**Food Addiction and Eating Addiction: Scientific Advances and Their Clinical, Social and Policy Implications**

Special Issue Editors

**Adrian Carter Tracy Burrows Charlotte Hardman**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Special Issue Editors* Adrian Carter Monash University Australia

Tracy Burrows University of Newcastle Australia

Charlotte Hardman University of Liverpool UK

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Nutrients* (ISSN 2072-6643) (available at: https://www.mdpi.com/journal/nutrients/special issues/ Food Addiction Eating Addiction).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03936-358-2 (Pbk) ISBN 978-3-03936-359-9 (PDF)**

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**


**Kerry S. O'Brien, Rebecca M. Puhl, Janet D. Latner, Dermot Lynott, Jessica D. Reid, Zarina Vakhitova, John A. Hunter, Damian Scarf, Ruth Jeanes, Ayoub Bouguettaya and Adrian Carter**

The Effect of a Food Addiction Explanation Model for Weight Control and Obesity on Weight Stigma

Reprinted from: *Nutrients* **2020**, *12*, 294, doi:10.3390/nu12020294 ................... **159**

### **About the Special Issue Editors**

**Adrian Carter** (Ph.D.): Associate Professor Adrian Carter is an NHMRC Research Fellow and Director, Community Engagement and Neuroethics, Turner Institute for Brain and Mental Health, Monash University. He is also: Director, Neuroethics Program, ARC Centre of Excellence for Integrative Brain Function; Co-Chair, Neuroethics and Responsible Research and Innovation Committee, Australian Brain Alliance; Co-Editor-in-Chief, Neuroethics (Springer); and sits on the Board of Directors, International Neuroethics Society. His research examines the impact of neuroscience on our understanding and treatment of mental and neurological disorders. Dr Carter has been an advisor to the WHO, OECD, European Monitoring Centre for Drugs and Drug Addiction, and United Nations Office on Drugs and Crime.

**Tracy Burrows** (Ph.D.): Associate Professor Tracy Burrows is a NHMRC Research Fellow at The University of Newcastle and a researcher at the Hunter Medical Research Institute. She is recognized as a Fellow of the Dietitians Association of Australia. Her research focuses on understanding eating behavior with an interest in mental health populations, dietary assessment and weight management.

**Charlotte Hardman** (Ph.D.): Dr Charlotte Hardman is a Senior Lecturer in the University of Liverpool. She is also: Fellow of the Higher Education Academy (FHEA); Co-ordinator of the North West Network of the UK Association for the Study of Obesity; and Advisory Editor for the journal Appetite. Her research focuses on the psychology of food-related behaviour and the application of this knowledge to interventions for behaviour change. She has received research funding from UK Research and Innovation, the European Commission, and the Wellcome Trust. She has published over 60 peer-reviewed articles in high-impact journals such as JAMA Psychiatry, Nature Reviews Endocrinology and the International Journal of Obesity.

### *Editorial* **Food Addiction and Eating Addiction: Scientific Advances and Their Clinical, Social and Policy Implications**

### **Adrian Carter 1,\*, Charlotte A. Hardman <sup>2</sup> and Tracy Burrows <sup>3</sup>**


Received: 1 May 2020; Accepted: 14 May 2020; Published: 20 May 2020

There is a growing understanding within the literature that certain foods, particularly those high in refined sugars and fats, may have addictive potential for some individuals. Moreover, individuals who are overweight and have obesity display dietary intake patterns that resemble the ways in which individuals with substance use disorders consume addictive drugs. While food addiction is not yet recognized in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), there are many similarities with substance use disorders, and a growing acceptance that some forms of obesity should be treated as a food addiction. Despite growing research in this area, there remain many unresolved questions about the science of food addiction and its potential impact upon: how we treat overweight and overeating; stigmatization and discrimination of people who are overweight; internalized weight bias and treatment seeking; as well as policies to reduce excess weight and overeating.

This interdisciplinary special issue collects 10 articles, including reviews and original research, that further our understanding and application of the science of addictive eating. These papers span a broad range of areas, including basic science, clinical assessment tools, neural responses to addictive foods, as well as insights into future treatments and public health policies, and the possible stigma associated with food addiction.

### **1. Validation of Food Addiction Scales**

### *Validation of the Japanese Version of the Yale Food Addiction Scale 2.0 (J-YFAS 2.0)*

The Yale Food Addiction Scale (YFAS) is the most widely used diagnostic tool for food addiction, and has been translated into numerous languages, including Italian, French, German, Spanish, Arabic Chinese and Turkish. In this special issue, Khine and colleagues [1] describe the translation and validation of the Japanese version of the Yale Food Addiction Scale 2.0 (J-YFAS 2.0), carried out in 731 undergraduate students. The J-YFAS 2.0 has a one-factor structure and adequate convergent validity and reliability, similar to the YFAS 2.0 in other languages. Prevalence of J-YFAS 2.0-diagnosed mild, moderate, and severe food addiction was 1.1%, 1.2%, and 1.0% respectively.

### **2. Neural Responses Underlying Food Addiction**

### *2.1. Food Addiction Symptoms and Amygdala Response in Fasted and Fed States*

Pursey et al. [2] conducted a small pilot study to explore the association between food addiction symptoms and activationin the basolateral amygdala and central amygdala. 12 females, aged24.1 ± 2.6 years, completed two functional magnetic resonance imaging (fMRI) scans (fasted and fed) while viewing high-calorie food images and low-calorie food images. Food addiction symptoms were assessed using the Yale Food Addiction Scale. Participants had a mean BMI of 27.4 <sup>±</sup> 5.0 kg/m2, and food addiction symptom score of 4.1 ± 2.2. The results found a significant positive association, between food addiction symptoms, and higher activation of the left basolateral amygdala to high-calorie versus low-calorie foods in the fasted session, but not the fed session. There were no significant associations with the central amygdala in either session.

### *2.2. Increasing Chocolate's Sugar Content Enhances Its Psychoactive E*ff*ects and Intake*

This study by Caperson and colleagues explored the potential psychoactive effect of chocolate [3]. Participants consumed 5 g of a commercially available chocolate with increasing amounts of sugar (90% cocoa, 85% cocoa, 70% cocoa, and milk chocolate). After each chocolate sample, participants completed the Psychoactive Effects Questionnaire (PEQ) and the Binge Eating Scale (BES). Participants were also allowed to eat as much as they wanted of each of the different chocolates. Casperson et al. [3] found that the excitement subscale of the PEQ increased (relative to baseline) after the 90% cocoa. The Morphine–Benzedrine Group subscale (containing questions about wellbeing and euphoria) and the Morphine subscale (focusing on attitudes and physical sensations) increased after the 85th cocoa sample. This suggests incremental increases in the sugar content of chocolate has a psychoactive effect which enhances the addictive-like eating response.

### **3. Implications for Treatment**

### *3.1. Food Addiction: Implications for the Diagnosis and Treatment of Overeating*

The validity of a food addiction diagnosis remains controversial, despite a growing body of preclinical, neurobiological and clinical evidence supporting it. This literature review discusses the DSM-5 diagnostic criteria for substance use disorders, to summarize evidence for food addiction. Adams and colleagues [4] concluded that there is evidence to suggest that, for some individuals, food can induce addictive-type behaviors similar to those seen with other addictive substances. However, as several DSM-5 criteria have limited application to overeating, they argue that the term 'food addiction' is likely to apply only in a minority of cases. Research investigating the underlying psychological causes of overeating within the context of food addiction has also led to some novel treatment approaches, such as cognitive training tasks and neuro-modulation interventions.

### *3.2. Food Addiction in Eating Disorders and Obesity: Analysis of Clusters and Implications for Treatment*

This study by Jimenez-Murcia et al. [5] identified three distinct clusters of food addiction in those with eating disorders and obesity. The study was conducted in 234 participants who scored positive on the Yale Food Addiction Scale 2.0. Cluster 1, classified as "dysfunctional", was associated with the highest prevalence of other specified feeding or eating disorders and bulimia nervosa, as well as the highest eating disorder severity and psychopathology, and more dysfunctional personality traits. Cluster 2, classified as "moderate", was associated with high prevalence of bulimia nervosa and binge eating disorders, and moderate levels of eating disorder psychopathology. Cluster 3, classified as "adaptive", was characterized by high prevalence of obesity and binge eating disorders, low levels of eating disorder psychopathology and more functional personality traits. The authors suggest that the identification of types of food addiction traits may allow for more personalized treatment to improve outcomes.

### *3.3. Food Addiction Is Associated with Irrational Beliefs via Trait Anxiety and Emotional Eating*

Irrational beliefs are believed to be one of the prime causes of psychopathologies, including anxiety and depression. A study of 239 adults by Nolan and colleagues [6] investigated whether food addiction and emotional eating are associated with irrational beliefs. Questionnaires measuring food addiction, irrational beliefs, emotional eating, depression, trait anxiety, and anthropometry were assessed and reported. They found that irrational beliefs were significantly positively correlated with food addiction, emotional eating, depression and trait anxiety. Results also showed that irrational beliefs were associated with higher food addiction via higher trait anxiety and emotional eating acting in a serial pathway. As such, targeting irrational beliefs as a treatment in individuals who experience food addiction and emotional eating may be a reasonable approach for clinicians.

### *3.4. Fat Addiction: Psychological and Physiological Trajectory*

A number of recent studies have attempted to parse out the psychological and physiological etiology of food addiction. This review article by Sarkar et al. [7] examines the specific role of dietary fats in compulsive overeating. They review preclinical, psychological and clinical evidence to argue for the addiction to fat rich diets as a prominent subset of food addiction. They then discuss the clinical implications of "fat addiction" for society.

### **4. Associations between Food Addiction, Stigma and Public Policy**

### *4.1. Ethical, Stigma, and Policy Implications of Food Addiction: A Scoping Review*

This scoping review by Cassin and colleagues [8] examines the potential ethical, stigma and health policy implications of food addiction described in the current literature. Their findings suggest that the literature on potential ethical implications was mostly focused on debates regarding individualized responsibility and sources for blame. Potential stigma focused on evidence of internalized and externalized stigma when food addiction is used as the explanation for obesity. The policy implications of food addiction largely drew on comparisons with the historic regulation of the tobacco industry to manage food addiction in policy, and the current challenges in classifying foods in terms of their addictive potential.

### *4.2. Obesity Stigma: Is the 'Food Addiction' Label Feeding the Problem?*

There is significant debate around whether describing someone as addicted to food would increase or decrease weight-based stigma. Ruddock et al. [9] examined the effect of the food addiction label on stigmatizing attitudes towards an individual with obesity, and towards people with obesity more generally (i.e., general stigma). They presented the results of two online studies, where participants (n = 439, n = 523) read a short description about a woman described as 'very overweight'. They found that a food addiction label may exacerbate stigmatizing attitudes towards an individual with obesity. However, the label appears to have no effect on general weight-based stigma. Stigmatizing attitudes towards people with obesity also appeared to be more pronounced in individuals with low levels of addiction-like eating behaviors, compared to high levels of addiction-like eating.

### *4.3. The E*ff*ect of a Food Addiction Explanation Model for Weight Control and Obesity on Weight Stigma*

In the final paper of this special issue, O'Brien and colleagues [10] reported on two experimental studies examining the impact of a food addiction model of obesity and weight control on weight stigma. In both experiments, participants were randomized to receive one of two newspaper articles: one describing obesity as the result of a brain-based food addiction, and the other describing obesity as the result of diet and exercise. The food addiction explanation for weight control and obesity did not increase weight stigma, and resulted in lower stigma than the diet and exercise explanation, which attributes obesity to personal control. Their findings highlight the need for evidence-based health messaging about the causes of obesity, and the need for communications that do not exacerbate weight stigma.

**Author Contributions:** T.B. developed an initial draft of the manuscript that was updated by A.C. All authors revised the final manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Validation of the Japanese Version of the Yale Food Addiction Scale 2.0 (J-YFAS 2.0)**

### **May Thet Khine 1, Atsuhiko Ota 1,\*, Ashley N. Gearhardt 2, Akiko Fujisawa 1, Mamiko Morita 3, Atsuko Minagawa 3, Yuanying Li 1, Hisao Naito <sup>1</sup> and Hiroshi Yatsuya <sup>1</sup>**


Received: 9 February 2019; Accepted: 19 March 2019; Published: 22 March 2019

**Abstract:** The Yale Food Addiction Scale 2.0 (YFAS 2.0) is used for assessing food addiction (FA). Our study aimed at validating its Japanese version (J-YFAS 2.0). The subjects included 731 undergraduate students. Confirmatory factor analysis indicated the root-mean-square error of approximation, comparative fit index, Tucker–Lewis index, and standardized root-mean-square residual were 0.065, 0.904, 0.880, and 0.048, respectively, for a one-factor structure model. Kuder–Richardson α was 0.78. Prevalence of the J-YFAS 2.0-diagnosed mild, moderate, and severe FA was 1.1%, 1.2%, and 1.0%, respectively. High uncontrolled eating and emotional eating scores of the 18-item Three-Factor Eating Questionnaire (TFEQ R-18) (*p* < 0.001), a high Kessler Psychological Distress Scale score (*p* < 0.001), frequent desire to overeat (*p* = 0.007), and frequent snacking (*p* = 0.003) were associated with the J-YFAS 2.0-diagnosed FA presence. The scores demonstrated significant correlations with the J-YFAS 2.0-diagnosed FA symptom count (*p* < 0.01). The highest attained body mass index was associated with the J-YFAS 2.0-diagnosed FA symptom count (*p* = 0.026). The TFEQ R-18 cognitive restraint score was associated with the J-YFAS 2.0-diagnosed FA presence (*p* < 0.05) and symptom count (*p* < 0.001), but not with the J-YFAS 2.0-diagnosed FA severity. Like the YFAS 2.0 in other languages, the J-YFAS 2.0 has a one-factor structure and adequate convergent validity and reliability.

**Keywords:** food addiction; Japan; validation; Yale Food Addiction Scale 2.0

### **1. Introduction**

The idea of food addiction (FA) is receiving increased interest [1]. Evidence is emerging that certain types of foods (e.g., highly processed foods with high levels of refined carbohydrates and/or added fat) may be capable of triggering addictive-like eating behaviors (e.g., loss of control, withdrawal, and cravings) in some individuals, which can lead to significant impairment or distress, [2,3]. Obesity and eating disorders such as bulimia nervosa (BN), binge eating disorders (BED), along with psychiatric disorders such as depression, posttraumatic stress disorder, attention-deficit hyperactivity disorder, have been reported as potential correlates with FA [4–6]. Relevant pharmacological findings have been reported. Highly processed sweetened and fatty foods trigger a rewarding effect through the release of dopamine [7]. Repeated eating of hyper-palatable food down-regulates the dopaminergic response, resulting in impulsive and compulsive responses to food cues [8]. Food craving—an intense

desire to eat a specific food—activates the hippocampus, insula, and caudate nucleus, similar to drug craving [9]. On the other hand, there has been a lot of debate regarding the extent to which food can be addictive in the same way as drugs. Controversies exist, for instance, as to whether FA represents a specific construct as addiction that is distinct from other eating disorders, such as BED, and whether neurobiological changes underlying FA behaviors are sufficiently ascertained in humans [10,11].

The Yale Food Addiction Scale (YFAS) is the most commonly used measure to assess FA, although FA is not included in the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5) [12] and controversy exists regarding its definition [11]. The original YFAS applies the DSM 4th edition (DSM-IV) diagnostic criteria for substance dependence to the consumption of highly palatable foods (e.g., chocolate, ice cream, and pizza) [13,14]. Later, the scale was replaced with the Yale Food Addiction Scale 2.0 (YFAS 2.0) in response to the revision of the Substance-Related and Addictive Disorders criteria in the DSM-5 [15]. The YFAS 2.0 additionally introduced the following four diagnostic criteria: craving, use despite interpersonal or social consequences, failure in role obligations, and use in physically hazardous situations. It also introduced a severity classification. The YFAS 2.0 is available not only in English but also in German, French, Italian, Spanish, and Arabic [16–20]. The YFAS 2.0 exhibits good internal consistency, as well as convergent, discriminant, and incremental validity [15–20]. Associations of the YFAS-diagnosed FA with obesity, eating disorders, and psychiatric disorders have been accumulated. The YFAS 2.0-defined FA prevalence is supposed to draw a J-shape curve according to body mass index (BMI): 3.3–15.8% in healthy general populations [15–20], 17.2–47.4% in obese population [16,21], and 15.0% in underweight population [21]. Women and patients with eating disorders (BN and BED) and mental disorders (depression, sleep disturbance, and general psychiatric status) were more likely to have FA diagnosed with the YFAS 2.0 [15,17–19].

The current study aimed to validate the Japanese version of YFAS 2.0 (J-YFAS 2.0). Scant evidence regarding FA is available in Asia. The previous version of YFAS was translated into Chinese [22,23] and Malay [24]. Using these questionnaires, researchers reported that a FA diagnosis was assigned to 6.9–9.2% of Chinese teenage students [22,23] and 10.4% of Malay obese adults [24]. FA prevalence in Japan has not been reported so far, to the best of our knowledge. The YFAS 2.0 has not yet been translated into Asian languages. Development of the J-YFAS 2.0 enables examining the FA prevalence in Japan, comparing it with other countries and regions, and exploring the mechanism of FA. Referring to previous research [15–19], we hypothesized that (1) the J-YFAS has a one-factorial structure for the 11 J-YFAS 2.0 diagnostic criteria (structural validity); (2) underweight, overweight, obesity, uncontrolled and emotional eating, frequent desire to overeat, frequent snacking, and mood and anxiety disorders are associated with the J-YFAS 2.0-diagnosed FA (convergent validity); (3) cognitive restraint in eating is not associated with the J-YFAS 2.0-diagnosed FA (discriminant validity); and (4) the internal consistency is good for the 11 J-YFAS 2.0 diagnostic criteria (reliability).

### **2. Subjects and Methods**

### *2.1. Study Design*

We employed a cross-sectional design. All data were collected from a questionnaire survey. The present study was completed in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects established by the Ministry of Education, Culture, Sports, Science and Technology and the Ministry of Health, Labour and Welfare, Japan. We obtained the approval by the Ethics Review Committee of Fujita Health University, Japan (HM17-110 and HM18-155). All subjects provided their informed written consent for participation in the present study.

### *2.2. Subjects*

This study was conducted with a convenience sample of undergraduate students from a private medical and health science university in Japan. The authors (A.O., M.M., and A.M.) explained the study purpose and methods to the students in the classes. Paper-based questionnaires were then distributed. Of the 759 students to whom the questionnaires were distributed, 752 (99%) were returned. Those who did not provide informed consent (*n* = 2) and who did not fully complete the J-YFAS 2.0 (*n* = 18) were excluded from the analysis. One student who replied to experience desire to overeat 50 times per week was excluded as this reply was a significant outlier. Consequently, we retained the remaining 731 students (96%) as the subjects.

### *2.3. J-YFAS 2.0*

As with the YFAS 2.0 [15], the J-YFAS 2.0 is a 35-item self-administered questionnaire (Table S1). It assesses food consumption during the past 12 months. A Likert-scale ranging from 0 (never) through 7 (every day) is employed as a response option for each item. The items assess clinical impairment/distress and the following 11 diagnostic criteria: (1) eating larger amounts for a longer period than intended (consumed more than intended); (2) persistent desire or repeated unsuccessful attempts to quit eating (unable to cut down or stop); (3) spending considerable time or activity obtaining or eating food or recovering from eating (great deal of time spent); (4) giving up or reducing important social, occupational, or recreational activities due to eating (important activities given up); (5) continued eating despite knowledge of adverse consequences (use despite physical/emotional consequences); (6) development of tolerance (tolerance); (7) characteristic withdrawal symptoms (withdrawal); (8) continued eating despite interpersonal or social problems (use despite interpersonal/social problems); (9) failure to fulfil major role obligation at work, school, and home due to eating (failure in role obligation); (10) eating even in physically hazardous situations (use in physically hazardous situations); and (11) craving, strong desire, or urge for certain foods (craving). Each item is scored dichotomously based on the threshold determined by the YFAS 2.0 validation paper [15]. If any item that corresponds to the diagnostic criteria or clinical severity meets the clinical threshold, this criterion is endorsed. There are two scoring methods: the symptom count and the diagnostic threshold. For the symptom count scoring method, the diagnostic criteria for which the subjects meet are summed together. For the diagnostic threshold, the clinically significant impairment/distress criterion has to be met and two or more diagnostic criteria have to be met. The J-YFAS 2.0 FA diagnostic severity is classified as mild (2–3 criteria met plus impairment/distress), moderate (4–5 criteria met plus impairment/distress), and severe (6–11 criteria met plus impairment/distress).

For the development of the J-YFAS 2.0, the original English YFAS 2.0 [15] was translated into Japanese by the three Japanese authors (A.O., A.F., and H.Y.) and back-translated into English by an external professional translator who had no previous knowledge of the YFAS 2.0. Discrepancies between the back-translation and the original were resolved by consensus amongst the three Japanese authors and an American author (A.N.G.), who developed the original YFAS 2.0. We added two food examples, wagashi (Japanese traditional confectionery) and instant noodles (as salty snacks), in the introductory part, considering that food preference in Japan differs from western countries.

