Humans have evolved with the capacity to store energy in periods of food abundance, preparing for potential periods of scarcity (Rosenbaum and Leibel, 2010). Environmental and societal changes have made calorie-rich foods that are high in fat and sugar widely available. Combined with sedentary lifestyles and stressful life experiences (Swinburn and Egger, 2004), this has contributed to overeating and weight gain, which is now a major societal concern.

Female beauty standards, which emphasize an often unattainable slender and thin body (Engeln-Maddox, 2006), have a powerful influence on women’s self-evaluation (Fairburn, 2008). Sociocultural pressure to conform to these standards may lead to the overvaluation of weight and shape and body image dissatisfaction, resulting in feelings of shame, anger, anxiety or disgust. In an attempt to regulate these difficult emotions, individuals may engage in multiple eating, weight and shape-related behaviors, such as restrictive dieting, excessive exercise, misuse of laxatives, and overeating. They may also rely on several cognitive strategies, such as self-criticism, avoidance and rumination (Duarte et al., 2016; Sharpe et al., 2018). However, these mechanisms maintain disordered eating habits, exacerbate negative affect and are linked to psychopathology (Mellings and Alden, 2000; Nolen-Hoeksema, 2000; Gilbert and Irons, 2005; Verhaeghen et al., 2005; Dunkley and Grilo, 2007; Michael et al., 2007; Surrence et al., 2009; Barnes et al., 2010; Erskine and Georgiou, 2010; Rawal et al., 2010; Goss, 2011; Cowdrey and Park, 2012; Inzlicht and Schmeichel, 2012; Seidel et al., 2016; Lazarus and Shahar, 2018).

Self-criticism refers to a negative and harsh judgment of oneself based on physical appearance, behavior or personality (Gilbert et al., 2004). There have been only a few studies on the neural correlates of self-criticism, and the ones conducted have reported activation in the anterior cingulate cortex (ACC) and lateral prefrontal cortex (PFC) regions (Longe et al., 2010; Doerig et al., 2014), which are associated with executive function, error monitoring and behavioral inhibition (Wittfoth et al., 2009; Nejati et al., 2021). A heightened activation of the insula, amygdala, putamen, caudate nucleus and orbitofrontal cortex (OFC) have also been reported (Longe et al., 2010; Doerig et al., 2014, 2016), indicating difficulties in emotional and reward processing when exposed to self-critical stimuli.

Avoidance entails a refusal to experience unpleasant thoughts and feelings with attempts to avoid them (Hayes, 2004). Individuals who intentionally restrict their caloric intake to lose or maintain their weight tend to suppress thoughts about palatable food (Van Gucht et al., 2014). However, this self-control strategy may backfire and lead to an increase in food consumption (Erskine and Georgiou, 2010). Neuroimaging studies have shown the involvement of the dorsolateral PFC (dlPFC) when trying to suppress such thoughts (Mitchell et al., 2007; Carew et al., 2015; Aso et al., 2016) and the ACC in the detection of conflicts during thought suppression (Mitchell et al., 2007). In a recent study, restrained eaters who suppressed thoughts about food showed reduced activity in the dorsal ACC in association with the choice for more high-calorie foods. This impacts conflict monitoring, suggesting that individuals’ self-control resources may be depleted, leading to food choices incongruent with their restriction goals (Zhang et al., 2021). Furthermore, high-trait chocolate cravers exhibited higher activation in the putamen and ventral striatum than those with a low-trait when exposed to chocolate images, indicating the involvement of reward-related striatal areas in craving. Interestingly, high-trait chocolate cravers reported a decreased in chocolate-related thoughts after receiving suppression instructions; however there was no differential neural activity between high-trait and low-trait chocolate cravers (Miedl et al., 2018). The OFC is also implicated in the devaluation of chocolate when frequent chocolate eaters have to avoid it (Lender et al., 2023).

Rumination is a repetitive cognitive process that involves focusing on the causes, symptoms and consequences of distress (Nolen-Hoeksema, 1991). Individuals who engage in ruminative thinking about their eating, weight and body shape experience negative emotions, which can maintain eating psychopathology (Rawal et al., 2010; Cowdrey and Park, 2012; Seidel et al., 2016). An fMRI study that induced rumination by asking participants to recall negative autobiographical memories showed activation in the subgenual ACC and medial PFC (Kross et al., 2009). Cooney et al. (2010) also found that rumination recruited limbic and prefrontal regions, such as OFC, ACC, dlPFC, and amygdala. Additionally, body and weight-related rumination was associated with activation in the ventral striatum, observed in patients with anorexia nervosa (Seidel et al., 2018; Erdman et al., 2020).

