ABSTRACT 

 

 

 

Title of Dissertation: RELATIONS BETWEEN LATENT 
EPISODIC MEMORY, NAP HABITUALITY, 
AND THE CORTEX DURING CHILDHOOD. 

  
Tamara Lynn Allard, Doctor of Philosophy, 
2023 

  

 

Thesis Directed By: 

 

Professor Tracy Riggins, 

Department of Psychology 

 

During childhood, episodic memory demonstrates marked improvements that are supported by the 

protracted development of the hippocampus and a larger network of cortical regions. To date, most 

research has focused on associations with the hippocampus in this age group. Few studies have 

explored the contribution of cortical regions and no studies have explored this longitudinally. Thus, 

the first aim of this dissertation was to examine the longitudinal co-development of cortical 

thickness and surface area in memory-related cortical regions with a latent episodic memory 

variable in 4- to 8-year-old children (N = 177). Findings, uncorrected for multiple comparisons, 

demonstrated that a thinner cortex in multiple episodic memory network regions (i.e., inferior 

frontal gyrus, inferior parietal sulcus, lingual gyrus, middle temporal gyrus, precuneus, lateral 

occipital cortex, superior frontal gyrus, superior parietal lobule, superior temporal gyrus, and 

temporal pole) at age 4 predicted more rapid improvements in memory performance from age 4 to 

6 years. Similarly, greater surface area in the precuneus and less surface area in the medial 

orbitofrontal gyrus at age 4 also predicted more rapid improvements in memory performance from 



 
 

age 4 to 6 years. Additionally, results revealed that several regions demonstrate parallel co-

development with latent episodic memory performance from age 4 to 8 years. Specifically, greater 

changes in cortical thickness and surface area of the entorhinal cortex were associated with greater 

changes in memory from age 4 to 6 years. Furthermore, cortical thickness of entorhinal cortex and 

surface area of anterior cingulate cortex, entorhinal cortex, inferior parietal sulcus, lingual gyrus, 

and superior temporal gyrus showed co-development with latent episodic memory from age 6 to 

8 years. Together, these findings suggest that cortical thickness and surface area of the episodic 

memory network support improvements in memory performance during childhood. However, 

these findings did not survive correction for multiple comparisons. Although age-related 

differences were one focus of this investigation, individual differences were another. Specifically, 

during childhood children transition away from afternoon napping. This transition has previously 

been associated with differences in memory consolidation abilities and hippocampal maturation. 

These associations suggest that habitual nappers require more regular sleep to consolidate 

memories due to an immature episodic memory network. However, limited work has examined 

these associations outside the hippocampus. Therefore, the second aim of this dissertation was to 

examine whether regions that support longitudinal memory development differ as a function of 

nap habituality (N = 44). Findings revealed significant differences in cortical thickness of right 

inferior frontal gyrus and surface area of lateral occipital cortex, such that non-nappers 

demonstrated a thinner cortex and greater surface area in these regions compared to nappers, 

though these findings did not survive correction for multiple comparisons. Thus, although there is 

some evidence that memory-related cortical regions may differ based on nap habituality, additional 

work is needed to support this claim.  Together this dissertation provides new data on the co-



 
 

development of memory with brain structure in the episodic memory network and identifies 

individual differences that may be associated with these brain structures. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



 
 

 

 
 
 
 
 
 
RELATIONS BETWEEN LATENT EPISODIC MEMORY, NAP HABITUALITY, AND THE 

CORTEX DURING CHILDHOOD. 
 
 
 

by 
 

 
 
 
 

 
Dissertation submitted to the Faculty of the Graduate School of the 

University of Maryland, College Park, in partial fulfillment 
of the requirements for the degree of 

Doctor of Philosophy 
2023 

 

 
 
 
 
 
 
 
 
Advisory Committee: 
 
Professor Tracy Riggins, Ph.D., Chair 
Associate Professor Elizabeth Redcay, Ph.D. 
Assistant Professor Rachel Romeo, Ph.D. 
Professor Rebecca Spencer, Ph.D. 
Associate Professor Donald Bolger, Ph.D., Dean’s Representative 
 

 
Tamara Lynn Allard 



 
 

 

 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

© Copyright by 
Tamara Lynn Allard 

2023 
 
 
 
 
 
 
 
 
 
 



ii 
 

Dedication 

 I dedicate this to my family whose steadfast support of my educational aspirations has 

made this dissertation a reality. To my husband, Ethan, thank you for your unwavering 

encouragement, patience, and belief in my abilities. Your support has sustained me through this 

degree. You are my rock and I love you. To my parents Bernard Allard and Ava Grajeda-Allard, 

because of you I am a critical thinker and a hard worker, qualities required for a Ph.D. Without 

you, none of this would have been possible. I know that you walked so I could run. Thank you. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



iii 
 

Acknowledgments 

There are many people who made my dissertation a possibility. It is my honor to recognize those 

individuals here. First and foremost, I would like to express my heartfelt gratitude to my advisor 

and committee chair, Dr. Tracy Riggins, who is the perfect example of what a graduate advisor 

should be. Tracy, thank you for consistently being the first to “put on your mentor hat” and for 

supporting my career ambitions. Furthermore, I would like to thank my dissertation committee, 

including Dr. Rebecca Spencer, Dr. Elizabeth Redcay, Dr. Rachel Romeo, and Dr. DJ Bolger, for 

their invaluable feedback and advice. Thank you for pushing me to be a better scientist. I would 

also like to thank Dr Greg Hancock whose patience, mentorship, and unique teaching abilities 

allowed me to implement the advanced analyses I used in this dissertation. Notably, this 

dissertation would not have been possible without the dedicated members (past and present) of 

the Neurocognitive Development Laboratory and the Somneuro Laboratory. I am especially 

grateful to Sanna Lokhandwala, Arcadia Ewell, Jade Dunstan, Brooke Kohn, Yuqing Lei, Dr. 

Christine St Laurent, and Dr. Morgan Botdorf. Also, I would like to give special recognition to 

Dr. Kelsey Canada whose mentorship and advice were instrumental for this dissertation. 

Moreover, this work was dependent on several grants. Thus, I would like to acknowledge the 

National Institute of Health, the National Science Foundation, and the Janet W. Johnson 

Foundation whose funding made this investigation a reality. I also want to recognize Archilline 

Tablada whose organization and expert administrative knowledge made graduating possible. 

Next, I would like to acknowledge the developmental area faculty and students who provided me 

with a supportive community that I could rely on. The support system I developed in this 

program was instrumental to my success. Additionally, I want to communicate my great 

appreciation for my family. Thank you so much to my parents, my husband, and of course my 



iv 
 

loving cat, Henry for supporting me through the highs and lows of graduate school. Lastly, I 

want to thank all the participants and their families for contributing the data that was used in this 

dissertation. Without your willingness to participate, none of this would have been possible. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



v 
 

Table of Contents 
Dedication ..................................................................................................................... ii 
Acknowledgments........................................................................................................ iii 
Table of Contents .......................................................................................................... v 
List of Tables ............................................................................................................. viii 
List of Figures ............................................................................................................... x 
List of Abbreviations ................................................................................................... xi 
Chapter 1: Introduction ................................................................................................. 1 

Memory Development................................................................................................................. 3 
Brain Development ..................................................................................................................... 5 
Sleep Development ..................................................................................................................... 7 
Associations Between Memory and the Brain ............................................................................ 8 
Associations Between Memory and Sleep ................................................................................ 12 
Associations Between the Brain and Sleep ............................................................................... 14 
Study Overview ......................................................................................................................... 16 

Chapter 2: Aims and Hypotheses ................................................................................ 20 
Aim 1 ......................................................................................................................................... 20 
Aim 2 ......................................................................................................................................... 21 

Chapter 3: Methods ..................................................................................................... 23 
Study 1....................................................................................................................................... 23 

Behavioral Tasks. .................................................................................................................. 26 
MRI. ....................................................................................................................................... 32 

Study 2....................................................................................................................................... 34 
Participants. ........................................................................................................................... 34 
Nap Habitually. ...................................................................................................................... 36 
24-hour Sleep Duration ......................................................................................................... 37 
MRI. ....................................................................................................................................... 37 

Chapter 4: Data Analysis ............................................................................................ 38 
Aim 1 ......................................................................................................................................... 38 

Selection of ROIs................................................................................................................... 38 
A Latent Measure of Memory Development. ........................................................................ 38 
Power Analysis. ..................................................................................................................... 39 
Outliers. ................................................................................................................................. 40 
Cross-sectional ROI Analysis. ............................................................................................... 41 
Cross-sectional Whole-brain Analysis. ................................................................................. 41 
Latent Growth Curve Modelling. .......................................................................................... 42 



vi 
 

Aim 2 ......................................................................................................................................... 49 
Preliminary Analysis. ............................................................................................................ 49 
ROI Analysis ......................................................................................................................... 49 
Vertex by Vertex Analysis. ................................................................................................... 50 

Chapter 5: Results ....................................................................................................... 51 
Aim 1: Preliminary Analysis ..................................................................................................... 51 
Aim 1 (Part 1): Cross-Sectional Associations between Memory and ROIs ............................. 52 

ROI Analysis. ........................................................................................................................ 52 
Whole Brain Analysis. ........................................................................................................... 56 

Aim 1 (Part 2): Memory Development ..................................................................................... 56 
Aim 1 (Part 2): Cortical Development ...................................................................................... 57 

ROI Development Fit Indices. ............................................................................................... 57 
ROI Intercept and Slope Variances. ...................................................................................... 58 
Cortical Thickness Development. .......................................................................................... 58 
Surface Area Development. ................................................................................................... 62 
Hemispheric Differences in Cortical Development. .............................................................. 67 
Associations between development of cortical thickness and surface area. .......................... 68 
Total Gray Matter Volume Development. ............................................................................. 70 

Aim 1 (Part 3): Co-development of Memory and the Cortex ................................................... 70 
Associations between the ROI Intercept and the Memory Intercept. .................................... 71 
Associations between the ROI Intercept and the Slope of Memory Development. .............. 72 
Associations between the Memory Intercept and the Slope of ROI Development. .............. 73 
Associations between the Slope of ROI Development and the Slope of Memory 
Development. ......................................................................................................................... 74 
Total Gray Matter Volume. ................................................................................................... 83 

Aim 2: Preliminary Findings ..................................................................................................... 83 
Aim 2: Associations between Nap Habituality and the Cortex ................................................. 84 

Selection of ROIs................................................................................................................... 84 
Associations between Cortical Thickness of Selected ROIs and Nap Habituality................ 85 
Associations between Surface Area of Selected ROIs and Nap Habituality. ........................ 87 
Whole Brain Associations with Nap Habituality. ................................................................. 88 
Specificity of ROI Findings. .................................................................................................. 88 

Chapter 6: Discussion ................................................................................................. 90 
Cross-Sectional Associations .................................................................................................... 91 
Memory Development............................................................................................................... 97 



vii 
 

Co-development of Latent Episodic Memory with the Episodic Memory Network ................ 97 
Associations between the Slope of ROIs with the Slope of Latent Episodic Memory. ........ 97 
Associations Between the ROI Intercept and the Slope of Latent Episodic Memory. ........ 100 
Associations Between the Latent Episodic Memory Intercept and the Slope of ROI. ........ 103 
What do these associations mean for memory? ................................................................... 105 

Specificity of Regional Findings ............................................................................................. 107 
Differences in Cortical Thickness and Surface Area Based on Nap Habituality .................... 107 
Limitations .............................................................................................................................. 108 
Future directions ...................................................................................................................... 109 
Conclusions ............................................................................................................................. 110 

Appendix A: Supplementary Table(s) ...................................................................... 112 
Appendix B: Literature Review ................................................................................ 113 
Bibliography ............................................................................................................. 153 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 



viii 
 

List of Tables 

Table 1. Peer-reviewed Articles that Examine Associations Between Memory and Cortical ROIs. 

