How Are You Feeling Today? Dynamic and Static Indices of Daily Affect Predict Psychological Adjustment One Year Later in a Multi-cohort, Longitudinal Investigation

It is well established that the presence of both acute and chronic stress pose risk for the development of affective symptoms and related disorders (Grant et al., 2004; Shields & Slavich, 2017). Yet how daily reports of emotion during those moments may or may not inform adjustment remains still less-well understood. Some evidence suggests that emotional responses drive key features of mental health trajectories, from adaptive and protective coping such as support seeking or physical exercise (Aurora et al., 2022; Nylocks et al., 2018) to maladaptive risk behaviors, such as substance use or binge eating (Coifman & Aurora, 2022; Johnson et al., 2013). Moreover, affective disorders broadly defined (e.g., depression, anxiety, stress, substance and eating disorders: Kring, 2008; Barlow et al., 2014) are characterized by high and sustained levels of negative affect and low but variable levels of positive affect (Barlow et al., 2014; Clark & Watson, 1991; Khazanov & Ruscio, 2016). Indeed, the capacity to generate and maintain positive emotional experiences is considered an etiological risk factor for psychiatric illness broadly, predicting risk over decades (Kendall et al., 2015). However, when considering the scientific meaning of emotion reported in-vivo, self-reports of emotion reflect not only the relative intensity -but- the manner in which emotional experiences are conceptualized (Barrett et al., 2001). For example, the Conceptual Act model of emotion suggests that emotions are constructed based on how an individual conceptualizes core affective phenomena, including arousal (e.g., based on interoceptive cues), valence, and contextual factors (Barrett, 2006). Hence, how reports of emotion are associated with long-term psychological outcomes may also be particularly relevant to clinical models of emotion and emotion-related risk. Indeed, the subjective experience (or conceptualization) of emotion are the primary target of dominant psychotherapeutic interventions for affective disorders, including Cognitive Behavioral Therapy (Beck, 1995); Dialectical Behavioral Therapy (Linehan, 1993); and Acceptance and Commitment Therapy (Hayes et al., 1999).

Methodological developments over the past two decades have facilitated increased research examining the dynamic nature of emotion in daily life. Daily emotion reports can be interpreted in terms of both static features, such as person-mean level, as well as dynamic indices, with increasing scientific emphasis placed on testing the relevance of dynamic indicators to psychological functioning (Trull et al., 2015). However, both static and dynamic indices of emotion tend to overlap (Dejonckheere et al., 2019a, 2019b), and research is needed to determine their unique contributions to health—as well as how they interact to predict longitudinal outcomes. Importantly, dynamic indices of affective conceptualization can capture affective phenomenon, such as emotion flexibility, that have been broadly attributed to psychological health, in ways not typically evident for static indices (Coifman & Summers, 2019). This is likely because dynamic constructs capture the individuals’ conceptualization of emotion with time in mind, indexing many moments of emotion report, across variable contexts, within one individual to derive an estimate.

