Download PDF
Abstract
Binge eating disorder (BED) is characterized by repeated episodes of binge eating accompanied by a loss of control. Although the neurobiological underpinnings of binge eating (BE) episodes are not fully understood, there are indications that variability in nucleus accumbens (NAcc) responses could lead to increased variability in food intake. Here, we assessed whether BED is associated with higher intra-individual variability in behavioral and neuroimaging indices of reward responses. To this end, patients with BED (n = 35, MBMI = 33.2 kg/m2 ± 6.8), participants with subsyndromal BED (n = 21, MBMI = 29.0 kg/m2 ± 7.2), and individuals without symptoms of binge eating (n = 23, MBMI = 32.3 kg/m2 ± 6.5) completed an effort allocation task with concurrent functional magnetic resonance imaging. In line with our hypothesis, we found that patients with BED had higher variability in subjective wanting ratings of food (F34,21 = 1.48, pboot = 0.024), but not effort exertion (F34,21 = 1.13, pboot = 0.30). Crucially, trial-by-trial variability in NAcc responses during the presentation of cues was associated with a higher BMI (b = 0.11, 95%CI [0.03, 0.19], BF10 = 11.1) and disinhibited eating (b = 0.19, 95%CI [0.01, 0.36], BF10 = 4.0) across groups, whereas NAcc variability was only marginally elevated in patients with BED (b = 0.12, 95%CI [−0.04, 0.29], BF10 = 1.2, P > 0|data = 88%). Our results support the idea that BMI and disinhibited eating are associated with more variable NAcc responses, which may contribute to the symptoms of BED. However, this association is only weakly indicative of clinical severity of BED.
Introduction
Binge eating disorder (BED) is the most prevalent eating disorder [1], affecting approximately 1.5% of women and 0.3% of men [2]. BED is characterized by repeated binge eating (BE) episodes, in which an objectively large amount of food is consumed, accompanied by experienced loss of control. Individuals with BED present with high rates of overweight and obesity [3], and mental and physical comorbidities [2, 3]. Whereas circumstances under which BE are more likely to occur have been documented (e.g., 4,5,6), the mechanisms underlying BE remain largely elusive.
Cognitive and motivational models of eating behavior posit that attentional biases are important in maintaining eating disorders including BED (e.g., 7,8,9,10). Such biases may be rooted in incentive sensitization [11]; through frequent encounters with food rewards, the motivational pathway is sensitized to food cues. Consequently, future encounters are associated with higher arousal and craving [9, 12]. In other words, repeated exposure to rewarding food is thought to lead to increased (food) reward responsiveness, which is reflected in altered neural activity when processing the (imminent) receipt of food cues [13]. In line with incentive-sensitization theory, fMRI studies found increased activation of the mesocorticolimbic system to food in individuals with BED [14, 15] while activity in regions that control food intake, such as the dorsolateral prefrontal cortex (dlPFC), is reduced [16]. Moreover, simulations of food-related reward behavior demonstrate that fluctuations in food reward responses could explain an increased variability in food intake [17]. Thus, variability in reward responsiveness could provide a more nuanced model of the contribution of reward processing to the etiology of BED.
To better conceptualize the inherent variability of eating behavior, intra-individual variance—in addition to differences in response amplitude—of behavioral and neural indices of reward processes, such as reward responsiveness, is highly promising [17]. Since phases of food intake alternate with phases of fasting [18, 19], BE episodes or excessive restraint may be reconceptualized as extremes of a distribution. In other words, if homeostatic signals that regulate food intake are overridden through hedonic eating (“disinhibited”), this variance increases. Thus, episodes of disinhibited eating and BE may conceivably increase variability [17]. This idea is supported by an association of the variability in the nucleus accumbens (NAcc) to milkshake receipt with greater variability in subsequent ad libitum food intake in a sample of participants without pathological eating behavior [20]. In this study, variable NAcc responses were associated with increased body mass index (BMI) and disinhibited eating. This complements previous work on the role of the NAcc, which suggests that it is involved in the anticipation of food rewards [11]. Moreover, altered connectivity of the NAcc potentially increases the risk of individuals experiencing aberrant (e.g. emotional) eating and eating disorders [21,22,23]. Hence, higher intra-individual variability in reward responsiveness may be associated with higher variability in food intake, including variability caused by BE episodes.
Beyond reward responsiveness, the NAcc is also critical for reward valuation [24], reflected in its involvement in the willingness to work for a reward [25]. It is therefore plausible that the effects of elevated variability in reward responsiveness extend beyond food intake. When viewed as an economic exchange, willingness to work for a specific reward reflects reward valuation [26, 27]; as the reward value increases, the effort to obtain it increases. The motivational processes that drive behavior to obtain (food) rewards are mediated by reward value through cost-benefit computations [27, 28]. Effort allocation tasks (EAT) capture such tradeoffs through the dynamic integration of perceived benefits against perceived costs (i.e., physical force; 28,29,30,31). These cost-benefit computations are associated with signaling in the dlPFC [32, 33], which is also known to contribute to effective cognitive control over food intake and has been shown to be attenuated in BED [16]. Since the expected reward value underlying these computations is not a set value, but rather a distribution of possible outcomes [34], fluctuations in the reward value account for a substantial variance in subsequent decision-making [25, 35, 36]. In line with this prediction, BMI was associated with higher variability in subjective wanting ratings [37]. Accordingly, animal studies demonstrate that uncertainty of reward receipt could exacerbate such variability; when rodents are provided food at insecure feeding schedules, binge-like eating behavior ensues [38, 39]. This uncertainty in availability was also associated with weight cycling [40]. As such, variability in the motivation to work for reward should be most pronounced when uncertainty about the difficulty—and therefore receipt of reward—is introduced in the parameters of the cost-benefit computation.
Based on previous evidence showing that altered reward processing is implicated in the etiology and maintenance of BED, we determined whether BED is associated with increased variability in reward valuation using comprehensive behavioral and neural indices. We indexed the variability of the reward value of food and monetary rewards in individuals with BED, subsyndromal BED and controls. Here, we recruited weight-matched controls because BED is associated with overweight and obesity [3]. In turn, this is associated with increased variability in NAcc reward responses [20]. By using a weight-matched control group, rather than a normal weight control group, we can determine the unique contribution of BED on variability in reward processing while estimating the dimensional effects of BMI and disinhibited eating across all groups.
We predicted that variability in task behavior reflects variability in neural reward signaling [20]. We expected variability to be exacerbated if the effort required was uncertain and manipulated the uncertainty of the effort requirements accordingly. We hypothesized that individuals with BED are characterized by increased variance in reward valuation, which is reflected in increased variability in effort exertion and in NAcc responses. Moreover, we expected increased BE symptomology to be associated with increased variability in reward valuation. Furthermore, we evaluated variability in the dlPFC, since this region is integral to cost-benefit computations. Therefore, behavioral variability could be associated with variability in control-related signals associated with the dlPFC function.
