Introduction
Post-traumatic stress disorder (PTSD) is a debilitating and tenacious condition, that involves a core impairment in the control or regulation of negative emotions (Karlsson & Sjöberg, Reference Karlsson and Sjöberg2009). Difficulties in emotion regulation are considered central to PTSD, because they predict the development and maintenance of the disorder, and are associated with more severe PTSD symptomatology (Boden et al., Reference Boden, Westermann, McRae, Kuo, Alvarez, Kulkarni and Bonn-Miller2013; Forbes et al., Reference Forbes, Tull, Rapport, Xie, Kaminski and Wang2020; Pencea et al., Reference Pencea, Munoz, Maples-Keller, Fiorillo, Schultebraucks, Galatzer-Levy and Powers2020).
The view on what constitutes emotion dysregulation in PTSD has shifted throughout the years. These shifts can be understood within a central engagement-disengagement regulatory classification that categorizes regulatory strategies as involving Engagement with emotional information processing and meaning-making v. Disengagement from emotional information processing and meaning avoidance (e.g. Parkinson & Totterdell, Reference Parkinson and Totterdell1999; Roth & Cohen, Reference Roth and Cohen1986; Thayer & Lane, Reference Thayer and Lane2000).
Traditional views suggested that PTSD individuals over-utilize disengagement regulatory strategies such as avoidance at the expense of engagement regulatory strategies that promote emotional processing (Aldao, Nolen-Hoeksema, & Schweizer, Reference Aldao, Nolen-Hoeksema and Schweizer2010; Bonanno & Burton, Reference Bonanno and Burton2013; Foa, Hembree, & Rothbaum, Reference Foa, Hembree and Rothbaum2007; Foa & Kozak, Reference Foa and Kozak1986; John & Gross, Reference John and Gross2004). However, later studies showed that in certain contexts, disengagement from stressful and traumatic events is associated with adaptive outcomes whereas engagement with emotional information processing is maladaptive (e.g. Bonanno, Keltner, Holen, & Horowitz, Reference Bonanno, Keltner, Holen and Horowitz1995; Chapman, Rosenthal, Dixon-Gordon, Turner, & Kuppens, Reference Chapman, Rosenthal, Dixon-Gordon, Turner and Kuppens2017; Coifman, Bonanno, Ray, & Gross, Reference Coifman, Bonanno, Ray and Gross2007; See Park, Reference Park2010 for a review).
These mixed findings led to a new conceptual understanding that adaptive regulation of negative affective events may require the use of disengagement regulatory strategies in certain contexts and engagement regulatory strategies in other contexts (e.g. Aldao, Sheppes, & Gross, Reference Aldao, Sheppes and Gross2015; Bonanno & Burton, Reference Bonanno and Burton2013; Bonanno, Papa, Lalande, Westphal, & Coifman, Reference Bonanno, Papa, Lalande, Westphal and Coifman2004 for review). This updated view suggests that individuals with PTSD may show reduced regulatory selection flexibility that manifests in reduced ability to choose engagement v. disengagement regulatory strategies in a manner that is sensitive to differing situational demands.
Despite the increasing conceptual agreement, empirical evidence for regulatory selection flexibility impairments in PTSD remains indirect. Specifically, one line of studies demonstrated that when PTSD individuals are instructed to execute engagement and disengagement strategies they show impaired execution flexibility (Bartholomew, Badura-Brack, Leak, Hearley & McDermott, Reference Bartholomew, Badura-Brack, Leak, Hearley and Mcdermott2017; Rodin et al., Reference Rodin, Bonanno, Rahman, Kouri, Bryant, Marmar and Brown2017). While important, these studies did not examine whether PTSD individuals fail to voluntarily select these strategies flexibly according to differing demands. A second line of studies showing that PTSD individuals self-report general impairments in the frequency of using regulatory strategies, did not assess active selection between strategies in different contexts (see Seligowski, Lee, Bardeen, & Orcutt, Reference Seligowski, Lee, Bardeen and Orcutt2015 for a meta-analysis).
To fill these gaps the present two-study investigation was set to demonstrate deficits in regulatory selection flexibility in two different populations with PTSD symptomology. In doing so we concentrated on perhaps the most fundamental regulatory selection phenomenon, concerning the ability of individuals to flexibly select between regulatory disengagement and engagement strategies in a manner that is sensitive to differing emotional intensity levels.
