When fed a standard laboratory monkey diet, low in sugar and fat and high in fiber (53, 65), subordinate females weigh significantly less than dominant animals (Fig. 2a). This decreased body weight is associated with reduced fat, but not lean, mass (Fig. 2a). Additionally, subordinate females have lower levels of circulating leptin and insulin, and higher levels of adiponectin compared to dominant females (65). The reduced body weight in subordinate females corroborates findings of diminished body weight in rodents exposed to repeated restraint stress (66), chronic variable stress (67), or exposure to a predator (68). These results in rodents are consistent with other data from rodents showing that exposure to chronic stressors (69, 70) or CRH administration (71–74) attenuates intake of these low calorie, laboratory diets.

The lower body weight and fat mass in subordinate animals could be due to lower intake of the LCD, greater energy expenditure, or both. Assessment of LCD intake over the course of 2 weeks in socially housed animals using the automated feeders yields no status differences in LCD intake. Subordinate females do not consume fewer calories of a LCD than dominant females in this dietary environment (Fig. 2b), suggesting that the status difference in body weight is due to greater energy expenditure in subordinate females compared to their dominant counterparts. Indeed, assessment of activity bouts with Actical accelerometers, as a surrogate measure of energy expenditure (75), in these group-housed animals, shows that subordinate animals are more active during daytime hours, suggesting that subordinates expend more overall energy over the course of a day than do dominant animals (Fig. 2c). Increased activity in subordinate animals could indeed be a consequence of the constant threat of aggression and harassment and the need to avoid and withdraw from higher ranking females. Nonetheless, these data indicate that lower body weights in subordinate females are associated with higher levels of activity.

The imposition of social subordination following new group formation (47) reduces body weight, and these lower body weights are sustained over time (76). It is important to emphasize that these observations occur in the presence of a standard LCD. The more relevant question as it pertains to the increase in emotional eating and obesity in humans is what happens to food intake and preference when the dietary environment is expanded to include highly palatable, high-calorie foods. Do socially subordinate females show a preference for a high caloric diet that is high in fat and sugar (HFSD)? And do they increase overall calorie intake in a rich dietary environment when a choice of food is available, analogous to the dietary environment in human societies?

Availability of a HFSD in combination with a LCD for just 2 weeks has a profound impact on caloric intake, most notably for subordinate females. Although both dominant and subordinate females prefer to consume the HFSD when given a choice, overall caloric intake is markedly different in dominant and subordinate animals during this diet condition. While dominant females consume similar amounts of total calories during both the choice condition and LCD-only conditions, subordinate females significantly increase total caloric intake when HFSD is offered concurrently with a LCD option. This augmented intake of calories upon HFSD availability in subordinate animals is greater than both subordinate baseline intake during the LCD-only condition and dominant female intake under the choice condition (Fig. 3). These data in monkeys corroborate findings from a number of rodent models (77–79) and data from humans (3, 5, 6), indicating that exposure to stress facilitates augmented calorie intake and preference for a high-calorie diet.

Despite having only 2 weeks of access to this rich dietary environment, the increase in caloric intake during a choice diet condition is associated with an increase in body weight in subordinate females (45). Because the data indicate that appetite in dominant animals is unaffected by dietary environment, it is plausible that the efficacy of satiety signals could be diminished and orexigenic signals augmented in subordinates (45), resulting in increased caloric intake among subordinate females during the choice diet condition. Taken together, these data indicate that the chronic psychosocial stress experienced by subordinate female monkeys increases total caloric intake only when these females are exposed to a rich dietary environment. These findings are directly relevant to the increase in emotional feeding and obesity in humans that occurs when the dietary environment includes highly palatable, high-calorie foods (21, 23).

While social status and stress exposure account for a significant proportion of variance in total caloric intake when a HFSD is available, there is nonetheless variability in feeding behavior within each social status category in this dietary environment. Animals classified as dominant show some variability in HFSD intake during its availability (Fig. 5), whereas females categorized as subordinate show a greater amount of variance in HFSD intake. This variability among members of each social status category indicates that other variables interact with social status to contribute to this feeding phenotype. Genetic factors confer individual differences and contribute to increased individual vulnerability to stress-induced disorders (85). An example of this is the polymorphic region in the promoter of the serotonin reuptake transporter, whose short allele variant has been linked to increased susceptibility to depression (85). Likewise, a polymorphism in the dopamine D2 receptor gene is linked to increased incidence of obesity (86). Thus, polymorphisms in genes regulating feeding behavior and stress axis reactivity could interact with the environment and account for the variability observed in emotional feeding within subordinate monkeys (87–89). Furthermore, epigenetic changes in gene expression due to life experiences and the environment likely facilitate individual variability in emotional feeding and other stress-induced phenotypes (90). Importantly, social subordination in female rhesus monkeys provides us with a unique model to differentiate how these factors contribute to the feeding phenotypes observed in both dominant and subordinate animals.

