Introduction
The worldwide prevalence of overweight and obesity is high with 37% of adults experiencing overweight or obesity (Ng et al.,
2014). Being overweight places individuals at risk of cardiovascular diseases, various forms of cancer, diabetes mellitus type II and musculoskeletal disorders (Lim et al.,
2013). While there may be medical causes involved, most of the variance in Body Mass Index (BMI) can be attributed to behavioral factors (Ravussin & Bogardus,
2000) that result in an energy imbalance (i.e., more energy is consumed than expended). Thus, to achieve weight loss, one needs to reduce caloric intake, increase physical activity, or do both. However, in practice, the prevention and treatment of obesity is not as straightforward. While many individuals may engage in weight loss attempts (De Ridder et al.,
2014) about 80% of overweight dieters are not able to maintain their weight loss in the long run (Wing & Phelan,
2005). Given the considerable health implications of obesity, it is important to find out what determines treatment success and successful weight loss.
Maintaining a healthy weight requires self-control (Hofmann et al.,
2009a,
b). According to dual-process theories of self-control a healthy lifestyle depends on the balance between two competing systems: the impulsive and the reflective (Hofmann et al.,
2009a,
b; Strack & Deutsch,
2004). Behaviors relating to weight gain such as the overconsumption of palatable, energy-dense foods are the result of bottom-up impulses (i.e., the impulsive system) that are not sufficiently regulated via top-down cognitive control processes (i.e., the reflective system; Hofmann et al.,
2009a,
b; Strack & Deutsch,
2004). Executive function lies at the heart of cognitive self-control (Hofmann et al.,
2012). Executive function is an umbrella term that refers to three main cognitive functions: (1) working memory—maintaining and updating relevant information; (2) inhibition—inhibiting prepotent impulses, and (3) shifting—rapidly and efficiently adapting to different situations (Miyake et al.,
2000). Executive function allows for goal-directed action, and is important for planning and monitoring behavior, suppressing undesired responses, resisting temptations and creating alternatives, which are all highly relevant for engaging in and maintaining health behaviors, such as maintaining a diet in order to achieve weight loss (Dohle et al.,
2018; Hall & Marteau,
2014; Hofmann et al.,
2012). For example, working memory is important for the regulation of food intake by keeping long-term goals active and down regulating cravings for immediate desires (e.g. Hofmann et al.,
2008) while inhibition assists in the suppression of automatic impulses to consume tasty, high-calorie foods (e.g. Guerrieri et al.,
2012; Hofmann et al.,
2009a,
b). Shifting has been hypothesized to be important in the selection of alternative means to pursue diet goals (e.g. Dohle et al.,
2018; Hofmann et al.,
2012), and has been found to be predictive of the extent to which individuals translate their healthy eating intentions into eating behavior (Allan et al.,
2011).
In line with dual-process accounts, weaker executive function appears to be related to elevated BMI (for reviews see Fitzpatrick et al.,
2013; Prickett et al.,
2015; Smith et al.,
2011; Vainik et al.,
2013; Yang et al.,
2018). However, the associations between BMI and each individual facet of executive function are not consistently found. Regarding inhibition, while it appears that the association between obesity and weaker inhibition is quite robust (for reviews see Bartholdy et al.,
2016; Lavagnino et al.,
2016), some studies only found differences in performance between individuals with obesity and healthy weight controls on a food-specific inhibition task (e.g. Houben et al.,
2014; Nederkoorn et al.,
2012), and others found no differences in task performance (e.g. Hendrick et al.,
2012; Loeber et al.,
2012; though note that in the latter differences on self-report measures were significant). The association between working memory and BMI is also inconsistent. For example, some studies demonstrate a difference in working memory between individuals with obesity and those with a healthy weight (Maayan et al.,
2011; Stingl et al.,
2012), while other studies find no such difference (Ariza et al.,
2012; Gonzales et al.,
2010). Similarly, significant differences in shifting between individuals with obesity and those with a healthy weight have been found in a number of studies (e.g. Fagundo et al.,
2012; Lokken et al.,
2010; Maayan et al.,
2011), though not in all studies (Ariza et al.,
2012; Mobbs et al.,
2011).