Multiple studies have repeatedly demonstrated that healthy individuals show regulatory selection flexibility manifested in an increased preference to select distraction over reappraisal as intensity increases from low to high intensity (Sheppes, Reference Sheppes and Gawronski2020 for a review). Specifically, in low-intensity contexts, healthy individuals strongly select to engage with emotional information and reinterpret its negative meaning via reappraisal, which is both effective in modulating mild emotional reactions, and more beneficial than a distraction for long-term adaptation (e.g. Thiruchselvam, Blechert, Sheppes, Rydstrom, & Gross, Reference Thiruchselvam, Blechert, Sheppes, Rydstrom and Gross2011). However, in high-intensity contexts, healthy individuals strongly select to disengage their attention via distraction, which effectively blocks potent emotional information and provides short term benefits (e.g. Shafir, Schwartz, Blechert, & Sheppes, Reference Shafir, Schwartz, Blechert and Sheppes2015).
Relevant yet scarce support for the importance of flexible strategy selection in the context of trauma comes from a single study that found that exclusively among non-PTSD firefighters with impaired regulatory selection flexibility, higher traumatic exposure was associated with higher PTSD symptoms (Levy-Gigi et al., Reference Levy-Gigi, Bonanno, Shapiro, Richter-Levin, Kéri and Sheppes2016).
In the current two-study investigation, Study 1 examined whether relative to college students with low post-traumatic symptoms, students with high post-traumatic symptoms that meet the clinical cutoff for PTSD would show reduced regulatory selection flexibility. Study 2 sought to critically extend the reduced regulatory selection flexibility findings of Study 1 in a clinical population of women with PTSD due to childhood sexual abuse (CSA). Emotion dysregulation is considered a strong predictor of CSA-PTSD (Ullman, Peter-Hagene, & Relyea, Reference Ullman, Peter-Hagene and Relyea2014), and it accounts for severe functional, and interpersonal impairments as well as to higher risk for ensuing psychopathology (Browne & Finkelhor, Reference Browne and Finkelhor1986; Cloitre, Miranda, Stovall-McClough, & Han, Reference Cloitre, Miranda, Stovall-McClough and Han2005; Kim & Cicchetti, Reference Kim and Cicchetti2010; Zlotnick et al., Reference Zlotnick, Zakriski, Shea, Costello, Begin, Pearlstein and Simpson1996). However, prior empirical evidence (e.g. Coffey, Leitenberg, Henning, Turner, & Bennett, Reference Coffey, Leitenberg, Henning, Turner and Bennett1996; Ehring & Quack, Reference Ehring and Quack2010; Griffing et al., Reference Griffing, Lewis, Chu, Sage, Jospitre, Madry and Primm2006; Poole, Dobson, & Pusch, Reference Poole, Dobson and Pusch2017) is restricted to impaired ability to execute different strategies, thus lacking crucial evidence regarding regulatory selection flexibility impairments for CSA-PTSD individuals.
To test our hypotheses, we validated a modified version of a classic performance-based regulatory selection paradigm (Sheppes, Reference Sheppes and Gawronski2020 for review). In the modified paradigm participants are exposed to high and low negative intensity word stimuli and they behaviorally select whether they want to regulate their emotions using disengagement distraction or engagement reappraisal. Our use of high and low-intensity emotional words instead of pictorial stimuli previously used in the classic regulatory selection paradigm (c.f., Sheppes, Scheibe, Suri, & Gross, Reference Sheppes, Scheibe, Suri and Gross2011), bypasses the requirement to use highly explicit and concrete traumatic content (e.g. mutilation pictures) in this vulnerable population (Kindt & Brosschot, Reference Kindt and Brosschot1997; Öhman & Soares, Reference Öhman and Soares1994; Wikström, Lundh, Westerlund, & Högman, Reference Wikström, Lundh, Westerlund and Högman2004).
Hypotheses in both studies were identical. Compared to non-clinical individuals, individuals that meet the clinical cutoff for PTSD (Study 1) and women with PTSD due to CSA (Study 2) will show reduced regulatory strategy selection flexibility, manifested in a diminished increase in distraction (over reappraisal) preference as word stimuli intensity increases from low to high.
Methods
Below we report how we determined our sample size, all data exclusions, all manipulations, and all measures that were collected in both studies. Study 1 was approved by the Institutional Review Board (IRB) and Study 2 was approved by the Medical Center Ethics (Helsinki) Committee.