Table 2. Summary of Aims and Hypotheses. 

Table 3. Available MRI Sample and Average Age of Participants in Each Age Block for Study 1 

Table 4. Number of Children Contributing Behavioral and Neuroimaging Data for Study 2. 

Table 5. Cross-Sectional Associations Between Performance on the Composite Memory 

Variable and Selected ROIs. 

Table 6. Growth Parameters for the Latent Episodic Memory Variable by Cohort. 

Table 7. Fit Indices for Cortical Thickness Growth Models. 

Table 8. Fit Indices for Surface Area Growth Models. 

Table 9. Growth Parameters for Cortical Thickness in The Left Hemisphere. 

Table 10. Growth Parameters for Cortical Thickness in The Right Hemisphere. 

Table 11. Growth Parameters for Surface Area in The Left Hemisphere. 

Table 12. Growth Parameters for Surface Area in the Right Hemisphere. 

Table 13. Fit Indices for Bivariate Growth Curve Models Examining Associations Between ROI 

Thickness and the Latent Episodic Memory Variable. 

Table 14. Associations Between Memory and Bilateral ROI Thickness Parameters in the 

Younger Cohort. 

Table 15. Associations Between Memory and Bilateral ROI Thickness Parameters in the Older 

Cohort. 

Table 16. Associations Between Memory and Bilateral ROI Surface Area Growth Parameters in 

the Younger Cohort. 

Table 17. Associations Between Memory and Bilateral ROI Surface Area Growth Parameters in 



ix 
 

the Older Cohort. 

Table 18. Fit Indices for Lateralized Bivariate Growth Curve Models Examining Associations 

Between ROI Thickness and the Latent Episodic Memory Variable 

Table 19. Fit Indices for Lateralized Bivariate Growth Curve Models Examining Associations 

Between ROI Surface Area and the Latent Episodic Memory Variable. 

Table 20. Associations of Memory with Left ROI Thickness Parameters in the Younger Cohort. 

Table 21. Associations of Memory with Left ROI Thickness Parameters in the Older Cohort. 

Table 22. Associations of Memory with Right ROI Thickness Parameters in the Younger Cohort. 

Table 23. Associations of Memory with Right ROI Thickness Parameters in the Older Cohort. 

Table 24. Associations of Memory with Left ROI Surface Area Parameters in the Younger 

Cohort. 

Table 25. Associations of Memory with Left ROI Surface Area Parameters in the Older Cohort. 

Table 26. Associations of Memory with Right ROI Surface Area Parameters in the Younger 

Cohort. 

Table 27. Associations of Memory with Right ROI Surface Area Parameters in the Older Cohort. 

Table 28. Associations of Memory with Total Gray Matter Volume Parameters. 

Table 29. Differences in Cortical Thickness of Selected ROIs Based on Nap Status. 

Table 30. Differences in Surface Area of Selected ROIs Based on Nap Status. 

Table 31. Hemispheric Differences in Cortical Thickness and Surface Area of Selected ROIs. 

 

 



x 
 

List of Figures 

Figure 1. Schematic depiction for the timing of data collection in Study 1. 

Figure 2. Schematic depiction for the timing of data collection in Study 2. 

Figure 3. Conceptual Depiction of the Latent Growth Models used to Analyze the Development 

of ROIs 

Figure 4. Conceptual Diagram of Linear Second-order Latent Growth Curve Models for Episodic 

Memory in the 4-year-old Cohort. 

Figure 5. Conceptual Diagram of Bivariate Growth Curve Model Assessing Co-development of 

Episodic Memory and Selected ROIs. 

Figure 6. Associations Between Cortical Thickness and the Composite Memory Variable. 

Figure 7. Associations Between Surface Area and the Composite Memory Variable. 

Figure 8. Developmental Trajectory of Latent Episodic Memory Performance from 4 to 8 years. 

Figure 9. Development of Cortical Thickness and Surface Area by Hemisphere. 

Figure 10. Examples of Developmental Trajectories by Subject in Two Selected Cortical 

Regions. 

Figure 11. Associations Between the Slopes of Cortical Thickness and Surface Area 

Figure 12. Differences in Inferior Frontal Gyrus Based on Nap Status. 

Figure 13. Differences in Lateral Occipital Cortex Based on Nap Status. 

 



xi 
 

List of Abbreviations 

ACC = Anterior Cingulate Cortex 

AG = Angular Gyrus 

CA1 = Cornu Ammonis 1 

DG/CA2-4 = Dentate Gyrus/Cornu Ammonis 2-4 

dlPFC = Dorsolateral Prefrontal Cortex 

DPC = Dorsal Parietal Cortex 

DMN = Default Mode Network 

ERC = Entorhinal Cortex 

ICV = Intracranial Volume 

IFG = Inferior Frontal Gyrus 

IPS = Inferior Parietal Sulcus 

LOC = Lateral Occipital Cortex 

MTL = Medial Temporal Lob 

PCC = Posterior Cingulate Cortex 

PFC = Prefrontal Cortex 

PHG = Parahippocampal Gyrus 

pIPS = Posterior Intraparietal Sulcus 

PRC = Perirhinal Cortex 

SPL = Superior Parietal Lobule 

ROI = Region of Interest 

vlPFC = Ventrolateral Prefrontal Cortex 

VPC = Ventral Parietal Cortex 



1 
 

Chapter 1: Introduction 

Fifty years ago, Tulving (1972) suggested that the ability to remember past events is a 

distinct form of declarative memory. Specifically, he noted that episodic memory is different from 

semantic memory due to its autonoetic quality that requires a person to perform mental time travel 

(Tulving, 1984, 2002). Today, most groups agree that episodic memories are specific and detailed 

memories for past events, that include information about event content (i.e., what), location (i.e., 

where), and timing (i.e., when). 

Across childhood and into early adulthood, episodic memory demonstrates marked 

improvements  (Canada et al., 2021, 2022; Lee et al., 2016; Picard et al., 2012; Riggins, 2014; Yim 

et al., 2013). These improvements are supported by the protracted development of the 

hippocampus (Botdorf et al., 2022; Canada et al., 2021; Lee et al., 2020; Riggins et al., 2018), an 

area strongly implicated in memory (Scoville & Milner, 1957). However, work in adults has 

demonstrated that episodic memory is also supported by a broad network of cortical regions 

including the MTL, PFC, and parietal cortex (e.g., Brewer et al., 1998; Buckner et al., 1999; 

Davachi et al., 2003; Lavenex & Amaral, 2000; Nyberg et al., 2000; Stern et al., 1996; Stevens et 

al., 2008; Wagner et al., 1998). Like the hippocampus, these regions undergo protracted structural 

development during childhood (Frangou et al., 2022; Gogtay et al., 2004) suggesting they may 

also play a role in the development of episodic memory abilities.  

Although limited, some work in developmental populations has supported this hypothesis 

by demonstrating cross-sectional relations between memory performance and the cortex during 

childhood (see Ghetti & Bunge, 2012; Østby et al., 2012; Schommartz et al., 2023). These studies 

show that, for most regions, cortical thickness is negatively associated with episodic memory 



2 
 

performance across early to late childhood (e.g., ages 4 to 12 years). However, no study has 

examined these associations longitudinally. This is problematic because cross-sectional studies 

cannot provide information about change over time. In other words, studies that only assess one 

time point cannot make claims about whether changes in the brain support changes in memory. 

Furthermore, past work has demonstrated that cross-sectional and longitudinal studies sometimes 

result in different outcomes, with cross-sectional studies often underestimating relations between 

the brain and memory during development (Keresztes et al., 2022). Thus, the first aim of this study 

is to examine longitudinal co-development of memory performance and structural measures of 

cortical ROIs that support memory. 

During early childhood children transition from biphasic sleep to monophasic sleep (i.e., 

cessation of the afternoon nap; Staton et al., 2020). These developmental shifts in memory, nap 

habits, and brain development are likely related. For example, some have theorized that habitual 

nappers may require more regular sleep due to an inefficiency in sleep-dependent memory 

consolidation caused by an immature episodic memory network (Lokhandwala & Spencer, 2022; 

Mason et al., 2021; Spencer & Riggins, 2022). Evidence in early childhood supports this 

hypothesis. Namely, habitual nappers do worse on an episodic memory task following a wake 

period compared to their non-napping counterparts and they demonstrate less hippocampal 

maturity (Kurdziel et al., 2013; Riggins & Spencer, 2020). Yet, no work has examined whether 

these findings generalize to other memory-related brain regions. Thus, the second aim of this 

dissertation is to examine associations between nap habituality and cortical regions that support 

episodic memory. 

Exploring longitudinal brain development and how it relates to changes in memory and 

sleep requires knowledge of multiple literatures. Below I review literature examining the 



3 
 

developmental trajectory of episodic memory performance, structural brain maturation, and sleep 

habits across childhood (e.g., age 3 to 12 years). Investigations of infancy, toddlerhood, and 

adolescence are beyond the scope of this dissertation (see instead Galván, 2020; Johnson et al., 

2020; Kopasz et al., 2010; Mason et al., 2021). Next, I will examine associations between 

memory and structural measures of brain maturation, including cortical thickness and surface 

area. These measures are thought to reflect synaptogenesis, gyrification, and myelination in 

cortical grey matter (Cafiero et al., 2019; Huttenlocher, 1979; Rakic, 2009). Therefore, they are 

proxies for neural development that have been regionally associated with memory performance 

during early to middle childhood (e.g., Schommartz et al., 2023). Notably, this review will not 

directly focus on measures of brain function because brain structure and function develop on 

distinct time scales for most brain regions (Gilmore et al., 2015). Finally, I will review relations 

between nap habitually, memory, and the brain. There is a burgeoning body of work examining 

the association between memory development, the brain, and sleep physiology (e.g., sleep 

architecture and microstructure). However, these measures are also beyond the scope of this 

dissertation and will not be covered (see instead Lokhandwala & Spencer, 2022). I will conclude 

by describing the study design and aims of this dissertation. 