Emotional inertia is a dynamic index that captures the tendency for emotions to be reported as persistent or rigid over time and is measured by indexing the within-person, serial, autocorrelation of reported negative—or—positive affect (Koval et al., 2016; Kuppens et al., 2009). For example, high inertia in reports of negative affect from moment-to-moment likely suggest persistent inflexibility in the appraisal of emotional events and experiences across contexts, with greater negative emotional inertia predicting lower wellbeing and risk for psychopathology (Koval et al., 2016; Kuppens et al., 2009). Emotional inertia may reflect a disconnection between the conceptualization of emotional experiences and the typical variability in contextual demands that is present in daily life (Coifman & Summers, 2019) as the individual’s emotional state is resistant to change even when contexts shift (Kuppens et al., 2009). Indeed, research has shown that context-insensitive emotional responses (e.g., low sadness in response to sad stimuli, low positive affect in response to positive stimuli) are a characteristic of psychiatric disorders, like depression (Rottenberg et al., 2005), as well as maladjustment in times of stress (Coifman et al., 2012). Although the negative effects of negative emotional inertia are well-established, research on the role of positive emotional inertia in wellbeing and psychological functioning has been mixed. Prima facie, persistent positive emotions might be a sign of psychological health, rather than inflexibility, as positive emotions tend to be highly adaptive (Fredrickson, 2001; Gruber et al., 2013), even when expressed in negative contexts (Coifman & Bonanno, 2010; Harvey et al., 2016; Papa & Bonanno, 2008). Indeed, some research suggests that greater positive emotional inertia predicts fewer depression symptoms in the general population and greater response to treatment in patients with depression (Höhn et al., 2013). However, other research on positive emotional inertia has found an association with negative indicators of well-being, including increased depressive symptoms (e.g., Koval et al., 2016). The adaptive (or maladaptive) role of positive emotional inertia may depend on individual-level factors, such as trait level or dispositional tendencies to appraise experiences. Importantly, positive emotional inertia indices can reflect either persistent high or persistent low levels of positive emotion over time. Thus, within a person, positive emotional inertia is most likely only adaptive when mean levels of positive emotion are also high. However, limited research has tested interactions between dynamic emotion indices (e.g., emotional inertia) and more static indices (e.g., mean intensity of positive or negative affect).

Emotion polarity is an alternative dynamic index which captures the momentary within-person association of negative to positive emotional experiences (Rafaeli et al., 2007). Highly consistent with clinical models of dichotomous thinking (Beck, 1995), polarized reports of emotional experience reflect a rigid appraisal of an experience as either all bad or all good. A broad literature has examined how this phenomenon is typically more prevalent when individuals are under greater stress, varying both within and between individuals (Dowd et al., 2010; Zautra et al., 2001). Importantly, highly polarized emotional reports during stress can predict maladaptive outcomes including increased depression eighteen months following bereavement (Coifman et al., 2007) and are characteristic of specific high-risk patient groups, predicting increased reliance on maladaptive behaviors (i.e., substance use, binge eating, self-injury, and risky sexual behavior in Borderline Personality, Coifman et al., 2012). Generally, individuals demonstrate less polarity in their emotional conceptualization during periods of low perceived stress or when psychological resources are abundant (Reich et al., 2001; Zautra et al., 2001). There are also key individual differences and evidence that some individuals can maintain low polarity even during heightened stress (Coifman et al., 2007). However, it may also be that polarized responses reflect active coping during acute periods of stress. For example, elevated reports of both negative and positive affect are associated with greater cardiovascular recovery following an acute stressor in lab (Dowd et al., 2010) and daily sampling evidence suggests that polarized reports of negative and positive affect may facilitate coping, oscillating widely when needed, gradually abating over days and weeks (e.g., in bereavement: Bisconti et al., 2004). Indeed, polarized affective responses experienced in daily life may be a key component of psychological health and wellbeing. However, their influence on long-term mental health outcomes, particularly when other indices of affect reporting are considered, remains unclear.

