Metacognitive confidence independently predicts body image delusionality in anorexia and bulimia nervosa

Abstract

Objectives

Body image beliefs in eating disorders (EDs) are often characterized by high conviction and resistance to corrective evidence. Although perceptual distortion has been widely implicated, body image delusionality may also depend on higher-order belief evaluation processes. We examined whether confidence in body-size judgments is associated with body image delusionality beyond perceptual misestimation.

Methods

Ninety-seven adult female patients with EDs, 51 with anorexia nervosa (AN) and 46 with bulimia nervosa (BN), were assessed using the Brown Assessment of Beliefs Scale (BABS). Body-size perception was evaluated with the Contour Drawing Rating Scale, including current and ideal size estimates, body dissatisfaction (the absolute current–ideal discrepancy), misestimation (the signed discrepancy from clinician ratings), and confidence in body-size judgments. Associations were examined using non-parametric analyses with false discovery rate correction for the body-size perception correlations, and hierarchical regression models.

Results

Delusional body image beliefs were more frequent in AN than BN (29.4% vs 2.2%). At the bivariate level, delusionality was associated with current size ratings, body dissatisfaction, misestimation, and confidence. In hierarchical regression analyses, perceptual indices explained little variance in delusionality. After entering confidence in the final step, confidence emerged as the sole independent predictor and substantially increased explained variance (final model R² = 0.34).

Conclusions

Body image delusionality in EDs appears more strongly related to metacognitive confidence than to perceptual distortion. These findings support a dimensional account of delusionality and suggest that confidence in body-size judgments may contribute to persistent distorted body image beliefs, although the cross-sectional design precludes causal inference.

1. Introduction

Lack of insight is a central clinical feature of eating disorders (EDs), particularly anorexia nervosa (AN), where individuals often deny the severity of low weight, restrictive behaviors, and associated medical risks (Konstantakopoulos et al., 2012; Mountjoy et al., 2014). Impaired insight complicates treatment engagement and adherence and has been associated with poorer short- and long-term outcomes in AN (Halmi et al., 1979; Saccomani et al., 1998; De Young et al., 2022). Although insight deficits were historically described in psychotic disorders (David, 2020), reduced illness awareness and diminished recognition of the pathological nature of body-related beliefs are now recognized as central to the phenomenology and clinical management of EDs (Hartmann et al., 2013; Konstantakopoulos et al., 2020; Gawron et al., 2025).

Impaired insight in EDs is closely related to the degree of conviction with which body weight and shape beliefs are held. When such beliefs are rigid or delusional in intensity, they substantially contribute to poor illness awareness and resistance to treatment. Investigating the cognitive and evaluative processes through which body image beliefs are formed and maintained is therefore central to understanding impaired insight in EDs.

Body image beliefs in EDs span a continuum from intact insight through overvalued ideas to delusional convictions (Konstantakopoulos et al., 2012). This variability has been captured using dimensional approaches to belief evaluation, most notably the Brown Assessment of Beliefs Scale (BABS) (Eisen et al., 1998), which has been widely applied in ED research. Studies show that a substantial proportion of individuals with AN endorse overvalued or delusional body image beliefs, with higher rates in the restricting subtype than in the binge–purge subtype (Steinglass et al., 2007; Hartmann et al., 2013; Mountjoy et al., 2014; Konstantakopoulos et al., 2012). In contrast, individuals with bulimia nervosa (BN) more commonly exhibit overvalued ideas, whereas fully delusional beliefs are uncommon (Konstantakopoulos et al., 2012; McKenna et al., 2014). Cross-diagnostic comparisons using the BABS further indicate that individuals with AN exhibit a level of delusional intensity intermediate between obsessive–compulsive disorder and first-episode psychosis (Camprodon-Boadas et al., 2023). Importantly, delusionality in this context does not imply psychosis but refers to the degree to which specific body-related beliefs are strongly endorsed despite contrary evidence.

