Metacognitive ability is associated with reduced emotion suppression

Abstract

Emotion regulation and metacognition are two key self-regulatory capacities that contribute significantly to psychological functioning. Emotion regulation involves the strategies individuals use to influence their emotional experiences, whereas metacognition refers to the capacity to reflect on and monitor one’s own mental processes. Both constructs have been the focus of extensive research and are associated with important cognitive and behavioral outcomes. Although theoretical models propose that self-monitoring is a fundamental component of emotion regulation, the empirical link between emotion regulation strategies and metacognitive ability remains poorly understood. This preregistered study (N = 200) investigated the association between self-reported emotion regulation strategies, specifically reappraisal and suppression, and metacognitive ability, which we assessed using a visual discrimination task paired with confidence ratings. We also included self-reported emotional intelligence and rumination as control variables. Our findings indicate that better metacognitive ability is associated with lower use of suppression, but shows no relationship with reappraisal. These results suggest that metacognitive processes may play a meaningful role in shaping how individuals regulate their emotions.

Introduction

Metacognitive ability is associated with reduced emotion suppression

Emotion regulation involves modulating one’s emotional responses through various cognitive or behavioral strategies¹. Decades of research has confirmed that different strategies produce markedly different outcomes; notably, the tendency to engage in cognitive reappraisal is generally associated with healthier affective profiles, better interpersonal relationships, and greater well-being, whereas frequent suppression predicts experiencing fewer positive emotions, worse social support, and lower life satisfaction².

Effective emotion regulation relies on the monitoring and modification of internal states. Self-monitoring plays an important role in Gross’ Extended Process Model of Emotion Regulation as individuals need to attend to their emotional experience and regulation efforts in real time, assess whether the chosen strategy is effective, and adjust accordingly³. Accurate monitoring is particularly important for strategy selection and strategy efficacy beliefs because it facilitates drawing on the past performance of regulatory strategies to inform better strategy choices^3,4. Effective self-monitoring may also support the development of regulatory expertise over time, as individuals accumulate knowledge about which strategies work best in different contexts^5–7.

Different emotion regulation strategies vary considerably in their cognitive and monitoring demands, which may explain why metacognitive ability relates to their use. Cognitive reappraisal involves reinterpreting the meaning of an emotion-eliciting situation to alter its emotional impact¹. This process requires substantial cognitive resources, including working memory capacity to maintain alternative interpretations, cognitive flexibility to generate multiple perspectives, and executive control to override initial appraisals^8,9. Neuroimaging evidence indicates that reappraisal recruits prefrontal regions associated with cognitive control and evaluative processing¹⁰. Successfully implementing reappraisal also requires accurate monitoring of one’s emotional state to determine when reappraisal is needed and whether the attempted reappraisal is effectively modulating the emotional response.

Expressive suppression, by contrast, involves inhibiting the outward expression of emotion while the emotional experience continues internally². Although suppression may appear simpler than reappraisal, it imposes its own cognitive costs, such as maintaining suppression requires sustained monitoring of one’s expressive behavior and continuous executive control to inhibit emotional displays¹¹. Research has shown that suppression depletes cognitive resources and impairs memory for information encountered while suppressing¹². Importantly, suppression often proves ineffective at actually reducing the subjective experience of emotion and may even intensify physiological arousal¹. Individuals who more accurately monitor the ineffectiveness of their regulatory efforts may be more likely to abandon ineffective strategies like suppression in favor of alternatives.

The theoretical link between self-reflection and emotion regulation strategy selection has been explored in several frameworks. Aldao and Nolen-Hoeksema proposed that individuals develop regulatory repertoires partly through learning which strategies prove effective across different contexts¹³. This learning process inherently requires metacognitive monitoring—tracking which strategies work and when. Similarly, the process model of emotion regulation³ explicitly identifies “monitoring” as a key phase in the regulatory cycle, though the specific metacognitive processes involved remain understudied. If metacognitive ability supports accurate tracking of regulatory effectiveness, individuals with better metacognitive skills should be more likely to select strategies that genuinely work and avoid those that consistently fail.

Metacognition refers to “thinking about thinking” and is typically assessed using confidence ratings elicited during a cognitive task. These cognitive tasks tend to be “cold” emotionally neutral tasks. Although metacognitive ability is best known as a predictor of performance on “cold” cognitive tasks e.g.,¹⁴, the same self-reflective monitoring and control mechanisms may also operate in affective contexts, potentially shaping how individuals regulate their emotions. In essence, metacognition could be seen as a monitoring mechanism that provides input to the regulation cycle, supporting self-awareness, self-evaluation, and control over emotional responses^15,16.