Memory Development 

Episodic memory, unlike other forms of declarative memory, is thought to be a recent 

evolutionary development. As a result, it is comparatively late developing, early deteriorating, and 

more sensitive to neural deficits (Tulving, 2002). Longitudinal research from the past decade 

supports this theory, showing protracted development from early to middle childhood (Allard et 

al., 2023; Canada, Pathman, et al., 2020; Canada et al., 2022; Lee et al., 2016; Riggins, 2014). For 

example, previous studies have demonstrated improvement in memory abilities for feature binding 



4 
 

(e.g., source memory; Allard et al., 2023; Cheke & Clayton, 2015; Drummey & Newcombe, 2002; 

Lee et al., 2016; Lorsbach & Reimer, 2005; Riggins, 2014; Sluzenski et al., 2006), fine-grained 

details (i.e., pattern separation; (Canada et al., 2019; Ngo et al., 2017), and temporal order of events 

across early to late childhood (Canada, Pathman, et al., 2020).  

However, not all episodic memory abilities develop at the same rate. Some demonstrate 

linear trajectories (Canada, Pathman, et al., 2020), while others show non-linear trajectories 

((Allard et al., 2023; Riggins, 2014). Further, different developmental trajectories can be observed 

within the same task depending on the variable assessed (Allard et al., 2023; Lee et al., 2016). For 

example, one study found that hit rates (i.e., the ability to accurately identify a previous connection 

as old) on a feature binding task are relatively early developing, while false alarm rates (i.e., the 

inability to recognize a novel connection as new) are relatively late developing (Allard et al., 2023). 

This dynamic could suggest that these memory tasks reflect development of additional cognitive 

abilities. For example, late development of false alarms in Allard (2023) could suggest changes in 

process that support encoding, like attention.  This reveals a key limitation of episodic memory 

research, that no single task is process pure. However, latent measures of episodic memory may 

be a potential solution. Specifically, using structural equation modeling (SEM), researchers can 

create a construct-level measure of episodic memory (i.e., a latent variable) by capturing the shared 

variance from a variety of tasks assessing varying memory abilities. Theoretically, this shared 

variance between tasks represents an episodic memory signal, while the associated error represents 

noise from uncommon sources that may be attributed to other cognitive abilities. The resulting 

latent variable is likely a better measure of actual episodic memory performance. 

In children, a handful of studies have successfully utilized latent techniques to assess 

associations between latent episodic memory and age (Canada et al., 2022; Cheke & Clayton, 



5 
 

2015). These studies have demonstrated that a battery of episodic memory tasks featuring surface 

differences can be used to create a latent measure of episodic memory in children as young as age 

4 years (Canada et al., 2022; Cheke & Clayton, 2015) and that the episodic memory construct 

remains consistent across childhood (Canada et al., 2022). In other words, previous work has 

demonstrated that a latent episodic memory variable can be created that assesses the same memory 

construct at ages 4, 5, 6, 7, and 8 years. Further, results from a longitudinal study show that these 

latent measures of episodic memory do undergo linear age-related increases from early to middle 

childhood (Canada et al., 2022). However, associations between latent measures of memory and 

brain development during childhood remain unclear. 

Brain Development 

Research in adults demonstrates that several cortical regions support memory performance 

across the lifespan (e.g., Brewer et al., 1998; Buckner et al., 1999; Davachi et al., 2003; Lavenex 

& Amaral, 2000; Nyberg et al., 2000; Stern et al., 1996; Stevens et al., 2008; Wagner et al., 1998). 

These included several subregions in the medial temporal cortex (i.e., parahippocampal gyrus, 

entorhinal cortex, and perirhinal cortex), prefrontal (i.e., IFG, medial orbitofrontal, superior 

frontal, rostral middle frontal gyrus, and ACC), parietal (i.e., precuneus, posterior parietal, and 

inferior parietal sulcus), and occipital cortices (lateral occipital gyrus and lingual gyrus). 

Moreover, all of these regions undergo some form of change across childhood, with varying 

trajectories (Ducharme et al., 2015; Frangou et al., 2022; Gogtay et al., 2004; Lenroot et al., 2007; 

Mills et al., 2014; Raznahan et al., 2011; Tamnes et al., 2017; Wang et al., 2019; Wierenga et al., 

2014). 



6 
 

Global gray matter demonstrates an overall increase from age 4 to the onset of puberty, 

then gradually thins into adulthood (Gogtay et al., 2004; Lenroot et al., 2007). For example, 

Lenroot and colleagues examined total gray matter volume in participants aged 3 to 27 years and 

found that volumes peaked at age 8.5 years (Lenroot et al.,2007). Gray matter volume can be 

accessed via two different measures, cortical surface area, and cortical thickness. These two 

measures are thought to reflect different aspects of brain development. Specifically, early increases 

in cortical thickness are thought to reflect synaptogenesis, whereas later cortical thinning is thought 

to reflect synaptic pruning (Huttenlocher, 1979). In contrast, increases in cortical surface area 

reflect folding and gyrification due to the division of neural stem cells and intra-cortical 

myelination, whereas later decreases in cortical surface area reflect decreases in folding and 

gyrification due to apoptosis (Cafiero et al., 2019; Rakic, 2009).  

Given cortical thickness and surface area measure separate processes, they do not co-vary 

(Cafiero et al., 2019; Im et al., 2008) and they demonstrate unique developmental trajectories 

(Hutton et al., 2009; Li et al., 2013; Lyall et al., 2015; Wierenga et al., 2014). For example, in 

Carfiero and colleagues (2019), the authors demonstrated that age-related effects on cortical 

thickness and surface area were distinct in all but 2 cortical regions (i.e., lingual gyrus and calcarine 

sulcus) when examining a sample of 5- to 7-year-old children. 

While trajectories for cortical thickness and cortical surface are distinct, both demonstrate 

non-linear change from birth to puberty. For example, cortical surface area increases across early 

to middle childhood, peaks between 8 and 10 years, and then decreases into adulthood (Ducharme 

et al., 2015; Mills et al., 2014; Raznahan et al., 2011; Wierenga et al., 2014). Similarly, early 

increases in cortical thickness eventually give way to later decreases in late childhood and 

adolescence (Lenroot et al., 2007; Raznahan et al., 2011). However, it is debated when cortical 



7 
 

thickness reaches its peaks (Tamnes et al., 2017; Walhovd et al., 2017). Specifically, some suggest 

that cortical thickness peaks in late childhood (e.g., 8 to 10 years; Lenroot et al., 2007; Raznahan 

et al., 2011; Shaw et al., 2007) while others suggest that thinning has already begun in early 

childhood (i.e., prior to age 5 years; Ducharme et al., 2016; Gogtay et al., 2004; Mills et al., 2014; 

Wierenga et al., 2014). In spite of this conflicting evidence, several studies have demonstrated that 

cortical thickness peaks before to surface area (Raznahan et al., 2011; Wierenga et al., 2014).  

Importantly, development changes in cortical thickness and surface are not uniform, rather 

sub-regions demonstrate maturation at distinct rates (Gogtay et al., 2004; Raznahan et al., 2011; 

Tamnes et al., 2017; Wierenga et al., 2014). For example, cortical thinning begins in the dorsal 

parietal cortices and then spreads rostrally, caudally, and laterally over the frontal, occipital, and 

temporal cortices (Gogtay et al., 2004). Importantly, a meta-analysis including over 17,000 

subjects aged 3 to 90 years demonstrated that all cortical regions appear to exhibit the greatest 

thickness during childhood (e.g., 3 to 10 years) except for the entorhinal cortex, the temporopolar 

cortex, and the anterior cingulate cortices which peak in middle adulthood (Frangou et al., 2022). 

In contrast, most regions show peak cortical surface area values between ages 8 and 13 years 

(Wierenga et al., 2014). 

Sleep Development 

During childhood, sleep undergoes significant changes. Specifically, the average amount 

of sleep children receive over a 24-hour period changes with age (Bathory & Tomopoulos, 2017; 

Galland et al., 2012; Iglowstein et al., 2003; Matricciani et al., 2012). According to a meta-analysis 

conducted by Galland et al., (2012), infants sleep an average of 12.7 hours per day. This decreases 

to 11.5 hours per day for preschool-aged children (aged 4 to 5 years; Bathory & Tomopoulos, 



8 
 

2017). Past age 5 years, overall 24 hour sleep time continues to decreases by approximately 5.9 

minutes each year until age 12 years (Galland et al., 2012). 

From infancy through early childhood, children also take progressively fewer naps and the 

duration of naps decreases (Iglowstein et al., 2003; Kurth et al., 2016; Ohayon et al., 2004; 

Weissbluth, 1995). For example, infants take upwards of 4 naps per day lasting an average of 3 

hours per nap. In contrast, children over 5 years take only one nap per day lasting an average of 1 

hour per nap (Galland et al., 2012; Iglowstein et al., 2003; Staton et al., 2020; Weissbluth, 1995). 

Eventually, these trends lead to nap cessation sometime in early childhood. Specifically, whereas 

97% of children under 2 years nap at least one day per week, by age 3 years, 67% of children nap 

and by age 6 years, 6% of children nap (Staton et al., 2020).  

Associations Between Memory and the Brain 

Extensive research in adults has demonstrated that the hippocampus plays a critical role in 

supporting the formation and consolidation of episodic memories (Davachi et al., 2003; 

Eichenbaum, 2004; Lavenex & Banta Lavenex, 2013; Scoville & Milner, 1957). In children, cross-

sectional studies have demonstrated that structural measures of the hippocampus are related to 

episodic memory during development (see Botdorf et al., 2022 for review). Moreover, a handful 

of longitudinal studies have demonstrated that volumetric changes in the hippocampus across 

childhood are associated with improvements in episodic memory performance (Canada et al., 

2021; Lee et al., 2020). Together these findings suggest that hippocampal development plays a 

role in memory improvement across childhood. Research in adults also demonstrates that episodic 

memory abilities are supported by a broad network of neocortical regions outside the 

hippocampus. These include several subregions in medial temporal (i.e., parahippocampal gyrus, 



9 
 

entorhinal cortex, and perirhinal cortex), prefrontal (i.e., anterior cingulate cortex, orbitofrontal 

cortex, inferior frontal gyrus, superior frontal gyrus, and rostral middle frontal gyrus), parietal (i.e., 

precuneus, posterior parietal, and inferior parietal sulcus), and occipital cortices (lateral occipital 

and lingual gyrus). Importantly, there is both theoretical and empirical evidence to support the role 

of these regions in episodic memory formation and retrieval. 

