Introduction
Sequential transformation is a form of emotional processing in which opposing affective states are activated to alleviate lingering states of distressing emotions (Greenberg, 2017). This intervention strategy does not follow traditional cognitive models, where emotions such as anger and sadness are understood exclusively as symptoms. However, emotion-focused theory suggests such emotions could also serve as processes and be used to leverage change. This notion has now been empirically supported across a range of treatment approaches (Pascual-Leone, 2018) and recent developments are exploring it in cognitive and behavioral approaches (e.g., Nardone et al., 2024). Opposing affective states refer to emotions with action tendencies that conflict and cannot be completed simultaneously (Shen & Bigsby, 2010; Welling, 2012). For example, sadness and anger are considered opposing due to their conflicting action tendencies to either withdraw (i.e., sadness; Veenstra et al., 2017) or approach (i.e., anger; Burton et al., 2013). Accordingly, sadness and anger have been the focus of many studies examining the use of opposing emotion in alleviating distress. Research demonstrates that feelings of anger may reduce in intensity through the activation of sadness, whereas feelings of sadness may decrease in intensity as anger is activated (e.g., Rizvi et al., 2011, Zhan et al., 2015).
Activating Sadness to Attenuate Anger
The use of sadness to attenuate anger has primarily been examined using experimental mood inductions. In an experiment involving undergraduate students, anger was induced by providing participants with fictious negative feedback on a writing task (Zhan et al., 2015). Participants were then randomly assigned to view a movie clip that induced sadness, fear, or a neutral mood. Those guided to feel sadness reported a greater decrease in their initial intensity of anger than participants in the fear or neutral mood induction (Zhan et al., 2015). In a similar experiment, Zhan et al. (2017b) examined whether acute stress impacts the effect of sadness in reducing anger. Participants were randomly assigned either to a stress condition or non-stress condition. Prior to mood induction protocols, only participants in the stress condition engaged in a cold pressor test to induce physiological stress (Ulrich-Lai & Herman, 2009). Across both conditions, anger was then induced using fictitious negative feedback, and measured by physiological arousal via skin conductance, additional increases in cortisol, and self-report of changes in the intensity of experienced anger. Thereafter, participants were asked to view a movie clip that either induced sadness or required participants to engage in a cognitive reappraisal task. Independent of whether acute stress was elicited, participants who experienced sadness after anger reported significantly lower cortisol levels and self-report measures of anger intensity than those who had completed a cognitive reappraisal or a neutral mood induction task (Zhan et al., 2017b).
Additional research has examined whether the temporal sequence of anger and sadness impacts changes in anger intensity. The most promising of these seems to be experiencing anger-before-sadness as a procedure for attenuating anger. Lutz and Krahé (2018) found that after an anger induction, a sadness induction task was associated with lower levels of aggressive behavior than a control task, regardless of whether sadness was induced before or after anger. Rochman and Diamond (2008) examined physiological arousal associated with anger, among individuals experiencing lingering anger towards an attachment figure. Drawing from emotion-focused therapy, the target emotions were induced by asking participants to verbalize feelings of anger or sadness as they narrated specific interactions with the attachment figure. In contrast to the findings of Lutz and Krahé (2018), physiological arousal increased when participants experienced anger first and sadness second, but not when these emotions were experienced in the reverse sequence. In keeping with theories of emotion-focused therapy, greater emotional arousal was believed to indicate greater resolution of lingering anger towards the attachment figure (Greenberg & Foerster, 1996; Greenberg & Malcom, 2002).
Activating Anger to Counteract Sadness
Another body of literature suggests that activating anger can counteract feelings of sadness. Zhan et al. (2017a) compared the effects of anger and joy, which were elicited by movie clips, in reducing sadness that participants were experiencing in relation to recent personal event. The results showed a greater reduction in the intensity of sadness among participants guided to subsequently experience anger, relative to those in the joy condition (Zhan et al., 2017a). Rizvi et al. (2011) also examined the role of anger in counteracting sadness in a sample of participants with comorbid borderline personality disorder and substance abuse disorder. Participants received opposite action training, a dialectical behavior therapy skill that involves first identifying one’s current emotion and its associated action tendency, and subsequently engaging in an opposite action tendency (Rizvi & Linehan, 2005). Participants reported an immediate reduction in the intensity of the presenting sadness and a subsequent reduction in depressive symptoms 14 days after starting training (Rizvi et al., 2011). Although participants were not explicitly guided to express anger, results suggest that opposite actions may reduce sadness intensity.