### *2.4. Variables for Convergent and Discriminant Validity*

### 2.4.1. Body Mass Index (BMI)

Each subject self-reported their current and highest attained BMI. The questionnaire included a table indicating BMI from the weights and heights so that the subjects could choose their BMI from the following options: <16.0, 16.0–16.9, 17.0–18.4, 18.5–22.9, 23.0–24.9, 25.0–29.9, and ≥30.0 kg/m2. No one chose <16.0 kg/m2 for their current or highest attained BMI.

It was reported that Japanese tended to under-report their body weights and the tendency was more prominent among those with high BMI than those with low BMI [25]. We arranged the categorical response options for BMI to minimize the shame that subjects may feel for self-reporting their actual BMI.

### 2.4.2. Three-Factor Eating Questionnaire Revised 18-Item Version (TFEQ R-18)

The TFEQ R-18 is a self-assessment tool used to measure the following three types of eating behaviors: Cognitive restraint, uncontrolled eating, and emotional eating [26]. Cognitive restraint is a control over food intake in order to influence body weight and body shape [26]. Uncontrolled eating is a tendency to overeat food with the feeling of being out of control [27]. Emotional eating is a tendency to eat in response to negative emotions [27]. The higher the score is, the greater the levels of cognitive restraint, uncontrolled eating, and emotional eating are. We chose the corresponding items for the current study from the Japanese version of the original 51-item TFEQ [28].

### 2.4.3. Desire to Overeat

No validated questionnaire was available in Japanese to evaluate binge eating frequency. Thus, we asked the frequency of desiring to overeat with a single question, "How many times per week did you feel you wanted to eat more even after eating quite a lot of food during the last two hours?" The subjects filled in the number of the times.

### 2.4.4. Snacking Frequency

A frequency of snacking (eating and drinking outside of breakfast, lunch, or dinner) was self-reported. No validated questionnaire was available in Japanese to evaluate the frequency of snacking. Thus, we developed a single question, "How many days per week are you snacking?" for this evaluation. The subjects chose one of the following options: none, 2–3 days, 4–5 days, and almost every day. The snack included foods and drinks that contained any calories. Zero-calorie drinks, such as coffee and tea without milk and sugar, and vitamin and mineral supplements were excluded from the snack.

### 2.4.5. Kessler Psychological Distress Scale (K6)

The Japanese version of K6 was used as an indicator of mood and anxiety disorders [29]. A K6 score of 13 or greater was regarded as having such disorders.

### *2.5. Statistical Analyses*

Confirmatory factor analysis (CFA) was conducted to assess the one-factor structure for the 11 J-YFAS 2.0 diagnostic criteria. Clinically significant impairment/distress was not included in this CFA analysis. The model fit was evaluated with the root-mean-square error of approximation (RMSEA), comparative fit index (CFI), Tucker–Lewis index (TLI), and standardized root-mean-square residual (SRMR). For assessing the reliability, internal consistency was calculated for the 11 J-YFAS 2.0 diagnostic criteria with Kuder–Richardson's α (KR-20). Convergent and discriminant validity was examined with chi-square test, *t*-test, analysis of variance (ANOVA), and Spearman's rank correlation. We examined whether the current and highest attained BMI, TFEQ R-18 cognitive restraint, uncontrolled eating, and emotional eating scores, frequency of desire to overeat, snacking frequency, and K6 score were associated with the J-YFAS 2.0-diagnosed FA. Not only the presence and severity (mild, moderate, and severe) but also the symptom count was used as the J-YFAS 2.0-diagnosed FA index, given the small numbers of subjects diagnosed as having FA. We could not apply the chi-square test to examine the associations of BMI, high K6 score, and the snacking frequency with the J-YFAS 2.0-diagnosed FA severity, since more than 20% of all cells had an expected frequency of less than five. Effect size indices were calculated [30–32]. Subjects with missing responses were excluded from the corresponding analyses. SPSS version 23.0 (IBM, Armonk, NY, USA) and Amos Version 23.0 (IBM, Chicago, IL, USA) were used for statistical calculations.

### **3. Results**

### *3.1. Subjects' Characteristics*

Most subjects were women (78.5%, *n* = 574) (Table 1). The mean (standard deviation) age was 20.8 (1.8) years. The years and majors included fourth-year medical technology students, first- to fourth-year nursing students, and third-year medical students. Around 80% of the subjects reported normal-weight BMI, 18.5–24.9 kg/m2.



SD: standard deviation. There were missing responses for sex (*n* = 1), age (*n* = 1), current BMI (*n* = 1), K6 (*n* = 4), the TFEQ R-18 cognitive restraint (*n* = 8), uncontrolled eating (*n* = 12), and emotional eating (*n* = 2), desire to overeat (*n* = 1), and snacking frequency (*n* = 1). \* Highest attained BMI means the highest weight ever (when not pregnant) during the lifetime.

### *3.2. CFA and Internal Consistency*

The RMSEA, CFI, TLI, and SRMR were 0.065, 0.904, 0.880, and 0.048, respectively. One diagnostic criterion (failure in role obligation) indicated a factor loading of 0.31 (Table 2). The other diagnostic criteria had factor loadings of 0.41 or higher. The KR-20 was 0.78 for the 11 diagnostic criteria.


**Table 2.** Diagnostic criteria of the Japanese version of the Yale Food Addiction Scale 2.0 (*n* = 731).

\*\*\* *p* < 0.001, calculated with confirmatory factor analysis.

### *3.3. J-YFAS 2.0-Diagnosed FA Prevalence*

The mean J-YFAS 2.0-diagnosed FA symptom count was 0.84 (SD = 1.61; range = 0–11). The proportions of the subjects who met the threshold for each diagnostic criterion ranged from 2.9–17.0% (Table 2). A total of 24 (3.3%) subjects were regarded as having FA: 8 (1.1%) received a mild, 9 (1.2%) a moderate, and 7 (1.0%) a severe FA diagnosis using the J-YFAS 2.0 (Table 1). All subjects who were diagnosed as FA with the J-YFAS 2.0 were women. Sex was significantly associated with the J-YFAS 2.0-diagnosed FA (*p* = 0.004, Fisher's Exact Test).

### *3.4. Convergent and Discriminant Validity*

For convergent validity, neither the current nor the highest attained BMI was associated with the presence of J-YFAS 2.0-diagnosed FA (Table 3). The highest attained BMI was associated with the J-YFAS 2.0-diagnosed FA symptom count, while the current BMI was not (Table 4). The effect size was small for the association between the highest attained BMI and the J-YFAS 2.0-diagnosed FA symptom count—the η<sup>2</sup> was 0.02. A high K6 score and snacking frequency were significantly associated with the J-YFAS 2.0-diagnosed FA presence and the J-YFAS 2.0-diagnosed FA symptom count (Tables 3 and 4, respectively). TFEQ R-18 uncontrolled eating and emotional eating scores and desire to overeat were significantly associated with the J-YFAS 2.0-diagnosed FA presence, severity, and symptom count (Tables 5–7, respectively).

For discriminant validity, there was a significant association between the TFEQ R-18 cognitive restraint score and the J-YFAS 2.0-diagnosed FA presence (Table 5). Its effect size was small—the Cohen's d was 0.44. There was no significant association between the cognitive restraint score and the J-YFAS 2.0-diagnosed FA severity (Table 6). We found a significant correlation between the cognitive restraint score and the J-YFAS 2.0-diagnosed FA symptom count (Table 7). Its Spearman's rank correlation coefficient was 0.143.


**Table 3.** Associations of body mass index (BMI), the Kessler Psychological Distress Scale (K6) score, and snacking frequency with the J-YFAS 2.0-diagnosed food addiction (FA) absence/presence.

Chi-square test was used. The numbers of missing responses were as follows: current BMI (*n* = 1), K6 score (*n* = 4), and snacking frequency (*n* = 1). \* Highest attained BMI means the highest weight ever (when not pregnant) during the lifetime.



Analysis of variance (for BMI and snacking frequency) and *t*-test (for K6 score) were used. FA symptom counts are shown as mean (standard deviation). The numbers of missing responses were as follows: Current BMI (*n* = 1), K6 score (*n* = 4), and snacking frequency (*n* = 1). <sup>a</sup> Pairwise differences were of *p* < 0.05 (Bonferroni corrected). \* Highest attained BMI means the highest weight ever (when not pregnant) during the lifetime.


**Table 5.** Associations of the 18-item Three-Factor Eating Questionnaire (TFEQ R-18) scores and frequency of desiring to overeat with the J-YFAS 2.0-diagnosed food addiction (FA) absence/presence.

*t*-test was used. TFEQ R-18 scores and frequency of desiring to overeat are shown as mean (standard deviation). The numbers of missing responses were as follows: TFEQ R-18 cognitive restraint, *n* = 8 (7 from FA absent, 1 from FA present); uncontrolled eating, *n* = 12 (11 from FA absent, 1 from FA present); emotional eating, *n* = 2 (all from FA absent); and desire to overeat, *n* = 1 (from FA absent).

**Table 6.** Associations of the 18-item Three-Factor Eating Questionnaire (TFEQ R-18) scores and frequency of desiring to overeat with the J-YFAS 2.0-diagnosed food addiction (FA) severity.


Analysis of variance was used. TFEQ R-18 scores and frequency of desiring to overeat are shown as mean (standard deviation). The numbers of missing responses were as follows: TFEQ R-18 cognitive restraint, *n* = 8 (7 from FA absent, 1 from Moderate FA); uncontrolled eating, *n* = 12 (11 from FA absent, 1 from Moderate FA); emotional eating, *n* = 2 (all from FA absent); and desire to overeat, *n* = 1 (from FA absent). <sup>a</sup> Pairwise differences were of *p* < 0.05 (Bonferroni corrected). 1 = No FA, 2 = Mild FA, 3 = Moderate FA, 4 = Severe FA.

**Table 7.** Spearman's rank correlation coefficients among the J-YFAS 2.0-diagnosed food addiction (FA) symptom count, the 18-item Three-Factor Eating Questionnaire (TFEQ R-18) scores, and frequency of desiring to overeat (*n* = 731).


\*\*\* *p* < 0.001. The numbers of missing responses were as follows: TFEQ R-18 cognitive restraint (*n* = 8), uncontrolled eating (*n* = 12), emotional eating (*n* = 2), and desire to overeat, (*n* = 1).

### **4. Discussion**

We examined the J-YFAS 2.0's properties in a sample of healthy undergraduate students in Japan. The J-YFAS 2.0 had a one-factor structure and adequate convergent validity and reliability, like the YFAS 2.0 in other languages [15–20], whereas our results were not the same as hypothesized with regard to the associations of BMI and cognitive restraint in eating with the J-YFAS 2.0-diagnosed FA. The J-YFAS 2.0-diagnosed FA prevalence was 3.3% in our subjects. Similar findings were reported from Italian and Spanish young healthy samples [18,19].

A one-factor structure was confirmed for the J-YFAS 2.0, which is the same as for the English, German, French, Italian, and Spanish YFAS 2.0 [15–19]. Our findings did not strictly meet the Hu and Bentler criteria, i.e., RMSEA ≤ 0.06, CFI ≥ 0.95, TLI ≥ 0.95, and SRMR ≤ 0.08 [33]. However, one or more of the four indices do not often meet the criteria [34]. The French version of the YFAS 2.0 showed a CFI of 0.887 and RMSEA of 0.083 [17]. There is the criticism that the Hu and Bentler criteria may be too stringent [35]. Our fit indices did not deviate substantially from the Hu and Bentler criteria. We thus retained the one-factor structure of the J-YFAS 2.0. Regarding the reliability of the J-YFAS 2.0, KR-20 was 0.78. This suggests the acceptable internal consistency of the J-YFAS 2.0.

The J-YFAS 2.0-diagnosed FA prevalence was 3.3% in this study. A similar prevalence was observed in other developed countries: Italy (3.4%) and Spain (3.3%) [18,19]. The subjects' characteristics of these three studies bear some resemblance, which might account for the similar prevalence. They were mainly young (aged about 20) and normal-weight people. Like our study, the Italian study collected the subjects from a medical school. About 80% of the subjects were female in both the Spanish and our sample. On the other hand, a web-based survey found a much higher YFAS 2.0-diagnosed FA prevalence, 9.7%, among German-speaking university students with the similar age and BMI [16]. This could imply that not only biological characteristics but also cultural differences are associated with YFAS 2.0-diagnosed FA, although it is possible that the web-based survey received considerable attention from those with FA and obtained their participation. Similar to previous reports in the U.S. [15] and Italy [18], women exhibited a significantly greater YFAS 2.0-diagnosed FA prevalence than men in our study. This suggests a sex difference in YFAS 2.0-diagnosed FA, which should be further investigated in future studies.

The YFAS 2.0-diagnosed FA prevalence was high in overweight, obese, and underweight people in the U.S., Germany, France, Italy, Spain, and Egypt [15–21]. Contrary to these findings, both the current and highest attained BMI did not demonstrate an explicit association with the J-YFAS 2.0-diagnosed FA in the present study. We only found that the subjects with the highest attained BMI of 25.0–29.9 had a greater J-YFAS 2.0-diagnosed FA symptom count than those with the highest attained BMI of 17.0–24.9. However, its effect size was small—the η<sup>2</sup> was only 0.02. One possible reason for the finding might be the small numbers of our subjects with overweight, obesity, and extreme underweight. Current overweight and obesity were declared only by nearly 4% of the subjects. This reflected the fact that Japan has a much lower prevalence of overweight and obesity than other countries where the YFAS 2.0 has been validated [36]. Perhaps, some subjects in our study could have under-reported their BMI [37], although we arranged the categorical response options for BMI to minimize the shame that subjects may feel for self-reporting their actual BMI. Consequently, the low prevalence of overweight and obesity exerted a floor effect, diminishing the association between BMI, especially overweight and obesity, and the J-YFAS 2.0-diagnosed FA. Another possible reason is that the causes to affect BMI are multifactorial and different by region. We did not examine all causes that potentially affected BMI more strongly than FA. For instance, some researchers pointed out that social norms (pressure) might drive the young Japanese women's desire for slimming [38–40]. They could have more impact on BMI than FA among our subjects. Our subjects involved medical, nursing, and medical technology students. They must have a greater knowledge of health, nutrition, and exercise than the normal population, which may have skewed the association between BMI, especially the current BMI, and the J-YFAS 2.0-diagnosed FA. Development of the J-YFAS 2.0 improves the examination of FA in Japan where the prevalence of obesity is much lower than the western countries [36]. This may help elucidate our understanding of the impact of FA on body weight. Although FA was initially applied to understanding obesity, controversy remains over how much FA explains obesity [10,11].

Other variables hypothetically related to the convergent validity of the J-YFAS 2.0 showed significant associations with the J-YFAS 2.0-diagnosed FA as we expected. The TFEQ R-18 uncontrolled eating and emotional eating scores and desire to overeat were associated with the J-YFAS 2.0-diagnosed FA presence, severity, and symptom count in our study, as hypothesized based on the previous studies [15–19]. One study limitation is that we could not assess binge eating itself which was positively and moderately associated with the YFAS-diagnosed FA [41]. However, our findings regarding the desire to overeat and snacking would suggest the relationship between compulsive eating and FA. A desire to overeat forms a part of binge eating. Highly processed sweetened foods, which has been reported to be potentially related to FA [42–44], are often chosen for snacking in Japan [45]. We found that a high K6 score, which implied mood and anxiety disorders, is associated with the presence and higher symptom count of the J-YFAS 2.0-diagnosed FA. Some previous studies showed associations of psychopathological disorders [19] and depressive symptoms [18] with the YFAS 2.0-diagnosed FA. A recent systematic review suggested a positive, moderate association of the YFAS-diagnosed FA with depression and anxiety [41]. Our finding was consistent with them.

Regarding the discriminant validity, we hypothesized that cognitive restraint in eating was a different entity from FA, referring to the idea that the YFAS 2.0 does not simply measure an intention and a failure to restrict food consumption [15,16]. Our findings exhibited an inconsistency in the association of cognitive restraint in eating with the FA presence, severity, and symptom count. In our sample, the association between cognitive restraint and the J-YFAS 2.0-diagnosed FA would not be so strong even if the association existed. Previous findings are also inconsistent regarding the association. It was reported in France and Italy that the YFAS 2.0-diagnosed FA was associated with a high level of cognitive restraint [17,18]. The Italian researchers mentioned the possibility that addictive-like eating and restricting food consumption could coexist in patients with anorexia nervosa [18]. We could not ascertain this possibility in our study since we did not examine whether the subjects suffered from anorexia. Further research would be necessary to examine the role of anorexia in the association between cognitive restraint and FA.

There are several limitations to the interpretation of our findings. First, the current study employed a convenience sample that was dominated by young, under- and normal-weight, female, healthy undergraduate students. For a generalization of the present findings, the J-YFAS 2.0 should be tested for different-age groups, obese individuals, and patients with eating disorders. Second, we were not able to include all kinds of validity and reliability. For instance, we did not address incremental validity and test-retest reliability. Third, we used our original questions to assess the desire to overeat and frequency of snacking. For instance, the Binge Eating Scale (BES) [46] and the Eating Behavior Patterns Questionnaire (EBPQ) [47] are the validated tools to evaluate binge eating and snacking, respectively. We did not use them since they were not translated into and validated in Japanese. This may limit the comparison of our results with the others. Finally, as FA has not yet been recognized in the DSM-5, we could not define the standard of psychiatrist-diagnosed FA.

As mentioned in the introduction, the conceptual construct of FA and the neurobiological changes underpinning it remain controversial [10,11]. Development of the J-YFAS 2.0 would facilitate research on FA in Japan where prevalence of overweight and obesity is much lower than the western countries [36]. This would contribute to specifying the conceptual construct of FA and the neurobiological changes related to FA.

### **5. Conclusions**

The J-YFAS 2.0 had a one-factor structure and adequate convergent validity and reliability, like the YFAS 2.0 in other languages. Further studies are necessary to confirm the discriminant validity of the J-YFAS 2.0.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/11/3/687/s1, Table S1: The Japanese version of the Yale Food Addiction Scale 2.0 (J-YFAS 2.0) (written in Japanese).

**Author Contributions:** M.T.K., A.O., A.N.G., and H.Y. designed the study plan. A.O., A.N.G., A.F., and H.Y. translated the English YFAS 2.0 into Japanese. M.T.K., A.O., A.F., M.M., A.M., Y.L., H.N., and H.Y. collected the data. M.T.K., A.O., and H.Y. analyzed the data and drafted the manuscript. A.N.G., A.F., M.M., A.M., Y.L., and H.N. critically reviewed and approved the manuscript. A.O. and H.Y. obtained the fund.

**Funding:** This research was funded by Fujita Health University. The funder had no role in the design of the study, the collection, analyses, or interpretation of data, writing the manuscript, or the decision to publish the results.

**Acknowledgments:** Data collection and processing were supported by Shunsuke Omura, Nozomi Furukawa, Yukino Onuma, Sho Nakahama, and Keigo Yamada.

**Conflicts of Interest:** The authors declare no conflict of interest. The content does not present the official views of the affiliations to which the authors belong.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Food Addiction Symptoms and Amygdala Response in Fasted and Fed States**

**Kirrilly M. Pursey 1,2, Oren Contreras-Rodriguez 3, Clare E. Collins 1,2, Peter Stanwell <sup>1</sup> and Tracy L. Burrows 1,\***


Received: 5 May 2019; Accepted: 4 June 2019; Published: 6 June 2019

**Abstract:** Few studies have investigated the underlying neural substrates of food addiction (FA) in humans using a recognised assessment tool. In addition, no studies have investigated subregions of the amygdala (basolateral (BLA) and central amygdala), which have been linked to reward-seeking behaviours, susceptibility to weight gain, and promoting appetitive behaviours, in the context of FA. This pilot study aimed to explore the association between FA symptoms and activation in the BLA and central amygdala via functional magnetic resonance imaging (fMRI), in response to visual food cues in fasted and fed states. Females (*n* = 12) aged 18–35 years completed two fMRI scans (fasted and fed) while viewing high-calorie food images and low-calorie food images. Food addiction symptoms were assessed using the Yale Food Addiction Scale. Associations between FA symptoms and activation of the BLA and central amygdala were tested using bilateral masks and small-volume correction procedures in multiple regression models, controlling for BMI. Participants were 24.1 <sup>±</sup> 2.6 years, with mean BMI of 27.4 <sup>±</sup> 5.0 kg/m<sup>2</sup> and FA symptom score of 4.1 <sup>±</sup> 2.2. A significant positive association was identified between FA symptoms and higher activation of the left BLA to high-calorie versus low-calorie foods in the fasted session, but not the fed session. There were no significant associations with the central amygdala in either session. This exploratory study provides pilot data to inform future studies investigating the neural mechanisms underlying FA.

