Investigating the effectiveness of instructing emotion regulation flexibility to individuals with low and high anxiety

ABSTRACT Background and Objectives Psychopathology has been associated with a deficit in emotion regulation (ER) flexibility – the ability to flexibly utilize ER strategies that are appropriate to situational demands. Yet, whether anxious individuals can be taught ER flexibility, or whether ER flexibility is effective in managing negative affect, remains unknown. We investigated the impact of instructed ER flexibility on emotional responding among individuals with differing levels of anxiety. Design and Methods Participants (N = 109) were taught two ER strategies (reappraisal, distraction) and randomized to be instructed in either flexible or inflexible ER while viewing images that differed in negative emotional intensity. Results When averaged over anxiety, or for participants with low anxiety, negative affect did not differ between conditions. However, among anxious participants, those in the flexible regulatory conditions – those who were instructed to flexibly switch between strategies – reported lower negative affect than those in the inflexible Reappraisal only condition, but not the Distraction only condition. The effectiveness of the two flexible conditions did not significantly differ. Conclusions Anxious individuals benefitted from being instructed in either ER flexibility or distraction. This finding supports literature on the adaptiveness of distraction and provides preliminary evidence linking instructed ER flexibility and improved emotional responding.

The ability to manage one's emotional responses when confronted with daily stressors is a core component of adaptive psychological functioning.Over the past two decades, emotion regulation (ER) has garnered increasing empirical attention as a potential transdiagnostic factor underpinning psychopathology (Aldao et al., 2010).ER is a dynamic process where individuals modify their emotional responses to meet situational demands (Gross, 1998).There is a growing consensus within this field that adaptive ER relates more to one's ability to flexibly utilize a variety of ER strategies in response to changing contextual demands, rather than simply the tendency to inflexibly rely on a single putatively adaptive ER strategy (Aldao et al., 2015;Bonanno & Burton, 2013).In recognition of this, the concept of ER flexibility has recently gained empirical interest.ER flexibility, or regulatory flexibility, refers to one's capacity to flexibly select and implement ER strategies that are appropriate to dynamic situational demands (Aldao et al., 2015;Bonanno & Burton, 2013).Although a relatively recent term, regulatory flexibility is conceptually aligned with a number of existing theoretical models that argue for the centrality of flexible responding to adaptive functioning, such as coping flexibility (Cheng, 2001), expressive flexibility (Bonanno et al., 2004), psychological flexibility (Kashdan & Rottenberg, 2010), and affective flexibility (Malooly et al., 2013).
One key facet of ER flexibility is regulatory selection flexibilitythe ability to select between different ER strategies in response to dynamic contextual demands (Sheppes, 2020).While there are a multitude of ways in which contextual demands can differ between situations, research has robustly demonstrated that the emotional intensity of a situation is an important contextual demand influencing regulatory behavior (see Matthews et al. (2021) for a meta-analysis, see Sheppes (2020) for a review).Specially, it has been demonstrated, over a range of emotion-eliciting contexts (e.g., negative images, words, sounds and electric shocks), that healthy individuals exhibit a strong preference for selecting distraction 1 when confronted with high intensity emotional stimuli, and conversely, a strong preference for reappraisal 2 when confronted with low intensity emotional stimuli (Matthews et al., 2021;Sheppes, 2020).Additionally, this pattern of strategy selection has also been associated with larger Late Positive Potential modulationan electro-cortical indictor of successful regulation (Ilan, et al., 2019).In further support of this pattern of regulatory selection, distraction has been found to be effective in regulating affect, particularly in high intensity contexts, while reappraisal has been found to be effective in regulating affect in low intensity contexts (Shafir et al., 2016;Sheppes, Brady et al., 2014;Sheppes & Meiran, 2007).Here, distraction is thought to be appropriate for high intensity contexts because it allows participants to disengage their attention from the stimulus early on, to prevent emotional processing and the subsequent development of strong negative emotions.Conversely, in low intensity contexts, when negative emotionality is expected to be lower, reappraisal is considered more appropriate because it allows participants to stay engaged with the situation while reinterpreting the meaning of the emotional information as a way to downregulate the subsequent emotional response (Sheppes, 2020).Accordingly, to operationalize this aspect of ER flexibility, Sheppes et al. (2011) developed an experimental paradigm; the regulatory selection paradigm.This paradigm involves participants viewing a series of negative stimuli (half of low emotional intensity, and half of high emotional intensity) and selecting whether they wish to use reappraisal or distraction to reduce their emotional distress when viewing each stimulus.Here, higher ER flexibility is operationalized as a greater tendency to select distraction in response to high intensity emotional situations, and reappraisal in response to low intensity emotional situations (Fine et al., 2021;Levy-Gigi et al., 2016).
