Using objective measures to evaluate the role of physical activity as a compensatory mechanism of emotion regulation

Emotion regulation (ER) is the process of altering one’s emotional experience or expression (i.e., emotion type, frequency, or intensity) to better align with one’s goals (Gross, 1998). Successful ER, which may include decreasing negative or increasing positive emotions, is associated with emotional well-being (Balzarotti et al., 2016; Quoidback et al., 2010) and physical health (Butler, 2011; Cloitre, et al., 2019; Trindade et al., 2018). For example, there are widely replicated effects showing that emotion regulation success is associated with reduced symptoms of depression (Jorrman & Siemer, 2011; Wirtz et al., 2014), and increased measures of well-being (Cutuli, 2014; Gross & John, 2003a), including greater longevity (Xu & Roberts, 2010) and life satisfaction (Jiang et al., 2022; Kornienko & Rudnova, 2023).

One well-studied ER strategy that has been associated with better mental health and well-being is cognitive reappraisal: the process of re-framing one’s thoughts and interpretations in order to change one’s emotional experience or response (Gross, 1998, 2015). In particular, cognitive reappraisal has been associated with increased positive affect (Andreotti et al., 2013; Mauss et al., 2007), self-esteem (Gross & John, 2003b; John & Gross, 2004), and many additional factors indexing positive well-being (Ripenhausen et al., 2022). Alternatively, expressive suppression is a response-focused ER strategy that involves the inhibition of ones’ experience, behavior, and physiological response to an emotion-eliciting event to prevent an emotional experience or expression (Gross, 2001). Compared to cognitive reappraisal, suppression requires less cognitive resources, occurs later in the emotion regulation process (after an emotion has been elicited), and has been associated with maladaptive outcomes for emotional well-being, including increased negative emotions (Srivastava et al., 2009), reduced emotion recognition (Yan et al., 2022), and increased anxiety severity (Al-Wardat et al., 2024).

Interestingly, physical activity (PA) can been considered a behavioral form of ER (Al-Wardat et al., 2024; Daniela et al., 2022; Zhang et al., 2019), although the majority of this work has used self-reported measures of PA. Similar to cognitive reappraisal, self-reported PA has been associated with enhanced emotional well-being, as indexed by reduced symptoms of depression and anxiety (De Mello et al., 2013; McMahon et al., 2017), increased positive emotions (Landers & Arent, 2007; Pasco et al., 2011), and higher self-esteem (Zamani Sani et al., 2016). Previous research has also found associations between greater amounts of PA and lower emotion dysregulation (Rezaie et al., 2023), as well as cognitive processes that support ER (e.g., inhibitory controland executive functions; Guiney et al., 2019; Langois, et al., 2013; Salas-Gomez et al., 2020).

Only a few studies have investigated cognitive reappraisal specifically in relation to PA. These studies found greater self-reported habitual PA was associated with reappraisal success (Giles et al., 2017; Ligeza et al., 2019; Perchtold-Stefan et al., 2020) and a greater tendency to reappraise in daily life (Wu et al., 2022). Alternatively, suppression and PA may have independent associations with well-being, with one study reporting that PA may be a more effective alternative than suppression for inhibiting emotional reactivity without hindering the experience and expression of emotions (Al-Wardat et al., 2024). However, only weak-to-moderate associations have been found between subjective and more objective measures of PA (Prince et al., 2008). These differences may be, in part, due to biases and subjectivity within self-reported PA measures, leading to inaccuracies in the data. Thus, the link between PA and ER needs to be confirmed with more objective measures.

Other work has examined acute effects of exercise on ER. In line with self-reported PA findings, this work largely reports positive effects of exercise on ER, such that, following a bout of exercise, participants perform better in a reappraisal task (Giles et al., 2018; Liu et al., 2022), and show improved cognitive and emotional outcomes (Bernstein & McNally, 2017; Lucas et al., 2012; Rebar et al., 2015; Reed & Ones, 2006). Relatedly, Bernstein and McNally (2018) found that individuals who reported greater difficulty with ER experienced better emotional recovery after a bout of exercise, suggesting that the exercise may have mitigated the typical effect of having ER difficulties on delaying emotional recovery. However, experimental studies are limited by ecological validity, and more objective measures of activity are required to determine that previous results were not a product of methodological limitations or artifacts. Further, these relationships with emotion regulation have most often been found in the context of moderate-to-vigorous intensity exercise (rather than light intensity exercise, such as walking or stretching; Nakagawa et al., 2020). Given that PA—encompassing all activities that exert energy (World Health Organization, 2024 )—is distinct from exercise (e.g., time course, intent, activity type), there may be differences between PA-related and exercise-related effects on cognitive reappraisal.