As opposed to these defensive cognitive processes, research has found the benefits of self-reassurance and self-compassion. These are linked with the ability to be more understanding, warm, encouraging and supportive toward oneself when facing setbacks (Gilbert et al., 2004, 2017). When individuals are exposed to signals and stimuli such as touching, stroking, holding, soft voice tone, and facial expressions, this activates the soothing affect regulation system and the parasympathetic system, inducing feelings of safeness and contentment (Gilbert, 2014; Carter et al., 2017; Kim et al., 2020b). Research on this topic has highlighted the link between self-reassurance and self-compassion with wellbeing and interpersonal functioning (MacBeth and Gumley, 2012; Yarnell and Neff, 2013) and a reduction in psychopathology (Raes, 2010; MacBeth and Gumley, 2012; Krieger et al., 2013; Petrocchi et al., 2019). Self-compassion can help address negative body image, shame, self-criticism and self-disgust in individuals with eating difficulties (Adams and Leary, 2007; Breines et al., 2014; Ferreira et al., 2014; Kelly and Carter, 2015; Braun et al., 2016; Serpell et al., 2020; Marques et al., 2021). It can also contribute to weight management and appropriate nutrition through healthier eating behaviors (Albertson et al., 2015; Sirois et al., 2015; Rahimi-Ardabili et al., 2018). A recent meta-analysis of compassion with sixteen fMRI studies identified several brain regions related to compassion—middle frontal gyrus, bilateral inferior frontal gyrus (IFG), bilateral insula, ACC, medial frontal gyrus, and basal ganglia/thalamus—which are involved in reward and emotional processing (Kim et al., 2020a).

Taken together, current evidence of the brain regions implicated in self-criticism, avoidance, rumination, and self-reassurance or self-compassion suggests a possible overlap, with consistent activations observed in dorsal and ventral PFC, ACC, insula, dorsal striatum (caudate and putamen), and OFC. However, there is a paucity of studies specifically investigating these cognitive strategies related to eating, weight and body image concerns. To address this gap, in the present study, a new fMRI paradigm was designed to investigate the neural correlates of self-criticism, avoidance, rumination and self-reassurance related to eating psychopathology. During the experimental task, participants were presented with audio sentences related to each cognitive process with eating, weight, and body-related content, while viewing food and body images.

We aimed to investigate whether there were differences in neural activity involved in self-criticism, avoidance, rumination and self-reassurance scripts and their relation to behavioral profiles. We also sought to explore whether such strategies lead to differential patterns of brain activation. Based on previous literature, we expected that self-criticism, avoidance, rumination, and self-reassurance would be associated with the recruitment of superior and inferior prefrontal areas, OFC, dorsal striatum and regions of the salience network, e.g., the insula and ACC. Specifically for the self-reassurance and self-criticism contrast, we expected to observe activation changes in lateral PFC, ACC, caudate, and insula, similar to previous studies (Longe et al., 2010; Kim et al., 2020b). In sum, we aim to identify the neural circuitry underlying distinct but related cognitive strategies featuring self-criticism, avoidance, rumination and self-reassurance regarding eating, weight and body image concerns.

The present study aimed to investigate the neural underpinnings of cognitive strategies of self-criticism, avoidance, rumination and self-reassurance related to eating and body concerns using a novel fMRI task. Also, we further examined the associations between neural correlates of each condition with self-reported psychological measures.

The findings showed significant activations in the PFC (particularly SFG and IFG), insula and basal ganglia (putamen and caudate) for all conditions. These results are consistent with previous literature on the neural correlates of defensive strategies and self-reassurance/self-compassion (Longe et al., 2010; Brühl et al., 2014; Doerig et al., 2014; Carew et al., 2015). Contrast analysis between conditions did not reveal any significantly different activations, indicating that a similar neural network is activated at similar levels, as captured by GLM analysis, across all conditions. However, each condition was found to have a distinct pattern of processing in each area of interest, as shown by the RSA. While the standard GLM compares mean activations across conditions, RSA captures multivariate patterns of neural activity and compares whether those patterns are similar or dissimilar across conditions (Kriegeskorte, 2008). This reveals that self-criticism, rumination, avoidance and self-reassurance may be encoded differently over time in the same brain region, underlying distinct mental operations or cognitive demands.