Importantly, emerging research has found an association between psychopathology and deficits in ER flexibility.Experimental research, using the aforementioned regulatory selection paradigm, found a link between posttraumatic stress disorder (PTSD) and impaired ER flexibility (Fine et al., 2021;Levy-Gigi et al., 2016).Specifically, Fine and colleagues (2021) found that university students with high levels of PTSD as well as child sexual assault survivors with PTSD exhibited significantly lower ER flexibility than matched controls.Further, Levy-Gigi and colleagues (2016) found that ER flexibility moderated the relationship between trauma exposure and PTSD among a group of warexposed firefighters.Specifically, the well-established dose-response relationship between trauma exposure and PTSD, where greater trauma exposure predicts more severe PTSD, was only evident among firefighters with low regulatory flexibility.Conversely, for firefighters with high regulatory flexibility, there was no significant correlation between trauma exposure and PTSD, suggesting that greater ER flexibility was protective against the development of PTSD following trauma.Taken together, these findings suggest that psychopathology, particularly anxiety-related disorders, may be characterized by deficits in ER flexibility, and that such impairments in flexibility are associated with more severe symptomatology.
In light of emerging findings linking psychopathology to deficits in ER flexibility, this necessarily begs the question; are deficits in ER flexibility a mechanism driving psychopathology?And if so, can ER flexibility be taught as a way to improve emotional functioning and clinical symptoms?Insight into whether ER flexibility is amenable to psychological intervention, and effective in promoting psychological functioning, could aid in identifying a novel and viable clinical target for treatments.Despite this, while some studies have explored the positive emotional and neural consequences and correlates for participants who spontaneously demonstrate greater ER flexibility (Dorman Ilan et al., 2019;Fine et al., 2021;Levy-Gigi et al., 2016;Shafir et al., 2016), no study to our knowledge has sought to evaluate the effectiveness of instructed ER flexibility in managing emotional responding.A likely reason for this gap is the lack of a preclinical evidence-base from which to extrapolate.In particular, the absence of an experimental paradigm through which ER flexibility can be manipulated has hindered advancements in our understanding of the mechanistic role of ER flexibility in the context of psychopathology.Experimental research is a critically important preclinical line of inquiry, allowing hypothesized mechanisms to be manipulated in controlled settings to infer the causal impact of such variables on psychological outcomes (Haslam & McGarty, 2008).
While existing experimental studies investigating the association between ER flexibility and PTSD (Fine et al., 2021;Levy-Gigi et al., 2016), using the regulatory selection task (Sheppes, 2014;Sheppes et al., 2011), have established a valuable line of inquiry, three key gaps remain.First, the regulatory selection task provides a way to measure ER flexibility, rather than a way to manipulate this process.Accordingly, causal relationships between variables cannot be inferred.This could only be investigated using a between-group design where participants are instructed to use either a flexible or inflexible ER approach, and subsequent psychological distress is measured.Second, it has been noted that extant regulatory selection studies have almost exclusively relied on IAPS images (Lang et al., 1997) as the experimental emotional stressor (Matthews et al., 2021).Though a widely used and standardized image battery, the IAPS comprise relatively old-fashioned images whose content may lack realism or relevance for contemporary audiences.Though sparsely studied, there is some evidence that the nature of the emotional content of the stressor may influence regulatory selections (DeMarco et al., 2015;Matthews et al., 2021;Tahlier et al., 2013).Consequently, the use of images that depict more contemporary and personally-relevant stressful events, such as those related to the recent Coronavirus pandemic, may increase the ecological validity of the experimental task.Indeed, one study recently adapted the regulatory selection task to include low and high intensity sentences about Coronavirus-related threats and found that a predominant regulatory preference for distraction, when reading high intensity sentences, and reappraisal, when reading low intensity sentences, was replicated (Shabat et al., 2021).Third, research suggests that ER flexibility differs between clinical and non-clinical samples (Fine et al., 2021), and thus the benefit of being instructed in ER flexibility may be most evident among individuals with greater clinical symptoms.Such findings suggest that it may be important to consider the moderating role of psychological symptoms when assessing the effectiveness of instructing ER flexibility.