Empirical studies indicate that body image delusionality in EDs is associated with central attitudinal features such as drive for thinness and shape and weight concerns, and has also been linked to restrictive eating behaviors (Steinglass et al., 2007; Konstantakopoulos et al., 2012; Şenay & Yücel, 2023; De Young et al., 2022). However, delusional intensity is not consistently related to body mass index, illness duration, or overall severity (Hartmann et al., 2013; Konstantakopoulos et al., 2012; De Young et al., 2022), and findings regarding its association with perceptual body dissatisfaction are mixed (Barton et al., 2022; Şenay & Yücel, 2023). Other mechanisms, such as cognitive inflexibility or altered learning processes, may also contribute to the formation and maintenance of these beliefs. Moreover, delusionality in EDs does not appear to be explained by psychosis-typical reasoning biases such as jumping to conclusions (McKenna et al., 2014).

Evidence from psychosis research implicates metacognitive and belief-evaluation processes in the formation and maintenance of delusions (Moritz & Woodward, 2006; Morrison et al., 2011; Ward & Garety, 2019). In particular, delusions have been linked to abnormal self-evaluation of judgments, including exaggerated confidence in errors and reduced openness to disconfirmatory evidence (Moritz et al., 2005; Garety & Freeman, 2013). Moreover, associations between metacognitive beliefs and delusionality have been reported in obsessive–compulsive disorder (Önen et al., 2013), suggesting that higher-order belief evaluation processes may contribute to delusional intensity across mental disorders (Phillips et al., 2012). Metacognitive interventions further suggest that targeting belief-evaluation processes can reduce delusional symptoms in psychosis (Penney et al., 2022).

In ED research, emphasis has traditionally been placed on perceptual distortion and affective components of body image disturbance. However, empirical findings regarding objective body-size misestimation are heterogeneous and vary across paradigms (Mölbert et al., 2017; Prnjak et al., 2022; De Simone et al., 2025). Even when perceptual distortion is observed, it does not consistently differentiate individuals with rigid or delusional body-image beliefs from those with less fixed beliefs (Mountjoy et al., 2014; Barton et al., 2022). Comparatively little attention has been paid to how individuals evaluate and endorse their perceptual judgments. Recent experimental work indicates slower updating of body size beliefs in acute AN following corrective feedback, alongside reduced updating of interoceptive self-efficacy beliefs despite intact performance (Saramandi et al., 2022; Saramandi et al., 2024). Whether delusionality in EDs is more closely related to the magnitude of perceptual distortion or to the confidence with which distorted judgments are held remains unclear.

Against this background, the present study investigated the delusionality of body image beliefs in EDs using a dimensional approach, focusing on the relative contributions of body-size perception and metacognitive confidence. We compared individuals with AN and BN in the frequency and severity of delusional and overvalued body image beliefs and examined associations with clinical variables and body-size perception measures, including perceptual estimates, misestimation, and confidence. Building on models of belief evaluation, which propose that delusionality depends on how individuals appraise the reliability and certainty of their own judgment, we investigated whether confidence – conceptualized as a metacognitive confidence judgment – is associated with delusionality beyond perceptual distortion and misestimation. We hypothesized that higher confidence in subjective body-size judgments would be more strongly associated with body-image delusionality than with perceptual distortion itself. Importantly, our focus was on overall confidence in body-size judgments rather than on metacognitive efficiency or confidence–accuracy correspondence.

2. Methods

2.1. Participants and procedures

Ninety-seven participants, 51 patients with AN and 46 patients with BN, all female, were consecutively recruited following admission to the Eating Disorders Unit of the Eginition University Hospital. All patients were native Greek speakers, aged 18-45, and were diagnosed with either AN or BN according to DSM-5 criteria by a psychiatrist specializing in EDs. In the AN group, 29 had the restricting type of the illness (AN-R), and 22 had the binge-purge type (AN-BP). Exclusion criteria for participation in the study were: intellectual disability, concurrent comorbidity with substance abuse-related disorders, obsessive-compulsive disorder, or body dysmorphic disorder, current major depressive episode, and a history of psychosis.

