Food in general, and fatty foods in particular, have obtained intrinsic reward value throughout evolution. This reward value results from an interaction between exteroceptive signals from different sensory modalities, interoceptive hunger/satiety signals from the gastrointestinal tract to the brain, as well as ongoing affective and cognitive processes. Further evidence linking food to emotions stems from folk psychology (‘comfort foods’) and epidemiological studies demonstrating high comorbidity rates between disorders of food intake, including obesity, and mood disorders such as depression.
This review paper aims to give an overview of current knowledge on the neurophysiological mechanisms underlying the link between (fatty) foods, their reward value, and emotional responses to (anticipation of) their intake in humans. Firstly, the influence of exteroceptive sensory signals, including visual, olfactory (‘anticipatory food reward’), and gustatory (‘consummatory food reward’), on the encoding of reward value in the (ventral) striatum and of subjective pleasantness in the cingulate and orbitofrontal cortex will be discussed. Differences in these pathways and mechanisms between lean and obese subjects will be highlighted. Secondly, recent studies elucidating the mechanisms of purely interoceptive fatty acid-induced signaling from the gastrointestinal tract to the brain, including the role of gut peptides, will be presented. These studies have demonstrated that such subliminal interoceptive stimuli may impact on hedonic circuits in the brain, and thereby influence the subjective and neural responses to negative emotion induction. This suggests that the effect of foods on mood may even occur independently from their exteroceptive sensory properties.
Aims: To review the human neuroscientific literature on the neurophysiological mechanisms underlying the link between (fatty) foods, their reward value, and emotional impact.
Methodology: Narrative (non-systematic) literature review.
Results & Take Home Message: Fatty foods obtain their intrinsic reward value and are inextricably linked to emotions, not only through exteroceptive sensory pathways (visual, olfactory, gustatory), but likely also through purely interoceptive homeostatic gut-brain signaling pathways impacting on brain reward and hedonic systems. Further research is needed to elucidate the mechanisms underlying the latter influence, particularly the role of gut peptides. This is not only important to increase our understanding of the role of emotions in disorders of food intake such as obesity and eating disorders, but also to elucidate the fundamentally homeostatic nature of emotions and their disorders.
Glossary of Terms
‘Exteroceptive’: pertaining to stimuli that originate from outside the body or to the sensory receptors or pathways on which they impact.
‘Interoceptive’: pertaining to stimuli that originate from within the body or to the receptors or pathways on which they impact.
‘Metabolic hormones’: general term for chemical substances which control various functions of the body's metabolism such as energy intake and expenditure.
‘Gut peptides’: more specific term for hormones produced by the gastrointestinal tract in the presence or absence of nutrients, thereby affecting gastrointestinal function and signaling homeostatic/metabolic information from the gastrointestinal tract to the brain through neural and humoral routes.
‘Affective value of food’: the hedonic quality (pleasure or displeasure) of a food, or in other words, the pleasantness that is experienced when being exposed to a certain food.
Food ‘wanting’: motivation to obtain food reward triggered by exposure to reward-predicting food cues (e.g. sight or smell).
Food ‘liking’: the hedonic reaction or affective response (i.e. pleasure) triggered by consumption of a food.
‘Incentive salience’: a motivational ‘wanting’ for a reward, attributed by the brain to reward-predicting cues. This makes the cue and its associated reward more attractive, thereby motivating to engage in behavior to obtain the reward.
‘Anticipatory food reward’: food reward elicited by exposure to reward-predicting stimuli (e.g. food images or cues announcing real food intake such as sight or smell).
‘Consummatory food reward’: food reward elicited by the actual consumption of food.
‘Gut-brain axis’: the bidirectional neurohumoral communication system between the gastrointestinal tract and the brain, including afferent & efferent autonomic neural pathways and endocrine signaling mechanisms such as the stress hormone system (hypothalamo-pituitary-adrenal axis) and gut peptides secreted by the gastrointestinal tract in response to presence or absence of nutrients.
‘Free fatty acids’: carboxylic acids with long aliphatic tails, which could be either saturated or unsaturated. They are called free fatty acids when they are incorporated in other molecules such as triglycerides or phospholipids.
‘Homeostatic emotions’: According to Craig's and Damasio's theories, emotions are firmly rooted in homeostasis and therefore crucially dependent on homeostatic interoceptive signaling. This led Craig to coin the term ‘homeostatic emotions’.