Study 1
Participants
As part of a standard departmental procedure, the first-year undergraduate student cohort (n = 317) signed informed consent and completed a battery of self-report measures at the beginning of the academic year, including the Post Traumatic Checklist (PCL-5, without criterion A) that assesses post-traumatic stress symptoms and constitutes our main group factor. In addition, students completed the Patient Health Questionnaire (PHQ-9) and the State-Trait-Anxiety Inventory (STAI) that assess depression and anxiety symptoms, respectively, in order to obtain common comorbidity measures. Students, unaware of the reason they were contacted, were invited to take part in the present study if their post-traumatic stress symptoms levels met a pre-defined clinical cutoff for PTSD (High PTS group: PCL-5 > 33, c.f., Rubin, Boals, & Berntsen, Reference Rubin, Boals and Berntsen2008; Weathers et al., Reference Weathers, Litz, Keane, Palmieri, Marx and Schnurr2013) or if they had minimal post-traumatic stress symptoms (Low PTS group: PCL-5 < 5). This relatively known ‘extreme group’ categorization design was chosen to maximize symptomatology differences (c.f., Azriel, Lazarov, Segal, & Bar-Haim, Reference Azriel, Lazarov, Segal and Bar-Haim2020; Shelby, Golden-Kreutz, & Andersen, Reference Shelby, Golden-Kreutz and Andersen2008; Vail, Goncy, & Edmondson, Reference Vail, Goncy and Edmondson2019). From the large cohort, we identified 22 individuals that met the PCL clinical cut-off (high PTS group). To match the size of the high trauma group we chose 22 individuals with the lowest PCL scores (low PTS group). All participants had normal or corrected to normal vision, and were native Hebrew speakers, because understanding and implementing complex cognitive emotion regulation strategies require high verbal proficiency (c.f., Sheppes, Reference Sheppes and Gross2014). For participation, students received academic credit or monetary compensation (~45 USD).
Procedure
Approximately 1–2 months following the mass testing, participants signed a written informed consent, completed the PCL-5 (including criterion A), followed by performing the modified performance-based regulatory selection paradigm. One week later participants completed the regulatory selection paradigm again in order to examine its test re-test reliability.Footnote †Footnote 1
Clinical instruments
Post-Traumatic Checklist (PCL-5) – A 20 item self-administered inventory that indexes PTSD symptoms in the past month and is strongly recommended for the assessment of PTSD in undergraduate populations with mixed civilian trauma exposure (Adkins, Weathers, McDevitt-Murphy, & Daniels, Reference Adkins, Weathers, McDevitt-Murphy and Daniels2008). Responses are rated on a scale of 0–4 and are summed to a total score. Cronbach's α in the current sample was α = 0.77.
Patient Health Questionnaire (PHQ-9) – A 9-item self-administered inventory indexing each of the DSM-IV depression criteria on a scale ranging from 0 to 3 (Kroenke, Spitzer, & Williams, Reference Kroenke, Spitzer and Williams2001). Cronbach's α in the current sample was α = 0.85.
State-trait Anxiety Inventory (STAI) – A 20 item self-administered inventory of trait anxiety on a 4-point scale (Spielberger, Reference Spielberger1983). Cronbach's α in the current sample was α = 0.78.
Modified performance-based regulatory selection word paradigm
Stimuli: 40 negative emotional words in Hebrew were selected from an Effective Norms for Hebrew Words databaseFootnote 2 (Armony-Sivan, Cojocaru, & Babkoff, Reference Armony-Sivan, Cojocaru and Babkoff2014). Low negative intensity words (n = 20, M arousal = 4.8; s.d. = 0.61, M valence = 2.5, s.d. = 0.6) differed significantly from high negative intensity words (n = 20, M arousal = 7; s.d. = 0.48, M valence = 1.9, s.d. = 0.47) in arousal, t(38) = −12.2, p < 0.001, and valence, t(38) = 3.67, p < 0.001. Low and high-intensity words were matched in word length, t(38) = 0.44, p > 0.66, and prevalence, t(38) = 0.58, p > 0.56. Emotional words included diverse negative content (e.g. ‘poverty’, ‘boredom’/ ‘death, ‘rape’, low/high intensity respectively) and were matched across the two intensity categories when possible. Previous studies with similar arousal and valence differences between low and high-intensity stimuli have demonstrated differential levels of emotional-response activation (e.g. Bradley, Codispoti, Cuthbert, & Lang, Reference Bradley, Codispoti, Cuthbert and Lang2001; Shafir et al., Reference Shafir, Schwartz, Blechert and Sheppes2015) and differential regulatory preferences (e.g. Sheppes et al., Reference Sheppes, Scheibe, Suri and Gross2011; Sheppes, Brady, & Samson, Reference Sheppes, Brady and Samson2014a, Sheppes et al. Reference Sheppes, Scheibe, Suri, Radu, Blechert and Gross2014b). Importantly, to provide further validation for our stimulus intensity categorization, at the onset of the study participants were presented with all words and rated their level of negative experience on a Likert scale (1 = not negative at all, 9 = extremely negative). As expected high-intensity words (M = 5.84 s.d. = 1.47) were rated as more negative than low-intensity words (M = 4.44 s.d. = 1.16), F(42) = 161.5, p < 0.001.