All the interviews and questionnaires were administered to each participant on the same day. Basic demographic data were obtained, and participants’ weight and height were measured on the day of testing. The Eginition University Hospital Ethics Committee approved the research protocol. All procedures were conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent for their participation in the study.

2.2. Measures

Demographic and clinical data included age, education, illness duration, and lowest adult lifetime BMI as calculated from adult height and lowest adult lifetime weight. BMI was calculated from the height and weight measured at the time of recruitment.

2.2.1. Assessment of body image delusionality

The Brown Assessment of Beliefs Scale (BABS) (Eisen et al., 1998) is a semi-structured interview that estimates the degree of delusional thinking dimensionally during the past week related to a dominant belief and also provides cutoff points to categorize beliefs as delusional (i.e., lacking insight) and non-delusional with poor insight (overvalued) or with intact insight. The first six items measure the following components of delusionality: (1) conviction (how certain the patient is about the accuracy of the belief), (2) perception of others’ views (the patient’s ability to accurately assess the beliefs of others), (3) explanation of differing views (the patient’s explanation for the difference between his/her and others’ views of the belief), (4) fixity of beliefs (how unshakable the patient’s conviction about the belief is), (5) attempt to disprove beliefs (how actively and frequently the patient tries to disprove or reject his/her belief), (6) insight into the symptomatic nature of the belief (whether the patient recognizes that the belief has a psychiatric/psychological cause). Each item is rated from 0 to 4, with higher scores indicating poorer insight. All item scores are summed to reach a total score, ranging from 0 to 24. An additional item assessing referential thinking is not included in the total score.

We used BABS to assess patients’ beliefs about their body weight and shape that might interfere with eating, as has been used in previous studies for the assessment of body image delusionality in EDs (Steinglass et al., 2007; Konstantakopoulos et al., 2012; Hartmann et al., 2013; McKenna et al., 2014; Mountjoy et al., 2014). Higher total scores on the BABS indicate poorer insight. A total score ≥ 18 plus a score of 4 on item 1 indicates delusional beliefs, and a total score ≥ 13 plus a score ≥ 3 on item 1 is the threshold for overvalued ideas (poor insight) (Konstantakopoulos et al., 2012).

2.2.2. Body-size perception measures

The Contour Drawing Rating Scale (CDRS) (Thompson & Gray, 1995) was used as a body image assessment tool. The CDRS uses schematic figures of nine female drawings ranging from underweight to overweight. The participants were asked to rate their ideal figure and their current size. Body dissatisfaction was computed as the absolute difference between the current and ideal size ratings (|current − ideal|), quantifying the magnitude of the current–ideal discrepancy irrespective of its direction.

To evaluate the accuracy of participants’ body size estimation, we asked each patient’s treating clinician to rate the patient’s body size on the CDRS. Misestimation was defined as the signed difference between the patient’s and the clinician’s rating on the CDRS (patient − clinician), with positive values indicating overestimation and negative values underestimation of one’s own body size, and was used as a measure of body image distortion. We use the terms body dissatisfaction and misestimation throughout to align variable labels with the constructs they are intended to capture. To examine confidence about performance, participants were also asked to indicate on a visual scale how confident they were in the correctness of their choice for CDRS. Participants indicated their confidence from 0 (not at all confident) to 100% (extremely confident).

2.2.3. Symptom assessment

The Eating Attitude Test, 26-item version (EAT-26) (Garner et al., 1982) was used to measure eating-related pathology. The Bulimic Investigatory Test Edinburgh (BITE) (Henderson and Freeman, 1987) measures specifically bulimic symptomatology consisting of two – symptoms and severity – subscales. The Eating Disorders Inventory (EDI) (Garner et al., 1983) is a 64-item questionnaire that assesses a broad spectrum of psychological and behavioral aspects of EDs. The Greek versions of EAT-26, EDI (Simos, 1996), and BITE (Konstantakopoulos, 2013) have proven validity and reliability.