**Keywords:** Food addiction; Yale Food Addiction Scale; functional magnetic resonance imaging; basolateral amygdala

### **1. Introduction**

There is increasing scientific interest in the possible role of "food addiction" (FA) underlying particular patterns of overeating, dietary relapse and weight gain in vulnerable individuals. Neuroimaging techniques, such as functional magnetic resonance imaging (fMRI), have provided insight into this phenomenon in humans. Visual food cues and consumption of palatable, energy-dense foods have been shown to activate reward-related brain circuits in humans in a similar way to substances of abuse [1]. Despite accumulating evidence supporting FA as a phenomenon in preclinical [2] and behavioural research [3], few studies have investigated the potential underlying neural substrates in humans. Many studies have used obesity as a proxy for addictive-like eating in lieu of a recognised assessment tool for FA, such as the Yale Food Addiction Scale (YFAS) [4]. Using obesity as a proxy for

FA could result in inconsistent neuroimaging findings as it is unclear as to the proportion of participants truly affected by addictive-like eating in these samples.

In one neuroimaging study using a recognised FA assessment tool, YFAS symptoms were associated with greater activation in response to a visual milkshake cue in brain areas encoding the reward value of foods and craving (amygdala, anterior cingulate cortex (ACC), medial orbitofrontal cortex (OFC), and the dorsolateral prefrontal cortex (DLPFC)) in young females [5]. In another study, while main effects were found activation in the amygdala, OFC, nucleus accumbens and inferior frontal cortex in response to the taste of sugar-sweetened beverages, no relationships were observed between YFAS symptoms and neural response in male and female adolescents [6]. The divergence in previous studies may be related to the limited range of food cues used, which have often been restricted to sweetened beverages [5,6]. These previous studies have also not studied participants in different motivational states (i.e., fasted and fed), which is important given the differences in responsivity in reward-related networks in these states [7,8].

The amygdala has been implicated in previous FA neuroimaging research [5] and plays a role in regulating the hedonic impact of salient stimuli and coordinating appetitive behaviours, with studies reporting selective sensitivity of the amygdala to food cues in the fasted state [9,10]. The amygdala has also been shown to integrate interoceptive states and sensory cues along the ventral visual stream [9] and has been implicated in drug cue reactivity and drug craving [11,12]. While the amygdala may play an important role in the context of FA, no studies have explored the potential role of distinct subregions of the amygdala. The basolateral amygdala (BLA) is of particular interest as it has been shown to drive external cues to the hypothalamic feeding centres in both humans and rats [13,14], consistent with the role in processing high-level sensory input and stimulus-value associations in humans [15]. In animal studies, the BLA has been implicated in reward-seeking behaviours in response to food-related stimuli [16] and relapse to food seeking [17]. In the satiated state, the BLA has also been associated with eating in the absence of homeostatic needs in rats [18] as well as predicting weight gain susceptibility in males and females [13]. In addition, the central amygdala has been reported to have a role in increasing reward saliency, modulating food consumption and promoting appetitive behaviours in mice [19]. While these previous studies of the BLA and central amygdala were not conducted in FA populations specifically, these findings suggest that there is a need to study the subregions of the amygdala in relation to FA in different motivational states.

This pilot study aimed to explore the association between YFAS assessed FA symptoms and activation in the central and BLA, assessed via functional MRI, in response to visual food cues in fasted and fed states. It was hypothesised that FA symptoms would be associated with greater activation in the BLA in response to high-calorie vs low-calorie foods in both the fasted and fed states, based on the findings of previous neuroimaging research [7,9,10].

### **2. Materials and Methods**

### *2.1. Participants*

Australian females aged 18–35 years were recruited to the current study from an existing pool of participants who had completed an online FA survey, which aimed to determine FA prevalence and associations with dietary intake in Australian adults [20]. At the end of the survey, participants could exit the survey with no further contact or could volunteer to be recontacted for future research. Full details regarding the FA survey are published elsewhere [20]. This study was approved by the University of Newcastle Human Research Ethics Committee (Approval number H-2012-0419) and was conducted in accordance with the ethical standards of the 1964 Helsinki Declaration.

Participants were eligible for the current study if they were female, aged 18–35 years, lived within a one-hour proximity to the imaging facility (Newcastle, NSW, Australia) and elected to be contacted for future research at the end of the online survey. Eligibility for this pilot study was restricted to females only in order to reduce potential inter-person variation in appetite and neural activation to visual food cues associated with sex-related differences [21]. Seventy-seven participants from the original survey volunteered for future research and were recontacted via email to participate in the fMRI component of the study. Of those recontacted, 35 responded that they were interested in participating in the fMRI study and were screened via telephone. Exclusion criteria for the current study included pregnancy, body mass >150 kg due to weight limitations for the MRI scanner, contraindications to MRI, left handedness, pre-existing medical or Axis 1 disorders, disordered eating behaviour, medications affecting appetite, history of substance abuse or head injury with loss of consciousness, allergy to any beverage ingredients, risk of adverse medical events as a result of fasting (e.g., diabetes), or inability to refrain from cigarette smoking. Seventeen participants did not meet the inclusion criteria for the current study, and five participants were not able to be contacted for screening, resulting in a final sample of thirteen participants. Written informed consent was obtained from all participants in the current study.

### *2.2. Procedures*

The study procedure is outlined in Figure 1. Participants attended a single session where they underwent two fMRI scans of the brain, the first in the fasted state and the second in the fed state. Four hours prior to the first scan, participants consumed a standardised pre-fast meal (237 mL Ensure Plus; 1485 kJ/355 kcal). Participants then fasted for four hours, excluding water, to capture the hunger-state experienced when approaching the next meal. Participants were instructed to avoid caffeinated and alcoholic beverages, and smoking for twelve hours prior to the scan. Compliance with the pre-scan protocol was checked via verbal self-report upon arrival at the imaging facility prior to scanning. The session included anthropometric measurements, a demographic survey, hunger and image ratings, and the first of two fMRI scans. Following the first scan, participants drank a second Ensure Plus drink with compliance monitored by the research team. Participants again completed the hunger and image ratings before undertaking a second fMRI scan approximately 45 min following the completion of the second meal replacement beverage. Participants were not able to be scanned at the same time of day due to scheduling at the imaging facility.

**Figure 1.** Flow diagram of the study procedure.

### *2.3. Measures*

Demographics: Demographic data including sex, age, indigenous status, marital status and highest qualification were collected. Current and previous dieting history [22] was also assessed.

Yale Food Addiction Scale: The YFAS is a 25-item questionnaire which assesses addictive-like eating according to the diagnostic criteria for substance dependence. The YFAS has been shown to have adequate psychometric properties [4]. The tool includes two scoring options, a symptom score from 0–7 based on the DSM criteria for substance dependence, as well as a diagnosis of FA if ≥3 symptoms are reported as well as clinical impairment or distress. Cronbach α for the current sample was 0.93.

Anthropometrics: Height was taken using a BSM370 stadiometer, while weight and body composition (% body fat, fat free mass) was assessed using the InBody720 bioelectrical impedance analyser (Biospace, Seoul, Korea) using a standardised protocol. Body mass index (BMI) was subsequently calculated.

Hunger and Image Ratings: Participants completed a set of ratings related to hunger and appetite, as well as rating a subset of images shown to them in the MRI scanner (*n* = 20; *n* = 10 high-calorie, *n* = 10 low-calorie), using a 10 cm visual analogue scale. Participants completed the set of ratings in the fasted state at baseline (prior to Scan 1) and in the fed state (prior to Scan 2). Hunger and

appetite ratings included self-reported (i) hunger, (ii) satiety, (iii) fullness and (iv) prospective food consumption [23]. For the image ratings, participants were asked to rate the (i) appeal of the food, (ii) desire to eat the food, (iii) whether the food increased their appetite and (iv) perceptions of prospective food consumption amount related to each of the food images.

### *2.4. Imaging Paradigm*

Food images were chosen from a standardised database [24] and foods representative of the Australian food environment. Images were matched for brightness, complexity, resolution and size. Two groups of images were created according to nutrient composition, (i) appetising high-calorie foods (e.g., chocolate, chips, pizza, ice cream) and (ii) low-calorie foods (fruits and vegetables). The selected images were informed by the foods commonly reported in the original FA online survey. Food images were piloted on a sample of six university students independent of the study sample (mean age 27 years) regarding recognisability, familiarity and appeal. Those foods with low pilot ratings for recognisability and familiarity were excluded (*n* = 27). A final sample of 150 single food images were selected for the study including *n* = 75 high-calorie and *n* = 75 low-calorie foods. Nutritional composition of the food images can be found in Table S1.

Visual stimuli were presented in a block design format using Presentation software (Neurobehavioral Systems, Inc, Berkley, USA) and NordicNeuroLab sync box (NordicNeuroLab AS, Bergen, Norway), and consisted of two 18 min 20 s runs. Each run consisted of 15 epochs each of (i) low-calorie foods, (ii) high-calorie foods and (iii) fixation cross. Within each 20 s epoch of food images, five images were presented for four seconds each. A 20 s central fixation cross was presented as a control, and each epoch was separated by a four second blank screen. Each run consisted of 440 volumes over the scanning period. Images were presented onto an MRI compatible screen via a projector, and viewed via a mirror attached to the head coil. Standardised instructions to remain still focus on the visual stimuli were provided by the radiographer during the scan.

### *2.5. Imaging Data Collection*

Structural and task-related functional scans were acquired using a Siemens 3-tesla MRI (Siemens AG, Germany). Structural images were acquired with a three-dimension (3D) magnetisation-prepared rapid gradient-echo (MP-RAGE) sequence with the following parameters: TE = 3.5 ms, TR = 2 s, 7◦ excitation flip angle, 160 slices with 1mm isotropic resolution. Task-related functional MR images were acquired using a T2\*-weighted gradient-echo echo planar imaging (EPI) pulse sequence. The parameters were TE = 24 ms, TR = 2.5 s, 90◦ excitation flip angle, 47 axial slices 3 mm thick with a 0 mm interslice gap, in-plane resolution of 3 × 3 mm, covering from vertex to cerebellum. Axial slices were angled to the anterior cranial floor for to limit distortions that may impact on imaging of the orbito-frontal region.

### *2.6. Statistical Analysis*

Participant characteristics, hunger and image ratings were analysed descriptively using Stata13. Differences in hunger and image ratings between the first and second scan were assessed using paired t-tests. Task-related functional MR images were processed and analysed using MATLAB version R2017a (The MathWorks Inc, Natick, Mass, USA) and Statistical Parametric Software (SPM12; The Welcome Department of Imaging Neuroscience, London, UK). Preprocessing steps involved motion correction, spatial normalization and smoothing using a Gaussian filter (FWHM 8 mm). The realigned functional sequences were coregistered to each participant's anatomical scan, which had been previously coregistered and normalized to the SPM-T1 template. Normalization parameters were then applied to the coregistered functional images, which were then resliced to a 2 mm isotropic resolution in Montreal Neurological Institute (MNI) space.

First-level (single-subject) SPM contrasts images were estimated for each of the fasted and fed scans for the following task main effects of interest: high-calorie > low-calorie foods, and low-calorie > high-calorie foods. Changes in brain activation between sessions were computed by the creation of the following contrasts: (Fasted scan (high-calorie > low-calorie foods) – Fed scan (high-calorie > low-calorie foods)) and (Fed scan (high-calorie > low-calorie foods) – Fasted scan (high-calorie > low-calorie foods)). The same contrasts were created to explore the between session effects on the activation of low-calorie > high calorie foods. Three regressors were used in these analyses to model conditions separately corresponding to high-calorie and low-calorie foods, and the fixation cross. Furthermore, to correct for subtle in-scanner movements from volume-to-volume scans, we identified the outlier scans present in the realigned task-related functional scans as determined using the CONN toolbox (Whitfield-Gabrieli and Nieto-Castanon, 2012). For each participant, the actual removal of outlier scans was completed by entering the subject-specific variables identifying the outlier scans (i.e., one regressor per outlier) in the first-level models as covariates of no interest, so these outlier scans were removed from these and subsequent analyses. We excluded one subject that had 19.8% of outlier scans. A hemodynamic delay of 4 s and a high-pass filter (1/120 Hz) were considered. The resulting first-level contrasts images were then carried forward to subsequent second-level random-effects (group) analyses. One-sample t-tests were used to assess the main task effects in the fasted and fed sessions. Multiple regression models were used to assess associations between FA symptoms and whole-brain activation under the main task effects in the fasted and fed sessions, and the change between sessions. These analyses were controlled for the subject-specific number of outlier scans, BMI, hunger and dieting when the association with FA symptoms was explored. Results were considered significant when surviving a *p* > 0.001 and a determined cluster-extent using Monte Carlo simulations implemented in the AlphaSim thresholding approach [25] (94 voxels and 135 voxels for the fasted and fed sessions, respectively incorporating a 2×2×2mm grey matter mask of 128,190 voxels). However, we specifically tested for associations between FA symptoms and the activation of the basolateral and central amygdala using bilateral masks and small-volume correction procedures in the multiple regression models. To that end, we defined bilateral basolateral and centromedial amygdala masks comprising 3.5 mm radial spheres cantered at left basolateral (x = −26, y = −5, z = −23) and right BLA (x = 29, y = −3, z = −23), and the left centromedial amygdala (x = −19, y = −5, z = −15) and right centromedial amygdala (x = 23, y = −5, z = −13) following previous research [26,27].

### **3. Results**

Participant characteristics can be found in Table 1. The mean YFAS symptom score for the group was 4.1 ± 2.2 (range 1–7), with six participants classified as having a YFAS FA diagnosis. Hunger and image ratings are presented in Table 2. Hunger ratings were not significantly different between the fasted and fed scans (*p* = 0.13); however, ratings of fullness and satisfaction increased, and prospective food consumption decreased (all *p* < 0.05). Image ratings were significantly reduced from the fasted to fed scans (*p* < 0.05). FA symptoms were not associated with hunger or image ratings (*p* > 0.05)



Data is presented as mean ± SD unless otherwise specified.


**Table 2.** Hunger, appetite and image ratings in the fasted and fed states.

Ratings completed using a 10 cm visual analogue scale. Results are presented as mean ± SD.

### *Brain Activation and Association with FA Symptoms*

Visual cues of high-calorie vs low-calorie foods activated a cluster in the right amygdala, the left hippocampus, the fusiform gyrus, and in bilateral occipital cortex in the fasted session, and the left amygdala in the fed session (Table 3, Figure 2). No significant activations were recorded in the contrast low-calorie vs high-calorie foods in the fasted and fed sessions.

**Table 3.** Brain regions showing significant activation in response to the sight of high-calorie vs low-calorie foods in the fasted and fed sessions.


Coordinates are given in Montreal Neurological (MNI) Atlas space. The results for the fasted session surpassed a *p* < 0.001 and a cluster size (CS) of 94 voxels and 135 voxels, for the fasted and fed session, respectively. The t-values refer to the comparison of the activation of each of the listed brain regions to the high-calorie vs low-calorie food images.

**Figure 2.** Increased brain activation to "high-calorie vs low-calorie" foods during the fasted and the fed sessions. Right side of the figure corresponds to the right hemisphere in the coronal views. In sagittal views, the upper figure corresponds to the right hemisphere, whereas the lower figure corresponds to the left hemisphere. Colour bar displays t-values for the comparison of activation in response to high-calorie vs low-calorie food cues.

No significant associations with FA symptoms were found using whole-brain corrections, but a significant association between the FA symptoms and higher activation of the left BLA to high-calorie vs low-calorie foods in the fasted session (x = −26, y = −4, z = −26, t = 3.45, 11 voxels, *p*SVC-FWE<0.05 = 0.042) were identified using small-volume correction procedures. Importantly, this association was found regardless of the body mass index of the participants, and it also remained statistically significant after controlling for hunger reported by the participants during the fasted session and current dieting status. No significant associations were identified between FA symptoms and the activation of the central amygdala in the fasted session, and the activation of the basolateral and central amygdala in the fed session (all *p*SVC-FWE >0.05). The change in the magnitude of activation from the fasted to the fed sessions of the left BLA showed a significant association with FA symptoms, controlling for BMI (x = −26, y = −4, z = −26, t = 3.77, 8 voxels, *p*SVC-FWE<0.05 = 0.027) (Figure 3).

**Figure 3.** Association between the activation of the left basolateral amygdala and food addiction traits, (**a**) location of the basolateral (green) and central (blue) amygdala seeds, used as a mask for small-volume corrections; (**b**) left basolateral amygdala significantly associated with food addiction traits during the fasted session (*p*SVC-FWE <sup>&</sup>lt;0.05 = 0.042); (**c**) scatter plot represents the correlation between the change in the activation of the left basolateral amygdala from the fasted to the fed sessions (y-axis) and food addiction traits (x-axis) (x = −26, y = −4, z = −26, t = 3.77, *p*SVC-FWE<0.05 = 0.027).

### **4. Discussion**

This is the first study to use fMRI to investigate activation in distinct subregions of the amygdala in response to visual food cues in relation to FA symptoms in both fasting and fed states. In line with the hypothesis, activation in the BLA in response to high-calorie foods versus low-calorie foods was found to be associated with higher FA symptoms in the fasted state, which is consistent with previous research [7,9,10]. Furthermore, participants with greater FA symptoms showed higher activation in the BLA in the fasted than in the fed scans to the sight of high-calorie vs low-calorie foods. One possible mechanism consistent with this is that the amygdala boosts activation in the ventral stream [9,28], increasing food salience in the hunger state. Individual differences in BLA response in the hunger state to appetizing, high-calorie foods may therefore be associated with susceptibility to overeating.

The findings of the current study suggest that those with higher FA symptoms may be vulnerable to environmental food cues of energy-dense, nutrient-poor, high-calorie food when hungry compared to those with lower FA symptoms. This may lead those vulnerable individuals to greater desire to eat, hence leading to food seeking and consumption, which is consistent with previous research [16,17]. It is possible that FA may play a mediating role in non-homeostatic eating, dietary relapse and other longer-term issues such as weight gain. This may be an important consideration in future interventions targeting FA, suggesting a regular meal pattern that uses strategies to avoid long periods of fasting may be warranted. This may also suggest that further consideration to reducing environmental food cues (e.g., food advertisements) in treatment approaches may be warranted for those vulnerable to addictive-like eating.

The second hypothesis was not supported, with no significant associations found between FA symptoms and BLA activation in the fed state. This is divergent previous work investigating the role of the amygdala in eating in the absence of hunger and weight gain susceptibility [13]. However, the study by Sun and colleagues [13] did not recruit a sample to investigated FA, in contrast to the current study. These differences may also be due to the study samples (i.e., inclusion of males in the previous study) and use of different food cues (i.e., taste vs sight). Alternatively, this may be related to the standardised meal not significantly reducing self-reported hunger from baseline in the current study, although fullness was increased, and prospective food consumption was decreased. Of note, the standardised meal would have contributed a relatively small proportion of total daily energy intake (TEE) for participants (<30%). Future studies should therefore consider the use of a standardised meal with greater contribution to %TEE.

The strengths of the current study include the use of a recognised assessment tool for FA and investigation of participants in fasted and fed states. Food images were informed by previous research and selected from a standardised database [24], which was found to be a limitation of previous studies [21]. This current exploratory study is limited by the small sample size and inclusion of females only. Additionally, participants could not be scanned at the same time of day, however, a standardised meal replacement was provided to reduce inter-individual variation. Self-reported hunger was not significantly different between the fasted and fed states, however, other measures of appetite were significantly reduced. Future studies should consider standardised meals with greater calorie content, to ensure hunger is significantly different between the two conditions. As the scans were not conducted in a counterbalanced order, it is possible that the lack of associations may be attributed to a habituation effect. A further limitation is the lack of control condition in the cue reactivity task, and the addition of a control condition should be considered in future studies. Although participants were asked about disordered eating behaviours during the screening process, a clinical interview to identify eating disorder was not included in the baseline assessment. Future studies should use a clinical eating disorder assessment to better control for underlying disordered eating. This study used the original YFAS tool as the updated YFAS 2.0 tool had not been released at the time of the study. Future imaging studies should consider the use of the YFAS 2.0, which aligns with the DSM-5 criteria [29]. The current study also investigated associations between neural activation and FA symptoms, however, future

studies may consider analysing according to the dichotomous YFAS diagnosis and in those with higher FA symptoms to better understand addictive-like eating.

### **5. Conclusions**

This study demonstrates that activation of the BLA, which has been linked to reward-seeking behaviours and susceptibility to weight gain, was associated with FA symptoms in the fasted state. This study provides pilot data to inform future studies with larger sample sizes investigating the neural mechanisms associated with FA.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/11/6/1285/s1, Table S1: Nutritional composition of food image groups per 100 g.

**Author Contributions:** Conceptualization, K.M.P., T.L.B., P.S. and C.E.C.; Analysis, O.C.-R., K.M.P.; Writing—Original Draft Preparation, K.M.P.; Writing—Review and Editing, T.L.B., O.C.-R., C.E.C., P.S.; Funding Acquisition, T.L.B. All authors have contributed to the development of the manuscript and have approved the final version.