In light of promising developments suggesting the relevance of ER flexibility to psychological functioning, the current study sought to develop an experimental paradigm to test the effectiveness of instructed ER flexibility in reducing negative affect in response to emotional stressors, and to explore whether the effectiveness of instructed ER flexibility may differ between individuals with varying levels of anxiety.To develop the instructed ER flexibility paradigm, we drew on Sheppes, Scheibe et al.'s (2014) regulatory selection paradigm as well as studies using instructed ER manipulations (Campbell-Sills et al., 2006;Nickerson et al., 2017).In our paradigm, participants were allocated to one of four conditions and presented with images that differed in negative emotional intensity (high vs. low).Participants were instructed on which ER strategy to use while viewing each image, in line with their condition.There were two inflexible conditions, characterized by an inflexible reliance on a single strategy regardless of contextual changes: Reappraisal only and Distraction only.To create a fully crossed design, there were two flexible conditions where participants were guided in switching between distraction and reappraisal in line with changing contextual demands (i.e., changes in the emotional intensity of the images).These participants were instructed to either Distract during High intensity images, and Reappraise Low intensity images (Flexible -DHRL), which mirrors the pattern of strategy selection found to be adaptive in prior ER flexibility studies (Fine et al., 2021;Levy-Gigi et al., 2016), or adopt the inverse pattern of switching between strategies; Reappraise High intensity images and Distract during Low intensity images (Flexible -RHDL).Additionally, anxiety symptoms were indexed and a battery of negative images depicting the Coronavirus pandemic were developed for use in the experimental paradigm.Although a fully crossed design was implemented for completeness, the flexible condition of greatest interest to the aims of the present study was the Flexible -DHRL condition.We hypothesized that (1) participants in the Flexible -DHRL condition would report lower negative affect than participants in the inflexible conditions (Reappraisal only or Distraction only), and (2) participants high in anxiety would especially benefit from being in the Flexible -DHRL condition compared to the inflexible conditions.Finally, as an exploratory question, given prior research pointing to the adaptiveness of using distraction in high intensity emotional contexts and reappraisal in low intensity emotional contexts, we sought to test whether the Flexible -DHRL condition (which follows this strategy preference) would result in lower negative affect than the Flexible -RHDL condition.

Participants
One hundred and fifty adults were recruited via Amazon's Mechanical Turk (MTurk) in June 2020.MTurk samples have been shown to provide high-quality data (Kees et al., 2017;Shapiro et al., 2013), and be more representative of the general population than student samples or alternate internet-based samples (Buhrmester et al., 2011).Eligibility criteria required participants to: reside in the United States, have completed at least 100 Human Intelligence Tasks (HITs), and have a HIT approval ratio (HAR) of at least 95%.
To ensure a high level of data quality, the present study utilized a number of widely-recommended response validity indicators (Chandler et al., 2020;Chmielewski & Kucker, 2020) to identify inattentive responders, "farmers" (individuals using server farms to bypass MTurk location restrictions) and "bots" (computer programs that pose as participants).Participants' IP addresses, time taken to complete the study, and responses to a reCAPTCHA question were screened.Four participants were excluded for failing the reCAPTCHA item.Within the study, participants were required to complete one forced-choice attention check ("Have you read the above instructions?"Yes/No) and one open-ended bot check ("Enter your initials in the box below to indicate that you have read the above information"), where responders who provided a response that was inconsistent with the instructions were excluded (e.g., typing the author's MTurk ID or name, or typing the word "initials").Seven responders were excluded for this reason.Additionally, our study included sixteen openended validity checks that occurred after each of the eight training and practice items (described below) and additionally eight times throughout the experimental phase ("Please briefly describe what you thought about while viewing the image you just saw").In line with past research (Chmielewski & Kucker, 2020), the open-ended items were screened for: (1) repetitive or incorrect responses, such as the repetition of the same word across all items or answers that clearly contradicted the experimental instructions, (2) ungrammatical or nonsensical responses, or (3) suspicious responses, such as phrases that repeated portions of the question or were plagiarized from the internet.Responders were excluded if they failed at least half of the open-ended item checks.Thirty responders (20%) were excluded from the present study for this reason.Open-ended validity checks are one of the most effective ways to identify fraudulent responders, and an exclusion rate of 20% is to be expected (Chmielewski & Kucker, 2020).The final sample consisted of 109 participants.

Measures
Demographics.Participants provided their age, gender, ethnic background, and educational attainment.