The Body Shape Questionnaire (BSQ) (Cooper et al., 1987) is a self-report questionnaire comprising 34 questions assessing concerns about body shape, in particular the phenomenal experience of “feeling fat”. The items refer to the subjects’ state over the past four weeks, and higher scores indicate greater body dissatisfaction. The Greek version of BSQ has been shown to have good convergent validity and test-retest reliability (Tsiakou, 2022).

2.3. Statistical analysis

The normality of distribution was examined by means of the Shapiro–Wilk tests. The majority of the examined variables did not follow a normal distribution, and therefore, non-parametric tests were used. Kruskal–Wallis H and Mann–Whitney U tests were conducted to evaluate differences between groups as appropriate. Effect sizes are reported as r = |Z|/√N for Mann–Whitney comparisons and epsilon-squared (ε²) for Kruskal–Wallis tests. Spearman’s rho correlation coefficients were computed between the BABS total score and (a) the demographic and clinical variables and (b) the five CDRS-derived measures (current size, ideal size, body dissatisfaction, misestimation, and confidence), separately in the AN group, the BN group, and the whole sample. The five CDRS correlations in each sample were treated as the a priori, hypothesis-driven family of tests and were corrected for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure at q = 0.05; correlations with demographic and clinical variables were exploratory and are reported uncorrected. Patients with delusional beliefs, overvalued ideas, and intact insight according to the BABS cutoffs were compared on the CDRS measures using Kruskal–Wallis tests followed by Dunn’s pairwise comparisons with Bonferroni adjustment. To examine the contribution of body-size perception measures to delusionality, we used hierarchical multiple regression with the BABS total score as the dependent variable. Predictors were entered in a theoretically motivated order reflecting increasing levels of cognitive processing: the subjective perceptual estimate (current size) in the first step, body dissatisfaction in the second, misestimation in the third, and confidence, as the metacognitive belief-evaluation measure, in the final step. To verify that conclusions regarding incremental explained variance did not depend on entry order, we additionally examined the variance explained by confidence entered alone and the increment associated with the three perceptual indices added beyond confidence. Furthermore, to address the potential conceptual overlap between the confidence measure and the conviction component of the BABS, the hierarchical regression was repeated with the BABS total score excluding item 1 (conviction) as the dependent variable. Collinearity and residual statistics were within the acceptable range. The alpha level was set at 0.05. Statistical analyses were performed using IBM SPSS Statistics version 29.

With N = 97 and α = 0.05 (two-tailed), the study had 80% power to detect a small-to-moderate overall effect (f² ≈ 0.13) in the final hierarchical regression model including four predictors (body-size perception measures), and small incremental effects (f² ≈ 0.085) for the addition of a single predictor.

3. Results

3.1. Sample characteristics

Table 1 shows the demographic and clinical characteristics of the sample, including total BABS score and body-size perception measures. No statistically significant differences were found between groups with respect to age and years of education, as well as between AN and BN in disease duration. As expected, the current mean BMI and the lowest lifetime BMI were significantly lower in AN than in BN patients, whereas BN patients scored significantly higher on BITE symptoms and severity subscales than AN patients. There was no significant difference between groups in EAT-26, EDI, and BSQ.

Table 1. Demographic and clinical characteristics and body-size perception measures of anorexia nervosa (AN) and bulimia nervosa (BN).

