Food-Cue Reward Processing

Understanding the neural underpinnings of hedonic processes is particularly relevant for BED, as the overconsumption of high-fat and high-sugar foods during binges suggests alterations in reward sensitivity in this population. To date, most neuroimaging studies in BED examine food-cue reactivity; neural responses are investigated as individuals are exposed to palatable food stimuli in the scanner (Table 1). The first neuroimaging study in BED applied SPECT in 8 females with BED, and also included 2 control groups: an obese non-BED group and a lean control (LC) group.Reference Karhunen, Vanninen, Kuikka, Lappalainen, Tiihonen and Uusitupa ¹⁸ Relative to both of these groups, a food-exposure task produced greater regional cerebral blood flow (rCBF) to frontal and prefrontal regions in the BED group. Additionally, this prefrontal activity was linked to increased hunger feelings in the BED group, but not in the control groups. Consistent with the SPECT findings, an fMRI studyReference Geliebter, Ladell, Logan, Schneider, Sharafi and Hirsch ¹⁷ also reported increased prefrontal activation to food stimuli in obese females with BED. This study was also one of the first to distinguish between lean and obese individuals with BED. Notably, lean females with BED did not show any significant prefrontal differences relative to the control groups. While these results were obtained in a very small sample (n=5 per group) and are still preliminary, they nonetheless hint at activation differences related to conjoint obesity and binge-eating status.

A food-cue stimuli presentation during fMRI by Schienle et al Reference Schienle, Schafer, Hermann and Vaitl ¹⁰ also reported increased prefrontal activity; food pictures elicited significantly greater medial orbitofrontal cortex (OFC) activity in the BED group. Notably, contrasts were performed relative to both lean and overweight control groups, but also to a bulimia nervosa (BN) group (purging type), with a similar degree of bingeing and disorder duration. Not only did BED individuals report significantly greater reward sensitivity, but this measure correlated positively with medial OFC activity, further supporting the idea of increased sensitivity to food reward in the BED group. The OFC constitutes a secondary taste cortex,Reference Rolls, Yaxley and Sienkiewicz ^{21 ,} Reference Baylis, Rolls and Baylis ²² but is also part of an extensive system encoding subjective values of a variety of rewards.Reference Levy and Glimcher ²³ Increased OFC recruitment suggests alterations in value representation; this is further supported and linked to correlations with reward sensitivity. Structural differences are also observed: increased gray-matter volume is reported in BED relative to LC groups, particularly in medial OFC and anterior cingulate areas.Reference Schäfer, Vaitl and Schienle ²⁰ Given the importance of the OFC in guiding choice behavior, misrepresentations of value signals could have detrimental effects on decision-making processes.

Few neuroimaging studies to date have examined negative valence processing in BED individuals. However, the Schienle et al study specifically examined the neural substrates in response to disgust pictures; BED individuals showed significantly reduced activity in OFC and insula areas relative to LC participants.Reference Schienle, Schafer, Hermann and Vaitl ¹⁰ Although valence ratings did not differ between groups, reduced neural responses in insular and lateral OFC areas suggest, among other possibilities, potential alterations in disgust responsiveness in the BED group.Reference Schienle, Schafer, Hermann and Vaitl ¹⁰ Examining responses to negative valence stimuli is particularly relevant to binge-eating syndromes, where responses to aversive qualities of food or satiety signals may be altered. An important future direction will be to clarify how eating restraint relates to appetitive and non-appetitive stimuli.

Findings of OFC alterations in BED are consistent with the role of this brain area in coding for the subjective motivational value of reinforcers, including food (for reviews see Kringelbach,Reference Kringelbach ²⁴ Peters and Buchel,Reference Peters and Buchel ²⁵ and RollsReference Rolls ²⁶ ). Multiple fMRI studies demonstrate how OFC activity increases in response to an appetitive stimulus, and decreases as the stimulus becomes less rewarding or aversive (for example, when eating chocolate beyond satietyReference Small, Zatorre, Dagher, Evans and Jones-Gotman ^{27 ,} Reference Breiter, Aharon, Kahneman, Dale and Shizgal ²⁸ ). Some research also differentiates further localization of function within different OFC subregions, with reward value coded in medial areas and negative or punishing stimuli signaled in more lateral areas.Reference Kringelbach and Rolls ²⁹ By processing salience attribution and the relative reward value of a reinforcer, the OFC contributes importantly to decision-making and guiding goal-directed behavior. In this way, alterations in OFC signaling could have significant influences on choice behavior.