Exposure to Coronavirus-related stressors.The Coronavirus Stressor Survey (CSS; McLean & Cloitre, 2020) is a six-item questionnaire used to assess exposure to Coronavirus-related stressors, including illness, hospitalization, having a job that requires possible exposure, lost income, increased domestic responsibilities or difficulty obtaining basic goods and services due to the Coronavirus.Participants were asked to rate whether each stressor had (a) happened to me personally, (b) happened to someone close to me, or (c) doesn't apply to me.A total count of the stressors that each participant had directly and/or indirectly experienced was derived for the present study.
Anxiety symptoms.The 7-item anxiety subscale of the Depression, Anxiety and Stress Scale (DASS-21; Lovibond & Lovibond, 1995) was used to assess anxiety symptom severity.Established clinical cut-offs were used to group participants into either a "low" or "high" anxiety group (Lovibond & Lovibond, 1995).Participants who scored in the Normal to Mild range were categorized as "low" and participants who scored in the Moderate to Extremely Severe range were categorized as "high" anxiety.This scale has been widely used to screen for anxiety symptoms within community samples (Henry & Crawford, 2005), and demonstrated excellent internal consistency in the present study (α = .927).
Negative Affect.Negative affect was measured using a single self-report item administered following the presentation of each image.Participants were asked to rate how negative they felt on a 9point Likert-type scale, where 1 = not negative at all, 9 = very negative (Sheppes, Scheibe et al., 2014).A mean score was calculated to index overall negative affect.
Emotional stimuli.Stimuli were 30 Coronavirus-related images.To ensure that there was an equal number of low and high negative emotional intensity images, a pilot study was conducted.A battery of 100 images depicting a wide range of economic, social and healthrelated experiences of the Coronavirus pandemic was sourced from Bing and Google Images.A community sample of 92 American adults (Mean age = 41.49[SD = 12.55; range 21-73]; 62% male; 77.2%European American) were recruited from MTurk and rated a randomized subset of the image battery (60 of the 100 images) according to valence (1 = very unpleasant; 9 = very pleasant), arousal (1 = low; 9 = high) and Coronavirus-relatedness (1 = not at all related; 9 = strongly related).Following this, 15 low intensity and 15 high intensity images were selected based on normative valence and arousal ratings (Valence Low = 3.61; Arousal Low = 4.34; Coronavirusrelatedness Low = 7.42; Valence High = 2.43; Arousal High = 5.80; Coronavirus-relatedness High = 7.56).As predicted, paired samples t tests indicated that the two groups of images significantly differed in valence, t(91) = 14.50, p < .001,and arousal, t(91) = −13.20,p < .001,but not in Coronavirus-relatedness, t(91) = −1.53,p = .130.These mean arousal and valence ratings were consistent with IAPS and borderline personality disorder-specific image batteries that have been used in prior regulatory selection research (Sauer et al., 2016;Sheppes et al., 2011).The selected images depicted Coronavirus-related medical scenes, deaths and burials, economic impacts (e.g., food shortages, unemployment and business closures), social impacts (e.g., isolation and physical separation), and behaviors (e.g., people flouting social distancing and personal protective equipment recommendations).Appendix A in the Supplemental Materials outlines the mean valence, arousal and Coronavirus-relatedness ratings for each image.
Manipulation check.Open-ended manipulation checks were used to ensure that participants were implementing the ER strategy that they were instructed to use.Participants were asked to, "Please briefly describe what you thought about while viewing the image you just saw", eight times throughout the experimental phase.We screened for responses that clearly contradicted the experimental instructions (e.g., reappraising despite being instructed to distract), or demonstrated a gross misapplication of the ER strategy (e.g., doing something that did not resemble either ER strategy).Exclusion criteria was failing 50% or more of these manipulation checks, however no participant met this threshold.This is consistent with past research using the regulatory selection paradigm, which found that participants closely adhered to experimental instructions (Levy-Gigi et al., 2016;Sheppes et al., 2011).

Procedure
The current study was conducted online using Qualtrics.Following informed consent, participants answered demographic questions and completed the CSS and DASS-21 anxiety subscale.Similar to the procedure of Sheppes et al.'s (2011) regulatory selection task, our study comprised a teaching phase, a practice phase, and an experimental phase.In the teaching phase, participants received instructions on distraction and reappraisal via written text and instructional videos (see Appendix B of Supplemental Materials).Each instructional video was approximately 2 min and 30 s in length and included a brief explanation of the strategy as well as a demonstration of how to use the strategy when viewing an image that was representative of those shown in the experimental phase.Following each instructional video, participants practiced implementing the target strategy while viewing one low intensity and one high intensity image.The order in which participants learnt the two ER strategies was counterbalanced.
