Food choices are made in the brain, integrating a multitude of neural and hormonal signals reflecting internal state and the environment [
43]. The brain does not reach full maturity until 21 years of age. Furthermore, not all brain areas mature at the same rate; relatively greater changes have been reported in the prefrontal cortex (PFC) compared with the other brain regions between the age of 8 and the early 20s for synaptogenesis [
44], gray matter reduction [
45], myelination increase [
46], and resting level metabolism [
47]. Areas in the PFC, such as its lateral areas, mediate the capacity to voluntarily inhibit desire for a short-term reward in favor of a (larger) long-term reward [
48] and are thus important for self-control. As people grow old, there are gradual structural changes such as decreases in gray matter density and synaptic pruning and cell shrinkage [
49].
Neural Correlates of Anticipation to Food
The process of food choice starts with the anticipation phase, when food or food-related cues are perceived or thought of. Upon perception of a food cue, multiple processes occur in the brain such as preparation for food ingestion and food evaluation [
43,
50]. Examining brain responses to food cue exposure helps to elucidate the mechanisms underlying eating behavior. This is supported by studies showing that brain reactivity to food cues predicts things like future weight gain in adolescent girls [
51] and women [
52], food choice [
53,
54], snack consumption [
55], weight status [
56], and outcome in a weight-loss program [
57]. When normal-weight individuals look at food pictures compared with non-food pictures, areas in the appetitive brain network become active. This network centers around four interconnected brain regions: (1) the amygdala and hippocampus, (2) the orbitofrontal cortex (OFC) and ventromedial prefrontal cortex (vmPFC), (3) the striatum, and (4) the insula [
50,
58]. Furthermore, brain areas involved in attention and visual processing (lateral occipital complex) are consistently more active in response to food compared with non-food pictures [
50].
Functional neuroimaging has provided a means to investigate on a neural level whether overweight and obese individuals are more sensitive to food cues (see, e.g., Schachter’s externality hypothesis, which states that obese people are more reactive to external food cues and less sensitive to internal hunger and satiety signals than normal-weight individuals [
59]) and may thus exhibit greater anticipatory brain activation upon food cue exposure. Indeed, overweight and obese individuals have increased activation in response to food cues in regions associated with cognitive evaluation of salient stimuli (OFC, dorsomedial prefrontal cortex; dmPFC, anterior cingulate cortex; ACC), motor responses (precentral gyrus) and explicit memory (parahippocampal gyrus), when compared with normal-weight individuals. Additionally, they have reduced activation in regions linked to cognitive control (dorsolateral prefrontal cortex; dlPFC) and interoceptive awareness (insular cortex) compared to normal-weight individuals [
60•]. Furthermore, hunger state has a differential effect on obese than on normal-weight individuals. When hungry, obese individuals show greater activation in areas involved in emotion and memory (amygdala/hippocampus), and reduced activation in areas involved in interoception (insula) than those with normal-weight. When satiated, obese individuals have greater activation in reward areas (caudate body/striatum), areas associated with cognitive evaluation of salient stimuli (dmPFC), and attention (supramarginal gyrus) than normal-weight individuals [
61]. Thus, overweight and obese individuals may have a stronger anticipatory response to food in areas involved in evaluation and memory and a lower response in areas important for cognitive control and interoception. Food-related brain responses of overweight and obese people may be differentially affected by satiation as they may have a higher reward response than normal-weight people when satiated. This may make them more likely to eat even when they are not hungry.
In response to food cues, children most consistently activate the same areas as adults do, which are part of the appetitive brain network [
62•]. There are some indications that children may not activate areas important for cognitive control (ventrolateral prefrontal cortex; vlPFC), but there are not enough studies in children to properly establish this [
62•]. Only a handful of studies have looked at the difference in brain activation in response to food cues between normal-weight and overweight children. When comparing overweight and obese with normal-weight children, the former show higher activation during food anticipation in areas involved in cognitive control (dlPFC, vlPFC), interoception (insula), and cognitive evaluation of salient stimuli (OFC, ACC) [
51,
63‐
65]. Overweight and obese children deactivate areas involved in visual attention (the middle occipital and fusiform gyrus), memory (the hippocampus and parahippocampal gyrus), and reward (the caudate/striatum) compared with normal-weight children [
63]. In summary, children may have less inhibitory activation during food anticipation. Few studies have been done in overweight children and results appear to contradict those in adults, as children with overweight have a higher response in areas involved in cognitive control and interoception when compared with normal-weight children while the opposite is found in adults. Intriguing as this finding may be, given the small number of studies and large age ranges of children studied (8–18 years), future studies should directly compare normal and overweight children and adults. So far, no studies have addressed the neural correlates of food anticipation in older adults or elderly.