Why are fatty foods so tasty and why do we crave them, especially when feeling down? The inherent pleasure associated with eating and the undeniable appeal of fat have evolved by natural selection to provide the necessary motivation to acquire adequate nutrient and energy supply in an environment where food is scarce.[1, 2] Moreover, dietary fat is not only the most concentrated source of nutritional energy, it also contains essential fatty acids and fat-soluble vitamins instrumental to growth, development, and general well-being. This explains why the intrinsic reward value and palatability of food as well as the pleasure induced by its intake or even the anticipation thereof are usually linked to its energy density and hence fat content. However, this biological mechanism, favorable for maintenance of homeostasis and reproduction in environments where food sources were limited and unreliable, now seems to be our worst enemy when it comes to resisting the abundance of palatable, energy-dense foods (usually high in fat) present in affluent societies. Furthermore, the increase in consumption of fatty foods is considered to be one of the primary culprits of the current obesity epidemic, with dietary fat accounting for well over 30% of the total calorie intake in a Western diet. Indeed, high-calorie, fatty foods are readily and cheaply available in today's environment, and these are easily overindulged beyond homeostatic needs. This may be explained by an excessive hedonic drive induced by their naturally rewarding properties and hence their ability to reduce stress and negative mood, which are also increasingly prevalent in Western societies. This link is further supported by clinical and epidemiological observations of high overlap between mood disorders (e.g. depression) and disorders of food intake (e.g. obesity), and the folk psychology notion of ‘comfort foods’. Therefore, gaining insight into the neural mechanisms underlying the rewarding properties of fatty foods and the emotional responses they evoke is imperative. This will not only allow better understanding of the maladaptive patterns of eating behavior that are central to the current obesity epidemic, but also of the fundamentally homeostatic nature of emotion and its disorders. In this review article, we will give an overview of recent human research on these neural mechanisms. Firstly, we will discuss how fatty foods obtain their intrinsic reward value and evoke emotional responses through ‘exteroceptive’ (e.g. visual, olfactory and gustatory) sensory signals, which are relatively well established. Less is known, however, about the putative influence of subliminal, purely ‘interoceptive’ fat-induced signals, through gut-brain interactions, on these hedonic responses. Nonetheless, recent studies have started to elucidate how these signals are processed in the brain, including whether and how they may impact on reward and emotional circuits. In the second part of this review, we will discuss these studies in more detail.
Exteroceptive Food-related Signals, Reward, and Emotion
Homeostatic and hedonic control of food intake
Normal eating behavior is controlled by a complex interplay between two parallel signaling systems regulating energy balance and emotional–motivational processes, respectively. The homeostatic system, responsible for energy balance regulation, comprises peripheral endocrine and metabolic signals acting on hypothalamic and brainstem nuclei to modulate energy intake and expenditure according to the body's energy resources and needs (Fig. 1). These homeostatic-metabolic controls of feeding behavior are integrated with a large neural system implicated in hedonic control of food intake that attributes rewarding properties to foods, thereby enabling these foods to generate emotional responses. This brain reward system or hedonic circuit encompasses midbrain regions (ventral tegmental area [VTA] and substantia nigra [SN]) and their dopaminergic projections to the ventral (nucleus accumbens [NAc]) and dorsal striatum (putamen, caudate), as well as the amygdala, orbitofrontal cortex (OFC), anterior cingulate cortex (ACC), and insula[1, 8, 9] (Fig. 1). Integration of interoceptive homeostatic signals reflecting nutritional status, exteroceptive food-related sensory signals, including sight, smell, taste, and texture, and ongoing cognitive–affective processes ultimately determines the rewarding properties and affective value of food, resulting in appetite and eating behavior.[10-12] In general, these homeostatic and hedonic circuits are interconnected and exchange information, but under certain conditions (e.g. stress) or in particular disorders (e.g. obesity, eating disorders, depressvion) this delicate balance may get disrupted.[13-15]
Sensory processing, integration, and affective value of fatty foods
Exteroceptive food-related sensory information, including visual, olfactory, gustatory, and somatosensory signals, is integrated in the brain together with satiety signals (including gastric distension and ‘metabolic hormones’ including ‘gut peptides’ produced by the gastrointestinal tract in response to the presence/absence of nutrients) to build a complex representation of the food, and to evaluate its reward value, regulate feeding behavior, and generate affective responses.[16, 17] Besides food seeking behavior and meal initiation, the attribution of reward and affective value might also influence meal termination, by driving consumption throughout the meal until satiation signals begin to prevail. However, research on the exact impact of the hedonic circuitry on meal termination is rather limited, probably partly because of the fact that both phases of ingestive behavior are inextricably linked. Investigating the potentially distinct impact of hedonic circuitry on meal initiation and termination is therefore difficult and falls beyond the scope of this article, although the topic will be briefly addressed later in this review.