In the practice phase, the experimental procedure was explained and participants were introduced to the negative affect rating scale.At no stage in the study were participants advised that images varied in negative emotional intensity.Participants completed four practice trials, which mirrored the trial structure of the experimental phase (see Figure 1).For each trial, participants first previewed an image for 1 s, where they were instructed to attend to the image.Next, large text instructed participants on which strategy to use when the image returned (participants were advised to click the "next" button when they had read the instructions and were ready to proceed to the image).The image then re-appeared for 6 s while participants attended to the image and implemented their instructed strategy.Immediately following this, participants rated their negative affect.Every participant completed two distraction practice trials and two reappraisal practice trials, each comprising a low and high intensity image.
Participants were randomized to one of four conditions: (1) Reappraisal only, (2) Distraction only, (3) Flexible -DHRL, (4) Flexible -RHDL.The experimental phase comprised 30 image trials (15 low intensity trials and 15 high intensity trials).Images were arranged into 10 "blocks', where each block comprised 3 images of the same emotional intensity.The images were arranged as such to limit the amount of times that participants in the ER flexible conditions were required to switch between the strategies, thereby reducing the possible additional cognitive load of being in a flexible condition.The order of the images, and blocks, were randomly generated, and then fixed across conditions (see Appendix C of Supplemental Materials for the image order).Participants in the inflexible conditions (Reappraisal only and Distraction only) were instructed to use the same target strategy for every image, regardless of image intensity.Participants in the Flexible -DHRL condition were instructed to distract during high intensity images and reappraise low intensity images, while participants in the Flexible -RHDL condition were instructed to the do the inverse.Immediately following each training and practice trial, and eight times throughout the experimental phase, participants were presented with the manipulation check item.At the completion of the experiment, participants were debriefed and paid US$5.00.The University of New South Wales Human Research Ethics Committee provided ethics approval.

Data analysis
Statistical analyzes were conducted using SPSS version 27.To identify potential between-group (anxiety group and instructed ER condition) differences in demographic characteristics, chi square tests and one-way ANOVAs were calculated.Next, a 4 (instructed ER condition) x 2 (anxiety group) x (2) (emotional intensity) ANCOVA, with planned comparisons, was conducted to test the study's hypotheses.Instructed ER condition (Reappraisal only, Distraction only, Flexible -DHRL, or Flexible -RHDL) and anxiety group (low or high) were entered as between-subject factors, and emotional intensity (low and high) was entered as a within-subject factor.Demographic variables that significantly differed between anxiety groups or instructed ER conditions were entered into the model as covariates.A post-hoc sensitivity power analysis for a between-subjects ANCOVA (with planned comparisons) conducted using GPower 3.1.9.7 indicated that our sample size was sufficient to detect a medium effect (f = 0.27) with power (1-β) of .80 and α of .05.

Sample characteristics
Sociodemographic characteristics are provided in Table 1.Mean anxiety symptoms across each of the four experimental conditions are presented in supplemental Table D.
Overall ANCOVA model.To test our hypotheses, planned comparisons were computed.However, for completeness, the results of the overall ANCOVA model are presented in Appendix E of the online Supplemental Materials.Of note, there was a significant main effect of anxiety (p < .001),where participants with high anxiety reported more negative affect (M = 6.01) than participants with low anxiety (M = 4.01).Additionally, there was a significant main effect of emotional intensity, such that negative affect was higher for high intensity images (M = 5.61) than low intensity images (M = 5.61), indicating that the experimental manipulation of image intensity was successful.

Planned contrasts
Hypotheses 1: Testing whether the instructed ER flexible condition (Flexible -DHRL) resulted in lower negative affect than the inflexible conditions.Hypothesis 1 was not supported (p = .Hypotheses 2: Testing whether the instructed ER flexible condition (Flexible -DHRL) resulted in a greater reduction in negative affect than the inflexible conditions for participants with high anxiety.Hypothesis 2 was partially supported (see Figure 2).For participants with high anxiety, negative affect differed depending on instructed ER condition (trending significance, p = .066;η p 2 = .070).Among highly anxious participants, those in the Flexible -DHRL condition reported significantly lower negative affect than those in the inflexible Reappraisal only condition (p = .049;M(Reappraisal only) = 6.65;M(Flexible -DHRL) = 5.69).However, the Flexible -DHRL and Distraction only conditions did not significantly differ in negative affect (p = .358;M(Distraction only) = 6.14).For participants with low anxiety, negative affect did not significantly differ between ER flexible vs inflexible conditions (p = .355;η p 2 = .032;M(Flexible -DHRL) = 4.44; M(Reappraisal only) = 3.88; M(Distraction only) = 3.59; p's from comparisons were .394and .157respectively).