Several studies have shown a relationship between self-report measures of metacognition and emotion regulation e.g.¹⁷. However, the validity of self-report measures of metacognition has been criticized for being illogical ¹⁸ and self-report measures of metacognition have been shown to negatively correlate with the accuracy of performance measures of metacognition such as confidence ratings¹⁹.

Emerging evidence supports a meaningful link between how adults regulate their emotions and how accurately they evaluate their own cognition. Rouault et al. surveyed nearly 1000 individuals and found that performance measures of metacognition on a visual dot discrimination task predicted psychopathology symptoms²⁰. Higher anxiety and depressive symptoms were associated with lower overall confidence (metacognitive bias) but heightened metacognitive Efficiency (how well confidence tracks performance). In contrast, a trait dimension characterized by compulsive, intrusive thoughts showed the opposite pattern: overconfidence coupled with lower metacognitive efficiency. Notably, these effects were specific to metacognition, there were no corresponding differences in objective task accuracy. This suggests that how well we monitor our performance may be an important predictor of emotional outcomes.

Current study

The current exploratory study examines whether metacognitive abilities measured during a visual dot discrimination task are associated with self-reported emotion regulation strategies, specifically cognitive reappraisal and expressive suppression. Leveraging an affectively neutral (“cold”) metacognitive task enables us to test whether core monitoring skills generalize beyond the cognitive domain; if such cold-task performance predicts how people regulate their emotions, this would demonstrate a valuable cross-domain transfer. At the same time, because different facets of metacognition (e.g., calibration versus global confidence bias) may influence regulation in distinct ways, we examine multiple indices to capture this nuance.

Research examining correlates of emotion regulation strategies has consistently found that habitual use of reappraisal tends to be associated with more positive psychological outcomes including better well-being, more positive affect, and stronger interpersonal relationships, whereas frequent use of suppression tends to be associated with less positive outcomes including lower well-being, reduced positive affect, and poorer relationship quality^2,21. While these patterns do not imply that reappraisal is universally adaptive or suppression universally maladaptive, as the effectiveness of any strategy depends on contextual factors such as emotional intensity, controllability, and social demands^22–24, they do suggest general tendencies in how these strategies relate to psychological functioning.

Based on these observed patterns and on theoretical models proposing that accurate self-monitoring supports effective emotion regulation³, we hypothesized that better metacognitive ability should be positively associated with greater use of cognitive reappraisal and negatively associated with expressive suppression. Additionally, we measured rumination and self-rated emotional intelligence as covariates to control for their potential confounding effects on the relationship between metacognition and emotion regulation, as both have previously been linked to metacognitive processes²⁵.

Method

Participants

Participants were recruited online using Prolific Academic (https://www.prolific.com), a platform enabling access to diverse community samples. Participants were pre-screened to be residing in an English-speaking country and fluent in English. Rather than recruiting a student sample, we purposefully sampled for gender parity (half male participants, half female participants) and substantial variation in age to obtain a diverse adult community sample. We determined our pre-registered sample size based on 80% power to detect a theoretically meaningful correlation (r = 0.20). A power analysis suggested a minimum sample of 191 was required. As preregistered, we recruited a total sample of 200 participants. Thirteen participants met our pre-registered exclusion criteria and were removed from the final analysis, leaving a final sample of 187 participants (49.20% female; Mage = 42.35, SD = 14.16, range = 18–78 years). This diverse adult community sample allows for greater generalizability of findings compared to studies relying primarily on undergraduate student populations.

Materials and procedure

All experiments were performed in accordance with relevant named guidelines and regulations. The protocol was approved by The University of Sydney Human Ethics Committee (2022/796). Informed consent was obtained from all participants. The study was programmed using jsPsych^26,27 and completed on participants’ personal computers. Participants completed the tasks in the order presented below.

Self-Rated Emotional Intelligence Scale SREIS²⁸ A 19-item self-report scale that assesses four dimensions of emotional intelligence, emotion perception, understanding, regulation, and expression, on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). An example item is “I am aware of the nonverbal messages other people send”.

Ruminative Response Scale RRS²⁹ A 10-item self-report measure assessing the tendency to engage in ruminative thought patterns when feeling down. Responses were made using a 5-point scale from 1 (almost never) to 5 (almost always). An example item is “Think about all your shortcomings, failings, faults, mistakes”.

