à SE à CF -0.01 0.01 [-0.02,
0.01]
0.442 -
0.004
0.01 [-0.03,
0.02]
0.690
c. PA à CF 0.03 0.03 [-0.04,
0.09]
0.457 0.02 0.04 [-0.05,
0.09]
0.579

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE -0.01 0.05 [-0.11,
0.09]
0.798 -0.01 0.06 [-0.12,
0.11]
0.882

72
b. SE à Flanker -3.27 1.54 [-6.28, -
0.27]
0.033* -3.56 1.61 [-6.72, -
0.41]
0.027*
c’. PA à Flanker 1.11 0.93 [-0.70,
2.91]
0.230 1.14 0.96 [-0.73,
3.02]
0.231
a*b. PA à SE à
Flanker
0.04 0.17 [-0.29,
0.38]
0.799 0.03 0.21 [-0.37,
0.43]
0.882
c. PA à Flanker 1.15 0.87 [-0.56,
2.86]
0.187 1.18 0.94 [-0.68,
3.02]
0.213

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE -0.03 0.05 [-0.13,
0.07]
0.532 -0.02 0.06 [-0.14,
0.10]
0.714
b. SE à LNS 0.05 0.07 [-0.08,
0.18]
0.411 0.06 0.06 [-0.06,
0.18]
0.323
c’. PA à LNS -0.04 0.03 [-0.10,
0.02]
0.194 -0.04 0.03 [-0.10,
0.01]
0.135
a*b. PA à SE à LNS -0.002 0.003 [-0.01,
0.004]
0.546 -
0.001
0.004 [-0.01,
0.01]
0.705
c. PA à LNS -0.04 0.03 [-0.10,
0.02]
0.187 -0.04 0.03 [-0.10,
0.01]
0.136

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à SE -0.03 0.06 [-0.13,
0.08]
0.635 -0.02 0.07 [-0.15,
0.11]
0.777
b. SE à Latent Cog 0.17 0.11 [-0.04,
0.39]
0.108 0.00 0.04 [-0.09,
0.09]
0.996
c’. PA à Latent Cog 0.03 0.03 [-0.04,
0.09]
0.411 0.01 0.02 [-0.03,
0.05]
0.583
a*b. PA à SE à Latent
Cog
0.00 0.01 [-0.02,
0.01]
0.625 0.00 0.00 [0.00,
0.00]
0.996

73
c. PA à Latent Cog 0.02 0.04 [-0.05,
0.09]
0.521 0.01 0.02 [-0.02,
0.04]
0.568

PA = All light physical activity; SE = Sleep efficiency; MoCA = Montreal cognitive assessment; CVLT =
California Verbal Learning Test long delay score; CF = Complex Figure Test long day; Flanker = Inhibitory
control measured via Flanker; LNS = Letter-number sequencing; Bold rows indicate a significant
relationship (p<.05); Covariates = Age, Sex, & Education.

Appendix Table 5: Global Gray Matter CBF as Proposed Mediator – Moderate Activity Only
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.05 0.07 [-0.10,
0.19]
0.531 0.07 0.08 [-0.09,
0.23]
0.374
b. CBFà MoCA 0.07 0.03 [0.01,
0.13]
0.019* 0.10 0.04 [0.03,
0.18]
0.009*
c’. PA à MoCA -0.02 0.02 [-0.06,
0.03]
0.452 -0.02 0.02 [-0.06,
0.02]
0.282
a*b. PA à CBF à
MoCA
0.003 0.01 [-0.01,
0.01]
0.527 0.01 0.01 [-0.01,
0.03]
0.396
c. PA à MoCA -0.01 0.02 [-0.06,
0.03]
0.568 -0.01 0.02 [-0.06,
0.03]
0.507

