Introduction

There are relatively few evidence-based treatments and high relapse rates across eating disorder (ED) diagnoses, in part due to a limited understanding of how ED symptoms are maintained [1,2,3]. One symptom commonly endorsed by patients with an ED that is hypothesized to maintain behaviors over time is the subjective feeling of fatness [4]. Feeling fat refers to a subjective feeling of being overweight that does not always correspond to actual body weight [4, 5]. Despite widespread inclusion in ED assessments and theoretical models [4], limited empirical work has explored this construct. For example, Enhanced Cognitive Behavior Therapy includes a section on addressing feeling fat in treatment [4], which suggests that one’s experience of feeling fat is related to the experience of uncomfortable emotions or bodily sensations (e.g., feeling hot, full, depressed). Patients are asked to identify and track what emotions or bodily sensations they are really feeling in moments of feeling fat, such as depression or fullness [4]. However, of concern is the fact that there is currently limited empirical support for this widely used clinical recommendation. Expanding upon the literature regarding correlates of feeling fat is, therefore, crucial to either support current clinical treatment recommendations for feeling fat or provide evidence-based research for new treatment recommendations, especially as feeling fat holds significant clinical relevance. For example, one study found that decreases in feeling fat over the course of treatment were related to weight maintenance post-treatment in an anorexia nervosa sample [7]. However, research is needed to identify what factors relate to feeling fat and how feeling fat may relate to altered body awareness and affective features of EDs. Altogether, a deeper empirical understanding of feeling fat would yield insights relevant to theoretical models of EDs [4] and may inform the development of treatment approaches to target this symptom.

Research shows that perception of one’s own body weight, as opposed to actual weight status, relates to decreased psychological well-being [8]. Feeling fat is conceptualized as a distinct component of body image that is commonly experienced by those with EDs [9] and correlates with a broad range of symptoms, such as dietary restraint and eating concerns [10, 11]. Further, initial findings from ecological momentary assessment research suggest that feeling fat fluctuates in real time in relation to ED symptoms [12]. However, the question of what factors relate to this subjective sensation remains largely unanswered. In the current study, we built upon prior work on the correlates of feeling fat by exploring specific affective and interoceptive constructs associated with this experience.

It may be the case that feeling fat is the result of deficits in interoceptive awareness (IA), the perception and integration of signals related to body states [13]. IA is related to the ability to detect and distinguish between bodily signals (e.g., hunger and satiety), or between body sensations and emotional states [13]. Research has hypothesized that due to IA deficits, those with an ED may feel fat as a result of factors unrelated to food consumption, such as tight clothing or increased body temperature [4,5,6, 14, 15]. IA deficits might also explain why patients with anorexia nervosa feel fat despite low caloric intake and weight loss [9, 16]. Although research suggests that individuals with EDs are more likely to experience difficulties with IA and interpreting body states [9,10,11,12,13,14, 17,18,19], to date, only one study has explored the relationship between interoception and feeling fat [20]. This study’s results found that feeling fat was significantly related to poorer self-reported interoceptive awareness in healthy females [21].

In addition to potential alterations in IA, recent work suggests that feeling fat is linked to affective constructs, including depressive symptoms and guilt [4,5,6, 9,10,11,12,13,14,15]. Generalized negative affect, depressive symptoms, and guilt have all been uniquely linked to engagement in eating disordered behaviors [21,22,23,24,25]. Cognitive misattributions about one’s body (e.g., thinking “I am fat”) may be triggered by negative mood states, especially for those who overvalue weight and shape [4]. In a recent study exploring affective correlates of feeling fat in a clinical sample, depressive symptoms uniquely predicted feeling fat, over and above guilt, shame, anxiety sensitivity, and negative affect [15]. These affective constructs have implications for treatment, as targeting negative affect or symptoms of depression may decrease feeling fat over the course of treatment [15].

Alexithymia, a difficulty identifying and describing emotions [26], is a related, yet distinct construct that may also be linked to interoceptive awareness and feeling fat. Prior theoretical work posits that feeling fat may stem from an inability to label affect, and subsequently, it is recommended to ask an eating disorder patient during treatment to substitute “fat” with an emotion unrelated to weight, thus targeting their alexithymia [6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27]. Additionally, one group of researchers found that self-reported alexithymia and feeling fat were significantly associated in healthy females [20]. However, the association between alexithymia and feeling fat has not yet been replicated.