Emotion Regulation Questionnaire ERQ² A 10-item self-report assessment measuring the tendency to regulate one’s emotions using cognitive reappraisal and expressive suppression. Responses were made on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). An example item is “When I am feeling negative emotions, I make sure not to express them”.

Visual Discrimination Task²⁰ To measure metacognitive abilities we used a visual discrimination task which is perhaps the most common way of measuring metacognition³⁰. In the task, participants judged which of two briefly presented squares contained more dots. Each stimulus displayed two black squares containing a number of white dots, see Fig. 1. The baseline number of dots was fixed at 313. On each trial, one square contained an incrementally greater number of dots, where the increment was adjusted based on performance using a logarithmic staircase (starting log value = 4.25), where the task became more difficult (i.e. smaller dot difference) after two correct responses or easier after one incorrect response. This staircase procedure was done to ensure that performance across participants was approximately matched by adaptively modifying the difference in dots between the two squares to ensure that an optimal difficulty level across participants was maintained. Performance on the task is measured using a d’ metric derived from a signal-detection theory approach.

Fig. 1.

An example trial from the visual discrimination task.

The task consisted of two phases: a practice phase with 25 trials and a test phase comprising five blocks of 42 trials each (210 total test trials). During practice, participants received feedback (green for correct, red for incorrect) and did not provide confidence ratings. In the test phase, participants rated their confidence after each judgment but received no feedback. Confidence ratings were collected on a 6-point scale from 1 (Guessing) to 6 (Certain) using a horizontal slider. In the test phase, the participant’s selected box was outlined in blue without correctness feedback.

Metacognitive ability

Three measures of metacognitive ability were drawn from the visual discrimination task:

Metacognitive sensitivity reflects how well a person’s confidence distinguishes correct from incorrect responses. It is estimated using a signal detection theory (SDT) model—specifically, meta-d’—which fits confidence data from two-alternative forced-choice tasks³¹.

Metacognitive efficiency quantifies metacognitive sensitivity relative to perceptual performance. It is calculated as the ratio of meta-d’ to d’, where meta-d’ reflects the level of perceptual sensitivity that would optimally produce the observed confidence ratings^32,33. A value of meta-d’/d’ = 1 indicates optimal metacognitive use of available information.

Metacognitive bias was also assessed by calculating participants’ average confidence across the confidence ratings. We refer to this as a “bias” because the staircase procedure ensures participants’ performance is matched, thus any difference in average confidence reflects a bias in how participants are using the confidence ratings³⁴.

Transparency and openness

We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study relevant to the current research questions. All online supplemental materials, data, analysis code, and research materials are available on the Open Science Framework (https://osf.io/f8kd9). This study was preregistered (https://aspredicted.org/65d5-vnyt.pdf). We report all relevant studies and analyses that were conducted.

Results

Bi-variate correlations

Descriptive statistics and correlations are reported in Table 1. Focusing on the relationship between metacognition and emotion regulation: metacognitive sensitivity and metacognitive efficiency were both negatively correlated with expressive suppression, such that those with better metacognition were less likely to suppress their emotions. In contrast, neither was significantly associated with cognitive reappraisal. Average confidence was positively associated with both cognitive reappraisal and expressive suppression. Metacognitive bias was also significantly positively associated with rumination and self-rated emotional intelligence.

Table 1.

Means, standard deviations, and correlations.

	M (SD)	d’	Sensitivity	Efficiency	EI	Reappraisal	Suppression	Rumination
Bias	4.15 (0.87)	0.03	− 0.15*	− 0.17*	0.31***	0.23**	0.24***	0.22**
d’	1.28 (0.16)	–	0.11	− 0.14	− 0.03	0.13	0.09	0
Sensitivity	0.96 (0.5)		–	0.96***	− 0.05	0.01	− 0.24**	− 0.09
Efficiency	0.75 (0.39)			–	− 0.05	− 0.04	− 0.25***	− 0.07
EI	3.16 (0.47)				–	0.39***	0.07	− 0.05
Reappraisal	2.91 (0.68)					–	0.23**	− 0.01
Suppression	2.21 (0.91)						–	0.26***
Rumination	12.32 (6.65)							–

* p < 0.05, ** p < 0.01, *** p < 0.001. d’, performance on the visual discrimination task; EI, self-rated emotional intelligence.