	AN (n=51)	BN (n=46)	U	P	r
	Median (IQR)
Age (years)	23.00 (8.00)	23.50 (10.00)	985.5	0.176	0.14
Education (years)	15.00 (2.00)	16.00 (2.00)	1143.5	0.824	0.02
BMI (kg/m2)	16.43 (1.77)	19.91 (2.63)	61.0	< 0.001	0.82
Lowest lifetime BMI (kg/m2)	15.41 (3.06)	18.00 (2.07)	161.5	< 0.001	0.74
Duration of disease (years)	5.00 (6.00)	6.00 (7.25)	1060.5	0.417	0.08
EAT-26	41.00 (20.00)	36.00 (9.00)	1324.5	0.093	0.17
BITE
Symptoms	16.00 (12.00)	23.00 (5.00)	505.5	< 0.001	0.47
Severity	6.00 (8.50)	10.00 (4.00)	621.0	< 0.001	0.38
EDI total	68.00 (46.50)	80.00 (22.50)	933.5	0.202	0.13
BSQ	88.00 (32.00)	97.50 (27.50)	1047.0	0.364	0.09
BABS total	14.00 (9.00)	12.00 (5.00)	1377.0	0.140	0.15
CDRS
Current size	3.00 (1.00)	3.00 (2.00)	737.5	0.001	0.33
Ideal size	2.00 (1.00)	2.00 (1.00)	1041.0	0.309	0.10
Body dissatisfaction	1.00 (1.00)	1.00 (1.00)	844.5	0.012	0.25
Misestimation	1.00 (2.00)	1.00 (2.00)	1170.0	0.985	0.00
Confidence	70.00 (30.00)	60.00 (20.00)	1400.5	0.096	0.17

IQR: interquartile range, BMI: body mass index, EAT-26: Eating Attitude Test, BITE: Bulimic Investigatory Test Edinburgh, EDI: Eating Disorders Inventory, BSQ: Body Shape Questionnaire, BABS: Brown Assessment of Beliefs Scale, CDRS: Contour Drawing Rating Scale. Body dissatisfaction: absolute difference between current and ideal size ratings, |current − ideal|. Misestimation: signed difference between the participant’s and the clinician’s rating of the participant’s current size (positive values indicate overestimation). p-values were estimated from the Mann–Whitney U test; r = |Z|/√N is the corresponding effect size.

Thirty-nine patients (40.2%) had received mental health care in the past, and twenty-three (23.7%) were taking medication (all of them antidepressants and seven of them benzodiazepines). Thirty-two patients (33%) had a history of mental disorder other than EDs, and 10 (10.3%) had a history of substance abuse.

3.2. Delusionality of body image beliefs in AN and BN

As shown in Table 1, there was no significant difference between AN and BN patients in the BABS total score.

According to BABS cutoffs, 29.4% of AN patients had delusional body image beliefs (lack of insight), and 25.5% had overvalued ideas (poor insight). Only one BN patient was classified as delusional (2.2%), and 39.1% of BN patients had poor insight. These differences in frequencies between AN and BN were statistically significant (χ2 = 13.15, p = 0.001). This apparent discrepancy reflects the dimensional nature of the BABS, whereby similar total scores can mask differences at the extreme end of conviction captured by categorical cutoffs. Delusional beliefs were found in 44.8% of AN-R patients and in 9.1% of AN-BP patients, while overvalued ideas were found in 22.7% of each AN subgroup. The differences between AN-R and AN-BP were also statistically significant (χ2 = 10.12, p = 0.006).

3.3. Body-size perception measures in AN and BN

AN patients scored significantly higher on the CDRS rating of their current body size than BN patients (r = 0.33) and also showed significantly greater body dissatisfaction (r = 0.25). No significant difference between the two groups was found regarding the ideal size or misestimation. Moreover, although AN patients tended to report higher confidence than BN patients, this difference did not reach statistical significance.

3.4. Associations of body image delusionality

As shown in Table 2, the BABS total score was negatively correlated with illness duration in the BN group (rho = −0.34), indicating that longer illness duration was associated with lower delusionality, and negatively correlated with the BITE symptoms score in the AN group and the whole sample. Moreover, there was a weak but significant positive correlation between BABS total and BSQ scores in the whole sample. No other significant correlation was found between any demographic or clinical variable and the BABS total score in either patient group.

Table 2. Bivariate correlations of body image delusionality in anorexia nervosa (AN), bulimia nervosa (BN), and the whole sample - Spearman’s correlation coefficients (rho).