Experimental paradigm: Participants first learned how to implement disengagement-distraction and engagement-reappraisal (three examples for each instruction) and then practiced (six trials) choosing between them with an instruction to base their decision on the strategy which they assume would be more effective in reducing their negative emotional experience in response to each stimulus (c.f., Sheppes et al., Reference Sheppes, Scheibe, Suri and Gross2011, Reference Sheppes, Brady and Samson2014a, Reference Sheppes, Scheibe, Suri, Radu, Blechert and Gross2014b). Distraction instructions involved disengaging attention by producing unrelated neutral thoughts (i.e. visualizing daily activities or geometric shapes) (e.g. Shafir, Thiruchselvam, Suri, Gross, & Sheppes, Reference Shafir, Thiruchselvam, Suri, Gross and Sheppes2016, Reference Shafir, Guarino, Lee and Sheppes2017). Reappraisal instructions involved engaging with the processing of the emotional stimuli, but reinterpreting their negative meaning (i.e. thinking about less negative aspects of the situation or that the situation will improve over time) (Gross, Reference Gross and Gross2014, Reference Gross2002). In addition, participants were not allowed to form reality challenge reappraisals (i.e. interpret emotional events as unreal), since these reappraisals function as a form of disengagement (see Qi et al., Reference Qi, Li, Tang, Zeng, Diao, Li and Hu2017; Sheppes et al., Reference Sheppes, Brady and Samson2014a, Reference Sheppes, Scheibe, Suri, Radu, Blechert and Gross2014b). Adherence to regulatory instructions during these phases was examined by asking participants to verbalize strategies out loud, during which corrective feedback was provided as needed.
The actual task consisted of 40 trials (divided into 2 equally long blocks, separated by a short break), during which words of low and high emotional intensity were presented in a random order, with the restriction that no more than two trials of the same emotional intensity category repeat in sequence. Each trial (see Fig. 1) began with a 500 ms fixation cross, followed by a 1000 ms preview of the emotional word. Then participants viewed a choice screen where they consciously indicated their regulatory selection between two fixed options- distraction or reappraisal by pressing a keyboard button that corresponded to each strategy (assignment of a button to the strategy was counterbalanced across trials), similarly to classic decision-making paradigms (e.g. Marewski & Schooler, Reference Marewski and Schooler2011, for review). Following a reminder, cue preparing the participants to perform their chosen strategy (500 ms), the same word stimulus was presented again for 5000 ms, during which participants implemented their chosen strategy. The offset of each word was followed by a 1-to-9 Likert scale in which participants reported their level of negative emotional experience in response to the word (1 = ‘not negative at all’, 9 = ‘extremely negative’).Footnote 3
Illustration of a trial structure in the Modified Regulatory selection paradigm in which the participant saw a high emotional intensity word and selected disengagement distraction (ms = milliseconds).
Data analysis
To examine our main predictionFootnote 4 regarding reduced regulatory selection flexibility among high relative to low PTS groups we employed a 2 × 2 mixed analysis of variance (ANOVA) with Group (High and Low PTS) as a between-subject factor, and Intensity (High and Low) as a within-participant factor, with the percentage of trials for which distraction was chosen (over reappraisal) as the dependent variable (Sheppes, Reference Sheppes and Gawronski2020 for review). The expected two-way interaction was decomposed in a follow-up analysis that examined whether the high relative to low PTS group showed reduced regulatory selection flexibility manifested in a smaller increase in distraction (over reappraisal) preference from low to high intensity. For all analyses, we provide model fit estimates that include partial eta square and F-value.
Results
Demographics and reliability checks
Demographic and psychopathological characteristics by the group are presented in Table 1. Before addressing our main research question we wished to establish the internal and test-retest reliability indices of the modified regulatory selection word paradigm. First, meeting the standard acceptable value of the Kuder-Richardson 20 index (KR-20 = 0.5) in tasks with 20 or less binary items (Field, Reference Field2009; Dall'Oglio et al., Reference Dall'Oglio, Rossiello, Coletti, Caselli, Ravà, Di Ciommo and Pasqualetti2010; Hinton, McMurray, & Brownlow, Reference Hinton, McMurray and Brownlow2014), our low intensity [KR-20 = 0.71, 95% confidence interval (CI) 0.57–0.82] and high intensity (KR-20 = 0.53, 95% CI 0.31–0.71) showed adequate internal reliability. Second, participants' test-retest reliability indices showed a significant correlation across the weekly measurements for low intensity, r(42) = 0.38, p = 0.01, and high intensity, r(42) = 0.35, p < 0.001.