Exploratory hypothesis: Testing whether the Flexible -DHRL condition resulted in lower negative affect than the Flexible -RHDL condition.Hypothesis 3 was not supported.Negative affect did not significantly differ between the two ER flexible conditions, when averaged over anxiety and intensity (p = .897;M(Flexible -DHRL) = 5.07; M(Flexible -RHDL) = 5.02).Further, negative affect between the two flexible conditions did not significantly differ for participants in either the low or high anxiety groups (p's = .959and .787respectively).Moreover, the two flexible conditions yielded very similar patterns of results when each were compared to the inflexible conditions; among high anxiety participants, the Flexible -RHDL condition (just like the Flexible -DHRL condition) reported lower negative affect than the Reappraisal only condition (p = .025),but was not different to the Distraction only condition (p = .232).

Discussion
The current study developed a novel instructed ER flexibility paradigm to investigate the effectiveness of ER flexibility in reducing negative affect following exposure to emotional stressors.To our knowledge, this is the first study to experimentally manipulate ER flexibility to test its impact on emotional responses, as well as the first study to directly compare the effectiveness of instructed ER flexibility between individuals with differing levels of anxiety.Our first hypothesis was not supported; when averaged over anxiety levels, participants in the Flexible -DHRL condition, which instructed participants to distract during high intensity images and reappraise low intensity images, did not report significantly lower negative affect than participants in the inflexible conditions (Reappraisal only and Distraction only).However, our second hypothesis was partially supported; among participants with high anxiety, those in the Flexible -DHRL condition reported significantly lower negative affect than those in the Reappraisal only condition, while levels of negative affect did not significantly differ between the Flexible -DHRL and Distraction only conditions.Conversely, participants with low anxiety did not differ in negative affect between the Flexible -DHRL and inflexible conditions.The benefit of a flexible approach, compared to an inflexible reappraisal only approach, accords with prior experimental investigations, which found that participants who adopted a flexible ER approach had better overall psychological functioning (Fine et al., 2021) and that spontaneously switching between strategies to select reappraisal for low intensity images, and distraction for high intensity images, conferred better neuro-affective outcomes (Dorman Ilan et al., 2019;Shafir et al., 2016).However, in light of this research, it is unexpected that this effect was not observed globally in the present study, but rather, only among participants with high anxiety.
There are several possible explanations for our findings.First, in light of prior research finding anxiety disorders to be characterized by deficits in emotion regulation flexibility (Conroy et al., 2020;Fine et al., 2021), it stands to reason that such individuals had the most to gain from being guided in employing greater flexibility when confronted with emotional stressors.Considering that the habitual ER behaviors of anxious individuals are typically characterized by an inflexible reliance on strategies like suppression and rumination (Aldao & Nolen-Hoeksema, 2010;Campbell-Sills et al., 2014), being instructed to flexibly use reappraisal and distraction in response to stressors may have thus conferred a benefit for anxious participants precisely because it facilitated a substantial deviation from their habitual ER repertoire.Second, it may be relevant to consider that we also found a significant main effect of anxiety, such that individuals with high anxiety reported significantly greater negative affect than individuals with low anxiety, regardless of condition allocation.This finding is consistent with research suggesting that people with anxiety exhibit greater reactivity to emotional stressors (Campbell-Sills et al., 2014).Similarly, individuals with high anxiety also reported being exposed to more than double the number of Coronavirus-related stressors than individuals with low anxiety.Considering that the emotional stimuli used in our experimental paradigm were Coronavirus-related, it is possible that these images thus evoked greater negative affect due to their greater relevance for participants in the high anxiety group.Accordingly, it is possible that the higher levels of negative affect among our anxious participants created greater opportunities for subsequent down-regulation, provided they were given guidance on how to effectively do so.It is noteworthy however, that very little experimental research has investigated whether the effectiveness of instructed ER strategies differ depending on an individual's level of psychopathology or the relevance of the emotional stressor.As such, future research may benefit from comparing clinical and non-clinical samples, indexing baseline psychopathology, or manipulating stressor relevance to further investigate the moderating effect of these factors on the effectiveness of ER flexibility.