Actual consumption of hyperpalatable foods in the scanning environment remains difficult and has not yet been directly examined in a BED population. However, in a recent studyReference Filbey, Myers and Dewitt ¹² in compulsive overeaters (as assessed by the Binge-Eating ScaleReference Gormally, Black, Daston and Rardin ³⁰ ), tasting food provides consistent findings with those demonstrated to pictorial food cues. The receipt of high-calorie taste cues (such as chocolate milk) on the tongue also produces greater responses in OFC, striatal, and insula regions in compulsive overeaters relative to tasting water.Reference Filbey, Myers and Dewitt ¹² Analyses demonstrated how connectivity between the ventral striatum and other reward areas appeared stronger during high-calorie tastes versus water; moreover, this relationship was stronger with increasing binge-eating scores. As this study did not include a control group, this finding may simply represent the response between palatable versus neutral tastes. Nonetheless, this study represents an important direction in mechanistic investigations related to food-reward processing. Understanding basic associative learning mechanisms underlying food-reward pairing has implications for identifying therapeutic targets. For example, if high-calorie tastes alter connectivity in reward neurocircuitry in some overeaters, interventions might focus on limiting intake of such foods, particularly in those at risk for binge eating or obesity, including children, whose reward neurocircuitry is still developing. With increased knowledge of the underlying neurobiology, pharmacological interventions might target neural systems involved in reward-related learning. More broadly, public health campaigns might educate the public about neurological tendencies and potentially reduce stigma around these conditions.Reference Carnell, Gibson, Benson, Ochner and Geliebter ³¹

A recent study further applied a classification analysis to data from a 2009 studyReference Schienle, Schafer, Hermann and Vaitl ¹⁰ in which BED, obese non-BED (OB), BN, and healthy control (HC) participants viewed food, disgust, and neutral pictures during fMRI. The reanalysis demonstrates how neural correlates during food-cue processing might be used to discriminate between BED, BN, and non-disordered obese groups.Reference Weygandt, Schaefer, Schienle and Haynes ¹¹ Regions of interest (ROIs) included the anterior cingulate cortex (ACC), OFC, amygdala, insula, and striatum. Activity in insular, striatal, ACC, and OFC areas correctly classified participant groups with a decoding accuracy of around 70% in these areas. Of note, the ventral striatum provided the best separation between the BED group and the obese and BN groups, albeit on different sides of the brain. Thus, neural information encoded during food-cue processing may be used to discriminate between clinical conditions, thereby further supporting the diagnostic autonomy between different types of disordered eating, including BED. Notably, clinical condition for the 4 different groups (BED, OB, BN, and HC) could be decoded from reward-processing regions, particularly those implicated in motivational signaling during food-cue processing. This first study applying classification analyses in BED demonstrates a data-driven approach in which brain response patterns may be used not only to study underlying physiological disturbances but also to potentially characterize and diagnose specific psychiatric conditions.

In sum, food-cue studies provide evidence linking positive affective food-cue responses with prefrontal activity, in particular with OFC recruitment. Relationships between heightened responsiveness in the BED group (but not observed in other populations) with hunger and reward sensitivity measures support this area as a motivational marker of eating pathology in this group.