The purpose of the current study was to clarify the associations between IA, affective constructs, and subjective feelings of fatness. Gold-standard clinical treatments for EDs have suggested that the above constructs may underlie feeling fat, and thus recommend patient tracking of emotions and bodily sensation to replace “fat” with these experiences [4,5,6, 27]. However, empirical work on the correlates of feeling fat is essential to support the use of these recommendations in treatment. The first aim of this study was to address whether feeling fat relates to greater difficulties with IA using multimethod assessments of IA (self-report and behavioral physiological measurements). Second, we explored the connection between feeling fat and other affective constructs previously investigated in the literature: alexithymia, guilt, and depression. Finally, we explored which of these variables accounted for the most variance in feeling fat. We predicted that higher levels of feeling fat would be significantly associated with IA deficits. Additionally, we predicted that higher levels of feeling fat would be associated with higher levels of alexithymia, guilt, and depression. Finally, we hypothesized that these constructs would significantly contribute to variance in feeling fat and that IA, alexithymia, guilt, and depressive symptoms would uniquely account for variance in feelings of fatness.

Methods

Participants

Participants were recruited from University at Albany, State University of New York as part of a separate experimental study protocol that examined the effect of negative affect on loss of control (LOC) eating [28]. Half of the sample included individuals who self-reported engaging in LOC eating episodes at least once a week for the past 3 months, while the other half endorsed no LOC eating, as assessed by the Eating Disorder Diagnostic Scale for DSM-5 [29]. Participants either received a negative or neutral mood induction before participating in an in-lab meal, after which food consumption was measured. Results of the original study found no effect of mood induction on amount of food consumed. We analyzed the full sample because there were no significant differences in scores on the Eating Disorder Examination Questionnaire between groups. However, a t test revealed significant differences between group means on feeling fat, where those in the LOC eating group reported higher levels of feeling fat (M = 2.85, SD = 2.09) than the no LOC eating group (M = 1.89, SD = 2.00), t (124) = − 2.6, p = 0.009. The LOC eating group also reported significantly greater levels of depressive symptoms (M = 17.24, SD = 9.44) than the no LOC eating group (M = 9.89, SD = 8.05), t (125) = − 4.72, p = 0.000. Finally, the LOC eating group reported significantly greater levels of alexithymia (M = 52.72, SD = 13.60) compared to the no LOC eating group (M = 9.89, SD = 8.05), t (118) = − 0.30, p = 0.003. However, our multiple regression model accounting for group membership (LOC eating/no LOC eating) found that group membership did not contribute to variance in feeling fat or impact our results.

Measures

Eating Disorder Examination—Questionnaire 6.0 (EDE-Q; [30]): The EDE-Q assesses the severity and frequency of eating disorder pathology by asking participants to recall eating-related attitudes and behaviors over the past 28 days. It is comprised of 28 items and four subscales that measure the core features of EDs: restraint, shape, weight, and eating concerns. The 11th item of the EDE-Q was used for this study as a measure of feeling fat, which specifically asks participants: “On how many of the past 28 days have you felt fat?” Participants rate this on a seven-point scale ranging from zero (No Days) to six (Every Day). This item has been used as a measure of feeling fat in past research [10, 15].

Multidimensional Assessment of Interoceptive Awareness (MAIA; [31]): The MAIA is a 32-item scale that measures Interoceptive Awareness, which is the perception and integration of signals related to body states. ​​Scale items consist of 32 questions rated on a six-point Likert scale, with zero indicating a response of “Never” and five indicating a response of “Always.” Higher global scores indicate more IA. The MAIA has shown construct validity and reliability in eating disorder samples [17], and in our sample had good internal consistency (α = 0.88). Research using the MAIA has found a link between lower IA and ED symptom severity [17]. One group of researchers found that poorer interoceptive awareness as indicated by the MAIA was associated with feeling fat in a healthy community sample [20]. Studies using this measure have also found potential links between IA and affect. In a clinical sample of mixed ED patients, those with higher levels of IA had higher levels of emotion regulation [17].

Mental tracking method (MTM) heart-rate perception task: [32] The MTM is a behavioral measure of interoceptive awareness. During this task, participants are asked to count in time with their heartbeat for intervals of 25, 35, 45, and 55 sections with 30-s rest periods between each interval. After each interval, participants report the number of heartbeats they counted. Throughout this task, participants wore a heart-rate monitor. Scores were calculated using the following equation [32]: (1/4)*Σ[perceived heartbeats/(actual heartbeats – perceived heartbeats)]. Total scores fall within a range of zero and one, where zero indicates lower IA and 1 indicates higher IA levels.