**Funding:** This study was funded by a University of Newcastle, Faculty of Health and Medicine Pilot Grant obtained by T.L.B.

**Acknowledgments:** K.M.P. is supported by a Hunter Medical Research Institute Greaves Family Early Career Support Grant. O.C.-R. is funded by the postdoctoral contract "PERIS" (SLT006/17/00236) from the Catalan Government. T.L.B. is supported by University of Newcastle Brawn Research Fellowship.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Increasing Chocolate's Sugar Content Enhances Its Psychoactive Effects and Intake**

### **Shanon L. Casperson 1,\*, Lisa Lanza 2, Eram Albajri <sup>2</sup> and Jennifer A. Nasser 2,\***


Received: 29 January 2019; Accepted: 7 March 2019; Published: 12 March 2019

**Abstract:** Chocolate elicits unique brain activity compared to other foods, activating similar brain regions and neurobiological substrates with potentially similar psychoactive effects as substances of abuse. We sought to determine the relationship between chocolate with varying combinations of its main constituents (sugar, cocoa, and fat) and its psychoactive effects. Participants consumed 5 g of a commercially available chocolate with increasing amounts of sugar (90% cocoa, 85% cocoa, 70% cocoa, and milk chocolates). After each chocolate sample, participants completed the Psychoactive Effects Questionnaire (PEQ). The PEQ consists of questions taken from the Morphine-Benzedrine Group (MBG), Morphine (M,) and Excitement (E) subscales of the Addiction Research Center Inventory. After all testing procedures, participants completed the Binge Eating Scale (BES) while left alone and allowed to eat as much as they wanted of each of the different chocolates. We found a measurable psychoactive dose–effect relationship with each incremental increase in the chocolate's sugar content. The total number of positive responses and the number of positive responses on the E subscale began increasing after tasting the 90% cocoa chocolate, whereas the number of positive responses on the MBG and M subscales began increasing after tasting the 85% cocoa chocolate sample. We did not find a correlation between BES scores and the total amount of chocolate consumed or self-reported scores on the PEQ. These results suggest that each incremental increase in chocolate's sugar content enhances its psychoactive effects. These results extend our understanding of chocolate's appeal and unique ability to prompt an addictive-like eating response.

**Keywords:** chocolate; Addiction Research Center Inventory; sugar; craving; addictive-like eating; eating behavior

### **1. Introduction**

Chocolate holds a special status in our society. Indeed, it is one of the most loved and craved, but problematic, foods [1]. Consuming chocolate evokes pleasant feelings, reduces tension, and improves mood [2,3]. Furthermore, chocolate elicits unique brain activity compared to other high-sugar and high-fat foods, recruiting brain structures that respond to craving-inducing stimuli, and is therefore more likely to provoke an addictive-like eating response [4]. The particular combination of cocoa, sugar, and fat in chocolate may play important, yet distinct, roles in chocolate's unique ability to elicit an addictive-like eating response. Smit et al. [5] demonstrated a role of the main psychopharmacological active constituents of cocoa in producing psychostimulant effects but determined that other attributes, such as sweetness and texture, may be more important. In a prior study, we observed effects of the percent cocoa and sugar contents on "desire to consume more chocolate", while fat content trended towards significance for this effect [6]. Defeliceantonio et al. [7] demonstrated a supra-additive effect

of combining sugar and fat on food reward in humans, while others have demonstrated that the sugar component (in a sugar/fat combination) in particular is more effective at activating reward [8,9] and gustatory brain circuits [9]. Additionally, we have demonstrated that the highly reinforcing properties of sugar are difficult to overcome [10,11]. Taken together, chocolate's desirability appears to arise from the synergistic relationship among its components.

Fat and sugar are known to stimulate both the dopamine and the opioid neurotransmitter systems to regulate a food's rewarding potential [12–14]. The dopamine neurotransmitter system stimulates 'wanting', or the motivation to consume the food [12,14], whereas the opioid neurotransmitter system modulates the consumption of a desired food by amplifying 'liking', or the hedonic value, of the desired food [12,13]. Thus, the sight, smell, and taste of a highly palatable food such as chocolate work together to trigger motivational and hedonic reward mechanisms that result in the pursuit and consumption of the desired food. Subjective dopaminergic and opioidergic effects of food consumption can be differentiated using the Addiction Research Center Inventory (ACRI) [15]. We previously utilized the ARCI, specifically the Morphine-Benzedrine Group (MBG), Morphine (M), and Excitement (E) subscales, in a between-group design (groups identified by percent cocoa of the sample tasted), to provide indices of the psychoactive effects of chocolate that are associated with addictive-like eating [6]. We showed that simply tasting chocolate increases the number of positive responses on the MBG subscale of the ARCI, consistent with responses obtained on the MBG subscale after dopaminergic–opioidergic drug administration [16]. These results help explain chocolate's high reinforcing value, as foods that elevate feelings of euphoria have a greater reinforcing potential [17]. However, different individuals may specifically crave a particular combination of chocolate's main components (dark vs milk chocolate).

To continue our study of the psychoactive effects of chocolates varying in percent cocoa and sugar and fat content, we repeated our 2011 study using a within-subject design. Because some have postulated that "food addiction" is more of an "eating behavior addiction" similar to binge eating [18,19], we added the Binge Eating Scale (BES) [20] to our data collection. We hypothesized that self-reported scores on the MBG and E subscales of the ARCI would correlate with the particular combination of the main components (cocoa, sugar, and fat) of the chocolate and the amount of each of the different chocolates consumed after the tasting session. Consistent with our 2011 study, we expected no correlation between self-reported scores on the M subscale of the ARCI and the particular combination of chocolate's main components. In addition, we hypothesized that the amount of chocolate consumed after the tasting session would positively correlate with the increasing sugar content and decreasing cocoa and fat content of the chocolate. Furthermore, we hypothesized that BES scores would positively correlate with self-reported scores on the ARCI questionnaire and chocolate consumption.

### **2. Materials and Methods**

### *2.1. Participants*

Healthy adults (Table 1) were recruited from the greater Grand Forks, ND, and Philadelphia, PA, areas. Our participant population consisted of 57% non-overweight, 30% overweight, and 13% class 1 obese participants. Screening for study eligibility included height, weight, and a medical health history questionnaire. Exclusion criteria included: presence of food and non-food allergies; current status as a dieter; current or past metabolic illnesses (diabetes, renal failure, thyroid illness, hypertension); psychiatric, neurological, or eating disorders (schizophrenia, depression, Parkinson's Disease, Huntington's Disease, cerebral palsy, stroke, epilepsy, anorexia nervosa, or bulimia nervosa); taking prescription medications except for oral contraceptives or antihyperlipidemia agents. The study was approved by both the University of North Dakota and the Drexel University Institutional Review Boards and registered on clinicaltrials.gov as NCT03364413. Informed written consent was obtained from all participants prior to any study-related procedures.


**Table 1.** Participant characteristics. BMI, body mass index.

Values are means ± SD.

### *2.2. Experimental Procedures*

Participants reported to the Center at which they were recruited 3–4 h postprandial to determine the psychoactive effect of consuming chocolate with varying amounts of fat, sugar, and cocoa (Table 2). Participants were instructed to have a light meal (e.g., sandwich with side salad or cup of soup) and to then refrain from eating or drinking (except water) before reporting for their study visit. Upon arrival, participants rated their appetite using 10 cm visual analog scales and completed a baseline Psychophysical Effects Questionnaire (PEQ; based on the ARCI) [6]. Participants were then presented with 5 g of commercially available chocolate varying in cocoa, sugar, and fat concentrations (Table 2). Chocolates were tested in order from least to most amount of sugar, and the PEQ was completed immediately after each chocolate tasting. Participants completed the Binge Eating Scale (BES) after the tasting session. Participants at the Grand Forks site were then left alone and allowed to eat as much as they wanted of each of the different chocolates.

**Table 2.** Characteristics of each 5 g chocolate sample.


### *2.3. Questionnaires*

Subjective homeostatic and hedonic hunger ratings were assessed using 100 mm visual analogue scales. Questions asked included: 1. How hungry do you feel; 2. How strong is your desire to eat; 3. How full do you feel; 4. How satisfied do you feel; 5. How much do you think you could eat right now; 6. Would you like to eat something sweet; and 7. Would you like to eat something fatty.

The PEQ is composed of 30 questions taken from the ARCI Morphine-Benzedrine Group (MBG), Morphine (M), and Excitement (E) subscales that assess subjective dopaminergic and opioidergic effects of psychoactive drugs [15]. The MBG subscale contains questions that center on feelings of well-being and euphoria, which correlate with the activation of both the dopaminergic and the opioidergic neurotransmitter systems. The M subscale focuses on attitude and physical sensations, and the E subscale relates to physical and psychological feelings of excitement, both of which correlate with activation of the dopaminergic neurotransmitter system [16]. Participants completed the PEQ before and after tasting each of the different chocolates.

The BES is composed of 16 questions used to assess the presence of binge eating behavior [20]. The BES contains questions that assess both behavioral manifestations of binge eating and of the feelings that either cue or follow a binge episode. Participants completed the BES after completing all testing procedures.

### *2.4. Anthropometric Measurements*

Height, measured to the nearest 0.1 cm using a stadiometer (SECA Model 214, Hamburg, Germany), and body weight, measured using a calibrated digital scale to the nearest 0.1 kg (Fairbanks model 50735; Kansas City, MO, USA), were obtained at the end of the testing session.

### *2.5. Statistical Analysis*

General linear models were used to compare general participant characteristics (i.e., sex, age, body mass index (BMI), hunger, testing site) and ARCI subscale scores of each chocolate type (defined by the percent cocoa in the chocolate sample tasted) and the amount of chocolate consumed. Tukey's post hoc analysis was used to determine differences. The threshold of significance was set at alpha = 0.05. JMP V14 (SAS Institute, Inc., Cary, NC, USA) was used for all analyses.

### **3. Results**

### *3.1. Subjective Appetite Responses*

On a scale from 0 (not at all) to 100 (very much), participants' rating of hunger was 58 ± 21 and of the desire to eat was 66 ± 28, while fullness was 25 ± 24 and feeling of being satisfied was 31 ± 21. There was no main effect of hunger or fullness on the total number of positive responses and on the number of positive responses on the MBG and M subscales; however, there was a main effect of hunger (F(1,15) = 4.73, *p* = 0.046) and fullness (F(1,15) = 8.06, *p* = 0.012) on self-reported scores on the E subscale of the PEQ.

Participants also rated their desire to eat foods with a specific taste profile on a scale from 0 (not at all) to 100 (very much). Participants' rating of the desire to eat something sweet was 79 ± 14 and that of the desire to eat something fatty was 48 ± 27. There was no main effect of the desire to eat something sweet or fatty on the total number of positive responses or on the number of positive responses on the MBG, M, and E subscales.

### *3.2. Psychophysical Effects Questionnaire*

There was no main effect of testing site, sex, age, BMI, or BES on the total number of positive responses or on the number of positive responses on the MBG, M, and E subscales. Because of the direct correlation between the percent cocoa and sugar contents and the simultaneous changes of the fat and cocoa contents, we were not able to assess the direct effects of the individual chocolate components in this within-subject study design.

There was a significant main effect of chocolate type on the total number of positive responses (F(4,106) = 30.10, *p* < 0.0001) and on the number of positive responses on the MBG (F(4,106) = 15.83, *p* < 0.0001), M (F(4,106) = 10.34, *p* < 0.0001) and E (F(4,106) = 16.57, *p* < 0.0001) subscales. Tukey's post hoc analysis revealed slight differences in the effect of chocolate type on the total number of positive responses and on the number of positive responses on MBG, M, and E subscales.

The total number of positive responses (Figure 1) significantly increased after the consumption of all the different types of chocolate, with milk chocolate eliciting the greatest increase.

The number of positive responses on the MBG subscale (Figure 2) significantly increased after the consumption of the 85% cocoa chocolate and continued to increase in a dose-dependent manner in response to tasting each of the other chocolates. The number of positive responses on the MBG subscale after the consumption of the milk chocolate was significantly greater than after consumption of any of the other chocolates.

### **3 V\FKRSK \VLFDO( IIHF WV4 X HV WLRQQ DLUH**

**Figure 1.** Total Psychophysical Effects Questionnaire (PEQ) scores after the consumption of chocolates differing in cocoa, sugar, and fat content. Self-reported scores on the PEQ questionnaire are presented as box and whiskers plots with the line representing the median, the box representing the 25th to 75th percentiles, and the whiskers representing the minimum to maximum values. Least-squares means are Baseline: 8.26 ± 1.16 (SE), 95% CI [5.93, 10.58]; 90% cocoa: 10.87 ± 1.10 (SE), 95% CI [8.65, 13.08]; 85% cocoa: 12.37 ± 1.10 (SE), 95% CI [10.15, 14.58]; 70% cocoa: 13.70 ± 1.10 (SE), 95% CI [11.49, 15.91]; Milk: 15.87 ± 1.10 (SE), 95% CI [13.65, 18.08]. Levels not connected by the same letter are significantly different.

### **0 R US KLQ H % HQ ]HG ULQ H\* UR X S6 X E V FDOH**

**Figure 2.** Morphine-Benzedrine Group (MBG) subscale scores after the consumption of chocolates differing in cocoa, sugar, and fat content. Self-reported scores on the MBG subscale are presented as box and whiskers plots with the line representing the median, the box representing the 25th to 75th percentiles, and the whiskers representing the minimum to maximum values. Least-squares means are Baseline: 4.28 ± 0.71 (SE), 95% CI [2.86, 5.69]; 90% cocoa: 5.37 ± 0.66 (SE), 95% CI [4.03, 6.70]; 85% cocoa: 6.10 ± 0.66 (SE), 95% CI [4.77, 7.43]; 70% cocoa: 6.63 ± 0.66 (SE), 95% CI [5.30, 7.97]; Milk: 7.97 ± 0.66 (SE), 95% CI [6.63, 9.30]. Levels not connected by the same letter are significantly different.

The number of positive responses on the M subscale (Figure 3) did not increase after the consumption of the 90% cocoa chocolate. A significant increase above baseline was not observed until participants consumed the 85% cocoa chocolate, with no further increases for each sequential chocolate consumption.

**Figure 3.** Morphine (M) subscale scores after the consumption of chocolates differing in cocoa, sugar, and fat content. Self-reported scores on the M subscale are presented as box and whiskers plots with the line representing the median, the box representing the 25th to 75th percentiles, and the whiskers representing the minimum to maximum values. Least-squares means are Baseline: 0.78 ± 0.20 (SE), 95% CI [0.37, 1.19]; 90% cocoa: 1.00 ± 0.19 (SE), 95% CI [0.62, 1.38]; 85% cocoa: 1.47 ± 0.19 (SE), 95% CI [1.09, 1.84]; 70% cocoa: 1.57 ± 0.19 (SE), 95% CI [1.19, 1.94]; Milk: 1.57 ± 0.19 (SE), 95% CI [1.19, 1.94]. Levels not connected by the same letter are significantly different.

The number of positive responses on the E subscale (Figure 4) significantly increased after the consumption of the 90% cocoa chocolate. As with the MBG subscale, the number of positive responses on the E subscale continued to increase in a dose-dependent manner in response to each incremental increase in the chocolate's sugar content and decrease in the percent cocoa and fat content.

**( [FLWHP HQ W6 XE VFDOH**

**Figure 4.** Excitement (E) subscale scores after the consumption of chocolates differing in cocoa, sugar, and fat content. Self-reported scores on the E subscale are presented as box and whiskers plots with the line representing the median, the box representing the 25th to 75th percentiles, and the whiskers representing the minimum to maximum values. Least-squares means are Baseline: 3.51 ± 0.47 (SE), 95% CI [2.56, 4.46]; 90% cocoa: 4.57 ± 0.44 (SE), 95% CI [3.68, 5.45]; 85% cocoa: 4.80 ± 0.44 (SE), 95% CI [3.92, 5.68]; 70% cocoa: 5.53 ± 0.44 (SE), 95% CI [4.65, 6.42]; Milk: 6.27 ± 0.44 (SE), 95% CI [5.38, 7.15]. Levels not connected by the same letter are significantly different.

### *3.3. Chocolate Consumption*

There was a main effect of chocolate type on the amount of chocolate consumed (*p* < 0.0001). Overall, participants consumed significantly more milk chocolate (21 g ± 25 (SD)) than any other chocolate type (90% cocoa: 3 g ± 9; 85% cocoa: 2 g ± 4; 70% cocoa: 6 g ± 10 (SD)). There was no correlation between the amount of chocolate consumed and the total number of positive responses or the number of positive responses on the MBG, M, or E subscales. There was a main effect of hunger

(F(1,18) = 11.74, *p* = 0.003) on the total amount of chocolate consumed. Post hoc analysis revealed a main effect of hunger (F(1,17) = 8.64, *p* = 0.009) on the amount of milk chocolate consumed only. There was no main effect of BES score on the total amount of chocolate consumed.

### **4. Discussion**

The current study aimed to extend our previous findings on the psychoactive effects of consuming chocolate varying in cocoa, sugar, and fat concentrations using a validated ARCI "drug effects" questionnaire in a within-subject design. The questions used simultaneously reflect alterations in motivation, mood, sensation, and perception, and, therefore, provide insight into the interrelation of these variables and chocolate consumption [16]. Our data indicate a measurable psychoactive dose–effect relationship with each incremental increase in the chocolate's sugar content and decrease in the percent cocoa and fat contents. Overall, there were an inverse dose–effect relationship with cocoa concentration and fat content and a positive dose–effect relationship with the sugar content of a chocolate. In addition, our data indicate that the dose–effect relationship of the different chocolates was slightly different for each ARCI subscale. Thus, the present study is the first to demonstrate a dose-dependent relationship between self-reported scores on the PEQ and chocolate consumption.

Increased feelings of well-being, euphoria, and physical and psychological feelings of excitement after chocolate consumption are consistent with chocolate's ability to modulate both the opioid and the dopamine neurotransmitter systems. Both human and animal research has demonstrated the reinforcing potential and comforting and mood-ameliorating effects of chocolate [2,3,21–23]. Contrary to our hypothesis, we did not find an association between self-reported scores on the MBG subscale and chocolate consumption. These results are also inconsistent with our previous study in which self-reported scores on the MBG subscale were associated with an increased desire to eat more chocolate [6]. A possible explanation for these findings is that subjective appetite measurements were obtained prior to the tasting session rather than after; however, participants were allowed to consume as much of the different chocolates as they wanted after the tasting session. On average, participants consumed 8 ± 15 g more chocolate than what was provided to them for the tasting session. Another possible explanation for this difference is that the smell and tasting of four different chocolates in the same session as opposed to only one type of chocolate could have satiated the desire to eat more chocolate. Massolt et al. [24] demonstrated that not only eating but simply smelling chocolate (85% cocoa) suppresses appetite. Additionally, Sørensen and Astrup [25] demonstrated that dark chocolate (70% cocoa) increases satiety and decreases the desire to eat something sweet more than milk chocolate. In the current study, participants consumed 15 g of dark chocolate before tasting the milk chocolate, and this could have decreased their appetite for more chocolate.

The sugar content, which plays a key role in chocolate's pleasurable taste and texture, is important in determining chocolate's reinforcing potential [26]. Research has shown that the added sugar component of a food is greatly associated with its reinforcing value [8,9,27]. We have also shown that the highly reinforcing properties of sugar are difficult to overcome [10,11]. The activation of sweet taste receptors, the speed at which the information about a food is delivered from the chemosensory and somatosensory neurons in the mouth to the brain, and the magnitude of the activation of the food reward system govern the reinforcing and rewarding effect of sugar [28,29]. Low et al. [30] recently reported that the average concentration at which sugar can be differentiated from water is 9 mass percent (m%); however, interindividual variability is large (reported range from 2 m% to 32 m%). The sugar content of the chocolates provided in this study were 8 m%, 13 m%, 30 m%, and 48 m% for the 90% cocoa, 85% cocoa, 70% cocoa, and milk chocolates, respectively. Therefore, our finding that consuming the 85% cocoa, 70% cocoa, and milk chocolates resulted in significant increases in self-reported scores on the PEQ, and each subscale, above baseline is supported by the fact that all of these chocolates were above the 9 m% detection threshold.

In agreement with our prior study [6], consuming milk chocolate elicited a greater increase, compared to all the other chocolates, in the total number of positive responses as well as a greater

increase in positive responses on the MBG subscale. The sugar content of the milk chocolate, which was equivalent to the upper sucrose detection threshold reported by Low et al. [30], may explain these results, as well as the mass appeal of milk chocolate. Indeed, participants in the current study consumed an average of 21 ± 25 g of milk chocolate compared to 6 ± 10 g of the 70% cocoa and 2 ± 4 g of the 85% cocoa chocolates. Taken together, these results agree substantively with other indicators of reinforcement essential for motivational behavior [15,31] and support the "addictive-like" behavior response individuals can experience with chocolate, in particular milk chocolate, consumption.