	BABS total
	AN	BN	Whole sample
Age	−.15	−.19	−.16
Education	.13	−.22	−.03
BMI	.02	−.08	−.16
Lowest lifetime BMI	.15	−.01	−.05
Duration of disease	.04	−.34*	−.12
EAT-26	−.18	.12	−.05
BITE
Symptoms	−.35*	.05	−.25*
Severity	−.16	.01	−.16
EDI total	−.02	.24	.07
BSQ	.23	.22	.22*
CDRS
Current size	.54***†	.08	.28**†
Ideal size	.10	−.16	−.03
Body dissatisfaction	.22	.36*	,23*†
Misestimation	.49***†	.27	.38***†
Confidence	.73***†	.33*	.59***†

*p < 0.05, **p < 0.01, ***p < 0.001 (uncorrected). † Association remains significant after Benjamini–Hochberg false discovery rate correction applied within the a priori family of the five CDRS correlations in each sample; correlations with demographic and clinical variables were exploratory and are reported uncorrected. Body dissatisfaction: |current − ideal| size rating; Misestimation: participant − clinician rating of current size (positive values indicate overestimation).

Regarding the body-size perception measures, the BABS total score was positively correlated with current size ratings, body dissatisfaction, misestimation, and confidence in the whole sample, and all four associations remained significant after FDR correction. Of note, the correlation with confidence was the strongest. A similar pattern of correlations was observed in the two groups, with the following differences: (a) non-significant correlations with current size ratings and misestimation in the BN group, (b) a non-significant correlation with body dissatisfaction in the AN group, and (c) a strong correlation with confidence in the AN group. After FDR correction, the associations with current size, misestimation, and confidence remained significant in the AN group, whereas in the BN group, no CDRS correlation survived correction (body dissatisfaction and confidence: pFDR = 0.058). Associations between body-size perception measures and BABS item scores in AN, BN, and the whole sample are presented in the Supplementary material. Across samples, confidence correlated most strongly with conviction, explanation of differing views, fixity of beliefs, and attempts to disprove beliefs, and only weakly with perception of others’ views and referential thinking (Supplementary Tables S1–S3).

Additionally, we examined the relationships between body image delusionality and the body-size perception measures by comparing scores from CDRS between ED patients (whole AN and BN sample) with delusional beliefs, overvalued ideas, and intact insight according to BABS cutoffs (see Table 3). Patients with delusional or overvalued body image beliefs had higher levels of Misestimation and Confidence than patients with intact insight. Moreover, patients with delusional beliefs had significantly higher ratings of Current size than patients with intact insight. No other significant difference was found.

Table 3. Difference in body-size perception measures between patients with delusional (D), overvalued (O), and with intact insight (I) body image beliefs.

CDRS	D (n=16)	O (n=31)	I (n=50)	H	P	ε 2	Pairwise comparisons
	Median (IQR)
Current size	4.00 (1.00)	3.00 (2.00)	2.50 (1.25)	7.25	0.027	0.06	D > I
Ideal size	2.00 (2.75)	2.00 (0.00)	2.00 (2.00)	0.22	0.894	−	ns
Body dissatisfaction	1.00 (1.00)	1.00 (1.00)	1.00 (1.00)	6.30	0.043	0.05	ns
Misestimation	2.00 (1.75)	1.00 (1.00)	0.00 (1.00)	17.19	< 0.001	0.16	D, O > I
Confidence	80.00 (17.50)	70.00 (20.00)	60.00 (12.50)	40.04	< 0.001	0.41	D, O > I

IQR: interquartile range, CDRS: Contour Drawing Rating Scale, ns: non statistically significant. p-values were estimated from the Kruskal–Wallis H test; ε2 is the corresponding effect size. Pairwise comparisons were conducted with Dunn’s test, and significance values were adjusted by the Bonferroni correction for multiple tests. Body dissatisfaction: |current − ideal| size rating; Misestimation: participant − clinician rating of current size (positive values indicate overestimation).

3.5. The contribution of body-size perception measures to body image delusionality

We used hierarchical multiple regression analysis to examine the contribution of the body-size perception measures of the study to delusionality of body image beliefs in our sample; the model specification and the rationale for the order of predictor entry are described in Section 2.3.