Methods
Participants
Participants were recruited through the mailing lists of the University of Tübingen, the associated outpatient clinic and through (social media) advertisements posted in the surrounding area. An a priori power analysis based on previous work [20] indicated that a combined sample of N = 60 yields excellent power (1-β = 0.91) assuming medium to large effect sizes (r = 0.45). In total, 79 women participated in the first session, of which 35 fulfilled the criteria of BED (range BMI: 22.6 – 54.6), 21 experienced subsyndromal binge eating (subBED, range BMI: 20.1 – 42.5), and 23 individuals did not experience binge eating (no BE, range BMI: 22.2 – 44.9; Table 1). From this original sample, 59 participants also completed the fMRI session (several BED patients had MR contraindications and therefore only completed a behavioral session). For both sessions, groups were comparable regarding BMI and age.
Group membership was determined through the Eating Disorder Examination (EDE; 41). For the subBED group, all diagnostic criteria of BED according to DSM-5 needed to be fulfilled, with the exception that the frequency of BE episodes was lower than once a week for 3 months. Participants were excluded if they fulfilled the criteria for bipolar disorder, schizophrenia spectrum or other psychotic disorders during their lifetime, or alcohol/substance use disorder within the past six months as determined by the Structured Clinical Interview for DSM-IV (SCID; 42). Given high comorbidity (all listed in Table S1) rates with major depressive disorder [1], we only excluded participants with comorbid depression if they reported acute suicidality, or if antidepressive medication was started or altered in the past two months. We additionally excluded no BE and subBED participants who fulfilled the lifetime criteria for any other eating disorder.
The study was preregistered (NCT04184856) and approved by the institutional review board of the University of Tübingen (3939/2017BO2) and completed in line with the Declaration of Helsinki. Participants provided written informed consent before the diagnostic interview. For the full study, participants received €110 with potential additional winnings based on task performance
Experimental procedure
We assessed in- and exclusion criteria through a telephone screening (~25 min). Suitable participants subsequently partook in an online study and two laboratory sessions. The online part included questionnaires on aberrant eating behavior and psychopathology, among them the three factor eating questionnaire (TFEQ; 43), which measures cognitive restraint, dietary disinhibition and non-homeostatic eating. For the first session, we asked participants to eat approximately 1.5 h in advance (neither hungry nor full). After signing informed consent, participants completed a practice round of the grip force EAT (grEAT). Subsequently, we conducted a clinical interview. Participants then completed a food cue reactivity task (FCR, ca. 20 min; 44), before completing the grEAT (ca. 40 min; 45). Finally, participants took part in a taste test (ca. 20 min; 46). The first session lasted approximately 3.5 h.
For the second session, participants came to the lab after an overnight fast (i.e., no caloric intake >8 h prior to the session). Participants received a small breakfast and completed a reinforcement learning task [47]. They completed a second grEAT (ca. 30 min) and a food bidding task (ca. 20 min) with concurrent fMRI. In total, the second session lasted approximately 4.75 h. The current paper focuses on the grEAT, while the other tasks are reported elsewhere
Grip force effort allocation task
In line with previous work [37], we estimated variability in reward value through its influence on effort exertion. To this end, we combined the task from Neuser, Teckentrup [31] with grip force as input modality [29] and uncertaint difficulty in a subset of trials. For each participant, we determined the maximum grip force [31] and an exchange rate for points to money/kcal was adapted according to the ratio of the linear exertion slopes of exerted force for standardized amounts of money and snacks.
In the grEAT (Fig. 1d; details in the SI), participants worked for monetary and food rewards in low (one point per second) and high magnitude trials (10 points/s). To incorporate previous insights about uncertainty and BE, the difficulty was uncertain (no line shown) in half of the trials. In trials of the grEAT in the first session, participants additionally indicated on 0–100 visual analog scales (VAS) how much they exerted themselves, how much they wanted the reward
a Hypothesized differences in average reward sensitivity across groups. b Hypothesized differences in the distribution of reward sensitivity across groups. c Schematic depiction of high and low variability over trials for % grip force and NAcc signaling. d Overview of trials on the effort allocation task. Participants first see the reward cue, before the difficulty level is revealed. In the certain condition, a token counter is shown depicting the earned points so far. This counter is hidden in the uncertain condition because participants do not know the exact difficulty level. At the end of each trial, participants are shown the total number of points earned in the current trial (for details see Supplemental Information). e Schematic overview of the conditions of the grip force effort allocation task. Participants worked for small and large money and food rewards. The difficulty always ranged from 64 – 95% of the individual grip force maximum. In the certain condition, this was shown by a single line. In the uncertain condition, the entire range from 64 – 95% was indicated by a red box (Supplemental Information). BE = binge eating, BED = binge eating disorder, BMI = body mass index, NAcc = nucleus accumbens.
MRI data acquisition and preprocessing
fMRI data (~35 min, 1500 volumes, see SI) were acquired on a Siemens 3T Prisma scanner with a 64-channel head coil. Data was then preprocessed using fMRIprep [48] and smoothed (6 × 6 × 6 mm³). For confound correction in first-levels, we extracted the average white matter and CSF signal as well as the six motion regressors
Data analysis
Behavioral data
Using behavioral data from the first session, we assessed effort and wanting ratings. We defined effort as the percentage of the force of each participant’s individual maximum grip force. We computed the average effort per trial and used linear mixed-effect models as implemented in R (lmerTest) using group (no BE, subBED, BED), reward type (food, monetary), and reward magnitude (low, high), and their interactions as fixed effects with BMI and age as covariates. As random effects of participants, we used random intercepts and slopes for reward type and reward magnitude. To evaluate variability in two states where the costs of effort are known versus unknown, we estimated separate models for certain and uncertain trials.
In line with previous work [20, 37], we defined variability as trial-wise residuals (i.e., after accounting for systematic effects of conditions) of the relative effort exerted (“objective” value) and wanting ratings (subjective value) after each trial. We anticipated that differences in variability would show primarily in uncertain trials as they reflect the internal valuation of the rewards independent of the difficulty. Therefore, to test whether variability differed between groups, we bootstrapped (1,000 resamples) variance ratio tests (Var(BED/subBED)/Var(no BE)) on residuals of uncertain trials for food and monetary rewards [25]. To assess associations with BMI, we calculated the individual SD of the residuals and report bootstrapped Pearson correlation with BMI.
fMRI data
Statistical analyses of fMRI data were conducted using SPM12. The first-level general linear models included regressors for cue, work, and feedback phases (see SI). For the variability analyses, we extracted the adjusted signal (filtered and corrected for covariates) in the NAcc and dlPFC of the cue phase, when difficulty was not shown yet. Therefore, we include all trials in this analysis and only report results separated for certain and uncertain trials to highlight the robustness in Tables (S4-8). To determine whether alterations in brain response variability were specific to these ROIs, we additionally extracted the adjusted signal of the temporal lobe control regions (anterior: aMTL, posterior: pMTL, temporo-occipital: toMTL), which should be unrelated to reward processing. Therefore, any non-specific associations of BMI with increased brain response variability, for example, due to increased movement, should affect the signal in control regions as well. Regions of interest (ROI) were defined from the Harvard Oxford atlas [49] and combined across hemispheres. The dlPFC was derived from Neurosynth (meta-analysis with the term dlPFC, z-threshold = 5.2, smoothed with a 4x4x4 mm³ kernel).