Demographic and psychopathological characteristics by group
PTS, Post Traumatic Symptoms; PCL-5, Post Traumatic Checklist; PHQ-9, Patient Health Questionnaire; STAI-T, State-Trait-Anxiety Inventory.
**p ⩽ 0.01, ***p ⩽ 0.001.
Reduced regulatory selection flexibility in high relative to low PTS group
Replicating and extending prior regulatory selection findings obtained with images to word stimuli (see Sheppes, Reference Sheppes and Gawronski2020 for a review), we found a significant main effect of intensity, indicating that participants' preference for distraction over reappraisal increased as the emotional intensity increased from low (M = 36.47%) to high (M = 56.7%) intensity, F(1,42) = 56.15, p < 0.0001, $\eta _{\rm p}^2$
= 0.57.
Importantly, consistent with our main hypothesis, we found a significant two-way interaction between Group and Intensity, F(1,40) = 6.81, p = 0.01, $\eta _{\rm p}^2$
= 0.14, 95% CI 0.1–0.18 (See Fig. 2). Planned follow up analyses confirmed that the low PTS group demonstrated a robust Regulatory Selection Flexibility pattern, manifested in a 27% increase in distraction choice from low intensity (M = 31.59%, s.d. = 16.06%) to high intensity (M = 58.86%, s.d. = 15.88%), F(1,42) = −9.99, p < 0.000001. By contrast, the magnitude of the regulatory selection flexibility pattern in the high PTS group was less than half, and manifested in only a 13% increase in distraction choice from low intensity (M = 41.36%, s.d. = 2.03%) to high intensity (M = 54.54%, s.d. = 14.38%) F(1,42) = −7.79, p = 0.005.Footnote 5
Performance-Based Emotion Regulatory Selection Flexibility in Low and High PTS groups. Percentage signifies Regulatory Selection Flexibility. Error bars represent 95% CIs. ** p ⩽ 0.01, ***p ⩽ 0.001.
Study 2
Participants
Given the supporting findings in Study 1 and given the similar design in both studies, for Study 2 we were able to determine the sample size with a formal a-priori power analysis. Using G*Power (Campbell & Thompson, Reference Campbell and Thompson2012), applying the conventional power of 0.8, alpha of 0.05 and the observed effect size of the interaction from Study 1 ($\eta _{\rm p}^2$
= 0.14), the analysis pointed to a required sample size of 27 participants in each group in order to detect a reliable effect.
Study 2 was conducted as part of a larger study,Footnote 6 that included 31 female participants (M age = 34.20, s.d. = 7.49; range = 22–50), with a history of recurrent CSA and an ascertained diagnosis of PTSD on the Clinician-Administered PTSD Scale (CAPS-5), but with no history of neurological disorder, psychosis, or current substance dependence. To match the clinical sample size, 31 non-clinical matched female controls (M age = 31.51; s.d. = 6.43; range = 19–45) were recruited via electronic flyers, resulting in 62 participants for the final sample. Non-clinical participants completed the Life Events Checklist (LEC, self-report screening measure for trauma exposure) to ensure they did not meet criterion A and to specifically verify they were not exposed to sexual trauma (M LEC # trauma = 0.16, s.d. = 0.11) and the Post Traumatic Checklist (PCL-5) in order to confirm they did not meet the clinical cutoff for PTSD (PCL-5 < 33; see Weathers et al., Reference Weathers, Litz, Keane, Palmieri, Marx and Schnurr2013). In addition, all participants completed several self-report measures including the Beck Depression Inventory (BDI-II) and State-trait Anxiety Inventory (STAI), that assess depression and anxiety symptoms, respectively, in order to obtain a general estimate of comorbidity. All participants had normal or corrected to normal vision and were native Hebrew speakers (c.f., Sheppes, Reference Sheppes and Gross2014; Sheppes et al., Reference Sheppes, Scheibe, Suri and Gross2011).
Procedure
Female PTSD-CSA participants that were in ongoing treatment in an out-patient clinic provided written informed consent according to the Medical Center Ethics Committee guidelines followed by participating in a CAPS assessment by a certified psychologist. Within a week after ascertaining DSM-5 PTSD diagnosis and completing self-report measures, in a separate session, the modified regulatory selection paradigm was administered. Similarly, non-clinical control participants provided written informed consent, completed self-report measures, and then completed the modified regulatory selection paradigm.