Sensory signals from different receptor cells in the periphery (including, among others, taste and olfactory receptors) activate primary sensory cortices such as the anterior insula/adjoining frontal operculum for taste and the pyriform cortex for smell, where stimulus identity and intensity are encoded into stable representations, independent of hunger or motivational state.[19-22] Information from this first cortical tier is then conveyed to subcortical areas such as the amygdala and hippocampus as well as higher-order cortical areas including the insula and OFC for further multimodal sensory integration, and integration with stored information regarding former experience with different foods, to guide current and future food intake behavior. Obesity is associated with alterations in brain regions involved in sensory processing such as the operculum, insular taste cortex, and OFC. This may render obese subjects more sensitive to food-related sensory stimuli, and hence to the rewarding properties and affective responses generated by them. This might contribute to the risk for overeating in these patients.[24, 25]
The sensory perception of dietary fat is primarily determined by its texture and flavor, thereby making food more palatable and pleasurable; this has become evident due to increasing knowledge on the detection mechanisms for fat in the mouth obtained during the last decade. In addition, some animal work suggests that olfactory cues also contribute to oral fatty acid detection, but evidence in humans is still highly disputable. Furthermore, there is mounting evidence to support the notion of an oral taste or ‘gustatory’ system specific for fatty acids.[27, 28] It appears that dietary fat itself, which is almost completely consumed as triglycerides, is not an effective taste stimulus. Instead, complementary animal and human data indicate that the digestive products of fat, free fatty acids varying in chain length and saturation, act as taste cues. However, this finding has raised some issues regarding the probability of a fat taste in humans, as there is no evidence supporting the secretion of the enzyme lingual lipase, which breaks down triglycerides into fatty acids, in human saliva.[30, 31] However, it is important to note that several psychophysical reports indicate that humans can detect free fatty acids within the millimolar range, which corresponds to the concentrations naturally present in foods. Moreover, several putative fatty acid taste transduction mechanisms including, among others, the fatty acid transporter CD36 and G protein-coupled receptor GPRC120 have recently been identified in the membrane of human lingual papillae.[33, 34] Each of these transduction mechanisms appears to have unique ligand specificity and does not convey information about other sensory cues. However, although these data support the notion for a fat taste mechanism, they cannot be considered definitive. Most evidence is based on animal work and the evidence for the existence and functionality of a fatty acid transduction mechanism in human taste cells is still sparse. At the neural level, it has been shown that oral fat exposure activates the insular taste cortex, mid-insular somatosensory region, OFC, and ACC. The exact sensory properties responsible for this activation pattern are not yet fully understood, as evidence has reported both viscosity-dependent and -independent tactile effects.[35, 36]
Recent neuroimaging studies have begun to unravel some of the neural pathways through which the sensory experience of food in general, and fat in particular, leads to the generation of subjective pleasure. More specifically, studies in primates and humans suggest that portions of the (medial) OFC (perigenual) ACC, and perhaps the (anterior) insular cortex are involved in the encoding of the affective value or pleasantness of food.[18, 37, 38] For example, neurons within the OFC display, as opposed to those in the primary sensory cortices, a selective reduction in their responsiveness to a food which was already eaten to satiety, compared to their responsiveness to the first intake of this food. Moreover, neuronal activity within these regions correlates with subjective pleasantness induced by many different sensory stimuli or modalities including, among others, oral fat taste and texture.