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.05 0.07 [-0.10,
0.19]
0.531 0.07 0.08 [-0.09,
0.23]
0.374
b. CBFà CVLT 0.05 0.05 [-0.05,
0.16]
0.329 0.04 0.06 [-0.07,
0.15]
0.486
c’. PA à CVLT 0.02 0.02 [-0.02,
0.06]
0.275 0.03 0.02 [-0.02,
0.07]
0.224
a*b. PA à CBF à
CVLT
0.002 0.01 [-0.01,
0.01]
0.613 0.003 0.01 [-0.01,
0.01]
0.555
c. PA à CVLT 0.03 0.02 [-0.02,
0.07]
0.258 0.03 0.03 [-0.02,
0.08]
0.229

74
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.02 0.07 [-0.11,
0.14]
0.808 0.05 0.07 [-0.10,
0.19]
0.541
b. CBFà CF 0.07 0.08 [-0.09,
0.22]
0.407 0.08 0.09 [-0.08,
0.25]
0.329
c’. PA à CF 0.02 0.03 [-0.04,
0.08]
0.523 0.02 0.03 [-0.04,
0.08]
0.582
a*b. PA à CBF à CF 0.001 0.01 [-0.01,
0.01]
0.827 0.004 0.01 [-0.01,
0.02]
0.595
c. PA à CF 0.02 0.03 [-0.04,
0.09]
0.528 0.02 0.04 [-0.05,
0.09]
0.556

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.07 0.08 [-0.08,
0.22]
0.344 0.08 0.08 [-0.08,
0.23]
0.345
b. CBFà Flanker 2.01 1.25 [-0.43,
4.46]
0.107 1.99 1.48 [-0.91,
4.88]
0.178
c’. PA à Flanker 1.01 0.87 [-0.70,
2.71]
0.248 0.94 0.95 [-0.91,
2.79]
0.320
a*b. PA à CBF à
Flanker
0.14 0.18 [-0.58,
0.50]
0.432 0.15 0.23 [-0.29,
0.59]
0.502
c. PA à Flanker 1.15 0.88 [-0.58,
2.88]
0.193 1.09 0.92 [-0.71,
2.89]
0.235

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.04 0.08 [-0.11,
0.19]
0.564 0.04 0.08 [-0.13,
0.20]
0.673
b. CBFà LNS 0.06 0.05 [-0.03,
0.15]
0.172 0.09 0.05 [-0.01,
0.19]
0.089

75
c’. PA à LNS -0.04 0.03 [-0.10,
0.02]
0.144 -0.04 0.03 [-0.09,
0.01]
0.079
a*b. PA à CBF à LNS 0.003 0.01 [-0.01,
0.01]
0.617 0.003 0.01 [-0.01,
0.02]
0.670
c. PA à LNS -0.04 0.03 [-0.10,
0.02]
0.191 -0.04 0.03 [-0.10,
0.02]
0.148

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.05 0.07 [-0.09,
0.19]
0.502 0.04 0.09 [-0.13,
0.21]
0.626
b. CBF à Latent Cog 0.07 0.07 [-0.07,
0.21]
0.345 0.09 0.09 [-0.08,
0.26]
0.309
c’. PA à Latent Cog 0.01 0.03 [-0.06,
0.07]
0.806 0.00 0.04 [-0.07,
0.08]
0.962
a*b. PA à SE à Latent
Cog
0.00 0.01 [-0.01,
0.02]
0.627 0.00 0.01 [-0.01,
0.02]
0.651
c. PA à Latent Cog 0.01 0.04 [-0.06,
0.08]
0.753 0.01 0.04 [-0.08,
0.09]
0.898

PA = All moderate physical activity; CBF = Global gray matter cerebral blood flow; MoCA = Montreal
cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex Figure Test
long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number sequencing; Bold rows
indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Appendix Table 6: Hippocampal Cerebral Blood Flow as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.01 0.01 [0.00,
0.02]
0.099 0.01 0.01 [0.00,
0.02]
0.084
b. CBFà MoCA -0.48 0.29 [-1.05,
0.09]
0.098 -0.59 0.31 [-1.19,
0.02]
0.059
c’. PA à MoCA -0.01 0.01 [-0.03,
0.02]
0.701 0.00 0.01 [-0.03,
0.03]
0.992