Findings from our study also suggested a lack of a clear distinction between the effectiveness of an ER flexible approach (Flexible -DHRL) and the use of distraction only, regardless of emotional intensity.As participants in both the Flexible -DHRL and Distraction only conditions were instructed to use distraction when confronted with high intensity images, our finding might indicate that the use of distraction in high intensity contexts is especially helpful.Indeed, this explanation aligns with prior research suggesting that the use of distraction is more effective than reappraisal in managing negative affect, particularly in situations of high emotional intensity (Ilan, et al., 2019;Shafir et al., 2015;Sheppes, Brady et al., 2014;Sheppes & Meiran, 2007).
The present study also explored whether the two flexible conditions (Flexible -DHRL and Flexible RHDL) differed in effectiveness.Our exploratory hypothesis was not supported as negative affect did not significantly differ between the two ER flexible conditions.In fact, the Flexible -DHRL condition (i.e., the emotion regulation approach that was derived from prior research attesting to the greater adaptiveness of distracting during high-intensity images and reappraising low intensity images) actually yielded a similar pattern of results to the Flexible -RHDL condition (i.e., the condition where participants followed an inverse regulatory approach, using reappraisal for high-intensity images and distraction for low intensity images).Specifically, levels of negative affect between the two flexible conditions did not significantly differ for participants in either the low or high anxiety groups.Additionally, just like the pattern of results seen for the Flexible -DHRL condition, highly anxious participants in the Flexible -RHDL condition also reported lower negative affect than those in the inflexible Reappraisal only condition but their level of negative affect did not differ from those in the Distraction only condition.There are a number of potential explanations for this novel, and unexpected, finding.First, we based this hypothesis on prior research using the regulatory selection paradigm, which found that participants displayed a preference for distraction when confronted with high intensity images and reappraisal when confronted with low intensity images and that those who spontaneously adopted this pattern of ER choices had better subsequent affective functioning (Dorman Ilan et al., 2019;Shafir et al., 2016;Sheppes, Scheibe et al., 2014).This discrepancy may be the result of methodological differences between the current study and past research using Sheppes and colleagues' (2014) paradigm.Our instructed ER flexibility paradigm was developed to measure negative affect following the implementation of an instructed ER strategy, while the ER choice paradigm measured spontaneous ER choices.Each design had distinct objectives; the ER choice paradigm offered an ecologically valid assessment of habitual ER behavior, while our ER flexibility paradigm was intended to serve as a pre-clinical test of the efficacy of instructing ER flexibility to reduce psychological distress.The lack of support for our exploratory hypothesis may suggest that spontaneous and instructed ER behavior are distinct concepts.
Second, prior studies exploring the consequences of specific regulatory choices did not consider the potential moderating effect of pre-existing levels of anxiety.Thus, while these studies found that switching to reappraisal for low intensity images, and switching to distraction for high intensity images, was associated with larger LPP modulation (a neural indicator of successful regulation; Dorman Ilan et al., 2019), it is still possible that there may be some conditions in which alternate forms of flexibility (such as the inverse pattern of strategy selection) may be adaptive.Alternatively, the lack of difference in effectiveness between our two flexible conditions may indicate that, for anxious individuals, simply being flexible in one's ER responding, or being guided in accessing two regulatory strategies rather than reappraisal only, is more central to adaptive functioning than the specific way one matches ER strategies to each context.It is noteworthy however, that very little experimental research has investigated whether the effectiveness of instructed ER strategies differ depending on an individual's level of psychopathology or habitual patterns of ER, and the present study was the first to implement an instructed ER flexibility paradigm.Accordingly, further research is warranted to replicate our findings and disentangle some of the tentative explanations for our present findings.For instance, future research that includes both spontaneous and instructed ER paradigms would allow us to explore how pre-existing ER habits influence how effectively individuals can implement instructed ER flexibility.Furthermore, future research may benefit from comparing clinical and non-clinical samples or indexing baseline psychopathology to further investigate the moderating effect of psychopathology on the effectiveness of ER flexibility.