To address the need for more objective and ecologically valid measures of PA in the ER literature, this study examined accelerometer-derived measures of naturalistic accelerometer-derived PA (i.e., actigraphy) in relation to cognitive reappraisal and suppression in adults. While cognitive reappraisal was the primary ER strategy of interest, suppression was included as a comparison form of ER; given important differences between cognitive reappraisal and suppression on cognitive resources and well-being, examining both may clarify whether PA is differentially associated with cognitive reappraisal versus suppression, helping to identify which aspects of ER are most influenced by PA. Actigraphy data was collected with research-grade ActiGraphs®, which have been increasingly used in PA research due to their ability to capture naturalistic PA (Han & Wang, 2017) and strong correlation with doubly-labeled water (i.e., the current “gold standard” measurement of energy expenditure; Chomistek et al., 2017). Further, self-reported measures of PA have been found to over-estimate MVPA, showing both self-report and sex biases, and may be less accurate compared to accelerometry (Dyrstad et al., 2014). Building on previous work that established a role for moderate-to-vigorous exercise in ER (Bernstein & McNally, 2018; Giles et al., 2018; Liu et al., 2022), we focused on moderate-to-vigorous PA (MVPA). Moderate intensity activity can be defined as the exertion of energy through which one can talk—but not sing (i.e., speed walking, yard work), whereas during vigorous intensity activity, one cannot say more than a few words at a time (i.e., running or hiking uphill; Centers for Disease Control & Prevention, 2022). Due to the cumulative effects of PA on physical health (Saint-Maurice et al., 2018) and large overlap in potential benefits of PA for physical (Hoon Lee, et al., 2022) and mental health (Nakagawa et al., 2020), moderate and vigorous activity are often reported together.

ER was measured through self-reported tendency to use reappraisal in daily life (i.e., trait reappraisal), and compared to self-reported tendency to use suppression (i.e., trait suppression). Using these measures, we assessed whether engaging in greater MVPA in daily life is associated with more cognitive ER usage (i.e., reappraisal). To account for differences in cognitive ER ability and the role of PA in the face of emotional challenges, this relationship was additionally examined in the context of self-reported ER difficulties (which has been negatively associated with trait reappraisal; Sörman et al., 2022) and self-esteem, a non-ER related measure of emotional well-being that has previously been associated with self-reported PA (Zamani Sani et al., 2016).

Considering previous methodological limitations, we were interested in clarifying the role of MVPA in supporting emotional well-being. Given the impact of MVPA on executive functioning and emotional reactivity, it is possible that individuals who engage in higher amounts of MVPA also have higher trait reappraisal, and therefore, have the highest emotional well-being. Alternatively, PA may only act as a compensatory from of regulation, such that it supports emotional well-being in the face of less reliance on adaptive strategies, like cognitive reappraisal. Our hypotheses were two-fold. First, we predicted that PA would be positively associated with trait reappraisal, such that individuals who engaged in greater amounts of MVPA would report higher trait reappraisal. Second, we predicted that PA would act as a compensatory form of regulation, such that MVPA would buffer reliance on reappraisal to mitigate ER difficulties and to maintain high self-esteem. In other words, in individuals who engaged in less MVPA, trait reappraisal would be associated with higher self-esteem and lower ER difficulties, but in those that engaged in greater MVPA, the link between trait reappraisal and self-esteem or ER difficulties would be weaker or not significant. Considering that trait suppression requires fewer cognitive resources, we did not expect MVPA to moderate the relationships between trait suppression and emotional well-being (Al-Wardat et al., 2024).

Sample sizes and descriptive statistics are reported in Table 1. Variability in Ns between actigraphy and survey data arose due to date of inclusion in study protocols (i.e., pre/post-funding acquisition). A multiple imputation analysis confirmed no significant differences in outcome variables between original and imputed datasets, indicating that missing data did not affect our results. Welch's Two Sample independent t-tests were conducted to explore sex differences in quantity of MVPA. Results indicated no significant difference in average MVPA by sex (p = .255). However, average MVPA was inversely correlated with age, such that MVPA significantly decreased with increasing age across the sample (rho = − .18, p = .004). Additionally, there was a significant difference in average MVPA by race, such that White participants (n = 224), engaged in significantly higher MVPA compared to Asian participants (n = 8, t(15.09) = − 2.65, p = .018) and those who reported more than one race (n = 8; t(16.02) = − 3.77, p = .002). Similarly, the effect trended in the same direction for White compared to Black or African American (n = 6; t(13.16) = − 1.82, p = .092).

Engaging in higher amounts of MVPA may protect well-being (i.e., trait reappraisal), as indexed by ER difficulties. As expected, we found that higher trait reappraisal was associated with lower ER difficulties, and that ER difficulties and self-esteem were inversely related (Antunes et al., 2021; Kim et al., 2017; Sörman et al., 2022; Zamani Sani et al., 2016).