76
a*b. PA à CBF à
MoCA
0.00 0.00 [-0.01,
0.00]
0.218 -0.01 0.00 [-0.01,
0.00]
0.160
c. PA à MoCA -0.01 0.01 [-0.04,
0.02]
0.514 -0.01 0.01 [-0.03,
0.02]
0.722

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.01 0.01 [0.00,
0.02]
0.101 0.01 0.01 [0.00,
0.02]
0.084
b. CBFà CVLT -0.81 0.46 [-1.70,
0.09]
0.077 -1.09 0.44 [-1.94, -
0.23]
0.013*
c’. PA à CVLT 0.01 0.02 [-0.03,
0.04]
0.729 0.01 0.02 [-0.02,
0.04]
0.587
a*b. PA à CBF à
CVLT
-0.01 0.01 [-0.02,
0.00]
0.169 -0.01 0.01 [-0.02,
0.00]
0.108
c. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.930 0.00 0.01 [-0.03,
0.03]
0.937

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.01 0.01 [0.00,
0.02]
0.051 0.01 0.01 [0.00,
0.02]
0.061
b. CBFà CF -0.54 0.50 [-1.52,
0.44]
0.279 -0.45 0.58 [-1.58,
0.69]
0.440
c’. PA à CF 0.00 0.02 [-0.04,
0.04]
0.985 0.00 0.02 [-0.04,
0.04]
0.975
a*b. PA à CBF à CF -0.01 0.01 [-0.02,
0.01]
0.313 0.00 0.01 [-0.01,
0.01]
0.423
c. PA à CF -0.01 0.02 [-0.04,
0.03]
0.753 0.00 0.02 [-0.04,
0.03]
0.847

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p

77
a. PA à CBF 0.01 0.01 [0.00,
0.02]
0.069 0.01 0.01 [0.00,
0.02]
0.055
b. CBFà Flanker 0.05 1.03 [-1.98,
2.07]
0.964 -0.01 1.11 [-2.18,
2.17]
0.997
c’. PA à Flanker 0.13 0.05 [0.03,
0.22]
0.009* 0.12 0.05 [0.03,
0.21]
0.010*
a*b. PA à CBF à
Flanker
0.00 0.01 [-0.02,
0.02]
0.964 0.00 0.01 [-0.02,
0.02]
0.997
c. PA à Flanker 0.13 0.05 [0.03,
0.22]
0.008* 0.12 0.04 [0.03,
0.21]
0.007*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.01 0.01 [0.00,
0.02]
0.070 0.01 0.01 [0.00,
0.02]
0.113
b. CBFà LNS -0.28 0.36 [-0.98,
0.43]
0.442 -0.07 0.37 [-0.79,
0.65]
0.855
c’. PA à LNS -0.03 0.02 [-0.06,
0.00]
0.054 -0.03 0.01 [-0.06,
0.00]
0.034*
a*b. All PA à CBF à
LNS
0.00 0.00 [-0.01,
0.01]
0.499 0.00 0.00 [-0.01,
0.01]
0.853
c. All PA à LNS -0.03 0.02 [-0.06,
0.00]
0.047* -0.03 0.02 [-0.06,
0.00]
0.035*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.01 0.01 [0.00,
0.02]
0.075 0.01 0.01 [0.00,
0.02]
0.085
b. CBF à Latent Cog -0.85 0.61 [-2.05,
0.36]
0.168 -1.00 0.64 [-2.25,
0.26]
0.120
c’. PA à Latent Cog -0.01 0.03 [-0.06,
0.04]
0.760 0.00 0.03 [-0.05,
0.05]
0.904
a*b. PA à SE à Latent
Cog
-0.01 0.01 [-0.02,
0.01]
0.229 -0.01 0.01 [-0.03,
0.00]
0.160

78
c. PA à Latent Cog -0.02 0.02 [-0.06,
0.03]
0.464 -0.01 0.02 [-0.06,
0.03]
0.557