For example, in the MTL, the hippocampus receives and binds sensory information from 

two distinct neural pathways (i.e., what and where) that are thought to provide distinct event-

related details (Eichenbaum et al., 2012; Wixted & Squire, 2011). The “what” pathway carries 

item-related information from visual regions through the perirhinal cortex (PRC) and the lateral 

entorhinal cortex (ERC) to the hippocampus (Eichenbaum et al., 2012). In contrast, the “where” 

pathway carries contextual information from the parietal cortex and retrosplenial cortex through 

the parahippocampal gyrus (PHG) and medial ERC to the hippocampus (Eichenbaum et al., 2012). 

Therefore, it follows that PHG, ERC, and PRC would be associated with memory performance. 

In the PFC and PPC there are two hypothesized roles for subregions in memory 

performance based on functional connections to larger networks. Specifically, subregions that 

align with the Default Mode Network, including bilateral medial orbital frontal, medial frontal 

pole, and anterior cingulate cortex (ACC), angular gyrus (AG), posterior cingulate cortex (PCC), 

and precuneus support memory recall (Amlien et al., 2018; Miotto et al., 2020; Yu, Daugherty, et 

al., 2018). Whereas other subregions in the PFC and PPC that align with control and attention 

networks, including dorsolateral PFC (dlPFC) in the rostral middle frontal gyrus, ventrolateral 

PFC (vlPFC) in the inferior frontal gyrus (IFG), inferior parietal sulcus (IPS), and superior parietal 

lobule (SPL) support improved attention during encoding (Miotto et al., 2020; Tang et al., 2018; 

Wendelken et al., 2011). 



10 
 

Finally, in the occipital cortex, associations with memory are thought to reflect the 

reactivation of encoded information during recall due to the consolidation of memories from the 

hippocampus to the neocortex (e.g., Alvarez & Squire, 1994; Norman & O’Reilly, 2003). This 

theory is often called the Neurobiological Model of Memory. There is unique functional evidence 

in the occipital cortex that supports this theory in two subregions, the lingual gyrus and the lateral 

occipital cortex (LOC; Karanian & Slotnick, 2015; Rosen et al., 2018; Wing et al., 2015). The 

nature of Neurobiological Model of Memory theory means that findings are likely task specific 

and associated with the given role of each occipital region. 

In children, associations between memory and the larger episodic memory network are still 

an emerging area, with most structural studies surfacing in the last decade (see Table 1). Of these, 

most agree that cortical thickness of aforementioned subregions in the MTL, PFC, parietal, and 

occipital cortices is negatively associated episodic memory performance during early to late 

childhood. In other words, a thinner cortex in these ROIs is related to better memory performance. 

Additionally, one study found that thinning of subregions in the PFC mediates associations 

between memory performance and age from late childhood into adulthood suggesting that cortical 

thinning accounts for age-related differences in memory performance (Klijn et al., 2016). In 

contrast, little to no research has examined these dynamics in cortical surface area. One notable 

exception found positive associations between memory performance and cortical surface area in 

anterior right middle frontal gyrus and inferior frontal gyrus; in other words, greater surface area 

was related to better memory performance (Lyle et al., 2017). However, the study focused 

primarily on adolescents. In combination, these findings suggest that integrity of these cortical 

regions is related to individual differences in episodic memory performance. 

 



11 
 

Table 1 

Peer-reviewed Articles that Examine Associations Between Memory and Cortical ROIs. 

Study N Mean 
Age 
(years) 

Age 
Range 
(years) 

ROI(s) Memory Assessment(s) 

Amlien et al 
(2018) 

270 19.4 6 to 80 Isthmus Cingulate 
Lingual Gyrus 
Posterior Cingulate Cortex 
Precuneus 

Source Memory 

Bauer et al. 
(2019) 

66 7.34 5 to 8 ACC 
mPFC 
Hippocampus 

Self-Derivation through Integration 
(Stem Facts–Open Ended)  
Self-Derivation through Integration 
(Stem Facts–Total)  
Self-Derivation through Integration 
(Integration Facts–Open Ended)  
Self-Derivation through Integration 
(Integration Facts–Total) 

Chad-
Friedman et 
al (2021) 

63 4.23 & 
7.19 

4 to 7 SPL CMS 
Source Memory 

Fjell et al 
(2019) 

650 25.8 4 – 88 Hippocampus 
lPFC 

CVLT-C  
Rey-Osterrieth Complex Figure test 

Klijin et al., 
2016 

90 11.0 
(child 
sample) 

10 to 12, 
18, & 25 
to 32 

IFG Verbal Memory Task 

Guillery-
Girard et al 
(2013) 

30 11.31 6 to 23 Anterior Middle Temporal 
Gyrus 
dlPFC 
Hippocampus 
Supeior Temporal Cortex 
vlPFC 

What-Where-When Paradigm 

Keretesztes 
et al., (2017) 

70 9.8 6 to 14 ERC 
Hippocampus 

Source Memory 
MST 

Ostby et al 
(2012) 

107 13.9 8 to 19 Hippocampus 
mOFC 

Rey-Osterrieth Complex Figure test 

Schommartz 
et al., (2023) 

63 6.37 5 to 7  Hippocampus 
ERC 
IFG 
Inferior Parietal Sulcus 
LOC 
Lateral Orbitofrontal Cortex 
Medial Orbitofrontal Cortex 
Precuneus 
Rostral Middle Frontal 
Cortex 
SPL 

Object-Location Association Task 

Sowell et al 
(2001) 

35 N/A 7 to 16 Frontal Cortex 
MTL 

CVLT-C 
Rey-Osterrieth Complex Figure test 

Squeglia et 
al (2013) 

185 N/A 12 to 14 IPS 
SPL 

CVLT-C 

Yu et al 
(2018) 

120 13.6 5 to 25 dlPFC (Rostral Middle 
Frontal Cortex) 

CVLT-C 



12 
 

However, these previous studies have focused on one memory tasks, which, as described 

above also includes non-memory related attention and cognitive processes. No study to date has 

explored relations between brain structure and a latent variable of episodic memory. Thus, it 

remains unclear whether these findings are attributable to the development of other cognitive 

functions. Furthermore, there is a lack of research examining longitudinal co-development of 

episodic memory and these regions. Therefore, the first aim of this dissertation was to examine 

the longitudinal co-development of extrahippocampal regions that support memory and a latent 

variable of memory during early to middle childhood. Specifically, I hypothesize that 

longitudinal changes in cortical thickness/surface area will be associated with longitudinal 

changes in memory performance. 

Associations Between Memory and Sleep 

In early childhood, evidence demonstrates that an afternoon nap has a positive impact on 

memory abilities including emotion memory (Kurdziel et al., 2018), temporal order memory 

(Lokhandwala & Spencer, 2021); verbal memory (Esterline & Gómez, 2021; Giganti et al., 

2014; Spanò et al., 2018; H. Wang et al., 2022; Williams & Horst, 2014) and memory 

generalization (Sandoval et al., 2017; H. Wang et al., 2022). Furthermore, several studies have 

found that the memory benefit generated by an afternoon nap persists after overnight sleep 

(Kurdziel et al., 2013; Kurdziel et al., 2018; Lokhandwala & Spencer, 2021; Sandoval et al., 

2017; Spanò et al., 2018; Williams & Horst, 2014) with some effects persisting 1 week later 

(Williams & Horst, 2014).  

Some work has even suggested that napping and overnight sleep interact to produce a 

positive memory effect (Kurdziel et al., 2018). For example, Kurdziel and colleagues (2018), 



13 
 

tested whether an afternoon nap or an overnight sleep bout had an impact on memory 

performance across a 24-hour period. In the study, there were two conditions: the first was a nap 

condition and the second was a wake condition lasting the same duration. They found that there 

were memory deficits due to a missed afternoon nap, but only after an overnight sleep bout. 

Thus, the effects of an afternoon nap and overnight sleep have an interactive effect on memory 

performance. 

In addition, there is evidence that the positive effect of an afternoon nap are not specific 

to early childhood. Specifically, a meta-analysis of 54 studies examining relations between 

cognition and afternoon naps in samples ranging from early childhood to mid-adulthood found 

that an afternoon nap has a positive effect on declarative memory and that these effects are not 

moderated by age (Leong et al., 2022). In short, an afternoon nap is good for everyone, 

regardless of age. However, work examining nap habituality demonstrates that the effects of 

missing an afternoon nap are not similar across age.  

Past work looking at nap habituality in early childhood demonstrates that habitual 

nappers who miss their afternoon nap display more extreme memory decay on episodic memory 

tasks than non-nappers (Esterline & Gómez, 2021; Kurdziel et al., 2013; Kurdziel et al., 2018). 

Additionally, these memory differences are still present 24 hours later (Kurdziel et al., 2013; 

Kurdziel et al., 2018). For example, Kurdziel and colleagues (2013) examined memory 

performance across nap and wake session with both habitual napper and non-nappers. They 

found that habitual nappers, but not non-nappers, took a significant hit to their memory following 

the wake session compared to the nap session. Further, memories that were lost during the 

afternoon wake session were not recovered during overnight sleep. These findings suggest that 

an afternoon nap is critical for memory consolidation in habitual nappers. 



14 
 

Associations Between the Brain and Sleep 

During sleep, memories are thought to be consolidate from the hippocampus to the cortex 

making them less vulnerable to decay (Rasch & Born, 2013). Importantly, children who are 

habitual nappers may be required to consolidate memories more often due to immaturity of 

memory regions. For example, habitual nappers may have a less mature hippocampus requiring 

more regular memory consolidation due to insufficient storage capacity (Spencer & Riggins, 

2022). Evidence for this hypothesis comes from studies that show there are significant 

differences in both hippocampal subregions (Allard et al., Under Review) and hippocampal 

subfields (Riggins & Spencer, 2020) based on nap habituality. 

However, research examining associations between brain structure and nap habituality 

has not extended beyond the hippocampus. Thus, it is unclear if these associations are specific to 

the hippocampus or if they generalize to cortical regions associated with memory. In other 

words, no research to date has examined whether the maturation of cortical memory regions is 

also related to the nap transition. It is possible that, in addition to changes in the hippocampus, 

the cortex also matures in a way that “allows for” the nap transition.  However, changes in the 

cortex could also be a downstream result of the nap transition.  Regardless, it would be 

interesting to determine if regions other than the hippocampus show differences around the nap 

transition, then a mechanism can be further explored.  