The finding that consuming chocolate containing 90% cocoa increased the total number of positive responses and the number of positive responses on the E subscale is of interest, given its reported "bitter" taste. The bioactive compounds found in dark chocolate may explain these positive results. In two double-blind, placebo control studies, Smit et al. [5] demonstrated that the amount of methylxanthines (theobromine and caffeine) found in dark chocolate can produce psychostimulant effects. They found that the consumption of both encapsulated cocoa powder and a "typical portion" of dark chocolate increases energetic arousal ("energetic", "alert", etc. versus "tired", "sluggish"). This is consistent with our results demonstrating chocolate's ability to increase physical and psychological feelings of excitement. Interestingly, when examined separately, the amount of theobromine found in a "typical portion" of chocolate (200–300 mg) does not appear to play a psychopharmacological role [32], whereas the amount of caffeine (25–35 mg) is well above the previously reported stimulatory threshold (12.5 mg [33]). The amount of caffeine consumed from the 90% cocoa chocolate sample in the current study was 9 mg, well below the stimulatory threshold. It may be that the theobromine found in chocolate enhances the stimulatory effect of caffeine. Thus, it is plausible that the combination of theobromine and caffeine found in chocolate provides an additive psychostimulant effect. While the methylxanthines in chocolate provide psychostimulant effects, Smit et al. [5], as well as Michener and Rozin [26], reported no difference in chocolate cravings between the ingestion of cocoa powder versus placebo, whereas consuming chocolate, including white chocolate (albeit to a lesser extent), immediately reduced chocolate cravings. As mentioned above, another explanation for this finding is that the sugar content of the 90% cocoa chocolate (8 m%) is well above the lowest and close to the average sucrose detection threshold previously reported [30]. Taken together, these results confirm that the combination of the main chocolate constituents is necessary to produce its psychoactive effects.

As expected, more milk chocolate was consumed than any of the other chocolates, indicating its reinforcing potential. This is consistent with the greater number of positive responses on the PEQ. However, contrary to our hypothesis, we did not find an association between the total amount of chocolate consumed or the total number of positive responses on the PEQ and BES scores. A potential explanation for this finding is all of our study participants had a score less than or equal to 17 and thus would be classified as non-binge eaters [20]. The narrow range in BES scores (2–17) of our participants may not have provided the heterogeneity needed to reliably determine a correlation between BES scores and chocolate consumption or self-reported PEQ scores. Further studies are needed to determine the potential correlation between BES scores and chocolate consumption or self-reported PEQ scores. Additionally, the low BES scores may also explain why we did not find an association between self-reported scores on the MBG subscale and chocolate consumption.

This study is not without limitations. The direct correlation between the percent cocoa and sugar content and the simultaneous changes of the fat content in the chocolate samples did not allow us to assess the independent psychoactive effects of each component. However, using commercially available chocolate provides a "real-world" value to our observations. Furthermore, tasting the chocolate samples sequentially in a relatively short amount of time may have produced an additive effect on our outcomes, such as habituation or sensory-specific satiety [34]. For this study, each participant consumed a total of 20 g of chocolate (5 g/sample) compared to the 12.5 g provided in our previous study [6] with similar PEQ results. Therefore, it does not appear that our within-subject study design had a significant impact on our results. Lastly, the participants in this study tended to be healthy-weight individuals (57% of the participants), thus, the results may not reflect the response of

overweight and obese individuals. Although we did not find a significant effect of BMI on self-reported scores on the PEQ or any of the subscales, research has shown that food reinforcement can vary significantly between lean, overweight, and obese individual [35–37] and individuals with an eating disorder [38,39]. Further research is needed to determine if there exist differential psychoactive effects of chocolate consumption between lean, overweight, and obese populations and individuals with an eating disorder.

Our prior work suggests that obese individuals who engage in binge eating may do so because of an inherent reward deficiency, resulting in overeating high-sugar foods. We previously found an increase in food reinforcement after the consumption of a liquid chocolate-based preload, indicating a "sensitization" response to foods that increase the dopaminergic response [39]. Davis et al. [38] found that obese individuals diagnosed with binge eating disorder have a "hyper-reactivity to the hedonic properties of food, coupled with the motivation to engage in appetitive behaviors." In women diagnosed with bulimia nervosa (BN), Bulik et al. [40] reported that food reinforcement decreases following deprivation. On the basis of these results, we posit that obese individuals with an eating disorder would demonstrate an overall increase in their self-reported scores on the PEQ due to reward sensitization. However, we would not anticipate a significant increase per se in obese individuals who do not exhibit a binge eating-related eating disorder.

Although questionnaires provide validated subjective measurements about food reward and eating behavior, the use of objective methodologies provide insight into the effect of highly palatable food on central dopamine activity and brain regions known to have a high density of dopamine neurons (positron emission tomography and functional magnetic resonance imaging, respectively). However, the cost of instrumentation, the expertise needed to operate the equipment and interpret the data, as well as the engineering and mechanical constraints of the scanners make objective measurements of the dopaminergic response to food challenging. A promising methodology, electroretinography (ERG), may provide a more efficient way to objectively assess central dopamine activity. ERG is a clinical ophthalmological procedure that overcomes the impediments of other neuroimaging methods. ERG records the electrical potential from the retina, which is dependent upon dopamine signaling [41,42], in response to light simulation. ERG has been used to show a negative correlation between self-reported cocaine craving and dopamine-mediated retinal signal [43,44] and a positive correlation between dopamine metabolite levels in the cerebrospinal fluid [45] and dopamine-mediated retinal signal. We have also used ERG to demonstrate a positive correlation between increased dopamine-mediated retinal signal with food stimulation and BES scores [46], consistent with positron emission tomography [47]. We are currently conducting research to extend those findings to chocolates varying in cocoa, sugar, and fat content. Results from this ongoing study should provide further evidence that ERG should be considered as a low-cost, non-invasive method for objective evaluation of the stimulating properties of food in conjunction with ARCI questionnaires.

### **5. Conclusions**

Chocolate's ability to modulate both the opioidergic and the dopaminergic systems, evident by the significant increase in self-reported scores on the PEQ, is consistent with research demonstrating chocolate's reinforcing potential and comforting and mood-ameliorating effects [2,3,21–23]. These results help explain chocolate's high reinforcing value, as foods that elevate feelings of euphoria, as indicated by an increase in the number of positive responses on the MBG subscales, have a greater reinforcing potential [17]. Given the measurable psychoactive dose–effect relationship observed in this study, we posit that chocolate's ability to provoke "addictive-like" eating behavior is initiated by its bioactive compounds, and each incremental increase in added sugar further enhances these effects.

**Author Contributions:** The authors' responsibilities were as follows—S.L.C. and J.A.N.: conceived the project, developed the overall research plan, oversaw the study, wrote and edited the manuscript, and had primary responsibility for final content; S.L.C., L.L., and E.A.: collected the data, analyzed the data, and edited the manuscript; and all authors read and approved the final manuscript.

**Funding:** This work was supported by the Agricultural Research Services of the United States Department of Agriculture #5450-51530-051-00D (S.L.C.) and a competitive pilot grant from the Clinical Translational Research Institute (CTRI/DUCOM) to J.A.N. The role of the funding sponsors was to approve the study and the submission of this manuscript for publication (USDA) and approve funding for the grant proposing this research (CTRI/DUCOM).

**Acknowledgments:** The authors would like to thank Clint Hall and Joshua Severud for their assistance with the implementation of the protocol, data collection, and data entry, and Wei DU, MD, MS (Chair, Department of Psychiatry, Drexel University College of Medicine (DUCOM), for his insights on the physiological/pharmacological effects of addictive substances.

**Conflicts of Interest:** The authors have nothing to disclose. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental status, religion, sexual orientation, genetic information, political beliefs, reprisal, or because all or part of an individual's income is derived from any public assistance program. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA's TARGET Center at (202) 720-2600 (voice and TDD). To file a complaint of discrimination, write to USDA, Director, Office of Civil Rights, 1400 Independence Avenue, S.W., Washington, D.C. 20250-9410, or call (800) 795-3272 (voice) or (202) 720-6382 (TDD). USDA is an equal opportunity provider and employer.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Discussion* **Fat Addiction: Psychological and Physiological Trajectory**

### **Siddharth Sarkar 1, Kanwal Preet Kochhar <sup>2</sup> and Naim Akhtar Khan 3,\***


Received: 9 October 2019; Accepted: 12 November 2019; Published: 15 November 2019

**Abstract:** Obesity has become a major public health concern worldwide due to its high social and economic burden, caused by its related comorbidities, impacting physical and mental health. Dietary fat is an important source of energy along with its rewarding and reinforcing properties. The nutritional recommendations for dietary fat vary from one country to another; however, the dietary reference intake (DRI) recommends not consuming more than 35% of total calories as fat. Food rich in fat is hyperpalatable, and is liable to be consumed in excess amounts. Food addiction as a concept has gained traction in recent years, as some aspects of addiction have been demonstrated for certain varieties of food. Fat addiction can be a diagnosable condition, which has similarities with the construct of addictive disorders, and is distinct from eating disorders or normal eating behaviors. Psychological vulnerabilities like attentional biases have been identified in individuals described to be having such addiction. Animal models have provided an opportunity to explore this concept in an experimental setting. This discussion sheds light on fat addiction, and explores its physiological and psychological implications. The discussion attempts to collate the emerging literature on addiction to fat rich diets as a prominent subset of food addiction. It aims at addressing the clinical relevance at the community level, the psychological correlates of such fat addiction, and the current physiological research directions.

**Keywords:** diet; fat; food addiction; obesity

### **1. Introduction**

Over the last half a century, many developing countries have seen rapid socio-economic development, resulting in a move from a traditional to a modern way of life, including changes in local dietary and culinary profiles [1–3]. Abundance and easy availability of food, especially the one that is rich in fat and carbohydrate, have resulted in changes of dietary patterns and preferences. Right from early childhood in developing brains, the exposure and imprinting to high sugar, high salt and high fat food (rich in saturated and trans-fat), which is cheap and easily available, are impacting the health of younger population. Trans fat may lead to its greater consumption than polyunsaturated fat, as the latter is more quicker than the former to trigger satiety [4]. The changes in dietary intake profile with cultural and societal transitions have gained traction [5]. The dietary profile and constitution have a role in the etiopathogenesis of lifestyle-related diseases like obesity, metabolic syndrome, coronary artery disease, gut motility disorders, psychosomatic, autoimmune as well as degenerative disorders. Major transition, noticed during the last couple of years, has been an increasing use of sugar, processed food, beverages, animal-fat based food rich in trans-fats that have impacted human health [6–10].

In the recent years, there is a growing interest in the concept of food addiction from both clinical and applied nutritional research perspectives [9,11–14]. The increase in obesity, and associated metabolic syndrome and diabetes mellitus have called into the questions about factors leading to genesis of obesity. The imbalance of energy intake has been proposed as one of the reasons of increasing prevalence of obesity, though there are other several factors, i.e., epigenetics, psychological trauma, use of medications and dieting, that may increase body weight gain. why do some individuals consume excess of certain types of food (including fat-rich food)? Hence, the phenomenon of addiction to food has been suggested to be one of the mechanisms. Food addiction can refer to a variety of substrates, but fat and sugars have been considered the typical prototype food items to which individuals develop addiction. The occurrence of distinct features of salience and inability to control intake of specific types of foods have been considered similar to addiction to other psychoactive substances. The adverse consequences of uncontrolled fat intake on the body metabolism have been documented [15,16]. This has implications in intervention modules for addressing this problem and promoting healthy lifestyles [17,18]. Yet, the understanding of fat addiction as a concept is still under evolution, and progress is being made to characterise and discern the psychology and physiology behind this condition.

Recent studies suggest that fat has its own metabolic, physiological and nutritional profiles, which are distinct from other macronutrients [19–21]. Fat accords palatability and organoleptic properties to food, and is consumed across all ages from infancy through adulthood to elderly. Evolutionarily, this confers survival benefit due to high energy density, more so in cultures with thrifty genotypes [22–24]. In recent years, the benefits of mono-unsaturated fatty acid and controlled amounts of saturated fat intake have been revisited, especially in the context of benefits in cognitive functioning, synaptic connectivity, and membrane stability for both brain and heart health [25,26].

In this context, taste for fat has been proposed as the sixth taste modality in recent years [27]. The interactions between fatty acids and specific receptors in taste bud cells elicit physiological changes that are implicated in dietary fat preference via the activation of tongue-brain-gut axis. This phenomenon has an implication in the genesis of obesity as oro-sensory detection of nutrients determines the 'liking' and 'wanting' of food products. It has been proposed that there are two components of eating behavior, represented in neuronal circuits, i.e., emotional (hedonic and affective) and metabolic (homeostatic). Obesity may arise due to the imbalance of these two eating motives.

Research on food addiction or eating addiction has not paid distinct attention to specific nutrients like dietary fat [9,10,12,13]. The concept of fat addiction would have important psychological determinants like motivation, depression, anxiety and reasoning that merit cautious evaluation. There is cognitive appraisal that makes an individual "like" and "want" a specific food product, and the reward obtained from the food is cognitively processed as well. Hence, the intertwined psychological and physiological aspects of addiction towards fat rich food must be considered and understood further. There is a lack of comprehensive synthesis of literature to provide an account of fatty food addiction. In this paper, we have aimed at providing an overview of the construct, the clinical relevance, the psychological correlates, and the current physiological research trajectory in the emerging area of fat addiction as a subset of food addiction. Wherever specific literature with regard to fat is not available, evidence related to food addiction will be alluded to. However, our main emphasis is to discuss about fat addiction as this phenomenon might lead to high dietary fat intake and, consequently, to obesity. The term "high fat" in this article would mean the diet where the calories brought by fat are more than 40% of total dietary calories as most of industrialized countries recommend to respect this limit. As we have mentioned in the title, our main emphasis is to shed light on fat addiction and we have excluded other addictive behaviors like sweet addiction.

### **2. The Construct of Fat Rich Food Addiction**

Addressing the issue of obesity would require improved knowledge of pathophysiological and neurobehavioral mechanisms. This would help better target behaviors which predispose individuals to obesity [28]. Schmidt and Campbell argue that disordered eating cannot remain "brainless" [29], and the "psychological constructs", that define aberrant consumptive patterns of food, are relevant. In this regard, addiction to food explains hedonistic excess and uncontrolled consumption of food items which are associated with adverse consequences.

Addiction towards fat rich diet relates to the overall definition of addiction. Addiction has been conceptualized as a maladaptive pattern of substance intake or behavior that signifies neurobiological changes and is associated with adverse consequences. The nosological systems providing nomenclature to diagnosis has moved on from abuse and dependence to substance use disorders in DSM5. The criteria based evaluation of the cluster of symptomatology helps provide with coherent account of the disorder, and categorize individuals who meet a threshold for diagnosis and consequent potential treatment.

### *2.1. Defining Fatty Food Addiction in the Context of Nutrient Intake*

Food addiction shares some of the commonalities with drug addiction like craving, bingeing and tolerance [30]. The DSM5 criteria for substance use disorders have been adapted and explored in the context of food addiction [31,32]. The 11 criteria for substance use disorders can be applicable to individuals with addiction to lipid dense foods (especially trans and saturated fat). The empirically supported criteria describe a substance (food) often taken in larger amounts or over a longer period of time than that was intended, persistent desire or unsuccessful efforts to cut down or control substance use (food), and continued use despite knowledge of having a persistent or recurrent physical or psychological problem. The plausible features include great deal of time being spent in activities necessary to obtain or use the substance (food) or recover from its effects, recurrent substance (food) use resulting in a failure to fulfil major role obligations, continued use despite having persistent or recurrent social or interpersonal problems, important social, occupational, or recreational activities are given up or reduced, and tolerance. What might be difficult to clearly clinically elicit are withdrawal (while differentiating from energy deficit), and recurrent use in physically hazardous situations. As with different substances, each of the criteria is endorsed to different extent by a sample of participants.

The diagnostic constructs related to food addiction include binge eating disorder and an eating disorder not otherwise specified. Binge eating disorder is characterised by repeated ingestion of eating in large amounts of food in a short amount of time, followed by intense guilt and attempts to either remove the food (by vomiting or using laxatives) or compensatory behaviors to increase the energy expenditure [33,34]. On the other hand, eating disorder not otherwise specified is a diagnostic rubric that resembles anorexia nervosa or binge eating disorder, but does not fulfil the diagnostic thresholds for these disorders. These disorders may have some overlap with food addiction from a phenomenological and behavioral perspective, but the constructs themselves are distinct. It has been seen that individuals with binge eating disorders have greater rates of food addiction, than expected by chance [35,36], though at the same time, not all individuals with binge eating disorders would have food addiction [36,37].The main point of divergence lies in the focus of the constructs: food addiction lays emphasis on the salience and loss of control of hedonic eating behaviors, while eating disorders are accompanied by intense immediate guilt after excessive food consumption and efforts are made to get rid of (effects of) the ingested food quickly.

### *2.2. Clinical and Epidemiological Implications of Addiction towards Fat*

While limited literature has looked at addiction to fat rich foods per se, there is enough evidence that has ascertained the occurrence rate and determinants of food addiction in the community and clinical samples [38,39]. The questionnaires used to assess food addiction generally incorporate fat as a component of food that the respondents are asked to think about, when they answer the questions. The Yale Food Addiction Scale is perhaps the most commonly used instrument for the assessment of food addiction. The weighted mean prevalence of food addiction according to this instrument was 19.9% [38]. The prevalence of food addiction was high in women with obesity [38]. Also, food addiction was higher in clinical samples, as compared to community samples [37,38]. Food addiction was high

in subjects that were either obese, or suffered from eating disorders. High scores of food addiction were associated with high depressive symptoms, food craving and impulsivity. Food addiction has not only been related to negative mood states, but also with poorer quality of life [40]. It has been seen that individuals with food addiction had higher dietary fat intake as compared to those without food addiction [41]. Similarly, Pursey et al reported that the subjects with high food addiction scores had high percentage of consumption of saturated fat [42]. Thus, food addiction provides a paradigm for identification of individuals with skewed dietary profiles with other psychological vulnerabilities, which might require concomitant attention.

Food addiction has also been studied in those individuals who have undergone bariatric surgery which is generally indicated for people with severe obesity [43]. The rates of food addiction in bariatric surgery population go down after the surgery. In one study, the proportion of individuals with food addiction reduced to 2% post-surgery from 32% pre-surgery [44]. Another long term follow-up suggested that the rates of food addiction reduced from 57.8% to 7.2% at 6 months and to 13.7% at 12 months after surgery [45]. In pre-operative cases of bariatric surgery, the dietary intervention is less effective in individuals with food addiction [46]. It has been seen that food addiction in bariatric surgery patients was associated with greater levels of depression, anxiety and binge eating episodes, though it did not predict the degree of weight loss. Thus, it seems that food addiction has some clinical prognostic influence with surgical intervention outcomes.

### *2.3. Measurement Approaches*

Currently the standard of practice for determination of food addiction has been the diagnostic cut-off from the Yale Food Addiction Scale (YFAS) [47]. The YEAS is a 25 item self–reported questionnaire based scale that assesses various features of food addiction. There are two items that assess for clinically significant impairment or distress. The instrument looks at the past year pattern of food intake and includes fatty foods like steak, bacon, hamburgers, cheeseburgers, pizza, and French fries as one of the representative group of foods that are mentioned in the questionnaire. The instrument has become standard of use in the field of food addiction. The instrument has adequate internal reliability, good convergent validity and good discriminant validity. The instrument has been adapted for use in children [48]. The instrument has also been translated into several other languages like Chinese, French and Malay [49–51]. A newer version of the scale (YFAS 2.0) has been developed considering the changes in conjunction with the DSM5 [52]. The instrument has been used in studies of epidemiology, etiology, nosology and interventions of food addiction. While the YFAS addressed food as a whole, assessing fat addiction separately may have implications for interventions. This could be in terms of the type of food products that are focused upon in the intervention modules that are developed. This would have also a corollary for the investigation procedures that can include assessment of salience and behavioral neuroplasticity (eye tracking and neuroimaging) for different types of food products (fat rich versus carbohydrate rich, sweet versus savory fatty food) that are implicated in food addiction.

Other self-reported scales and questionnaires for assessment of aspects of food addiction are also available and have been validated, though they rely on features like craving and eating patterns. These include Eating Behaviors Questionnaire [53], Food Cravings Questionnaire [54], Eating Behaviors Patterns Questionnaire [55], and Power of Food Scale [56]. Many of these questionnaires are self-reported, i.e., the individual reads through the questions and responds through them. The responses are thereafter graded and interpreted based upon the cut-offs from the population scores.