To calculate individual variability in anticipatory cue responses, we used linear mixed-effects models (fitlme, MATLAB v2023a). In this step, we partitioned brain responses in each trial into group-level and individual components. This eliminated common variance that is driven by condition-specific effects (e.g., in response to large vs. small rewards). Second, we estimated individual variability as the variance of individual participants’ random effects estimate per trial using BRMS [20]. To estimate trial-wise cue responses, we assigned all volumes a corresponding trial number, with a new trial starting once a new cue was shown. The models predicting activation in the ROIs included all task regressors from the first-level model (i.e., after convolution with the hemodynamic response function and filtering). To account for differences between trials, we included interactions of cue presentation with reward magnitude (centered), difficulty (centered), uncertainty (dummy-coded), and reward type (dummy-coded). The models included random intercepts and slopes for all predictors and a nested trial term (1 + Cue|ID:Trial) so that distinct deviations in activation from the group means are estimated for each trial and each individual.
As additional sensitivity analyses regarding the potentially confounding effects of movement, we re-estimated the linear mixed-effects models that estimate trial-wise cue responses including movement parameters and volume-wise framewise displacement to provide a more stringent control of motion and added participant-level average frame-wise displacement (ln-transformed) to models evaluating BMI effects. To determine whether the results were specific to the cue response or also held for feedback, we fit analogous models for feedback response (i.e., nested trial term (1+feedback|ID:Trial). Note that the feedback phase is more strongly affected by movement (Table S8-9) and more comparable to our previous work [20], providing a way to test whether the observed associations with NAcc variability are specific for cue events.
To determine the effects of group and BMI on the variability in cue responses, we estimated Bayesian mixed location-scale models with the trial residuals (i.e., deviation from the condition mean) as outcome. The model for the amplitude (location) only included a random intercept. The model for individual variance terms included the task conditions, reward type, reward magnitude, their interaction, and random intercepts and slopes for task conditions. We included BMI (z-standardized) and Group (dummy-coded) and their interaction with the task conditions as fixed effects. We estimated separate models for all 5 ROIs (NAcc, dlPFC, aMTL, pMTL, toMTL). Bayesian models are evaluated using the 95% credible interval that does not include 0 if an effect is significant. In addition, Bayes Factors (BF) quantify the evidence for or against the null hypothesis, where BF10 quantifies the evidence for the (undirected) alternative hypothesis.
In the current manuscript, we are focusing on differences in the variability of the reward response, as differences in amplitude of the anticipatory NAcc response have been reported elsewhere, showing a lower reward valuation (i.e., difference in NAcc response to high vs. low rewards) in BED compared to no BE but no association with BMI [45]. Crucially, we used the hierarchical estimation of trial coefficients chosen here and then analyzed them using the same linear mixed-effects model as for the behavioral session.
Statistical threshold and software
For prespecified hypotheses that either aimed to replicate our previous findings [20] that NAcc reward variability is increased in participants with a higher BMI and with higher self-reported disinhibited eating or predicted that participants with BED would show higher behavioral and NAcc variability as specified in the trial preregistration NCT04184856 [17], we used one-sided tests and Bayes factors (BF). For all other tests (i.e., associations with cue response variability in the dlPFC or interactions with task components), we used a two-tailed threshold of α ≤ 0.05 to interpret the significance of our findings. Correction for multiple comparisons was applied for post hoc analyses (i.e., reward type specific effects) using Bonferroni correction for two tests and across interactions with task conditions (i.e., reward type and reward magnitude) by using more stringent credible intervals (97.5%). We preprocessed the raw behavioral data in MATLAB v2021a. Linear mixed-effects models [50] were analyzed with lmerTest [51]. Location-scale models to investigate changes in variability were analyzed with brms [52] in R [53]. We used ggplot2 [54] and ggdist [55] for data visualization.
Results
No differences between groups for average effort and wanting in session one
To characterize average effort, we used separate mixed-effects models for certain and uncertain trials of the first session. As anticipated, participants exerted more effort in trials with a large vs. small reward at stake (certain: b = 14.3, t95.2 = 3.09, p < 0.001; uncertain: b = 18.7, t93.2 = 3.33, p < 0.001). Furthermore, they exerted more effort for money vs. food rewards (certain: b = 12.2, t91.4 = 5.01, p = 0.003; uncertain: b = 13.9, t89.0 = 3.09, p = 0.001). Likewise, participants reported wanting large rewards more than small rewards (certain: b = 25.3, t89.3 = 5.48, p < 0.001; uncertain: b = 26.5, t91.3 = 6.00, p < 0.001) and money more than food rewards (certain: b = 16.0, t77.3 = 3.25, p = 0.002; uncertain: b = 16.0, t85.9 = 2.98, p = 0.004). No other predictors were significant (see SI; 45).
Patients with BED have increased variability in wanting ratings of food
To evaluate differences in variability of behavior, we compared the trial-to-trial variability of effort maintenance and wanting in patients with BED using bootstrapped variance ratio tests (F-test). As hypothesized, in uncertain trials, patients with BED showed higher variability in wanting (Fig. 2) for food (F34,22 = 1.48, pboot = 0.024), but not money (F34,22 = 0.97, pboot = 0.56). No differences were observed for the subBED group (food: F21,22 = 0.95, pboot = 0.62, money: F21,22 = 1.00, pboot = 0.52). In contrast to the self-reported wanting, effort maintenance did not vary differentially between groups (ps > 0.11, Table S2). In certain trials, there were no associations with wanting. However, individuals with subBED showed consistently higher variability in effort maintenance (ps < 0.001, Table S2) and patients with BED showed higher variability in money trials (F34,22 = 1.67, pboot = 0.007). Against our expectation, BMI was not correlated with individual SD of effort maintenance (r = −0.06, 95%CIboot [−0.22, 0.08]) or wanting ratings (r = 0.07, 95%CIboot [−0.08, 0.23], Table S3).