Clinical instruments
Clinician-Administered PTSD Scale (CAPS): We used the gold standard structured clinician interview for assessing PTSD diagnosis and symptom severity. We administered a version of the CAPS that combines DSM-IV and DSM-5 criteria in order to maintain continuity between classifications (Friedman, Kilpatrick, Schnurr, & Weathers, Reference Friedman, Kilpatrick, Schnurr and Weathers2016; Hoge, Riviere, Wilk, Herrell, & Weathers, Reference Hoge, Riviere, Wilk, Herrell and Weathers2014, Reference Hoge, Yehuda, Castro, McFarlane, Vermetten, Jetly and Rothbaum2016). The CAPS contains explicit, behaviorally anchored probes for each of the 17 PTSD symptom criteria of the DSM-IV (on severity and frequency scale of 0–4), and 20 symptoms of the DSM-5 (on a severity scale of 0–4). Cronbach's α for the current sample was α = 0.73, α = 0.68 for CAPS-4 and-5 respectively.
Post Traumatic Checklist (PCL-5) and State-trait Anxiety Inventory (STAI) – See details in Study 1. Cronbach's α for the current sample was α = 0.937 and α = 0.81, respectively.
Beck Depression Inventory (BDI-II)–A 21 item self-administered inventory of depression symptoms and their respective intensity on a 4-point scale (Beck, Steer, & Carbin, Reference Beck, Steer and Carbin1988). Cronbach's α for the current sample was α = 0.86.
The Life Events Checklist (LEC) - The LEC is the self-report trauma assessment portion of the CAPS (Blake et al., Reference Blake, Weathers, Nagy, Kaloupek, Gusman, Charney and Keane1995; Weathers, Keane, & Davidson, Reference Weathers, Keane and Davidson2001) that assesses exposure to 16 traumatic events known to potentially result in PTSD. Items that were personally endorsed (‘the event happened to me’) receive a score of 1 and are summed to a total score.
Modified performance-based regulatory selection word paradigm
Stimuli: Word stimuli were identical to those used in Study 1. In order to provide further validation for stimuli categorization into low and high intensity, at the onset of the experimental procedure (identical to Study 1), participants were presented with all words and rated their level of negative experience on a Likert scale (1 = not negative at all, 9 = extremely negative). As expected, and replicating Study 1 findings, high-intensity words (M = 6.66, s.d. = 1.07) were rated as more negative than low-intensity words (M = 4.63, s.d. = 1.31), F(60) = 221.24, p < 0.0001.
Experimental paradigm, and procedure: Experimental paradigm and task procedure in Study 2 were identical to Study 1 except for the following changes. In the present study, in order to further verify regulatory choice adherence, 15% of the trials were randomly followed by a screen instructing participants to write a sentence describing how they implemented the strategy they chose (c.f., Sheppes et al., Reference Sheppes, Scheibe, Suri and Gross2011). A judge who was blind to participants' choices (i.e. participants' button presses) coded the sentences as distraction or reappraisal. As expected and congruent with prior findings, levels of agreement approached a perfect score (96.9% accuracy), indicative of adequate adherence (e.g. Levy-Gigi et al., Reference Levy-Gigi, Bonanno, Shapiro, Richter-Levin, Kéri and Sheppes2016; Sheppes et al., Reference Sheppes, Scheibe, Suri and Gross2011).
Data analysis
Data analysis in study 2 was identical to Study 1.
Results
Demographics and reliability checks
Demographic and psychopathological characteristics by the group are presented in Table 2. Before addressing our main research question, as in Study 1, we calculated the internal reliability index of the modified regulatory selection word paradigm. Meeting the standard acceptable value of the Kuder-Richardson 20 index (KR-20 = 0.5) in tasks with 20 or less binary items (Field, Reference Field2009; Dall'Oglio et al., Reference Dall'Oglio, Rossiello, Coletti, Caselli, Ravà, Di Ciommo and Pasqualetti2010; Hinton et al., Reference Hinton, McMurray and Brownlow2014), our low-intensity KR-20 = 0.68 (95% CI 0.56–0.79) and high-intensity KR-20 = 0.75 (95% CI 0.65–0.83) showed adequate internal reliability.
Demographic and psychopathological characteristics by group
PTSD, post-traumatic stress disorder group; CAPS, Clinician-Administered PTSD Scale; PCL-5, Post Traumatic Checklist; BDI-II, Beck Depression Inventory; STAI, State-Trait-Anxiety Inventory; LEC, # trauma types, Life Event Checklist.