### **3. Psychological Correlates of Addiction to Fat Rich Diets**

### *3.1. Attentional Biases and Cognitive Functioning*

Research has been carried out towards attentional biases and psychological processing in individuals with food addiction. Obese as compared to lean teens showed less activation of prefrontal regions (dorsolateral prefrontal cortex, ventral lateral prefrontal cortex) when trying to inhibit responses to high-calorie food images which suggest behavioral evidence of reduced inhibitory control [57]. Adults who had greater dorsolateral prefrontal cortex activation when instructed to "resist craving" after viewing food images had better weight loss success following gastric bypass surgery [58]. This suggests that visual cue induction paradigms have relevance to assessment of how food images are processed centrally.

Rodrigue et al. [59] compared those with higher and lower food addiction scores on cognitive processes of planning, inhibition, cognitive flexibility and error processing. The investigators found that high food addiction group differed from the low food addiction group only in terms of inhibition/cognitive flexibility scaled scores, but not in individual scores. The authors infer that though basic level processes are intact, individuals with higher food addiction scores experience greater difficulties in more challenging context where they had to simultaneously keep in mind to inhibit a behavior and switch their mind-set when the task required it. This might make it difficult for them to anticipate the long-term consequences of behavior. Also, individuals with symptoms of food addiction made more errors as the interference task became challenging, suggesting that those with food addiction might have greater difficulty in detecting and monitoring errors. Another study compared error monitoring among individuals with food addiction and healthy controls using the Eriksen flanker task [60]. The results suggested that food addiction group had higher number of errors on the flanker task, implying impaired performance monitoring and cognitive control, as seen with other addictions. In a study that included women with obesity, food addiction severity levels were negatively correlated with overall scores on the Iowa Gambling Task, which measures decision making capacity [61]. Also, those with food addiction had attentional deficits as reflected by more omissions and perseveration errors on the Continuous Performance Task. On the other hand, Blume et al. [62] compared response inhibition, attention, decision-making, and impulsivity among four groups of individuals, i.e., obesity and food addiction; obesity and binge eating disorder; obesity/food addiction and binge eating disorder; and obesity only. The authors did not find food addiction to be related to altered executive functioning.

Ruddock et al. [63] evaluated the attentional bias using eye tracking while showing pictures of chocolate among individuals with self-perceived food addiction in design that evaluated state factors like hunger or expectancy of reward or having food addiction. The authors found that the expectancy of receiving chocolate as reward was associated with attentional bias, while hunger state or having self-perceived food addiction was not associated with attentional bias toward food related cues. In another eye tracking paradigm, sad mood induction through showing of a video of child passing away with cancer was associated with attentional bias towards unhealthy food among those with food addiction, but such a change did not occur in those without food addiction [64]. This suggests that emotional cues may impel or prime those with food addiction towards specific food types. In another study, Gearhardt et al. [65] studied food-related visual attention and dwell time of food stuff among obese and overweight women. The authors reported that hunger was associated with attentional bias toward sweets, and trend level attentional bias towards fried (fatty) foods. On the other hand, hunger was associated with shorter dwell time on fried food. Taken together, literature suggests that hunger may be an important component that may influence attentional biases in individuals with food addiction. We acknowledge that though addiction has gained traction, fat addiction is an emerging concept, and nevertheless needs debate and discussion to inform lifestyle practices and research directions.

### *3.2. Craving and Liking*

Craving and liking are related, but represent distinct terms that are linked to food addiction. While craving refers to desire or urge to eat a food item, liking refers to qualitative and affective evaluation of food [66–68]. Liking for fat has been evaluated in a large web-based study to examine the determinants of dietary patterns and nutritional status [69]. The investigators reported that individuals with a strong liking for fat had high total energy and fat intake, and high consumption of saturated fats,

meat, butter, sweetened cream desserts and croissant-like pastries. Such individuals also consumed low quantities of fiber, fruits, vegetables and yogurt. It was highlighted that increased liking for fat, especially fat-and-salt liking, was associated with a lower intake of fruit and vegetables.

Gearhardt et al. [70] assessed craving for 180 food items among a sample of 105 obese or overweight women. The authors found that those with greater symptomatology of food addiction had higher craving ratings for fatty foods. However, as BMI increased, the craving decreased. In contrast to craving in this study, high fat content was not associated with high liking for food product, suggesting a dichotomy between craving and liking.

### **4. Understanding the Physiological and Neurobiological Processes of Fat Food Addiction**

There have been considerable advances in understanding the mechanisms of addiction for food rich in lipids. Some of them were conducted on animals, particularly rodent models. Other directions of research, for example, genetics and neuroimaging have explored the origin of addiction towards fat and other palatable foods in human participants [71,72]. The reward pathway (schematically shown in Figure 1) is intricately linked to understanding the addiction to fatty food, though some differences have been reported in food addiction and substance use disorders [73].

**Figure 1.** Schematic representation of the reward pathway. The figure shows the interplay between different neurons where the nucleus accumbens seems to be the central player, receiving the projection of dopaminergic, glutamatergic and opioidergic neurons. The model for food addiction might be quite different and is under examination [74].

### *4.1. Animal Models for Understanding the Addiction to Fat Rich Foods*

The advantage of animal models is that they are able to develop addiction to fat as the diets given in animal models are homogenous [75–77]. This is not possible in human studies. The high-fat diet that generally comprises of 45% of energy from lipids is used to trigger obesity in rodents [78]. However, none of the experimental high-fat diets resembles closely to human diet fatty acid composition, though they are efficient to induce obesity.

Initially, Avena et al. [79] developed a model for sugar addiction which showed patterns of binge eating, withdrawal symptoms, and neurochemical changes similar to those observed with opiate addiction. The phenotype of animal was created by restricting the frequency, duration or access to sugar. Subsequently, fat models of bingeing have been developed in conjunction with carbohydrates, wherein

corn oil is used as a reinforcing food item. However, the features of opiate-like withdrawal were not noted by Avena et al. [80] when animals were deprived of food after fat bingeing. This observation suggests that fat addiction may have different phenomenological aspects than the addiction to sugars. An alternate explanation could be that fat addiction might be more closely aligned to behavioral addiction like gambling disorder, while addiction to sugar rich food might be more closely aligned to substance use disorders. The development of animal models has the potential to advance the field substantially, by enabling to better understand the neurobiological alterations, and to assess the changes in bingeing behaviors with medications or other interventions [81]. Yet, one needs to be cognizant of the fact that translation of human behavior of food consumption is much more complex than animals, and is influenced by socio-economic and political environment, and the determinants like cost, availability and marketing. Furthermore, it is possible that modeling addiction in animals (especially rodents) might differ substantially from clinical situation [8]. It has been argued that simplistic experiments would need critical reflection about translational validity of patterns of eating behavior and food choice from animals to humans.

### *4.2. Neurotransmitters Including Dopamine*

Animal studies have suggested implication of dopamine in the nucleus accumbens in the rat model of addictive behaviors towards fat [82]. Hence, microdialysis samples were taken in this model before, during and after sham feeding with corn oil. The study found an increase in dopamine in the sham licking group leading to the inference that corn oil increases dopamine concentrations in the nucleus accumbens in a manner similar to those induced by sucrose. In another study, low concentration of non-esterified fatty acid (linoleic acid) increased the dopamine levels in the nucleus accumbens and amygdala in a manner equivalent to those resulting from corn oil in the brain's reward system [83].

Dela Cruz and colleagues studied the expression of c-Fos in reward circuit areas in rats which were exposed to sugars and fats [84,85]. The authors reported c-Fos like immunoreactivity after consumption of corn oil solutions, isocaloric glucose and fructose, in the dopaminergic mesotelencephalic nuclei (ventral tegmental area) and projections (infralimbic and prelimbic medial prefrontal cortex, basolateral and central-cortico-medial amygdala, core of nucleus accumbens as well as the dorsal striatum), but not in the nucleus accumbens shell. This signified transcriptional activation of the dopaminergic pathway with exposure to certain nutrients including fat.

Dela Cruz et al aimed at investigating whether dopamine antagonists (D1 receptor antagonist SCH23390 and D2 receptor antagonist raclopride) attenuated the development of fat conditioned flavour preference among rats [86]. These investigators reported that, as compared to sucrose, the D1 and D2 receptor antagonists were not able to attenuate the fat conditioned flavor preference. They further suggested that fat addiction in rats could possibly have distinct mechanisms than sugars, which involved the post ingestive phase.

The role of opioid receptors and fatty food addiction has also been explored [76]. It has been seen that after injection of morphine, a mu-opioid receptor agonist, rats preferred fats over carbohydrates when both were available. Intra-accumbens administration of opioid agonists increased the consumption of fats, and the effect was blocked by the administration of naltrexone, an opioid antagonist [87]. The opioid receptors have been implicated in not only the 'liking' process, but also the 'wanting' process of excessive food consumption, and the effects are blocked by opioid antagonists.

Endocannabinoid system is another neurotransmitter system studied in relation with animal model of excessive fat consumptive behavior. Ward et al. [88] studied male, cannabinoid (CB1) knockout mice which were trained to respond to the sweet reinforcer (Ensure) or corn oil. The authors suggest that CB1 receptor antagonism selectively attenuated reinstatement of responding for Ensure. Interestingly, the genetic deletion of the CB1 receptor did not attenuate reinstatement of corn-oil seeking. The authors suggest that either CB1 receptor system does not play an equivalent role in modulating conditioned seeking or corn oil may serve as a robust reinforcer. Additionally, Brissard et al. [89] found that invalidation of CB1R gene was related to lower levels of fat preference among mice, and

similar results were obtained after using rimonabant, a cannabinoid receptor antagonist. The authors reported that fat taste perception was mediated through calcium signaling and GLP-1 secretion in lingual taste bud cells. Peterschmitt et al. [90] looked at the link between the gustatory and the reward pathway with regard to fat intake. The authors observed that lipid taste perception was based upon the systematic activation of the major cerebral structures of the canonical gustatory pathway and was intricately linked to the reward pathway through the ventral tegmental area.

### *4.3. Neuroimaging Correlates*

Though literature exists on the neuroimaging correlates of obesity [91,92], studies on the neuroimaging of food addiction have gradually started to come up. Gearhardt et al. [93] assessed the blood oxygen level-dependent functional magnetic resonance imaging (fMRI) activation in response to receipt and anticipated receipt of palatable food (chocolate milkshake) among adolescent female participants. The investigators demonstrated that food addiction scores correlated with greater activation in the anterior cingulate cortex, medial orbitofrontal cortex and amygdala, consequent to anticipated receipt of food. The participants with high food addiction scores had enhanced activation of dorsolateral prefrontal cortex and caudate, but less activation in lateral orbitofrontal cortex in response to anticipated receipt of food. These findings underscore the similarity of food addiction to other types of addictions, especially in relation to involvement of the reward pathway.

Hsu et al. [94] assessed response inhibition and error processing among subjects with obesity and sweet food addiction by fMRI. Women with obesity and food addiction had a higher score for impulsivity and lower brain activation (processing response inhibition over the right rolandic operculum and thalamus) than controls. The activation during error processing over the left insula, precuneus, and bilateral putamen were higher in the subjects with obesity and sweet food addiction than controls. These findings suggest that women with obesity and sweet food addiction have impaired rolandic operculum activation.

A further study looked at the relationship of food addiction and functional connectivity in the brain during fasting and fed state [95]. The authors found that high number of symptoms of food addiction were associated with ventral caudate-hippocampus hyperconnectivity in the fasted scan only. However, a significant reduction of this connectivity was observed in the fed scans, suggesting that heightened connectivity in the ventral striatum during a fasted state corroborated reward prediction signals, further lending credence to the involvement of the reward pathway.

A schematic representation of the neurobiological relationship of fatty food intake, mediated through gustatory signaling and reward pathway, is presented in Figure 2.

**Figure 2.** Relationship of food intake and reward pathway. The figure shows that the gustatory memory for fat and its implication would depend on the cues coming from taste bud cells, localized in the lingual papillae, and vagal nerve information from intestinal lipid sensing. Both kinds of information will ascend to different parts of the brain via NTS. Hippocampus will be involved in the learning of palatability of fat, and communicate to VTA which is sending its afferences to frontal cortex, straitum and other parts of the brain. Indeed, the dopaminergic zone covers VTA and NA. NTS: nucleus tractus solitaris; HIPP: hippocampus; VTA: ventral tegmental area.

### *4.4. Genetics Underpinnings*

Several studies have also looked at the genetic associations of food addiction. A study evaluated whether a composite index of elevated dopamine signaling, a multilocus genetic profile score (MLGP) could segregate between those with food addiction and normal eating behavior [96]. The authors observed that MLGP score was high in subjects with food addiction, and it correlated positively with binge eating, food cravings, and emotional overeating. This finding supported the view that dopamine signaling genetic profile was different in subjects with food addiction.

Pedram et al. [97] studied food addiction in the Newfoundland population and observed the major allele A of rs2511521 located in DRD2 and the minor allele T of rs625413 located in TIR domain containing adaptor protein (TIRAP) to be significantly associated with food addiction. A study on the Asian American college students assessed the relationship of food addiction and a dopamine-resistant receptor (DRD2) polymorphism [98]. The authors reported that DRD2 A1 allele among Asian Americans (versus A2 allele) was associated with greater carbohydrate craving, but not fat craving. Cornelis et al. [99] presented genome wide analysis of food addiction in more than 9000 women with European ancestry. This study showed two loci significant at genome-wide level (17q21.31 and 11q13.4), but they did not have any obvious roles in eating behavior. The study did not find any candidate single nucleotide polymorphism or gene for drug addiction to be significantly associated with food addiction after correction for multiple testing.

There is accruing literature that suggests that reduced fat taste perception may contribute to increased fat consumption and, consequently, to obesity [100], and this might be influenced by the genetic polymorphisms. Studies from USA, Algeria and Tunisia seem to suggest that rs1761667-AA genotype of CD36 receptor is associated with obesity, and high thresholds for oro-sensory detection of dietary lipids [101–103]. Interestingly, Plesnik et al. [104] reported that another variant of CD36, i.e., rs1527483 SNP, was associated with greater body weight in young Czech participants. Thus, the taste threshold and preference for fat, mediated through specific genetic polymorphisms, may determine fat-eating behaviors that may lead to fat addiction.

### **5. Conclusions, Limitations and Future Directions**

Addiction to food products, especially those rich in fat has received attention in recent decades. Figure 3 depicts the overall associations and implications of addiction to fat replete diets. The construct of food addiction has undergone sufficient scrutiny, and means and measures have been developed to reliably assess this condition. Fat as a component of food addiction itself has yet to find its niche, but has possible implications for the control and prevention of obesity. Research has elaborated on the attentional biases and cognitive functioning in individuals with food addiction, and has pitched varied findings. Animal models of food addiction, especially those which have used fat as a substrate, have expanded the scope of the field and have given an armamentarium of options for understanding the condition and interventional choices. Neuroimaging and genetic studies have also progressed, enriching the field.

**Figure 3.** Schematic representation of addiction to fat.

Some of the limitations of the present paper should be born in mind while considering different observations. The present findings have not been synthesized as a systematic review, but are rather in the form of a narrative review. The advantage of narrative review is that broad range of findings can be presented to provide the reader with various dimension of the topic, but it may not be able to present all relevant literature in the field. The concept of food addiction, similar or different from other (substance/drug) addiction, as a continuum of behavioral addiction has been debated. Additionally, segregation of food addiction into a specific macronutrient based fat addiction may be difficult to operationalize clinically, as food products generally contain multiple elements together (fats, sugars and salt). Furthermore, food addiction as a concept has been criticized as pathologizing a normal behavior, and some researchers have questioned the validity and nomenclature of the construct [8].

Future research investigations are required to look at the stability of addiction to fat rich food over longitudinal course. The etiological understanding would be strengthened by foray into multi-modal assessment incorporating neuroimaging, genetic and psychological domains. Another aspect would be determining a threshold for fat composition in the food to qualify for fat addiction. Neurobiological studies would be strengthened if they incorporate the neuroimaging responses to palatable food taste, and cues (including visual cues). Neurobiological correlates of fat addiction and its persistence can be elicited by developing studies for individual nutrient components as well as combinations in sweet and savory food and linking it to markers of obesity. It would be pertinent to see how other neuropsychological functions like motivation, sensory processing in various domains and working memory interact with reward and homeostatic systems in controlling the various phenomenological aspects of fat addiction. Also, relationship of addictive behavior with intervention outcomes (for example, for obesity) needs to be looked into. The social impact of food addiction from a macro policy level, and the lived experience of the individual with 'food addiction' would help better understand the condition. Also, attempts to enhance the awareness of this condition and the harmful impact of trans-fat and saturated fat, coupled with greater funding to understand and address this issue would help both for primary and secondary prevention. Of course, the overarching aim would be to provide with relevant prevention for at-risk population, and suitable interventions for affected individuals.

**Author Contributions:** Conceptualization, S.S., K.P.K., and N.A.K.; review of literature, S.S., K.P.K., and N.A.K.; data curation, S.S.; writing—original draft preparation, S.S.; writing—review and editing, K.P.K. and N.A.K.; project administration, N.A.K.

**Funding:** This research received no external funding

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Food Addiction: Implications for the Diagnosis and Treatment of Overeating**

### **Rachel C. Adams 1,\*, Jemma Sedgmond 1, Leah Maizey 1, Christopher D. Chambers <sup>1</sup> and Natalia S. Lawrence <sup>2</sup>**


Received: 15 July 2019; Accepted: 21 August 2019; Published: 4 September 2019

**Abstract:** With the obesity epidemic being largely attributed to overeating, much research has been aimed at understanding the psychological causes of overeating and using this knowledge to develop targeted interventions. Here, we review this literature under a model of food addiction and present evidence according to the fifth edition of the Diagnostic and Statistical Manual (DSM-5) criteria for substance use disorders. We review several innovative treatments related to a food addiction model ranging from cognitive intervention tasks to neuromodulation techniques. We conclude that there is evidence to suggest that, for some individuals, food can induce addictive-type behaviours similar to those seen with other addictive substances. However, with several DSM-5 criteria having limited application to overeating, the term 'food addiction' is likely to apply only in a minority of cases. Nevertheless, research investigating the underlying psychological causes of overeating within the context of food addiction has led to some novel and potentially effective interventions. Understanding the similarities and differences between the addictive characteristics of food and illicit substances should prove fruitful in further developing these interventions.

**Keywords:** food addiction; overeating; obesity; impulsivity; reward sensitivity; cognitive training; neuromodulation

### **1. Introduction**

In 2003, obesity was declared a global epidemic by the World Health Organisation [1], and the prevalence of overweight and obesity in both developed and developing countries continues to increase [2,3]. In 2016, 39% of adults were estimated to be overweight and 13% to be obese [4]. Overweight and obesity present a substantial economic burden; in the UK, the total direct and indirect costs are expected to reach £37.2 billion by 2025 [5]. One of the common explanations for the increase in obesity over recent decades is the environment and, in particular, the availability of highly varied, palatable and fattening foods—which have been considered to be addictive [6–9]. While many individuals manage to resist these temptations and maintain a healthy weight, obese individuals have been shown to have a preference for such energy-dense foods compared to healthy-weight individuals [10–12]. The critical question is why some individuals are able to resist overeating while others cannot; what is the evidence for 'food addiction' and how can this be used to inform interventions for overeating.

The concept of 'food addiction' has been evident in the media and general public for some time and is gaining increasing interest in the scientific literature [13]. There are now numerous reviews discussing the diagnostic, neurobiological and practical aspects of food addiction, with arguments both for and against its utility and validity [14–20]. This surge of interest comes with the perspective that addiction can be conceptualised as a loss of control over intake for a particular substance or behaviour without the need to focus purely on psychoactive substances [21,22]. The fifth edition of the Diagnostic and Statistical Manual [23] acknowledged this shift in perspective, with the addition of gambling disorder as the first behavioural addiction. Acceptance of this disorder was based on evidence that gambling can produce behavioural symptoms that parallel those of substance addiction and can activate the same neural reward circuits as drugs of abuse [24,25]. There is now a large body of research documenting similar observations for overeating and obesity. Moreover, treatments developed for addictive disorders have also shown some efficacy for the treatment of obesity and overeating. These findings highlight how a model of food addiction may help us to understand elements of overweight/obesity beyond a simple lack of willpower and can also be used to inform effective interventions and policy [26–30].

Food addiction has not yet been recognised in the DSM; however, the similarities between some feeding and eating disorders and substance-use disorders (SUDs) have been acknowledged. These similarities include the experience of cravings, reduced control over intake, increased impulsivity and altered reward-sensitivity. Binge eating disorder (BED) and bulimia nervosa (BN) have been proposed as phenotypes that may reflect these similarities to the greatest extent [31–34]. Both BED and BN are characterised by recurrent episodes of binge eating in which large quantities of food are consumed in a short time accompanied by feelings of a lack of control, despite physical and emotional distress. Reports of food addiction have been shown to be particularly high amongst these individuals [32,35,36]. Food addiction has also been acknowledged with a standardised 'diagnostic' tool—the Yale Food Addiction Scale (YFAS) [37,38]. The YFAS is a questionnaire that parallels the diagnostic criteria for SUDs. The scale has so far been shown to exhibit good internal reliability as well as convergent, discriminant and incremental validity [37–40].