Understanding the unique role of daily emotion in long-term psychological health and adjustment is critical to further developing targets for treatment of many common and burdensome emotion-linked psychiatric conditions. Although many psychotherapies target moment-level conceptualization of emotion, including many “third-wave” cognitive and behavioral therapies (e.g., dialectical behavioral therapy; Dimidjian et al., 2016; Linehan, 1993), these techniques and their emphasis vary widely. Prior research has shown that emotion dynamics can predict the course of psychopathology. For example, daily emotion dynamics have been shown to predict the course of depressive symptoms (Wichers et al., 2010) and influence treatment responsiveness to antidepressants and psychotherapy (Husen et al., 2016; Wichers et al., 2012). However, how dynamic emotion indices uniquely influence long-term psychological health may also simply be attributable to dispositional tendencies that are commonly present in clinical samples. Increasingly, models of psychopathological risk and emotion-related disease states attribute a large proportion of the variance to common underlying general factors (Conway et al., 2022; Kotov et al., 2017). Indeed, relatively little research has considered how in-vivo emotional processes, including moment-level conceptualizations of emotional experience, can reflect and/or can be influenced by more stable dispositional tendencies that are known to drive risk and relapse for common affective disorders (Barlow et al., 2014). Considerable research has converged around the relative importance of trait negative affectivity (Shackman et al., 2016) in driving both momentary reports of emotion and in long-term processes of psychological adjustment. Indeed, evidence across disciplines has demonstrated the role of general affective tendencies in shaping patterns of emotion-related processing and health, even decades later (e.g., Behavioral Genetics: Gilman et al., 2015; Caspi et al., 2010; Neuro-imaging: Hostinar et al., 2017; Personality: Wrzus et al., 2021). In sum, dynamic and static reports of daily emotion may have important relevance to models of psychological health but this must also be considered in relation to dispositional factors. Unpacking these interacting relationships will further the understanding of emotion-related risk, and is essential for tailoring treatments to specific patients, and to develop more effective tools to assess and identify individuals at high risk.

In this investigation, we examined emotional experiences reported in daily life as predictors of psychological health one year later, in two representative community samples of U.S. adults. Using a multi-cohort longitudinal design, we investigated how dynamic conceptualizations of emotion, operationalized as emotion polarity, NE and PE inertia, and static indices of affect intensity and variability, extracted from daily reports of affect over 10 days, were associated with psychological symptoms, loneliness, and psychological wellbeing 1 year later. Results from both cohorts indicated that dynamic indices of affect were associated with long-term psychological health, but that they only rarely accounted for unique variance in outcomes when controlling for static indices of affect intensity. Rather, mean negative daily affect predicted higher reports of symptoms and mean positive daily affect tended to predict lower reports of loneliness and higher reports of wellbeing. The one dynamic emotion indicator that was an exception was that greater PE inertia predicted lower loneliness and higher wellbeing, above and beyond static indices (as well as trait anxiety) in cohort 1. However, in cohort 2, the protective effect of PE inertia was moderated by mean intensity of positive emotion, exerting benefits only for those with high intensity PA across days. Our results also demonstrated that although trait anxiety was the strongest and most consistent predictor of maladaptive longitudinal outcomes (as expected), daily reports of affect, particularly static, but in some cases dynamic (PE inertia), did still have unique, albeit small, associations with outcomes. Because daily affective experience, and in particular the ways in which individuals understand their affective experience, is a common focus in contemporary psychotherapies, these findings have clear relevance as they relate to targets for intervention and models of risk.

Although some effects were robust and consistent across samples (e.g., effects of trait anxiety, mean NA, and mean PA), it is possible that some of the subtler discrepancies (e.g., PE inertia) are due to contextual differences related to the COVID-19 pandemic. Indeed, we tested these questions at two distinct timepoints: the initial months in which lockdowns were in place across the country (March and April 2020) and during the first winter peak in cases (January and February 2021). Therefore, results could be due to differences in the experience of stress, or in the amount of social interaction participants were engaging in during these unique pandemic contexts.

Recent methodological and theoretical innovations have led affective scientists to emphasize the importance of emotion flexibility, with a focus on the role of individual conceptualizations of affective experience (Coifman & Summers, 2019). With this framework in mind, we investigated if reports of more flexible affective experiences (e.g., lower NE inertia and less polarized reports of negative and positive emotion) would be adaptive for long-term mental health outcomes when considered against static indices of affect—and—stable and robust predictors of outcomes.