To date only one study has applied PET to examine specific neurotransmitter systems in BED. Wang et al Reference Wang, Geliebter and Volkow ¹⁹ conducted a [¹¹C]raclopride scan investigating dopaminergic functioning with a therapeutic dose (20 mg) of methylphenidate (MPH) in obese individuals with and without BED. This drug has previously been shown to increase striatal dopamine (DA) release in HC participants during food stimulation; therefore, MPH may be used to gauge DA alterations during food stimulation across OB and BED participants. A food stimulation task (including both olfactory and gustatory cues) produced significantly increased extracellular DA levels in the caudate nucleus in BED individuals, relative to a non-BED obese group. In the BED group, caudate activity further correlated with higher binge-eating scores, but not body mass index (BMI), which was matched across groups. This result suggests a relationship between DA systems and eating pathology. Given the importance of the dorsal striatum in motivation and habit formation, this relationship between DA levels and binge-eating pathology is suggestive of this neurotransmitter’s role in coding for motivational, rather than consummatory, properties of food reward. This relationship is also consistent with the positive relationship observed between OFC activity and reward sensitivity scores during a food-cue fMRI studyReference Schienle, Schafer, Hermann and Vaitl ¹⁰ ; this prefrontal reward–sensitivity relationship with food cues could further reflect ensuing effects from DA striatal activation.Reference Wang, Geliebter and Volkow ¹⁹ While ventral striatal activity is attributed a role in reward prediction,Reference Knutson, Adams, Fong and Hommer ³² more dorsal striatal areas are implicated in habit formation and automatic behaviors.Reference Vanderschuren, Di Ciano and Everitt ³³ Thus, it would be of interest to examine if a similar relationship occurs in lean BED individuals, or those experiencing escalation in bingeing. Nonetheless, these findings demonstrate how BED and non-BED obese groups may demonstrate distinct patterns of dopaminergic transmission with caudate function related to BED pathophysiology.

Generalized Reward Processing

To date, only one fMRI study has specifically examined non-food reward processing using the monetary incentive delay task (MIDT).Reference Balodis, Kober and Worhunsky ¹³ Examining cognitive mechanisms beyond food cues represents an important area in BED research; alterations in basic cognitive processing (eg, generalized reward processing) may relate to vulnerability and maintenance factors in BED (see Table 2 for summary). The MIDT employs monetary rewards, rather than food-cue rewards, to parse anticipatory from outcome phases of reward. Understanding anticipatory–outcome distinctions is particularly relevant to obesity research, as anticipatory processing may relate particularly to food intake.Reference Epstein and Leddy ³⁴ On the MIDT, anticipatory processing distinguished obese BED from non-BED obese groups with decreases in the ventral striatum noted in the BED group, versus increased recruitment in the non-BED obese group. Divergent striatal recruitment during reward processing between BED and non-BED obese groups is consistent with ensemble coding findings reported by Weygandt et al,Reference Weygandt, Schaefer, Schienle and Haynes ¹¹ who found that the left ventral striatum provided the best differential diagnostic separation between these 2 groups. These findings lend further support to the idea of the ventral striatum playing an important role in the pathophysiology of the disorder, given the critical role of this brain region in goal-directed behaviors and affective state.Reference Carlezon and Wise ^{35 –} Reference Ito, Robbins and Everitt ³⁷ These results are also consistent with blunted anticipatory processing reported in other disorders, which is characterized by problems of self-regulation, including alcohol dependence,Reference Beck, Schlagenhauf and Wustenberg ³⁸ pathological gambling,Reference Balodis, Kober, Worhunsky, Stevens, Pearlson and Potenza ³⁹ and attention-deficit hyperactivity disorder.Reference Strohle, Stoy and Wrase ⁴⁰

Outcome processing on the MIDT demonstrated generalized hyporesponsiveness to non-food cues in the BED group; relative to non-BED obese and LC groups, outcome processing produced diminished OFC and insula activation.Reference Balodis, Molina and Kober ¹⁴ Similar blunted prefrontal and insular activity has previously been noted during palatable food consumption in BN.Reference Bohon and Stice ⁴¹ It is also noteworthy that patients with fronto-temporal dementia, a neurodegenerative disease resulting in atrophy patterns in the striatum, as well as frontal, insular, and temporal cortices, often develop compulsive overeating.Reference Woolley, Gorno-Tempini and Seeley ⁴²

Overall, this first study examining monetary reward processing in BED demonstrated diminished fronto-striatal processing of rewards and losses during both anticipatory and outcome processing, specifically in areas relevant to reward processing and self-regulation. Similar patterns of activation to monetary cues of both wins and losses suggests that fronto-striatal signaling is less valenced in BED, relative to the other comparison groups, although more study of negative valence processing is necessary. Hypofunctioning of frontostriatal circuitry in this population may represent a neural precursor contributing to the development of BED, where an individual may overeat to stimulate a sluggish reward system. Alternatively, patterns of food exposure may lead to changes such as those observed in BED. The differences in OFC and insular areas noted in contrasts between both LC and obese groups suggest alterations in interoceptive awareness, given the important role of these areas in homeostasis and in updating on the motivational state of an organism,Reference Small ^{43 –} Reference Paulus, Rogalsky, Simmons, Feinstein and Stein ⁴⁵ although this possibility warrants further direct examination.