In this review, we first discuss the DSM-5 diagnostic criteria for SUDs to summarise evidence for food addiction. These criteria are defined as 'a cluster of cognitive, behavioural and physiological symptoms' [23]. More specifically, the following categories are considered: impaired control, social impairment, repeated use despite negative consequences and physiological criteria. However, it should be noted that the physiological criteria of tolerance and withdrawal—for which there is less evidence in relation to food—are not necessary for a diagnosis of SUD. The DSM-5 also states that although changes in neural functioning are a key characteristic of SUDs, the diagnosis is based on a pathological pattern of behaviours. Hence, we discuss the diagnostic criteria initially, followed by a review of neurobiological evidence. We then explore the question of how this information can be, and has been, applied to interventions for overeating.

### *1.1. Impaired Control*

Taking larger amounts of the substance for longer periods than intended has been cited as one of the most commonly reported symptoms in overweight/obese and BED individuals [41,42]. Excessive and uncontrolled eating also forms the definition of binge eating in BED [23]. Although bingeing can be a planned behaviour, it has been shown that planned binges still result in a greater intake than initially intended [41]. Binge eating has also been documented in non-clinical samples [43,44]; however, in these individuals, occasions of impaired control are more likely to reflect unintentional snacking and excessive portion sizes [8,41,45].

Unsuccessful efforts to restrict food intake are also well documented, with many dieters failing to maintain their diet or even gaining weight in the long term [46–51]. In their paper reviewing evidence for refined food addiction (i.e., processed foods with high levels of sugars or sweeteners, refined carbohydrates, fat, salt and caffeine), Ifland et al. [52] report that 'Every refined food addict reports a series of attempts to cut back on eating. They have used a variety of techniques' (pg. 521). Curtis and Davis [41] also report similar anecdotes in women with BED who describe avoiding certain trigger foods to control their binges.

The third criterion of time spent obtaining, using and recovering from substance use also translates to BED and BN. These individuals may spend a lot of their time thinking about, engaging in and recovering from binge episodes. As mentioned earlier, bingeing is often a planned behaviour which may require a great deal of effort to purchase and store foods ready for a binge episode [41]. In addition, the criteria for BED emphasise the time spent bingeing, with the number of binge episodes per week determining the severity of the disorder [23]. Moreover, these individuals often experience physical and emotional distress following a binge eating episode. Recovery from food consumption has also been reported in self-identified food addicts with references to feeling sleepy or 'hung-over' [52,53].

Although evidence for food addiction directly related to the DSM-5 diagnostic criteria for impaired control is largely anecdotal, there is a considerable amount of empirical evidence for an association between overeating/obesity and impaired control generally. Two aspects of self-regulatory failure that are particularly pertinent in the case of substance use and overeating are impulsivity and reward sensitivity [54–56].

### 1.1.1. Impulsivity

Although impulsivity is a multi-faceted construct, it can be defined broadly as the tendency to think and act without sufficient forethought, which often results in behaviour that is discordant with one's long-term goals. The role of impulsivity in SUDs is well documented [55,57–60]. Many studies have reported higher impulsivity levels with increasing substance use across a wide range of questionnaires and behavioural tasks, and for a variety of different substances [61–66]. For example, Noël et al. [67] performed a series of behavioural tasks assessing the ability to suppress irrelevant responses (response inhibition) and irrelevant information (proactive interference) in a group of detoxified alcohol-dependent individuals and matched healthy controls. They found a statistically significant group difference for all three tests assessing response inhibition but no differences for proactive interference.

Impulsivity has also been implicated in overeating and obesity [54,68–71]. Overweight/obese individuals score higher on self-reported [72–74] and behavioural measures of impulsivity [75–77], whereas those high in self-control have been shown to be less likely to give in to temptation [78–80] and are more likely to maintain a healthy diet and engage in physical exercise [81–83] Impulsivity scores have also been shown to predict poor food choices [84] and correlate positively with food consumption [85–87]. For example, Guerrieri et al. [87] found that, in a sample of healthy-weight women, those with higher impulsivity scores ate more candy during a 'bogus' taste test than those with lower impulsivity scores. Churchill and Jessop [88] also showed a predictive relationship between impulsivity and snacking on high-fat foods over a two-week period. Scores on the YFAS have also been associated with various measures of impulsivity, such as motor and attentional impulsivity, mood-related impulsivity and delay discounting [89,90].

### 1.1.2. Reward Sensitivity

A heightened general sensitivity to reward has also been linked to both substance use and overeating [69,77,91–93]. In the food literature, self-report measures of reward sensitivity have revealed associations with BMI, food craving and preferences for foods high in fat and sugar [93–95]. Using two behavioural tasks, Guerrieri et al. [69] measured reward sensitivity and response inhibition in children aged 8–10. They subsequently measured food intake in a bogus taste test when the foods were either varied or monotonous. Their results revealed that reward-sensitive children consumed significantly more calories than non-reward sensitive children only when the food was varied. There was no effect of response inhibition on food intake, nor any interaction with variety; however, unlike reward sensitivity, deficient response inhibition was associated with being overweight. The authors suggested that reward sensitivity may play a causal role in overeating, whereas deficient inhibitory control may be more of a maintaining factor. This fits well with findings from a study demonstrating a role of reward sensitivity in the early onset of heroin use and a role of impulsivity in escalating use [92,96].

There is also evidence to suggest that reward sensitivity may decrease with more prolonged or established overeating, with studies showing anhedonia, or hypo-sensitivity to reward, in obese participants [97–100]. For example, Davis et al. [97] demonstrated that although overweight women were more sensitive to reward than healthy-weight women, those who were obese were significantly *less* reward sensitive than overweight women. Importantly, the earlier mentioned association between reward sensitivity and increased BMI was found in a sample of mainly healthy-weight women, with only 1% classified as obese [93]. Although there is a great deal of evidence to suggest that sensitivity to reward plays a role in substance abuse and overeating, the causal direction of this relationship remains unclear. On the one hand, increasing reward sensitivity may lead to overeating by increasing motivation towards pleasurable activities, such as consuming energy-dense foods that elicit dopamine and opioid activation. On the other hand, decreased reward sensitivity may cause individuals to seek out rewarding activities as a form of 'self-medication' in order to boost dopamine functioning (i.e., addictive behaviour is the result of a 'reward deficiency syndrome') [101,102]. These two arguments, and the relevant neuroimaging literature, are discussed further below (see the Neurobiological Similarities section below) and in more detail by Burger and Stice [103].

Burger and Stice [103] offer several theories for how these two causal directions combine to explain obesity. They propose that high sensitivity to reward may initially cause individuals to over-consume palatable foods, but this sensitivity is then modified over time as the brain's reward system adapts and shows divergent changes in food motivation ('wanting') versus hedonic pleasure ('liking'). According to Robinson and Berridge's [104–106] incentive-sensitisation theory, repeated intake results in an increased incentive value for these foods and their associated cues, which may be subjectively experienced as excessive wanting or craving. Moreover, this theory argues that with repeated presentations of palatable foods, the hedonic pleasure derived from consuming the food will decrease due to neural habituation, while the anticipation of reward increases. Hence, a vicious cycle emerges in which the individual will experience less pleasure from the food ('liking'), but will simultaneously experience an increased desire ('wanting') for the food, driving further food seeking and consumption [107–109] (see Figure 1). The experience of intense cravings is the third criterion of impaired control and is another symptom of substance addiction that can be readily applied to overeating and obesity.

**Figure 1.** The proposed cycle of 'food addiction'. Initial vulnerability for the over-consumption of palatable food is marked by increased impulsivity and reward sensitivity, as well as a diminished capacity for inhibitory control. As a consequence of overconsumption, individuals experience tolerance, craving and withdrawal, along with a range of social, emotional and behavioural difficulties such as weight stigmatisation and feelings of guilt and shame. With repeated consumption of these foods, the individual is likely to habituate to the hedonic properties of the food, resulting in reduced enjoyment or liking. These changes are also accompanied by an increased desire or 'wanting' for the food [104–108]. In an attempt to relieve these symptoms, the individual 'self-medicates' by increasing food consumption, which can result in compulsive or binge eating behaviour, thus creating a cycle of addiction. It should be noted that the extent to which each of these mechanisms is experienced varies considerably across individuals. In particular, initial vulnerability to addiction may be related to individual differences in reward sensitivity, impulsivity and inhibitory control [110–113].

### *1.2. Craving*

The term 'food craving' typically refers to an intense desire to consume a specific food [114,115]. Food cravings appear to be very common with reports of 100% of young women and 70% of young men experiencing a craving for at least one food in the past year [116,117]. The most commonly reported craved food is chocolate, although cravings for carbohydrates and salty snacks are also common [118–122]. The prevalence of food cravings has prompted the development of several standardised questionnaires that measure food cravings with a good degree of internal consistency and construct validity [123–127], including a specific questionnaire just for chocolate (Attitudes to Chocolate Questionnaire) [128]. Recurrent food cravings are of interest in relation to food addiction as they have been associated with binge eating, increased food intake and increased BMI [124,127,129–132]. Increased reports of food craving have also been demonstrated in individuals who score highly on measures of self-reported food addiction [133–135] and those with BED and BN [136–138]. Furthermore, just as drug craving is associated with an increased likelihood of relapse [139–141], food craving has been linked to poor dieting success [142–144].

Further support for the similarity between drug and food craving is evident in the findings of cue-reactivity research. The aphorism that cravings are most likely to occur in the presence of substance-related stimuli has been well documented, with cue-exposure paradigms showing significant effects of drug-related cues on self-reported and physiological measures of craving [145–148]. Similarly, exposure to food cues has also been shown to increase food cravings [149,150] and a recent systematic review of 45 studies (involving 3292 participants) concluded that 'food cue-reactivity' (physiological, neural and subjective reward-related responses to food cues) reliably and prospectively predicts both

energy intake and weight gain, particularly over the longer-term, accounting for ~11% (7%–26%) of variance in these outcomes [129]. Food cue-induced craving is especially prevalent among binge eaters and those with BED [151,152] in whom it has been correlated with binge eating frequency and BMI [153]. It is possible, therefore, that certain individuals are more susceptible to cue-induced cravings, and also that this susceptibility may transfer across different substances. Both Mahler and de Wit [147] and Styn et al. [148] found a significant correlation between cue-induced cigarette craving and cue-induced food craving in smokers, suggesting a common mechanism. Cue-induced craving is also believed to strengthen with repeated consumption, fueling the vicious circle shown in Figure 1.

### *1.3. Social Impairment*

Overeating and obesity have been associated with poor social functioning, especially among children and adolescents. When assessing quality of life with child and parent-proxy reports, social functioning is significantly lower for obese compared to healthy-weight children and is inversely correlated with BMI [154–156]. Poor social functioning in overweight children may be partly due to the overt victimisation and teasing experienced as a direct result of their weight status [157,158]. Hayden-Wade et al. [159] found that the degree of teasing experienced by overweight children was positively correlated with loneliness, an increased preference for isolative activities and a lower preference for social activities. This preference for being alone, along with the emotional difficulty of being victimised, fuels a vicious cycle as these circumstances are likely to promote further overeating and binge-eating—which, in turn, leads to increased weight gain and further teasing [42,160] (see Figure 1).

Weight stigmatisation may also affect interpersonal friendships and romantic relationships in adulthood with reports of discriminatory attitudes and behaviours in occupational [161,162] and romantic settings [158,162,163]. For example, Chen and Brown [164] reported that when making sexual choices about a partner, both male and female college students ranked an obese individual as the least liked. In a study focusing on the psychosocial correlates of food addiction, Chao et al. [165] found that, compared to control participants, those who met the YFAS criteria scored lower on physical, mental and social aspects of health-related quality of life. Social impairments were related to self-esteem, sexual life, public distress and work. Interpersonal problems have also been associated with binge eating—a relationship which is likely to be bidirectional [166,167].

### *1.4. Repeated Use Despite Negative Consequences*

It has been noted that due to its increase in prevalence and associated comorbidities, obesity now appears to be a greater threat to the burden of disease than smoking [168]. The physical and psychological effects of overweight and obesity are well documented and include, but are not limited to, depression, an increased risk of diabetes, hypertension, cardiovascular disease and some cancers [169–177]. With pervasive warnings regarding the consequences of overeating, from the media, government, and the medical profession, it seems fair to assume that most overweight and obese individuals are aware of the negative outcomes associated with their dietary behaviour [41,52]. Critically, even those who have undergone weight loss treatment often fail to lose weight or gain weight following intervention [46,48,50,51]. Continued overeating also occurs in those who have received bariatric surgery with patients showing continued snacking and poor food choices [178,179]. There is, therefore, considerable evidence to support continued overeating despite negative consequences.

### *1.5. Physiological Criteria*

Tolerance to a substance occurs when the same amount of the substance has an increasingly diminished effect with repeated use. This effect usually results in escalated use as the individual increases their dosage in order to recreate the original experience. There is some evidence of food tolerance in animal models of sugar addiction. Rats given intermittent and excessive access to sugar solution increase their intake significantly over time, and this is accompanied by neurochemical

changes that are similar to those seen in drug abuse [180,181]. In humans, there is some indication that tolerance to sugar may occur in the first few years of life. The effectiveness of sucrose as an analgesic in young infants is reported to diminish after 18 months of age as sugar consumption increases [182–185]. The possibility of such early tolerance to palatable foods and the methodological difficulties of diet restriction in humans makes finding empirical evidence of tolerance in adults difficult and unlikely. However, statistics indicating increased consumption and portion sizes for these foods provide indirect evidence of tolerance to high-fat/high-sugar foods at a population level [52,186], and also at an individual level based on anecdotal reports. For example, Pretlow [42] found that 77% of overweight poll respondents reported eating more now than when they originally became overweight. Furthermore, in response to a follow-up question asking why they believed that they ate more, 15% indicated that they were less satisfied by food. Hetherington et al. [109] also found that when participants were provided with chocolate for three weeks, they increased their intake over time while simultaneously reporting a reduction in food liking.

Withdrawal is the second physiological criterion for substance abuse and is defined by the presence of physical or psychological symptoms in response to substance deprivation, or the use of the substance in order to relieve these symptoms. Evidence of withdrawal has also been found in the aforementioned animal models of sugar addiction. Under conditions of sugar deprivation, these animals show withdrawal symptoms similar to those seen with morphine and nicotine withdrawal, including physical symptoms of teeth chattering, forepaw tremor, head shaking and reduced body temperature [187,188] as well as increased aggression [189] and anxiety [190]. There are also anecdotal reports of withdrawal-like symptoms in humans, including persistent cravings and negative affects when attempting to reduce food intake [42,191], as well as the tendency to eat to avoid the emotional symptoms associated with withdrawal such as fatigue, anxiety and depression [52]. Using the YFAS, withdrawal symptoms (such as agitation, anxiety, or other physical symptoms) have been reported in up to 50% of individuals with obesity and BED [35].

### **2. Neurobiological Similarities between Palatable Foods and Drugs of Abuse**

Just as altered brain functioning has been reported in SUDs, overeating and obesity have also been associated with changes in the neural processing of the motivational properties of food. This includes changes in systems coding the hedonic and rewarding aspects of the substance, as well as the systems involved in controlling these motivations [103,192–194]. Volkow and colleagues [195–199] have proposed a common model for addiction and obesity that involves two neural circuits that are both modulated by dopamine—increased reward sensitivity and diminished inhibitory control [70].

### *2.1. Neurobiology of Reward Sensitivity*

Addictive drugs directly affect the mesolimbic dopamine system (MDS), which is thought to mediate the processing of motivational salience, pleasure and reward [200]. Animal studies have shown that, similar to drugs of abuse, palatable foods are capable of triggering dopamine release in the nucleus accumbens (NAc) and ventral tegmental area (VTA) [181,201–203]. Furthermore, activity in the MDS has been linked to the amount of food ingested and its rewarding properties [204,205]. However, distinct patterns of neuronal firing in the NAc to food and illicit substances have also been reported [206,207]. Increased activation of this reward system has also been shown in human participants during the presentation of food cues and meal consumption [96,208–211]. For example, Stoeckel et al. [212] demonstrated that when viewing images of high-calorie foods, obese women showed significantly greater activation in a number of regions associated with reward, compared to healthy-weight women. Obese participants have also demonstrated increased responsivity to food in gustatory and somatosensory regions [213,214], suggesting a heightened sensitivity to palatable food that may contribute to overeating and obesity.

Although an increased sensitivity to reward may initially drive individuals to consume calorific foods, it has been speculated that compulsive eating may develop as the pleasure derived from these foods diminishes with increased tolerance (see Figure 1). It has been argued that, just as with drugs of abuse, the chronic consumption of such rewarding foods may cause the downregulation of dopamine receptors in order to compensate for their overstimulation [215–217]. Decreased striatal dopamine receptor availability has frequently been observed in individuals with substance addictions [218–222], whereas increased receptor availability has been shown to have a protective role against alcoholism [223,224]. It has also been shown that striatal D2 receptor availability is significantly lower in severely obese individuals compared to controls and is significantly and negatively correlated with BMI [99,100].

It has been argued, therefore, that a reduction in dopamine receptor availability may subsequently cause or exacerbate overeating as a form of 'self-medication' in which the individual attempts to compensate for a diminished experience of reward [100,225–227] (see Figure 2). For example, Geiger et al. [228] found that rats fed on a cafeteria-style diet showed reduced baseline levels of mesolimbic dopamine activity. This activity was stimulated by cafeteria foods but not by their regular chow, thus suggesting that a preference for palatable food may develop as a consequence of its ability to increase dopamine release compared to other, less palatable, foods. Animal studies have also demonstrated causal effects of D2 receptor agonists and antagonists on overeating. The administration of D2 antagonists has been shown to increase meal size, meal duration and body weight, whereas treatment with D2 agonists can reduce hyperphagia and prevent weight gain [229–231]. The effects of such pharmaceutical interventions in humans, however, have been fairly mixed. The use of antipsychotic medication which blocks D2 receptors is typically associated with weight gain [232] and some D2 agonists have been found to reduce body weight [233]. A recent trial, however, found no effect of the dopamine agonist cabergoline on preventing weight regain [234,235] and there is some evidence that D2 agonists can promote weight gain in patients with anorexia nervosa [236]. More encouragingly, studies with gastric bypass patients have demonstrated increased D2 receptor availability following weight loss, indicating that the effects of overeating on dopamine receptor downregulation may be reversible [237–239].

**Figure 2.** The proposed cycle of 'food addiction' including the role of dopamine. When palatable food is consumed, the brain releases the hormone dopamine (alongside other neurotransmitters such as opioids). Over time, this increase in dopamine leads to the downregulation of dopamine receptors, causing individuals to experience a reduction in pleasure during palatable food consumption. This decrease in pleasure, combined with symptoms of tolerance, craving, withdrawal and other social, emotional and behavioural difficulties, results in the individual engaging in compensatory behaviour by increasing food consumption. As a consequence, food consumption may become compulsive, thus creating a cycle of food addiction.

### *2.2. Neurobiology of Inhibitory Control*

Dopamine receptor availability in obese individuals has also been shown to correlate positively with metabolism in prefrontal regions involved in inhibitory control (specifically the dorsolateral prefrontal cortex [DLPFC], medial orbitofrontal cortex [mOFC] and anterior cingulate gyrus, as well as the somatosensory cortices) [99]. Similar findings have been observed in healthy-weight participants, who demonstrated a positive correlation between dopamine receptor availability and inhibitory control performance on the stop-signal task [240]. Volkow et al. [99] hypothesised that altered dopamine functioning may play a role in overeating not only through altering the rewarding properties of food but also by reducing inhibitory control. A significant negative correlation between BMI and prefrontal activity has also been reported [75,241,242] along with reduced prefrontal activation following a meal in obese men and women [243–245]. Conversely, successful dieting has been positively associated with frontal activation [246–249].

In a study of healthy women, Lawrence et al. [96] reported an association between food cue reactivity in the NAc and later snack consumption [117]. They also found that this reactivity was associated with increased BMI for individuals who reported low self-control. The authors proposed a 'dual hit' of increased reward motivation and poor self-control in predicting increased food intake [250]. Similarly, reductions in frontal grey matter volume have also been linked to increased BMI, poor food choices and related deficits in executive functioning [251–258]. These findings are reflective of a growing literature on the cognitive dysfunction associated with drug abuse and obesity, although research indicates that the causal relationship is bidirectional [76,259–263].

Although it has been hypothesised that overeating is initially caused by a hyper-responsive reward circuitry and maintained by the subsequent degradation of this system [103], there is also evidence to suggest that some individuals may be genetically vulnerable to an impaired capacity for reward and inhibitory control. Genetics studies have revealed that both drug users and obese individuals have a significantly greater prevalence of the *Taq*I A1 allele polymorphism which can cause a 30%–40% reduction in striatal D2 receptors [213,264–269]. In addition, this polymorphism has been associated with behavioural measures of impulsivity and low reward sensitivity [270–272]. It has also been linked to low grey matter volume in the anterior cingulate cortex (ACC) [273], an area which is believed to be involved in executive control and reward expectancy [240,274,275], and has been shown to be active during resistance of cigarette craving [276]. Together these findings demonstrate that overeating and SUDs may share a common neurobiological mechanism involving altered dopamine functioning that subsequently disrupts mechanisms involved in reward sensitivity and inhibitory control.