Contrary to our expectation, there were no significant activations when contrasting self-criticism with self-reassurance, which is different from what previous studies have shown (Longe et al., 2010; Kim et al., 2020b). It is possible that the same neural regions process these strategies along a dimensional spectrum. Another possible reason for this divergence might be related to differences in experimental designs. While previous studies asked participants to imagine being self-critical or self-reassuring and what those thoughts would be, our task provided participants with predetermined content for these thoughts. Our aim with this design choice was to reduce variability in task performance and improve the overall validity of the experimental task.

Our task activated several key frontal regions, including dlPFC, ventrolateral PFC (VLPFC) and OFC, that have been implicated in inhibitory control (Aron et al., 2004; Phillips et al., 2008) and cognitive emotion regulation (Goldin et al., 2008; Ochsner et al., 2012). The dlPFC is responsible for downregulating the emotional response by inhibiting limbic regions and insula (Beauregard, 2007; Ochsner and Gross, 2008; Ochsner et al., 2012), while the vlPFC is responsible for emotional appraisal and connects directly with the dlPFC to regulate emotion via the anterior middle cingulate gyrus (Kohn et al., 2014). In line with Longe et al. (2010) findings, we found a positive correlation between inadequate self scores and dlPFC activation during both self-criticism and avoidance conditions, which suggests that highly self-critical individuals tend to engage more in error processing and inhibitory processes (Wittfoth et al., 2009; Nejati et al., 2021). This attempt to correct or solve an error and difficulties in inhibiting the repetitive cycle of thoughts could also explain the association found between self-reported rumination and the increased activation of the dlPFC in the rumination condition.

Activation of the OFC may reflect the important role of this region in emotion processing and subjective emotional experiences related to food and body stimuli. OFC and the orbitofrontal part of the IFG are both connected to the supracallosal ACC, areas related to punishment and reward, meaning that OFC computes the reward value and then sends the information to the ACC to modulate goal-related actions (Rolls et al., 2020a). The OFC activation in self-reassurance condition was negatively correlated with self-reassurance scores, indicating that difficulties in calming the self might require additional OFC recruitment in an attempt to modulate the affective state. This finding could have significant implications for therapeutic interventions, where promoting a self-reassured dialog may help in emotion processing (Gilbert and Procter, 2006).

We found activation in the anterior insula and ACC, which are known to be two hubs of the salience network. The anterior insula receives and integrates interoceptive-autonomic information to, along with ACC, guide individual responses and decisions (Seeley et al., 2007; Craig, 2009; Seeley, 2019). The co-activation of these regions occurs regardless of whether the salient stimuli are internal or external, and irrespective of their negative or positive valence (Bartra et al., 2013). Several studies have shown altered connectivity in the salience network in patients with psychiatric disorders like anorexia and bulimia nervosa, binge eating, schizophrenia, anxiety and depression (White et al., 2010; Pannekoek et al., 2013; Mcfadden et al., 2014; Sha et al., 2019; Stopyra et al., 2019). Therefore, it would be interesting to conduct the present task in patients with eating disorders or obesity to examine the activity in the salience network regions under cognitive processes related to eating, weight and body image. Interestingly, Longe et al. (2010) reported activation in insula only for the self-reassurance condition, but not for self-criticism. However, Doerig et al. (2014), which used individually tailored unpleasant stimuli, found insula activation in self-criticism. In our study, the anterior insula is found to be recruited in all four conditions, which suggests the salience of food and body-related stimuli to the self, thus greater emotional monitoring is involved. Nonetheless, the correlational analyses revealed a significant negative correlation between insula activation in the self-reassurance condition and self-compassion scores. Self-compassion involves being sensitive to our own suffering and having a commitment to alleviating it (Gilbert, 2014), and the role of the insula in the perception of the body’s internal state may contribute to the embodied nature of compassion.