Interestingly, MVPA was not directly correlated with trait reappraisal. Although this finding was contrary to our predictions, limited research has examined the link between MVPA and trait reappraisal, making it unclear whether a direct relationship should be expected. While one previous study reported a positive association between the two (Nakagawa et al., 2020), another found an indirect association between self-reported MVPA and trait reappraisal through affective response during MVPA (Gürdere et al., 2024). The absence of a direct association between MVPA and trait reappraisal suggests that their respective relationships with well-being may operate through separate mechanisms, rather than a shared pathway. Here, we found that MVPA moderated the relationship between trait reappraisal and ER difficulties. In other words, there was a stronger inverse relationship between trait reappraisal and ER difficulties in people who were less physically active. In contrast, this effect was weaker with increasing PA, ultimately becoming non-significant. Contrary to our hypotheses, parallel effects were not found for self-esteem, indicating that the effect of MVPA on well-being may be specifically related to cognitive emotion regulation-specific processes.

These findings provide support for our hypothesis that MVPA may act as a compensatory form of ER, through protecting emotional well-being in people who use reappraisal less frequently. Previous work on self-reported PA supports this theory as well; engaging in higher amounts of PA has been associated with the use of more adaptive forms of reappraisal (i.e., positive reinterpretation reappraisal; Perchtold-Stefan et al., 2020), as well as higher trait reappraisal (Wu et al., 2022). Although distinguishable from PA, bouts of exercise have additionally been found to boost cognitive reappraisal success (Giles et al., 2018), and emotional recovery (Bernstein & McNally, 2018) in a dose-dependent manner. These benefits may be sustained with higher MVPA—or the accumulation of bouts of high-intensity activity—thereby increasing tolerance to future stress (Perchtold-Stefan et al., 2020). Reappraisal quality (i.e., detail, uniqueness, and plausibility) has also been found to be an important factor in the success of reappraisal in reducing negative affect (Southward et al., 2022), highlighting a potential pathway from exercise to reappraisal success, and therefore trait reappraisal.

As MVPA did not moderate the relationship between trait suppression and ER difficulties, or trait suppression and self-esteem, our findings were specific to reappraisal. Although speculative, this may indicate that the influence of PA on ER operates via its impact on cognitive resources. Indeed reappraisal has been associated with better executive functioning compared to suppression (Lantrip et al., 2015). PA is also thought to influence cognitive processes (Zaehringer et al., 2018), such as cognitive flexibility and working memory capacity (Lerche et al., 2018; Shi et al., 2022; Zhao et al., 2024). However, more work is needed to clarify whether these functions explain the relationship between PA and trait reappraisal (Toh & Yang, 2022).

Several important limitations are noted. First, due to the cross-sectional nature of this study, directionality of these effects cannot be established. Future longitudinal and experimental work is needed to investigate causal mechanisms through which MVPA may support trait reappraisal in individuals with ER difficulties, including emotional reactivity, cognitive functions involved in reappraisal, or additional variables associated with trait reappraisal. These include exposure to stress (Perchtold-Stefan et al., 2020), PA context (i.e., leisure time versus occupational PA; Sofi et al., 2007), and PA-related ER efficacy (Sudeck et al., 2018). Second, as our sample was predominantly White and Non-Hispanic/Latina(o), our findings may not be generalizable to all races and ethnicities.

Third, there are several limits to actigraphy data that are important to note. Actigraphy data were collected for a minimum of five days, which, while standard in the literature, may not fully capture participants' typical activity levels. Future research could employ longer windows to enhance the representativeness of estimates. Further, according to Pedišić & Bauman (2015), actigraphy data may not accurately represent energy expenditure on all movements, including biking and weightlifting. Due to increased sensitivity to upper body motion (e.g., during cooking, cleaning, or even expressive gestures), research has shown that wrist-worn actigraphy tends to yield higher MVPA estimates than hip-worn devices, especially in free-living conditions (e.g., Rowlands et al., 2019; Skrede et al., 2017). Indeed, the MVPA reported in our sample was high (3.62 h per day, on average). Notably, the MVPA cut point we used—based on Montoye et al. (2020)—included moderate activities such as 'picking up light items,' 'brisk walking,' 'treadmill walk,' 'stairs,' and 'short walk.' As these activities may not be perceived as ‘exercise,’ they may be classified as light activity in self-report data (Esliger & Tremblay, 2007).

Additionally, it is possible that our sample engaged in above average levels of PA. Although we did not collect employment data, the Midwest area in which data were collected contains a high percentage of manual labor jobs (25–26% of the workforce, according to the U.S. Bureau of Labor Statistics (2023)). These elevated activity levels should be considered when interpreting the generalizability of the findings, as the observed associations may not extend to more sedentary populations. Finally, cut point thresholds vary person-to-person, likely leading to some misclassification between moderate and light intensity.

To reduce these ambiguities, future research could code and control for activity type, as well as employment type, and should aim to replicate these results in samples with more typical levels of physical activity. Future studies could also investigate continuous measures of intensity and additional sociodemographic factors that have been found to be associated with watch-wearing adherence and compliance to study protocols (Pedišić & Bauman, 2015).

This study contributes to a body of literature disentangling the complexities of the relationship between PA and ER. Our results provide further confirmation that PA is an important component of emotional—as well as physical—health, particularly for individuals with relatively lower use of reappraisal who struggle with cognitive emotion regulatory difficulties.