**p ⩽ 0.01, *** p ⩽ 0.001.
Reduced regulatory selection flexibility in PTSD relative to non-clinical group
Replicating prior and Study 1 findings, we found a significant main effect of intensity, indicating that participants' preference for distraction over reappraisal increased as the emotional intensity increased from low (M = 28.75%) to high (M = 51.56%) intensity, F(1,60) = 93.53, p < 0.001, $\eta _{\rm p}^2$
= 0.609.
Importantly, consistent with our main hypothesis, Study 2 extended Study 1 findings to a clinically diagnosed PTSD sample, and demonstrated converging evidence in showing a significant interaction between Group and Intensity, F(1,60) = 10.07, p = 0.002, $\eta _{\rm p}^2$
= 0.114, 95% CI −0.21 to 0.51 (See Fig. 3). Planned follow up analyses confirmed that the non-clinical group demonstrated a robust Regulatory Selection Flexibility pattern, manifested in a 30% increase in distraction choice from low intensity (M = 23.32%, s.d. = 14.13%) to high intensity (M = 53.61%, s.d. = 2.01%), F(1,60) = −9.12, p = 0.0001. By contrast, the magnitude of the regulatory selection flexibility pattern in the PTSD group was half in size, manifested by only a 15% increase in distraction choice from low intensity (M = 34.19%, s.d. = 2.18%) to high intensity (M = 49.51%, s.d. = 2.05%), F(1,60) = −4.52, p = 0.0002.
Performance-Based Emotion Regulatory Selection Flexibility in non-Clinical and PTSD groups. Percentage signifies Regulatory Selection Flexibility. Error bars represent 95% CIs. ** p ⩽ 0.01, *** p ⩽ 0.001.
General discussion
Despite a growing conceptual agreement that adaptive regulation involves flexibly matching emotion regulatory strategies to situational demands, empirical evidence of reduced Regulatory Selection Flexibility in PTSD is lacking. The present study demonstrated for the first-time reduced performance-based Regulatory Selection Flexibility in two different populations with PTSD symptoms. This impairment was manifested in reduced ability to flexibly choose engagement v. disengagement regulatory strategies in a manner that is sensitive to differing affective intensity demands. Specifically, Study 1 modified a performance-based regulatory selection paradigm using low and high-intensity affective word stimuli, and showed adequate internal reliability and significant test-retest reliability. Importantly, Study 1 confirmed hypotheses in showing that relative to college students with low PTS symptoms, students with high PTS symptoms presented reduced regulatory flexibility that was manifested in a smaller increase in distraction (over reappraisal) preference from low to high intensity. Extending Study 1 findings, Study 2 investigated a CSA-PTSD population that its hallmark deficit is emotional dysregulation. Mirroring findings from Study 1, Study 2 showed that relative to non-clinical women, women with a diagnosis of CSA-PTSD showed reduced regulatory flexibility that was demonstrated in a smaller increase in distraction (over reappraisal) preference from low to high intensity.
Taken together, findings from both studies provide important empirical support for the conceptual notion that PTSD individuals lack adaptive emotion regulation that requires the use of disengagement regulatory strategies in high-intensity contexts and engagement regulatory strategies in low-intensity contexts (e.g. Aldao et al., Reference Aldao, Sheppes and Gross2015; Bonanno & Burton, Reference Bonanno and Burton2013; Bonanno et al., Reference Bonanno, Papa, Lalande, Westphal and Coifman2004 for review). Specifically, regulatory selection flexibility entails that in low-intensity contexts, individuals would predominantly select to engage with emotional information and reinterpret its negative meaning via reappraisal (e.g. Thiruchselvam et al., Reference Thiruchselvam, Blechert, Sheppes, Rydstrom and Gross2011), whereas in high-intensity contexts, individuals would predominantly select to disengage attention via distraction (Sheppes, Reference Sheppes and Gawronski2020 for review).
Diverting from this healthy pattern, reduced regulatory selection flexibility in PTSD involves failing to maximize the benefits of selecting disengagement distraction to manage high-intensity events and or engagement reappraisal to cope with low-intensity events. Specifically, overly selecting disengagement regulation in low-intensity contexts precludes the long-term benefits of engaging with and making meaning of affective situations, and overly selecting engagement regulation in high-intensity contexts precludes the short-term benefits of warding off overwhelming negative emotions via disengagement strategies.