The results also show a significant activation in the basal ganglia, specifically bilateral putamen and the left caudate nucleus. The dorsal striatum activation is known for its activation in studies related to the processing of reward and punishment processing (Pagnoni et al., 2002; Delgado et al., 2003). As also suggested by Longe et al. (2010), the defensive strategies elicited in this study may be associated with a form of self-punishment, as strategies focused on food and body concerns can induce feelings of inferiority and self-deprecation. However, we also found that these structures were activated in the self-reassurance condition, suggesting that negative emotional states due to defensive strategies may at least overlap the positive reward associated with a self-reassured response. Neurobehavioral correlational analyses revealed that individuals who scored higher on food thought suppression exhibited higher activation in the left putamen in the self-reassurance condition. This finding may indicate that when one repeatedly attempts to control their thoughts about food to control food intake, the elicited self-reassurance response (a new habitual response to be acquired) may compete with cognitive control processes, and thus enhance putamen activation (Robertson et al., 2015). Conversely, when individuals higher in self-compassion engage in a self-reassured dialog, they may not appraise the situation as interfering and the task may not require as much mental effort and emotional regulation, diminishing putamen activation (Otto et al., 2014; Robertson et al., 2015; André et al., 2019).

We observed a distinct pattern of lateralization during the execution of the task. The activated regions were predominantly located in the left hemisphere, which is strongly associated with language processing and inner speech production (Knecht, 2000; Morin and Hamper, 2012). As we asked participants to engage in internal dialog, this pattern of activation suggests that language processing plays a crucial role in the cognitive mechanisms engaged in our experimental task.

In contrast to the 1-back task, the four cognitive conditions revealed stronger activation in the SFG, IFG, and OFC, while the remaining brain areas of interest were equally involved in both cognitive strategies and the working memory task. The stronger activation of the PFC during the cognitive strategies is consistent with previous research that suggests the involvement of higher-level cortical structures in emotion regulation (Dixon et al., 2017; Berboth and Morawetz, 2021). This finding suggests that specific brain regions, particularly areas of the PFC, have a distinct role in cognitive strategies, which supports the use of this paradigm to assess eating and body-related cognitive strategies.

The present study may be limited by its relatively low sample size and it is essential to interpret the reported findings in consideration of this factor, and in particular that replication is needed. This is mitigated by the fact that fMRI studies have effect sizes which still allowed us in this case to employ random effects analysis. Moreover, we implemented correction for multiple comparisons. Nevertheless, we acknowledge that for generalization future studies with larger sample sizes will be needed. Also, our results are confined to a population of healthy individuals. Furthermore, the task involves participants evaluating their emotional states, implicating the monitoring of their internal states, which could influence neural activity. The correlation analysis was mainly exploratory, and while it provides valuable insights, further research is warranted to replicate the results observed in this study. The assessment of participants’ engagement in the cognitive processes is an inherent limitation which was mitigated by online experiential debriefing, where participants rated their performance in the engagement task.

Considering the tendency to use defensive cognitive strategies to cope with difficult emotions (Aldao et al., 2010; Prefit et al., 2019; Rodrigues et al., 2022), it would be of particular interest in future studies to test this paradigm with a sample of individuals with eating difficulties. Such study could lead to a deeper understanding of these self-relating cognitive processes and highlight potential differences between self-reassurance and defensive strategies. Moreover, it could provide better insight into the associations between eating psychopathology and brain activations found in our study. It would be interesting to examine the neural impact of a compassion-based intervention on individuals with eating disorders (Goss and Allan, 2010, 2014; Goss, 2011). Compassion approaches, designed to address self-criticism and shame, and promote the activation of the soothing and contentment system, could potentially alter the response patterns of the neural circuitry underlying those processes.

The present study is the first to examine the neural correlates of self-relevant cognitive processes such as self-criticism, avoidance, rumination and self-reassurance linked to food and body concerns. The four cognitive processes activated a similar set of brain regions such as superior and inferior frontal gyri, OFC, insula, dorsal striatum, and ACC, but with a differential pattern across conditions as shown by RSA. Neurobehavioral correlational analyses revealed quite distinct patterns across conditions, supporting the dissimilarity analysis. These findings contribute to a growing body of research on these regulation processes and encourage future studies to better understand how these processes differ between them.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by the Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Coimbra. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

CM: Formal analysis, Conceptualization, Investigation, Visualization, Writing – original draft. AS: Methodology, Resources, Software, Validation, Visualization, Writing – review and editing. JC: Conceptualization, Investigation, Writing – review and editing. JD: Investigation, Methodology, Software, Writing – review and editing. PC: Conceptualization, Investigation, Supervision, Writing – review and editing. KG: Investigation, Supervision, Writing – review and editing. AP: Investigation, Supervision, Validation, Writing – review and editing. MC-B: Conceptualization, Data curation, Funding acquisition, Investigation, Project administration, Supervision, Validation, Visualization, Writing – review and editing.