[23, 39, 40] In addition, the OFC is a crucial site for integration of all these sensory modalities with various peripheral homeostatic hunger/satiety-related signals (conveyed through the nucleus of the solitary tract, via thalamic and possibly hypothalamic nuclei), which modulate the representation of the ‘affective value’ of the food. Of course, the OFC is not operating on its own, but connects to other cortical and subcortical brain areas, particularly the ACC, perirhinal and entorhinal cortices, as well as the hippocampal formation and the amygdala. It is within these areas that the representation of previous experiences with food is stored and continuously updated with contextual information. The importance of these representations lies in the emotional component that is inherent to them, reflecting the rewarding or pleasurable aspect of the foods that are represented. It is this emotional component that motivates people to engage in behavior to obtain and consume these palatable foods. For example, functional neuroimaging studies in humans have shown that exposure to fat in the mouth activates key reward/hedonic areas, including the insula, anterior cingulate cortex, and amygdala, and this activation may be enhanced in obesity.[35, 41]
The multiple aspects of food reward: ‘wanting’, ‘liking’, and mesolimbic reward pathways
The literature on food reward makes an important distinction between the pleasure derived from eating food and appetitive motivation (incentive motivation or a disposition to eat), or according to the influential theory of Berridge, ‘liking’ and ‘wanting’ of food, respectively.[42, 43] This division between food liking and wanting is primarily based on animal research showing that separate neural mechanisms may underlie the expression of affective responses to food consumption or liking on one hand and the propensity to eat or wanting on the other.
Food ‘wanting’ or, more extremely, craving is often elicited by encounters with (palatable) food and food-related cues, and is highly dependent on meso(cortico)limbic dopaminergic pathways.[42, 43] Dopamine (DA) is generally considered to be the key neurotransmitter implicated in attributing reward value to food, acting mainly through projections from the VTA to the NAc shell, the so-called mesolimbic dopaminergic pathway.[1, 44, 45] In addition, dopaminergic circuits in the dorsal striatum (caudate and putamen), OFC, ACC, and amygdala, as well as the lateral hypothalamus, are also involved in the rewarding aspects of food. The ‘wanting’ hypothesis of reward proposes that, relative to normal conditions when DA circuits maintain a constant DA tone, encounters with reward-predicting cues (e.g. palatable foods or food-related cues) result in increased firing of DA neurons, leading to phasic DA release. This increase in DA signaling will then attribute ‘incentive salience’ to the rewarding stimulus, and potentially induce conditioned learning of the stimulus-reward association, hence motivating the organism to engage in targeted behavior (i.e. eat).[48, 49] It is important to note that, although numerous preclinical and clinical studies have linked feeding and food cues to motivational aspects of the mesolimbic DA reward circuitry, results have been conflicting.[50, 51] Activation of DA systems, as quantified by electrophysiological, microdialysis, or voltammetric measures, is triggered in animals by encounters with food and secondary reinforcers for this incentive. However, positron emission tomography (PET) data related to either DA receptor availability or DA release in both normal-weight and obese people are inconsistent and positive results often rely on the use of pharmacological stimulation of the DAergic transmission system.[53, 54] Thus, the message emerging from PET ligand studies in humans is not as unequivocal as commonly assumed.
Food ‘liking’ or the pleasurable affective response triggered by tasting palatable foods, has been associated with opioid activation in specific limbic forebrain structures or ‘hedonic hotspots’, including the medial shell of the NAc and the posterior ventral pallidum. Interestingly, other neurotransmitter systems within these limbic structures (e.g. the endocannabinoid system and GABA-ergic neurotransmission) may also affect food liking, suggesting a complex interplay of neurotransmitter systems in determining affective responses to food.
Finally, it should be noted that translating the differentiation between these two distinct, though highly entangled, reward components from animal to human research has been problematic, as it has been proven to be very difficult to capture true wanting separately from true liking using rating scales, tasks, and tests.