a No group differences in the standard de of the residuals in trial-wise effort maintenance. b Patients with BED show higher trial-wise SDs of food wanting ratings (F34,21 = 1.48, pboot = 0.024) compared to participants without binge eating (no BE)
BMI is associated with greater variability of NAcc cue signals
Next, we sought to replicate our previous finding of an increased variability of the NAcc reward response with a higher BMI. In line with previous findings, a higher BMI was associated with more variable NAcc cue responses across trials (b = 0.11, 95% credible interval, CI one-sided test [0.04, 0.17], BF+0 = 352.0, Fig. 3c). This association was more pronounced for small rewards (97.5%CI [−0.17, −0.01], BF10 = 5.3, Table S4)
In addition, we hypothesized that patients with BED would show a higher variability beyond the effects of BMI. The variability was numerically higher for subBED (b = 0.08, 95%CI one-sided test [−0.08, 0.20]) and BED (b = 0.10, 95%CI one-sided test [−0.04, 0.24], Fig. 3a-b), although the one-sided 95% credible intervals still included 0. Nonetheless, the one-sided posterior probability of a positive effect was 79% and 89% with moderate to large evidence (BF+0subBED = 3.4, BF+0BED = 7.4) for a positive effect, indicating that the data provide support for an additive effect of BED beyond BMI in the expected direction. Fig. 4.
a Individual distributions of NAcc cue responses ordered by BMI and split by group (no binge eating (no BE), subsyndromal binge eating disorder (subBED), and binge eating disorder (BED). b Variability of NAcc cue responses is higher in the BED group beyond differences in BMI (directed posterior probability = 89%, b = 0.10, BF+0 = 7.4). However, the undirected 95%CI [−0.06, 0.26] includes 0. Distributions of trial residuals from all participants of the groups. c Variability of NAcc cue responses is higher in participants with a higher BMI beyond binge eating groups (b = 0.11, BF+0 = 352.0) and more pronounced for money (b = 0.06 of the interaction, 95%CI [0.01, 0.11], BF10 = 2.3). d Variability of NAcc cue responses is higher in participants reporting high levels of disinhibited eating beyond differences based on BMI (b = 0.19, 95%CI [0.04, 0.33], BF+0 = 53.7). Error bars depict 95% percentiles.
a Variability of dlPFC cue responses (distributions of all trial residuals from all participants) does not differ between the binge eating groups. b Variability of dlPFC cue responses is elevated in participants with higher BMI (b = 0.08, 95%CI [0.01, 0.15], BF10 = 2.7). c Variability of dlPFC cue responses does not differ depending on three factor eating questionnaire: disinhibited eating. Error bars depict 95% percentiles
Since patients with BED did not conclusively show enhanced NAcc variability, we sought to replicate the association with disinhibited eating as assessed by the TFEQ [20]. To this end, we exchanged the group variable by categories of disinhibited eating. In line with previous findings, variability in NAcc cue responses was higher in participants with high levels of disinhibited eating across the groups (b = 0.19, 95%CI [0.04, 0.33], BF+0 = 53.7, Fig. 3d, Table S5) in addition to the variance explained by BMI as a predictor.
Crucially, including individual SDs of wanting ratings and effort maintenance from the behavioral session showed that higher SDs of wanting were nominally associated with increased variability in NAcc cue responses as indicated by the posterior probability, P(SD_Wanting > 0|data)=89% for the directed test (BF+0 = 8.4; undirected: b = 0.05, 95%CI [−0.03, 0.14], BF10 = 0.5; Table S7). In contrast, higher SDs of effort maintenance were not associated with increased variability in NAcc cue responses according to the posterior probability, P(SD_RelEffort>0|data)=12%, (BF +0 = 0.1; undirected BF10 = 0.5, Table S6).
After replicating the association of higher variability in NAcc cue responses with BMI and disinhibited eating, we next evaluated effects in the dlPFC due to its putative role in top-down control. Again, a higher BMI was associated with more variable dlPFC cue responses across trials (b = 0.08, 95%CI [0.01, 0.15], BF10 = 2.7, Fig. 4, Table S4). However, patients with BED (b = 0.04, 95%CI [−0.118, 0.188], BF10 = 0.4) and individuals with higher disinhibited eating (b = 0.04, 95%CI [−0.13, 0.21], BF10 = 0.5, Fig. 4c) showed no differences in variability of dlPFC cue responses.
Exploratory and sensitivity analyses
To verify that associations of variability in cue responses with BMI are specific and not explained by mere differences in movement or noise, we performed the same analyses in control regions of the temporal lobe that are largely independent of bottom-up or top-down control signals. There was no association of BMI with cue response variability in the toMTL (b = 0.04, 95%CI [−0.04, 0.012], BF10 = 0.3) or pMTL (b = 0.02, 95%CI [−0.06, 0.010], BF10 = 0.2) and a weak association in the aMTL (b = 0.08, 95%CI [0.00, 0.15], BF10 = 1.3). Notably, all results only changed numerically when including the SD of cue responses in the aMTL to the models, indicating that variability differences in the NAcc and dlPFC are separable from global noise in cue responses (Table S7).
Since noise may not be homogeneous across the brain, we also added the average framewise displacement as a subject-level covariate to the models, in addition to accounting for volume-wise movement. The association between BMI and NAcc variability remained significant (b = 0.07, 95%CI one-sided test [0.00, 0.15], BF+0 = 18.5, Table S8), although it reduced the association. Follow-up tests indicated that the reduction of the effect was not significant (Δb = 0.033, 95%CI [−0.08, 0.15]). All other reported associations were unchanged by these analyses (Table S8). To determine whether higher NAcc variability was specific to the cue phase, we also extended the analysis to feedback responses. Notably, the feedback response of the NAcc was also more variable with a higher BMI (b = 0.16, 95%CI one-sided test [0.10, 0.22], BF+0 = 6000.0) and for patients with subBED (b = 0.08, 95%CI one-sided test [−0.05, 0.22], BF+0 = 5.3) and BED (b = 0.10, 95%CI one-sided test [−0.04, 0.23], BF+0 = 7.3), and those results only changed marginally when including average framewise displacement, suggesting that it exceeded motion-related confounding effects (Table S8-9). To determine whether use of medication, hormonal contraception, or symptom load (i.e., number of binges in the last month might) explain the results, we first evaluated their association with variability. Of note, use of any medication beyond contraception was associated with higher variability in wanting of food reward (r = 0.24, p = 0.038) as well as NAcc cue responses (r = 0.30 p = 0.020, Table S10). Next, we included those variables as subject-level covariates to the models. Crucially, the main results only changed marginally (Tables S11), indicating that they did not drive the associations of BMI (b = 0.11, 95%CI one-sided test [0.04, 0.17], BF+0= 234.5) and disinhibited eating (b = 0.09, 95%CI one-sided test [0.03; 0.15], BF+0 = 128.3) with variability in reward responses, although the association with BED was only anecdotal (b = 0.05, 95%CI one-sided test [−0.12; 0.22], BF+0 = 2.4).