What might explain the reduced regulatory selection flexibility in PTSD? One possible explanation suggests that PTSD individuals lack available cognitive resources, which manifest in the decreased choice of strategies that are effortful to implement such as engagement reappraisal (c.f., Milyavsky et al., Reference Milyavsky, Webber, Fernandez, Kruglanski, Goldenberg, Suri and Gross2019; Sheppes et al., Reference Sheppes, Brady and Samson2014a, Reference Sheppes, Scheibe, Suri, Radu, Blechert and Gross2014b). However, a general lack of resources can only partially explain the present findings. Specifically, the reduced regulatory selection flexibility in PTSD was indeed manifested in selecting less effortful reappraisal in low intensity, but PTSD individuals selected more reappraisal in high intensity. Therefore, it is possible that PTSD individuals may lack the cognitive resources required to alternate their regulatory selections between distraction and reappraisal to differing intensities of affective events.
The replicability and robustness of our findings, together with a firm theoretical background suggests that flexible regulatory selection may constitute an important underlying mechanism in PTSD. Accordingly, improving regulatory selection flexibility should be added to canonical clinical interventions that involve general efforts to improve regulatory selection (e.g. Berking, Ebert, Cuijpers, & Hofmann, Reference Berking, Ebert, Cuijpers and Hofmann2013; Linehan, Reference Linehan2015).
Despite the replicable findings and novelty of the present investigation, it is important to mention several limitations and future directions. First, although groups categorization was based on well-defined parameters of PTS symptoms (Study 1) and PTSD diagnosis (Study 2), we cannot fully determine whether regulatory selection flexibility is a specific mechanism for PTSD or related to more general psychopathology. While additional analyses (see online Supplemental Materials) that covaried depressive and anxiety symptoms provided preliminary support for reduced regulatory selection flexibility that is specific to PTSD, future studies should further examine potential moderators that are associated with PTSD, such as comorbidities, gender, trauma type and trauma history.
Second, a possible limitation relates to the cross-sectional design of our study which does not allow to test whether selection flexibility impairment is an antecedent or consequence of PTSD symptoms (see Kring, Reference Kring, Lewis, Haviland-Jones and Barrett2008, for a review). Specifically, regulatory selection flexibility can be antecedent to PTSD and hence may predict PTSD symptoms. Alternatively, it can be a consequence such that individuals with PTSD symptoms may become less emotionally flexible.
Third, the current study has some psychometric limitations. In self-report measures, we found relatively low Cronbach's alphas for PCL in Study 1 and CAPS-5 in Study 2. Nevertheless, obtaining similar regulatory selection findings in Study 2 where internal consistency of the PCL was high (α = 0.937) strengthens our confidence in the results of Study 1. In the performance-based Regulatory Selection paradigm, we found relatively low (albeit significant) test-retest correlations. Accordingly, conclusions regarding reduced regulatory selection flexibility in PTSD should be limited to the group level rather than the individual level (c.f., Berger, Reference Berger2006).
Fourth, while the present findings demonstrate reduced regulatory selection flexibility in PTSD, the affective consequences of this selection deficit cannot be accurately evaluated using our paradigm (see Footnote 3). Accordingly, future studies should consider combining the regulatory selection paradigm with a regulatory implementation paradigm (where participants are instructed which strategy to employ on each trial) that accurately assesses affective consequences. Of possible affective consequence measures, future studies should consider electrophysiological measures of regulatory success (e.g. late positive potentials) that have been proven to adequately reveal the consequences of different strategies across varying intensities (e.g. Shafir et al., Reference Shafir, Schwartz, Blechert and Sheppes2015).
Lastly, although the present study investigated the two most established regulatory engagement and disengagement strategies and the central emotional intensity situational factor (Sheppes, Reference Sheppes and Gawronski2020), future studies may consider testing other strategies along the engagement disengagement continuum as well and other cognitive and motivational factors to further establish regulatory selection flexibility impairments in PTSD.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291721004670.
Acknowledgements
The authors would like to thank the research team in Tel-Aviv Sourasky Medical Center, and especially Efrat Routledge and Marina Gordon for their major contribution in carrying out this research, including subjects' recruitment, behavioral and clinical assessments.
Author contributions
All authors sufficiently contributed and take responsibility for the content of the paper.
Financial support
This work was supported by the Israel Science Foundation (GS, Grant No. 1130/16) (MB, Grant No. 2107/17); the U.S. Department of Defense (TH, Grant No. W81XWH-16-C-019); National Institute of Psychobiology for Israel Young Investigator (LH); and the Brain and Behavior Foundation, NARSAD (LH, Grant No. 26302).
Conflicts of interest
None.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