Results from the current investigation suggested that more flexible conceptualizations of affective experiences—in the form of lower emotion polarity—were generally protective. However, when static indicators were also accounted for, those impacts fell away. This finding is surprising, given that past research has shown meaningful relationships between emotion polarity and psychological health even when controlling for mean affect intensity and variability (Dejonckheere et al., 2018, 2019a, 2019b). However, unique effects in past research were small, and more recent research did not find evidence for longitudinal associations between polarity and wellbeing (also during the pandemic; Dawel et al., 2023). Moreover, there are important methodological differences between the present investigation and past research worth noting. Specifically, past research has conceptualized emotion polarity as the outcome, rather than the predictor (e.g., Dejonckheere et al., 2018, 2019a, 2019b), which might explain some of these discrepancies. On the other hand, the COVID-19 pandemic created an unprecedented source of stress, and past research has suggested that less polarized affective experiences are hard to retain under stressful conditions (Coifman et al., 2012; Zautra, 2003) likely due to depleted psychological resources (Davis et al., 2004). Indeed, in now two investigations administered during the pandemic (that we are aware of), the relationship between polarity and wellbeing-related measures has not been robust when controlling for other factors (Dawel et al., 2023). To attempt to reconcile disparate findings in the literature, future research should consider moderators and mediators of the relationship between emotion polarity and wellbeing-related factors, including stressful contexts (e.g., Coifman et al., 2012) and emotion regulation capacity or flexibility (e.g., Dejonckheere et al., 2018), as well as the directionality of these relationships within a longitudinal framework.

Consistent with previous research, greater NE inertia predicted greater reports of symptoms, loneliness, and lower wellbeing (Koval et al., 2016; Kuppens et al., 2009) one year later. However, like the results of emotion polarity, these effects did not survive when controlling for static levels of NA and PA. Indeed, much of the past work on emotional inertia has not controlled for both mean levels of NA and PA, and thus has not accounted for the shared variance between these sometimes, interdependent constructs (Rafaeli et al., 2007). A recent investigation into the relative importance of dynamic indices of affect suggested that emotional inertia may have limited added value for the prediction of outcomes such as depression symptoms, borderline symptoms, and life satisfaction (Dejonckheere et al., 2019a, b). However, we would argue that these findings still have clinical importance. NE inertia reflects persistent reports of negative emotion across time, reflecting rigidity in the conceptualization of emotion. This rigid pattern of emotion is consistent with dominant models of emotion-related risk for affective disorders (i.e., Joormann & Gotlib, 2010). However, our results are also consistent with recent research (e.g., Dejonckheere et al., 2019a, b) that suggest NE inertia provides only limited incremental value in the prediction of longitudinal outcomes.

Unlike the NE inertia findings that were broadly consistent across cohorts, PE inertia did have interesting diverse effects by cohort. In cohort 1, greater PE inertia predicted lower loneliness and higher wellbeing, and these effects remained significant after controlling for (1) affect intensity, (2) affect variability, and even (3) trait anxiety, suggesting that the effects were robust. Hence, reporting persistent levels of positive affect during the baseline measurement was linked to better psychological health one year later. Interestingly, the effects were more complicated in cohort 2. Indeed, the inconsistency between cohorts 1 and 2 is consistent with the broader PE inertia literature (e.g., Höhn et al., 2013; Koval et al., 2016). These mixed findings may be the result of how emotional inertia is computed, where greater scores on the PE inertia index reflect only the persistence of reports from one moment to the next. Hence, greater PE inertia scores can reflect both persistent, but low, or persistent, but high reports of PA, making the interpretation of findings difficult. Thus, by testing PE inertia’s interaction with mean intensity levels of PA, the results gained added nuance. Specifically, in cohort 2, PE inertia was protective against symptoms only at higher mean intensity PA. These findings have clear clinical implications, suggesting that persistent elevated PE in daily experiences can be adaptive and is associated with psychological health and wellbeing. Although the test of moderation did not reach significance in cohort 1, the same relationships appeared to be present. Future research on PE inertia should continue to consider its interaction with PA intensity to better parse its effects on indices of adjustment and wellbeing, which can depend on overall PA intensity.