Aberrant anticipatory and consummatory food reward in obesity
An increasingly influential concept in the scientific community and in the popular media postulates that obesity and overweight are the result of a ‘food addiction’. This would imply that excessive intake of palatable foods can engender similar neuroadaptive changes in brain reward circuitries as chronic drug use does in drug addiction, thereby driving excessive food intake.[18, 50] However, although increasing evidence indicates that obesity and overweight are indeed associated with abnormal neural responses in several key hedonic areas, this is, to our opinion, not yet sufficient to classify all obese patients unequivocally as suffering from food addiction.[50, 58]
Functional brain imaging studies investigating the neural basis underlying (aberrant) eating behavior indicate a dissociation between the reward value of or affective responses to food-related stimuli (‘anticipatory reward’, i.e. food images or cues announcing real food intake) and actual food consumption (‘consummatory reward’).[59, 60] It is important to note that the distinction between anticipatory and consummatory reward is conceptually similar to that between, respectively, wanting and liking described in the previous paragraph, although not identical – discussion of the exact nature of the differences between these concepts falls beyond the scope of this paper. Results from neuroimaging studies in healthy subjects have been largely consistent, showing increased activation in hedonic areas including the dopaminergic midbrain (VTA, substantia nigra), insula, amygdala, striatum (NAc, nucleus caudatus, putamen), ACC, and anterior hippocampus during exposure to food cues (anticipatory food reward), which is accompanied by activation in the OFC and ventromedial prefrontal cortex to assign incentive motivation to food stimuli which in turn leads to feeding behavior.[61, 62] On the other hand, gustatory (anterior and mid insula, frontal operculum) and somatosensory (parietal operculum, Rolandic operculum) cortices were found to be more responsive in obese patients during actual food consumption.[62-64] These results might suggest that distinct brain areas are implicated in anticipatory and consummatory food reward, but there is in fact large overlap (particularly for the OFC and striatum).
Evidence emerging from neuroimaging studies comparing obese and healthy subjects is, however, strikingly inconsistent (Table 1). Nonetheless, most of these studies suggest that obese compared to lean individuals are characterized by increased anticipatory responses to viewing palatable food images or in anticipation of real food reward in key reward/hedonic regions, possibly accompanied by impaired prefrontal metabolism (inhibitory control regions) and increased activity in gustatory and somatosensory regions. This increased hedonic responsiveness during anticipation may lead to increasingly frequent eating, especially when confronted with highly palatable food cues. In contrast, actual food consumption in obesity appears to be associated with reduced reward system activation, particularly in the striatum, which might therefore promote overeating as an attempt to achieve the expected level of reward.[60, 65] As mentioned before and illustrated in Table 1, results vary across studies. It should be noted that there are differences in methodology (e.g. study paradigm, type of food stimulus), subject characteristics (e.g. BMI, age, sex), image acquisition, and type of analysis used in these neuroimaging studies, which could at least partially explain these discrepancies. Even with this in mind, the current lack of a consistent neurobiological profile for obesity may reflect the heterogeneous nature of the disorder, pointing out that it might not be possible to contain the altered brain reward function in obesity in one static model. Indeed, the brain activity pattern implicated in eating behavior is shaped by many different influences, including genetic, biological, and environmental factors, as well as emotions, cognitions, and eating behavior. For example, it is known that obese individuals are more emotionally reactive and report more emotional (over)eating than normal-weight people, although brain imaging studies investigating activation in affective and reward-related areas among emotional eaters are sparse. However, some recent evidence does suggests that emotional eating among lean subjects may be associated with enhanced activity in brain regions implicated in anticipatory and consummatory food reward, although only during a negative emotional state.
Table 1. Overview of functional neuroimaging studies on anticipatory and consummatory food reward in obese patients vs lean controls
Anticipatory food reward
Consummatory food reward
Cues announcing actual food delivery
↑, ↓, and ↔ indicate elevated, reduced, or no difference in brain activation in obese vs healthy subjects, representing significant group differences (↑/↓) or lack thereof (↔) found in functional neuroimaging (fMRI and PET) studies.
Taken together, these results indicate that future neuroimaging studies should take into account psychological (i.e. general emotional state, eating behavior, and comorbid eating disorders) and biological characteristics of participating subjects, rather than merely comparing obese and lean groups.
In this first part, we tried to give an overview of the mechanisms by which exteroceptive signals impact on the reward value of and emotional responses to foods in general and fatty foods in particular. However, there is also increasing evidence that purely interoceptive signals from the gastrointestinal tract in response to the presence or absence of nutrients can directly impact on food reward and emotion, beyond their influence on homeostatic circuits, as will be discussed in the second part of this review.