Several limitations merit acknowledgment.First, our paradigm, though experimentally rigorous, represents a narrow operationalization of ER flexibility.The current study operationalized ER flexibility as the use of a wider ER strategy repertoire (reappraisal and distraction) that was responsive to contextual changes (changes in image intensity).Accordingly, our findings may not be generalizable to all real-world instances of ER.Instead, our study represents the first attempt at operationalizing instructed ER flexibility, upon which future investigations can build and refine.Future replication of our paradigm using diverse operationalizations of ER flexibility is important to verify the robustness of our findings and to better approximate real-world ER behavior.Second, the inclusion of an image preview in the experimental trial procedure merits further consideration.While an image preview is a core component of the original regulatory selection paradigm (Sheppes, 2020) as a means for participants to preview an image prior to making their ER strategy choice, the necessity for this step in an instructed ER flexibility paradigm (where ER strategy use is instructed, rather than self-selected) is less clear.It is also possible that participants may have preemptively began using an ER strategy during this preview stage.Accordingly, to reduce this possibility, it would be fruitful for future replications of our instructed paradigm to consider omitting image previews from the trial procedure.Third, while our sample size (n = 109) was similar to past research examining ER flexibility using the ER selection paradigm (e.g., Levy-Gigi et al., 2016;Sauer et al., 2016), it is possible that there was insufficient power to detect small effects.Relatedly, the data analytic approach employed in the present study did not control for multiple comparisons, thus increasing the Type 1 error rate.Future research would benefit from using larger sample sizes, and more stringent statistical controls, to address these potential limitations.Third, the sensitivity of our measures of negative affect and anxiety symptoms could be improved.Negative affect was measured using a single self-report item in line with prior experimental studies (Sheppes, Scheibe et al., 2014).It would be valuable for future research to additionally include more objective measures of emotional reactivity, such as psychophysiological indexes like heart rate variability, as well as consider more specific emotional states, such as ratings of sadness, fear, anger and disgust.Anxiety symptoms were assessed using the DASS which, although a widely used self-report screener of clinical symptoms (Henry & Crawford, 2005), does not provide specific diagnostic information.The use of disorder-specific scales or clinician-administered diagnostic tools would allow for greater accuracy and specificity when assessing clinical symptoms.Additionally, as the experimental images depicted stressful Coronavirus-related events, it may have been valuable to additionally control for individual differences in Coronavirus-related anxiety.Although we controlled for group differences in Coronavirus stressor exposure, it would be fruitful for future studies to explore the influence of Coronavirusrelated anxiety on instructed ER flexibility.Finally, while several validity checks were employed to maximize data quality, the use of an online platform (MTurk) for data collection is a potential limitation of the current study.Replication in a strictly controlled lab-based setting is warranted.
The current study developed a novel experimental paradigm to test the effectiveness of instructed ER flexibility in managing negative affect in response to dynamic emotional stressors.Our findings offer preliminary evidence for a link between instructed ER flexibility and reduced emotional reactivity among anxious individuals.For participants with high levels of anxiety symptoms, being guided in flexibly using reappraisal and distraction was more effective in managing negative affect than just using either reappraisal or distraction by itself.Although tentative given the preliminary nature of our findings, ER flexibility may be an important clinical target in psychological interventions for individuals suffering from anxiety.For example, skills training in ER flexibility may be a viable mechanism of change to incorporate into existing well-supported interventions for anxiety, such as exposure-based therapy and Cognitive Behavior Therapy (CBT).Indeed, the principles of ER flexibility are already reflected in a number of existing ER-focused psychological treatments, such as Dialectical Behavior Therapy (DBT; Linehan, 1993) and Skills Training in Affect and Interpersonal Regulation (STAIR; Cloitre et al., 2002), which focus on enhancing an individual's capacity for emotional awareness and flexible application of learnt ER strategies in response to stressors.Future research would benefit from exploring whether existing well-supported anxiety treatments enhance ER flexibility, and if not, future research investigating the feasibility and accessibility of ER flexibility-based strategies in managing anxiety in real-life settings would be warranted as a first step in developing an ER flexibility-based intervention.As our findings offer initial evidence of a relationship between instructed ER flexibility and improved emotional responding for anxious individuals, further experimental research with clinical populations is warranted to verify the impact of ER flexibility on psychological functioning.

Notes
1. Distraction is an ER strategy that involves disengaging from an emotional stimulus by diverting one's attention away from the emotional stimulus, in order to reduce the intensity of the emotional response (van Dillen & Koole, 2007).2. Reappraisal is an ER strategy that involves engaging with an emotional stimulus, and changing the way one thinks about the emotional stimulus, in order to reduce the intensity of the emotional response (Gross & John, 2003).

Figure 1 .
Figure 1.Illustration of experimental trial structure in the instructed ER flexibility paradigm.ms = milliseconds.

Figure 2 .
Figure 2. Differences in negative affect ratings between instructed ER conditions for participants in the low and high anxiety groups.Note: Error bars represent ± 1 standard error (SE).For ease of interpretation, only the error bars for the flexible and the Reappraisal only condition are presented.a: p = .049,b: p = .025.