Discussion
BED is characterized by repeated episodes of BE with subjective loss of control. Even though such behavior in patients with BED could be seen as an expression of more variable food-related reward behavior compared to individuals without BED [17], most studies have so far focused primarily on potential differences in the amplitude of NAcc responses and its association with symptoms of BE. Here, we showed that variability in food wanting ratings, but not behavioral effort was increased in patients with BED compared to the control groups. We found that increased NAcc variability was positively associated with BMI, as well as TFEQ disinhibition. Against our expectations, this effect was not exacerbated by uncertain effort conditions. In addition to the effect of BMI, we also observed that patients with BED showed slightly elevated variability of NAcc responses as predicted, but the provided level of evidence was inconclusive. To conclude, our findings are the first to highlight the relevance of volatility in subjective experiences of food reward wanting in BED, whilst corroborating the connection between variability in NAcc response, BMI, and food-related disinhibition.
Previous work demonstrated increased variability in NAcc responses is associated with increased BMI and TFEQ disinhibition [20], and that BMI is associated with increased variability in reward learning [45]. Here, we replicated this association of variability in NAcc responses with BMI and disinhibition using a different task and focusing on a phase that resembles subjective value signals more narrowly: during anticipation. Whereas Kroemer, Sun [20] measured NAcc variability after milkshake receipt (consummatory response without cues), the current study shows variability in NAcc responses before participants worked for money and food rewards (i.e., by collecting reward points). Since the NAcc is thought to be involved in incentive salience and approach motivation [25, 56], this provides further indication that variability in NAcc responses may be involved in the regulation of reward-directed behavior. Although our original hypotheses focused on NAcc variability in association with BED, we did not find conclusive evidence in favor or against this hypothesis. Instead, the current results provide evidence that body weight and disinhibited eating are associated with neural adaptations in the NAcc [22], suggesting that beyond these associations, NAcc response variability does not specifically contribute to BED or subBED. Instead, individuals with BED may also show higher reward-related NAcc variability due to its comorbidity with overweight and obesity [3] and dimensional associations with disinhibition.
Indeed, our results extend the literature beyond the more commonly investigated NAcc response amplitude, which has been previously related to weight gain [57, 58] and snacking [59], but not convincingly with current BMI [60,61,62]. Opposed to signal averages, variability in neural and psychological parameters may generally allow for variability in reacting to (food) cues in the environment [63]. In addition, brain signal variability has been used to predict psychiatric treatment response, underscoring that it is more than just noise [64, 65]. Our study suggests that heightened variability in NAcc responses may be mechanistically associated with overweight and disinhibited eating.
Moreover, we find a similar association between dlPFC variability and BMI. Conceptually, this could be an indication of the variability of the reward signal of the NAcc, as the dlPFC receives information from the NAcc for cost-benefit integration [32]. Alternatively, volatility in top-down processing may additionally contribute to a predisposition for overweight, which is in line with inhibitory control deficits observed in overweight and obesity [66, 67]. However, the current study cannot directly disentangle the contribution of these top-down processes. Research directly focusing on response inhibition – for example, with the stop-signal go/no-go task – is thus required to determine whether the association between dlPFC activation and BMI is characterized by greater signal variability beyond lowered amplitudes that have been demonstrated before [16, 68, 69].
The current study could not conclusively support the hypothesis that heightened variability in NAcc responses is a specific characteristic of BE. Instead, we found that individuals with BED had a higher variability in their subjective wanting ratings for food compared to subBED or no BE. Although effort to work for reward and wanting are related [70, 71], we previously demonstrated unique correlates of variability in wanting ratings, not effort with BMI in a similar effort task [37]—despite high correlations between wanting ratings and behavior on the task [31]. One possible explanation of this difference could be that individuals with BED rely more strongly on heuristics or habitual responses to determine their willingness to expend effort [72], which would lower the influence of cost-benefit representations on effort expenditure [73] while maintaining variability in experienced benefits (i.e., wanting) of the reward. In favor of this interpretation, one study found that individuals with high BE pathology relied more strongly on reward magnitude in their effort expenditure for food rewards than individuals with low BE pathology [74]. Therefore, behavior of individuals with BED may be more driven by (external) task factors, but may show fewer behavioral fluctuations independent of task factors. Furthermore, our results could be interpreted as a weaker link between wanting and effort. Dissociations in reward processing have previously been found in substance use disorder [75], and may contribute to the experienced ambivalence towards food cues experienced by individuals with BED [76]. Therefore, variability in wanting ratings of food in BED may represent a component of subjective value that is separable from variability in effort. Since this dissociation was unexpected, future research may help understand how variable wanting ratings of food reward contribute to BED.
Despite the notable strengths of our study, the results need to be interpreted under the following considerations. First, our study did not include a measure of variability in food intake, as this would require many repeated sessions to derive good estimates. As such, only limited inferences can be made on the association between variability in NAcc responses and food intake although previous research demonstrated a correlation between variance in ad libitum food intake and variability in NAcc responses to milkshake [20]. Second, due to the controlled setting under which the experiments took place, we cannot account for variability in NAcc responses that naturally occurs over longer time periods and contexts. Two important fluctuating factors that contribute to BE are metabolic (e.g., 77, 78) and mood states (e.g., 6, 79). Our current study used standardized meals and metabolic states, and no mood induction, deliberately restricting such contributions. Naturalistic designs can better determine whether fluctuations in metabolic and mood states are associated with larger intraindividual variability in reward responses and subsequent BE and we have reported associations of BE with more variable behavior in a gamified reinforcement learning task [45]. In this regard, motivational fluctuations should also be taken into account, as they seem to generally influence reward and effort based processing [80]. Third, as only women participated in the current study, further research is required to extend the results to men. Furthermore, sex hormones may influence inter- and intra-sex variability in BE [81], which cannot be accounted for in the current study. Fourth, our control analyses indicate that a higher BMI is associated with increased movement, and that this association contributes non-significantly to the association between greater NAcc cue response variability and BMI. Although our additional analyses demonstrated that the relationship was not explained by movement, future studies may consider using additional means of correction to estimate associations between BMI and neural variability more precisely given the robust link between BMI and larger framewise displacement [82]. Finally, the current study did not account for illness duration. With regard to psychopathology, illness duration has been shown to influence neural mechanisms [83]. As we could not correct for this effect, the current study can only make general conclusions about BED and subBED [84], and future work is required to determine how these processes change as the disorder progresses. In the same vein, future research is required to determine the effect of changes brought about by experience with psychotherapeutic interventions, as treatment status was not assessed [84]. Likewise, although we explored the role of medication use, hormonal contraception and severity of binge eating and found limited evidence that they explained effects of BMI and disinhibited eating, there were associations of medication with variability in reward responses. This calls for future research to disentangle which medications are particularly relevant. Moreover, the reported comorbidities were varied (very few shared diagnoses) and occurred almost exclusively in the BED group so that we could not estimate the effects of specific mental comorbidities on variability. Therefore, larger trials including specific combinations of mental disorders are needed to determine their contribution to variability in reward responses.