Finally, mean levels of affect, reflecting broader estimates of mood during our 10-day assessment, remained consistent and robust predictors of longitudinal health, even when controlling for dispositional risk, operationalized as trait anxiety. Negative affectivity is well-established for its stable relationship with psychopathology and other indices of wellbeing across the lifespan (Griffith et al., 2010; Shackman et al., 2016). Thus, by controlling for trait anxiety, we were able to isolate the relative role of state emotion indicators reported in the diary. As expected, trait anxiety was indeed the most consistent predictor of all three outcomes. However, mean levels of NA and PA also exerted small but meaningful effects on the outcomes. Specifically, mean NA was most predictive of psychological symptoms, whereas mean PA was most predictive of loneliness and wellbeing. Given that the shared variance between dispositional negative affectivity and affect reports is considerable (here rs at approximately 0.50), this suggests that daily affect remains a useful target for interventions targeting adjustment processes over time and psychological health. Several clinical implications follow from the results of the current investigation. First, modern psychotherapy approaches (e.g., CBT) emphasize more diverse and flexible conceptualizations of emotional experience, consistent with our findings. Although dynamic indices (polarity and NE or PE inertia) were relatively less meaningful than affect intensity in our fully inclusive models of long-term psychological health, our findings suggest that daily conceptualizations of affective experience are still important and can predict important variance in outcomes one year later. However, our results suggest that interventions targeted at decreasing negative affect and increasing positive affect, at daily intervals, may be most effective. Indeed, research has shown that daily negative affect is associated with increases in maladaptive cognition and coping in patients with psychopathology (e.g., depression), which in turn maintains disease states, whereas daily positive affect is associated with more adaptive coping (Dunkley et al., 2017). In addition, the reduction of daily negative affect is a mechanism of symptom reduction for several treatments (e.g., CBT: McIntyre et al., 2019; Mindfulness-based stress reduction: Snippe et al., 2017) and the increased generation of positive emotion, a cornerstone of positive psychological intervention (e.g., Seligman et al., 2005).

Findings from the current study relating to PE inertia and mean PA have additional clinical implications. First, it has long been established that positive emotions promote physical health and wellbeing through several channels (Fredrickson, 2004). For instance, experiences of positive emotion serve to down-regulate (or “undo”) the negative effects of bouts of negative emotion (e.g., cardiovascular recovery; Behnke et al., 2023; Fredrickson et al., 2000). In addition, positive emotions foster social connectedness, which in turn, promotes increased positive emotion, access to social resources, and wellbeing (Kok & Fredrickson, 2010; Ong & Allaire, 2005). Moreover, daily positive emotions promote healthy behaviors (e.g., exercise, relaxation and hobbies) in both healthy and impaired (e.g., depression and anxiety) individuals, likely through increased motivation to approach goals (e.g., exercise, support-seeking; Aurora et al., 2022; Nylocks et al., 2018). Results from the current investigation build on these findings by showing that both mean PA and elevated PE inertia over 10 days of reports were protective against loneliness and were associated with greater psychological wellbeing one year later during a period of reduced access to social interaction (e.g., social distancing and lock-down orders).

Importantly, the findings relating to PE Inertia are consistent with broader evidence that it is not only the intensity of positive emotional experiences that matter, but the frequency of positive emotion that promote psychological health and wellbeing (Diener et al., 2009; Fredrickson & Joiner, 2018). For example, individuals who report a high frequency of positive moments are buffered against risk for psychopathology during times of high perceived stress (Bränström, 2013). Research targeting an increase in moments of positive emotion have shown promise for stress reduction (e.g., in medical and emergency personnel during COVID-19; Coifman et al., 2021) and findings are consistent with theories underlying behavioral treatments for disorders like depression (e.g., behavioral activation; Dimidjian et al., 2014). Importantly, there is also increasing evidence suggesting that positive emotion is linked to circadian regulation (Murray et al., 2002; 2009) and that the salutary motivational influences of positive emotion are limited to within a day, rather than across days (Aurora et al., 2022). Taken together, results from the current investigation highlight the importance of both frequent and persistent daily positive affect as a target for treatment to promote greater wellbeing and psychological health.