PA = All light and moderate physical activity; CBF = Hippocampal cerebral blood flow; MoCA = Montreal
cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex Figure Test
long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number sequencing; Bold rows
indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Appendix Table 7: Anterior Cingulate Gyrus Cerebral Blood Flow as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.662 0.00 0.01 [-0.01,
0.01]
0.680
b. CBFà MoCA 0.42 0.46 [-0.49,
1.33]
0.365 0.58 0.51 [-0.41,
1.57]
0.253
c’. PA à MoCA -0.01 0.01 [-0.04,
0.02]
0.463 -0.01 0.01 [-0.03,
0.02]
0.653
a*b. PA à CBF à
MoCA
0.00 0.00 [0.00,
0.01]
0.676 0.00 0.00 [0.00,
0.01]
0.684
c. PA à MoCA -0.01 0.01 [-0.04,
0.00]
0.514 -0.01 0.01 [-0.03,
0.02]
0.722

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.660 0.00 0.01 [-0.01,
0.01]
0.666
b. CBFà CVLT -0.50 0.42 [-1.32,
0.32]
0.232 -0.32 0.52 [-1.33,
0.70]
0.541
c’. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.991 0.00 0.02 [-0.03,
0.03]
0.972
a*b. PA à CBF à
CVLT
0.00 0.00 [-0.01,
0.00]
0.674 0.00 0.00 [0.00,
0.00]
0.729
c. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.930 0.00 0.01 [-0.03,
0.03]
0.937

79
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.854 0.00 0.01 [-0.01,
0.01]
0.913
b. CBFà CF 0.42 0.27 [-0.10,
0.95]
0.115 0.31 0.30 [-0.28,
0.90]
0.305
c’. PA à CF -0.01 0.02 [-0.04,
0.03]
0.731 0.00 0.02 [-0.04,
0.03]
0.838
a*b. PA à CBF à CF 0.00 0.00 [0.00,
0.01]
0.859 0.00 0.00 [0.00,
0.00]
0.914
c. PA à CF -0.01 0.02 [-0.04,
0.03]
0.753 0.00 0.02 [-0.04,
0.03]
0.847

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.676 0.00 0.01 [-0.01,
0.01]
0.755
b. CBFà Flanker -1.64 0.97 [-3.54,
0.27]
0.092 -2.00 1.00 [-3.95, -
0.04]
0.045*
c’. PA à Flanker 0.13 0.05 [0.03,
0.23]
0.008* 0.12 0.05 [0.03,
0.21]
0.009*
a*b. PA à CBF à
Flanker
0.00 0.01 [-0.02,
0.01]
0.661 0.00 0.01 [-0.02,
0.02]
0.751
c. PA à Flanker 0.13 0.05 [0.03,
0.22]
0.008* 0.12 0.04 [0.03,
0.21]
0.007*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.933 0.00 0.01 [-0.01,
0.01]
0.635
b. CBFà LNS 0.13 0.57 [-0.99,
1.25]
0.820 -0.33 0.59 [-1.50,
0.83]
0.576

80
c’. PA à LNS -0.03 0.02 [-0.06,
0.00]
0.045* -0.03 0.02 [-0.06,
0.00]
0.044*
a*b. All PA à CBF à
LNS
0.00 0.00 [0.00,
0.00]
0.939 0.00 0.00 [0.00,
0.00]
0.695
c. All PA à LNS -0.03 0.02 [-0.06,
0.00]
0.047* -0.03 0.02 [-0.06,
0.00]
0.035

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.734 0.00 0.01 [-0.01,
0.01]
0.477
b. CBF à Latent Cog -0.05 1.56 [-3.11,
3.01]
0.975 0.10 2.37 [-4.55,
4.74]
0.968
c’. PA à Latent Cog -0.02 0.03 [-0.08,
0.04]
0.554 -0.02 0.03 [-0.07,
0.04]
0.601
a*b. PA à SE à Latent
Cog
0.00 0.00 [-0.01,
0.01]
0.975 0.00 0.01 [-0.02,
0.02]
0.968
c. PA à Latent Cog -0.02 0.03 [-0.07,
0.04]
0.519 -0.01 0.02 [-0.06,
0.03]
0.523