Positive and negative affect schedule – extended version (PANAS-X; [33]). The PANAS-X is a reliable and valid self-report measure that assesses specific trait-level dimensions of positive and negative affect [33]. The six-item guilt subscale of the PANAS-X was used in this study to measure self-reported feelings of guilt based on how participants “generally feel.” This subscale has demonstrated high construct validity [33], and in our sample had excellent internal consistency (α = 0.91).

Beck Depression Inventory – II (BDI-II; [34]): The BDI-II is a 21-item self-report measure of depression in adolescents and adults. It assesses for the severity of depressive symptoms over the past 2 weeks and has demonstrated high reliability and construct validity [35]. In our sample, the BDI-II had good internal consistency (α = 0.89).

Toronto Alexithymia Scale – 20 (TAS-20; [36]): The TAS-20 is a 20-item self-report measure that assesses alexithymia on three subscales: Difficulty Describing Feelings; Difficulty Identifying Feelings, and Externally Oriented Thinking. Items are rated on a five-point Likert scale, with one indicating strong disagreement and five indicating strong agreement. The TAS-20 has good internal consistency and reliability, and adequate convergent and concurrent validity [36]. In our sample, internal consistency was good (α = 0.84).

Procedure

All procedures were approved by the IRB, and participants provided informed consent. Following completion of the EDE-Q, BDI-II, and MTM at baseline, participants were randomized to engage in either a neutral or negative mood induction task before participating in a taste test meal where they could eat as little or as much as they would like. After the taste test, participants completed the TAS-20, PANAS-X, and MAIA.

Analytic approach

Statistical analyses were conducted using R. First, we assessed that all assumptions for statistical tests were met. For our first and second aims, we evaluated the relationships between feeling fat, IA, and affective variables through zero-order correlation analyses. To address our third aim and evaluate which variables accounted for the most variance in feeling fat, we conducted a multiple regression analysis in R while accounting for group membership differences (LOC eating/no LOC eating); group membership was included in a separate block of the primary multiple regression analysis alongside two other covariates of age and BMI. Because there was likely to be multicollinearity between affective variables and the two IA measures, we also conducted a follow-up relative weights analysis looking at the unique contribution of depressive symptoms, alexithymia, guilt, and IA to feeling fat. Relative weights analysis is an approach that accounts for multicollinearity between variables, and in the current study was used to assess the relative importance of individual predictor variables in their contribution to feeling fat [37, 38]. The relative weights analysis was conducted using RWA Web Shiny App for Multiple Regression [38].

Results

Descriptive variables and zero-order correlates

Participants consisted of 128 female undergraduates between 18 and 29 years old (mean age = 19.3 years, SD = 1.95 years). All participants had BMIs above 17.5 kg/m2 (mean BMI = 23.93 kg/m2, SD = 4.13 kg/m2). The sample had a mean score of 1.51 on the EDE-Q (SD = 1.18). Women identified as White (54.7%), Asian (14.1%), Black (10.9%), Hispanic (12.5%), and Other (7.8%).

All variables were normally distributed with no significant outliers, and assumptions of homogeneity of variance and linearity were met. Means, standard deviations, and zero-order correlates for all variables are reported in Table 1. Feeling fat was significantly positively correlated with all affective variables and was significantly negatively correlated with self-report IA. Feeling fat was not correlated with age or behavioral IA as assessed by the Mental Tracking Method heartbeat task.

Table 1 Summary of zero-order correlations, means, and standard deviations
Full size table

Multiple regression

Assumptions for a linear regression were met. All model variables explained 56% of the variability in feeling fat (see Table 2). Results of a multiple regression analysis indicated that depressive symptoms, guilt, BMI, and IA deficits assessed by the MAIA accounted for significant variability in feeling fat, while behavioral IA assessed by the MTM and self-reported alexithymia did not significantly account for variability in feeling fat. Group membership (LOC eating/no LOC eating) and age also did not account for significant variability in feeling fat (see Table 2). An exploratory analysis controlling for mood condition in a separate block of the regression was conducted; mood condition did not account for significant variability in feeling fat (B = 0.131, p = 0.65). Tests of whether the data met the assumption of collinearity indicated that this was not a concern (all VIF < 1.5).

Table 2 Summary of multiple regression analysis for feeling fat, IA, and affective variables
Full size table

Relative weights analysis

The relative weights analysis revealed that depressive symptoms uniquely accounted for 19% of the variability in feeling fat, accounting for greater variability than any other variable (Table 3). Thus, depression emerged as a unique contributor to feeling fat above the other constructs measured in this study. Because BMI emerged as the most significant predictor of feeling fat in the regression analysis, a post hoc analysis was conducted to include BMI in addition to all original variables. Post hoc relative weights analysis revealed that BMI uniquely accounted for 25% of the variability in feeling fat, while depression uniquely accounted for 15% of the variability in feeling fat (Table 4).

