Brain principles of eating

While food clearly is essential to survival, it is the pleasure involved that makes eating worthwhile. While the members of the religious sect in Blixen’s novella may try hard to deny the pursuit of pleasure in its many forms, their well-being is ultimately strongly enhanced as they submit to Babette’s cooking, i.e. to the strong primal drive for pleasure. The evolutionary imperatives of survival and procreation are not possible without the principle of pleasure for the fundamental rewards of food, sex and conspecifics—and as such may well be evolution’s boldest trick [3]. The scientific study of pleasure, hedonia research, is dedicated to searching for the functional neuroanatomy of hedonic processing, taking its name from the ancient Greek for pleasure (ἡδονή; transl. hédoné) derived from the word for “sweet” (ἡδύς, transl. hēdús) [4].

In the novella, the sect’s initial food asceticism may stem from their religious beliefs but is guided by the basic homeostatic regulation of human eating behaviour [5], of which animal models have elucidated in great details the many subcortical circuits and molecules shared amongst mammals including humans [6-8]. Yet, as illustrated by the effects of Babette’s Feast, homeostatic processes are not solely responsible for human eating. This hedonic eating is difficult to suppress and is even more poignantly illustrated by the current worldwide obesity pandemic [9]. There is often very little well-being linked to such over-eating, with anhedonia—the lack of pleasure—being a prominent feature of affective disorders. From this public health perspective, it is imperative that we better understand the fundamental pleasure systems such that we find new and more effective ways of re-balancing the system and potentially reducing obesity which is threatening to undermine public health [10].

Eating can seem simple but at its most basic, human food intake is still rather complex. The procurement of food can be surprisingly difficult in the wide variety of often hostile climates inhabited by humans. Once food is available, the preparation and eating of food are also complex processes, involving a multitude of peripheral and central processes for carefully orchestrated acts requiring significant brain processing. The necessary, sophisticated motivational, emotional and cognitive processing are likely to have been main drivers for the evolution of large primate brains [11]. The brain principles underlying eating have been investigated for a long time in many mammalian species [6,12]. Here, the focus is on the pleasure component of human eating, which over the last decade has started to transform our understanding [13,14].

To understand pleasure in the brain, it is important to consider the main challenge for the brain which is to successfully balance resource allocation for survival and procreation [15]. In order to achieve this balance, different rewards compete for resources over time. In understanding the multi-faceted nature of pleasure, it can therefore be useful to consider the typical cyclical time course shared between all rewards with distinct appetitive, consummation and satiety phases [16,17] (Figure 1). The research has demonstrated that pleasure consists of multiple brain networks and processes and involves a composite of several components: “liking” (the core reactions to hedonic impact), “wanting” (motivational processing of incentive salience) and learning (typically Pavlovian or instrumental associations and cognitive representations) [18-21]. These component processes have discriminable neural mechanisms, which wax and wane during the cycle. The neural mechanisms of wanting, liking and learning can occur at any time during the pleasure cycle, though wanting processes tend to dominate the appetitive phase (and are primarily associated with the neurotransmitter dopamine), while liking processes dominate the consummatory phase (and are associated with opioids) [13]. In contrast, learning can happen throughout the cycle (and is thought to be associated with synaptic plasticity). A neuroscience of pleasure seeks to map the necessary and sufficient pleasure networks allowing potentially sparse brain resources to be allocated for survival.

This basic cyclical model of pleasure can be expanded into an elaborate multilevel model of food intake taken in account the episodic and tonic changes over time (Figure 2) [12]. The model links the pleasure cycle with the cyclical changes in hunger levels related to the initiation and termination of meals and the way food intake comes about through the interaction given signals from the body, e.g. from the brain, gut-brain, oral cavity, stomach and intestines, liver and metabolites and body mass.

The dual processes of satiation and satiety are central to the model and to the energy obtained by the associated meals [22]. Terminating eating is complex process, which is encapsulated by satiation [23], while satiety is the feeling of fullness that persists after eating to suppress further eating. These processes are controlled by a cascade of sensory, cognitive, post-ingestion and post-absorptive signals, beginning with the consumption of a food in a meal and continuing as the food is digested and absorbed.

The multilevel model of food intake describes the changes over time in A) pleasure, B) the levels of hunger, C) satiation/satiety cascade signals, D) origin of signals and signal carriers, E) brain processes, F) behavioural changes including those in the digestive system and G) general modulatory factors (Figure 2). Many of these changes have been described elsewhere, e.g. the mechanisms of the changes after the termination of a meal such as the gut-brain interactions, include signals from receptors in the digestive tract which are sensitive to calorie-rich nutrients (even in the absence of taste receptors) [24,25].

Here, however, the focus is on the processing principles involved primarily in the initiation and termination of a meal (Figure 3). The multisensory experience of food intake involves all the senses with different routes into the brain; from the distant processing of sight, sound and tactile of food to more proximal smell, taste and tactile (mouth-feel) processing. Smell is the most important determinant of the flavour of food and comes to the brain via orthonasal and retronasal pathways, experienced as we breathe in and out, respectively [26]. As demonstrated by the case with coffee, the subjective olfactory experience can feel very different from smelling the coffee in the cup to tasting the coffee in the mouth, which also relies on pure tastants (such as bitter) and mouth feel factors (such as the smoothness of the crema) (Figure 3A).