Regression

To control for self-rated emotional intelligence, we performed a series of regression models using each of the three metacognition measures as predictors of expressive suppression and cognitive reappraisal, see Supplementary Materials for model results. Controlling for self-rated emotional intelligence, metacognitive efficiency (β = − 0.25, p = 0.001), and metacognitive sensitivity (β = − 0.23, p = 0.001) negatively predicting suppression, while metacognitive bias positively predicted suppression (β = 0.24, p = 0.001). Thus, all measures of metacognition suggest that better metacognitive abilities are associated with lower suppression. After controlling for emotional intelligence, none of the metacognitive measures significantly predicted cognitive reappraisal. We performed an additional set of models to assess whether metacognition predicted rumination after controlling for self-rated emotional intelligence, however, only metacognitive bias predicted rumination, with higher bias associated with more rumination (β = 0.26, p = 0.001).

Discussion

Our results partially support the hypothesized link between metacognitive monitoring and emotion regulation. As predicted, individuals with higher metacognitive ability reported significantly lower use of expressive suppression, a generally maladaptive strategy³⁵. However, contrary to expectations, metacognitive ability was not associated with greater use of cognitive reappraisal. This pattern suggests a nuanced relationship: better self-monitoring may specifically help individuals avoid suppressing their emotions, but does not necessarily translate into more frequent use of reappraisal.

Evidence from recent studies provides converging support for a link between metacognitive processes and emotion regulation. For instance, Temircan³⁶ found that university students with stronger self-reported metacognitive skills tended to rely more on reappraisal and less on suppression. Similarly, a study of pre-service teachers reported a significant positive correlation between self-reported metacognitive awareness and overall emotion regulation capacity¹⁷. These findings, combined with our results, suggest that individuals who are more adept at reflecting on their own cognition may be more likely to use emotion regulation strategies that tend to be associated with positive outcomes in daily life and less likely to rely heavily on strategies that tend to be associated with negative outcomes. Notably, our study extends this literature by employing a performance-based metacognition measure rather than self-report; even so, the link to reduced suppression remained robust.

Contemporary theoretical models emphasize that emotion regulation effectiveness depends critically on flexible deployment of strategies matched to situational demands rather than habitual reliance on any single approach^22,24. From this perspective, better metacognitive monitoring may support emotion regulation not by increasing use of particular strategies per se, but by enhancing awareness of when current regulatory efforts are effective or ineffective. This heightened awareness could prompt strategy switching when needed. Indeed, there is evidence that regulatory flexibility (the ability to deploy different strategies across varying contexts) predicts psychological well-being above and beyond the simple frequency of using any particular strategy e.g.,³⁷. Our finding that metacognitive ability specifically predicts lower suppression may reflect accurate monitoring revealing suppression’s typical ineffectiveness: suppression often fails to reduce emotional experience while simultaneously impairing cognitive performance and social functioning^2,11.

The absence of a significant association between metacognitive ability and cognitive reappraisal deserves careful consideration. While prior research suggests that individuals with high self-awareness or related skills (such as emotional intelligence) tend to engage in reappraisal more frequently e.g.,³⁸, basic metacognitive sensitivity on a perceptual task did not predict reappraisal use in our sample. This pattern may reflect fundamental differences in what metacognitive monitoring facilitates. Accurate monitoring may be particularly useful for identifying when a strategy is ineffective (as suppression typically is) thereby motivating disengagement from that approach. However, increased use of reappraisal may require additional capacities beyond monitoring alone, such as cognitive flexibility to generate alternative interpretations, sufficient strategy knowledge to implement reappraisal effectively, or adequate motivation to engage in the cognitively demanding process of reframing³⁹.

This interpretation aligns with contemporary models emphasizing that effective emotion regulation involves not just selecting better strategies, but flexibly adjusting regulatory approaches based on contextual demands^5,24. From this perspective, metacognitive monitoring might support adaptive regulation primarily by facilitating disengagement from ineffective strategies rather than by driving increased use of alternatives. An individual with strong metacognitive skills might accurately perceive that suppression is failing to achieve their regulatory goals and consequently reduce its use, but this awareness alone may not translate into more frequent reappraisal unless other enabling factors are present. Future research employing experience sampling methods could examine whether metacognitive ability predicts flexible, context-appropriate deployment of different strategies rather than global increases in reappraisal frequency. Training studies could also test whether improving metacognitive insight (for instance, via mindfulness or feedback-based training) leads to more adaptive strategy selection in naturalistic emotion-eliciting contexts.

Although greater metacognitive sensitivity and efficiency were linked to lower suppression, our data also showed that higher mean confidence, an index of metacognitive bias, was positively related to both suppression and reappraisal. This suggests that feeling generally more certain about one’s judgments does not straightforwardly map onto adaptive regulation; instead, it may reflect a global confidence style that accompanies the habitual use of any strategy, adaptive or otherwise. However, it is also notable that the counterintuitive relationship between bias and reappraisal was not significant after self-rated emotional intelligence was controlled for as a covariate. Nonetheless, future work should examine how distinct facets of metacognition differentially predict emotion regulation.