PA = All light and moderate physical activity; CBF = Anterior Cingulate Gyrus cerebral blood flow; MoCA =
Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex
Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number sequencing;
Bold rows indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Appendix Table 8: Middle Frontal Gyrus Cerebral Blood Flow as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF -0.01 0.00 [-0.01,
0.00]
0.192 -0.01 0.00 [-0.01,
0.00]
0.048*
b. CBFà MoCA 0.46 0.42 [-0.35,
1.28]
0.264 0.73 0.46 [-0.18,
1.64]
0.115
c’. PA à MoCA -0.01 0.01 [-0.04,
0.02]
0.621 0.00 0.01 [-0.03,
0.03]
0.986
a*b. PA à CBF à
MoCA
0.00 0.00 [-0.01,
0.00]
0.321 -0.01 0.00 [-0.01,
0.00]
0.150

81
c. PA à MoCA -0.01 0.01 [-0.04,
0.02]
0.514 -0.01 0.01 [-0.03,
0.02]
0.722

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF -0.01 0.00 [-0.01,
0.00]
0.194 -0.01 0.00 [-0.01,
0.00]
0.049*
b. CBFà CVLT 0.72 0.43 [-0.11,
1.56]
0.090 1.12 0.49 [0.16,
2.07]
0.022*
c’. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.881 0.01 0.01 [-0.02,
0.04]
0.642
a*b. PA à CBF à
CVLT
0.00 0.00 [-0.01,
0.00]
0.291 -0.01 0.01 [-0.02,
0.00]
0.135
c. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.930 0.00 0.01 [-0.03,
0.03]
0.937

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF -0.01 0.00 [-0.01,
0.00]
0.190 -0.01 0.00 [-0.01,
0.00]
0.063
b. CBFà CF 0.83 0.55 [-0.24,
1.90]
0.128 0.87 0.59 [-0.28,
2.03]
0.138
c’. PA à CF 0.00 0.02 [-0.04,
0.04]
0.933 0.00 0.02 [-0.03,
0.04]
0.896
a*b. PA à CBF à CF 0.00 0.01 [-0.01,
0.01]
0.346 -0.01 0.01 [-0.02,
0.01]
0.269
c. PA à CF -0.01 0.02 [-0.04,
0.03]
0.753 0.00 0.02 [-0.04,
0.03]
0.847

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.00]
0.342 -0.01 0.00 [-0.01,
0.00]
0.077

82
b. CBFà Flanker -0.54 1.43 [-3.34,
2.27]
0.707 -0.86 1.71 [-4.21,
2.49]
0.615
c’. PA à Flanker 0.13 0.05 [0.03,
0.22]
0.007* 0.11 0.04 [0.03,
0.19]
0.006*
a*b. PA à CBF à
Flanker
0.00 0.01 [-0.01,
0.01]
0.693 0.01 0.01 [-0.02,
0.03]
0.600
c. PA à Flanker 0.13 0.05 [0.03,
0.22]
0.008* 0.12 0.04 [0.03,
0.21]
0.007*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.350 -0.01 0.00 [-0.01,
0.00]
0.207
b. CBFà LNS 0.68 0.26 [0.16,
1.19]
0.010* 0.40 0.29 [-0.17,
0.97]
0.173
c’. PA à LNS -0.03 0.02 [-0.06,
0.00]
0.051 -0.03 0.01 [-0.06,
0.00]
0.049*
a*b. All PA à CBF à
LNS
0.00 0.00 [-0.01,
0.00]
0.407 0.00 0.00 [-0.01,
0.00]
0.375
c. All PA à LNS -0.03 0.02 [-0.06,
0.00]
0.047* -0.03 0.02 [-0.06,
0.00]
0.035*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.01 [-0.01,
0.01]
0.523 -0.01 0.01 [-0.01,
0.01]
0.309
b. CBF à Latent Cog 0.68 0.51 [-0.33,
1.68]
0.185 0.90 0.61 [-0.30,
2.09]
0.140
c’. PA à Latent Cog -0.02 0.02 [-0.06,
0.03]
0.522 -0.01 0.02 [-0.05,
0.04]
0.675
a*b. PA à SE à Latent
Cog
0.00 0.00 [-0.01,
0.01]
0.534 0.00 0.01 [-0.01,
0.01]
0.357
c. PA à Latent Cog -0.02 0.02 [-0.06,
0.03]
0.457 -0.01 0.02 [-0.06,
0.03]
0.531

