Introduction

In terms of classical conditioning, food intake may be considered an unconditioned stimulus and its metabolic effects unconditioned responses. Cues that reliably signal food intake (e.g., the sight, smell, and taste of food) may start to act as conditioned stimuli that can trigger conditioned responses. This learned food cue reactivity can manifest in several objectively-measurable physiological responses (e.g., increases in salivary flow, heart rate) upon exposure to food or food cues, yet an essential element of food cue reactivity is a subjective experience of an intense desire to eat the food (i.e., craving) [1].

The conditioning model of binge eating proposes that individuals with binge eating display higher reactivity to food cues than individuals without binge eating, which ultimately leads to excessive food intake [2]. However, when examining studies that measured self-reported food craving, it appears that the assumption of larger food cue-induced increases in craving in individuals with binge eating (e.g., persons with bulimia nervosa [BN] or binge-eating disorder [BED]) has not been reliably established.

First, individuals with binge eating do indeed report higher trait food craving (i.e., experience more frequent and intense food cravings in general) and higher state food craving (i.e., experience more intense food craving in the moment of data collection) than individuals without binge eating [3, 4]. However, such studies cannot answer the question of whether individuals with BN or BED show higher food cue reactivity—in terms of food cue-induced increases in craving—than healthy controls. Second, some studies reported higher food craving in individuals with BN or BED than in healthy controls after a food cue exposure, but did not measure baseline levels of food craving [5, 6]. Thus, whether craving increases were larger in participants with BN or BED than in controls cannot be inferred from these studies. Third, studies that included a food cue exposure with pre- and post-exposure measurements of food craving and compared individuals with BN and/or BED with healthy controls did not find a significant interaction between groups and measurements, that is, food craving or desire to binge increased equally in all groups during food cue exposure [7–14].

In conclusion, while individuals with BN or BED report more frequent and intense food cravings in general than individuals without binge eating, there is no compelling evidence that these differences reflect an elevated reactivity to food cues in terms of larger increases of food craving intensity in response to food cues (which would explain their difficulties in controlling food intake). Therefore, we tested whether a group of individuals with BN or BED would demonstrate stronger increases in momentary food craving during a food cue exposure than individuals without binge eating. Previous null findings may be partially due to the fact that control participants were not screened for trait food craving levels, which would also be associated with elevated food cue reactivity [15, 16]. Thus, to ensure that control participants did not have any sub-threshold eating disorder symptoms or any eating-related psychopathology, we used a control group of individuals without binge eating and low levels of trait food craving.

Materials and methods

Participants

The study was approved by the institutional review board of the Medical Faculty of the University of Heidelberg. Participants provided written informed consent before commencing the study and received financial compensation for their participation in the study. Fourteen individuals with BN and 13 individuals with BED, who took part in a larger research project [17] and who were all recruited at the same study site, participated in this study (binge-eating group). Nineteen individuals without any eating disorder or elevated trait food craving scores were used as control group [18]. Sample characteristics are displayed in Table 1. In the binge-eating group, 66.7% of participants had normal weight (n = 18), 14.8% were overweight (n = 4), and 18.5% were obese (n = 5), according to the guidelines by the World Health Organization [19]. In the control group, 78.9% of participants had normal weight (n = 15), 15.8% were overweight (n = 3), and 5.3% were obese (n = 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sample characteristics as a function of group.

https://doi.org/10.1371/journal.pone.0204151.t001

All participants were recruited through websites, circular mails, advertising posters and advertisements in local and social media. Women and men were eligible for participation in the study if they were aged 18 years or above and had a body mass index ≥ 18.5 kg/m². Diagnostic assessment in the binge-eating group was based on the Structured Clinical Interview for DSM-5 Disorders [20]. Exclusion criteria for the binge-eating group were: medical (e.g., severe electrolyte abnormalities) or psychiatric (e.g., acute suicidality) instability, the need for immediate inpatient treatment, substance dependence, psychosis, bipolar disorder, borderline personality disorder, psychotropic medication use, severe learning disability or inability to speak fluent German, impacting on the person’s ability to complete study assessments. Use of selective serotonin reuptake inhibitors was not an exclusion criterion when medication was stable (i.e. at least 14 days of continuous use). Participants in the control group were invited for laboratory testing if they scored in the lower third on the German version of the Food Cravings Questionnaire–Trait [21] in a sample of n = 358 volunteers, did not report any binge eating episodes as assessed with the Eating Disorder Examination–Questionnaire (EDE–Q) [22] and had no current or prior psychotherapeutic or psychopharmacological treatment.