Our review, considering each of the DSM-5 criteria for SUDs in isolation, suggests that there is considerable evidence for food addiction. Whether an individual meets clinical diagnostic criteria under an SUD model, and the severity of the disorder, however, is dependent on an individual presenting a number of symptoms (mild: two to three symptoms; moderate: four to five symptoms; severe: six or more symptoms). Studies utilising the YFAS (which uses diagnostic criteria for SUDs) have certainly suggested that a substantial proportion of the general population meet the diagnostic cut-off for food addiction (15%–20%), with approximately 11% of the population being classified as 'severe' [38,276]. The prevalence of food addiction in those with BED and BN has been reported as much higher, with estimates of 92% for BED and 96%–100% for BN [32,277,278]. Acknowledging the potential prevalence of food addiction, we next discuss a range of treatments for overeating that have been informed by the similarities between SUDs and overeating.

### **3. Treatment Implications**

One of the greatest potential advantages of identifying the similarities between substance addictions and overeating is the development of effective interventions. The standard approach to weight loss, involving maintaining a healthy diet and physical exercise, is often associated with poor adherence rates and overall weight gain [46–51,279]. One possible reason for the ineffectiveness of dieting is that it is treating the outcome of overeating and not the underlying cause. Approaches that target increased impulsivity and reduced self-control may have more success. For example, Hall, Fong, Epp and Elias [280] showed that executive function on the go/no-go task (a measure of response inhibition) predicted unique variance for dietary behaviour and physical exercise, and also moderated the association between intentions and behaviour [117,281]. This suggests that individuals who are more capable of controlling their impulsive actions are more likely to successfully meet their goals. This also implies that techniques to improve such abilities may prove to be effective tools for aiding weight loss.

### *3.1. Cognitive Interventions*

Increased motivation for illicit substances has been associated with several cognitive biases including attentional biases [282–287], approach biases [283,284,288–290] and affective biases [291–294]. One method for reducing this motivation, therefore, has been to use training tasks that are designed to reduce these cognitive biases, and recently, these training tasks have been explored as potential interventions for overeating.

Heightened attentional biases towards food have been demonstrated across various populations, including those with disordered eating patterns [295–300] and those who are overweight or obese [301–304]. Just as the addiction literature has explored whether attentional biases can be manipulated to reduce substance intake, this approach has also been explored with food consumption, although with mixed results [305–309]. Hardman et al. [305] trained undergraduate students on the visual probe task to either attend or avoid images of cake and stationery. They found a modest increase in attentional bias for the attend-cake group but no effects of bias training, for either group, on hunger or food consumption, suggesting that any attentional biases with food may be particularly difficult to modify. Using a female-only sample, Kemps et al. [306] demonstrated significant effects of a similar dot-probe training on attentional bias and food consumption. Effects on attention were found to generalise to novel pictures, however, effects on food intake were specific to the trained food and were undermined by participants consuming more of an equally unhealthy novel food. More recent results hold some promise for attentional bias modification, indicating that training can be used to decrease immediate calorie consumption in overweight and obese women [310] and can increase the consumption of healthy foods [311]. Multiple training sessions have also demonstrated that effects can persist beyond the period of training for up to a week [312].

There is also a small body of evidence demonstrating an approach tendency towards food for individuals with disordered eating [295,313,314], high trait food craving [315] and those who are overweight or obese [316–318]. For example, Veenstra and de Jong [313] showed that those who scored highly on a measure of dietary restraint (a measure for the chronic, cognitive limitation of food intake) were significantly faster to move a manikin towards than away from images of food. Using a different measure of approach bias, Kemps, Tiggemann, Martin and Elliott [319] also found that participants who liked chocolate were significantly faster at pairing images of chocolate with approach words compared to avoid words. Furthermore, they also demonstrated that participants who were trained to pair images of chocolate with either approach or avoid words increased and decreased their approach bias, respectively. The approach group also demonstrated a significant increase in chocolate cravings. Although the avoid group showed a decrease in reported craving, this finding was not statistically significant from baseline [320]. Similar training protocols have also been shown to be effective at reducing action tendencies towards high-calorie foods and reducing trait- and cue-induced craving in participants with subclinical eating disorders [321].

The use of addictive substances can also be motivated by the positive affect associated with them; therefore, reducing such an affective bias should discourage their use. Studies using evaluative (or affective) conditioning have shown initial promise. Here, the evaluation of a conditioned stimulus (CS) can be modified by consistently pairing it with a valenced unconditioned stimulus (US) [322,323]. In the food literature, negatively valenced stimuli are typically used to reduce the implicit liking of unhealthy foods [324–328]. Further, evaluative conditioning has been found to lead to more favourable food

choices in some studies. For example, Walsh and Kiviniemi [329] found that participants were three times more likely to select fruit over a granola bar after receiving evaluative conditioning training where positive (relative to negative or neutral) words and images were paired with images of fruit. Similarly, Hollands et al. [325] showed that participants were more likely to select fruit over an unhealthy snack when snack images were repeatedly paired with negative body images compared to a blank screen. Interestingly, this effect was moderated by implicit attitudes towards the snack foods; participants with more favourable attitudes at baseline showed the greatest change in subsequent behaviour. Although these studies have involved healthy participants, this latter finding, in particular, suggests that evaluative conditioning may be an appropriate intervention for those with disordered eating who show strong preferences for unhealthy foods. However, the effects of evaluative conditioning training on reducing unhealthy food consumption are currently unclear. To date, only a limited number of studies have included post-intervention follow-up, and while some have found reduced consumption in the week following training [330], others have failed to show immediate effects [327,328,330]. It is likely that training effects are dependent on baseline strength of liking for specific foods [325,330], the specificity of the US used [331], and awareness of the CS-US contingencies [332–334]. To establish the therapeutic benefits of evaluative conditioning training, studies in overweight and disordered eating groups are required.

Another approach to cognitive training is to reduce such biases indirectly through tasks such as response inhibition training. Response inhibition refers to our ability to interrupt or override impulsive reactions in accordance with new information, and plays a key role in goal-directed behaviour [335–340]. Deficient response inhibition has been linked not only to the use of different addictive substances [70] but also to the severity of use [61,64,341], poor treatment outcomes [342] and likelihood of relapse [343]. Houben and Wiers [344] have also shown that positive implicit attitudes towards alcohol are only related to alcohol consumption when inhibitory control is low. These results suggest that an increased ability to inhibit responses may enable an individual to exert self-control over their behaviour, even when they possess strong implicit preferences [117].

Similar findings have also been replicated with overeating and obesity. Obese individuals have been shown to demonstrate less efficient response inhibition than their healthy-weight counterparts [69,251,345,346] and poor inhibitory control has been associated with increased unhealthy food consumption [86,347,348], high BMI [75,85,349,350], food cravings [351], unhealthy food choices [84,352] and binge-eating [353]. Moreover, as in the addiction literature, inhibitory control has also been shown to interact with implicit attitudes towards food, thus indicating that effective response inhibition may play a protective role against strong implicit preferences for unhealthy foods [80,117,250].

Simple tasks designed to train response inhibition to relevant cues or contexts have been shown to reduce gambling behaviour and alcohol consumption [354–359]—although available evidence suggests that the longevity of such effects may be limited [360]. These training tasks have also been adapted to train response inhibition to food stimuli and are showing encouraging effects across a range of eating-related behaviours including food consumption [361–365], food choices [365–371] and even weight loss [372–375]. For example, Lawrence et al. [376] trained participants to inhibit their responses towards either images of unhealthy snack foods (active group) or non-food items (control group). After four training sessions, they found that, compared to the control group, individuals in the active group showed reduced energy intake (220 kcal less per 24 h food diary) and reduced liking for unhealthy foods. Furthermore, participants in the active group showed significant weight loss; showing objectively measured weight loss of 0.7 kg after 2 weeks and self-reported weight loss of more than two kilograms after a six month follow up (2.66% decrease).

The cognitive training paradigms discussed above show promise but are currently in the early phases of testing. Before such training methods can be taken forward to clinical trials, researchers should further explore the effects of different experimental protocols with the aim of developing the most effective training techniques. One aspect of training that is likely to be important in determining

successful behavior change is training performance. For example, the proportion of successful inhibitions on an inhibition training task and accuracy during attentional bias training have both been shown to moderate efficacy [308,375]. To establish whether effects can be long-lasting, we need to consider repeated testing sessions, personalised training stimuli and combining training techniques to simultaneously reduce cognitive biases and increase executive control [312,369,377,378]. Understanding the mechanisms that underlie such effects could also prove crucial. For example, the effects of inhibition training on food consumption may be due to the devaluation of inhibited stimuli. Several studies have shown that repeatedly pairing a stimulus with the inhibition of a response can reduce how much the image is liked or how attractive it was perceived to be [371,376]. Such devaluation could be the result of action conflict or inherent links between avoidance and aversion [371,379–382]. Designing interventions that promote automatic associations between stimuli and action tendencies may, therefore, prove fruitful, especially if training is performed accurately, personalised and delivered across multiple sessions. Combining cognitive training tasks with prefrontal brain stimulation is another avenue worthy of investigation. Brain stimulation methods have the potential to augment learning effects [383] and can also be used to reduce food consumption and craving in isolation; these methods are discussed below.

### *3.2. Neuromodulation Interventions*

Non-surgical brain stimulation techniques have also been explored for their potential benefits in reducing craving and addictive behaviours by altering neural activity and increasing dopamine [384– 388]. The most commonly applied stimulation methods are transcranial magnetic stimulation (TMS) and transcranial direct current stimulation (tDCS). These methods are used in awake participants and are generally considered to be safe when administered within recommended guidelines [389–394].

The use of TMS involves the delivery of electromagnetic pulses that penetrate the skull to induce electric current in the underlying cortex and cause short-term changes in cortical excitability. The modulation of cortical excitability can last beyond the period of stimulation by delivering trains of pulses, a technique known as repetitive TMS (rTMS) [395]. When applied to the DLPFC, rTMS has been shown to effectively reduce cravings for cigarettes, alcohol and drugs of abuse, especially when applied for multiple sessions [396–400]. The DLPFC is an area involved extensively in inhibitory control [401–404] and stimulation of this region may act to boost self-control, potentially by increasing dopamine release in the caudate nucleus [405,406].

Reductions in substance craving have also been demonstrated with stimulation of the DLPFC using tDCS [114,407–412]. The use of tDCS involves the application of a weak (typically 1–2mA) direct electrical current to the scalp via a pair of electrodes. The effect of tDCS on brain activity is dependent on the stimulation polarity; anodal stimulation is thought to increase cortical excitability by neuronal depolarisation whereas cathodal stimulation is believed to decrease excitability by hyperpolarising neurons [413–418]. Long-lasting effects on resting membrane potential have been shown with longer stimulation durations, for example, 13 min of anodal tDCS has been shown to increase motor cortical excitability for up to 90 min [419]. Compared to TMS, tDCS is a weaker form of stimulation with fewer incidental artefacts and is considered to be safer and more appropriate for reliable double-blinding [387,420,421]. It is also thought that tDCS can be used to potentiate learning [422], and may effectively enhance the effects of the aforementioned cognitive interventions [378,423].

These stimulation methods are currently being investigated for their potential to reduce food craving and consumption [424–429]. Using rTMS to the left DLPFC, Uher et al. [430] found an increase in cue-induced craving for palatable foods in the group who experienced sham stimulation but not the active group. However, no effect was found on ad-libitum food consumption, although this result may have been due to the limited time period (5 min) causing participants in both groups to consume a large amount of calories. Using a similar methodology, Van den Eynde et al. [431] demonstrated an increase in craving scores in the sham group, but a decrease in craving scores for the active group in a sample of participants with bulimic-type eating disorders. In addition, active rTMS was associated

with a reduction in binge-eating episodes in the following 24 h period. However, blinding was only partially successful in this study with most participants correctly guessing whether they were receiving active or sham rTMS. In a later study, Barth et al. [432] used a within-subjects design with an improved sham condition in which they matched the perceived pain of active rTMS with scalp electrodes. They found an equal reduction in cravings for both conditions and attributed this effect to the experience of pain rather than prefrontal stimulation.

As mentioned, tDCS is believed to involve a more appropriately matched sham condition, especially when participants receive active stimulation for a short initial period [420,421]. When stimulating the DLPFC bilaterally using tDCS, Fregni et al. [433] found a significant increase in cue-induced craving, measured before and after stimulation, in the sham condition and a significant reduction when participants received anodal right/cathodal left stimulation. Compared to the sham condition, active stimulation was also associated with a reduction in food intake during an ad-libitum eating phase. Although the authors did not assess blinding in this study, they did report equal occurrences of mild adverse effects across conditions. Using the same montage, Goldman et al. [424] and Lapenta et al. [426] also found the same reduction in food craving, and an early meta-analysis revealed a medium effect-size favouring active over sham stimulation in the reduction of cravings [434]. However, as the number of studies utilizing tDCS in the exploration of its effects on food craving has increased, evidence of efficacy has weakened. A more recent meta-analysis, including eight experiments, found no effect of tDCS on food craving [386], and subsequent research, including a large pre-registered experiment, has also failed to replicate findings for both food craving and consumption [435,436].

Another neuromodulation intervention, which is worthy of a brief mention and gaining in popularity for the treatment of SUDs, is real-time fMRI (rt-fMRI) neurofeedback training. Neurofeedback training involves providing participants with feedback of their neural response to certain cues and instructing them to increase or decrease their response, so that they may gain volitional control over specific brain regions. In the treatment of SUDs, neurofeedback training typically involves increasing activity in control regions, such as the prefrontal cortex, or decreasing activity in regions associated with craving, such as the ACC. For example, it has been shown that decreasing activity in the ACC with rt-fMRI neurofeedback is significantly correlated with decreased nicotine craving in smokers [437–439]. Using a similar technique with electroencephalography (EEG) has also shown improvements in cravings, drug use and treatment outcomes for a range of different substances [440–443]. Although in its early days, the application of neurofeedback training to food consumption and obesity has already been proposed [444–446]; recent findings have also suggested that neurofeedback may be another method of decreasing activity in motivation- and reward-related regions [447] and increasing activity within critical prefrontal regions such as the DLPFC [448,449].

### *3.3. Therapeutic Interventions*

Therapeutic interventions such as Overeaters Anonymous and cognitive behavioural therapy have taken a more holistic approach to the treatment of obesity. Overeaters Anonymous (OA) is based directly on the 12-step programme developed by Alcoholics Anonymous. The OA organisation promotes the central belief that obesity is a symptom of 'compulsive overeating', which is an addictive-like illness with physical, emotional and spiritual components [450]. Individuals are required to acknowledge that compulsive overeating is beyond their willpower to overcome and, therefore, they must attempt to control their intake by avoiding certain foods and surrendering to a 'higher power'. Just like Alcoholics Anonymous, OA involves group meetings for individuals to share their feelings and experiences. Although the way in which this programme influences outcomes is unclear [53], the group meetings may act to alleviate feelings of isolation and instead foster a sense of community. As discussed earlier, due to the feelings of shame and guilt and the weight teasing experienced, overweight and obesity are associated with a preference for isolative activities [159]. This social isolation can subsequently exacerbate overeating, creating a vicious cycle [42,160]. It is possible, therefore, that OA acts to

break this cycle by providing a supportive and encouraging social environment. However, due to the anonymous nature of OA, there has been little research conducted on its efficacy and it is not understood exactly how OA affects overeating and the extent to which it may do so.

Cognitive behavioural therapy (CBT), on the other hand, is a therapeutic approach which is extensively informed by research. CBT requires patients to critically evaluate the thoughts, feelings and behaviours that result in maladaptive behaviour and then modify them through therapy. This therapy allows patients to recognise potential triggers and develop appropriate coping strategies. CBT interventions have been effective in the treatment of substance addictions [451] and have also demonstrated their potential in the treatment of obesity [452,453] and BED [453–456]. However, it has been argued that the success of treating overeating and BED with CBT refutes the food addiction model [30]. The food addiction model applied in OA requires complete avoidance of so-called trigger foods, thereby acting to increase dietary restraint, whereas a reduction in dietary restraint has been shown to moderate the increased effectiveness of CBT on binge eating in a sample of patients with BN [457]. The focus of CBT is to replace dysfunctional eating with more normalised eating behaviour, therefore, favouring moderation and flexibility rather than absolute restraint.

### **4. Conclusions**

As the prevalence of obesity continues to increase and traditional weight loss methods appear to be largely unsuccessful, researchers and clinicians have begun to consider the addictive potential of food. There is a substantial body of evidence demonstrating the similarities between addictive drugs and food on reward and control pathways in the brain and subsequent behaviour such as craving and impulsivity. There is also limited evidence to indicate that in some circumstances, overeating meets the physiological criteria of substance dependence, although more research is necessary to determine the validity of these symptoms in human participants. More research is also required for other behavioural criteria such as social impairment and repeated use despite negative consequences, as the evidence to date is largely anecdotal. However, meeting the physiological criteria for addiction is not necessary for a DSM diagnosis, and as food is a legal substance, just like caffeine, tobacco and alcohol, not all criteria associated with SUDs [23] readily translate to food addiction. Nevertheless, the criterion of withdrawal in SUDs has been associated with clinical severity and the number of symptoms that an individual endorses is used to determine the disorder's overall severity [23].

With a number of these criteria having a limited application to food addiction, a clinical diagnosis appears unlikely in most cases of overeating; however, using the YFAS, it has been estimated that approximately 11% of the general population meet the criteria for a 'severe' food addiction [38]. It should also be made clear that the concept of food addiction does not equate with obesity. Obesity is a multifactorial condition determined by genetic, environmental, biological and behavioural components. For the majority of cases, obesity is caused by a steady increase in excess energy intake and it is not characterised by a compulsive drive for food consumption. Instead, it is thought that the concept of food addiction applies most appropriately to individuals with BED and BN [31,32,277,278].

Despite there being considerable parallels between substance use and compulsive overeating, there is still some concern regarding the use and validity of the term 'food addiction', which is unlikely to apply to the majority of cases [17]. There is also concern over the use of such terminology in the wider social context and whether the term may do more harm than good. While most people would believe that an addiction model reduces individual responsibility, it has also been argued that attributing the problem to a minority of individuals also reduces corporate responsibility [28,458]. As the majority of the population would not be considered 'food addicts', there would be less pressure for the food industry to reduce marketing or to promote healthier alternatives. Likewise, any environmental interventions to reduce access and availability may also seem less critical with a food addiction model.

There are also implications of such terminology for the diagnosed individual. Obesity is already associated with significant social stigmatisation [157–159,161–164] and an additional 'addict' label, which may invoke stereotypes of a person who is untrustworthy and inferior [459], may only serve to

heighten the problem [460–462]. DePierre et al. [460] found that when an individual was labelled as an 'obese food addict' they were more stigmatised than when they received either label in isolation ('food addict' or 'obese'). However, a study investigating the effect of an addiction model on public perceptions found that it actually reduced stigma, blame and perceived psychopathology [463,464], suggesting that it may be beneficial in reducing weight-related prejudice. The 'addicted' individual described in the study was viewed as being less at fault for their weight. Although, it is unclear whether the fault then lies with the individual's biology (i.e., certain individuals are prone to becoming 'food addicts') or the industry that continues to promote potentially addictive foods. Although it is almost certain to be a combination of both entities, demonstrating that certain foods can be addictive should increase corporate responsibility and pressure on the food industry to regulate the availability, advertising and nutritional content of such palatable foods [9,458].

Despite these issues and concerns, it has also been acknowledged that for some individuals, 'food addiction' may be the most appropriate diagnosis for their symptoms and it may help to inform their treatment [34]. The available evidence suggests, therefore, that some individuals *are* capable of experiencing an addictive-type relationship with food, although the majority of individuals who compulsively overeat are unlikely to receive such a diagnosis. Considering the underlying causes of impulsive overeating has also led to the development of some exciting and potentially effective interventions. While there are differences between the addictive characteristics of food and illicit substances, there are many parallels that should not be ignored. These parallels have contributed greatly to our current knowledge of compulsive overeating and potential treatments. Both the similarities and differences should encourage more research, which is necessary to determine the extent and potential impact of such a disorder. Until then, the idea of 'food addiction' is expected to remain hotly debated [14,19,20].

**Author Contributions:** R.C.A. conceived, drafted and finalised the manuscript. J.S., L.M., C.D.C. and N.S.L. made substantial contributions to the manuscript drafts and final approval of the manuscript. All the authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

**Funding:** This research was supported by grants held by C.D.C. from the Biotechnology and Biological Sciences Research Council [BB/K008277/1] and the European Research Council [Consolidator grant 647893 CCT].

**Conflicts of Interest:** The authors declare no competing interests.

### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