As shown in Table 4, Current size ratings, entered alone, showed a significant effect, but the model explained only a small proportion (4%) of the BABS total score variance. When Misestimation was entered as a predictor in the third model, the effects of Current size ratings and Body dissatisfaction were not significant, while the effect of Misestimation was significant. This model explained 10% of the BABS total score variance. In the final model, only the effect of Confidence remained significant (ΔR² = 0.24, F change (1,92) = 35.46, p < 0.001), and the model explained 34% of the BABS total score variance. The incremental contribution of Confidence did not depend on entry order: entered alone, Confidence explained 34% of the variance (R² = 0.34), whereas the three perceptual indices added only ΔR² = 0.03 beyond Confidence (F change (3,92) = 1.63, p = 0.187). In the sensitivity analysis using the BABS total score excluding item 1 (conviction) as the dependent variable, Confidence remained the sole independent predictor in the final model (β = 0.47, p < 0.001, ΔR² = 0.19; Supplementary Table S4).

Table 4. The contribution of body-size perception measures to body image delusionality - hierarchical regression analysis.

Model summary*					Contribution of variables at each step
Adjusted R2	F	Model p	ΔR2	p (F change)	Predictors	ϐ	Predictor p
Step 1
0.04	4.90	0.029	0.05	0.029	Current size	0.22	0.029
Step 2
0.05	3.51	0.034	0.02	0.153	Current size	0.12	0.343
					Body dissatisfaction	0.18	0.153
Step 3
0.10	4.61	0.005	0.06	0.013	Current size	−0.09	0.518
					Body dissatisfaction	0.13	0.278
					Misestimation	0.34	0.013
Step 4
0.34	13.60	< 0.001	0.24	< 0.001	Current size	−0.13	0.294
					Body dissatisfaction	0.10	0.334
					Misestimation	0.20	0.088
					Confidence	0.53	< 0.001

* Dependent variable: BABS total score; β = standardized regression coefficient; ΔR² = increase in R² at each step, with the corresponding F-change test. Entering Confidence in the first step yielded R² = 0.34; the three perceptual indices added ΔR² = 0.03 (F change (3,92) = 1.63, p = 0.187) beyond Confidence, indicating that the incremental contribution of Confidence is independent of entry order.

4. Discussion

The present study examined the relative contributions of body-size perception and metacognitive confidence to body image delusionality in EDs. Consistent with previous studies using BABS, delusional and overvalued beliefs were more frequent in AN than in BN. While perceptual measures were associated with delusionality at the bivariate level, confidence in body-size judgments showed the strongest association and remained the only independent predictor in hierarchical analyses. These findings suggest that subjective certainty in body-size judgments may play a more central role in body image delusionality than perceptual distortion alone.

The distribution of delusional and overvalued beliefs in our sample is consistent with prior studies showing that delusional intensity is more frequent in AN—particularly in the restricting subtype—and less common in BN (Konstantakopoulos et al., 2012; McKenna et al., 2014; Barton et al., 2022; Kambanis et al., 2022). Distortions in body-size perception and elevated body dissatisfaction have long been regarded as central features of ED psychopathology and have been linked to poorer insight and greater delusionality of body image beliefs (Konstantakopoulos et al., 2012; Kambanis et al., 2022; Şenay & Yücel, 2023). In the present study, these perceptual measures were associated with delusionality at the bivariate level but not after confidence was entered into the model, suggesting that perceptual disturbance alone does not account for body image delusionality despite its established role in ED psychopathology.

The findings are consistent with belief-evaluation models developed in psychosis research, where exaggerated confidence in incorrect judgments contributes to delusions (Moritz et al., 2005; Moritz et al., 2014). In the present study, confidence was task-based, whereas delusionality was assessed independently with the BABS. The persistence of their association after controlling for perceptual indices suggests that evaluative processes may play a role in shaping body image delusionality beyond perceptual distortion. However, the study did not assess metacognitive efficiency or belief updating directly, and therefore, conclusions remain limited to associations. Notably, reasoning biases characteristic of psychosis, such as jumping to conclusions, do not appear to account for body image delusionality (McKenna et al., 2014).