Limitations and future directions

Several limitations warrant consideration when interpreting our findings. First, the correlational nature of our design precludes causal inferences about the relationship between metacognition and emotion regulation. While our results are consistent with the hypothesis that metacognitive ability influences regulatory strategy selection, alternative explanations remain plausible. For example, frequent use of certain emotion regulation strategies might influence metacognitive development, or third variables such as general self-awareness could independently predict both constructs. Experimental or longitudinal designs manipulating metacognitive ability or tracking its development alongside emotion regulation patterns would provide stronger evidence for directional relationships.

Second, we relied exclusively on self-report measures to assess emotion regulation strategies. While the Emotion Regulation Questionnaire (ERQ) is well-validated and widely used, self-reports may be subject to various biases including social desirability, limited introspective access, or theory-driven responding⁴⁰. Individuals may inaccurately report their regulatory strategy use due to poor metacognitive awareness of their own emotional processes^19,41. Future research could complement self-report measures with behavioral observations, experience sampling in daily life, or laboratory assessments of actual regulatory behavior to provide convergent evidence.

Third, we assessed metacognitive ability using a single “cold” perceptual task involving visual discrimination. While this approach has important advantages, including objective performance measurement and well-established analytical methods, it raises questions about generalizability to metacognition in affective contexts. The relationship between domain-general metacognitive ability and emotion-specific metacognition (meta-emotion) remains unclear⁴². Metacognitive monitoring of emotional states may involve distinct neural systems or cognitive processes compared to monitoring of perceptual decisions. Future work should examine whether our effects replicate using metacognitive tasks with emotional content or measures of meta-emotional awareness, which might show stronger or different associations with emotion regulation strategies.

Fourth, we focused on only two emotion regulation strategies (reappraisal and suppression) despite growing recognition that individuals employ a diverse repertoire of strategies in daily life^13,43. Other important strategies including distraction, acceptance, problem-solving, and social sharing each have distinct cognitive demands and may relate differently to metacognitive ability. For instance, acceptance-based strategies might show different patterns of association with metacognitive monitoring compared to cognitively demanding approaches like reappraisal. Examining a broader range of strategies would provide a more comprehensive understanding of how metacognition relates to emotion regulatory repertoires.

Finally, our sample consisted of English-speaking volunteers recruited through an online platform, primarily from Western, educated, industrialized, rich, and democratic (WEIRD) populations. While our diverse adult community sample represents an improvement over typical undergraduate samples, generalizability to other cultural contexts remains uncertain. Cultural norms strongly shape emotion regulation preferences and effectiveness⁴⁴, and cultural factors might moderate the relationship between metacognition and strategy selection. Cross-cultural research would clarify whether the link between metacognitive ability and reduced suppression generalizes across different cultural contexts or reflects Western norms emphasizing emotional expression.

Conclusion

These findings have important implications for theoretical models of emotion regulation. Contemporary frameworks such as Gross’s extended process model (2015) posit that effective regulation entails iterative monitoring of one’s emotional state and regulatory outcomes. Yet the monitoring phase of emotion regulation has been relatively understudied. Our results help address this gap by empirically linking a core monitoring capacity to the use of an emotion regulation strategies. This supports the notion that accurate self-monitoring can shape regulatory behavior, presumably by alerting individuals when their current strategy proves ineffective. Those with greater metacognitive acuity may be more attuned to the typical ineffectiveness of suppressing emotional expressions, thereby reducing reliance on this approach. Our findings suggest that metacognitive ability relates to reduced reliance on suppression, though it does not necessarily foster greater use of reappraisal. This highlights self-monitoring as a selective, yet potentially consequential, component in the emotion-regulation cycle and underscores the potential value of future research examining whether metacognitive training could help individuals refine how they manage their emotions.

Author contributions

K.S.D. conceived and designed the study, conducted data collection and analysis, interpreted the results, and wrote the manuscript. The author approved the final version of the manuscript and is accountable for all aspects of the work.

Funding

This research was funded by an Australian Research Council fellowship awarded to Kit S. Double (DE230101223).

Data availability

All online supplemental materials, data, analysis code, and research materials are available on the Open Science Framework (https://osf.io/f8kd9/).

Declarations

Competing interests

The authors declare no competing interests.