83
PA = All light and moderate physical activity; CBF = Middle Frontal Gyrus cerebral blood flow; MoCA =
Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex
Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number sequencing;
Bold rows indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Appendix Table 9: Superior Frontal Gyrus Cerebral Blood Flow as Proposed Mediator
Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.01]
0.595 0.00 0.00 [-0.01,
0.01]
0.843
b. CBFà MoCA 0.78 0.32 [0.15,
1.41]
0.015* 0.95 0.33 [0.31,
1.59]
0.004*
c’. PA à MoCA -0.01 0.01 [-0.04,
0.02]
0.431 -0.01 0.01 [-0.03,
0.02]
0.677
a*b. PA à CBF à
MoCA
0.00 0.00 [-0.01,
0.01]
0.599 0.00 0.00 [-0.01,
0.01]
0.843
c. PA à MoCA -0.01 0.01 [-0.04,
0.02]
0.514 -0.01 0.01 [-0.03,
0.02]
0.722

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.01]
0.593 0.00 0.00 [-0.01,
0.01]
0.846
b. CBFà CVLT 0.48 0.46 [-0.42,
1.39]
0.296 0.68 0.50 [-0.30,
1.65]
0.172
c’. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.874 0.00 0.01 [-0.03,
0.03]
0.907
a*b. PA à CBF à
CVLT
0.00 0.00 [0.00,
0.01]
0.641 0.00 0.00 [-0.01,
0.01]
0.848
c. PA à CVLT 0.00 0.02 [-0.03,
0.03]
0.930 0.00 0.01 [-0.03,
0.03]
0.937

Unadjusted for covariates Adjusted for covariates

84
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.01]
0.894 0.00 0.01 [-0.01,
0.01]
0.810
b. CBFà CF 0.72 0.37 [0.00,
1.44]
0.049* 0.71 0.38 [-0.04,
1.46]
0.063
c’. PA à CF -0.01 0.02 [-0.04,
0.03]
0.718 0.00 0.02 [-0.04,
0.03]
0.878
a*b. PA à CBF à CF 0.00 0.00 [-0.01,
0.01]
0.894 0.00 0.00 [-0.01,
0.01]
0.815
c. PA à CF 0.00 0.02 [-0.04,
0.04]
0.993 0.001 0.02 [-0.04,
0.04]
0.976

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.01]
0.463 0.00 0.01 [-0.01,
0.01]
0.917
b. CBFà Flanker -2.28 0.83 [-3.91, -
0.65]
0.006* -2.57 0.85 [-4.24, -
0.90]
0.003*
c’. PA à Flanker 0.13 0.05 [0.04,
0.23]
0.007* 0.12 0.05 [0.03,
0.21]
0.008*
a*b. PA à CBF à
Flanker
-0.01 0.01 [-0.02,
0.01]
0.452 0.00 0.01 [-0.02,
0.02]
0.917
c. PA à Flanker 0.13 0.05 [0.03,
0.22]
0.008* 0.12 0.04 [0.03,
0.21]
0.007*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.01]
0.631 0.00 0.01 [-0.01,
0.01]
0.711
b. CBFà LNS 0.91 0.45 [0.03,
1.78]
0.042* 0.75 0.40 [-0.04,
1.54]
0.063
c’. PA à LNS -0.03 0.01 [-0.06, -
0.01]
0.020* -0.03 0.01 [-0.06, -
0.01]
0.016*