Table 3 Relative importance weights
Full size table
Table 4 Post hoc analysis: relative importance weights
Full size table

Discussion

This study explored how self-reported feeling fat related to both behavioral and self-reported interoceptive awareness and self-reported affective constructs in a sample of undergraduate women. We found that feeling fat was significantly related to self-reported IA, guilt, depressive symptoms, and BMI, but was not significantly related to alexithymia or a behavioral measurement of IA. When testing the relative contributions of self-report variables to variability in feeling fat, depressive symptoms uniquely predicted feeling fat over other affective and interoceptive variables. Interestingly, post hoc analysis revealed that BMI played a significant role in feeling fat above depressive symptoms. Thus, while feeling fat may fluctuate throughout the day in a manner that does not correspond with actual body weight, the current findings suggest that individuals with a higher BMI may be more likely to experience feeling fat. This may be due to the effects of sociocultural stigma on those in larger bodies, making body size more salient to an individual. Overall, our results regarding contributions to feeling fat provide insight into this common symptom that has been relatively unexplored up to this point and may be impacted by targeting depressive symptoms in treatment.

First, correlational analyses found that feeling fat significantly correlated with lower self-reported IA, in addition to higher levels of alexithymia, guilt, and depressive symptoms. This supports recent findings that guilt [15] and depressive symptoms [10, 15] positively correlate with feeling fat. The current study extends this work by finding that self-reported IA is significantly associated with feeling fat. This is the second study to our knowledge to report an association between IA deficits and feeling fat [20]. Though alexithymia and feeling fat were positively correlated in the current study, alexithymia did not account for significant variability in feeling fat, suggesting that other variables such as depression relate more strongly to this symptom and thus may be more useful treatment targets [4]. Ultimately, an increased knowledge of the related correlates of feeling fat has implications for theoretical models of EDs and how symptoms relate to both body awareness and facets of emotional experiencing, which can inform potential treatment targets such as symptoms of depression.

Second, self-reported IA, guilt, and depressive symptoms all accounted for significant variance in feeling fat in our regression model. It is of note that self-reported IA accounted for 4% of the variance in feeling fat in the model, but the behavioral measurement of IA did not. The different relationship between feeling fat and two measurements of IA reinforces the importance of including multimethod measurements of IA in research. This lack of alignment between objective and subjective assessments of IA is consistent with prior literature [39, 40]. For example, one study found that despite similar objective performance on a heartbeat detection task, meditators rated their performance as superior compared to non-mediators [40]. Additionally, although those with an ED have decreased performance on heartbeat perception tasks [41, 42], this IA deficit is not necessarily related to feeling fat. It is thus possible that self-report IA captured a particular dimension of IA that uniquely relates to feeling fat, such as emotional reactivity to bodily sensations. The relationship of lower MAIA scores with feeling fat aligns with our findings that affective components, such as depression and guilt, also account for significant variance in feeling fat.

Third, when exploring the relative contributions of the variables of interest to feeling fat, depressive symptoms emerged as a unique predictor of the variance in feeling fat above IA and other affective constructs, consistent with prior research [4,5,6,7,8,9,10,11,12,13,14,15]. In the present study, we extended this work by finding that depressive symptoms predicted 19% of the variance in feeling fat, while IA deficits only predicted 4% of the variance in feeling fat. One explanation for this finding is that depressive feelings may contribute to a negative view of self, which may be expressed as feeling fat, particularly if body shape is salient to one’s identity. Additionally, interoceptive deficits are characteristic of depression, and thus feeling fat may be a secondary symptom to depression [43]. Future research is needed to understand the nature of this unique relationship. For example, ecological momentary assessment research might track whether feeling fat fluctuates in parallel with depressive symptoms. Findings that emerge about the relationship between depressive symptoms and feeling fat can also be incorporated into treatment. Endorsement of feeling fat during an intake assessment may alert clinicians to assess for depressive symptoms and focusing on these depressive symptoms in treatment may improve feeling fat [15]. Additionally, as EDs are often comorbid with depression [44], targeting feeling fat might be particularly useful for patients with these comorbidities. The current study’s findings also provide support for clinical interventions addressing feeling fat that are recommended in the Enhanced Cognitive Behavioral Therapy manual for eating disorders [4]. Namely, the fact that depressive symptoms emerged as a unique predictor of feeling fat lends support to this treatment’s claim that, for patients with an ED, the experience of feeling fat is the result of mislabeling emotions, such as depressive symptoms [4]. As self-report IA only accounted for 4% of the variance in feeling fat, the connection between depressive symptoms and feeling fat may be emphasized in treatment over physical sensations that increase body awareness [4]. For example, patients (especially those with depressive symptoms) may be asked to self-monitor times when they feel fat and emotions related to depression that they are experiencing at that time [4].