Homeostatic Interoceptive Fat Signaling, Reward, and Emotion
Subliminal intragastric fatty acid infusion as a model to study nutrient-induced interoceptive gut-brain signaling
The ‘gut-brain axis’ (the neurohumoral communication system between the GI tract and the brain) is an important part of an integrated interoceptive system and continuously conveys homeostatic information from the digestive tract to the brain, including signals induced by gastric distension and/or the sensing of nutrients in the gut, most of which are not consciously perceived.[77, 78] ‘Orexigenic’ gut peptides, such as ghrelin and ‘anorexigenic’ gut peptides including cholecystokinin (CCK), glucagon-like peptides (GLP-1, GLP-2), and peptide YY3-36 (PYY), among others, are released from different parts of the GI tract in response to the presence or absence of nutrients. These gut peptides are crucial players in the regulation of appetite via humoral and neural (through receptors on vagal afferents) gut-brain axis pathways. The reader is referred to excellent recent reviews on this topic for more extensive discussion on the different gut peptides, the mechanisms underlying their release and the pathways by which they convey homeostatic hunger/satiety signals to the brain.
Recent studies have used direct intragastric or intraduodenal infusion of free fatty acids in low doses (not inducing any conscious sensation) as a model to study brain responses to subliminal nutrient-induced interoceptive signals, without any confounding effects of exteroceptive food-related signals (e.g. visual, olfactory, gustatory).[80-82] Triglycerides are the main constituents of fat in our diet, but undigested triglycerides do not induce the release of CCK in humans. However, saturated free fatty acids, the breakdown products of triglycerides, induce CCK release depending on their chain length. Similarly, the effect of fatty acids on other gut peptides differs based on chain length: saturated fatty acids with a chain length of 12 carbon atoms (dodecanoic acid, C12) or more increase plasma levels of CCK, PYY, and GLP-2, and suppress plasma ghrelin levels.[82, 84]
Subliminal intragastric fatty acid infusion induces responses in homeostatic and hedonic brain regions through a CCK-dependent pathway
Cholecystokinin is one of the most important anorexigenic gut peptides, mostly released from the duodenum and jejunum; it delays gastric emptying and causes gallbladder contraction, from which it got its name.[86, 87] Animal studies have shown that ingestion of high-fat nutrients stimulates vagal afferent activation through CCK receptors as this effect was blocked by the administration of CCK receptor antagonists.
Recently, functional magnetic resonance imaging has been used to study brain responses to subliminal intragastric fatty acid infusion in vivo in humans. Lassman et al. observed activation of key homeostatic brain areas such as the hypothalamus, medulla oblongata (including the nucleus of the solitary tract), and pons (including the parabrachial nucleus) in response to a low, non-consciously perceived dose of intragastric dodecanoic acid (250 mL of a solution of 0.05 M) infused over 2 min after an overnight fast. The start of the response in these brain regions was observed within 2–4 min after the start of the infusion, and the activation continued up to 30 min after the infusion started, when scanning stopped. Moreover, the intragastric fatty acid infusion induced a significant increase in plasma CCK levels and the brain responses were abolished by pretreatment with the selective CCK-1 receptor antagonist, dexloxiglumide.[89, 90] Considering that CCK-1 receptors are expressed on vagal afferents and the timing of the brain responses, the intragastric fatty acid infusion most likely activates these brain regions through a CCK-1 receptor-dependent vagal pathway rather than a direct CCK-related humoral route. Interestingly, although not discussed in the article, activation was also found in the caudate head, part of the dorsal striatum, a key reward region. This may indicate that subliminal fatty acid-induced gut-brain signals not only influence the function of homeostatic circuitry, but also of reward-related circuits, a mechanism through which these signals may impact directly on emotions.