Taken together, findings suggest in BED heightened activation to food reward in reward neurocircuitry, but a decreased response to generalized (ie, non-food or specifically monetary) reward. Although direct comparison between these 2 types of reinforcers is still necessary, these early studies lend support to the idea that a reduced response to generalized rewards may represent a vulnerability factor to consume palatable foods in an effort to stimulate a reward system.

Inhibitory Control

A better understanding of the neural underpinnings of inhibition is particularly relevant to BED studies, given difficulties in this population in controlling food intake. Although no imaging study has specifically examined inhibitory processing in relation to food cues or intake in BED, one study has examined generalized cognitive control using the Stroop color-word interference task during fMRI.Reference Balodis, Molina and Kober ¹⁴ Relative to both a BMI-matched non-BED obese group and a LC group, the BED group showed reduced activity in the OFC, inferior frontal gyrus (IFG), insula, and temporal areas. Activity differences specifically appeared to be driven by the BED group that demonstrated reduced recruitment of these areas during incongruent trials. Measures of eating restraint also demonstrated a differential pattern of correlations with Stroop performance across the 3 experimental groups. Restraint scores in the BED group correlated negatively with OFC, insula, and IFG activity—brain areas heavily implicated in self-regulation, inhibition, and homeostatic regulation. Notably, these areas were also identified during disgust processing in the study by Schienle et al Reference Schienle, Schafer, Hermann and Vaitl ¹⁰ ; as such, these regions may contribute importantly to multiple facets of BED.

Conversely, Stroop performance in the non-BED obese group demonstrated a positive correlation between restraint scores and increased IFG and insula recruitment. Opposite correlational patterns across the BED and non-BED obese groups suggest that these groups may differ in both their restraint applications and the neural mechanisms underlying them.Reference Balodis, Molina and Kober ¹⁴ Given the role of the IFG, OFC, and insula in self-regulation, these findings intimate that BED individuals may be impaired in recruiting brain areas critical for inhibitory control. A better understanding of neural underpinnings of cognitive control in BED is important, as the choice to diet is cognitively mediated and involves maintaining long-term goals in mind while repeatedly discounting more proximal food cues.

Neuroimaging and BED Treatments

Linking neuroimaging with treatment outcomes in BED provides a means to examine mechanisms of change and recovery processes. A better understanding of BED pathophysiology could potentially guide the development or refinement of therapeutic methods. Applying neuroimaging to identify neurobiological factors linked to treatment response has only just begun in BED. A pilot study that examining generalized reward neurocircuitry recruitment related hypofunctioning frontostriatal areas to treatment outcome.Reference Balodis, Grilo and Kober ¹⁶ Prior to commencing treatment, BED participants completed the MIDT, which examines anticipatory–outcome monetary reward processing while undergoing fMRI. Individuals who still reported bingeing following treatment demonstrated reduced striatal and IFG recruitment during anticipatory reward processing,Reference Balodis, Grilo and Kober ¹⁶ relative to individuals who had stopped binge eating. This is consistent with other findings that have related reduced striatal response to food cues with weight gain.Reference Bohon and Stice ^{41 ,} Reference Pelchat, Johnson, Chan, Valdez and Ragland ⁴⁶ Importantly, individuals who ceased or persisted in binge eating did not differ in BMI or binge frequency at treatment onset. Therefore, this initial pilot study demonstrates how specific reward processing regions may provide therapeutic targets in the future. For example, IFG recruitment while viewing palatable food cues has previously been linked to sustained weight loss.Reference McCaffery, Haley and Sweet ⁴⁷ During outcome win processing, individuals who persisted in binge eating also showed reduced medial prefrontal cortex (mPFC) recruitment—an area linked to processing monetary reward outcomes, emotional arousal, and decision-making.Reference Haber and Knutson ^{36 ,} Reference Stice, Spoor, Bohon, Veldhuizen and Small ^{48 –} Reference Chambers, Taylor and Potenza ⁵⁰ Altogether, these findings suggest that reduced reward circuitry recruitment is associated with persistent bingeing in BED. The striatum and prefrontal areas are projection areas for DAReference Fiorillo, Tobler and Schultz ^{51 ,} Reference Robbins ⁵² ; to date, however, no study has specifically examined dopaminergic alterations in relation to BED treatment.