Studies have shown that sleep is related to the maturation of the cortex (Cheng et al., 

2021; Hansen et al., 2022; Taki et al., 2012). For example, the synaptic homeostasis hypothesis 

suggests that “sleep is the price we pay for neural plasticity” (Tononi & Cirelli, 2014). During 

sleep, the brain is disconnected from external stimuli allowing for enhanced synaptic up and 

downscaling. This process leads to stronger relevant connections and fewer irrelevant 



15 
 

connections. In addition to supporting general brain development, it is also hypothesized that this 

process supports the long-term storage of memories (Tononi & Cirelli, 2014).  

Evidence in humans shows that shorter 24-hour sleep duration has been linked with 

reduced total gray matter volume, cortical surface area, and cortical thickness across childhood 

and adolescence (Cheng et al., 2021; Hansen et al., 2022; Taki et al., 2012). For example, past 

work has shown that shorter 24-hour sleep duration is associated with decreased cortical volumes 

and surface area in the orbital frontal region, superior and middle frontal gyri, inferior and 

middle temporal gyri, precuneus, posterior cingulate cortex, ventromedial prefrontal cortex, and 

dlPFC of the rostral medial prefrontal cortex across the middle and into late childhood (Cheng et 

al., 2021). Additionally, Hansen and colleagues (2022) showed that shorter sleep duration during 

weekdays in adolescents was significantly associated with reduced cortical thickness in the left 

middle temporal gyrus, right postcentral, and right superior frontal cortices. All these regions 

have previously been implicated in memory performance. Thus, these findings suggest that sleep 

habits may be associated with cortical development of memory regions outside the hippocampus.  

Furthermore, the animal literature suggests that the process of synaptic regulation (i.e., 

downscaling and upscaling) can be observed across a single nap-like session (Maret et al., 2011; 

Yang & Gan, 2012). Specifically, research in 3-week-old mice (developmental equivalent to 

approximately 2.5 years in humans) demonstrates that both synaptogenesis and synaptic pruning 

occur during a single 2-hour sleep bout. Further, effects on synaptic pruning were above and 

beyond the effects of a wake period lasting the same duration (Yang & Gan, 2012). 

It is possible that habitual napping during childhood could be linked to the need for 

consistent synaptic pruning in cortical memory regions. Specifically, in addition to more regular 

memory consolidation, immature memory systems may also require more regular synaptic 



16 
 

regulations via pruning and synaptogenesis to enhance or re-organize memories. Thus, changes 

in nap habits may be preceded by the maturation of cortical regions that results in a less 

consistent need for synaptic regulation. As a result, the second aim of this dissertation was to 

explore whether there are differences in cortical thickness and surface area of memory-related 

brain regions based on nap habituality.  

Summary 

In summary, previous work demonstrates that episodic memory, brain regions that 

support memory and sleep habits all undergo developmental changes across childhood. Further, 

these changes are likely related. Specifically, with sleep, memories are consolidated from the 

hippocampus to the cortex. Thus, changes in sleep habits during early childhood are 

unsurprisingly related to both episodic memory performance and volumetric measures of the 

hippocampus. However, the literature demonstrates that the hippocampus is not the only brain 

region that supports episodic memory in early childhood. Limited work has demonstrated cross-

sectional associations between memory and a larger episodic memory network in childhood; 

however, longitudinal investigations are still lacking. Given previous associations between sleep 

habits and cortical regions included in the episodic memory network, it is possible that these 

regions also differ as a function of nap habitually. However, previous work has not examined 

differences in these other brain regions based on nap habituality.  

Study Overview 

The proposed dissertation aimed to accomplish two goals. The first goal was to 

investigate associations between episodic memory and extra hippocampal brain regions that 

support memory in a longitudinal sample of 4 to 8-year-old children. This was assessed using 



17 
 

both a cross-sectional whole-brain approach and a longitudinal co-development approach. These 

results allow for the direct comparisons of longitudinal and cross-sectional methods in the same 

sample. The second goal was to examine whether individual differences in these brain regions 

are explained by differences in nap habituality using a cross-sectional sample of 3 to 5-year-old 

children. Importantly, brain regions for the second goal were partially motivated by the literature 

and partially by the results of the previous assessments.  

To accomplish these goals, two separate data sets were used. The first data set (Study 1) 

utilized an accelerated longitudinal design with two cohorts assessed at three-time points. 

Specifically, the younger cohort was assessed at 4, 5, and 6 years of age. Whereas the older 

cohort was assessed at 6, 7, and 8 years of age (see Figure 1). Additionally, a cross-sectional 

sample of children were recruited at ages 5, 7, and 8 years. At each visit, subjects participated in 

an MRI scan and a battery of memory tasks that included two temporal order tasks, a source 

memory task, and an item-location binding task. This study was used to address the first aim of 

this dissertation, which was to examine longitudinal co-development of memory and cortical 

development in early to middle childhood. 

 

 

 

 

 

 



18 
 

Figure 1 

 Schematic depiction for the timing of data collection in Study 1. 

 

Note. Dark chevron represents an approximately 1-week delay. 

Unfortunately, Study 1 was not designed to assess sleep habits and measures of nap 

habituality are rudimentary. Therefore, I used a second data set (Study 2) to examine associations 

between memory and the during early childhood. Study 2 includes a cross-sectional sample of 3 

to 5-year-old children who were assessed at 3 separate visits, each one week apart. The first two 

visits were conducted in the child’s home, where they were tested on an item location binding 

task before and after both a nap and a wake session lasting the same duration. Nap and wake 

sessions were counterbalanced for order. During the third visit, children participated in an MRI 

scan where a T1 weighted image was collected (see Figure 2). Across the two-week testing 

period children also wore an actigraphy watch and parents completed a sleep diary to assess nap 

status. This study was used to assess the second aim of this dissertation, which was to examine 

differences in cortical thickness and surface area of regions that support memory development 

based on nap status during early childhood. 

 

 

Time Point 1 
(Age 4 to 8 

years)

Primacy/
Regency

Temporal 
Order

Item-Location 
Binding

Source 
Memory 
Encoding

Time Point 2 
(Age 5 or 7 

years)

Primacy/
Regency

Temporal 
Order

Item-Location 
Binding

Source 
Memory 
Encoding

Time Point 3 
(Age 6 or 8 

years)

Primacy/
Regency

Temporal 
Order

Item-Location 
Binding

Source 
Memory 
Encoding

Source 
Memory Test MRI

Source 
Memory Test MRI

Source 
Memory Test MRI

~1 week



19 
 

Figure 2 

 Schematic depiction for the timing of data collection in Study 2 

 

 

 

 

 

 

Note. Week A and Week B were counterbalanced for order. 

 

 

 

 

 



20 
 

Chapter 2: Aims and Hypotheses 

Aim 1: Examine relations between cortical thickness/surface area and memory 

performance during early to mid-childhood. 

Part I: Identify Memory-related ROIs using a Cross-sectional Sample of 4- to 8-year-olds. 

Hypothesis 1A: There will be significant negative associations between cortical 

thickness/surface area of ROIs (i.e., IFG, medial orbitofrontal cortex, superior frontal gyrus, 

rostral middle frontal gyrus, ACC, PCC, SPL, IPS, precuneus, ERC, PHG, temporal pole, 

superior temporal gyrus, middle temporal gyrus, lingual gyrus, LOC) and a composite memory 

variable such that greater cortical thickness/surface area will be associated with poorer memory 

performance. Hypothesis 1A will be assessed using both an a priori ROI analysis and an 

exploratory vertex by vertex whole brain analysis. 

Part 2: Examine Changes in ROIs and Latent Episodic Memory in 4-year-old and 6-year-old 

Longitudinal Cohorts. 

Hypothesis 1B: There will be age-related changes in cortical thickness/surface area of 

ROIs from ages 4 to 8 years. Hypothesis 1B will be assessed using latent growth curve modeling. 

Research Question 1A:  Are there age-related changes in cortical thickness/surface srea 

of ROIs linear or non-linear from age 4 to 8 years. Further, are age-related changes in cortical 

thickness/surface area of ROIs linear or non-linear with in the 4-year-old and 6-year-old cohorts. 

Research Question 1A will be assessed within cohorts using model comparison of latent growth 

curve models. Research Question 1A will be assessed between cohorts using t-tests of latent 

slopes and variances. 



21 
 

Hypothesis 1C: There will be linear age-related changes in Latent Episodic Memory. 

Hypothesis 1C will be assessed using second order latent growth curve modeling. 

Part 3: Examine Co-Development of ROIs and Latent Episodic Memory.  

Hypothesis 1D:  Longitudinal changes in cortical thickness/surface area will be 

associated longitudinal changes in memory performance. Hypothesis 1D will be assessed using 

bi-variate growth curve models. 

Hypothesis 1E:  cortical thickness/surface area at timepoint 1 will predict longitudinal 

changes in memory performance. Hypothesis 1D will be assessed using bi-variate growth curve 

models. 

Aim 2: Examine differences in cortical thickness/surface area as a function of nap status.   

Hypothesis 2A:  There will be differences in cortical thickness/surface area of ROIs that 

longitudinal co-development with memory performance based on nap status during early 

childhood after controlling potential confounds. Specifically, cortical thickness/surface area will 

be greater habitual nappers compared to non-nappers. Hypothesis 2A will be assessed using both 

an a priori ROI analysis and an exploratory vertex by vertex whole brain analysis. 

 

 

 

 

 



22 
 

Table 2  

Summary of Aims and Hypothesis. 

Aim Hypothesis 

1) Examine relations between 
cortical thickness/surface area and 
memory performance during early 
to mid-childhood.  
 
  

P1) Identify ROIs Cross-                  
sectionally 

H1A) There will be significant negative associations 
between cortical thickness/surfacer area of ROIs and 
a composite memory variable such that greater 
cortical thickness/surface area will be associated 
with poorer memory performance.  

P2) Examine Changes in 
ROIs and Latent Episodic 
Memory 

H1B) There will be age-related changes in cortical 
thickness/surface area of ROIs from ages 4 to 8 
years. 
 
RQ1A) Are age-related changes in cortical 
thickness/surface area of ROIs from ages 4 to 8 
years linear or non-linear? 
 
H1C) There will be linear age-related changes in 
Latent Episodic Memory. 

P3) Examine Co-
Development. 

H1D) Longitudinal changes in cortical 
thickness/surface area will be associated 
longitudinal changes in memory performance. 
 
H1E) Cortical thickness/surface area at timepoint 1 
will predict longitudinal changes in memory 
performance. 

2) Examine differences in cortical 
thickness/surface area as a 
function of nap status. 

H2A) There will be differences in cortical 
thickness/surface area of ROIs that co-develop with 
memory performance based on nap status during 
early childhood after controlling potential 
confounds. Specifically, cortical thickness/surface 
area will be greater habitual nappers compared to 
non-nappers. 

 



23 
 

Chapter 3: Methods 

To accomplish the objectives of this examination, I used two separate studies with pre-

existing data. Study 1 includes a longitudinal sample of 4 to 8-year-old children that utilized a 

cohort sequential design. This study assessed brain structure via a T1-weighted MRI scan and 

memory via four memory tasks (i.e., Item-location Binding, Primacy, Temporal Order, and 

Source Memory) at three time points approximately one year apart. This study was used 

primarily to address Aim 1. 

Study 2 includes a cross-sectional sample of 3 to 5-year-old children that explored 

associations between memory, sleep, and the brain during early childhood. This study assessed 

brain structure via a T1-weighted MRI scan and actigraphy to assess nap status. This study was 

used to address Aim two. In this chapter, I will describe the participants and methods used in 

each study. 

Study 1 - Hippocampal-Memory Network Development and Episodic Memory in Early 

Childhood. 

Participants. 

The sample collected in Study 1 was drawn from a longitudinal dataset investigating 

memory and brain development in 4 to 8-year-old children. Previous papers have examined 

memory and hippocampal development from early to mid-childhood in this sample (e.g., Allard 

et al., 2023; Canada et al., 2019; Canada et al., 2020; Canada et al., 2021; Canada et al., 2022; 

Riggins et al., 2018). However, to date, relations between memory and cortical thickness have 

not been examined. 



24 
 

Recruitment. Subjects for Study 1were recruited from the greater Baltimore-Washington 

area via the Infant and Child Studies Consortium, community advertisements, and word of 

mouth. Exclusion criteria included a diagnosis for a neurological condition, premature birth, 

developmental delays, or disabilities. 

Demographics and descriptive statistics. Participants for Study 1 are 200 4- to 8-year-

olds (100 were male). An accelerated longitudinal or cohort sequential design with three-time 

points was used. Specifically, two longitudinal cohorts were followed. The first cohort (61 

participants) was enrolled at age 4 years (4-year-old cohort) and subsequently tested at ages 5 

and 6 years. Participants enrolled in this younger cohort were over-sampled to ensure enough 

usable data was collected as there was also an MRI portion of the study that is not discussed 

here. The second cohort (41 participants) was enrolled at age 6 years (6-year-old cohort) and 

subsequently tested at ages 7 and 8 years. In addition, some children (N=98) participated at age 

5, 7, or 8 years only, providing cross-sectional data for one time-point. To reduce bias, children 

with partial data were included. Children with only one time-point were included in the cohort 

and time-point that corresponded most closely to their age. For example, data for a child who 

was 7 years old was considered part of time-point 2 for the 6-year-old cohort.  

 Importantly, this experimental design provides an overlapping time point at age 6, 

allowing for an approximation of the longitudinal trajectory of memory and cortical development 

from age 4 to 8 years (Duncan et al., 1996). Of the 200 children recruited, 184 (89 females) 

provided usable brain and memory data during at least one-time point with a total of 329 usable 

scans across all three-time points. Longitudinally, 102 provided memory data at one-time point, 

19 provided memory data at only two time points, and 63 provided data at all three-time points. 

Only 1 participant did not provide usable data at any of the three-time points and was excluded 



25 
 

from all analyses. Further, seven participants did not provide usable data at timepoint 1 but did 

provide usable data at timepoint 2 and/or timepoint 3. In the younger cohort, approximately 82 

participants provided usable data during at least one-time point and in the older cohort, 

approximately 102 participants provided usable data during at least one-time point. 

Table 3 

Available MRI Sample and Average Age of Participants in Each Age Block for Study 1 

  
MAGE (N MRI Data) 

  

Age group 4 years 5 years 6 years 7 years 8 years Total 

Cross-Sectional Sample 4.43 (46) 5.58 (29) 6.46 (41) 7.54 (30) 8.56 (31) 6.34 (177) 

4-year-old cohort 4.43 (46) 5.47 (44) 6.43 (42) 
   

6-year-old cohort 
  

6.46 (41) 7.55 (34) 8.52 (42) 
 

 

Note. Gold cells denote individuals providing MRI data for the 4-year-old cohort. Red cell 

denotes individuals providing MRI data for the 6-year-old cohort. MAGE= Mean age in years. 

 

Participants in this sample were 56% Caucasian, 13% Black, 5% Asian, and 19% 

Multiracial. Further, 15.5% of the subjects’ ethnicity was reported as Hispanic or Latino. 

Approximately 7% of parents did not reveal their child’s race and 4.5% did not reveal their 

child’s ethnicity. The sample consisted mostly of middle-to-high-income families (median = 

>$105,000, range = < $15,000 - >$105,000) with 4% of the sample not revealing their income. 

Moreover, 81% of recruited children had a parent or parents who achieved at least a four-year 

college degree. 



26 
 

Prior to data collection, all methods were approved by the University of Maryland 

Institutional Review Board. Additionally, all parents provided informed consent and children 

provided verbal (i.e., children younger than 7 years) or written (i.e., children older than 7 years) 

assent depending on their age. Participants received age-appropriate prizes and parents/guardians 

received monetary compensation. 

Behavioral Tasks.  

Findings from behavioral tasks collected in Study 2 have previously been reported 

elsewhere. Therefore, prior to each task description, citations for relevant lab publications are 

provided. 

Item-location binding task (Allard et al., 2023; Canada et al., 2021). To assess 

participants’ feature binding abilities, they completed an item-location binding task (see Allard et 

al., 2023; Lorsbach & Reimer, 2005). Importantly, the present task and therefore a similar 

description of this task was previously used in another publication by the present author (Allard 

et al., 2023). Specifically, a forced-choice (yes/no) item-location recognition task was adapted 

from Lorsbach and Reimer (2005). First, participants viewed a practice booklet that familiarized 

subjects with eight black-and-white line drawings of common images (e.g., a balloon, heart, fish, 

lion, kite, snowman, pumpkin, and a frog,), and a square 3 × 3 grid that would be used in the 

task. Next, the practice booklet was used to demonstrate an example of a potential test sequence. 

Specifically, for each trial, three different drawings were displayed sequentially in different grid 

squares and the participant was prompted to remember location of each picture on the grid. 

Finally, before starting a test trial, participants completed two practice trials to ensure their 

understanding of the task. 



27 
 

The task contained 16 target trials and 16 lure trials. Importantly, target and lure trials 

were presented in random order for a total of 32 trials. In target trials, the line drawing was 

presented in the same grid location during both the encoding and the testing phase. In lure trials, 

the line drawing was presented in a different grid location during the encoding and the testing 

phase. At the conclusions of each test trial, participants were asked to verbally respond "yes" if 

they believed the test trial was a target trial and to respond "no" if they believed it was a lure 

trial. The experimenter was responsible for recording all answers.  

At the onset of each trial, "Ready?" was presented on the test screen. This was followed 

by a one-second blank screen. Next, the 3 × 3 grid was presented, initiating the encoding phase. 

During the encoding phase, three different line drawings were presented sequentially in three 

distinct locations on the 3 × 3 grid. The grid remained on the screen for a total of 3 seconds, with 

one second for each image to be displayed in a new location. Following encoding, a 4-second 

delay was initiated before the test phase. At the end of the delay, children were presented with 

the test item (i.e., a line drawing in a location on the grid). Then children were asked to 

determine if they had seen that line drawing in that location on the grid. The test item remained 

on the screen until the child responded. Each trial was followed by a two-second interval. 

To assess performance on the item-location binding task, d’ was calculated and included 

as indicator in the latent episodic memory variable (Snodgrass & Corwin, 1988). Specifically, d’ 

is computed by Z transforming hit rates and false alarm rates. Then corrected false alarm rates 

are subtracted from corrected hit rates (i.e., d’ = Z (Hit Rate) – Z (False Alarm Rate)). 

Source memory task (Canada et al., 2021; Riggins et al., 2018). To further assess 

participants’ feature binding abilities, they completed a source memory task that assessed 

weather they could bind novel information to the context where it was first encountered (see 



28 
 

Canada et al., 2021; Drummey & Newcombe, 2002; Riggins, 2014). Importantly, unlike the 

other memory tasks, the source memory task was administered across 2 test sessions separated 

by approximately 1 week.  

During the first session, participants were taught some new information (e.g., “Glass is 

made from sand.”) from two separate sources, a male puppet named “Henry” and an adult female 

named “Abby.” To maintain consistency across participants, all facts were presented via video 

recording. Each child was presented with 12 facts in total, 6 from each source. Importantly, facts 

were presented in source-based blocks. In other words, participants learned all 6 facts from 

Henry followed by all 6 facts from Abby, or vice versa. The order of the blocks was then 

counterbalanced across all subjects. For this task, there were 3 lists of facts. Each list contained 

different, but similar facts (e.g., “Honey is the only food that never goes bad.” or “Grapes are the 

most popular fruit in the world”). Lists were randomly assigned across all subjects and 

longitudinal subjects were provided with a different list at each time point. At the onset of the 

task, participants were prompted to remember the facts and informed that they would be tested 

the following week. However, participants were not informed that they needed to remember the 

source of each fact. Before children were told a fact, an experimenter would ask if the child 

already knew the fact (e.g., “What is glass made from?”). When a child demonstrated they knew 

a fact, the fact was omitted and a different fact from was presented from the same source. Each 

source was assigned 8 possible facts from each list. Therefore, if a participant knew at least 3 

facts from one of the sources, that child received fewer facts in total (N = 4). 

During the second session, participants’ memory for the first session was assessed. This 

was accomplished using a 22-item trivia questionnaire provided verbally by the experimenter. 

Importantly, before starting the questionnaire participants were informed that they had learned 



29 
 

some of the answers the week before from either Henry or Abby. They were also told that they 

may have learned some of the answers elsewhere (e.g., from school) and that they may not know 

the answer to some questions. Of the 22 facts on the trivia questions, children had learned 6 from 

Abby, 6 from Henry, 5 were considered to be commonly known facts by children (e.g., “What do 

you use to brush your teeth?”), and 5 were considered to be uncommonly known facts by 

children (e.g., “What animal’s tongue is two times as long as its body?”). For each list, there 

were two possible random orders of the facts. These were counterbalanced across participants.  

For each question (e.g., “What is glass made from?”) participants were first given the 

opportunity to demonstrate free recall of the fact. When a participant expressed that they did not 

know the answer, a recognition option was given instead. For this option children were provided 

four multiple-choice answers (e.g., Chameleon, Dog, Ant Eater, Snake). Once an answer was 

provided, the experimenter asked who taught the child that fact. Again, subjects were first 

provided with a free recall opportunity. If the participant indicated they did not know who had 

taught them the fact, they were provided with another recognition option. For this option, 

children were provided with five multiple-choice answers (e.g., parent, teacher, Abby, Henry, or 

they just knew). Proportion correct answers for both fact and source questions were calculated 

and included as an indicator for the latent episodic memory variable.  

Order memory-recognition (primacy) (Canada et al., 2020; Canada et al., 2021). To 

assess recognition memory for temporal order, a primacy task was included. Specifically, this 

investigation utilized a modified version of a primacy task used in several other studies that 

examined primacy abilities in children (Alden, 1994; Mathews & Fozard, 1970). Prior to the 

primacy test phase, participants were provided with 2, 4 item practice lists to ensure they 

understood the task. During this practice phase, children were provided feedback on their 



30 
 

performance to help solidify the task instructions. During the test phase, participants were 

provided with four lists of images (2 lists included 8 items and 2 included 12 items). All lists 

were comprised of black and white line sketches of ordinary objects (e.g., a bicycle, a bear, an 

airplane, etc.). In this paradigm, participants were prompted to remember the order of the 

images. Each image was presented to the child sequentially with a verbal cue (e.g., bicycle, bear, 

airplane) at a rate of 1 image every 2 seconds. 

After the experimenter finished presenting a list, the participant was either immediately 

provided with two forced-choice questions concerning the order of two images in the list (e.g., 

“which image came first”) or they were given a distractor task (e.g., instructions to draw a 

picture) prior to the forced-choice questions. The purpose of the forced choice questions was to 

assess the child’s primacy judgment by determining which of two pictures presented in a list the 

participant remembered seeing first. The order of the lists and which lists were followed by a 

distractor task was counterbalanced across subjects. Each pair of images that were used for 

primacy judgments were presented one image apart in the list sequence. Additionally, one pair 

was taken from the first part of the list (i.e., the first 4 or first 6 items) and the second pair from 

the second part of the list (i.e., the last 4 or 6 items). Specifically, the paring of item 2 with item 

4 and item 5 with item 7 was always used for the 8-item list. Similarly, for one of the 12 item 

lists, item 3 was paired with item 5 and item 7 was always paired with item 9. However, on the 

other 12 item list, item 4 was paired with item 6 and item 8 was paired with item 10. This list 

design was used to avoid overlaps of already paired items with other items and to avoid using the 

first and last items in each list. After completion of the task, proportion of correct responses was 

included as the indicator variable in the latent episodic memory variable. 



31 
 

Temporal memory recall (Canada et al., 2020; Canada et al., 2021).  An ordered 

sequence recall task was used to assess children’s memory for the temporal order of events, (see 

Bauer et al., 2013; Canada et al., 2020; Pathman & Ghetti, 2014). To ensure children understood 

the task, they were shown a 4-item practice sequence (Yard). Next, subjects were randomly 

assigned to two 9-item test sequences (either Fair, Park, or Pet Shop) and the sequences were 

counterbalanced for order. Each sequence was introduced using a verbal cue related to the 

sequence (e.g., for the Yard sequence they would say “I’m going to show you how I work in the 

yard.”). Then, the child was shown laminated cards with each item for that the sequence which 

were introduced with a distinct verbale cue (e.g., “mow the lawn”). Importantly, no temporal or 

causal language cues were given (e.g., “first”, “finally”, “next”). Once all 9 items were 

introduced, the original verbal event cue was given a second time, (e.g., “That’s how I work in 

the yard”). Finally, the item cards were rearranged, and participants were asked to rebuild the 

sequence in its original form. Importantly, for one of the 9-item picture sequences, participants 

were given a randomly assigned distractor task (e.g., tic-tac- toe) between the presentation of the 

sequence and sequence reconstruction.  

After the participant reassembled the sequence, performance was recorded. Specifically, 

performance was scored on the number of contiguous pairs in the child’s reconstruction (two 

adjacent items placed in the same order as the original sequence). Thus, each sequence included 

8 potential contiguous pairs. Next, accuracy across the immediate and delayed reconstructions 

were compared. Results demonstrated that the reconstructions were similar. Therefore, the 

proportion of contiguous pairs recalled across both 9 item sequences was included in the 

construction of the latent episodic memory variable. 



32 
 

Intelligence. Measures of visual spatial and verbal IQ were collected using age-

appropriate subtests of the Wechsler Preschool and Primary Scale of Intelligence (WPPSI) (i.e., 

for 4 to 5 year olds) and the Wechsler Intelligence Scale for Children (WISC) (i.e., for 6 to 8 

year olds). Scaled scores for the block design (i.e., visual-spatial IQ) and word recognition 

(verbal IQ) subtests were used to determine whether associations between episodic memory and 

the brain were specific or attributed to general cognitive ability.  

MRI. 

Acquisition. Prior to MR data acquisition, all participants were screened for MRI 

contraindications (i.e., metal, claustrophobia) and completed training in a mock scanner that 

allowed the child to become comfortable with the scanner environment. The scanner used for 

both Study 1 and Study 2 was a Siemens 3.0-T scanner (MAGNETOM Trio Tim System, 

Siemens Medical Solutions, Erlangen, Germany) with a 32-channel coil. A high-resolution T1-

weighted magnetization-prepared rapid gradient-echo (MPRAGE) image sequence that consisted 

of 176 sagittal slices (.9 mm isotripc; 1900 ms TR; 2.32ms TE; 900ms inversion time; 9° flip 

angle; pixel matrix= 256 x 256) was used to acquired structural data.  

Processing. Structural T1-weighted image preprocessing consisted of smoothing, skull 

stripping, image registration, motion correction, and subcortical segmentation. Standard 

procedures such as cortical surface reconstruction, cortical, and subcortical segmentation to 

determine white-pial boundaries were conducted using FreeSurfer Version 5.1 for Study 1 and 

6.0 for Study 2 (surfer.nmr.mgh.harvard.edu; Fischl, 2012; Fischl et al., 2002). Importantly, in 

our lab, we previously examined the effects of using these two different versions of Freesurfer in 

the sample data set by processing cases in both version 6.0 in version 5.1 and then conducting 



33 
 

correlation analysis. Results from these analyses were similar and suggest that differences in 

Freesurfer versions do not drive the observed effects (Ewell et al., In Press; Fitter et al., 2022). 

After initial processing, accuracy of boundary lines was evaluated by trained editors. In 

the case of errors extending for more than 7 slices, such as inclusion of the skull from either the 

gray/white or pial line, editors made corrections. Corrections were made by first changing the 

watershed value within FreeSurfer to either enhance or decrease the skull strip. If the error still 

remained manual edits were made (Ducharme et al., 2016). 

After all corrections were made, a third independent editor rated each brain on a scale 

from 1 to 5, with 1 being perfect quality and 5 being unusable. Scans that were rated as unusable 

contained extreme banding from motion and unreliable white pial boundaries that were unclear 

to the naked eye. To retain maximum data, scans rated 4 or higher were included in this analysis. 

In Study 1, 14 brains were rated as 4 or higher. In Study 2, 2 brains were rated as 4 or higher. 

Importantly, images for all subjects that provided usable data at more than one timepoint were 

additionally processed with longitudinal pipeline (Reuter et al., 2012) in FreeSurfer. This 

pipeline conforms subject data from all available time points to a common subject-template 

(called the BASE). Then, information from each cross-sectional time point and the BASE is used 

to create a new set of longitudinally processed runs for each subject. As a result, each run is 

processed in reference to other timepoints from the same subject leading to less variability. 

Cortical thickness was then computed by determining the distance between boundary lines 

(Fischl & Dale, 2000). The Desikan-Killiany Atlas was used for cortical parcellation (Desikan et 

al., 2006).  

 



34 
 

Study 2 - Hippocampal Development and Sleep-Dependent Memory Consolidation in 

Preschoolers. 

Participants. 

The sample collected in Study 2 were drawn from a larger dataset investigating memory, 

sleep, and brain development in 3 to 5-year-olds. Previous papers have examined associations 

between emotion regulation, parenting, and 24-hour behaviors with structural measures of the 

cortex and hippocampus (e.g., Allard et al., Under Review; Ewell et al., In Press; Fitter et al., 

2022; St. Laurent et al., 2022). However, past work has not examined differences in cortical 

thickness or surface area based on nap status. 

Recruitment. All subjects recruited for Study 2 were considered to be typically 

developing children at the time of data collection and were living in the greater Baltimore-

Washington area. Further, participants for this study were recruited via the same channels 

described for Study 1. Exclusion criteria for this study included abnormal circadian function; any 

brain abnormality, brain trauma, psychiatric disorder, neurological disorder, learning disability, 

or developmental delay; a familial history of autism spectrum disorder; and premature birth prior 

to 35 weeks gestation. 

Demographics and Descriptive Statistics. Participants for Study 2 are a cross-sectional 

sample of 56, 3 to 5-year-old children who partook in a larger study assessing the effect of nap 

habits on memory and brain development in early childhood. Of the 56 participants that provided 

usable data during at least 1 time point, 46 provided a T1 scan. Of these, 2 scans were deemed 

unusable.  



35 
 

In this sample, 64.3% were Caucasian, 16.1% were Black, 7.1% were Asian, and 8.9% 

were Multiracial. Additionally, 11% of parents did not reveal their child’s race. Furthermore, 

8.6% of participants identified as Hispanic or Latino, 87.5% identified as not Hispanic or Latino, 

and 3.6% of parents did not reveal their child’s ethnicity. Importantly, this sample consisted 

mostly of middle-to high-income families (median = >$195,000, range = < $15,000 - 

>$195,000). Moreover, most (94.5%) of the recruited children had a parent or parents who 

achieved at least a four-year college degree.  

In Study 2, actigraphy and parent report was used to categorize children into three 

distinct nap status groups based on previous literature. The three groups are habitual nappers 

(i.e., children who nap 5 or more days per week), intermediate nappers (children who nap 2 to 4 

days per week), and non-habitual nappers (children who nap 2 or less days per week). Cut offs 

for these categories are based on past literature (Allard et al., Under Review; Desrochers et al., 

2016; Kurdziel et al., 2013; Kurdziel et al., 2018). Based on this criteria, Study 2 consisted of 14 

habitual nappers, 19 intermediate-nappers, and 11 non-nappers (see table 4).  

Table 4 

Number of Children Contributing Behavioral and Neuroimaging Data for Study 2 

Nap Status N (MRI) Mean (MRI age) Range (MRI age) 
Nappers 14 4.057925636 3.26 - 5.00 
Intermediate-Nappers 19 4.420423412 3.40 - 5.77 
Non-Nappers 11 4.361643836 3.21 - 5.82 
Total 44 4.279997628 3.21 - 5.82 
 

 



36 
 

Prior to data collection, all methods were approved by the University of Maryland 

Institutional Review Board. Additionally, parents provided informed consent and children 

provided verbal assent. Participants received age-appropriate prizes and parents/guardians 

received monetary compensation. 

Nap Habitually. 

Actigraphy. Two weeks prior to the MR scan, children were given an actigraphy watch 

that they were told to wear continuously over the following two weeks. These watches record 

environmental light exposure and participant movement levels allowing an experienced coder to 

differentiate sleep from wake. Furthermore, to verify estimates, parents were instructed to record 

event markers (via a button on the watch) just prior to sleep onset and just after sleep offset. If 

event markers were not present, sleep was scored manually. Importantly scoring of watches was 

conducted using Philips Respironics and a previously standardized protocol (Acebo et al., 2005).  

Nap status was calculated using the following formula: (total days napped/total days recorded)7. 

Of the 44 participants included, 31 provided usable actigraphy data (i.e., 3 or more days).  

Parent Report Measures. Two weeks prior to the MR scan, parents were provided with 

a sleep diary. It required parents to record all sleep bouts, including naps and overnight sleep, 

during the same two-week period that the child was instructed to wear the actigraphy watch. 

Average number of napping days was calculated using the same formula used for actigraphy. Of 

the 44 children that provided data for this examination, nap status for 10 was derived from sleep 

diaries. If a participant was missing actigraphy data and the sleep diary (N =3), a lab specific 

questionnaire was used (i.e., “How many days a week does your child nap?”). 



37 
 

24-hour Sleep Duration 

Average 24-hour sleep duration was calculated using actigraphy data by computing the 

amount of time between bedtime and wake-up time for naps and overnight sleep separately, then 

durations were averaged across all available days. Next, the average nap and overnight durations 

were summed. 

MRI. 

Acquisition. During an earlier visit that occurred in the child’s home, subjects were 

introduced to the MR environment via a fabric tunnel and an audio track that featured scanner 

sounds. Additionally, a book was read to each child that demonstrated the purpose of the “brain 

camera” and the order of events that would occur during their visit to the Maryland 

Neuroimaging Center. Importantly, the acquisition process at the neuroimaging center was 

identical to Study 1. 

Processing. T1 processing was identical to Study 1, with the sole exception that 

Freesurfer version 6.0 was used instead of version 5.1. Furthermore, cortical thickness and 

surface area were also acquired similar to Study 1. (surfer.nmr.mgh.harvard.edu; Fischl, 2012; 

Fischl et al., 2002). For the purpose of Aim 2, hippocampal volumes and ICV were additionally 

calculated using Freesurfer’s automated subcortical segmentation process. Then, automated 

hippocampal volumes were adjusted using the Automatic Segmentation Adapter Tool (ASAT; 

nitrc.org/projects/segadapter; Wang et al., 2012). 

 

 



38 
 

Chapter 4: Data Analysis 

Aim 1:  Examine relations between cortical thickness/surface area and memory 

performance during early to mid-childhood. 

Selection of ROIs.  

 Existing literature demonstrates that several cortical regions support episodic memory 

abilities during childhood (see Appendix B). Based on this literature, 16 ROIs were selected for 

this analysis using the Desikan-Killiany Atlas for cortical parcellation (Desikan et al., 2006). 

This atlas was chosen because it is the commonly used in the developmental memory literature 

and utilizing it allows for finding comparison across a variety of previous studies (e.g., Amlien et 

al., 2018; Bauer et al., 2019; Chad-Friedman et al., 2021; Fjell et al., 2019; Schommartz et al., 

2023; Squeglia et al., 2013; Yu et al., 2018). These regions include IFG, medial orbitofrontal 

cortex (mOFC), superior frontal gyrus, rostral middle frontal gyrus, ACC, PCC, SPL, IPS, 

precuneus, ERC, PHG, temporal pole, superior temporal gyrus, middle temporal gyrus, lingual 

gyrus, LOC. IFG and ACC are comprised of several subregions. However, past research has not 

found differential effects in these subregions, therefore values for cortical thickness and surface 

area were averaged across subregions to create a singular value per ROI. 

A Latent Measure of Memory Development.  

 A previous study published from our lab produced a latent episodic memory variable 

using confirmatory factor analysis (Canada et al., 2022). Specifically, this variable was created 

using data from Study 1 described in the methods section, and included the item location binding 

task, source memory task, primacy task, and temporal order memory task. This latent episodic 

memory variable demonstrated strong factorial invariance and trajectory convergence between 



39 
 

the 4-year-old and 6-year-old cohort (see Canada et al., 2022). Together, this suggests that this 

structure of latent episodic memory assesses the same latent variable across cohorts and ages. As 

a result, this variable was a good candidate for this analysis and was used as a measure of 

episodic memory performance. Further, to assess cross-sectional associations of the memory 

variable, a composite memory variable was extracted from the confirmatory factor analysis. This 

is critical as latent structures cannot be assessed in qdec. Importantly, when the latent memory 

variable is extracted, it is considered a “composite” variable and is no longer “latent.” Thus, 

when the variable is extracted, it is referred to it as the “composite memory variable.” When the 

variable is not extracted, it is referred to as the “latent episodic memory variable.” 

Power Analysis. 

To assess whether our sample was large enough to address Hypothesis 1A, a power 

analysis was conducted using G*Power version 3.1.9.6 (Faul et al., 2007), based on findings 

from Schommartz et al., (2023) (N = 63). This study found a moderate effect size of f2 = .14 

when examining associations between cortical thickness and episodic memory performance in 

children aged 5 to 7 years (Cohen, 1988). Based on these findings, a significance criterion of α = 

.05, and power = .80, the minimum sample size needed for this effect size is N = 156. Thus, the 

obtained sample size of N = 177 is adequate to test the study hypothesis.  

Power analysis in latent growth curve modeling typically reveals that large samples, often 

in the thousands, are required to conduct models with small degrees of freedom.  For example, I 

conducted a power analysis using quantpsy for the proposed latent growth curve model for 

Hypothesis 1B (Depicted in Figure 8; MacCallum et al., 2006; Preacher & Coffman, 2006). In 

the analysis, I assumed an α =.05. power =.80, and df = 1. Findings revealed that this model 



40 
 

would technically require a sample of 856 participants to detect large effects. However, previous 

work has demonstrated that changes in both the amygdala and the hippocampus can be detected 

using latent growth curve models with similar sample sizes (Canada, Botdorf, et al., 2020; 

Canada et al., 2021). This is likely because omnibus approaches do not account for model 

parameters. One solution to this, is Monte Carlo simulation that assess power using predicted 

effects sizes and variances (Muthén & Muthén, 2002). Therefore, to assess power for Hypothesis 

1B, I used a Monte Carlo simulation to assess whether a sample of 82 (i.e., N for the 4-year-old 

cohort) could detect changes in cortical thickness given a moderate effect size of 0.3 for the 

slope, assuming other common parameters suggested by Muthen and Muthen (2002). The result 

showed that the effect would be detected 88% of the time. However, note that I was unable to 

fully account for data missingness thus this is likely an over-estimation of power. 

Importantly, for Hypothesis 1C, no a priori power analysis was conducted, because we 

have previously demonstrated that this sample is large enough to detect changes in latent 

episodic memory during early childhood using a model with fewer degrees of freedom (Canada 

et al., 2022). Additionally, no power analysis was conducted for Hypothesis 1D because no 

previous analysis has examined co-development of memory and these cortical ROIs.  

Outliers. 

Past research has demonstrated that general linear models are highly susceptible to the 

effects of extreme outliers (Osborne & Overbay, 2004). Further, previous work shows that 

extreme outliers that are at least 3 standard deviations above or below the grand mean can 

significantly bias findings. For example, extreme outliers can mask true effects, even when 

values are naturally occurring in the sample (Barnett & Lewis, 1994). Thus, prior to analysis, 



41 
 

extreme outliers were identified using a box plot method and excluded if values were greater 

than 3 standard deviations above the mean. The same process was used in Study 1 and Study 2. 

Cross-sectional ROI Analysis. 

To assess cross-sectional associations between the composite memory variable and 

cortical ROIs, data from the first wave of Study 1 (N = 177) was used. Specifically, separate 

lateralized regressions were conducted for each cortical measure (i.e., thickness and surface area) 

to predict performance on the composite episodic memory variable. All analyses controlled for 

age and sex. Additionally, IQ was explored as potential covariate to ensure that associations 

between episodic memory and the cortex were not attributed to general cognitive abilities. 

Further, to ensure that findings were specific to regions associated with memory performance, 

associations between the composite episodic memory variable and total gray matter volume were 

also assessed.  

Given the exploratory nature of this analysis, all significant findings are reported. 

However, this approach will yield 64 separate comparisons, thus, p values were adjusted based 

on the false discovery rate to control for the likelihood of false positives. However, both are 

reported. By reporting both criteria, results can be evaluated using both liberal (uncorrected) and 

conservative (corrected) approaches.  

Cross-sectional Whole-brain Analysis. 

Following the ROI analysis, a whole brain vertex-by-vertex analysis was employed to 

confirm previously identified regions and to investigate previously unidentified regions. In 

Freesurfer’s QDEC application, linear regressions examining associations between cortical 

thickness, surface area, and the composite memory variable were conducted controlling for age, 



42 
 

sex. Additionally, Monte Carlo simulations were utilized to correct for multiple comparisons 

across both hemispheres and estimate appropriate cluster sizes (Hagler et al., 2006). All analyses 

utilized a minimum threshold of p<.05. 

Latent Growth Curve Modelling.  

To best characterize and assess longitudinal growth in episodic memory performance and 

the brain, latent growth curve models were used. These are a specific subclass of Structural 

Equation Models that were selected because of their ability to effectively assess change over time 

(McArdle, 2009). Additionally, these models are robust to potential ceiling and floor effects and 

allow for the comparison of slopes between cohorts. To reduce bias in estimates and address 

missing data, full information maximum likelihood (FIML) estimation was utilized. All latent 

analyses were conducted in Mplus (v8; Muthén & Muthén, 2017). 

Longitudinal Development of