To summarize, we have demonstrated that variability in NAcc responses to reward cues is associated with BMI and dietary disinhibition, and that subjective wanting of food cues was more variable in patients with BED compared to participants without BE. Combined, these findings demonstrate that trial-to-trial variability in the reward system as indexed by NAcc responses forms a mechanism that is associated with overweight and uncontrolled eating and that this association goes beyond typically investigated measures of average reward values. Future research should thus focus on how variability in reward processes interacts with other well-established factors contributing to overeating and BE, such as satiation and mood.
Data availability
Trial-wise data (estimated cue responses and behavior) is available at https://osf.io/tewpn/?view_only=e35bb7a5f9574cc4b8faa54ed2979d01
Code availability
Trial-wise data (estimated cue responses and behavior) and analysis code are available at https://osf.io/tewpn/?view_only=e35bb7a5f9574cc4b8faa54ed2979d01
References
Kessler RC, Berglund PA, Chiu WT, Deitz AC, Hudson JI, Shahly V, et al. The prevalence and correlates of binge eating disorder in the World Health Organization World Mental Health Surveys. Biol Psychiatry. 2013;73:904–14
Keski-Rahkonen A. Epidemiology of binge eating disorder: prevalence, course, comorbidity, and risk factors. Curr Opin Psychiatry. 2021;34:525–31
Agüera Z, Lozano-Madrid M, Mallorquí-Bagué N, Jiménez-Murcia S, Menchón JM, Fernández-Aranda F. A review of binge eating disorder and obesity. Neuropsychiatrie. 2021;35:57–67
Stein RI, Kenardy J, Wiseman CV, Dounchis JZ, Arnow BA, Wilfley DE. What’s driving the binge in binge eating disorder? A prospective examination of precursors and consequences. Int J Eat Disord. 2007;40:195–203
Mathes WF, Brownley KA, Mo X, Bulik CM. The biology of binge eating. Appetite. 2009;52:545–53
Svaldi J, Werle D, Naumann E, Eichler E, Berking M. Prospective associations of negative mood and emotion regulation in the occurrence of binge eating in binge eating disorder. J Psychiatr Res. 2019;115:61–68
Williamson DA, White MA, York-Crowe E, Stewart TM. Cognitive-behavioral theories of eating disorders. Behav Modif. 2004;28:711–38
Svaldi J, Tuschen-Caffier B, Peyk P, Blechert J. Information processing of food pictures in binge eating disorder. Appetite. 2010;55:685–94
Schmitz F, Naumann E, Trentowska M, Svaldi J. Attentional bias for food cues in binge eating disorder. Appetite. 2014;80:70–80
Stojek M, Shank LM, Vannucci A, Bongiorno DM, Nelson EE, Waters AJ, et al. A systematic review of attentional biases in disorders involving binge eating. Appetite. 2018;123:367–89
Berridge KC. ‘Liking’ and ‘wanting’ food rewards: Brain substrates and roles in eating disorders. Physiol Behav. 2009;97:537–50
Schmitz F, Naumann E, Biehl S, Svaldi J. Gating of attention towards food stimuli in binge eating disorder. Appetite. 2015;95:368–74
Leenaerts N, Jongen D, Ceccarini J, Van Oudenhove L, Vrieze E. The neurobiological reward system and binge eating: A critical systematic review of neuroimaging studies. Int J Eat Disord. 2022;55:1421–58
Schienle A, Schäfer A, Hermann A, Vaitl D. Binge-eating disorder: Reward sensitivity and brain activation to images of food. Biol Psychiatry. 2009;65:654–61
Weygandt M, Schaefer A, Schienle A, Haynes J-D. Diagnosing different binge-eating disorders based on reward-related brain activation patterns. Hum Brain Mapp. 2012;33:2135–46
Lavagnino L, Arnone D, Cao B, Soares JC, Selvaraj S. Inhibitory control in obesity and binge eating disorder: A systematic review and meta-analysis of neurocognitive and neuroimaging studies. Neurosci Biobehav Rev. 2016;68:714–26
Neuser MP, Kühnel A, Svaldi J, Kroemer NB. Beyond the average: The role of variable reward sensitivity in eating disorders. Physiol Behav. 2020;223:112971
Morton GJ, Cummings DE, Baskin DG, Barsh GS, Schwartz MW. Central nervous system control of food intake and body weight. Nature. 2006;443:289–95
Ferrario CR, Labouèbe G, Liu S, Nieh EH, Routh VH, Xu S, et al. Homeostasis meets motivation in the battle to control food intake. J Neurosci. 2016;36:11469–81
Kroemer NB, Sun X, Veldhuizen MG, Babbs AE, de Araujo IE, Small DM. Weighing the evidence: Variance in brain responses to milkshake receipt is predictive of eating behavior. Neuroimage. 2016;128:273–83
Wang Y, Tang L, Wang M, Wu G, Li W, Wang X, et al. The role of functional and structural properties of the nucleus accumbens subregions in eating behavior regulation of bulimia nervosa. Int J Eat Disord. 2023;56:2084–95
Samara A, Li Z, Rutlin J, Raji CA, Sun P, Song S-K, et al. Nucleus accumbens microstructure mediates the relationship between obesity and eating behavior in adults. Obesity. 2021;29:1328–37
Domingo-Rodriguez L, Ruiz de Azua I, Dominguez E, Senabre E, Serra I, Kummer S, et al. A specific prelimbic-nucleus accumbens pathway controls resilience versus vulnerability to food addiction. Nat Commun. 2020;11:782
Floresco SB. The nucleus accumbens: an interface between cognition, emotion, and action. Annu Rev Psychol. 2015;66:25–52
Kroemer NB, Guevara A, Ciocanea Teodorescu I, Wuttig F, Kobiella A, Smolka MN. Balancing reward and work: Anticipatory brain activation in NAcc and VTA predict effort differentially. Neuroimage. 2014;102:510–9
Zald DH, Treadway MT. Reward processing, neuroeconomics, and psychopathology. Annu Rev Clin Psychol. 2017;13:471–95
Niv Y, Daw ND, Joel D, Dayan P. Tonic dopamine: Opportunity costs and the control of response vigor. Psychopharmacology (Berl). 2007;191:507–20
Kroemer NB, Burrasch C, Hellrung L (2016): To work or not to work: Neural representation of cost and benefit of instrumental action. In: Studer B, Knecht S, editors. Progress in Brain Research: Elsevier, pp125-57
Meyniel F, Sergent C, Rigoux L, Daunizeau J, Pessiglione M. Neurocomputational account of how the human brain decides when to have a break. Proc Natl Acad Sci. 2013;110:2641–6
Meyniel F, Safra L, Pessiglione M. How the brain decides when to work and when to rest: Dissociation of implicit-reactive from explicit-predictive computational processes. PLoS Comput Biol. 2014;10:e1003584