Strengths and limitations

The current study is the first to examine both interoceptive awareness and depressive symptoms as they relate to feeling fat. This study employed a multimethod assessment of interoceptive awareness, incorporating both subjective and objective measurements, to capture nuances of the construct’s correlation with feeling fat.

This study had several limitations. All analyses for the current study were cross-sectional and thus could not establish causation. Future studies using both longitudinal and experimental designs are needed to test these theoretical links and establish causation. The sample also consisted of undergraduate females rather than a clinical ED sample. However, our findings regarding the association between depressive symptoms and feeling fat replicate a prior study that used a clinical sample [15], and half of our participants reported engagement in LOC eating. An additional limitation is that our measurement of feeling fat was based on only one item on the EDE-Q. However, this item is the most commonly used measure of feeling fat in research [9]; because there is no standard measure for this construct, we had limited options for assessment. Given the importance of feeling fat in EDs, future work might explore the development of a multi-item scale to assess this construct or the use of ecological momentary assessment to track feeling fat as it fluctuates in real time. Furthermore, while a heartbeat task served as an objective measure of IA in this study, there are measures that use awareness of other bodily systems such as gastrointestinal systems (e.g., Water Load Task, ingestion of a vibrating capsule) that might be more relevant to EDs and feeling fat [45,46,47,48]. While the current study found no relationship between feeling fat and heartbeat perception, future research could investigate the relationship between feeling fat and gastrointestinal objective measures of IA that are more relevant to EDs. An additional limitation is the fact that the measured constructs were not collected at one time point; namely, guilt, alexithymia, and self-report IA were collected after a mood induction and food consumption. However, other than objective IA, measures were not collected as state-level variables. We assessed feeling fat over the past month as opposed to a present state-based feeling (“On how many of the past 28 days have you felt fat?”). Additionally, the PANAS-X, which was one of the measures administered following the mood induction, measured trait levels of guilt (“how do you generally feel”). The MAIA, our subjective measure of interoceptive awareness, also asks participants to respond based on “how often each statement applies to you generally in daily life.” Alexithymia, the final measure administered after the experimental condition, was also assessed as a trait-level variable. Importantly, our depressive symptom measure, which emerged as the most unique predictor of feeling fat, was assessed at baseline along with feeling fat. Finally, BMI emerged as a confounding variable that significantly contributed to feeling fat. It is possible that the sociocultural stigma that individuals in larger bodies face plays a larger role in feeling fat than depressive symptoms, and thus may provide a risk for this common eating disorder symptom. However, it must be noted that underweight individuals also commonly experience feeling fat. Further research is warranted to explore the role that BMI plays in feeling fat.

What is already known on this subject?

Though feeling fat is a commonly reported symptom in eating disorder patients that is included in current theoretical and treatment models, its correlates remain poorly understood. Theoretical work suggests that feeling fat may relate to deficits in IA, and one prior study found that IA deficits assessed by self-report were correlated with feeling fat. Another prior study found depressive symptoms to uniquely predict feeling fat.

What this study adds

This study extends past work to include a consideration of IA in feeling fat. The current study found that deficits in self-reported IA accounted for significant variability in feeling fat, while objective IA did not relate to feeling fat. Depressive symptoms uniquely accounted for feeling fat, over and above IA and other affective constructs. Endorsement of feeling fat during an intake assessment may alert clinicians to screen for depressive symptoms, and focusing on these depressive symptoms in treatment may improve feeling fat and improve ED treatment outcomes.

Conclusions

This study explored the relation between feeling fat and IA in addition to a range of affective constructs. Depressive symptoms, guilt, and self-reported IA accounted for significant variability in feeling fat, with depression emerging as the variable accounting for the greatest proportion of variance in feeling fat. Future work might use an experimental design to test the theory that for ED patients with interoceptive deficits, depressed affective states contribute to feeling fat. Additionally, ecological momentary assessment research might examine how feeling fat fluctuates throughout the day along with depressed mood. Ultimately, an improved understanding of how IA and affective constructs contribute to feeling fat may not only strengthen current models of what we know about EDs, but also contribute to treatment, given data suggesting that decreases in feeling fat predict weight maintenance after treatment [7]. An increased understanding of symptomatology such as feeling fat is thus crucial to improve our current understanding and treatment of ED symptoms.