Neural Correlates of Food Choice
To date, the neural correlates of food choice have been studied relatively little. Various tasks and designs have been used to investigate aspects of the brain processes behind food decisions. These studies mostly use single or dual food choice paradigms [
53,
66‐
73], willingness to pay for different foods [
74‐
77], or auction paradigms [
78]. However, tasks, types of choices, stimuli, and participant characteristics vary greatly between studies. In the decision-making process, the different attributes of the stimuli (e.g., taste, healthiness, size, and packaging) are valued, weighed, and integrated into a single stimulus value [
79,
80••]. Neuroimaging studies have consistently shown that this stimulus value is encoded in the vmPFC, both for food and non-food (e.g., monetary) items [
70‐
77,
81]. For a comprehensive review on the neurocomputational perspective of dietary choice see Rangel [
80••].
In the context of overconsumption, it is interesting to investigate how healthiness of food impacts the food choice process. To elucidate what happens in the brains of people motivated to make healthy choices, dieters can be examined. When dieters successfully make healthy choices, the value signal encoded in the vmPFC is increased by the healthiness of the choice option. During healthy choice, vmPFC activation is modulated by the dlPFC when self-control is necessary (e.g., when refusing an unhealthy, but tasty food) [
72]. In dieters that do not successfully exercise self-control, the value signal in the vmPFC only reflects taste, while in successful self-controllers it incorporates both taste and health. Intriguingly, these neural mechanisms underlying successful self-control can be activated by merely asking people to consider the healthiness of the food. When considering healthiness, the vmPFC value signal incorporates the health aspects of the food even in individuals without an explicit health goal. Furthermore, the vmPFC signal is again modulated by the dlPFC, and they make healthier choices [
73]. In everyday life, a health cue might come in the form of a health label used in marketing (such as “high in calories” or “low fat content”). When labels like this are shown alongside food in a food choice task, the healthiness of the foods is encoded in the amygdala (emotion) [
66]. Interestingly, there is a negative coupling between amygdala and dlPFC when these health labels are shown [
66]. The difference between the neural responses to health considerations and health labels may be caused by the fact that the health labels were shown more implicitly compared with the explicit instruction to consider healthiness. Alongside health labels, health information is commonly encountered in the shape of nutritional value tables on food packaging. However, a more graphic design, a traffic light system, has been proposed as an alternative and is more effective in promoting healthy choices [
82]. When the neural responses to this traffic light label are compared with text-based nutritional information, red traffic light signaling (for unhealthy foods) activates the dlPFC, and there is increased coupling between dlPFC and vmPFC [
83•].This suggests that explicitly asking to attend to healthiness or a graphic health label leads to different neural processing than implicitly showing a health label. This should however be further examined.
An interesting way to look at the effect of caloric content and tastiness of foods is to make choice-pairs based on liking. When people choose a high calorie product over a low calorie product, while they are sated and they have rated the foods as equally tasty, the superior temporal sulcus, a brain area involved in processing biological relevant information is activated [
69]. This suggests that even when motivation to eat is low, the brain still tracks caloric value. Choice-pairs can furthermore be made challenging by design, by pairing a liked high calorie food with a less liked low calorie food. Weight-concerned women, who are trying to limit their energy intake but are generally unsuccessful in this, show lower activation in the anterior cingulate cortex, an area involved in valuation and conflict monitoring when making challenging choices, and accordingly fail to choose in line with their dieting goal [
67].
To our knowledge, the effects of weight status or age on the neural correlates of food choice have not yet been examined. However, since the dlPFC is among the last brain regions to mature, the self-control system may be underdeveloped in children, which would make healthy food decisions more challenging for them. Furthermore, lower dlPFC activation in overweight/obese adults during food anticipation suggests that they may have poorer self-control.
In conclusion, there is a growing body of work on the neural correlates of food choice. Valuation activity in the vmPFC appears to be mostly related to tastiness in normal-weight individuals. When considering the healthiness of the food, or attending to graphic health labels, health value is encoded in the dlPFC and positively modulates vmPFC activation. More implicit health information is encoded in the amygdala and negatively coupled with dlPFC activation. Even when satiated, the brain tracks caloric content during choice, and the lack of conflict-related brain activation may cause self-control to fail in weight-concerned women. Future studies should expand this by exploring the role of weight status and age on healthy decision-making.