Study strengths include the pre-registered, multi-cohort longitudinal design using two large, nationally representative U.S. samples stratified by key demographic variables. Another strength is the use of a 10-day daily diary which allowed us to measure both static and dynamic indices of affective experience in daily life. By doing this, we were able to directly compare mean levels of NA and PA (as well as variability in affect) with dynamic indices that reflect clinically relevant conceptualizations of emotional experience as predictors of long-term psychological health, a necessary step in research focusing on dynamic indices (Dawel et al., 2023; Dejonckheere et al., 2019a, b). Furthermore, by considering affect within the context of dispositional negative affectivity, we were able to evaluate the unique effects of state-level affect indicators, including how one conceptualizes emotional experience, a common target in psychotherapy. In addition, because we considered multiple indices of emotion flexibility—an uncommon approach—we were able to examine the covariation between indices, as well as how they differ in their predictions of psychological outcomes. Finally, we investigated both positive (psychological wellbeing) and negative (psychological symptoms and loneliness) indicators of psychological health.

Limitations include single diaries per day, which may have limited our estimates of the dynamic indices of flexibility (polarity and inertia). In addition, our investigation focused on broad dimensions of psychological functioning and the unique effects of dynamic indices were relatively small when predicting these distal outcomes. Further explication of dynamic emotion indices could be oriented to more proximal outcomes (e.g., behaviors in daily life) as has been demonstrated in some prior work (Coifman et al., 2012). Another key limitation relates to our lack of baseline measurement of the longitudinal outcomes (symptoms, loneliness, and wellbeing). To attempt to compensate for this limitation, we included a robust and stable indicator of trait negative affectivity in trait anxiety (Shackman et al., 2016). By controlling for trait anxiety in our models, we likely accounted for some of the stable variance in the outcomes. This nevertheless is an important limitation and these findings warrant replication in future more conventional longitudinal design frameworks. Finally, only a portion of the samples completed the diary or follow-up portions of the study, and there was evidence that those who opted out of them were more distressed, least educated, and more racially diverse participants. This may be due in part to the unique circumstances of this investigation during key moments of the COVID-19 pandemic. Relatedly, because we do not have data from before the COVID-19 pandemic that could serve as a control group, it is unclear what impact the context had on the results. Therefore, both the pattern of attrition and the broader context of the investigation (e.g., COVID-19 pandemic) are important features that constrain the ability to generalize from these results and suggest additional replication is needed.

This investigation aimed to test how daily reports of emotion reported in two distinct periods of the COVID-19 pandemic were predictive of psychological symptoms, loneliness, and wellbeing one year later. By measuring emotion reports in daily life, we were able to test both conventional static measures of negative and positive affect and dynamic measures (polarity and inertia) reflecting specific types of cognitive conceptualizations of emotional experience as predictors of long-term psychological health one year later. Moreover, we tested the relative importance of daily affect when also considering a robust indicator of dispositional risk for emotion-related dysfunction. Results suggest that both stable and dynamic indices were associated with outcomes one year later, predicting small but meaningful variance. In particular, PE inertia was the sole dynamic index predicting unique variance in outcomes after accounting for all other factors. Results were interpreted within the context of common psychotherapeutic targets and suggest continued focus on daily affect experiences, including focus on shifting intensity levels (reducing negative and increasing positive daily emotions). However, our results also shed light on how to consider the dynamic ways in which individuals conceptualize experience, targeting less polarized and less persistent reports of negative affect, and more elevated and persistent reports of positive affect. Finally, these results highlight the importance of daily experiences of emotion during periods of stress in long-term adjustment when considering a range of outcomes, including psychological symptoms, loneliness, and well-being.