85
a*b. All PA à CBF à
LNS
0.00 0.00 [-0.01,
0.01]
0.615 0.00 0.00 [-0.01,
0.01]
0.710
c. All PA à LNS -0.03 0.02 [-0.06,
0.00]
0.047* -0.03 0.02 [-0.06,
0.00]
0.035*

Unadjusted for covariates Adjusted for covariates
Effect
b SE
CI
(95%)
p b SE
CI
(95%)
p
a. PA à CBF 0.00 0.00 [-0.01,
0.01]
0.395 0.00 0.01 [-0.01,
0.01]
0.441
b. CBF à Latent Cog 1.08 0.49 [0.11,
2.04]
0.029* 1.11 0.52 [0.09,
2.12]
0.033*
c’. PA à Latent Cog -0.02 0.02 [-0.05,
0.01]
0.153 -0.02 0.02 [-0.05,
0.01]
0.241
a*b. PA à SE à Latent
Cog
0.00 0.01 [-0.01,
0.01]
0.425 0.00 0.01 [-0.01,
0.02]
0.461
c. PA à Latent Cog -0.02 0.02 [-0.05,
0.02]
0.297 -0.01 0.02 [-0.05,
0.02]
0.413

PA = All light and moderate physical activity; CBF = Superior Frontal Gyrus cerebral blood flow; MoCA =
Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF = Complex
Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number sequencing;
Bold rows indicate a significant relationship (p<.05); Covariates = Age, Sex, & Education

Aim 3 Supplemental Tables

Appendix Table 10: Aim 3 Model Comparisons – Light Activity Only
With AR effect Without AR effect Superior model with
Covariates
Dependent DIC R
2
DIC R
2
DIC
difference
DIC R
2

Latent Cog 13364.73 0.359 13369.1 0.192 -4.366 13360.16 0.345
MoCA 12430.825 0.325 12407.663 0.298 23.162 13150.469 0.395
CVLT 12393.917 0.329 12393.772 0.189 0.145 NA NA

86
MoCA = Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF =
Complex Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number
sequencing; CBF = Global gray matter cerebral blood flow; DIC = Deviance information criterion (models
with values at least 5 units smaller are preferable); Covariates = Age, Sex, & Education

Appendix Table 11: Aim 3 Model Comparisons – Moderate Activity Only

MoCA = Montreal cognitive assessment; CVLT = California Verbal Learning Test long delay score; CF =
Complex Figure Test long day; Flanker = Inhibitory control measured via Flanker; LNS = Letter-number
sequencing; CBF = Global gray matter cerebral blood flow; DIC = Deviance information criterion (models
with values at least 5 units smaller are preferable); Covariates = Age, Sex, & Education

CF 12416.397 0.37 12431.757 0.248 -15.36 13187.352 0.436
Flanker 12513.322 0.298 12518.212 0.241 -4.89 13347.496 0.336
LNS 11890.684 0.316 11887.255 0.292 3.429 13131.05 0.384
CBF 12133.508 0.34 12150.554 0.21 -17.046 13064.512 0.501
With AR effect Without AR effect

Superior model with
Covariates
Dependent DIC R
2
DIC R
2
DIC
difference
DIC R
2

Latent Cog 8384.444 0.264 8393.813 0.157 -9.369 8375.183 0.414
MoCA 7480.25 0.325 7464.781 0.202 15.469 8187.623 0.246
CVLT 7445.849 0.518 7460.764 0.165 -14.915 8177.578 0.553
CF 7485.96 0.276 7506.243 0.156 -20.283 8217.075 0.383
Flanker 7601.643 0.238 7602.888 0.185 -1.245 NA NA
LNS 7119.226 0.262 7120.446 0.194 -1.22 NA NA
CBF 7270.293 0.36 7286.75 0.155 -16.457 8097.241 0.592

Abstract (if available)