The association between confidence and delusionality is unlikely to reflect mere conceptual overlap with the conviction component of the BABS. The two measures address different objects: confidence concerns a specific perceptual judgment, whereas BABS conviction concerns the certainty with which the dominant body-related belief is held despite contrary evidence. Accordingly, excluding conviction from the delusionality score left the findings essentially unchanged, and confidence was related broadly to the evaluative dimensions of delusionality rather than selectively to conviction.

Converging findings within ED research further support a role for metacognitive processes in insight impairment. Arbel et al. (2013) showed that metacognitive monitoring and control predicted poor insight beyond basic cognitive performance, Konstantakopoulos et al. (2020) found that body image delusionality was more closely associated with higher-order cognitive and social-cognitive processes than with general illness severity, and Saramandi et al. (2022, 2024) demonstrated slower updating of body-related beliefs. The present findings extend this literature by identifying metacognitive confidence as a domain-specific belief-evaluation process associated with body image delusionality, without implying a global metacognitive deficit or a psychotic process.

In clinical practice, individuals with EDs often resist medical or perceptual feedback about their body size. The present findings suggest that such resistance may reflect the high certainty with which body-related judgments are endorsed, contributing to limited responsiveness to corrective information. Slower updating of body-size beliefs (Saramandi et al., 2022), together with alterations in reinforcement learning and cognitive flexibility (Lao-Kaim et al., 2015; Dougherty et al., 2024), is consistent with this interpretation. Accordingly, attending to the confidence with which patients hold body-related beliefs—rather than focusing solely on their accuracy—may clarify persistent disagreement and inform interventions targeting confidence calibration, uncertainty tolerance, and openness to alternatives (De Young et al., 2022).

The confidence–delusionality association appeared stronger in AN than in BN, where none of the CDRS associations survived correction for multiple testing. While supporting a dimensional account of delusionality within EDs, these findings suggest that the underlying mechanisms may differ in magnitude across diagnoses.

Although Confidence emerged as the sole independent predictor, the final model accounted for approximately one-third of the variance in delusionality, leaving a substantial proportion unexplained; additional processes, such as cognitive inflexibility, altered learning, and social-cognitive mechanisms, are likely to contribute. Future research should examine whether elevated confidence prospectively predicts persistence of body image delusionality, treatment response, or relapse, and experimentally test whether modifying confidence alters responsiveness to disconfirmatory body-related feedback. Integrating confidence with other belief-evaluation measures may further distinguish shared from domain-specific mechanisms of delusionality, in line with multidimensional metacognitive impairments reported in schizophrenia-spectrum disorders (Rouy et al., 2021).

Several limitations should be acknowledged. First, the cross-sectional design precludes causal inference: associations, not mechanisms, are demonstrated. Second, the single task-specific confidence rating does not capture metacognitive efficiency, calibration, monitoring, or belief updating, and cognitive flexibility was not assessed, warranting caution regarding broader metacognitive claims. Third, body dissatisfaction and misestimation are coarse ordinal difference scores with restricted range and thus lower reliability, whereas confidence is a fine-grained scale; its apparent superiority may therefore partly reflect measurement precision. Fourth, clinician ratings, the criterion for misestimation, are not an objective anthropometric measure and may introduce bias. Fifth, one quarter of participants were receiving medication (exclusively antidepressants, seven also benzodiazepines, no antipsychotics); an effect on confidence, insight, or delusionality cannot be excluded. Sixth, despite adequate post hoc power, the sample is small for regression analyses with several correlated constructs. Finally, the sample comprised only adult female AN or BN patients from a single specialized center, without healthy controls, limiting generalizability to males, adolescents, community samples, and other diagnoses (e.g., binge eating disorder or other specified feeding or eating disorders).

In conclusion, body image delusionality in EDs appears more strongly associated with confidence in body-size judgments than with perceptual distortion per se. These findings support a dimensional account of delusionality and identify belief evaluation processes as a relevant component of impaired insight in eating disorders.

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.