The inconsistency in these findings can perhaps be explained by the variability in the measures used to assess these constructs. There are many cognitive tasks reported to measure the domains of executive function (Miyake et al.,
2000), and as a result, there is little consistency in methodology and results both within and across different domains of executive function. Thus, when studying executive function, it is imperative that a clear rationale for the selection of measures is presented (Etnier & Chang,
2009). For the current study, representative and commonly used tasks were selected: the
n-back task (Kirchner,
1958) to measure working memory, the Stop-Signal Task (Logan et al.,
1997) to measure inhibition (general and food-specific), and the Trail Making Test (Reitan,
1958) to measure shifting. In addition to cognitive performance-based tasks, a self-report measure of executive function was also included, as both types of measures appear to capture different aspects of executive function: behavioral tasks seem to measure the efficiency of cognitive abilities while self-report measures seem more related to goal achievement in daily life (Toplak et al.,
2013).
Most studies on the relationship between executive function and BMI are cross-sectional in nature and examine differences between individuals of different weights. Examining whether executive function prospectively predicts weight gain and weight loss may be more useful in terms of weight loss intervention design. However, currently, few studies have examined the association between executive function and successful weight loss during a weight-loss intervention. Nederkoorn et al. (
2007) showed that children with obesity who displayed weaker inhibition skills lost less weight during a multidisciplinary residential treatment for obesity. In line with this, Manasse et al. (
2017) found that weaker general inhibition (though not food-specific inhibition) was associated with less weight loss during treatment. Hege et al. (
2013) showed that differences in brain activity during a working memory task predicted successful weight loss during behavioral treatment, suggesting that the ability to encode or retrieve food and weight loss goals may contribute to the regulation of eating behavior. A more extensive prospective study, examining working memory, inhibition, shifting and planning, showed that poorer performance on a shifting task and more impulsive reactions on an inhibition task were associated with less weight loss after 8 weeks (Galioto et al.,
2016). This study included an extensive intervention program conducted over a short period of time using a relatively small sample. Weaker executive function has also been associated with less weight loss following bariatric surgery, possibly due to less adherence to post-operative guidelines for diet and physical activity (Spitznagel et al.,
2013a,
b). Thus, individual differences in executive function within individuals with obesity may explain why some individuals succeed during behavioral treatment and some do not. Currently, evidence for the role of executive function in the prediction of weight loss during treatment is insufficient.
Delay discounting is another construct that could be important in weight regulation, and was therefore included in this research. Delay discounting is the decline in value of a reward according to how temporally distal that reward is. Someone with a tendency to choose immediate over delayed rewards is considered to display greater delay discounting and therefore be more impulsive (Ainslie,
1975). As described earlier, impulsivity or a lack of self-control is counteractive to the maintenance of a healthy weight. Successful weight regulation requires that someone keeps their long term health in mind and does not succumb to tempting food stimuli. However, unhealthy eating seems associated with a focus on immediate benefits, and less concern with future consequences (e.g. Dassen et al.,
2015). A tendency to discount future rewards has been related to obesity (for a review see Barlow et al.,
2016), though note that this association has not been consistently found (e.g. Feda et al.,
2015; Nederkoorn et al.,
2006). A preference for immediate rewards could reduce the success of an intervention, while the ability to delay gratification might facilitate successful weight loss and its maintenance. Evidence for the association between delay discounting and successful weight loss is limited. In a family-based obesity treatment, children who displayed high discounting and considered food to be highly reinforcing lost less weight (Best et al.,
2012), whereas Manasse et al. (
2017) did not find delay discounting to be predictive of weight loss during a standard behavioral treatment.
Thus, evidence linking the specific facets of executive function and delay discounting to BMI and weight loss is inconsistent. Further, few studies have included all three facets of executive function in a single study. Therefore, the aim of the current study was twofold. First, we aimed to clarify the relationships between each facet of executive function and BMI, by comparing executive function performance, self-reported executive function, and delay discounting between individuals with obesity and healthy weight controls. We hypothesized that individuals with obesity would show weaker performance, and report lower scores, on executive function tasks and measures, and display more delay discounting, relative to the healthy weight controls. The second aim of this study was to examine whether executive function and delay discounting would predict weight loss following a multidisciplinary weight loss intervention. We hypothesized that individual differences in executive function and delay discounting among individuals with obesity at the start of a weight loss treatment would predict changes in BMI, such that better inhibition, working memory and shifting, and less delay discounting, would be associated with more weight loss controlling for weight at baseline. Determining whether baseline executive function or delay discounting predicts treatment success would provide additional targets for intervention.