Exogenous ghrelin modulates homeostatic and hedonic brain responses to subliminal intragastric fatty acid infusion
Ghrelin, the most important orexigenic hormone, is a 28-amino acid peptide, mainly released from X-cells in the gastric corpus during the interdigestive period, and acts as the endogenous ligand for the growth hormone secretagogue receptor. Jones et al. started continuous intravenous ghrelin infusion (or saline as a control condition) 30 min before the intragastric fatty acid administration, which was performed as in the previously discussed study, after an overnight fast. Firstly, IV ghrelin infusion alone (that is, before the intragastric fatty acid administration) activated regions involved in the homeostatic control of food intake such as hypothalamus, medulla, and pons, but also hedonic regions such as the ventral midbrain (VTA), amygdala/hippocampus, and (anterior) insula. This may be the neural mechanism underlying the orexigenic effects of ghrelin in the fasted state, an interpretation that is further supported by an fMRI study demonstrating that IV ghrelin enhances anticipatory reward responses to exteroceptive food cues (images of high-calorie foods) in the amygdala, OFC, anterior insula, and striatum. Secondly, intragastric fatty acid infusion alone activated homeostatic and some hedonic regions, confirming the earlier results of Lassman et al.,  but this effect was abolished by prior IV ghrelin infusion. Similarly, ghrelin deactivated the same regions in the postprandial state, as shown in a separate experiment by Jones et al. reported in the same paper. Thus, ghrelin attenuates the nutrient-induced signal in homeostatic and hedonic circuits, which may lead to increased food intake to compensate for the attenuated hedonic responses to fat-induced interoceptive signals. Taken together, these findings indicate that ghrelin not only interferes with the function of homeostatic regions, but also with hedonic circuits. This is in line with a growing body of animal evidence demonstrating effects of even peripheral ghrelin injection on reward and emotional circuitry. However, behavioral animal data on the bidirectional link between ghrelin and stress/emotion are ambiguous, as some studies demonstrate a stress-reducing effect of ghrelin, while others show a stress-increasing effect; human literature is completely lacking.
PYY subacutely activates homeostatic and hedonic brain circuits
PYY is a 36-amino acid anorexigenic gut peptide released postprandially from the ileum and colon. Studies have shown that plasma PYY levels raise ~30 min after (subliminal) fatty acid infusion. Although PYY might also be responsible for the fairly rapid brain responses of the intragastric fatty acid infusion in the studies discussed above, due to its local influence on vagal nerve terminals even before any increase in plasma PYY, the peptide probably plays a more important role in delayed brain responses rather than in the early stage, given the fact that the subliminal fat signals in early stage were abolished by CCK receptor antagonists.
It was shown that PYY inhibits food intake in rodents and in humans,[97, 98] but little was known about its role in homeostatic and/or hedonic regulation of food intake until Batterham et al. showed that during IV infusion of PYY, plasma levels covary with activity in homeostatic (hypothalamus, parabrachial nucleus) and hedonic regions (VTA, striatum, OFC). In their study, exogenous PYY or saline was infused intravenously in healthy volunteers in the fasted state. Functional brain images were taken during the infusion, and food consumption was measured afterward. Interestingly, in the PYY infusion condition, activation of hedonic regions such as the OFC predicts 77% of the variance in calorie intake, while homeostatic hypothalamic signals account for less than 1% of this variance. On the contrary, subsequent food intake is correlated with activation of the homeostatic hypothalamus instead of hedonic regions (OFC, VTA) after saline infusion. Thus, the presence of PYY, mimicking the fed state, switches food intake regulation from homeostatic to hedonic.
Interoceptive fat signaling and emotion
It was the American psychologist William James and the Danish physician–psychologist Carl Lange who first took into account the influence of bodily signals on psychological states in general and emotions in particular. The James-Lange theory that bodily changes evoke the feeling of emotions was later challenged by the American physiologist Walter Cannon, who argued that emotions are generated in subcortical brain regions rather than by bodily changes. However, in the past two decades, interest in the James-Lange theory was revived by recent discoveries in neuroscience.
In their similar, though not identical theories, Bud Craig and Antonio Damasio conceptualize emotions as firmly rooted in homeostasis and therefore crucially dependent on homeostatic interoceptive signaling, which led Craig to coin the term ‘homeostatic emotions’. In Craig's point of view, the feelings generated by interoceptive signals, reflecting the survival needs of the human body, are constitutive of emotions. Interoceptive sensory information reflecting the physiological state of the entire body is transmitted along homeostatic afferent fibers through the dedicated lamina I pathway in the spinal cord to the parabrachial nucleus in the brainstem and the thalamus. These are relay stations en route to the insular cortex, where conscious (emotional) sensations (that is, feelings) are generated, and the ACC, which generates the motivational component of the homeostatic emotions. Damasio conceptualizes an emotion as a set of corrective actions in response to internal and/or external sensory stimuli, which is attenuated or enhanced by feedback of the mapped body states, leading to conscious emotional feelings.