The mechanisms responsible for the increased motivation to consume calories in the presence of a highly palatable diet in individuals experiencing chronic stress remain uncertain. A possible explanation for altered feeding behavior is altered sensitivity to appetite and satiety signals under conditions of chronic stress. An example of this is the increased sensitivity to ghrelin observed in subordinate females, and not dominant females, that is linked to increased caloric intake and LHPA dysregulation in subordinate animals (91). Leptin signaling might also be disrupted by social subordination as HFSD intake in subordinate females increases circulating leptin levels while still maintaining increased caloric intake (56). Thus, changes in sensitivity to appetite and satiety signals can occur in concert with one another and depend on the dietary environment. Future studies are necessary to delineate how psychosocial stress exposure and activity of the LHPA axis influence the regulation of orexigenic and satiety signals in complex dietary environments to dysregulate feeding behavior.

Another possible mechanism is that comfort food ingestion diminishes activation of the stress response, as has been shown in rodents (22, 78, 92, 93) and in high-stress premenopausal women (94). However, because glucocorticoids are a critical component in initiating emotional feeding (95–97), a reduction in stress hormone responsivity by comfort foods is likely not a sustaining factor that maintains this phenotype. In contrast, the availability of a HFSD in female macaques increases the diurnal rhythm of cortisol (45) and HFSD consumption augments cortisol responsivity to an acute social separation stressor, regardless of social status (45, 56). These data in subordinate female monkeys not only support data from rodents with access to a HFSD (98–102), but also are consistent with clinical data linking enhanced LHPA activity to measures of central adiposity (103–105). This increased LHPA activity linked to ingestion of HFSD in subordinate monkeys supports the notion that emotional feeding is a behavior that results in the reinforcement and continued motivation to engage in emotional feeding through an increase in glucocorticoids and LHPA activation.

Because ingestion of HFSDs is linked to the activity of the LHPA axis and LHPA axis activity can modulate behavior (53, 91), it could be possible that consumption of these highly palatable diets might affect mood and socioemotional behavior. While some studies in rodents and in rhesus monkeys have shown that availability of a HFSD reduces aggression and anxiety-like behavior (45, 92, 101, 106, 107), other data from monkeys show that HFSD availability has no effect on socioemotional behavior (56). Additionally, while the lack of behavioral effects upon HFSD availability could indicate that HFSD intake does not affect behaviors, it is more plausible that the increased LHPA activation associated with HFSD availability is actually having an adverse or neutral effect on socioemotional behaviors in this particular social context. Indeed, if comfort food ingestion actually increases stress hormone responsivity, consumption of these diets may actually be anxiogenic. Further studies are necessary to determine how HFSD availability and ingestion affect social and emotional behaviors in this model and other social contexts. Additionally, determining how diet affects behavioral responses to acute threatening situations is important to better understand why individuals engage in emotional eating.

A final hypothesis for sustained emotional feeding involves the reward pathways. Chronic psychosocial stress exposure in humans increases individual vulnerability for addictive phenotypes (108), including psychostimulant abuse (109), by reducing dopamine D2 receptors in mesolimbic regions and producing a hypodopaminergic condition in corticolimbic regions of the brain that are important for reward processing (110–112). Intake of calorically dense diets reduces dopamine D2 receptors in these reward regions of the brain in obese humans (113, 114) and in some animals (113). Additionally, studies in rats and in humans indicate that HFSD availability activates reward circuitry in forebrain structures (115–117). Because social subordination in macaques also results in a reduction in dopamine D2 receptor binding potential (61, 118), it is likely that emotional eating in subordinate animals may act as a self-prescribed treatment for increasing the activation of an already dysfunctional reward system (56). Furthermore, the long-lasting effects of HFSD exposure on caloric intake among subordinate monkeys in a healthy dietary environment could be based on a continuing drive to engage a hypoactive dopaminergic system in the absence of a HFSD (56).