One of the first pharmacological neuroimaging studiesReference Cambridge, Ziauddeen and Nathan ¹⁵ examined actions of an opioid antagonist on food-cue responsivity in obese individuals with moderate binge-eating symptoms. While selectively blocking mu-opioid receptors, the antagonist GSK1521498 reduced high-fat and high-sugar food intake.Reference Chamberlain, Mogg and Bradley ^{53 ,} Reference Ziauddeen, Chamberlain and Nathan ⁵⁴ Using a double-blind, placebo-controlled, parallel-group design, this antagonist reduced activity in pallidum-putamen areas as individuals viewed highly palatable food-cues, without affecting subjective liking of the cues. The therapeutic efficacy of this drug may link to motivation-hedonic distinctions previously mentioned; the opioid-receptor antagonist may reduce motivation for food while leaving the subjective reward value of food unaffected. In particular, the pallidum/putamen is highlighted as an opioid hedonic hotspot for reward,Reference Castro and Berridge ⁵⁵ which highlights the motivation–hedonic relationship. These early neuroimaging studies therefore demonstrate evidence for divergent neural systems related to motivational and hedonic systems, and that targeted treatments may be possible and effective for BED.

Future Directions and Clinical Implications

To date, neuroimaging studies in BED have included multiple control groups, including BMI-matched non-BED obese individuals, non-BED binge-eating groups (eg, BN) with comparable degree of binge-eating frequency and disorder duration, and LC groups. Nonetheless, the majority of neuroimaging studies to date are predominantly in females; therefore, future studies with larger groups could examine potential gender-related differences. Additionally, most studies have only used cross-sectional designs, making it difficult to disentangle causes and consequences. Longitudinal studies are needed to investigate these processes and how specific factors (eg, increasing weight or escalating binge frequencies) may relate to neurobiological features. More generally, it will be important to understand the neural substrates underlying processes as eating behaviors shift from pleasurable to more compulsive. While multiple investigations now demonstrate alterations in IFG areas in BED, few studies have examined the development of aversive states and how negative valence relates to inhibition and restraint in this population. Nonetheless, it is noteworthy that frontostriatal associations with motivational measures often occur in the BED group (eg, reward sensitivity, hunger, or bingeing), rather than in non-BED groups, and support the idea of alterations here as motivational markers of pathology in BED.

The findings highlighted in this review give insight into potential biomarkers in striatal and OFC areas in BED. While dopaminergic projection sites suggest potential clinical targets for this neurotransmitter, pharmacological neuroimaging studies are only just beginning. Anticipatory–hedonic distinctions identified in neuroimaging research already demonstrate how targeting motivational processes may prove to be critical in the treatment of BED and might eventually serve to inform or refine intervention methods. These specific neurobiological alterations may prove central in understanding the mechanisms and guiding targeted treatments for BED.

Disclosures

Dr. Iris Balodis has nothing to disclose. Dr. Grilo has the following disclosures: Shire, consultant/advisor, speaker bureau, consulting fees; Sunovion, consultant/advisor, consulting fees; American Psychological Association, editor, honoraria; Guilford Press, author, book royalty; Taylor & Francis, author, book royalty; NIH-NIDDK, principal investigator, grants to Yale University; NIH-NIDDK, principal investigator, grant K24 DKO70052; CASA Columbia, senior scientist, percent salary; CME Entities, lecturer, honoraria; Academic Entities, lecturer, honoraria. Dr. Potenza has the following disclosures: Shire, consultant, consulting fees; INSYS, consultant, consulting fees; NCRG, researcher, grant to university; NIH, researcher, grants to university; Mohegan Sun Casino, unrestricted gift to university; Gambling Entities, consultant, consulting fees; Legal Entities, consultant, consulting fees; CT DMHAS/The Connection, psychiatric consultant, consulting fees; CT DMHAS, researcher, gambling research support for CMHC; Academic Entities, lecturer, honoraria; Publishers, editor, honoraria and royalties; Grant Agencies, reviewer, payment.