Abstract Due to the lack of disease-modifying treatments for dementia, elucidating ways to prevent or delay disease onset is a research priority. Physical activity has emerged as a promising intervention target, but there is still uncertainty regarding the mechanisms through which physical activity confers benefits on brain health and cognition. The current study investigated the mediating effects of sleep efficiency and global gray matter cerebral blood flow (CBF) measured via arterial spin labeling (ASL) on the relationship between physical activity and cognitive ability in a sample of older adults (N=52, mean age=66.5, 67% Female). Using mediation models, we tested whether average sleep efficiency and CBF mediated the effect of physical activity on cognitive ability. To address temporal relationships, random effects models were used to investigate the effect of daily physical activity on the subsequent night’s sleep efficiency. Dynamic structural equation models were used to investigate whether day-to-day fluctuations in physical activity were associated with cognitive ability and CBF. Contrary to our hypotheses, higher levels of physical activity were not associated with better cognitive ability, directly or via our proposed mediators. Random effects models revealed no association between daily physical activity levels and sleep efficiency. Although day-to-day patterns of physical activity were not associated with cognition or CBF, Bayesian model selection favored models that included their effect for outcomes of visual memory and CBF. These findings suggest that free-living levels of physical activity in sedentary older adults may not be associated with sleep, CBF, or cognitive ability, and that daily patterns of physical activity may provide important information when looking at brain health and cognitive outcomes.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Contributions of dynamic cerebrovascular function to cognitive decline and dementia: development and validation of a novel neuroimaging approach

PDF

Arterial stiffness and resting-state functional connectivity in older adults

PDF

Relationships between lifetime chronic stress exposure, vascular risk, cognition, and brain structure in HIV

PDF

A virtual reality exergaming system to enhance brain health in older adults at risk for Alzheimer’s disease

PDF

Indexing cerebrovascular resistance in cognitive decline and Alzheimer's disease

PDF

Deficit accumulation index as a predictor of cognitive outcomes in HIV+ and HIV− adults

PDF

Association between Blood Biomarkers of Alzheimer’s Disease and Cognition in Postmenopausal Women

PDF

The association between self-reported physical activity and cognition in elderly clinical trial participants

PDF

The protective role of progenitor cells: cell proliferation relates to cognition and white matter microstructure

PDF

Association between endogenous sex hormone levels and cognition: the Women’s Isoflavone Soy Health (WISH) trial

PDF

Cross-sectional association of blood pressure, antihypertensive medications, MRI volumetric measures and cognitive function scores in an aging population

PDF

Subclinical carotid atherosclerosis, psychosocial measures, and cognitive function in middle- to older-aged adults

PDF

Long-term blood pressure variability across the clinical and biomarker spectrum of Alzheimer’s disease

PDF

The role of blood pressure variability in cognitive decline, cerebrovascular disease and Alzheimer’s disease

PDF

Associations between inflammatory markers and change in cognitive endpoints

PDF

Associations between longitudinal loneliness, epigenetic age, and dementia risk

PDF

Psychophysiological assessment of cognitive and affective responses for prediction of performance in arousal inducing virtual environments

PDF

The role of music training on behavioral and neurophysiological indices of speech-in-noise perception: a meta-analysis and randomized-control trial

PDF

Relationship of blood pressure and antihypertensive medications to cognitive change in the BVAIT, WISH, and ELITE clinical trials

PDF

Associations between physical activities with bone mineral density in postmenopausal women

Asset Metadata

Creator Fenton, Laura (author)

Core Title Relationships between physical activity and cognitive ability in older adults: investigating the mediating effects of sleep efficiency and cerebral blood flow

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychology

Degree Conferral Date 2021-08

Publication Date 07/22/2021

Defense Date 07/19/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aging,cerebral blood flow,cognition,OAI-PMH Harvest,physical activity,Sleep

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Pa, Judy (committee chair), Beam, Christopher (committee member), Han, Duke (committee member), Thames, April (committee member)

Creator Email lefenton@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15616764

Unique identifier UC15616764

Legacy Identifier etd-FentonLaur-9821

Document Type Thesis

Format application/pdf (imt)

Rights Fenton, Laura

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu