University of Southern California Dissertations and Theses

The acute impact of glucose and sucralose on food decisions and brain responses to visual food cues

(USC Thesis Other)

The acute impact of glucose and sucralose on food decisions and brain responses to visual food cues

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE ACUTE IMPACT OF GLUCOSE AND SUCRALOSE ON FOOD
DECISIONS AND BRAIN RESPONSES TO VISUAL FOOD CUES
By
Xiaobei Zhang
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Completion of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY )
August 2021
Copyright 2021 Xiaobei Zhang

ii
Dedication

To my family.

iii
Acknowledgements
I have been fortunate enough to be surrounded by a community of mentors and friends
who inspired and supported me throughout my PhD journey, without whom this thesis would not
have been possible.
First and foremost, I would like to express my deepest gratitude to my advisor, Dr. John
Monterosso, for his immense patience and the guidance he provided. He has reviewed countless
versions of the current work and helped me to clarify and sharpen my ideas with his astute and
prompt feedback, always with tremendous enthusiasm. During my graduate school years, John
not only gave me the freedom to pursue my research interest, but also supported me to go to
academic conferences to expand my vision. I am also extremely grateful that he always instilled
confidence in me by encouraging me to reach my full potential and assisting me to achieve my
goals. During my PhD years, his faith in my ability to conduct research has been my greatest
source of motivation in my battle against my inner voice of self-doubt. Without his resolute
belief in my capabilities and his tireless efforts in guiding me, I couldn’t have come this far. His
rigorous approach to research and his genuine care for students’ well-being have inspired me to
pursue an academic career and have provided me with an excellent model that I can only hope to
emulate in the future. My admiration for him and my gratitude to him run deep.
I am grateful as well to my committee members, Drs. Richard John, Antoine Bechara,
Katie Page, and Shan Luo for their constant support and guidance throughout my graduate
school studies. My interest in decision-making was initially sparked by Antoine Bechara’s
research in affective decision-making when I was in college. Taking Richard John’s research
design and decision-making theories courses further extended my interest and equipped me with
essential knowledge to conduct research in this field. My collaboration with Katie Page’s lab

iv
opened the door to the interdisciplinary study of food decisions, which is both interesting and
challenging. I owe many thanks to Katie Page for her keen insights and discerning advice in
guiding me through this new realm. I respect her expertise and I am thankful that she always
went out of her way to support me. Shan is not only my mentor but also my dear friend. She is
my great influence and invaluable source of support. They both have become my extraordinary
female role models, and their presence has allowed me to envision my own potential.
My heartfelt thanks also go to Dr. Jonas Kaplan, who generously and patiently advised
me on multivariate pattern analysis as soon as I approached him. He is genuinely brilliant. His
guidance sped up my self-learning process enormously and was critical in making the current
work possible. I would also like to express my gratitude to the administrative staff (especially
Twyla Ponton and Jennifer Vo) in the Psychology Department at the University of Southern
California for their efforts in supporting graduate students.
I want to thank my previous and current lab mates. I am indebted to Dr. Eustace Hsu,
who has helped me tremendously in many ways, from navigating a new culture when I first came
to the U.S., to teaching me advanced statistical methods to helping me oversee my own project. I
also thank Dr. Andrew James Melrose, whose dissertation data was used in my very first
publication. I share many fond memories with Nina Christie of the parties and girls’ nights; I
have always admired her cheerful character and keen interest in research. Milad Kassaie has
always impressed me with his wide range of knowledge, no matter what topics are discussed. In
addition, this thesis would not have been possible without the help of my undergraduate research
assistants: Olivia de Santis, Vanya Vojvodic, and Qiongwen Cao. Their assistance in data
collection and preprocessing are sincerely appreciated, and their dedication in pursuing their own
dreams also inspired and resonated with me.

v
I would also like to express a special thank you to my dearest friends: Jiyoung Lee,
Zhiqin Chen, and Huazuo Shao. They are always there for me through the good times and bad
times. They know who I am and where I have been, and they accept who I have become. I
cherish these true friendships that have stood the test of time and distance, and I am also
genuinely proud to witness their growth and achievements in life.
All that I am and hope to be, I owe to my parents. Even though they have been thousands
of miles away during my graduate school years, they never stopped supporting and caring for
me. No words can adequately convey my appreciation to my beloved parents for their
unconditional love, devotion, and sacrifice. I am forever grateful for them.

vi
Table of Contents

Dedication…………………………………………………………………….…....ii
Acknowledgements………………………………..………………….………….. iii
Table of Contents………………………………………………..………….….…. vi
List of Tables……………………………………………………......…...……..… vii
List of Figures…………………………………………………….…...…....….... viii
Abstract……………………………………………………………………...……. ix
Chapter 1: Introduction…………………………………….………...……………..1
Chapter 2: Brain Food-Cue Responses and Food Valuation…………..………….11
Chapter 3: Food Valuation and Nutrient Tracking.…..……………...……..……...36
Chapter 4: Food Value Construction and the Effect of Sucralose and
Glucose on Brain Nutrient Tracking…………………………..………..……..….. 38
Chapter 5: General Discussion………………………………….………..……..…55
References…………………………………………...……………………………71
Appendix…………………………………………………………….……………93

vii
List of Tables

Table 1. …………………………………………………………………………….23
Table 2. ………………………………………………………………………….…33
Table 3. …………………………………………………………………….………35
Table 4. …………………………………………………………………….………45

viii
List of Figures
Figure 1. ……………………………………………………………………..……..13
Figure 2. ………………………………………………………..…………………..16
Figure 3. ………………………………………………………………………..…..21
Figure 4. ………………………………………………………………………..…..25
Figure 5. ………………………………………………………………..…………..25
Figure 6. …………………………………………………..………………………..27
Figure 7. ………………………………………………..…………………………..29
Figure 8. …………………………………………………..………………………..30
Figure 9. …………………………………………………..………………………..31
Figure 10. ………………………………………………………..…...…………….32
Figure 11. ……………………………………………………………..…...……….40
Figure 12. ……………………………………………………………..…..…….….46
Figure 13. ……………………………………………………………..…..………..47
Figure 14. ……………………………………………………………..…..…….….48
Figure 15. ……………………………………………………………..…..………..49
Figure 16. ………………………………………………………………..…..……..50
Figure 17. ………………………………………………………………..…..……..51
Figure 18. ………………………………………………………………..….……...53
Figure 19. ………………………………………………………………..….……...54

ix

Abstract
Obesity has become a global problem. In the United States, obesity and overweight have
become the second leading cause of preventable death. People in this obesogenic society are
faced with unprecedented challenges of unavoidable food cue exposure due to food marketing.
Maintenance of healthy weight requires consistent good consumption decisions despite this
challenge, and so understanding the neural processes associated with food decision-making is
crucial to understanding obesity. One potential strategy used in an effort to maintain or lose
weight is to use non-nutritive sweeteners (NNSs) to replace sugar. However, neither the acute
nor chronic effects of NNS consumption on food valuation and food decisions is well
understood.
The current study examined how acute ingestion of NNS (sucralose) and of glucose
influence 1) brain food-cue responses and food valuation, and 2) brain nutrient-tracking
functionally associated with food value signaling. Food decisions were examined 55 minutes
post-consumption using a food bid task in which participants bid on visually depicted food items,
simultaneous to functional Magnetic Resonance Imaging (fMRI). Thirty food items orthogonal
in carbohydrate and fat content were selected from a standardized dataset. Twenty-eight
participants completed 3 sessions after overnight fast, distinguished only by the consumption at
the start of the session of 300mL cherry flavor water with either 75g glucose, 0.24g sucralose or
no other ingredient.
First, the result suggests an attenuation of central nervous system (CNS) signaling
associated with food valuation, and reduced bids on food 55min after glucose and sucralose

x
consumption. Specifically, there was a significant attenuation of signal increase within the a
priori region of interest (ROI) after sucralose compared to water (P <.05). Activity after glucose
did not differ significantly from either of the other conditions in the ROI, but an attenuation in
signal was observed in the parietal cortex, relative to the water condition.
In addition, the current study design allowed examination of how glucose and sucralose
ingestion influence subsequent brain nutrient tracking for visually presented foods. Utilizing
multivariate pattern analysis (MVPA), I found that carbohydrate information was tracked by a
cluster of voxels within the lateral orbitofrontal cortex (OFC). During food decisions, this
carbohydrate-tracking cluster evidenced increased association (functional connectivity) with the
area of the medial OFC that tracked the overall value of foods. Moreover, food devaluation after
sucralose and glucose was associated with decreased activity in this carbohydrate tracking
cluster, and with reduced accuracy of MVPA in the classification of whether a depicted food was
high or low in carbohydrates based on the activity pattern within the lateral OFC.
Taken together, these data suggest an acute appetite suppression effect of sucralose (and
less surprisingly, of glucose) that is partly due to the devaluation of available carbohydrates. It is
worth noting that the present findings appeared to be sensitive to the timing of consumption, so
further studies that consider other timepoints post ingestion will be important. Furthermore, these
findings do not address the potential chronic (as opposed to acute) effects of sucralose
consumption on food decisions.

1
Chapter 1: Introduction
Obesity has become a global problem. In part, the problem is a function of societal
success in reducing poverty. At the turn of the 20
th
century, approximately 80% of the world’s
population lived in conditions of extreme poverty. By the 2010’s, less than 10% of the world was
living in extreme poverty (Roser & Ortiz-Ospina, 2013). In addition to relative abundance, the
modern environment in many places is obesogenic, with unavoidable food cue exposure, like
food commercials and advertisements depicting calorie dense, and often highly processed food.
Rates of overweight and obesity have grown so sharply in recent decades that they are now
considered the greatest source of preventable mortality (Heart et al., 1998; Masters et al., 2013).
As weight control turned into a struggle for more and more people, various strategies
have been developed to win the tug of war of appetite and weight management. Diet plan
restricting caloric intake or limiting the intake of carbohydrate (e.g. low carbohydrate ketogenic
diet) have been reported to contribute to successful weight loss (Paoli, 2014). Sugar is a type of
carbohydrate with high hedonic appeal (Ganley, 1989). Accumulating evidence demonstrates
that higher consumption of sugar can lead to a series of severe health problems like obesity,
diabetes, and cardiovascular disease (V. S. Malik et al., 2010). Non-nutritive sweeteners (NNSs),
first discovered in 1878, are used to reduce caloric intake and experience sweetness at the same
time. The usage of NNS is increasingly popular. An analysis of U.S. household food and
beverage purchases between 2002 and 2018 reported that the consumption of products
containing NNS alone or in combination with caloric sugar increased, as the consumption of
products sweetened only with caloric sugar declined (Dunford et al., 2020).
Sucralose, discovered in 1976 and approved by the Food and Drug Administration (FDA)
for use in 1998 (Whitehouse et al., 2008), is the most commonly used NNS in the world (“Global

2
Zero-Calorie Sweetener Market Projected to Be Worth USD 2.84 Billion by 2021: Technavio,”
2017). However, like other NNSs, use of sucralose is controversial since it is not clear how
sucralose influences appetite, food consumption, and long-term health. In this dissertation I
examine how sucralose and glucose (a simple sugar) acutely influence food decisions and brain
responses to visual food cues. In addition, utilizing multivariate pattern analysis (MVPA), I
explore how each (sucralose and glucose) acutely effects brain response to the carbohydrate
content of available food. Since sweet taste in the evolutionary environment was reliably paired
with carbohydrate intake, it is possible that sucralose intake could acutely lead to reduced
appetite for high carbohydrate food, or perhaps food in general, despite the fact that it is not a
source of energy. I begin by introducing the basic relevant theories and concepts, followed by the
rationale of my study.
Food cue activity
The mere sight of tasty food can trigger a strong desire for eating. This conditioned
response is a form of food cue reactivity that leads to more food consumption and potential
weight gain (Jansen, 1998; Preedy et al., 2011). The hypothalamus has been strongly implicated
in appetite signaling. Cells in the hypothalamus sense and respond to changes in metabolic
signals, and activity in the lateral section of the hypothalamus increases during exposure to food
cues (Coons et al., 1965; Coons & Cruce, 1968). In functional neuroimaging research, visual
food cues (especially in the context of food deprivation) induces the activation in a network of
brain regions that includes the ventral striatum (Schienle et al., 2009; St-Onge et al., 2005),
orbitofrontal cortex (OFC) (Schienle et al., 2009; Stoeckel et al., 2008), amygdala and
hippocampus (S. Malik et al., 2008; Schienle et al., 2009; Schur et al., 2009), insula (S. Malik et
al., 2008; Schienle et al., 2009; St-Onge et al., 2005), and hypothalamus (Cornier et al., 2007;

3
Killgore et al., 2003). Neural response to cues have been indicated to predict future eating
(Lawrence et al., 2012; Lopez et al., 2014), weight gain (Demos et al., 2012; Stice et al., 2010;
Yokum et al., 2014), and risk of obesity (Stice et al., 2011). In addition, weight status appears to
be a moderator in brain response to visual food cues, and a heightened cue response in reward
and gustatory regions has been illustrated in obese people compared to healthy weight
individuals (Carnell et al., 2012; Pursey et al., 2014).
Food valuation
Throughout our daily life, decisions need to be made between mutually exclusive
alternatives. According to one widely use approach, decision-making in these cases includes
separate valuation of alternatives, including a “common currency” value-representation that
allows decisions even when the alternatives are in different domains, for example, staying in bed
for ten more minutes or getting up to go to work (Clithero & Rangel, 2013; D. J. Levy &
Glimcher, 2012). Brain substrates of motivation are complex and include prominent
contributions from many sectors including the insula cortex and dorsolateral prefrontal cortex.
However, in the context of explicit decision-making, the brain areas that have been most
consistently implicated in tracking overall value (common currency) are the ventromedial
prefrontal cortex (vmPFC) and the ventral striatum (Bartra et al., 2013; D. J. Levy & Glimcher,
2012). Both these regions track and represent reward values irrespective of its identity or type,
ranging from snack foods (I. Levy et al., 2011; Plassmann et al., 2007; Pogoda et al., 2016), to
pastime activities (Gross et al., 2014), attractive faces (J. O’Doherty et al., 2003), and social
interactions (Lin et al., 2011; Ruff & Fehr, 2014). Accumulating evidence revealed that OFC,
along with adjacent medial prefrontal cortex (PFC), plays a critical role in subjective value
tracking during the decision-making process (Clithero & Rangel, 2013; Grabenhorst & Rolls,

4
2011; Padoa-Schioppa & Assad, 2006; Rich & Wallis, 2016; Rudebeck & Murray, 2014). In
addition to functional imaging research, a voxel-based lesion-symptom mapping study,
examining individuals with focal brain tissue damage, also demonstrated that functional-
anatomical networks of value-based decision-making include OFC and vmPFC (Gläscher et al.,
2012).
Food valuation is a crucial and fundamental part of everyday food decision-making, and a
dysfunctional food valuation process has been reported to be linked with obesity and eating
disorders (Carnell et al., 2012; Foerde et al., 2015). OFC activation has been implicated both in
direct appetite signaling when viewing food pictures (Simmons et al., 2005) and during the stage
of food decision-making in which the incentive value of the food stimuli are encoded (Gottfried
et al., 2003). The specific process of food decision-making appears to be strongly dependent on
processing in the medial OFC, which can track value information independent of the identity of
the food. For example, Barron and his colleagues found that medial OFC activity tracks value,
whether value information is acquired from a direct or from an imagined consumption
experience (Barron et al., 2013). However, unlike the medial OFC, the lateral OFC has been
shown to track value information in an attribute-specific way. In one study demonstrating
attribute-specific value, lateral OFC activity was predictive of values for savory and sweet food
odors with MVPA (Howard et al., 2015). Another study using functional magnetic resonance
imaging (fMRI) and a repetition suppression paradigm demonstrated that responses in lateral
OFC encode reward-identity representations and are linked to distinct predictive stimulus (Klein-
Flügge et al., 2013).

5
The construction of food value
Though the brain substrates of food value signals have been well established, less is
known about how those food value signals are constructed. A recent study attempted to uncover
the underlining mechanisms of value construction in food decision-making by taking into
account the constituent nutritive attributes of a food item (Suzuki et al., 2017). According to
these researchers, subjective food values can be predicted based on the value of the constituent
nutrients (e.g., fat, carbohydrates, etc.) contained in the food (Suzuki et al., 2017). The entire
orbital surface has been reported to track value information, with most prominent signaling in the
medial portion of it (Chib et al., 2009; Clithero & Rangel, 2013; Grabenhorst & Rolls, 2011; D.
J. Levy & Glimcher, 2011; McNamee et al., 2013). However, sensory inputs from the visual,
olfactory, auditory, gustatory, and somatosensory systems are primarily received in the lateral
portion of the OFC (Öngür & Price, 2000). MVPA of fMRI data (Suzuki et al., 2017) indicate
that neural patterns of food value can be decoded in both medial and lateral OFC, whereas only
the lateral OFC reflects the constituent nutritive attributes. Furthermore, evidence from
functional connectivity analyses suggests that nutritive attributes tracking in lateral OFC is
integrated in medial OFC to form food value (Suzuki et al., 2017).
The effects of glucose and sucralose
Glucose is the main circulating sugar in the blood. Both plasma glucose levels and insulin
(which is released in response to rising glucose) have been shown to alter signaling in reward
pathways at biologically relevant levels (Page & Melrose, 2016). Using fMRI combined with a
stepped hyperinsulinemic, euglycemic–hypoglycemic clamp technique, hypoglycemia (relative
to euglycemia) is reliably associated with greater activation during food-cue exposure in the
insula and striatum, along with higher hunger ratings. Circulating level of glucose has been

6
found to modulate brain reward circuitry (increased food-cue induced activity in the insula and
striatum) and associated subjective reports of food motivation (Page et al., 2011). Related
evidence linking insulin sensitivity to hunger and hunger signaling has been observed in studies
utilizing oral glucose intake (Van Vugt et al., 2014). Food-cue responses in normal weight
participants was attenuated after ingestion of glucose (dissolved in water) in the basal ganglia
and paralimbic regions, as were ratings of subjective hunger (Kroemer et al., 2013). However,
Heni and colleagues (Heni et al., 2014) did not observe significant attenuation of food-cue
responsivity following glucose ingestion, suggesting the possibility that methodological variance
(e.g., timing of intake relative to cue exposure) may be important.
NNSs are sugar substitutes which imbue foods with sweet taste without adding calories
or triggering a glycemic response (Nehrling et al., 1985). The health consequences of chronic
NNS consumption remains controversial (de Ruyter et al., 2012; Higgins & Mattes, 2019;
Pepino, 2015; Porikos & Pi-Sunyer, 1984; Raben et al., 2002; Rogers et al., 2016; Swithers,
2013). It has been argued that chronic NNS consumption disrupts learned responses that
normally contribute to glucose and energy homeostasis and may drive metabolic dysregulation
and increase the risk of obesity, diabetes, metabolic syndrome, and cardiovascular disease
(Pepino, 2015; Swithers, 2013). The “sweet uncoupling” hypothesis has been used to explain
why NNS consumption could lead to metabolic impairment (Davidson et al., 2011; Foletto et al.,
2016; Swithers et al., 2012). By decoupling sweet taste receptor activity (normally a reliable cue
of sugar consumption) from a subsequent rise in nutrient availability, learning processes that
ordinarily support appetite regulation are disrupted (Swithers, 2013; Swithers et al., 2012). A
recent study offered a new perspective on the effect of NNS by showing that the consumption
of sucralose together with carbohydrate, but not without carbohydrate, could impair glucose

7
metabolism (Dalenberg et al., 2020). However, an obesogenic impact of NNS has not been
convincingly established. A review of studies investigating the relation of NNSs and weight gain
and obesity concluded that the hypothesized link between NNS and obesity lacked empirical
support (Fernstrom, 2015). Indeed several well-designed intervention studies showed that
chronic and covert NNS substitution could reduce energy intake and body weight (de Ruyter et
al., 2012, 2012; Porikos & Pi-Sunyer, 1984; Raben et al., 2002). A systematic review with meta-
analysis also reported NNSs use in place of sugar leads to reduced energy intake and reduced
body weight (Rogers et al., 2016). The NNS sucralose, which is commonly used in the world
food supply (Grotz & Munro, 2009), appears to be particularly promising with regard to appetite
suppression. A recent 12-week intervention study comparing the effects of NNSs and sucrose
found decreased energy intake with sucralose consumption (but not with other NNSs) in
overweight and obese individuals (Higgins & Mattes, 2019).
The current study
While the acute effects of sucralose consumption on appetite signaling may diverge with
chronic sucralose ingestion effects, delineation of acute effects can provide clues regarding
chronic effects. While some studies assessed the immediate gustatory responses and neural
activity after consumption of sucralose (Dalenberg et al., 2020; van Opstal et al., 2019), no
study, to my knowledge, has directly examined the effect of acute sucralose intake on brain
activity during food decisions. By including water, glucose and sucralose conditions within the
same study, I intended to distinguish acute effects of sweet taste (present in both sucralose and
glucose conditions) from the effects of the caloric load (present only in the glucose condition).
In addition to assessing acute effects of sucralose and glucose on food decisions and
associated brain activity, the present study was designed to replicate and extend the

8
aforementioned finding from Suzuki et al. 2017 on value construction. Suzuki and colleagues
(2017) reported that food value was well-explained by an additive combination of four nutrients:
fat, carbohydrates, protein and vitamins (not salt and sugar). However, there are probably
interactions between nutrients, for example salt and sugar hardly go together. In addition to that,
fat and carbohydrates (including sugar) may be correlated among commercial food items (e.g.,
candy and desserts are high in both). Since multicollinearity might decrease the degree to which
nutrient-specific brain responses can be identified, avoiding this issue was an important objective
of the present study.
In order to build on prior neuroimaging work looking at food decision-making (especially
Suzuki et al. 2017), the present study incorporated two novel features: 1) I included food stimuli
with orthogonal carbohydrate and fat content, so that signaling specifically linked to each could
be optimally identified; and 2) I included multiple testing sessions with different preloads
(glucose, sucralose, and water) before the food valuation task in order to manipulate the
metabolic states, as well as (possibly) the carbohydrate-specific value signaling. This
manipulation allowed us to examine how different metabolic states impact food-cue activity and
food decisions as well as how an anticipated reduction in appetite for a specific nutrient
(carbohydrate) altered brain activity during food decision-making.
In the present neuroimaging study, participants bid money on visually depicted food. For
each food, the participant’s bid could determine whether the food was available upon completion
of the task. The task thus included both food-cue exposure, and food decision-making. To
investigate brain food-cue reactivity in the decision-making task, I utilized a food-cue mask
inclusive of the regions that have consistently been found activated when participants view food
pictures. I first examined the impact of study drinks on food bids and brain food-cue response. I

9
hypothesized that relative to water, the consumption of glucose and possibly sucralose would
attenuate MRI signal in regions that track food valuation and would lead to lower bids on food
items. Identifying neural correlates of acute glucose and sucralose consumption during food
choice may provide clues that inform the understanding of appetite regulation and obesity.
Moreover, gaining a better understanding of underlying brain function and how this relates to
motivated behaviors may also provide insights that extend to other disease processes with
overlapping neural pathways, such as drug addiction (Tang et al., 2012). I then assessed the
effects of study drinks on food valuation by evaluating changes in carbohydrate tracking in the
brain during food valuation. The effect of sucralose on nutrient tracking and the food valuation
process was of particular interest. After sucralose ingestion, the highly rewarding sweet taste
could be delivered in the absence of extra calories. If participants altered their food valuation
after sucralose consumption, potential mechanisms related to nutrient tracking (in particular,
carbohydrate tracking) could be investigated using these data. MVPA was used to decode value
and nutrient signals in brain since both signals have been revealed to be multivariate in nature.
Based on the previous findings, I predicted that glucose and sucralose preloads would influence
food valuation either by 1) reducing the constituent signaling in the sector of the lateral OFC
sensitive to the presence of carbohydrate nutrients, or 2) by reducing the connectivity between
this region and the portion of the medial OFC tracking overall value.
Whether prolonged NNS exposure alters physiological responses to caloric sweeteners in
humans remains unclear. One study reported that regular diet soda drinkers possessed altered
reward processing of sweet taste (Green & Murphy, 2012). To control for the effects of past
NNS consumption on food cue activity and food decisions, dietary recall reports were obtained
for all participants. Although the study was small, it allowed pilot data exploring possible

10
interactions between study variables and participants’ weight status (obese vs. non-obese) on
food valuation and brain activity.
It should be noted that the study was also designed to measure brain response to very
brief (subliminal) presentation of food-cues, which were presented before some trials following
the method used by (Sato et al., 2017). The intent in including subliminal primes was so that if I
did observe study drink effects on appetite signaling during decision-making, the subliminal
primes might allow us to address whether metabolic state affected the rapid reward-orienting
response to stimuli (Childress et al., 2008; Sato et al., 2019), or was limited to the slower central
nervous system (CNS) signaling associated with bidding on available food. However, perhaps
due to the small number of subliminal presentations in this study compared to past reports, I did
not observe any effect of subliminal presentations, even collapsing across all conditions.
Therefore, I do not discuss the subliminal primes further (though for completeness, they are
presented in Appendix).

11
Chapter 2: Brain Food-Cue Responses and Food Valuation
Materials and Methods
Participants
Study recruitment was done primarily through flyers placed near the campus of the
University of Southern California. Participants’ characteristics are depicted in the result section.
Due to excessive motion artifact or incomplete coverage, imaging data could not be included
from 1 glucose session, 1 sucralose session, and 2 water sessions. To reduce variance across
sessions related to hormonal change, female participants (with one exception) completed all
three scanning sessions in the window between 15 and 22 days post-start of their most recent
menstruation (presumed luteal phase, though not confirmed by bioassay) (Dye & Blundell, 1997;
Krishnan et al., 2016). In MVPA combining all three sessions, two participants (two men with
BMI of 39.5 and 25.3) were removed due to incomplete runs on one of the sessions. In MVPA
comparing classification performance between different drink conditions, only participants who
completed two runs of the food bid task each day were included.
Participants gave written informed consent to all experimental procedures approved by
the Institutional Review Board of the University of Southern California. The study was
conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the
Ethics Committee of the University of Southern California (UP-16-00413).
Stimuli
In each administration of the Food Bid Task, participants were presented with 30 visually
depicted food items selected from “Food.pics”, a freely available database of 568 food images,
each with associated normative data (Blechert et al., 2014). Normative data include rating scores

12
(derived from approximately 2000 adults) of palatability, desire to eat, complexity,
recognizability, and valence, as well as nutrient information for the item (mg of protein, fat &
carbohydrates, kcal). The 30 images used throughout the study were selected in two steps. First,
the research team identified 70 items among the 2000 with high clarity and familiarity to
participants across a range of ages and backgrounds. Next permutation analysis was used to
select 30 items from the set of 70 with minimal correlation between fat and carbohydrate content.
Specifically, a script was written to randomly select 1000 sets of 30 items from the 70, compute
the attribute collinearity of carbohydrate and fat for each set, and then retain the set with the
lowest collinearity for use in the experiment.
Procedures
Participants reported to Dana and David Dornsife Neuroimaging Center at USC between
8 and 9 am after an overnight fast. The start time for each of the three sessions, never varied for a
participant by more than 30 minutes. Weight was measured at each session. MRI was performed
using a 3-T Siemens MAGNETOM Tim/Trio scanner (Munich, Bavaria, Germany) with a 32-
channel head-coil. Prior to consumption of any drink, participants were trained on the procedures
used in the study and provided session baseline information including ratings of mood and
hunger. Next participants ingested 300 mL of water with a zero-calorie, mild cherry flavoring
either 1) with no other ingredient (“Water”), 2) mixed with 0.24g sucralose (“Sucralose”), or 3)
mixed with 75 g glucose (“Glucose”). Drinks were prepared by the study coordinator. The
concentration of sucralose (2 millimoles per liter (mmol/L) ) is similar to the diet soda (Pepino et
al., 2013). The rest of the study team and the participant were blind to drink type. The order of
drinks used in the three sessions was approximately balanced (Water first for 10 participants,
Glucose first for 10 participants, and Sucralose first for 8 participants). Participants were

13
instructed to consume drinks in less than two minutes. Immediately after consumption,
participants rated the pleasantness and sweetness of the drink by rating scale (1-10). Testing on
the Food Bid Task began approximately 55 minutes after consumption. Prior to the Food Bid
Task, participants completed one or more fMRI tasks that had no connection to food, which are
not discussed in this report. Sessions were no fewer than 2 days apart, and no greater than 30
days apart. In order to increase valuation of depicted food items, study sessions continued for 30
minutes after completion of the scan, and participants were made aware that their only
opportunity to eat would be if they bid enough (see below) on a randomly selected food item.
Figure 1 provides the timeline of measures included in this report.

Figure 1
Timeline of the Study

Note. “~” sign was used as the approximation of the time points and FMRI BID task refers to functional magnetic
resonance imaging food bid task.

Appetite Rating
Participants reported on their appetite three times during each session: 1) at the start of
the session (prior to consumption of the study drink), 2) just before entering the scanner
(approximately 5 mins after completing consumption of the study drink), and 3) upon completing
the scanning session (approximately 90 mins after completion of the study drink). For this report,

14
I consider only responses to the questions, “How hungry do you feel right now?” and “How
much do you want to eat something sweet?” Participants responded to these questions on a scale
from 0 to 100 with 0 representing “not at all” and 100 representing “a lot.”
Past NNS usage and Dietary Intake Assessment
Five (on average) 24-hour dietary recalls on non-consecutive days during a two-month
period were reported for each participant, in which detailed food and beverage consumption over
the previous 24 hours was recorded. Nutritional Data System for Research (NDSR) software
(Feskanich et al., 1989) was used to analyze the dietary data. The plausibility and quality of the
dietary recall were assessed by using the methods implemented in a previous study (Clark et al.,
2020). Past NNS intake was calculated by averaging daily intake of acesulfame potassium
(aceK), sucralose, saccharin, aspartame, or any combination thereof, in milligrams (mg). If
participants’ average daily intake of NNS was above 0, they were categorized as NNS users.
Participants with an average daily NNS intake equal to 0 were categorized as NNS non-users.
Food Bid Task
The food bid task was adopted from Suzuki et al. (Suzuki et al., 2017). The task utilized
the Becker–DeGroot–Marschak (BDM) auction method (Becker et al., 1964) to elicit
participants’ valuation for food items. Participants were endowed with $5 that, if not spent,
would be given as bonus compensation. In each trial of this task, the participant selected a bid
($0, $1, $3 or $5) indicating the amount that they were willing to pay for the depicted food. At
the end of the experiment, the computer randomly selected one of the trials from the imaging
session to be implemented (i.e., making that trial “real”, and others not). If an item was randomly
chosen, an associated “price” was randomly generated for it (either $0, $1, $3, or $5, with equal
probability). If the participant’s bid was lower than the price, then they did not receive the food

15
(and instead kept the full $5 endowment). If the participant’s bid was equal to or greater than the
price, the price (not the bid) was deducted from their $5 endowment and the participant received
the food item (plus any remaining amount from the $5). For example, if the bid amount for an
item was $3, and the price for the item was $1, the participant would receive the food, and $4
(the amount remaining from the $5 endowment after paying the price). Participants were
instructed that the market price for all food items was approximately $6. Participants went
through several practice examples until comfortable with the procedure. The BDM auction
method is incentive-compatible in the sense that the optimal strategy for participants is to always
bid the amount that is closest to their true willingness to pay for the depicted item (Becker et al.,
1964). Participants were explicitly instructed both the mechanisms of the auction task and the
optimal strategy. A procedure calculating the expected utility based on the current design of the
BDM auction task later revealed that the optimal way to bid would be rounding down to the
nearest available bid.
Within each imaging session, participants bid on each of the 30 food items during each of
two task runs. Thus, in total, participants made a bid two times for each food item during each
session. Prior to half of trials the depicted food was first subliminally presented, but as noted
above, I saw no evidence that this impacted brain activity or behavior even when all conditions
were combined, and so investigating differences between conditions was not warranted (though
is included in Appendix for completeness). The general timeline for each trial is presented in
Figure 2. A blank white screen with fixation cross was presented for a jittered duration, with a
mean of 3.5 sec and exponential distribution. This was followed by supraliminal presentation of
the food for that trial which remained visible for 3 sec. Next the participant indicated a bid on the
item, within 3 additional seconds, by pressing the key on a keypad that corresponded to the

16
intended dollar amount. Importantly, although participants could not enter bids until after the
visual food-cue disappeared, the requirement to bid was predictable and so it is likely that
participants were formulating their bids while the food was visually presented. Therefore, I did
not separate these periods with a jitter in the design phase (they occur in temporal lock-step) and
instead treat the period beginning with cue-presentation and ending with bid entry as a general
“food valuation period” in analyses. Mappings between keys and bid amounts were randomized
across trials in order to dissociate the bid amount from the spatial information. The bid the
participant made was visually presented in the center of the screen immediately after the
participant’s keypress (feedback phase, 0.5 s). At the end of each trial, a blank white screen with
fixation cross was presented during an intertrial interval (ITI phase), again with duration jittered
using an exponential distribution with a mean of 3.5 sec.

Figure 2
Food Bid Task Trial Structure

17
Attribute-rating task (outside the MRI scanner)
Following the procedure of Suzuki et al. (2017), after all three imaging sessions were
completed, participants estimated the nutrient content for each depicted food item, see
Appendix for details. Participants were not informed that they would be providing these
estimates prior to the task. In addition, participants were asked to report their guess for the
market price for each food item. The order of the questions was randomized across participants.
MRI imaging parameters
Participants laid supine on a scanner bed, viewing stimuli through a mirror that is
mounted on the head coil. Blood oxygen level-dependent (BOLD) response was measured via an
echo planar imaging (EPI) sequence with prospective acquisition correction (PACE). Acquisition
parameters during the functional acquisition were as follows: TR=2s, TE=25ms, flip angle=90,
resolution=3mm
3
isotropic, 64 x 64 matrix in FOV=192mm. A total of 41 axial slices, each 3mm
in thickness, were acquired in an ascending interleaved fashion to cover the whole brain. The
slices were tilted to align with each individual’s anterior commissure - posterior commissure line
plane, to minimize signal dropout in the orbitofrontal cortex (Deichmann et al., 2003).
Anatomical images were collected using a T1-weighted three-dimensional magnetization
prepared rapid gradient echo with TI=900ms, TR=1.95s, TE=2260ms, flip angle=90 degrees,
resolution=1mm, 256 x 256 matrix in FOV=256mm. This high-resolution structural image was
used for alignment and normalization of each individual’s brain into standard MNI space.
Subsequently, EPI images were standardized by applying the transformation used to normalize
each participant’s high-resolution image.

18
Neuroimaging Preprocessing
Preprocessing of fMRI data was carried out utilizing several tools from the Oxford
University Centre for Functional MRI of the Brain Software Library (FMRIB) (Jenkinson et al.,
2012; Smith et al., 2004; Woolrich et al., 2009). Head movement was corrected in three
dimensions using MCFLIRT (Motion Correction using the brain software library’s Linear Image
Registration Tool) (Jenkinson et al., 2002). Six motion parameters were added into the general
linear model (GLM) to explain variance in signal related to head motion. FMRI files were pre-
processed using motion correction, high-pass filtering (100 s), and spatial smoothing with a
Gaussian kernel of full width at half-maximum = 5 mm. No spatial smoothing was used in brain
data pre-processing for all MVPA analyses (in Chapter 4). Functional data were first mapped to
each participant’s anatomical image and then registered into standard space (Montreal
Neurological Institute, MNI) using affine transformation with FMRIB’s Linear Image
Registration Tool to the avg152 T1 MNI template.

Data analyses
The effect of sucralose and glucose on participants’ willingness to pay and brain food cue
activity
Primary analyses used linear mixed-effects models (LMMs; (Baayen et al., 2008)). This
approach allowed us to take the full response patterns into account, without averaging over
individual items or conditions. F and P values were obtained using the lmerTest package
(Kuznetsova et al., 2017), which uses the Satterthwaite approximation for degrees of freedom.
Pairwise comparison using Holm-Bonferroni procedure adjustment was done using emmeans
function in the emmeans package (Lenth et al., 2018) when significant effects were shown.

19
Repeated measures ANOVA (Analysis of variance) was used when analyzing the appetite
scores.
Appetite analysis
Before analyzing the appetite for sweet food and hunger score, I first replaced the 38
missing values (7.5% of the scores were missing/incomplete and missing). Since I had no reason
to expect a relationship between the study manipulation and missing data, missing data were
imputed using predictive mean matching (pmm) method from the mice package in R (Buuren &
Groothuis-Oudshoorn, 2010), taking the averaged imputed value over 50 imputations. However,
I also carried out analyses using the data without imputation. Both one-way and two-way
repeated measures ANOVA were used to test the baseline appetite score difference and
measurement time and drink effects on Appetite scores changes from baseline. Drink type,
measurement time and the interaction between measurement time and drink type were the fixed-
effects terms.
BID task data analysis
In the primary LMM analysis of bids during the Food Bid Task, participants’ bids for the
food items were the dependent variable, the fixed effect was drink type (water, sucralose or
glucose) and the random effects included random intercept and slope for the drink type by
participant. Baseline hunger and appetite for sweet food were included as covariates in order to
minimize the impact of pre-study drink session variance in appetite (e.g., a participant that
happened to be hungrier when arriving to their water session than their glucose session). BMI
and gender were additionally included as covariates since past research has shown that BMI and
gender could be moderators of cue reactivity (Field & Duka, 2004; Tetley et al., 2009). Past NNS
usage was also included as a covariate.

20
Neuroimaging data analyses
Data were processed using the fMRI Expert Analysis Tool (FEAT) version 6.0. In
addition to six motion parameters, I included nuisance regressors for time points corresponding
to motion outliers for both models using the FSL motion outliers program
(http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSLMotionOutliers), which defined outlier time points using
the upper threshold of the 75th percentile plus 1.5 times the interquartile range. Temporal
derivatives and temporal filtering were added to increase statistical sensitivity. Inter-trial interval
periods were not modeled, and therefore provided the implicit baseline for analyses. For imaging
analyses, I was primarily interested in the effects of study drink on 1) overall activity during food
valuation, and 2) activity tracking participant bids during food valuation.
The first part of the analyses focused on the food valuation period, which began with the
presentation of the food picture for the trial and ended when a bid was recorded (see Figure 2).
The first GLM included 1) food valuation period unweighted and 2) food valuation period
weighted by bid for the food. Trials in which participants did not respond were modeled with a
separate a regressor of no interest, as were the regressors for subliminal priming (see Appendix).
Region of Interest (ROI) analyses
ROI analyses were carried out in order to examine several possible effects of the
manipulation. Below I describe these ROIs, which were directed at 1) examining response to
depicted food stimuli within brain areas known to be active during visual food-cues (food-cue
ROI, see Figure 3), and 2) examining functional connectivity with the portion of the medial
orbital frontal cortex (OFC) (previously implicated in value-tracking) that overlapped with bid-
related activity in my analysis (overlapped medial OFC ROI). The details of the second

21
overlapped medial OFC ROI and functional connectivity result, as well as the related discussion,
are included in Appendix.
Food-cue ROI
In order to examine responses in brain areas sensitive to food-cues, I utilized an a priori
ROI mask, derived based on previous published reports localizing responses to food-cues as
identified in a meta-analysis (Tang et al., 2012). This “food-cue mask” included reported
coordinate peaks from eight regions within the slice position coverage: left OFC, left ventral
striatum, left amygdala, left hippocampus, bilateral anterior insula, bilateral middle insula,
bilateral precuneus, and left postcentral gyrus. For each region (with the exception of the ventral
striatum) I drew 4-mm-radius spheres around the peak voxel reported in the meta-analysis (see
detailed information in Tang et al, 2012). For the ventral striatum, a 2-mm-radius sphere was
drawn because of its smaller size (see Figure 3).

Figure 3
Food-cue A Priori Mask

22
In ROI analyses of food-cue responses, individual signal changes (beta values from
statistical models) were extracted separately for each participant during each session. LMM
analysis was done to test the drink effect on brain signal change for the ROI described above
with drink type as a fixed effect and participant entered as a random intercept with baseline
hunger, appetite for sweet food, BMI, gender and past NNS usage as covariates.
Whole-brain analyses
In addition to ROI analyses, whole brain analyses were carried out. Group analysis with
multiple sessions for each subject were performed in FEAT using a mixed-effects model, with
FSL's FLAME1 option. In accordance with the model discussed above, group-level paired t-test
analyses were carried out (sucralose vs. water, glucose vs. water and sucralose vs. glucose) with
FEAT using a mixed-effects model. For these I utilized FSL's FLAME1 option with outlier
deweighting. All statistical maps were cluster corrected for multiple comparisons (cluster height
threshold: Z > 3.1; cluster significance: P < 0.05).

Results
Participants Characteristics
Participants were right-handed, nonsmokers, weight stable for at least 3 months, non-
dieters, not on any medication (except oral contraceptives), with normal or corrected-to-normal
vision, and no history of diabetes, eating disorders, or other significant medical diagnoses.
Twenty-eight volunteers (14 females, mean age = 25.36 ± 4.74, range 19 – 36 years old) with no
history of eating disorders, diabetes, or other major medical illnesses participated in the study
(see baseline characteristics of subjects included in the final analyses in Table 1). Seven
participants (25%) had a body mass index (BMI) index in the normal range (18.5 to <25kg/m
2
),

23
14 participants (50%) had BMI in the overweight range (25 to <30kg/m
2
) and 7 participants
(25%) had BMI in the obese range (>=30 kg/m
2
) (classification based on World Health
Organization criteria (WHO Consultation on Obesity (1999: Geneva & Organization, 2000)). All
28 participants completed two runs of the Food Bid Task on each of three days (with the
exception of one run for one participant on one session due to a time constraint).

Table 1
Characteristics of Subjects Included in the Final Analyses
Characteristic Mean ± SD or N (%)
Gender

Male 14 (50%)
Female 14 (50%)
Age (years) 25.36 ± 4.74
1

BMI (kg/m
2
) 27.61 ± 5.02
1

Ethnicity

Caucasian 6 (21%)
Black or African American 7 (25%)
Hispanic or Latino 4 (14%)
Asian 10(36%)
Other 1 (4%)
Education (degree)

Bachelor’s 18 (64%)
Graduate school level 9 (32%)
High school 1 (4%)
Past Non-Nutritive sweetener (NNS) usage

NNS User 11(40%)
NNS Non-User 17(60%)
1
Values are means± SDs (standard deviations). BMI—body mass index

Study drink rating
Participants rated both glucose (t(54) = 2.69, P = 0.03) and sucralose (t(54) = 2.62, P =
0.03) as more pleasant than water. Participants also rated both glucose (t(54) = 15.65, p<0.0001)

24
and sucralose (t(54) = 14.96, p<0.0001) as sweeter than water. Participants reported similar
ratings of the pleasantness (t(54) = 0.07, P = 0.94) and sweetness (t(54) = 0.69, P = 0.49) of the
glucose and sucralose drinks.
Appetite rating
There were no baseline difference in hunger (F(2,54) = 0.13, P = 0.88) or appetite for
sweet food (F(2,54) = 1.27, P = 0.29)) prior to drink consumption. In addition to pre-drink
appetite ratings, as noted above, post-drink appetite ratings were acquired approximately 5
minutes after drink consumption, and then again approximately 90 minutes after drink
consumption. Post drink appetite ratings were analyzed as difference scores relative to pre-drink
ratings. Two by three repeated measures ANOVAs were carried out to examine the effects of
measurement time (appetite change at first and second post-drink assessment points), and drink
effects on changes from baseline.
For hunger score change from baseline (Figure 4), a significant main effect of
measurement time (F (1,27) = 50.02, P < 0.001), a marginally significant main effect of drink (F
(2,54) = 2.80, P = 0.07) and a significant interaction effect between measurement time and drink
(F (2,54) = 4.39, P = 0.02) were found. The first hunger score change (approximately 5 minutes
after drink consumption) did not differ significantly between drinks (all P values of paired t
comparisons with Holm–Bonferroni method multiple comparison correction were larger than
0.1). The second hunger score (approximately 90 minutes after drink consumption and after
completing the Food Bid Task) increased less after glucose than after sucralose (t(88.7) = -3.08,
P < 0.01) or water (t(87.5) = -3.08, P < 0.01) and no hunger difference was found between water
and sucralose consumption (t(88.7) = 0.28, P = 0.78). For appetite for sweet food (Figure 5),

25
only a significant main effect of measurement time (F (1,27) = 12.95, P = 0.001) was observed,
with the second sweet food craving score change significantly higher than the first change.
Figure 4
Hunger Score Changes from Baseline

Figure 5
Appetite for Sweet Scores Change from Baseline

26
Food Bid Task
The Food Bid Task began approximately 55 minutes after consumption of the study
drink. Across all participants, 4.3% of the trials (221/5110) were excluded from the task analysis
due to the absence of responses within the allotted time. There was a marginally significant
(trend) drink effect on participants’ bids for depicted food (F (2, 18.55) = 3.34, P = 0.058). Post-
hoc comparisons indicated that this overall marginally significant trend was driven by marginally
significant trends towards lower bids after glucose relative to water (z = -2.27, P = 0.07) as well
as after sucralose relative to water (z = -2.22, P = 0.07). No significant difference was found
between sucralose and glucose (z = 0.06, P = 0.95), each with Holm-Bonferroni adjustment for
multiple comparisons. The mean bid in the water condition was $1.93, in the glucose condition
was $1.68 (13.0% lower than water condition), and in the sucralose condition was $1.61 (16.6%
lower than water condition). BMI was positively associated with willingness to pay (Beta= 0.06,
P = 0.03). No effect of gender, baseline hunger, baseline appetite for sweet food or past NNS
usage was observed.
In a secondary (post hoc) analysis, I removed items for which participants bid “0” on all
the 6 experimental runs (3 drinks, with 2 runs per drink), treating these as missing values. I
reasoned that these items may have had no value (perhaps the participant did not like or want to
eat the food even if free) and so their inclusion could have reduced power to detect changes
between conditions. With these items removed (mean of 3.0 out of 30 food items per participant)
the drink effect on bids was significant overall (F (2, 20.29) = 3.70, P = 0.04).
The effect of study drink on food-cue network ROI activity during food valuation
Beta values were extracted from the a priori food-cue ROI (Tang et al., 2012) discussed
above (all clusters combined). A significant drink effect was found for beta-values within the

27
network during food valuation (F(2,32.19) = 4.57, P = 0.02). In post-hoc comparisons, beta-
values were observed to be significantly lower after sucralose relative to water (t (23.3) =-2.93, P
= 0.02). Beta-values did not differ significantly between glucose and water (t(22.8) = 1.69, P =
0.21) nor between glucose and sucralose (t(24.1) = 1.65, P = 0.21). Pre-drink appetite for sweet
food (which was included as a covariate in the model) significantly predicted beta values overall,
with increased signal when participants had higher baseline appetite score for sweet food (  =
0.24, P = 0.03). No significant effect of BMI, Gender or past NNS usage was found. See figure 6
for mean beta-values within the food-cue mask in each of the drink conditions.

Figure 6
Signals Change during Food Valuation in Food-cue Region of Interest (ROI) by Drinks

Note. * P value < 0.05. CI refers to confidence interval.

28
In order to allow visual comparison of the main effect of glucose and sucralose (each
relative to water) on primary dependent variables (bids and brain signal within the food-cue ROI)
along with appetite scores, I first normalized all dependent variables as z-scores, and then plotted
the 95% confidence intervals for all comparisons. These are presented in Figure 7, with order
matching the temporal sequence of the acquisition of the measure. Each of these main effects is
described above, but the conversion to z-scores and the ordering in temporal sequence allows
main effects to be better visualized. As can be seen, 55 minutes post glucose consumption
(relative to water), food bids were significantly lower, as were ratings of hunger 90 minutes post
glucose consumption. In addition, again relative to water, food bids were lower 55 minutes post
sucralose consumption, as was activity recruited within the food-cue network during the study
task. The only significant difference between glucose and sucralose was that the second hunger
score (approximately 90 minutes after drink consumption) increased less after glucose than after
sucralose (t (88.7) = -3.078, p < 0.01).

29
Figure 7
95% Confidence Intervals for the Effect of Glucose and Sucralose (relative to water) on
Standardized (z-score) Dependent Variables

Note. Four bars in each drink condition refer to hunger score (about 5 min after drink ingestion), willingness to pay
for the food (about 55 min after drink ingestion), brain activity during food valuation within the a priori food cue
mask (about 55 min after drink ingestion), and final rating of hunger (about 90 min after drink ingestion). In general,
intervals that do not intersect 0 indicate P < 0.05 for comparison with water condition. However, these confidence
intervals do not reflect Holm–Bonferroni adjustment used in analyses.

Whole Brain Analyses of Activity during Food Valuation Period
Overall increase in brain activity during food valuation (relative to rest) across drinks
was, as expected, quite extensive (see Figure 8 ). It included the network of regions previously
linked to food-cues: lateral OFC, dorsolateral prefrontal cortex (dlPFC), left ventral striatum, left
amygdala, left hippocampus, bilateral anterior insula, bilateral middle insula, bilateral precuneus,
and left postcentral gyrus), and visual cortex, as well as the fronto-parietal network which is

30
generally active during decision-making tasks (Andersen & Cui, 2009; Chaisangmongkon et al.,
2017; Chand & Dhamala, 2017).

Figure 8
Regions with Increased Signal during Food Valuation

Drink effects on whole brain activity associated with food valuation
In secondary whole brain analyses, I compared neural activity during food valuation for
each drink condition (glucose vs. water, sucralose vs. water, and glucose vs. sucralose). The only
cluster in which a significant activity difference between the glucose and water condition (see
Figure 9) was present was a cluster in the left parietal lobe (partially overlapping the postcentral
gyrus) in which activity after glucose was significantly diminished relative to water. This cluster

31
overlapped the network of regions that were generally active during food valuation . For the
comparison of sucralose and water (see Figure 10), significantly lower activity was observed
after sucralose in a set of regions that included left dlPFC, visual cortex, frontal gyrus, and
cingulate (mostly posterior), precuneous, supplementary motor cortex, and frontal operculum,
each of which overlapped with the areas showing general increase in activity during food
valuation. No significant differences were observed in the comparison between glucose and
sucralose conditions.

Figure 9
Regions with Significant Decreased Activity after Glucose (relative to water)

32

Correlation between nutritive attributes and bid difference between drinks
The difference in bids for each food item between different drink conditions are
presented in Table 2. No effect of gender (F(1,18.97) = 0.71, P = 0.41) and BMI (F(1,20.49) =
0.71, P = 0.12) was observed.

Figure 10
Regions with Significant Decreased Activity after Sucralose (relative to water)

33
Table 2
Bid Difference in Cents Between Drinks

(Bid difference from overall mean)
Food item Mean Bid Water
Sucralose Glucose
Sundae $2.33
+$0.56
-$0.24 -$0.33
Filled Chocolates $1.73
+$0.51
-$0.13 -$0.39
Cheese and Cold Meat Platter $1.73
+$0.31
-$0.28 -$0.02
Apple $1.36
+$0.30
-$0.27 -$0.04
Waffle with Whipped Cream $2.23
+$0.39
-$0.16 -$0.23
Sushi $2.05
+$0.32
-$0.20 -$0.13
Tomatoes $0.84
+$0.25
-$0.25 -$0.01
French Fries $2.33
+$0.34
-$0.14 -$0.20
Gummi Candy and Licorice
Mix
$1.15
+$0.27
-$0.20 -$0.07
Bowl of Rice $1.16
+$0.28
-$0.18 -$0.11
Pizza (With Mushrooms) $3.22
+$0.34
-$0.12 -$0.22
Crackers $1.10
+$0.24
-$0.17 -$0.06
Nuts (Cashews) $1.76
+$0.20
-$0.19 -$0.02
Cheese Platter $1.84
+$0.21
-$0.17 -$0.05
Roast Beef $3.03
+$0.15
-$0.22 +$0.06
Pizza (With Salami) $3.33
+$0.24
-$0.12 -$0.11
Doughnut / Donut Jam $1.84
+$0.25
-$0.06 -$0.20
Salad Plate $2.02
+$0.13
-$0.17 +$0.04
Loaf of Bread $1.29
+$0.06
-$0.20 +$0.13
Popcorn $1.46
+$0.17
-$0.08 -$0.09
Crisp Bread $0.80
+$0.08
-$0.16 +$0.07

34
Chocolate Muffin $1.75
+$0.10
-$0.12 +$0.01
Broccoli $0.92
+$0.16
$0.01 -$0.17
Peanuts $1.04
+$0.02
-$0.07 +$0.06
Strawberries $2.83
-$0.01
-$0.08 +$0.10
Banana $1.19
-$0.01
-$0.07 +$0.07
Toast $1.33
-$0.05
-$0.11 +$0.17
Opened Chips Bag $1.44
+$0.09
+$0.04 -$0.12
Croissants $2.28
-$0.10
+$0.03 +$0.06
Green Asparagus $0.82
-$0.09
+$0.16 -$0.06

Finally, I carried out correlational analyses directed at identifying attributes (based on
available normative data of perceivers’ judgments for the stimulus set) of the 30 food pictures
that were associated with how glucose and sucralose impacted bids. Although not corrected for
multiple comparisons, there was some evidence that glucose (relative to water) led to reduced
bids more for foods high in sugar, high in calories, and high in fat. While a similar pattern was
observed with regard to sucralose, correlation coefficients were generally lower and non-
significant (see Table 3).

35
Table 3
Association between Bid Difference of Drinks and Food Attributes
Food Attribute
Water Bid –
Sucralose Bid
Water Bid –
Glucose Bid
Sucralose Bid –
Glucose Bid
palatability 0.12 0.21 0.18
healthiness -0.28 -0.41
-0.3
(P = 0.11)
familiarity -0.03 0.01 0.07
fat 0.26
0.4
2
(P = 0.03)
0.31
1

(P = 0.097)
vitamin -0.24
-0.33
1
(P = 0.08)
-0.22
sodium 0.03 0.07 0.08
calorie
0.35
1

(P = 0.058)
0.44
2
(P = 0.02)
0.25
carb 0 0.1 0.18
sugar
0.3
(P = 0.11)
0.38
2
(P = 0.04)
0.23
protein 0.12 -0.06 -0.27
1
P value <0.1.
2
P value <0.05.

36
Chapter 3: Food Valuation and Nutrient Tracking
Correlational analyses suggested that food devaluation after sucralose and glucose
consumption might be associated with the devaluation of certain nutrients such as sugar.
Specifically, food value was decreased more for high sugar items after sweet preloads ingestion,
suggesting that a devaluation in sugar was underlying food devaluation. This finding was
consistent with my hypothesis that glucose and sucralose could devalue food through the
devaluation of carbohydrates (though it should be noted that reductions in bids after glucose and
sucralose consumption were not significantly correlated with carbohydrate content). I am
especially interested in sucralose's effect on food devaluation since it does not contain any
calories. Based on Suzuki’s valuation construction theory, I hypothesized that a sucralose
preload could influence food value 1) by changing the constituent signaling in the sector of the
lateral OFC sensitive to the presence of carbohydrate nutrients; or 2) by changing the
connectivity between this region and the medial OFC. These two possibilities are not mutually
exclusive and could both contribute to food devaluation. It is worth noting that carbohydrate
tracking, instead of sugar tracking, was evaluated in my study for two reasons. First, the
orthogonalized carbohydrate and fat content allowed me to find brain correlates of both nutrients
with higher statistical power. Sugar content, in contrast, likely covaried with other nutrients.
Second, in prior work using a similar paradigm, sugar content was not as strong a predictor of
food value as carbohydrate content (Suzuki et al., 2017). Therefore, I will focus on carbohydrate
tracking in the brain. However, brain tracking of sugar (subjective ratings of sugar) was also
investigated for exploratory purposes. In the next part, I first evaluated the hypothesis that food
bids are related to nutrient information, and then I evaluate study hypotheses regarding the
relationship between nutrient and value tracking in the brain using MVPA, univariate analysis,

37
and functional connectivity tests. Results comparing all study drinks are presented, though I
primarily focus on the effect of sucralose on brain activity related to food valuation and nutrient
tracking.

38
Chapter 4: Food Value Construction and the Effect of Sucralose and
Glucose on Brain Nutrient Tracking
Data analyses
Linear regression models predicting food value with objective elemental nutritive attributes
I conducted a series of linear regression models to select the best model for predicting
food value, as a function of the quantity of constituent nutrients (carbohydrate, fat, and protein).
The prediction accuracy of models including all combinations of these predictors was assessed
using leave-one-item-out cross-validation, and the performance of seven models (seven
combinations) was compared. Regression analysis was firstly run with one item out of 30 left out
for each model and each participant. The predicted food value of the left-out item was then
calculated based on the regression coefficients obtained from the above linear regression. The
previous steps were repeated for each of the 30 food items. The performance of each model was
calculated by averaging the correlations between predicted and actual food values among all
participants.
The same linear regression analyses and model selection procedure was implemented to
explore whether subjective ratings of nutrients and overall caloric content could predict food
value better than actual nutritive attributes. Paired t-tests were used to compare the performance
difference of the best prediction model in each group.
Multivariate pattern analysis (MVPA)
Classification analyses were conducted to examine whether food value and nutrient
information could be decoded from brain signal patterns. Searchlight analyses were used to

39
localize where in the brain the value and nutrient information can be decoded. All analyses
methods were adapted from Suzuki’s study (Suzuki et al., 2017).
A GLM was constructed for each run, and participants’ brain responses to each food item
were entered into the pattern analysis as classification samples. The GLM model was adapted
from Suzuki’s study and included the following exploratory variables (EVs): thirty EVs of the
food items during the food-viewing period (3 s), four EVs of different bid prices during the food
bidding period (reaction time), feedback EV during the feedback period (0.5 s), keypress EV (0
s), and missing trials EV. Trials in which participants did not respond, the subliminal priming
period, and six motion parameters were modeled as regressors of no interest.
Classification analysis in decoding food value and nutrient attribute factors
Classification labels of food value and nutrient attributes (actual carbohydrate, fat, and
protein content of the food) were first created by splitting the sample into “high” (above median)
and “low” (below or equal to median) values. The median-split was based on all samples across
runs within each participant. A linear support vector machine (SVM) classifier (Chang & Lin,
2011), as implemented in PyMVPA (Hanke et al., 2009), was trained to classify food value and
nutrient information. Classification accuracy was assessed with a leave-one-run-out cross-
validation for each of the six runs and the results were averaged for six runs. A balancer function
in PyMVPA was used to ensure that the number of samples within each label was equalized for
each run. This classification analysis procedure was repeated 1000 times to obtain the
classification accuracy for that participant.
The classification analysis was performed within anatomically defined ROIs of medial
and lateral OFC (see Figure 11)), as defined by the automated anatomical labeling (AAL)
database (Tzourio-Mazoyer et al., 2002). To determine whether classification accuracy was

40
above chance level, mean classification accuracy across all participants in each ROI was
compared with the chance level (50%) using one-tailed one-sampled t-tests. Though the study
focused on decoding actual macronutrients (carbohydrate, fat, protein) in the brain, decoding was
also performed for subjective estimates of sugar content in the depicted food, since actual sugar
content information was not available.

Figure 11
Anatomically Defined Medial (blue) and Lateral (red) OFC based on AAL Database

Additional classification analysis in decoding nutrient attributes
Next, to evaluate whether nutrient information can be decoded independent of food value,
I regressed out the effects of value from nutrient attribute for each participant and each run and
then calculated the residuals. Classification labels were created based on the residuals with the
same cross-validation procedure described above. The mean classification accuracy across all
participants, after each drink consumption, was compared with the chance level (50%) using

41
one-tailed one-sampled t-tests. This was an important step because higher carbohydrate items
tended to be valued more highly by participants.
Searchlight analysis
Searchlight analysis was conducted with a radius of three voxels within the ROIs in the
lateral and medial OFC. A SVM classifier was trained to classify food value and nutrient
information with a leave-one-run-out cross-validation in every possible sphere (searchlight)
within the ROIs. The significance of clusters in the overlapped searchlight accuracy maps was
evaluated using nonparametric permutation testing with FSL’s Randomise tool (Winkler et al.,
2014), which models a null distribution of expected accuracies. The searchlight accuracy maps
were thresholded using threshold-free cluster enhancement (TFCE)(Smith & Nichols, 2009).
Psychophysiological interaction (PPI) analysis
A GLM was constructed containing EVs indicating the food-viewing period (3 s), the
food-bidding period (reaction time), feedback period (0.5 s), and timing of the keypress (0 s).
The EVs of the food-viewing period depicting food value and nutrient information were
orthogonalized to the food-viewing period EV. Trials in which participants did not respond, the
subliminal priming period, and six motion parameters were modeled as regressors of no interest.
Brain signal changes were extracted from the nutrient (total carbohydrate) tracking ROI
identified by the searchlight analysis (spheres with a radius of 3 voxels centered at the peak
voxel) during the food-viewing period for each participant during each session.
A second GLM was constructed for the PPI analysis, including all the EVs in the
aforementioned GLM model. The EV indicating the food-viewing period was the psychological
factor. The time-series of mean activity in the carbohydrate tracking ROI was the physiological
EV. The interaction between this time series and the food-viewing EV served as the test

42
regressor for the presence of a psychophysiological interaction (PPI factor). For each participant,
parameter estimates of the PPI factor were estimated at the medial and lateral OFC value
tracking ROIs identified in the searchlight analyses (spheres with a radius of 3 voxels centered at
the respective peak voxels).
To examine whether the functional connectivity was significant within the carbohydrate
tracking and value tracking regions identified in medial and lateral OFC, mean parameter
estimates of the PPI factor within each value-tracking ROI were extracted for each participant,
and then the samples values were compared with 0 using one-tailed one-sample t-tests.
Drink effect on mean brain activity in carbohydrate tracking region
LMM was used to identify the effect of drink type on brain signal change in the
carbohydrate tracking region. Brain signal change extracted from the first GLM model in the
carbohydrate tracking ROI was our dependent variable, the fixed effect was drink type (water,
sucralose, or glucose), and the random effect included a random intercept for each participant.
Baseline hunger, baseline appetite for sweet food, BMI, gender, and NNS consumption history
were included as covariates.
Drink and past NNS usage effect on decoding accuracy of carbohydrate
To examine whether drink preloads could interact with past NNS usage to influence
decoding accuracy of carbohydrates, classification performances of carbohydrates tracking in
lateral OFC were compared between different drink sessions. The same classification analysis as
mentioned previously was used except that only two runs (instead of six runs) of data were
included using the procedure of leave-one-run-out cross-validation.
I first examined whether NNS users and non-users rated the sweetness and pleasantness
of preloads differently. This step could also imply whether past NNS usage could potentially

43
lead to perceptual differences of study drinks. LMM was conducted with sweetness or
pleasantness ratings as the dependent variable, and the fixed effects were drink type (water,
sucralose, or glucose), past NNS usage (NNS user or NNS non-user), and interaction between
drink type and past NNS usage. The random effect included random intercept by participant.
Baseline hunger, appetite for sweet food, gender, and BMI were included as covariates.
Next, in the LMM analysis of decoding accuracy during the Food Bid Task, participants’
carbohydrate decoding accuracy was the dependent variable, and the fixed effects were drink
type (water, sucralose, or glucose), past NNS usage (NNS user or NNS non-user), and interaction
between drink type and past NNS usage. The random effect included random intercept by
participant. Baseline hunger, appetite for sweet food, gender, and BMI were included as
covariates. Since the sweetness and pleasantness ratings could possibly imply participants’
baseline perception or sensitivity to different features of the same study drink, they were also
included as covariates.
Drink effect on functional connectivity between carbohydrate tracking region and medial
OFC value tracking region
LMM was used to identify the effect of drink type on the functional coupling between the
carbohydrate tracking region and the value tracking region in medial OFC. Parameter estimates
of the PPI regressor estimated at the targeted ROI were the dependent variable, the fixed effect
was drink type (water, sucralose, or glucose), and the random effect included a random intercept
for participant. Baseline hunger, baseline appetite for sweet food, BMI, gender, and past NNS
usage were included as covariates.
Exploratory analysis of the effect of drink type and weight status on willingness to pay
The data is underpowered to detect the effect of participants’ weight status (obese vs.

44
non-obese) and the interaction effect of drink type and obesity; thus, the following analysis is for
exploratory purpose only. Linear mixed models were used to identify the interaction effect of
drink type (water, glucose, and sucralose) and weight status (obese and non-obese) on
willingness to pay. Participants’ bids for the food items were the dependent variable, and drink
type, weight status, and the interaction between weight status and drink type were the fixed-
effects terms. Drink type was also a random slope nested within a random intercept participant
term, taking into account intra-individual variability. Baseline hunger, baseline appetite for sweet
food, gender, and past NNS usage were included as covariates.

Results
Linear regression models predicting food value with objective elemental nutritive attributes
A series of linear regression analyses were performed to predict food value with nutritive
attributes of the food's actual carbohydrate, fat, and protein content. The objective and subjective
ratings of the nutrients (carbohydrate, fat, and protein) and total calories were highly correlated
(See Appendix). The prediction accuracy was calculated using leave-one-item-out cross-
validation, and all combinations of nutritive attributes were assessed and compared to select the
model with the best prediction accuracy. When using the actual nutritive information of the food,
the best model with the highest prediction accuracy included carbohydrates as the only predictor,
and the prediction accuracy was significantly higher than the chance level (mean = 0.44, t(25) =
9.72, P < 0.0001 ). Even after Bonferroni correction of multiple comparisons, the model could
still predict food value with better than chance accuracy. The model performance was calculated
by averaging the correlations between the predicted food value and actual food value among all
participants based on the leave-one-item-out cross-validation procedure. The predicted food

45
value of the left-out item was calculated based on the regression coefficients obtained from the
linear regression using the other 29 items. The performance of models using subjective ratings of
nutritive attributes as predictors were also reported in Table 4.

Table 4
Model Performance

Rank Exploratory variables Performance
Linear regression 1 Total carbohydrate 0.37
Objective total
nutrient
2 Total fat 0.27

3 Total carbohydrate, fat 0.23

4 Total protein 0.21

5 Total protein, fat 0.19

6 Total protein, fat, carbohydrate 0.18

7 Total protein, carbohydrate 0.16
Linear regression 1 Subjective carbohydrate 0.44
Subjective
estimates
2 Subjective protein 0.43

3 subjective protein, carbohydrate 0.37

4 Subjective carbohydrate, fat 0.31

5 Subjective protein, carbohydrate, fat 0.3

6 Subjective fat 0.3

7 Subjective protein, fat 0.28

The prediction models predicting food value with the food's subjective or actual caloric
content were also assessed, and no significant difference in performance accuracy was found.
See detailed comparison results in Appendix.

46
Decoding subjective value in the OFC
Using MVPA, I found that subjective value could be decoded within the anatomically
defined medial (mean = 0.53, t (25) = 1.93, P = 0.03) and lateral (mean = 0.54, t(25) = 2.53, P <
0.01) OFC better than chance (50%) across all drink sessions during the food-viewing period
(see Figure 12).

Figure 12
Lateral and Medial OFC Decoded Food Value for 26 Participants

Note. In each box and whisker plot, significant results (P < 0.05) were indicated with *. T test vs. 50% (LOFC
(lateral OFC): mean = 0.54, t(25) = 2.53, P < 0.01; MOFC (medial OFC): mean =0.53, t(25) = 1.93, P = 0.03)

47
Searchlight analysis revealed that subjective value (high vs. low) could be decoded in
both the medial and lateral OFC (Figure 13, P < 0.05, family-wise error rate corrected)

Figure 13
Regions that Represent Subjective Value Information of the Food

Note. Medial OFC is blue, Lateral OFC is green. Peak voxels: Montreal Neurological Institute coordinates (MNI):
MNI x, y, z = -30, 60, -10 and 24, 58, -8 (P < 0.05) for lateral OFC (Green); MNI x, y, z = 6, 60, -4 for medial OFC
(blue). Decoding accuracy maps obtained from the searchlight analyses, thresholded at P < 0.05 (Family-wise error
rate corrected).

Decoding nutrient information in the OFC
The same MVPA procedure was applied to evaluate whether nutrient information (actual
fat, carbohydrate, and protein) could be decoded in medial OFC or lateral OFC during the food-
viewing period. It was found that only carbohydrate information could be decoded in the OFC,
see detailed testing results for fat and protein in Figure 14. As expected, carbohydrate
information could only be detected above chance level in lateral OFC (mean = 0.519, t(25) =
3.58, P < 0.001), but not in medial OFC (mean = 0.505, t(25) = 0.66, p = 0.26). Importantly,

48
carbohydrate information could still be decoded in lateral OFC (mean = 0.518, t(25) = 3 .01, P =
0.003) after regressing out the food value (bids) from carbohydrate information. This result
indicated that lateral OFC could decode carbohydrate information independent of value.
Exploratory analysis implied that sugar information (based on participants’ ratings) could not be
decoded in the lateral OFC (mean = 0.496, t(25) = -0.40, P = 0.65) nor the medial OFC (mean =
0.498, t(25) = -0.24, P = 0.59). Similarly, Actual fat and protein information could not be
decoded in either lateral or medial OFC (see Figure 14), and this was also true after regressing
out value.

Figure 14
Carbohydrate (Carb) Information could be Decoded in Lateral OFC but Not Medial OFC

Note. In each box and whisker plot, significant results (P < 0.05) were indicated with *. T test vs. 50% : Fat/Lateral
OFC (mean = 0.496, t(25) = -0.38, P = 0.65); Fat/Medial OFC (mean = 0.499, t (25) = -0.18, P = 0.57);
Protein/Lateral OFC( mean = 0.499, t(25) = -0.20, P = 0.58); Protein/Medial OFC (mean = 0.497, t(25) = -0.56, P =
0.71).

Using the same searchlight procedure described above, it was confirmed that
carbohydrate information during the food-viewing period could be decoded from activity within
subregions of the lateral OFC (Figure 15. P < 0.05, family-wise error rate corrected).

49
Figure 15
Region within which Carbohydrate Content of Food can be Decoded

Note. Peak voxels: Montreal Neurological Institute coordinates (MNI): MNI x, y, z= -40, 52, 6. Decoding accuracy
maps obtained from the searchlight analyses, thresholded at P < 0.05 (Family-wise error rate corrected).

Functional connectivity between carbohydrate tracking and value tracking regions in OFC
To identify regions in which value tracking may have been affected by carbohydrate
tracking activity, functional connectivity between the carbohydrate tracking region and two value
tracking regions in lateral and medial OFC were evaluated separately. Brain activity in the
carbohydrate tracking region was functionally coupled with the value tracking region in medial
OFC (mean = 0.10, t(151) = 1.95, P = 0.03), but not significantly in lateral OFC (mean = 0.04, t
(151) = 1.37, P = 0.09).
Drink comparison
To investigate whether drink preloads could influence food valuation by changing
carbohydrate signaling in lateral OFC, both the mean brain signal change (see Figure 16), and

50
the prediction accuracy of carbohydrate information in lateral OFC (see Figure 17 and Figure 18)
were compared between drink conditions.
Drink condition comparison of activity within carbohydrate tracking region
When brain mean signal change was evaluated (see Figure 16), a significant drink effect
on brain signal change in the carbohydrate tracking region was observed (F(2,119.50) = 3.37, P =
0.04), with significantly less signal change after sucralose consumption compared to water
(t(117) = -2.59, P = 0.03). No difference was found between water and glucose conditions
(t(122) = 1.54, P = 0.25), or between glucose and sucralose conditions (t(120) = 1.05, P = 0.30),
after adjusting for baseline hunger, appetite for sweet food, Gender, BMI and past NNS usage.

Figure 16
Brain Signal Change in Carbohydrate Tracking Region

51
Drink condition comparison of carbohydrate decoding accuracy
Paired T-tests were used to compare carbohydrate decoding accuracy in lateral OFC
between different drink conditions (see Figure 17). The decoding accuracy after water ingestion
was significant higher compared with glucose (t(22) = 2.17, P = 0.04). No significant difference
was observed between water and sucralose conditions (t(22) = 1.25, P = 0.22), or between
glucose and sucralose conditions (t(23) = -0.72, P = 0.48).

Figure 17
Carbohydrate decoding accuracy in lateral OFC

Carbohydrate decoding accuracy in lateral OFC, taking into account past NNS usage
Participants' ratings of pleasantness and sweetness of drinks were examined first. A
significant drink effect (F (2,40.35) = 133.79, P = 2e-16) and a marginally significant interaction
effect between drink type and past NNS usage were observed (F (2, 39.78) = 3.02, P = 0.06) in
participants’ sweetness ratings of the drink preloads. NNS users’ sweetness ratings of water were

52
lower than NNS non-users (t (44.9) = -2.18, P = 0.03). For pleasantness score, there was only a
significant drink effect (F (2,46.63) = 4.14, P = 0.02), with higher pleasantness ratings for
glucose (t(46.9) = 2.41, P = 0.04) and sucralose (t(45.5) = 2.58, P = 0.04) relative to water. No
main effect of past NNS usage, or significant interaction between drink type and past NNS usage
was found on pleasantness ratings.
When comparing prediction accuracy of carbohydrate information in lateral OFC (see
Figure 18), a marginally significant drink effect (F (2,54) = 3.12, P = 0.05) and a significant
effect of past NNS usage (F (1,54) = 4.99, P = 0.03) were found. The decoding accuracy of
carbohydrate was marginally significantly lower in glucose (t (53.8) = -2.43, P = 0.06) and
sucralose (t (54) = -2.02, P = 0.09) conditions relative to water. No difference was found
between glucose and sucralose (t (39.1) = 1.00, P = 0.32). In addition, NNS users had a higher
decoding accuracy than NNS non-users in general (t (17.5) = 2.22, P = 0.04). These results were
after adjusting for baseline hunger, appetite for sweet food, Gender, BMI, the pleasantness, and
sweetness ratings of the preload drinks.

53
Figure 18
Carbohydrate Decoding Accuracy in Lateral OFC

Note. SE stands for standard error.

Drink condition comparison of connectivity between carbohydrate tracking region and value
tracking regions
In order to evaluate whether drink preloads could impact food value signaling by
influencing the functional connectivity between the carbohydrate tracking region of the lateral
OFC and the value tracking region within the medial OFC, functional connectivity was
compared between drinks, and no significant drink effect was detected. Beta values in functional
connectivity between medial OFC and the identified carbohydrate tracking region did not differ
between drinks (F(2,115.55) = 0.92, P = 0.40) after adjusting for baseline hunger score, baseline
appetite for sweet food, gender, BMI, and past NNS usage.
Interaction effect of obesity and drink type on willingness to pay
Participants’ weight status appeared to interact with the drink type in models predicting

54
willingness to pay for food items. A significant main effect of drink type (F(2,20.41) = 5.93, P =
0.01) and a marginally significant interaction effect (F(2,20.29) = 3.08, P = 0.07) of drink type
and weight status (non-obese vs. obese) was found on food valuation (see Figure 19).
Specifically, the observed willingness to pay drink condition difference was mainly driven by
obese people’s higher bids after water. Obese people were willing to pay more after drinking
water (mean difference = 0.75, z = 2.29, P = 0.02) than non-obese people. A larger effect of
sucralose and glucose on decreasing food value was found in obese people compared to non-
obese individuals. Obese people had lower bids after glucose (mean difference = 0.72, z = -3.18,
P = 0.004) and sucralose (mean difference = 0.63, z = -2.41, P = 0.03) consumption (vs. water).
No significant effect of baseline hunger and appetite for sweet food, gender, BMI, and past NNS
usage was observed.

Figure 19
Interaction Effect of Obesity and Drink Type on Willingness to Pay

Note. SE stands for standard error.

55
Chapter 5: General Discussion
My study aimed to investigate the effects of acute ingestion of glucose and of sucralose on
subjective hunger, food valuation, and associated brain activity. In interpreting the results, it is
important to keep in mind both the baseline metabolic state (overnight abstinence) and the timing
of assessments relative to consumption. I will first discuss the results of brain food-cue responses
and food valuation, and then discuss the results of MVPA of brain nutrient tracking and the
related value construction process.
Food cue activity and food valuation
Primary Findings Related to Acute Glucose Consumption
Relative to water, glucose intake was associated with 1) significantly reduced change in
subjective hunger 90 minutes after consumption (t(87.5) = -3.08, p < 0.01) but not 5 minutes
after consumption (t(87.5) = 0.74, P = 1)), marginally significant reduction in monetary bids on
foods approximately 55 minutes after consumption (z = -2.27, P = 0.07), and reduced signal
change during food valuation within a small cluster of the left parietal cortex (which was part of
the extensive map in which signal was elevated during food valuation with all conditions
combined). Within the food-cue ROI, signal change was not significantly attenuated after
glucose (t (22.8) = -1.69, P = 0.21) though the pattern of results was in the expected direction.
The general pattern of findings in the glucose consumption condition is consistent with
prior research. One study that directly examined the food-cue reactivity after glucose ingestion
revealed decreased in-scanner food-cue induced hunger and desire for food scores compared
with water (Luo et al., 2015). Participants in that study also indicated lower amounts of money
that they were willing to pay for food after glucose ingestion. Moreover, a meta-analysis showed
that low level of blood glucose increases the willingness to pay for food (Orquin & Kurzban,

56
2016). Evidence has shown that ingestion of glucose elicits a decrease in food-cue reactivity in
brain regions associated with hunger (Page et al., 2013; Smeets et al., 2005; van Opstal et al.,
2019). While the area within the left parietal lobe in which reduced activity post-glucose
(relative to water) was observed is not implicated in appetite signaling, it has been implicated in
visual attention orientation (Fan et al., 2002) and food reward during visual food-cue
presentation (Yokum et al., 2011). Indeed, attention related activity in the parietal lobe in
response to food stimuli has been reported to be greater among individuals with higher BMI.
Therefore, the observed attenuation within the parietal cortex after glucose consumption could
reflect a decrease in attention to food-cues.
Primary Findings Related to Acute Sucralose Consumption
Relative to water, sucralose intake was associated with marginally significant reduction
in monetary bids on foods approximately 55 minutes after consumption (z = -2.22, P = 0.07), and
reduced activity within the a priori food-cue ROI (t(23.3) = -2.93,P = 0.02) and several
additional regions including the visual cortex, dorsolateral prefrontal cortex and posterior
cingulate. Collectively, the findings provide evidence that at approximately 1-hour post
consumption, sucralose reduces CNS activity associated with food valuation. This acute appetite
suppression effect may contribute to aforementioned recent evidence that consumption of
sucralose sweetened beverages can result in decreased energy intake throughout a 12-week
intervention and reduced BMI after the intervention (compared with baseline) in overweight and
obese individuals (Higgins & Mattes, 2019).
Although the mechanism of the observed diminished food-cue ROI activity after acute
sucralose consumption is not clear, it is likely that receptors in the mouth or gut that are normally
sensitive to sugars play a role. Sweet taste perception of both sugars and NNSs is peripherally

57
mediated by the T1R3 and T1R2 receptors on the tongue (Nitschke et al., 2006), though only the
T1R3 appears to be sensitive to sucralose (Damak et al., 2003). Sucralose has high binding
affinity (lower dissociation constant) to T1R3 receptor compared to glucose (Nie et al., 2005).
The upper gut has receptors that respond to sweetness, leading to satiety hormone release. It has
been found that both sucralose and glucose can induce GLP-1 release in a human L cell line (in
vitro)(Jang et al., 2007). Study in vivo, however, demonstrated no effect of oral sucralose on
GLP-1 secretion (Ford et al., 2011; Nichol et al., 2018; Yunker et al., 2020). Though NNSs were
initially considered to be without glycemic (Nehrling et al., 1985) and metabolic effects, more
recent evidence suggests that NNSs may have metabolic effects (Pepino, 2015).
Comparison of Acute Sucralose vs. Acute Glucose Consumption
Contrary to the expectation that glucose consumption relative to sucralose consumption
would cause greater reduction in food bids and attenuation in brain activity associated with food
decision making, no significant differences were observed. It is possible that differences would
have been observed had the Food Bid Task been administered at a different time point. It is
worth noting that appetite ratings, were not different between these conditions five minutes after
the study drink consumption, but were significantly lower during the glucose session 90 minutes
after study drink consumption (see timeline in Figure 1).
MVPA analyses
A correlational analysis indicated that the sugar content of foods presented during the Bid
Task was marginally significantly correlated (P = 0.11) with WTP drop after sucralose (relative
to water) and significantly correlated (P = 0.04) with WTP drop after glucose (relative to water).
In other words, the sweetened study drinks tended to reduce bids more for foods with higher
sugar content. This result is in line with the hypotheses that food devaluation after sucralose and

58
glucose might partly be from the devaluation of carbohydrates. Food devaluation after sucralose
was of special interest because it implied a potential acute beneficial effect of consuming NNS.
Carbohydrate rather than sugar (a specific sub-category of carbohydrate) content of available
foods was used in follow-up analyses because carbohydrate and fat contents were orthogonalized
within all food items. However, an exploratory analysis using subjects’ sugar ratings of the food
was also conducted to directly examine whether drink preloads, especially sucralose, could
impact sugar tracking in the brain. It was revealed that participants’ beliefs of the sugar content
of the food couldn’t be decoded in my study, even after regressing out food value from the sugar
value in MVPA. In Suzuki’s study, including sugar instead of carbohydrates as one of the
predictors of food value reduced the accuracy of the prediction model (Suzuki et al., 2017). In
light of these results, it appears that, at least given the set of stimuli used in this study, the effects
of sugar and of NNS are more evident with regard to signaling linked to the carbohydrate content
of visually presented foods than for the more specific sugar content of those foods. While it is
possible that brain localization of nutrient tracking for sugar per se (and not carbohydrates
generally) will be identified in future work, I did not observe it in the present study. Therefore, I
interpret carbohydrate tracking as reflecting the brain tracking of general carbohydrates,
inclusive of the sugar content of the food.
Food valuation and nutrient tracking
To evaluate whether the devaluation of food is from the devaluation of carbohydrates, I
first attempted to verify that food value was associated with elemental nutrient attributes of the
food. I have demonstrated that food bids could be predicted from the actual carbohydrate content
of the food when using behavioral data. And MVPA showed that both medial and lateral OFC
could decode subjective value information and only lateral OFC could decode nutrient

59
information (specifically carbohydrate content). Significant functional coupling during the food-
viewing period was observed between the identified carbohydrate-tracking region and the
identified value-tracking region in the medial OFC. These results are consistent with the
possibility that information coded in lateral OFC, which seems to at least in part be information
about carbohydrate content, influences value signals in medial OFC. In Suzuki's study (Suzuki et
al., 2017), food value was computed based on the weighted sum of four different nutrients:
carbohydrate, fat, protein and vitamin. Though the basic idea that nutrient value could be
transformed into the overall value of the food in medial OFC was supported, only carbohydrate
information was successfully decoded in lateral OFC in my MVPA study. I will address the
potential reasons underlying this discrepancy in detail later.
Food devaluation and devaluation of carbohydrate
With the basic value construction process confirmed, I then tested whether food
devaluation associated with glucose and sucralose consumption was related to the devaluation of
carbohydrates. I observed evidence supporting this hypothesis. Specifically, the strength of
carbohydrate tracking signal within the lateral OFC carbohydrate-tracking cluster was marginally
diminished after glucose and sucralose (relative to water). I did not observe any drink effect on
the functional connectivity between this region and the value-tracking region of the medial OFC.
Utilizing traditional univariate ROI analysis and MVPA, mean brain signals and brain
patterns were separately examined, and then the drink effect was evaluated separately with two
approaches. MVPA focuses on the distributed patterns of brain activity and has higher sensitivity
in detecting value-related signals that vary across participants and voxels compared with
traditional univariate analysis (DiFeliceantonio et al., 2018). The nutrient tracking region was not
detected when the univariate approach was implemented in my dataset, and it was possibly due

60
to the heterogeneous nature of the brain representation of elemental nutrients. On the other hand,
voxel-wise analyses are more sensitive to subject-level variability in mean activation across an
ROI, and MVPA lacks this kind of sensitivity (Davis et al., 2014). Taking advantage of the
strengths of both methods, MVPA and univariate analysis were combined to elucidate how
elemental nutrients signaling in the brain could be altered by ingestion of glucose and sucralose.
Based on brain signals in the lateral OFC nutrient tracking region, I observed evidence
consistent with the hypothesis that food devaluation after sucralose was related to the specific
devaluation of carbohydrates. In particular, I observed attenuated mean brain signals in
carbohydrate tracking region in lateral OFC after sucralose relative to water (t(117) = -2.59, P =
0.03). The signal change was also directionally lower in the glucose condition (vs. water), but the
comparison was not statistically significant (t(122) = 1.54, P = 0.25).
When MVPA was used to compare carbohydrate decoding accuracy in lateral OFC, the
decoding accuracy was marginally attenuated in both glucose (t(53.8) = -2.43, P = 0.06) and
sucralose (t(54)= -2.02, P = 0.09) conditions relative to water. No difference was found between
glucose and sucralose (t(39.1) = 1.00, P = 0.32). In general, NNS users’ decoding accuracy of
carbohydrate was higher than that of NNS non-users (t(17.5) = 2.22, P = 0.04).
The results utilizing two different approaches both supported the hypothesis that
sucralose ingestion could lead to acute food devaluation by decreasing carbohydrate sensitivity
to the food items depicted, at least shortly after consumption. Mean brain signal comparison
suggested similar effects in the glucose condition, though these did not cross the threshold for
statistical significance.

61
Sensory-specific satiety (SSS) and Sweet satiation
The data suggested that food devaluation after glucose and sucralose ingestion was partly
due to the devaluation of carbohydrate, specifically the attenuated carbohydrate sensitivity in
lateral OFC carbohydrate tracking region. The devaluation seems to be nutrient-specific based on
the MVPA. While the only nutrient significantly decoded in lateral OFC was carbohydrate, it is
possible that value attenuation related to drink consumption was related to changes in signaling
associated with other macronutrients. It should be noted that sucralose doesn't contain any
carbohydrates. However, since sweet taste is normally a signal of carbohydrate content, it is
possible that sucralose could temporarily induce devaluation of carbohydrates.
With that said, the mechanisms behind this “nutrient-specific” devaluation could be the
effect of sensory-specific satiety (SSS), which reflects the loss of appetite, or the declined liking
or wanting, for previously consumed food (Rolls et al., 1981; Weijzen et al., 2009). SSS has been
shown to occur similarly in obese and normal-weight people (Snoek et al., 2004). SSS for food
occurs within 2 min after consumption and could last as long as 60 min after consumption
(Hetherington et al., 1989). The immediacy of the satiating effect indicates that the sensory
properties like the pleasantness of appearance, smell, texture and taste of the food, rather than of
internal post-ingestive consequences, produces the effect. I did find an immediate decrease in
appetite for sweet food 5 min after the consumption, but no additional appetite score was
recorded until 90 min after the consumption. In addition, though appetite for sweet food (5 min
post consumption) was directionally lower after sweet drink preloads compared to water, the
difference is not statistically significant. At 90 minutes post-drink, appetite for sweet food did
not differ across conditions.

62
In this study, a similar effect of SSS (devaluation of carbohydrate) was found both after
glucose and sucralose, suggesting that sweet taste might play an essential role in SSS. One study
showed that sweet carbohydrates in a meal could suppress the appetite for sweet food and non-
sweet carbohydrates could suppress the appetite for something savory 2 hours after the meal (de
Graaf et al., 1993), suggesting that sweet taste, not the specific nutrient carbohydrate content
contributes to the SSS. In fact, most well-designed studies illustrated that the degree of SSS of
sweet stimuli does not seem to depend on energy content(Miller et al., 2000; Raben et al., 2002;
Rolls et al., 1988, 1989), though evidence supporting the opposite also exists (Weijzen et al.,
2009).
Several studies investigated how NNS consumption could influence SSS. One study
showed that jello sweetened with aspartame could produce SSS (reduced appetite for jello
relative to uneaten food), the same as jello sweetened with sucrose (Rolls et al., 1988).
Furthermore, the awareness of the caloric content of the foods did not influence SSS (Rolls et al.,
1989). A recent study (Rogers et al., 2020) revisited SSS by testing how real-life LCS drinks
(blackcurrant squash, orange squash and diet cola) could influence sweet food consumption.
Findings from this study provided evidence that LCS drink consumption could lead to acute
appetite suppression for sweet food (but not for non-sweet food). These results indicate that SSS
can be generalized from food-specific to taste-specific, from reducing appetite for eaten food to a
more general appetite suppression towards food with similar taste (sweet). While food-specific
SSS persisted for at least two hours, the generalized taste-specific satiety was more transient.
Also, the author reported that diet cola could reduce food simultaneously consumed compared
with regular cola with only a 20% energy compensation (Rogers et al., 2020).

63
This sensory-specific perspective is in line with the rationale of how food value could be
constructed. If nutrient information was encoded primarily based on the sensory input arriving at
the lateral portion of the OFC, signaling in this region could also be influenced by the sensory
properties of the drink preloads. The brain mechanisms of SSS have not been investigated
thoroughly, but one study (O’Doherty et al., 2000) showed that human olfactory sensory-specific
satiety is related to activation of a region in the orbitofrontal cortex. It should be noted, however,
that only five participants were included in that investigation of food-specific SSS.
If it is the sweet taste, in sucralose at least, that suppresses the appetite for sweet food,
one might argue that it would be more informative to look at how drink preloads influence sugar
tracking rather than carbohydrate tracking in the brain. However, using the same MVPA
procedure utilized elsewhere in the analysis, no sugar-tracking regions were identified in either
the medial nor lateral OFC (based on subjective ratings of sugar content). It is possible that the
collinearity between sugar and other nutrients in the selected food samples diminished the
capacity to decode sugar content based on brain activity patterns. Alternatively, it may be that
carbohydrate content is generally more effectively decoded using MVPA than is the sugar (a
sub-category of carbohydrate). Follow-up studies should address this issue by orthogonalizing
objective sugar content and other nutrients to investigate whether sugar tracking is present in the
OFC, and whether that could be impacted by NNS and NS.
Together, this study suggested that glucose and sucralose decreased response to the
carbohydrate content of visually depicted foods. As mentioned before, no other nutrient
information could be detected in lateral OFC, so it is not conclusive whether glucose and
sucralose influence food valuation through the devaluation of other nutrients, or a generally
decreased appetite towards all food. When comparing the scores of hunger and appetite for sweet

64
food after the drink preloads in this study, the general hunger and appetite for sweet food were
suppressed immediately after both sweet drinks. While the two appetite scores were significantly
different between glucose and sucralose 90 min after consumption, suggesting time-course
differences both in the suppression of overall hunger as well as in the suppression of appetite for
sweet food. This is not surprising since this delay allowed post-ingestive signaling sufficient time
to influence appetite.
The role of past NNS usage
The perception of the sensory characteristics of the study drinks and carbohydrate
decoding accuracy was different between NNS users and non-users. Specifically, NNS users
rated water as less sweet (t(44.9) =- 2.18, P = 0.03) and had a higher carbohydrate decoding
accuracy (t(17.5) = 2.22, P = 0.04) compared with NNS non-users.
Both NNS users and NNS non-users displayed attenuated carbohydrate sensitivity after
consuming sweet drinks relative to water, suggesting that the SSS was possibly caused by the
sweet taste. Interestingly, NNS users demonstrated a generalized heightened carbohydrate
sensitivity across all drink conditions compared with NNS non-users. NNS users also considered
the water preload as sweeter than NNS non-users. In this study, carbohydrate content detected
through multivariate brain patterns contributes to the valuation process. Hence the heightened
carbohydrate sensitivity could potentially lead to increased value signaling, and possibly more
food intake. A previous rodent study has reported a similar effect of habitual NNS exposure
leading to more food intake, increased weight gain and adiposity compared with a control group
fed with food sweetened with glucose(Swithers et al., 2009; Swithers & Davidson, 2008). One
possible mechanism is the decoupling of sweet taste from the subsequent metabolic
consequences, which could lead to energy regulation deficits and guide food intake

65
inappropriately. The potential detrimental effects of habitual NNS consumption seem to
contradict the finding that sucralose can induce SSS. However, I consider the beneficial satiating
effect of NNS to be acute and short-term. It is still inconclusive whether habitual consumption of
NNS leads to increased appetite, over-compensation of food and weight gain (Appleton &
Blundell, 2007; Foltin et al., 1988; Mattes, 1990; Naismith & Rhodes, 1995; Van Wymelbeke et
al., 2004). If there are adverse effects of NNS on appetite regulation, it will be important to
identify the amount and duration of NNS consumption that leads to dysregulation.
Other mechanisms of drink effect on nutrient valuation and related food valuation process
Besides the SSS explanation, possible mechanisms influencing food-cue activity
mentioned previously could also impact food valuation and nutrient valuation through the
suppression of general appetite. The health consequences of NNS consumption are especially
controversial. Based on the result of this study, the sensory property of sweet taste itself seems to
have an impact on brain activity during food decision-making. Sweetness can also activate sweet
taste receptors in the arcuate nucleus in hypothalamus and can lead to a dose-dependent decrease
in food intake when sucralose is intracerebroventricularly injected in mice (1–3 hours after the
injection) (Kohno, 2017; Kohno et al., 2016).
Possible explanations of why only carbohydrate could be decoded
In the original food valuation construction paper (Suzuki et al., 2017), food value was
computed based on the weighted sum of four different nutrients: carbohydrate, fat, protein and
vitamin. Though the basic idea that nutrient value could be transformed into the overall value of
the food in medial OFC was supported, only one nutrient, carbohydrate, was successfully
decoded in lateral OFC. Here I consider several possible explanations behind this discrepancy,
and also provide a simplified general summary.

66
First, the orthogonalized carbohydrate and fat content in the food samples might
influence which combination of elemental nutrients would best predict the food value. In
Suzuki’s study, the nutrient content of depicted food was not orthogonalized, and in some
cases, was highly correlated (Suzuki et al., 2017). This study’s food selections avoided the issue
of collinearity which allowed us to separate brain correlates of carbohydrate and fat with a higher
power. This is critical in my study since the final goal is to evaluate how different "sugar"
preloads could influence carbohydrate tracking in the brain.
Second, the "sugar" preloads might bias participants’ valuation through either the altered
attentional focus towards carbohydrates, or the priming effect of the preload itself, by ingesting
"sugary" drinks that potentially contains carbohydrates. Priming in different modalities such as
visual, olfactory, or cognitive primes (Gaillet et al., 2013; Manippa et al., 2019; Mas et al., 2019;
Papies, 2016; Wilson et al., 2016) could influence people’s food choices and food-related
behaviors. In addition, imaging studies have shown evidence that attentional focus can modify
brain food-cue responses (Franssen et al., 2020). The authors reported that, in overweight
females, attentional focus on the hedonic components of the food (vs. neutral) was related to
stronger neural response to high-calorie food in regions associated with food-cue processing
(e.g., medial OFC). Another FMRI study (Hare et al., 2011) revealed that focusing on the health
aspect of foods could lead to healthier food choices and increased brain activity associated with
self-control in non-dieting individuals. Moreover, the study design could induce a priming effect
of carbohydrate. Though the participants were not guided to direct their attention or focus to any
specific nutrient factors, it is plausible that their attention was shifted to the carbohydrate
component during food valuation by the sweet drink ingestion. To parse out the priming effect of

67
sweet drinks on food valuation, I redid the prediction model using the willingness to pay data
from the water study session alone. Still, carbohydrate remained the best predictor. However,
there is no way the awareness of sweet drinks could be completely removed since on 2/3 of
water sessions, participants had consumed sweet drinks in previous sessions. These findings have
led to the question of whether direct attention towards other constituent nutrients (fat, protein,
vitamin, etc.) through verbal guidance, cue priming, or direct consumption could alter people's
food decisions. In our current society where the ketogenic diet and the protein-based diet are
becoming popular, it would be interesting to evaluate whether consuming food that contains a
primary nutrient source (fat or protein) would influence food valuation and food choice later.
Third, participants with different metabolic states and weight status might differ in their
brain food valuation process. In this study, all participants came in after at least a 12-hour fast,
but subjects in Suzuki's study only fasted for 3 hours. Different drinks ingested before the food
valuation task could also lead to variation in metabolic states. A difference in metabolic state
might change participants' perception of the food choice. Instead of considering the overall
nutritional balance when evaluating food items, people in energy deficit might put greater weight
on the essential nutrients that can help them to survive. It was proposed that throughout human
evolution, survival depended heavily on plant foods with high starch quantities (Hardy et al.,
2015). About 40–75% of modern dietary energy intake is from carbohydrates (“Carbohydrates in
Human Nutrition. Report of a Joint FAO/WHO Expert Consultation,” 1998). The basic
functioning of the human body, including the brain (Fonseca-Azevedo & Herculano-Houzel,
2012), kidney, red blood cells (Mulquiney & Kuchel, 1999) and reproductive organs, requires a
steady supply of glycemic carbohydrate (Mulquiney & Kuchel, 1999). The brain itself accounts

68
for 20–25% of basal metabolic expenditure in adults (Fonseca-Azevedo & Herculano-Houzel,
2012). It is possible that in metabolic deficit, the value of carbohydrates (relative to other
nutrients) increases. Finally, it is important to note that (unlike Suzuki et al 2017) obese people
were included in the study sample. It has been reported that obese individuals exhibit different
food-related brain signaling compared with healthy people (Castellanos et al., 2009; Mehl et al.,
2019; Scharmüller et al., 2012). Hence including obese people in this study could potentially
impact the results obtained.
John P. O’Doherty, one of the authors in the food value construction paper on which my
study is based, commented later that “The active construction of value from a weighted
combination of underlying features naturally endows the decision-making agent with the
capability to…. flexibly change the weights assigned to attribute features based on changes in
internal motivation/homeostasis and/or external context (J. P. O’Doherty et al., 2021).”
Therefore, a simplified mechanism, underlying why only carbohydrate information was
significantly decoded in my study, could be that the specific context of the study led to more
weight assigned to carbohydrates relative to other nutrients.
Limitations
Several limitations of this study are worth noting. First and perhaps most significantly,
the sample size is relatively small. Small sample studies can lead to inconsistency, and serious
problems of low replicability of neuroimaging results (Button et al., 2013). I have made my data
accessible using the standardized “Brain Imaging Data Structure” (BIDS) format (Gorgolewski
et al., 2016) to facilitate future meta-analyses that may more definitively address this issue.
Second, the sample was quite heterogeneous in BMI (Green & Murphy, 2012). While this
heterogeneity might be a strength in a large study where associations with these variables could

69
be explored, the modest sample size here does not provide sufficient power to do this. The
heterogeneity may have resulted in loss of power to detect effects that would be evident in a
more homogenous sample. A third limitation of my experimental design is that we did not record
subjective hunger (appetite scale) for the 75-min window in which participants were completing
the neuroimaging portions of the protocol. Given the dynamic nature of post-ingestive signaling
relevant to hunger, regular assessment of appetite during this period would be informative.
Future directions
Several questions remained to be addressed in the future. First, the current study focused
on the acute effect of sucralose but not the chronic effect. In drug research, drug acute and
chronic effects are often quite different, and in some cases, directionally opposite. For example,
the psychoactive drug MDMA (known as "Molly" or "ecstasy") acutely causes euphoria, but
chronic use can cause the opposite -sustained depression, possibly due to neuroadaptation to the
drug (McCardle et al., 2004). Similarly, it has been found that sucralose treatment by exposing
mouse hypothalamic cells for 45 mins reduces the expression level of the sweet taste receptor in
the hypothalamus (Ren et al., 2009). So, it would be important to examine the long-term effect of
sucralose. Second, participants’ weight status seemed to interact with drink preloads and
influence their willingness to pay. The exploratory analysis in this study revealed that obese
people's lower bids for food after glucose and sucralose were the main driver for the drink
difference. Future studies should investigate the influence of obesity on glucose and sucralose
effect on food decisions with a larger sample. Third, individuals with different sweetness
sensitivities have different SSS for sweetness (Han et al., 2017) and different brain responses to
sweet food odors (Han et al., 2020). Future studies could examine how sweetness sensitivity
could change the relation between sweet drink preloads and food decisions. Finally, metabolic

70
and physiological data could be collected to help us understand the underlying mechanisms of
the current findings. Specifically, to my knowledge, no imaging studies in humans have been
done to directly examine the brain response to gut stimulation by NNS and NS.

71
References

Andersen, R. A., & Cui, H. (2009). Intention, action planning, and decision making in parietal-
frontal circuits. Neuron, 63(5), 568–583.
Appleton, K. M., & Blundell, J. E. (2007). Habitual high and low consumers of artificially-
sweetened beverages: Effects of sweet taste and energy on short-term appetite.
Physiology & Behavior, 92(3), 479–486.
Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed
random effects for subjects and items. Journal of Memory and Language, 59(4), 390–
412.
Barron, H. C., Dolan, R. J., & Behrens, T. E. (2013). Online evaluation of novel choices by
simultaneous representation of multiple memories. Nature Neuroscience, 16(10), 1492.
Bartra, O., McGuire, J. T., & Kable, J. W. (2013). The valuation system: A coordinate-based
meta-analysis of BOLD fMRI experiments examining neural correlates of subjective
value. Neuroimage, 76, 412–427.
Becker, G. M., DeGroot, M. H., & Marschak, J. (1964). Measuring utility by a single-response
sequential method. Behavioral Science, 9(3), 226–232.
Blechert, J., Meule, A., Busch, N. A., & Ohla, K. (2014). Food-pics: An image database for
experimental research on eating and appetite. Frontiers in Psychology, 5, 617.
Button, K. S., Ioannidis, J. P. A., Mokrysz, C., Nosek, B. A., Flint, J., Robinson, E. S. J., &
Munafò, M. R. (2013). Power failure: Why small sample size undermines the reliability
of neuroscience. Nature Reviews Neuroscience, 14(5), 365–376.
https://doi.org/10.1038/nrn3475

72
Buuren, S. van, & Groothuis-Oudshoorn, K. (2010). mice: Multivariate imputation by chained
equations in R. Journal of Statistical Software, 1–68.
Carbohydrates in human nutrition. Report of a Joint FAO/WHO Expert Consultation. (1998).
FAO Food and Nutrition Paper, 66, 1–140.
Carnell, S., Gibson, C., Benson, L., Ochner, C. N., & Geliebter, A. (2012). Neuroimaging and
obesity: Current knowledge and future directions. Obesity Reviews, 13(1), 43–56.
Castellanos, E. H., Charboneau, E., Dietrich, M. S., Park, S., Bradley, B. P., Mogg, K., &
Cowan, R. L. (2009). Obese adults have visual attention bias for food cue images:
Evidence for altered reward system function. International Journal of Obesity, 33(9),
1063–1073. https://doi.org/10.1038/ijo.2009.138
Chaisangmongkon, W., Swaminathan, S. K., Freedman, D. J., & Wang, X.-J. (2017). Computing
by robust transience: How the fronto-parietal network performs sequential, category-
based decisions. Neuron, 93(6), 1504–1517.
Chand, G. B., & Dhamala, M. (2017). Interactions between the anterior cingulate-insula network
and the fronto-parietal network during perceptual decision-making. Neuroimage, 152,
381–389.
Chang, C.-C., & Lin, C.-J. (2011). LIBSVM: A library for support vector machines. ACM
Transactions on Intelligent Systems and Technology (TIST), 2(3), 1–27.
Chib, V. S., Rangel, A., Shimojo, S., & O’Doherty, J. P. (2009). Evidence for a common
representation of decision values for dissimilar goods in human ventromedial prefrontal
cortex. Journal of Neuroscience, 29(39), 12315–12320.

73
Childress, A. R., Ehrman, R. N., Wang, Z., Li, Y., Sciortino, N., Hakun, J., Jens, W., Suh, J.,
Listerud, J., & Marquez, K. (2008). Prelude to passion: Limbic activation by “unseen”
drug and sexual cues. PLoS One, 3(1), e1506.
Clark, K. A., Alves, J. M., Jones, S., Yunker, A. G., Luo, S., Cabeen, R. P., Angelo, B., Xiang,
A. H., & Page, K. A. (2020). Dietary Fructose Intake and Hippocampal Structure and
Connectivity during Childhood. Nutrients, 12(4), 909.
https://doi.org/10.3390/nu12040909
Clithero, J. A., & Rangel, A. (2013). Informatic parcellation of the network involved in the
computation of subjective value. Social Cognitive and Affective Neuroscience, 9(9),
1289–1302.
Coons, E. E., & Cruce, J. A. F. (1968). Lateral hypothalamus: Food current intensity in
maintaining self-stimulation of hunger. Science, 159(3819), 1117–1119.
Coons, E. E., Levak, M., & Miller, N. E. (1965). Lateral hypothalamus: Learning of food-
seeking response motivated by electrical stimulation. Science, 150(3701), 1320–1321.
Cornier, M.-A., Von Kaenel, S. S., Bessesen, D. H., & Tregellas, J. R. (2007). Effects of
overfeeding on the neuronal response to visual food cues–. The American Journal of
Clinical Nutrition, 86(4), 965–971.
Dalenberg, J. R., Patel, B. P., Denis, R., Veldhuizen, M. G., Nakamura, Y., Vinke, P. C., Luquet,
S., & Small, D. M. (2020). Short-Term Consumption of Sucralose with, but Not without,
Carbohydrate Impairs Neural and Metabolic Sensitivity to Sugar in Humans. Cell
Metabolism, 31(3), 493-502.e7. https://doi.org/10.1016/j.cmet.2020.01.014

74
Damak, S., Rong, M., Yasumatsu, K., Kokrashvili, Z., Varadarajan, V., Zou, S., Jiang, P.,
Ninomiya, Y., & Margolskee, R. F. (2003). Detection of sweet and umami taste in the
absence of taste receptor T1r3. Science, 301(5634), 850–853.
Davidson, T. L., Martin, A. A., Clark, K., & Swithers, S. E. (2011). Intake of high-intensity
sweeteners alters the ability of sweet taste to signal caloric consequences: Implications
for the learned control of energy and body weight regulation. Quarterly Journal of
Experimental Psychology (2006), 64(7), 1430–1441.
https://doi.org/10.1080/17470218.2011.552729
Davis, T., LaRocque, K. F., Mumford, J. A., Norman, K. A., Wagner, A. D., & Poldrack, R. A.
(2014). What do differences between multi-voxel and univariate analysis mean? How
subject-, voxel-, and trial-level variance impact fMRI analysis. NeuroImage, 97, 271–
283. https://doi.org/10.1016/j.neuroimage.2014.04.037
de Graaf, C., Schreurs, A., & Blauw, Y. H. (1993). Short-term effects of different amounts of
sweet and nonsweet carbohydrates on satiety and energy intake. Physiology & Behavior,
54(5), 833–843. https://doi.org/10.1016/0031-9384(93)90290-V
de Ruyter, J. C., Olthof, M. R., Kuijper, L. D. J., & Katan, M. B. (2012). Effect of sugar-
sweetened beverages on body weight in children: Design and baseline characteristics of
the Double-blind, Randomized INtervention study in Kids. Contemporary Clinical Trials,
33(1), 247–257. https://doi.org/10.1016/j.cct.2011.10.007
Deichmann, R., Gottfried, J. A., Hutton, C., & Turner, R. (2003). Optimized EPI for fMRI
studies of the orbitofrontal cortex. Neuroimage, 19(2), 430–441.

75
Demos, K. E., Heatherton, T. F., & Kelley, W. M. (2012). Individual differences in nucleus
accumbens activity to food and sexual images predict weight gain and sexual behavior.
Journal of Neuroscience, 32(16), 5549–5552.
Dunford, E. K., Miles, D. R., Ng, S. W., & Popkin, B. (2020). Types and Amounts of
Nonnutritive Sweeteners Purchased by US Households: A Comparison of 2002 and 2018
Nielsen Homescan Purchases. Journal of the Academy of Nutrition and Dietetics,
120(10), 1662-1671.e10. https://doi.org/10.1016/j.jand.2020.04.022
Dye, L., & Blundell, J. E. (1997). Menstrual cycle and appetite control: Implications for weight
regulation. Human Reproduction (Oxford, England), 12(6), 1142–1151.
Fan, J., McCandliss, B. D., Sommer, T., Raz, A., & Posner, M. I. (2002). Testing the efficiency
and independence of attentional networks. Journal of Cognitive Neuroscience, 14(3),
340–347. https://doi.org/10.1162/089892902317361886
Fernstrom, J. D. (2015). Non-nutritive sweeteners and obesity. Annual Review of Food Science
and Technology, 6, 119–136. https://doi.org/10.1146/annurev-food-022814-015635
Feskanich, D., Sielaff, B. H., Chong, K., & Buzzard, I. M. (1989). Computerized collection and
analysis of dietary intake information. Computer Methods and Programs in Biomedicine,
30(1), 47–57.
Field, M., & Duka, T. (2004). Cue reactivity in smokers: The effects of perceived cigarette
availability and gender. Pharmacology, Biochemistry, and Behavior, 78(3), 647–652.
https://doi.org/10.1016/j.pbb.2004.03.026
Foerde, K., Steinglass, J. E., Shohamy, D., & Walsh, B. T. (2015). Neural mechanisms
supporting maladaptive food choices in anorexia nervosa. Nature Neuroscience, 18(11),
1571.

76
Foletto, K. C., Melo Batista, B. A., Neves, A. M., de Matos Feijó, F., Ballard, C. R., Marques
Ribeiro, M. F., & Bertoluci, M. C. (2016). Sweet taste of saccharin induces weight gain
without increasing caloric intake, not related to insulin-resistance in Wistar rats. Appetite,
96, 604–610. https://doi.org/10.1016/j.appet.2015.11.003
Foltin, R. W., Fischman, M. W., Emurian, C. S., & Rachlinski, J. J. (1988). Compensation for
caloric dilution in humans given unrestricted access to food in a residential laboratory.
Appetite, 10(1), 13–24.
Fonseca-Azevedo, K., & Herculano-Houzel, S. (2012). Metabolic constraint imposes tradeoff
between body size and number of brain neurons in human evolution. Proceedings of the
National Academy of Sciences, 109(45), 18571–18576.
https://doi.org/10.1073/pnas.1206390109
Ford, H. E., Peters, V., Martin, N. M., Sleeth, M. L., Ghatei, M. A., Frost, G. S., & Bloom, S. R.
(2011). Effects of oral ingestion of sucralose on gut hormone response and appetite in
healthy normal-weight subjects. European Journal of Clinical Nutrition, 65(4), 508.
Franssen, S., Jansen, A., van den Hurk, J., Roebroeck, A., & Roefs, A. (2020). Power of mind:
Attentional focus rather than palatability dominates neural responding to visual food
stimuli in females with overweight. Appetite, 148, 104609.
Gaillet, M., Sulmont-Rossé, C., Issanchou, S., Chabanet, C., & Chambaron, S. (2013). Priming
effects of an olfactory food cue on subsequent food-related behaviour. Food Quality and
Preference, 30(2), 274–281. https://doi.org/10.1016/j.foodqual.2013.06.008
Ganley, R. M. (1989). Emotion and eating in obesity: A review of the literature. International
Journal of Eating Disorders, 8(3), 343–361.

77
Gläscher, J., Adolphs, R., Damasio, H., Bechara, A., Rudrauf, D., Calamia, M., Paul, L. K., &
Tranel, D. (2012). Lesion mapping of cognitive control and value-based decision making
in the prefrontal cortex. Proceedings of the National Academy of Sciences, 109(36),
14681–14686.
Global Zero-Calorie Sweetener Market Projected to Be Worth USD 2.84 Billion by 2021:
Technavio. (2017, April 25). Yahoo Finance.
Gorgolewski, K. J., Auer, T., Calhoun, V. D., Craddock, R. C., Das, S., Duff, E. P., Flandin, G.,
Ghosh, S. S., Glatard, T., & Halchenko, Y. O. (2016). The brain imaging data structure, a
format for organizing and describing outputs of neuroimaging experiments. Scientific
Data, 3, 160044.
Gottfried, J. A., O’doherty, J., & Dolan, R. J. (2003). Encoding predictive reward value in human
amygdala and orbitofrontal cortex. Science, 301(5636), 1104–1107.
Grabenhorst, F., & Rolls, E. T. (2011). Value, pleasure and choice in the ventral prefrontal
cortex. Trends in Cognitive Sciences, 15(2), 56–67.
Green, E., & Murphy, C. (2012). Altered processing of sweet taste in the brain of diet soda
drinkers. Physiology & Behavior, 107(4), 560–567.
Gross, J., Woelbert, E., Zimmermann, J., Okamoto-Barth, S., Riedl, A., & Goebel, R. (2014).
Value signals in the prefrontal cortex predict individual preferences across reward
categories. Journal of Neuroscience, 34(22), 7580–7586.
Grotz, V. L., & Munro, I. C. (2009). An overview of the safety of sucralose. Regulatory
Toxicology and Pharmacology, 55(1), 1–5.

78
Han, P., Keast, R. S. J., & Roura, E. (2017). Salivary leptin and TAS1R2/TAS1R3
polymorphisms are related to sweet taste sensitivity and carbohydrate intake from a
buffet meal in healthy young adults. British Journal of Nutrition, 118(10), 763–770.
https://doi.org/10.1017/S0007114517002872
Han, P., Mohebbi, M., Seo, H.-S., & Hummel, T. (2020). Sensitivity to sweetness correlates to
elevated reward brain responses to sweet and high-fat food odors in young healthy
volunteers. NeuroImage, 208, 116413. https://doi.org/10.1016/j.neuroimage.2019.116413
Hanke, M., Halchenko, Y. O., Sederberg, P. B., Hanson, S. J., Haxby, J. V., & Pollmann, S.
(2009). PyMVPA: A python toolbox for multivariate pattern analysis of fMRI data.
Neuroinformatics, 7(1), 37–53.
Hardy, K., Brand-Miller, J., Brown, K. D., Thomas, M. G., & Copeland, L. (2015). The
Importance of Dietary Carbohydrate in Human Evolution. The Quarterly Review of
Biology, 90(3), 251–268. https://doi.org/10.1086/682587
Hare, T. A., Malmaud, J., & Rangel, A. (2011). Focusing Attention on the Health Aspects of
Foods Changes Value Signals in vmPFC and Improves Dietary Choice. Journal of
Neuroscience, 31(30), 11077–11087. https://doi.org/10.1523/JNEUROSCI.6383-10.2011
Heart, N., Lung, Institute, B., Diabetes, N. I. of, Digestive, & Diseases (US), K. (1998). Clinical
guidelines on the identification, evaluation, and treatment of overweight and obesity in
adults: The evidence report. National Heart, Lung, and Blood Institute.
Heni, M., Kullmann, S., Ketterer, C., Guthoff, M., Bayer, M., Staiger, H., Machicao, F., Häring,
H.-U., Preissl, H., & Veit, R. (2014). Differential effect of glucose ingestion on the neural
processing of food stimuli in lean and overweight adults. Human Brain Mapping, 35(3),
918–928.

79
Hetherington, M., Rolls, B. J., & Burley, V. J. (1989). The time course of sensory-specific
satiety. Appetite, 12(1), 57–68. https://doi.org/10.1016/0195-6663(89)90068-8
Higgins, K. A., & Mattes, R. D. (2019). A randomized controlled trial contrasting the effects of 4
low-calorie sweeteners and sucrose on body weight in adults with overweight or obesity.
The American Journal of Clinical Nutrition, 109(5), 1288–1301.
https://doi.org/10.1093/ajcn/nqy381
Howard, J. D., Gottfried, J. A., Tobler, P. N., & Kahnt, T. (2015). Identity-specific coding of
future rewards in the human orbitofrontal cortex. Proceedings of the National Academy
of Sciences, 201503550.
Jang, H.-J., Kokrashvili, Z., Theodorakis, M. J., Carlson, O. D., Kim, B.-J., Zhou, J., Kim, H. H.,
Xu, X., Chan, S. L., Juhaszova, M., Bernier, M., Mosinger, B., Margolskee, R. F., &
Egan, J. M. (2007). Gut-expressed gustducin and taste receptors regulate secretion of
glucagon-like peptide-1. Proceedings of the National Academy of Sciences of the United
States of America, 104(38), 15069–15074. https://doi.org/10.1073/pnas.0706890104
Jansen, A. (1998). A learning model of binge eating: Cue reactivity and cue exposure. Behaviour
Research and Therapy, 36(3), 257–272.
Jenkinson, M., Bannister, P., Brady, M., & Smith, S. (2002). Improved optimization for the
robust and accurate linear registration and motion correction of brain images.
Neuroimage, 17(2), 825–841.
Jenkinson, M., Beckmann, C. F., Behrens, T. E., Woolrich, M. W., & Smith, S. M. (2012). Fsl.
Neuroimage, 62(2), 782–790.

80
Killgore, W. D., Young, A. D., Femia, L. A., Bogorodzki, P., Rogowska, J., & Yurgelun-Todd,
D. A. (2003). Cortical and limbic activation during viewing of high-versus low-calorie
foods. Neuroimage, 19(4), 1381–1394.
Klein-Flügge, M. C., Barron, H. C., Brodersen, K. H., Dolan, R. J., & Behrens, T. E. J. (2013).
Segregated encoding of reward–identity and stimulus–reward associations in human
orbitofrontal cortex. Journal of Neuroscience, 33(7), 3202–3211.
Kohno, D. (2017). Sweet taste receptor in the hypothalamus: A potential new player in glucose
sensing in the hypothalamus. The Journal of Physiological Sciences: JPS, 67(4), 459–
465. https://doi.org/10.1007/s12576-017-0535-y
Kohno, D., Koike, M., Ninomiya, Y., Kojima, I., Kitamura, T., & Yada, T. (2016). Sweet Taste
Receptor Serves to Activate Glucose- and Leptin-Responsive Neurons in the
Hypothalamic Arcuate Nucleus and Participates in Glucose Responsiveness. Frontiers in
Neuroscience, 10. https://doi.org/10.3389/fnins.2016.00502
Kriegeskorte, N., Simmons, W. K., Bellgowan, P. S., & Baker, C. I. (2009). Circular analysis in
systems neuroscience – the dangers of double dipping. Nature Neuroscience, 12(5), 535–
540. https://doi.org/10.1038/nn.2303
Krishnan, S., Tryon, R., Welch, L. C., Horn, W. F., & Keim, N. L. (2016). Menstrual cycle
hormones, food intake, and cravings. The FASEB Journal, 30(S1), 418.6-418.6.
https://doi.org/10.1096/fasebj.30.1_supplement.418.6
Kroemer, N. B., Krebs, L., Kobiella, A., Grimm, O., Vollstädt-Klein, S., Wolfensteller, U.,
Kling, R., Bidlingmaier, M., Zimmermann, U. S., & Smolka, M. N. (2013). (Still)
longing for food: Insulin reactivity modulates response to food pictures. Human Brain
Mapping, 34(10), 2367–2380.

81
Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. (2017). lmerTest package: Tests in linear
mixed effects models. Journal of Statistical Software, 82(13), 1–26.
Lawrence, N. S., Hinton, E. C., Parkinson, J. A., & Lawrence, A. D. (2012). Nucleus accumbens
response to food cues predicts subsequent snack consumption in women and increased
body mass index in those with reduced self-control. NeuroImage, 63(1), 415–422.
https://doi.org/10.1016/j.neuroimage.2012.06.070
Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. (2018). Emmeans: Estimated
marginal means, aka least-squares means. R Package Version, 1(1), 3.
Levy, D. J., & Glimcher, P. W. (2011). Comparing apples and oranges: Using reward-specific
and reward-general subjective value representation in the brain. Journal of Neuroscience,
31(41), 14693–14707.
Levy, D. J., & Glimcher, P. W. (2012). The root of all value: A neural common currency for
choice. Current Opinion in Neurobiology, 22(6), 1027–1038.
Levy, I., Lazzaro, S. C., Rutledge, R. B., & Glimcher, P. W. (2011). Choice from non-choice:
Predicting consumer preferences from blood oxygenation level-dependent signals
obtained during passive viewing. Journal of Neuroscience, 31(1), 118–125.
Lin, A., Adolphs, R., & Rangel, A. (2011). Social and monetary reward learning engage
overlapping neural substrates. Social Cognitive and Affective Neuroscience, 7(3), 274–
281.
Lopez, R. B., Hofmann, W., Wagner, D. D., Kelley, W. M., & Heatherton, T. F. (2014). Neural
predictors of giving in to temptation in daily life. Psychological Science, 25(7), 1337–
1344. https://doi.org/10.1177/0956797614531492

82
Luo, S., Monterosso, J. R., Sarpelleh, K., & Page, K. A. (2015). Differential effects of fructose
versus glucose on brain and appetitive responses to food cues and decisions for food
rewards. Proceedings of the National Academy of Sciences, 112(20), 6509–6514.
https://doi.org/10.1073/pnas.1503358112
Malik, S., McGlone, F., Bedrossian, D., & Dagher, A. (2008). Ghrelin modulates brain activity
in areas that control appetitive behavior. Cell Metabolism, 7(5), 400–409.
Malik, V. S., Popkin, B. M., Bray, G. A., Després, J.-P., & Hu, F. B. (2010). Sugar-sweetened
beverages, obesity, type 2 diabetes mellitus, and cardiovascular disease risk. Circulation,
121(11), 1356–1364. https://doi.org/10.1161/CIRCULATIONAHA.109.876185
Manippa, V., van der Laan, L. N., Brancucci, A., & Smeets, P. A. M. (2019). Health body
priming and food choice: An eye tracking study. Food Quality and Preference, 72, 116–
125. https://doi.org/10.1016/j.foodqual.2018.10.006
Mas, M., Brindisi, M.-C., Chabanet, C., Nicklaus, S., & Chambaron, S. (2019). Weight Status
and Attentional Biases Toward Foods: Impact of Implicit Olfactory Priming. Frontiers in
Psychology, 10. https://doi.org/10.3389/fpsyg.2019.01789
Masters, R. K., Reither, E. N., Powers, D. A., Yang, Y. C., Burger, A. E., & Link, B. G. (2013).
The Impact of Obesity on US Mortality Levels: The Importance of Age and Cohort
Factors in Population Estimates. American Journal of Public Health, 103(10), 1895–
1901. https://doi.org/10.2105/AJPH.2013.301379
Mattes, R. (1990). Effects of aspartame and sucrose on hunger and energy intake in humans.
Physiology & Behavior, 47(6), 1037–1044.
McCardle, K., Luebbers, S., Carter, J. D., Croft, R. J., & Stough, C. (2004). Chronic MDMA
(ecstasy) use, cognition and mood. Psychopharmacology, 173(3), 434–439.

83
McNamee, D., Rangel, A., & O’doherty, J. P. (2013). Category-dependent and category-
independent goal-value codes in human ventromedial prefrontal cortex. Nature
Neuroscience, 16(4), 479.
Mehl, N., Morys, F., Villringer, A., & Horstmann, A. (2019). Unhealthy yet Avoidable—How
Cognitive Bias Modification Alters Behavioral and Brain Responses to Food Cues in
Individuals with Obesity. Nutrients, 11(4), 874. https://doi.org/10.3390/nu11040874
Miller, D. L., Bell, E. A., Pelkman, C. L., Peters, J. C., & Rolls, B. J. (2000). Effects of dietary
fat, nutrition labels, and repeated consumption on sensory-specific satiety. Physiology &
Behavior, 71(1–2), 153–158.
Mulquiney, P. J., & Kuchel, P. W. (1999). Model of 2,3-bisphosphoglycerate metabolism in the
human erythrocyte based on detailed enzyme kinetic equations: Equations and parameter
refinement. The Biochemical Journal, 342 Pt 3, 581–596.
Naismith, D. J., & Rhodes, C. (1995). Adjustment in energy intake following the covert removal
of sugar from the diet. Journal of Human Nutrition and Dietetics, 8(3), 167–175.
Nehrling, J. K., Kobe, P., McLane, M. P., Olson, R. E., Kamath, S., & Horwitz, D. L. (1985).
Aspartame use by persons with diabetes. Diabetes Care, 8(5), 415–417.
https://doi.org/10.2337/diacare.8.5.415
Nichol, A. D., Holle, M. J., & An, R. (2018). Glycemic impact of non-nutritive sweeteners: A
systematic review and meta-analysis of randomized controlled trials. European Journal
of Clinical Nutrition, 72(6), 796–804. https://doi.org/10.1038/s41430-018-0170-6
Nie, Y., Vigues, S., Hobbs, J. R., Conn, G. L., & Munger, S. D. (2005). Distinct Contributions of
T1R2 and T1R3 Taste Receptor Subunits to the Detection of Sweet Stimuli. Current
Biology, 15(21), 1948–1952. https://doi.org/10.1016/j.cub.2005.09.037

84
Nitschke, J. B., Dixon, G. E., Sarinopoulos, I., Short, S. J., Cohen, J. D., Smith, E. E., Kosslyn,
S. M., Rose, R. M., & Davidson, R. J. (2006). Altering expectancy dampens neural
response to aversive taste in primary taste cortex. Nature Neuroscience, 9(3), 435–442.
https://doi.org/10.1038/nn1645
Norman, K. A., Polyn, S. M., Detre, G. J., & Haxby, J. V. (2006). Beyond mind-reading: Multi-
voxel pattern analysis of fMRI data. Trends in Cognitive Sciences, 10(9), 424–430.
https://doi.org/10.1016/j.tics.2006.07.005
O’Doherty, J. P., Rutishauser, U., & Iigaya, K. (2021). The hierarchical construction of value.
Current Opinion in Behavioral Sciences, 41, 71–77.
O’Doherty, J., Winston, J., Critchley, H., Perrett, D., Burt, D. M., & Dolan, R. J. (2003). Beauty
in a smile: The role of medial orbitofrontal cortex in facial attractiveness.
Neuropsychologia, 41(2), 147–155.
Öngür, D., & Price, J. L. (2000). The organization of networks within the orbital and medial
prefrontal cortex of rats, monkeys and humans. Cerebral Cortex, 10(3), 206–219.
Orquin, J. L., & Kurzban, R. (2016). A meta-analysis of blood glucose effects on human decision
making. Psychological Bulletin, 142(5), 546–567. https://doi.org/10.1037/bul0000035
Padoa-Schioppa, C., & Assad, J. A. (2006). Neurons in the orbitofrontal cortex encode economic
value. Nature, 441(7090), 223.
Padoa-Schioppa, C., & Cai, X. (2011). Orbitofrontal cortex and the computation of subjective
value: Consolidated concepts and new perspectives. Annals of the New York Academy of
Sciences, 1239, 130.
Page, K. A., Chan, O., Arora, J., Belfort-Deaguiar, R., Dzuira, J., Roehmholdt, B., Cline, G. W.,
Naik, S., Sinha, R., Constable, R. T., & Sherwin, R. S. (2013). Effects of fructose vs

85
glucose on regional cerebral blood flow in brain regions involved with appetite and
reward pathways. JAMA, 309(1), 63–70. https://doi.org/10.1001/jama.2012.116975
Page, K. A., & Melrose, A. J. (2016). Brain, hormone and appetite responses to glucose versus
fructose. Current Opinion in Behavioral Sciences, 9, 111–117.
https://doi.org/10.1016/j.cobeha.2016.03.002
Page, K. A., Seo, D., Belfort-DeAguiar, R., Lacadie, C., Dzuira, J., Naik, S., Amarnath, S.,
Constable, R. T., Sherwin, R. S., & Sinha, R. (2011). Circulating glucose levels modulate
neural control of desire for high-calorie foods in humans. The Journal of Clinical
Investigation, 121(10), 4161–4169.
Paoli, A. (2014). Ketogenic Diet for Obesity: Friend or Foe? International Journal of
Environmental Research and Public Health, 11(2), 2092–2107.
https://doi.org/10.3390/ijerph110202092
Papies, E. K. (2016). Health goal priming as a situated intervention tool: How to benefit from
nonconscious motivational routes to health behaviour. Health Psychology Review, 10(4),
408–424. https://doi.org/10.1080/17437199.2016.1183506
Pepino, M. Y. (2015). Metabolic effects of non-nutritive sweeteners. Physiology & Behavior,
152(Pt B), 450–455. https://doi.org/10.1016/j.physbeh.2015.06.024
Pepino, M. Y., Tiemann, C. D., Patterson, B. W., Wice, B. M., & Klein, S. (2013). Sucralose
affects glycemic and hormonal responses to an oral glucose load. Diabetes Care, 36(9),
2530–2535.
Peters, J., & Büchel, C. (2010). Neural representations of subjective reward value. Behavioural
Brain Research, 213(2), 135–141.

86
Plassmann, H., O’Doherty, J., & Rangel, A. (2007). Orbitofrontal cortex encodes willingness to
pay in everyday economic transactions. Journal of Neuroscience, 27(37), 9984–9988.
Pogoda, L., Holzer, M., Mormann, F., & Weber, B. (2016). Multivariate representation of food
preferences in the human brain. Brain and Cognition, 110, 43–52.
Porikos, K. P., & Pi-Sunyer, F. X. (1984). Regulation of food intake in human obesity: Studies
with caloric dilution and exercise. Clinics in Endocrinology and Metabolism, 13(3), 547–
561. https://doi.org/10.1016/s0300-595x(84)80037-7
Preedy, V. R., Watson, R. R., & Martin, C. R. (2011). Handbook of behavior, food and nutrition.
Springer Science & Business Media.
Pursey, K. M., Stanwell, P., Callister, R. J., Brain, K., Collins, C. E., & Burrows, T. L. (2014).
Neural Responses to Visual Food Cues According to Weight Status: A Systematic
Review of Functional Magnetic Resonance Imaging Studies. Frontiers in Nutrition, 1.
https://doi.org/10.3389/fnut.2014.00007
Raben, A., Vasilaras, T. H., Møller, A. C., & Astrup, A. (2002). Sucrose compared with artificial
sweeteners: Different effects on ad libitum food intake and body weight after 10 wk of
supplementation in overweight subjects. The American Journal of Clinical Nutrition,
76(4), 721–729. https://doi.org/10.1093/ajcn/76.4.721
Ren, X., Zhou, L., Terwilliger, R., Newton, S. S., & de Araujo, I. E. (2009). Sweet taste
signaling functions as a hypothalamic glucose sensor. Frontiers in Integrative
Neuroscience, 3, 12. https://doi.org/10.3389/neuro.07.012.2009
Rich, E. L., & Wallis, J. D. (2016). Decoding subjective decisions from orbitofrontal cortex.
Nature Neuroscience, 19(7), 973.

87
Rogers, P. J., Ferriday, D., Irani, B., Hei Hoi, J. K., England, C. Y., Bajwa, K. K., & Gough, T.
(2020). Sweet satiation: Acute effects of consumption of sweet drinks on appetite for and
intake of sweet and non-sweet foods. Appetite, 149, 104631.
https://doi.org/10.1016/j.appet.2020.104631
Rogers, P. J., Hogenkamp, P. S., de Graaf, C., Higgs, S., Lluch, A., Ness, A. R., Penfold, C.,
Perry, R., Putz, P., Yeomans, M. R., & Mela, D. J. (2016). Does low-energy sweetener
consumption affect energy intake and body weight? A systematic review, including meta-
analyses, of the evidence from human and animal studies. International Journal of
Obesity (2005), 40(3), 381–394. https://doi.org/10.1038/ijo.2015.177
Rolls, B. J., Hetherington, M., & Burley, V. J. (1988). Sensory stimulation and energy density in
the development of satiety. Physiology & Behavior, 44(6), 727–733.
https://doi.org/10.1016/0031-9384(88)90053-4
Rolls, B. J., Laster, L. J., & Summerfelt, A. (1989). Hunger and food intake following
consumption of low-calorie foods. Appetite, 13(2), 115–127.
https://doi.org/10.1016/0195-6663(89)90109-8
Rolls, B. J., Rolls, E. T., Rowe, E. A., & Sweeney, K. (1981). Sensory specific satiety in man.
Physiology & Behavior, 27(1), 137–142.
Roser, M., & Ortiz-Ospina, E. (2013). Global Extreme Poverty. Our World in Data.
https://ourworldindata.org/extreme-poverty
Rudebeck, P. H., & Murray, E. A. (2014). The orbitofrontal oracle: Cortical mechanisms for the
prediction and evaluation of specific behavioral outcomes. Neuron, 84(6), 1143–1156.
Ruff, C. C., & Fehr, E. (2014). The neurobiology of rewards and values in social decision
making. Nature Reviews Neuroscience, 15(8), 549.

88
Sato, W., Kochiyama, T., Minemoto, K., Sawada, R., & Fushiki, T. (2019). Amygdala activation
during unconscious visual processing of food. Scientific Reports, 9(1), 7277.
https://doi.org/10.1038/s41598-019-43733-2
Sato, W., Sawada, R., Kubota, Y., Toichi, M., & Fushiki, T. (2017). Homeostatic modulation on
unconscious hedonic responses to food. BMC Research Notes, 10(1), 511.
https://doi.org/10.1186/s13104-017-2835-y
Scharmüller, W., Übel, S., Ebner, F., & Schienle, A. (2012). Appetite regulation during food cue
exposure: A comparison of normal-weight and obese women. Neuroscience Letters,
518(2), 106–110.
Schienle, A., Schäfer, A., Hermann, A., & Vaitl, D. (2009). Binge-eating disorder: Reward
sensitivity and brain activation to images of food. Biological Psychiatry, 65(8), 654–661.
Schur, E. A., Kleinhans, N. M., Goldberg, J., Buchwald, D., Schwartz, M. W., & Maravilla, K.
(2009). Activation in brain energy regulation and reward centers by food cues varies with
choice of visual stimulus. International Journal of Obesity, 33(6), 653.
Simmons, W. K., Martin, A., & Barsalou, L. W. (2005). Pictures of appetizing foods activate
gustatory cortices for taste and reward. Cerebral Cortex, 15(10), 1602–1608.
Smeets, P. A. M., de Graaf, C., Stafleu, A., van Osch, M. J. P., & van der Grond, J. (2005).
Functional MRI of human hypothalamic responses following glucose ingestion.
NeuroImage, 24(2), 363–368. https://doi.org/10.1016/j.neuroimage.2004.07.073
Smith, S. M., Jenkinson, M., Woolrich, M. W., Beckmann, C. F., Behrens, T. E., Johansen-Berg,
H., Bannister, P. R., De Luca, M., Drobnjak, I., & Flitney, D. E. (2004). Advances in
functional and structural MR image analysis and implementation as FSL. Neuroimage,
23, S208–S219.

89
Smith, S. M., & Nichols, T. E. (2009). Threshold-free cluster enhancement: Addressing
problems of smoothing, threshold dependence and localisation in cluster inference.
Neuroimage, 44(1), 83–98.
Snoek, H. M., Huntjens, L., van Gemert, L. J., de Graaf, C., & Weenen, H. (2004). Sensory-
specific satiety in obese and normal-weight women. The American Journal of Clinical
Nutrition, 80(4), 823–831. https://doi.org/10.1093/ajcn/80.4.823
Stice, E., Yokum, S., Bohon, C., Marti, N., & Smolen, A. (2010). Reward circuitry responsivity
to food predicts future increases in body mass: Moderating effects of DRD2 and DRD4.
Neuroimage, 50(4), 1618–1625.
Stice, E., Yokum, S., Burger, K. S., Epstein, L. H., & Small, D. M. (2011). Youth at risk for
obesity show greater activation of striatal and somatosensory regions to food. Journal of
Neuroscience, 31(12), 4360–4366.
Stoeckel, L. E., Weller, R. E., Cook III, E. W., Twieg, D. B., Knowlton, R. C., & Cox, J. E.
(2008). Widespread reward-system activation in obese women in response to pictures of
high-calorie foods. Neuroimage, 41(2), 636–647.
St-Onge, M.-P., Sy, M., Heymsfield, S. B., & Hirsch, J. (2005). Human cortical specialization
for food: A functional magnetic resonance imaging investigation. The Journal of
Nutrition, 135(5), 1014–1018.
Suzuki, S., Cross, L., & O’Doherty, J. P. (2017). Elucidating the underlying components of food
valuation in the human orbitofrontal cortex. Nature Neuroscience, 20(12), 1780–1786.
https://doi.org/10.1038/s41593-017-0008-x

90
Swithers, S. E. (2013). Artificial sweeteners produce the counterintuitive effect of inducing
metabolic derangements. Trends in Endocrinology and Metabolism: TEM, 24(9), 431–
441. https://doi.org/10.1016/j.tem.2013.05.005
Swithers, S. E., Baker, C. R., & Davidson, T. L. (2009). General and persistent effects of high-
intensity sweeteners on body weight gain and caloric compensation in rats. Behavioral
Neuroscience, 123(4), 772–780. https://doi.org/10.1037/a0016139
Swithers, S. E., & Davidson, T. L. (2008). A role for sweet taste: Calorie predictive relations in
energy regulation by rats. Behavioral Neuroscience, 122(1), 161.
Swithers, S. E., Laboy, A. F., Clark, K., Cooper, S., & Davidson, T. L. (2012). Experience with
the high-intensity sweetener saccharin impairs glucose homeostasis and GLP-1 release in
rats. Behavioural Brain Research, 233(1), 1–14.
https://doi.org/10.1016/j.bbr.2012.04.024
Tang, D. W., Fellows, L. K., Small, D. M., & Dagher, A. (2012). Food and drug cues activate
similar brain regions: A meta-analysis of functional MRI studies. Physiology & Behavior,
106(3), 317–324. https://doi.org/10.1016/j.physbeh.2012.03.009
Tetley, A., Brunstrom, J., & Griffiths, P. (2009). Individual differences in food-cue reactivity.
The role of BMI and everyday portion-size selections. Appetite, 52(3), 614–620.
https://doi.org/10.1016/j.appet.2009.02.005
Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N.,
Mazoyer, B., & Joliot, M. (2002). Automated anatomical labeling of activations in SPM
using a macroscopic anatomical parcellation of the MNI MRI single-subject brain.
Neuroimage, 15(1), 273–289.

91
van Opstal, A. M., Kaal, I., van den Berg-Huysmans, A. A., Hoeksma, M., Blonk, C., Pijl, H.,
Rombouts, S. a. R. B., & van der Grond, J. (2019). Dietary sugars and non-caloric
sweeteners elicit different homeostatic and hedonic responses in the brain. Nutrition
(Burbank, Los Angeles County, Calif.), 60, 80–86.
https://doi.org/10.1016/j.nut.2018.09.004
Van Vugt, D. A., Krzemien, A., Alsaadi, H., Frank, T. C., & Reid, R. L. (2014). Glucose-
induced inhibition of the appetitive brain response to visual food cues in polycystic ovary
syndrome patients. Brain Research, 1558, 44–56.
Van Wymelbeke, V., Beridot-Therond, M. E., de La Gueronniere, V., & Fantino, M. (2004).
Influence of repeated consumption of beverages containing sucrose or intense sweeteners
on food intake. European Journal of Clinical Nutrition, 58(1), 154–161.
Walton, M. E., Behrens, T. E., Noonan, M. P., & Rushworth, M. F. (2011). Giving credit where
credit is due: Orbitofrontal cortex and valuation in an uncertain world. Annals of the New
York Academy of Sciences, 1239(1), 14–24.
Weijzen, P. L., Smeets, P. A., & de Graaf, C. (2009). Sip size of orangeade: Effects on intake
and sensory-specific satiation. British Journal of Nutrition, 102(7), 1091–1097.
Whitehouse, C. R., Boullata, J., & McCauley, L. A. (2008). The Potential Toxicity of Artificial
Sweeteners. AAOHN Journal, 56(6), 251–261.
https://doi.org/10.1177/216507990805600604
WHO Consultation on Obesity (1999: Geneva S., & Organization W. H. (2000). Obesity:
Preventing and managing the global epidemic : report of a WHO consultation. World
Health Organization. https://apps.who.int/iris/handle/10665/42330

92
Wilson, A. L., Buckley, E., Buckley, J. D., & Bogomolova, S. (2016). Nudging healthier food
and beverage choices through salience and priming. Evidence from a systematic review.
Food Quality and Preference, 51, 47–64. https://doi.org/10.1016/j.foodqual.2016.02.009
Winkler, A. M., Ridgway, G. R., Webster, M. A., Smith, S. M., & Nichols, T. E. (2014).
Permutation inference for the general linear model. Neuroimage, 92, 381–397.
Woolrich, M. W., Jbabdi, S., Patenaude, B., Chappell, M., Makni, S., Behrens, T., Beckmann,
C., Jenkinson, M., & Smith, S. M. (2009). Bayesian analysis of neuroimaging data in
FSL. Neuroimage, 45(1), S173–S186.
Yokum, S., Gearhardt, A. N., Harris, J. L., Brownell, K. D., & Stice, E. (2014). Individual
differences in striatum activity to food commercials predict weight gain in adolescents.
Obesity (Silver Spring, Md.), 22(12), 2544–2551. https://doi.org/10.1002/oby.20882
Yokum, S., Ng, J., & Stice, E. (2011). Attentional bias to food images associated with elevated
weight and future weight gain: An fMRI study. Obesity, 19(9), 1775–1783.
Yunker, A. G., Patel, R., & Page, K. A. (2020). Effects of Non-nutritive Sweeteners on Sweet
Taste Processing and Neuroendocrine Regulation of Eating Behavior. Current Nutrition
Reports, 1–12.

93
Appendix

Attribute-rating Task (Outside the MRI Scanner)
After all three imaging sessions were completed, participants rated their estimates of
nutrient content for each food item. Participants were not informed that they would provide these
estimates prior to the task.
For each food item, participants were asked:
(1) how high is the item in fat?
(2) how high is the item in carbohydrates?
(3) how high is the item in protein?
(4) how high is the item in vitamins?
(5) how high is the item in sugar?
(6) how high is the item in sodium (salt)?
(7) how high is the item in calories?
(8) how familiar is the item?
(9) how healthy is the item?
(10) how palatable is the item?
Participants were instructed to indicate their guess about the nutrient by moving the slider
density from “not at all” to “very much”. In addition, participants were asked to report their
guess for the market price for each food item. The order of the questions was randomized across
participants.
Overlapped Medial OFC ROI
In order to examine study manipulation effects on value-tracking activity, I focused on
the medial orbitofrontal cortex, based on the extensive literature establishing its importance both

94
in the context of food valuation(Plassmann et al., 2007; Suzuki et al., 2017) and of valuation
more generally (Padoa-Schioppa & Cai, 2011; Peters & Büchel, 2010; Walton et al., 2011).
Because the medial OFC is large and heterogeneous, I limited the ROI to the overlap between
the anatomically defined medial OFC based on the AAL database (Tzourio-Mazoyer et al., 2002)
and the cluster-map identifying bid-tracking activations during food valuation (across conditions
to avoid “double dipping” confound, see (Kriegeskorte et al., 2009)). An exploratory
psychophysiological interaction (PPI) analysis was then performed in which I used the time-
series of mean activity in this cluster as a seed to predict activity throughout the rest of the brain
during the food valuation period vs. rest. A Group-level statistics images were thresholded with a
cluster-forming threshold of z > 3.1 and a Bonferroni corrected cluster probability of p < 0.05.
Three group-level paired-t analyses (sucralose vs. water, glucose vs. water and sucralose vs.
glucose) were performed in FEAT using a mixed-effects model, with FSL's FLAME1 option
with outlier deweighting.
Bid-correlated Brain Activity during Food Valuation Period
Regions in which brain activity during food valuation was positively associated with bids
across drinks are presented in Figure A1.

95
Figure A1
Brain Regions in which Activity was Positively Associated with Bids

Many of the regions associated with bid (greater signal on trials in which participants bid
more money) overlapped the food valuation period main effect contrast map (Figure A1). This
included bid-correlated clusters in the OFC, visual cortex, cingulate, paracingulate gyrus, frontal
pole, frontal gyrus, thalamus, caudate, brain stem, hippocampus, putamen, and accumbens.
Exceptions to this overlap with the general food valuation activity map were a large cluster
within the medial OFC and small bilateral clusters in the frontal pole, which tracked food bids
but which were not generally active during food valuation relative to rest. No significant
differences were observed between drink sessions on the association between bids and brain
activity.

96
Because the medial OFC has been implicated in valuation, and was prominent in the map
of bid tracking, I carried out an exploratory PPI in which I 1) identified clusters in which activity
was more correlated with the seed during food valuation, and 2) investigated whether there were
regions in which connectivity with the bid-tracking seed significantly differed based on drink
condition. For example, it might be the case that regions preferentially sensitive to particular
qualities of depicted food (e.g., sweetness) would differ in their association with the medial OFC
seed as a function of drink. As shown in Figure A2, I identified regions more correlated with the
orbitofrontal seed during the food valuation period than in other periods of the task. Significant
functional connectivity with the medial OFC was observed in a bilateral network of regions that
included the caudate, anterior insula, and nucleus accumbens, as well as part of the frontal pole,
part of lateral OFC. In contrasts of PPI results between drink conditions, no significant
differences were observed either in the whole brain nor in the areas showing positive functional
connectivity.

Figure A2
Positive Connectivity with the Medial OFC during Food Valuation

97
As expected, activity correlated with participants’ bids was observed throughout regions
previously linked to appetite but was especially prominent in a large cluster of the medial
orbitofrontal cortex. The overlapped medial OFC was not part of the network that was generally
recruited during the task, but its association with value is in keeping with an extensive literature
in neuroeconomics (Padoa-Schioppa & Cai, 2011; Peters & Büchel, 2010; Suzuki et al., 2017;
Walton et al., 2011). Based on psychophysiological interaction analysis (PPI) I identified the
overlapped medial OFC cluster to be functionally connected during food valuation to a network
of regions that included bilateral caudate, anterior insula, nucleus accumbens, the frontal pole,
and part of lateral OFC. I observed no statistical evidence that drink consumption altered
functional connectivity with the overlapped medial OFC.
Subliminal Priming Data Analyses
Since recent work suggests response to very brief (subliminal) presentation of food-cues
is sensitive to metabolic state (Sato et al., 2017), and includes a distinct neural pathway driving a
response in the amygdala (Sato et al., 2019), I included subliminal food primes embedded within
the task. My analyses focused on the food valuation period, and the GLM included the following
regressors: 1) Subliminal food-cue priming; 2) Subliminal checkerboard priming; 3) food
valuation period unweighted and 4) food valuation period weighted by bid for the food (de-
meaned relative to all bids in the run). Trials in which participants did not respond were modeled
with a fifth regressor (treated as a regressor of no interest).
To examine possible effects of study drink on subliminal priming related brain activity,
ROI analyses were carried out in order to examine amygdala response to subliminal primes.
Brain signals were extracted from bilateral amygdala, which was taken from the probabilistic
Harvard-Oxford Cortical Structural Atlas with probability larger than 50%, for the subliminal

98
prime period. Individual signal changes (beta values from statistical models) were extracted
separately for each participant during each session. LMM analysis was done to test the drink
effect on brain signal change for the ROI described above with drink type as a fixed effect and
participant entered as a random intercept with baseline hunger, appetite for sweet food, BMI and
gender as covariates.
Mean beta values within the amygdala (bilateral) were calculated for the subliminal
prime period contrast between the presence vs. absence of the 33ms food-cue presentation.
Collapsing across all conditions, I observed no evidence of differentiation within the amygdala
based on the presence of the subliminal prime (t (27) = 1.26, P = 0.22). Nor did I see any
difference between drink conditions in this contrast (F (2,131) = 1.50, P = 0.23). I do not
consider the effect of subliminal prime in further analyses.
Although I intended to use response to subliminal primes to examine whether drink
effects were present for the rapid orienting response previously documented in the amygdala for
food (Sato et al., 2019) and other reward cues (Childress et al., 2008), I did not observe any
signal in the amygdala associated with the subliminal food primes even when all sessions were
combined. This was likely because the relatively low number of subliminal presentations.
Correlation of Subjective and Objective Ratings of Carbohydrate, Fat and Protein
Participants’ ratings of carbohydrate, fat and protein were all significantly correlated
with the objective ratings (P < 0.05), see Figure A3 description for mean correlations and t-test
results.

99
Figure A3
Correlations between the Subjective and the Actual Nutritive Attributes

Note. In each box and whisker plot, significant results (P < 0.05) were indicated with *. T test (Carb (carbohydrate):
mean = 0.53, t(25) = 23.9, P = 2.2e-16; fat: mean = 0.47, t(25) = 20.97, P = 2.2e-16; protein: mean = 0.40, t(25) =
12.03, P = 6.846e-12; calories: mean = 0.36,t(25) =13.80, P = 3.414e-13).

Predicting Food Value with Caloric Content
It turned out that prediction models with objective carbohydrate content, subjective
carbohydrate content, objective caloric content, and subjective caloric content as predictor could
all significantly predict food value at above chance level, see Figure A4 for mean prediction
accuracy and t-test results. Furthermore, the regression model that included subjective ratings of
carbohydrate (t(25) = 1.15, P = 0.26), objective (actual) overall caloric content (t(25) = -1.37, P =
0.18), and subjective beliefs of caloric contents (t(25) = -0.63, P = 0.53) all had similar
prediction accuracy in predicting food value compared with prediction model including actual
carbohydrate content.

Carb Fat Protein Calories
0.0 0.2 0.4 0.6 0.8 1.0
data$nutrient
Correlation subjective vs. objective
* * * *

100
Figure A4
Prediction Accuracy of the Food Value in Each Regression Model

Note. Performance was assessed by the cross-validated correlation between the predicted and actual values. In each
box and whisker plot, significant results (P < 0.05) were indicated with *. T test (Obj_Carb (objective carbohydrate):
mean = 0.37, t(25) = 7.64, P = 5.337e-08; Sub_Carb (subjective carbohydrate): mean = 0.44, t(25) = 9.72, P =
5.675e-10; Obj_Calories (objective calories): mean = 0.28, t(25) = 6.30, P =1.363e-06; Sun_Calories (subjective
calories): mean = 0.33, t(25) = 9.97, P = 3.427e-10.

Obj_Carb Sub_Carb Obj_Calories Sub_Calories
0.0 0.2 0.4 0.6 0.8 1.0
data$nutrient
Correlation model vs. actual value
* * * *

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Behabioral and neural evidence of state-like variance in intertemporal decisions

PDF

Homeostatic imbalance and monetary delay discounting: effects of feeding on RT, choice, and brain response

PDF

Behavioral and neural evidence of incentive bias for immediate rewards relative to preference-matched delayed rewards

PDF

Effects of sugar and non-nutritive sweetener consumption on neural processing, ingestive behavior, and appetite regulation

PDF

Value-based decision-making in complex choice: brain regions involved and implications of age

PDF

Sequential decisions on time preference: evidence for non-independence

PDF

Reward substitution: how consumers can be incentivized to choose smaller food portions

PDF

Insula activity during safe-sex decision-making in sexually risky men suggests negative urgency and fear of rejection drives risky sexual behavior

PDF

Validation of a neuroimaging task to investigate decisions involving visceral immediate rewards

PDF

Heart, brain, and breath: studies on the neuromodulation of interoceptive systems

PDF

Effects of western dietary factors during early life on glucose metabolism, the gut microbiome, and neurocognition

PDF

Food deserts and perceptions of food access in urban low-income areas

PDF

Classically conditioned responses to food cues among obese and normal weight individuals: conditioning as an explanatory mechanism for excessive eating

PDF

Brain and behavior correlates of intrinsic motivation and skill learning

PDF

The evolution of decision-making quality over the life cycle: evidence from behavioral and neuroeconomic experiments with different age groups

PDF

The interplay between social connection and substance use

PDF

The effect of present moment awareness and value intervention of ACT on impulsive decision-making and impulsive disinhibition

PDF

A multiattribute decision model for the selection of radioisotope and nuclear detection devices

PDF

The impact of treatment decisions and adherence on outcomes in small hereditary disease populations

Asset Metadata

Creator Zhang, Xiaobei (author)

Core Title The acute impact of glucose and sucralose on food decisions and brain responses to visual food cues

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Degree Conferral Date 2021-08

Publication Date 07/12/2021

Defense Date 06/03/2021

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

Tag carbohydrate,fat,fMRI,food decision,food valuation,glucose,non-nutritive sweetener,nutrient tracking,OAI-PMH Harvest,sucralose

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Monterosso, John (committee chair), Bechara, Antoine (committee member), John, Richard (committee member), Luo, Shan (committee member), Page, Kathleen (committee member)

Creator Email 2365103343@qq.com,xiaobeiz@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC16255862

Unique identifier UC16255862

Legacy Identifier etd-ZhangXiaob-9724

Document Type Dissertation

Format application/pdf (imt)

Rights Zhang, Xiaobei

Type texts

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

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

Repository Name University of Southern California Digital Library

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

Repository Email cisadmin@lib.usc.edu