On the basis of this conceptual and neuroanatomical framework, Van Oudenhove et al. tested whether changes of interoceptive bodily states triggered by subliminal fat-induced signaling interact with exteroceptively generated conscious emotional states. Either sad or neutral emotion was induced by a combination of listening to validated classical music pieces and viewing of sad or neutral facial expressions. Shortly after the start of the emotion induction, fatty acid was infused into the stomach, using an identical paradigm as the studies[81, 89] discussed earlier. Hunger scores and mood ratings were collected before, during, and after the intervention. The effect of negative emotion induction on both subjective mood ratings and hunger scores was attenuated after fatty acid infusion. Further, key hedonic brain regions, including midbrain, dorsal and ventral striatum, right hippocampus and left ACC, as well as homeostatic areas such as hypothalamus and brainstem, were activated by both fatty acid infusion (confirming earlier results from Lassman et al. and from studies using negative emotion induction). However, a combination of fatty acid infusion and negative emotion induction attenuated the brain responses in these regions. These results were interpreted as an attenuation of the neural and behavioral effects of negative emotion by intragastric fatty acid infusion in healthy humans.
The work by Van Oudenhove et al. suggests that the effect of (fatty) foods on mood may even occur through purely interoceptive mechanisms, independently from their exteroceptive sensory properties. However, the mechanism underlying this interaction remains to be discovered. More specifically, it is unclear whether it is the fatty acid-induced signals that alter the brain responses to negative emotion (‘bottom-up’), or the negative emotion that alters the responses to fat at the level of the brain or even at the level of the gut, as stress/negative emotion has been shown to alter ghrelin levels in animal and ghrelin was found to attenuate the brain response to intragastric fatty acid (‘top-down’). In any case, gut peptides are the prime candidate mediators, but Van Oudenhove et al. did not assess plasma gut peptide levels in their study. Therefore, it would be interesting to study whether and how the levels of different gut peptides would be influenced by emotion induction, subliminal fat infusion, and their interaction to further unravel the mechanisms underlying the effect of subliminal fatty acid-induced interoceptive signals on hedonic circuits and responses. Furthermore, most of the human studies on fat signaling and emotion are limited to healthy volunteers. As these mechanisms are potentially highly relevant for obesity, eating disorders, functional dyspepsia, and depression, further studies in these patient groups are warranted.
Evolution shaped the intrinsically rewarding properties of fatty foods, and thereby their ability to evoke pleasurable emotional responses. In this review paper, we provided an overview of recent neuroscientific research on the neurophysiological mechanisms determining the reward value of and affective responses to fatty foods. The neural pathways through which exteroceptive food-related signals (visual, olfactory, gustatory) impact on the encoding of reward value in the (ventral) striatum and subjective ‘food pleasantness’ in the cingulate and orbitofrontal cortex are relatively well established. On the other hand, however, results of functional neuroimaging studies comparing food reward responses between obese and lean subjects are somewhat heterogeneous. Nonetheless, most data suggest an increased anticipatory food reward in combination with a blunted consummatory reward response in obesity, potentially leading to overeating as a compensatory mechanism to achieve the expected level of reward. Furthermore, recent evidence suggests a potentially important impact from purely interoceptive ‘homeostatic’ gut-brain signals on reward and emotion which occurs independently of exteroceptive sensory properties such as taste. More specifically, (an)orexigenic gut peptides may not only evoke responses in circuits involved in homeostatic control of food intake, such as the hypothalamus, but also influence functioning of hedonic circuits, through neural (i.e. vagal afferents) and potentially humoral routes. However, more research is needed to unravel the complex underlying mechanisms. Taken together, the emerging evidence on the role of both exteroceptive and interoceptive pathways is not only important to increase our understanding of the role of emotions in disorders of food intake such as obesity and eating disorders, but also to understand the fundamentally homeostatic nature of emotions and their disorders.
Lukas Van Oudenhove is a research professor of the KU Leuven Special Research Fund (Bijzonder Onderzoeksfonds, BOF). Nathalie Weltens is funded by a research grant from the Research Foundation-Flanders (FWO-Vlaanderen) to Jan Tack & Lukas Van Oudenhove. Dongxing Zhao is funded by a starting grant of the KU Leuven Special Research Fund (BOF) to Lukas Van Oudenhove.