Discussion
The main aim of this study was twofold: (1) to examine whether individuals with obesity would show weaker performance on executive function measures and display more delay discounting relative to healthy weight controls, matched (group level) on age, gender and education level, and (2) whether executive function and delay discounting would be predictive of weight loss during a subsequent multidisciplinary weight loss treatment. This study included both behavioral and self-reported measures of the three main facets of executive function (Diamond,
2013; Miyake et al.,
2000) and delay discounting in one study. Results show that individuals with obesity displayed less efficient general and food-specific behavioral inhibition, relative to healthy weight controls and that they reported weaker executive functioning in daily life. Individuals with obesity did not display weaker behavioral working memory or shifting than healthy weight controls. Regarding the prediction of weight loss, behavioral working memory was the strongest predictor of change in BMI, besides gender and education level. In addition, more difficulties in daily life with respect to inhibition as indicated on the inhibit subscale of the Behavioral Rating Inventory of Executive Functioning contributed marginally significantly to the prediction of BMI loss.
Performance on both the general and food-specific Stop-Signal Task differed between individuals with obesity and healthy weight controls, with individuals with obesity displaying less efficient inhibition. These results are consistent with those of Bartholdy et al. (
2016); however, they differ from Houben et al. (
2014), who only found an association between elevated BMI and weaker performance on a food-specific Stop-Signal Task. The inconsistency in findings may be explained by the different samples used in each study. The current sample included individuals with obesity rather than predominantly individuals who were overweight. It may be the case that differences in general inhibition are only apparent at the higher end of the weight spectrum. Previous research did show promising results of food-specific inhibition training on food intake and weight loss (Houben & Jansen,
2015; Lawrence et al.,
2015), whereas studies in which general inhibition was targeted were not successful (Allom et al.,
2016a).
Individuals with obesity reported more problems experienced in daily life as indicated on the Behavioral Rating Inventory of Executive Functioning (though note that self-reported shifting did not reach significance), in line with our hypothesis. While weaker working memory was reported by individuals with obesity, behavioral working memory as measured with the 2-back task did not differ between the weight groups. Behavioral tasks and ratings of executive functioning have been suggested to reflect different underlying constructs (Allom et al.,
2016b; Toplak et al.,
2013). Thus, based on the current results, obesity is not associated with working memory and shifting as measured by behavioral tasks, though individuals with obesity do seem to experience less successful goal pursuit in daily life.
The current results indicate that behavioral inhibition and self-reported executive function are important in preventing obesity, as they appear to be a risk factor for weight gain (e.g. Dohle et al.,
2018; Hofmann et al.,
2008). However, it is uncertain whether these factors indeed cause obesity, or obesity causes poorer executive function, or whether there is another factor explaining both. There are indications that impaired executive function leads to impaired self-regulation (e.g. higher food intake, less exercise), rendering individuals with deficient executive function more predisposed to becoming obese (Dohle et al.,
2018; Hofmann et al.,
2012). However, there is also evidence that obesity leads to impaired cognitive functioning via reduced blood flow to the areas of the brain that control executive function or abnormalities in glucose and insulin regulation (Boeka & Lokken,
2008; Smith et al.,
2011). Conversely, the results of a recent meta-analysis indicated a significant positive effect of weight loss on executive function (Veronese et al.,
2017). This suggests that the association between executive function and obesity is bidirectional (Kanoski & Davidson,
2011; Sellbom & Gunstad,
2012). To shed more light on this relationship, future research should study these factors in prospective designs including pre- and posttests or experimental designs including a manipulation of executive function.