Neuser MP, Teckentrup V, Kühnel A, Hallschmid M, Walter M, Kroemer NB. Vagus nerve stimulation boosts the drive to work for rewards. Nat Commun. 2020;11:3555
Chong TTJ, Apps M, Giehl K, Sillence A, Grima LL, Husain M. Neurocomputational mechanisms underlying subjective valuation of effort costs. PLoS Biol. 2017;15:e1002598
Soutschek A, Tobler PN. Causal role of lateral prefrontal cortex in mental effort and fatigue. Hum Brain Mapp. 2020;41:4630–40
Lowet AS, Zheng Q, Matias S, Drugowitsch J, Uchida N. Distributional reinforcement learning in the brain. Trends Neurosci. 2020;43:980–97
Findling C, Skvortsova V, Dromnelle R, Palminteri S, Wyart V. Computational noise in reward-guided learning drives behavioral variability in volatile environments. Nat Neurosci. 2019;22:2066–77
Peters J, Büchel C. The neural mechanisms of inter-temporal decision-making: understanding variability. Trends Cogn Sci. 2011;15:227–39
van den Hoek Ostende MM, Neuser MP, Teckentrup V, Svaldi J, Kroemer NB. Can’t decide how much to EAT? Effort variability for reward is associated with cognitive restraint. Appetite. 2021;159:105067
Myers KP, Majewski M, Schaefer D, Tierney A. Chronic experience with unpredictable food availability promotes food reward, overeating, and weight gain in a novel animal model of food insecurity. Appetite. 2022;176:106120
Ormaechea P, Boakes RA. Unpredictability of access to a high fat/high sugar food can increase rats’ intake. Physiol Behav. 2023;266:114182
Kreisler AD, Garcia MG, Spierling SR, Hui BE, Zorrilla EP. Extended vs. brief intermittent access to palatable food differently promote binge-like intake, rejection of less preferred food, and weight cycling in female rats. Physiol Behav. 2017;177:305–16
Hilbert A, Tuschen-Caffier B Eating disorder examination: Deutschsprachige Übersetzung. Verlag für Psychotherapie; 2006
Wittchen H-U, Wunderlich U, Gruschwitz S, Zaudig M SKID I. Strukturiertes Klinisches Interview für DSM-IV. Achse I: Psychische Störungen. Interviewheft und Beurteilungsheft. Eine deutschsprachige, erweiterte Bearb. d. amerikanischen Originalversion des SKID I. Göttingen. Hogrefe; 1997
Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res. 1985;29:71–83
Müller FK, Teckentrup V, Kühnel A, Ferstl M, Kroemer NB. Acute vagus nerve stimulation does not affect liking or wanting ratings of food in healthy participants. Appetite. 2022;169:105813
Kühnel A, Zietz J, Theuer J, Neuser MP, Teckentrup V, Müller FK, et al. Dense sampling of choices links high learning rates to obesity and low reward sensitivity to binge eating. medRxiv. Article 2025.2006.2019.25329936 [Preprint]. 2025 https://www.medrxiv.org/content/10.1101/2025.06.19.25329936v1
Schulz C, Klaus J, Peglow F, Ellinger S, Kühnel A, Walter M, et al. Blunted anticipation but not consummation of food rewards in depression. Cell Rep Med. 2026;7:102796
Kühnel A, Teckentrup V, Neuser MP, Huys QJM, Burrasch C, Walter M, et al. Stimulation of the vagus nerve reduces learning in a go/no-go reinforcement learning task. Eur Neuropsychopharmacol. 2020;35:17–29
Esteban O, Markiewicz CJ, Blair RW, Moodie CA, Isik AI, Erramuzpe A, et al. fMRIPrep: a robust preprocessing pipeline for functional MRI. Nat Methods. 2019;16:111–6
Desikan RS, Ségonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage. 2006;31:968–80
Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67:48
Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest Package: Tests in linear mixed effects models. J Stat Softw. 2017;82:1–26
Bürkner P-C. brms: An R package for Bayesian multilevel models using stan. J Stat Softw. 2017;80:28
R Core Team R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2024
Wickham H ggplot2: Elegant graphics for data analysis. Springer-Verlag New York; 2016
Kay M. ggdist: Visualizations of distributions and uncertainty in the grammar of graphics. IEEE Trans Vis Comput Graph. 2024;30:414–24
Salamone JD, Correa M. The mysterious motivational functions of mesolimbic dopamine. Neuron. 2012;76:470–85
Demos KE, Heatherton TF, Kelley WM. Individual differences in nucleus accumbens activity to food and sexual images predict weight gain and sexual behavior. J Neurosci. 2012;32:5549–52
Geha PY, Aschenbrenner K, Felsted J, O’Malley SS, Small DM. Altered hypothalamic response to food in smokers1 2 3. Am J Clin Nutr. 2013;97:15–22
Lawrence NS, Hinton EC, Parkinson JA, Lawrence AD. Nucleus accumbens response to food cues predicts subsequent snack consumption in women and increased body mass index in those with reduced self-control. Neuroimage. 2012;63:415–22
García-García I, Horstmann A, Jurado MA, Garolera M, Chaudhry SJ, Margulies DS, et al. Reward processing in obesity, substance addiction and non-substance addiction. Obes Rev. 2014;15:853–69
Ziauddeen H, Farooqi IS, Fletcher PC. Obesity and the brain: how convincing is the addiction model?. Nat Rev Neurosci. 2012;13:279–86
Brooks SJ, Cedernaes J, Schiöth HB. Increased prefrontal and parahippocampal activation with reduced dorsolateral prefrontal and insular cortex activation to food images in obesity: A meta-analysis of fMRI studies. PLoS ONE. 2013;8:e60393
Sanborn AN, Zhu J-Q, Spicer J, León-Villagrá P, Castillo L, Falbén JK, et al. Noise in cognition: Bug or feature?. Perspect Psychol Sci. 2025;20:572–89
Mansson KNT, Waschke L, Manzouri A, Furmark T, Fischer H, Garrett DD. Moment-to-Moment brain signal variability reliably predicts psychiatric treatment outcome. Biol Psychiatry. 2022;91:658–66
Tsikonofilos K, Kumar A, Ampatzis K, Garrett DD, Mansson KNT. The promise of investigating neural variability in psychiatric disorders. Biol Psychiatry. 2025;98:195–207
Houben K, Nederkoorn C, Jansen A. Eating on impulse: The relation between overweight and food-specific inhibitory control. Obesity. 2014;22:E6–E8
Svaldi J, Naumann E, Trentowska M, Schmitz F. General and food-specific inhibitory deficits in binge eating disorder. Int J Eat Disord. 2014;47:534–42
Alatorre-Cruz GC, Downs H, Hagood D, Sorensen ST, Williams DK, Larson-Prior L. Effect of obesity on inhibitory control in preadolescents during stop-signal task. An event-related potentials study. Int J Psychophysiol. 2021;165:56–67