Results

The binge-eating group was older and had higher EDE–Q scores than the control group (Table 1). Regarding state food craving, main effects of group (F_(1,44) = 4.78, p = .034, η_p² = .098) and measurement (F_(1,44) = 51.6, p < .001, η_p² = .540) were qualified by a significant group × measurement interaction (F_(1,44) = 4.29, p = .044, η_p² = .089). During the food cue exposure, state food craving increased in both the binge-eating group (t₍₂₆₎ = 6.58, p < .001, d = 0.64) and the control group (t₍₁₈₎ = 3.95, p = .001, d = 0.40). However, while groups did not differ before the food cue exposure, the binge-eating group had higher state food craving than the control group after the food cue exposure (Table 1; Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Scores on the Food Cravings Questionnaire–State as a function of group (controls vs. individuals with bulimia nervosa or binge-eating disorder) before and after food cue exposure.

Error bars represent the standard error of the mean. Asterisks indicate p < .05.

https://doi.org/10.1371/journal.pone.0204151.g001

Note that the age difference between groups was driven by participants with BED (M = 35.2 years, SD = 13.5) who were older than participants with BN (M = 25.2 years, SD = 6.62, t_(17.2) = 2.41, p = .028, d = 0.95). As individuals with BED are commonly older than individuals with BN on average, it would have been inappropriate to use age as covariate in the analysis of variance [28]. Thus, we examined effects after excluding the three oldest participants (>50 years old) such that the binge eating and control group did not differ in age any more (t_(31.8) = 1.49, p = .147, d = 0.42). Using these age-matched groups, the group × measurement interaction was still significant (F_(1,41) = 5.45, p = .025, η_p² = .117).

Discussion

Numerous studies showed that individuals with BN or BED experience more frequent and more intense food cravings than persons without binge eating [3–6]. Yet, food cue-induced craving did not differ in individuals with and without binge eating in the majority of studies [7–14], raising the question whether cue reactivity is a useful concept to explain binge eating.

In the current study, individuals with and without binge eating did not differ in their momentary food craving prior to food cue exposure. While it has been found that high trait food craving scores (which individuals with BN or BED have) relate to higher state food craving scores even without being exposed to food cues, this association is rather small [29]. Thus, a difference in state food craving between individuals with and without binge eating at baseline may only be detected in larger samples.

During food cue exposure, state food craving intensity significantly increased in both groups in the current study. However, individuals with BN or BED showed significantly larger food cue reactivity in terms of self-reported craving. Of note, the food cue exposure induced craving even in the control group who was explicitly recruited to have low trait food craving scores. Therefore, we speculate that the lack of finding interactive effects of groups and measurements in previous studies may be due to participant selection. For instance, several previous studies only examined analogue samples, participants with sub-threshold BN/BED, or very small samples [7–9, 11, 13]. Moreover, previous study designs may not have been sensitive enough to detect differences in food cue-induced craving because high calorie foods appeal to all humans and, likewise, induce craving in most individuals (as was demonstrated by craving changes in our low trait food craving control group, in line with previous findings [15]). Taken together, this suggests that differential effects of food cue exposure may only be observed when groups are clearly separable; not only with regard to eating disorder pathology but also with regard to other relevant aspects such as trait food craving scores.

Some methodological considerations limit interpretation of the current findings. First, the sample was predominantly composed of non-obese women. Thus, findings may not be applicable to men or individuals with obesity. Second, because of the small number of individuals with BN and BED, they were combined to one binge-eating group. However, some differences between individuals BN and BED have been noted, for example, with regard to binge eating episodes characteristics [30, 31]. Thus, future studies may reveal differences in food cue reactivity between these groups as well. Third, we did not include a control group of high trait food cravers without binge eating or any other eating-related psychopathology. Recruiting such a control group may be hard to achieve because of the large overlap between trait food craving scores and binge eating tendencies [21]. Yet, as elevated food cue reactivity in high trait food cravers has been previously demonstrated [15], we speculate that such a control group would show similar cue-induced increases in food craving to individuals with BN or BED.

Conclusions

Our data are in line with classical conditioning accounts which propose that food cues represent conditioned stimuli that trigger a conditioned response (e.g., food craving) and that these associations are stronger in individuals with binge eating than in healthy individuals. Although we did not assess actual food intake in the current study, this elevated food cue reactivity may ultimately increase the likelihood to engage in binge eating [2, 32]. As food craving increased in individuals with low trait food craving scores in the current study as well, previous null results might be explained by methodological considerations such as not screening control participants for trait food craving.

Supporting information