Rather than focusing on weight gain, the present study focused on examining the prediction of weight loss. Behavioral working memory was the strongest predictor of a decreased BMI after treatment, suggesting an important role of working memory in weight loss success. Specifically, working memory may help to keep weight loss goals in an active state, so someone can efficiently monitor whether their food intake is in line with their weight loss goals (Boutelle & Kirschenbaum,
1998). This is in line with dual process models that suggest that the relative influence of each system on self-regulation differs as a function of working memory such that individuals with better working memory display more goal-directed behavior (e.g. Hofmann et al.,
2008). Our results are in line with Hege et al. (
2013), who showed the predictive value of behavioral working memory for successful weight loss during a lifestyle treatment. Hege et al. (
2013), however, used a food-specific working memory task, which could be even more predictive in the self-regulation of weight loss given the role of working memory in food cue monitoring (Meule,
2016). Based on these results, targeting weaker working memory seems a promising target for intervention (Jansen et al.,
2015). Preliminary results of working memory training indicate short-term effects of training on food intake, though long-term effects on weight loss have not yet been established (Dassen et al.,
2018; Houben et al.,
2016).
Self-reported inhibition was a significant predictor of weight loss, with participants who reported less efficient inhibition in daily life displaying a smaller BMI change. The behavioral measures (e.g. general and food-specific Stop-Signal Task), in contrast, were not related to weight loss. This is contrary to our hypothesis and not in line with previous research (e.g. Galioto et al.,
2016; Manasse et al.,
2017; Nederkoorn et al.,
2007). Shifting as measured with the Trail Making Test did not differ between individuals with obesity and healthy weight controls, and was also not predictive of subsequent weight loss. Importantly, many executive function tasks were originally developed as indicators of brain damage, including the Trail Making Test (Reitan,
1958). Therefore, it is possible that this task was not sensitive enough to pick up subtle impairments in shifting ability (Fitzpatrick et al.,
2013). Task impurity is also a common problem in this field of research, as tasks may be thought to measure a particular facet of executive function but the execution of any task may also involve diverse cognitive systems in addition to the targeted facet (Jurado & Rosselli,
2007). The role of shifting in eating behavior is relatively unexplored (Dohle et al.,
2018). Shifting ability may be beneficial to weight loss, helping individuals to switch strategy when the current strategy is suboptimal, though this ability could also allow individuals to easily switch from their weight loss goal to more tempting immediate options (Hofmann et al.,
2012). However, in the current study, no association of behavioral or self-reported shifting with weight loss was found.
Delay discounting did not differ between individuals with obesity and healthy weight controls, and also did not add significantly to the prediction of weight loss, though results were in the expected direction. In the current study, we measured delay discounting with the Monetary Choice Questionnaire, which uses money as a reinforcer. However, it has been suggested that a food-specific delay discounting task would be more predictive of BMI and weight loss (e.g. Rasmussen et al.,
2010). Current results provide no evidence for a direct relationship of delay discounting with weight loss. Importantly, we followed participants until 6 months after the start of their weight loss intervention. As weight regain is one of the biggest problems in weight loss treatment (Ikeda et al.,
2005), it would be interesting to continue to monitor participants after finishing the treatment, to examine the potential role of executive function and delay discounting in the maintenance of weight loss (Gettens & Gorin,
2017). It has been suggested that self-regulation becomes even more important when the intensive treatment ends, as the external regulation coming from the treatment program also ends at this point (Halberstadt et al.,
2013).
In conclusion, executive function appears to be weaker in individuals with obesity, in line with the results of a recently published meta-analysis (Yang et al.,
2018). We discovered weaknesses on self-report measures and both general and food-specific behavioral inhibition. Executive function is also predictive of weight loss, with behavioral working memory being associated with a greater decrease in BMI following treatment, and a marginally significant contribution of self-reported inhibition was found. A significant contribution of shifting or delay discounting to the prediction of weight loss could not be established. Thus, it seems that differences in some facets of executive function between individuals with obesity and healthy weight individuals exist, but other facets of executive function seem responsible for weight loss success. Based on current results, behavioral inhibition can differentiate between obese and healthy weight individuals, while behavioral working memory predicts weight loss. Thus, the impact of specific executive functions may differ between different weight-related behaviors (Gettens & Gorin,
2017), and future research should distinguish between the role of each facet of executive function in weight gain and weight loss, as this would have different implications for which facets of executive function are important for prevention of obesity, and which facets would be important for intervention. Based on the current results, it would be interesting for future studies to explore whether adding a cognitive screening at the start of a weight loss trajectory could be an objective way to identify individuals who might benefit from additional executive function training to facilitate optimal weight. These results suggest new avenues for the improvement of current behavioral treatments, and may increase future treatment success.