Chen S, Jia Y, Woltering S. Neural differences of inhibitory control between adolescents with obesity and their peers. Int J Obes. 2018;42:1753–61
Lopez-Persem A, Rigoux L, Bourgeois-Gironde S, Daunizeau J, Pessiglione M. Choose, rate or squeeze: Comparison of economic value functions elicited by different behavioral tasks. PLoS Comput Biol. 2017;13:e1005848
Vinckier F, Rigoux L, Kurniawan IT, Hu C, Bourgeois-Gironde S, Daunizeau J, et al. Sour grapes and sweet victories: How actions shape preferences. PLoS Comput Biol. 2019;15:e1006499
Hartogsveld B, Quaedflieg CWEM, van Ruitenbeek P, Smeets T. Decreased putamen activation in balancing goal-directed and habitual behavior in binge eating disorder. Psychoneuroendocrinology. 2022;136:105596
Hardwick RM, Forrence AD, Krakauer JW, Haith AM. Time-dependent competition between goal-directed and habitual response preparation. Nat Hum Behav. 2019;3:1252–62
Racine SE, Horvath SA, Brassard SL, Benning SD. Effort expenditure for rewards task modified for food: A novel behavioral measure of willingness to work for food. Int J Eat Disord. 2019;52:71–78
Ostafin BD, Marlatt GA, Troop-Gordon W. Testing the incentive-sensitization theory with at-risk drinkers: Wanting, liking, and alcohol consumption. Psychol Addict Behav. 2010;24:157–62
Leehr EJ, Schag K, Brinkmann A, Ehlis A-C, Fallgatter AJ, Zipfel S, et al. Alleged approach-avoidance conflict for food stimuli in binge eating disorder. PLoS ONE. 2016;11:e0152271
Manasse SM, Espel HM, Forman EM, Ruocco AC, Juarascio AS, Butryn ML, et al. The independent and interacting effects of hedonic hunger and executive function on binge eating. Appetite. 2015;89:16–21
Haedt-Matt AA, Keel PK. Hunger and binge eating: A meta-analysis of studies using ecological momentary assessment. Int J Eat Disord. 2011;44:573–8
Leehr EJ, Krohmer K, Schag K, Dresler T, Zipfel S, Giel KE. Emotion regulation model in binge eating disorder and obesity – a systematic review. Neurosci Biobehav Rev. 2015;49:125–34
Hewitt SRC, Norbury A, Huys QJM, Hauser TU. Day-to-day fluctuations in motivation drive effort-based decision-making. Proc Natl Acad Sci. 2025;122:e2417964122
Mikhail ME, Anaya C, Culbert KM, Sisk CL, Johnson A, Klump KL. Gonadal hormone influences on sex differences in binge eating across development. Curr Psychiatry Rep. 2021;23:74
Beyer F, Prehn K, Wüsten KA, Villringer A, Ordemann J, Flöel A, et al. Weight loss reduces head motion: Revisiting a major confound in neuroimaging. Hum Brain Mapp. 2020;41:2490–4
Han S, Zheng R, Li S, Zhou B, Jiang Y, Fang K, et al. Altered structural covariance network of nucleus accumbens is modulated by illness duration and severity of symptom in depression. J Affect Disord. 2023;324:334–40
Frank GKW, Favaro A, Marsh R, Ehrlich S, Lawson EA. Toward valid and reliable brain imaging results in eating disorders. Int J Eat Disord. 2018;51:250–61
Acknowledgements
We thank Dana Wentz, Jacob Schwab, Juliane Zietz, Jennifer Piloth, Lilith Irtel von Brenndorff, Fee Arnold, and Vincent Koepp for help with data acquisition as well as Vanessa Teckentrup for support in preprocessing of the MRI data. The study was supported by the Else Kröner-Fresenius Stiftung grants 2017_A67 and 2024_EKEA.149, and DFG KR 4555/7-1, KR 4555/9-1, and KR 4555/10-1. We acknowledge support by the Open Access Publishing Fund of the University of Tübingen
Funding
Open Access funding enabled and organized by Projekt DEAL
Author information
Authors and Affiliations
Department of Psychiatry and Psychotherapy, Tübingen Center for Mental Health, University of Tübingen, Tübingen, Germany
Mechteld M. van den Hoek Ostende, Monja P. Neuser, Thomas Dresler & Nils B. Kroemer
Department of Psychology, Tübingen Center for Mental Health, University of Tübingen, Tübingen, Germany
Mechteld M. van den Hoek Ostende & Jennifer Svaldi
Section of Medical Psychology, Department of Psychiatry and Psychotherapy, University Hospital Bonn, University of Bonn, Bonn, Germany
Anne Kühnel & Nils B. Kroemer
German Center for Mental Health (DZPG), partner site Tübingen, Tübingen, Germany
Anne Kühnel, Thomas Dresler, Jennifer Svaldi & Nils B. Kroemer
LEAD Graduate School & Research Network, University of Tübingen, Tübingen, Germany
Thomas Dresler
German Center for Diabetes Research (DZD), Neuherberg, Germany
Nils B. Kroemer
Authors
- Mechteld M. van den Hoek OstendeView author publications
Search author on:PubMed Google Scholar
- Anne KühnelView author publications
Search author on:PubMed Google Scholar
- Monja P. NeuserView author publications
Search author on:PubMed Google Scholar
- Thomas DreslerView author publications
Search author on:PubMed Google Scholar
- Jennifer SvaldiView author publications
Search author on:PubMed Google Scholar
- Nils B. KroemerView author publications
Search author on:PubMed Google Scholar
Contributions
NBK and JS were responsible for the study concept and design. MPN & MH collected data under supervision by NBK. NBK conceived the method and AK & MH processed the data. AK & MH performed the data analysis and NBK contributed to analyses. MH, AK, & NBK wrote the manuscript. All authors contributed to the interpretation of findings, provided critical revision of the manuscript for important intellectual content and approved the final version for publication
Ethics declarations
Competing interests
The overarching study in which this research took place included a research cooperation with Boehringer Ingelheim to analyze ghrelin levels. These data are not part of the current manuscript
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations
Supplementary information
Supplemental Material (download DOCX )
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
van den Hoek Ostende, M.M., Kühnel, A., Neuser, M.P. et al. Obesity is associated with greater variability of reward signals in the nucleus accumbens.
Transl Psychiatry16, 370 (2026). https://doi.org/10.1038/s41398-026-04172-6
Received:04 November 2025
Revised:18 May 2026
Accepted:08 June 2026
Published:20 July 2026
Version of record:20 July 2026
DOI
:https://doi.org/10.1038/s41398-026-04172-6