This sensory information about food is coming from receptors in the body, typically the eyes, ears, nose and oral cavity and gets processed in the primary sensory cortices of the brain. The topology of these regions are remarkably similar between people with vision (red) always processed in the back of the brain, audition (dark blue) processed in regions of the temporal cortex, touch (light blue) in somatosensory regions and olfaction (orange) and taste (yellow) in frontal regions (Figure 3B). Importantly, unlike the other senses, olfactory processing is not processed via the thalamus which may explain the hedonic potency of odours [27]. Note that it is important that we are able to identify a food stimulus independently of whether we are hungry or sated, and accordingly, sensory information in primary sensory cortices is remarkably stable and not modulated by motivational state.

The sensory information is further integrated in multisensory areas before it is evaluated for reward value in the pleasure system. Here, the processing depends on prior memories, expectations and state and may give rise to brain activity which gives rise involuntary pleasure-evoked behaviour (such as licking of lips or soft moaning) and, at least in humans, subjective pleasure (Figure 3C).

Neuroscience has started to map the pleasure system in many species. This has been shown to include a number of important regions such as pleasure hotspot regions in subcortical areas of the brain such as the nucleus accumbens and ventral pallidum [28,29]. Manipulations of these regions with opioids have been shown to causally change pleasure-elicited reactions [13]. Other regions involved in pleasure have been found using human neuroimaging in the orbitofrontal, cingulate, medial prefrontal and insular cortices [30-37]. The pleasure system does not act in splendid isolation but is of course embedded within much larger brain networks. We are beginning to understand the metastable nature as well as the topological and functional features of these networks using advances in network science and graph theory together with advanced whole-brain computational models [38,39].

Computational processing principles for eating

Overall, eating has been demonstrated to rely on at least five fundamental processing principles: 1) hunger and attentional processing; 2) motivation-independent discriminative processing of identity and intensity; 3) learning-dependent multisensory representations; 4) reward representations of valence and 5) representations of hedonic experience [12,40]. In the following, these are briefly described.

Motivation-independent processing of identity and intensity

It is vital that reliable sensory food information is provided for the brain to guide ingestion decision-making. Eating has to be controlled very carefully since erroneous evaluation of the sensory properties of foods can potentially be fatal if ingesting toxins, microorganisms or non-food objects. Mammals have been shown to have brainstem reflexes (stereotypical for each basic taste) that are based on rudimentary analyses of the chemical composition, and which are not altered, even by the loss of all neural tissue above the level of the midbrain [46]. Eating-related behaviours in humans and other animals can usefully be described as a strategy to maintain a balance between conservative risk-minimising and life-preserving strategies (exploitation) with occasional novelty seeking (exploration) in the hope of discovering new, valuable sources of nutrients [47].

The sensory information about the identity and intensity of a food—sometimes called a flavour object—reaching the primary sensory cortices appears to be motivation-independent [48]. This principle has been demonstrated by neurophysiological and neuroimaging experiments using five basic pure tastes of salt, bitter, sour, sweet and umami to locate the primary taste area in humans in the bilateral anterior insula/frontal operculum [49-53] (Figure 4). Please note that one study has reported changes in activity in the primary taste cortex by expectancy [54]; but unfortunately, the authors did not publish the exact coordinates of their putative primary taste cortex. It is thus difficult to trust this finding which is further undermined by visual inspection of the published figure, which clearly shows that the authors’ purported primary taste cortex is located significantly posterior in the medial insular cortex, in contrast to the anterior insular primary taste region reported above and in all other careful neuroimaging taste studies.

Reward representations of sensory stimuli

Subsequent to establishing motivation-independent representations and multisensory representations of information about a food, affective valence is assigned, helping to guide prediction and decision-making. Again, pure taste serves as a good example with a neuroimaging study finding a dissociation between the brain regions responding to the intensity of the taste and its affective valence [64]. Another study found that subjective ratings of taste pleasantness correlated with activity in the medial orbitofrontal cortex (medial OFC) and in the anterior cingulate cortex [65] but, importantly, not with activity in the primary taste region, which was motivation-independent. Further evidence comes from experiments using orthonasal olfaction to show dissociable encoding of the intensity and pleasantness of olfactory stimuli, with the intensity encoded in the amygdala and nearby regions, and the pleasantness correlated with activity in the medial OFC (Figure 5A) and anterior cingulate cortex [66-68].

These reward-related findings in the medial OFC cohere with neuroimaging studies using other rewards. One study found a correlation between activity in the medial OFC with the amount of monetary wins and losses [69] (Figure 5B). Similarly, the subjective experience of methamphetamine over minutes was found to correlate with activity in the medial OFC [70] (Figure 5C). Even studies on the much shorter timescales of milliseconds have found activity in the medial OFC related to the reward of images of cute babies [71] (Figure 5D). These results point to the unity of reward-related activity in the pleasure system across many different rewards, which in turn suggest a system with a common currency of reward. Such a system would make it easier to decide and choose between different rewards.