Similarly, research on the Affect Phobia model of short-term dynamic therapy (McCullough-Vaillant, 1997) suggests that activating opposing emotion states can be used to resolve persistent sadness (Schanche et al., 2011). The Affect Phobia model (McCullough-Vaillant, 1997) assumes that activating affects, which are emotion states that have approach-oriented action tendencies (e.g., assertive anger, self-compassion), counteract inhibitory affects, which are emotion states with withdrawal-oriented action tendencies (e.g., sadness, shame, pain; Malan, 2001; Menninger, 1958; Schanche et al., 2011). Using this model, Schnache et al. (2011) examined the emotions that preceded recovery from self-criticism during therapy for Cluster C personality disorders. Regardless of whether participants were assigned to short-term dynamic therapy or cognitive therapy, those who reported an increase in self-compassion over the course of treatment were more likely to experience a decrease in inhibitory affect and an increase in activating affect during treatment (Schanche et al., 2011). Although anger was not exclusively examined, the findings support a broader notion that approach-oriented emotion states may be effective in defusing the intensity of sadness.
At least one study has identified that the temporal sequence of sadness-before-anger as a procedure for attenuating the problem of intense sadness. Within the context of emotion-focused therapy, Choi et al. (2016) examined sequences of expressed emotion among clients who were successfully treated for self-critical depression (i.e., sadness). Among clients who experienced a substantial increase in self-esteem during treatment, the most frequent naturally occurring pattern of emotion in psychotherapy was first sadness accompanied by the articulation of unmet existential needs, followed by an expression of anger (Choi et al., 2016).
Limitations of Current Literature
Existing studies examining the temporal sequence of anger and sadness have generally examined either anger or sadness intensity. Additional research is needed to determine whether such sequences are uniquely, or equally, effective in attenuating both anger and sadness intensity. Most studies have also relied on experimental mood induction protocols with limited external validity. Further studies are needed to examine the impact of opposing emotions in the context of everyday lived experiences.
Present Study
Through a pre-post experimental design involving multiple groups, the present study examined whether resolution of lingering anger and sadness about an interpersonal interaction depends on the sequence in which anger and sadness are experienced. Two groups were identified: participants with predominantly lingering anger related to an interpersonal interaction and participants with predominantly lingering sadness related to an interpersonal interaction. Participants completed a written intervention in one of two randomly assigned conditions (i.e., anger-before-sadness condition or sadness-before-anger condition), which differed only by the order in which participants were guided to feel anger and sadness. The written interventions employed are similar to those used in previous research informed by emotion-focused theory (e.g., Kramer & Pascual-Leone, 2016; Rohde et al., 2015). Such interventions have been associated with becoming more engaged in working on one’s problem (e.g., “problem activation”; Rohde et al., 2015) and activation of target emotions including anger (Kramer & Pascual-Leone, 2016) and sadness (Rohde et al., 2015). By examining changes in both anger and sadness intensity, in the context of emotions stemming from lived experience, this study aims to address the limitations of past research.
Hypotheses
Participants will experience a greater reduction of their presenting emotional concern when they express the presenting emotion first and an opposing emotion second (as contrasted with the inverse sequence).
(1)
Among participants who present primarily with lingering anger, those guided to experience anger first and a sadness second will report a greater reduction in anger intensity than those guided to feel these emotions in the reverse order (i.e., sadness first, then anger).
(2)
Among participants who present primarily with lingering sadness, those guided to feel sadness first and anger second will report a greater reduction in sadness intensity than those guided to feel the reverse sequence (i.e., anger first and then sadness).
Method
Participants and Eligibility
All potential participants completed a pre-screen questionnaire that asked whether they (a) were feeling either especially angry or especially sad because of (b) an interaction with an attachment figure (e.g., parent, current or past romantic partner, sibling, close friend), which (c) had occurred more than 6 months ago. Participants were eligible for the study if they reported feeling more sadness than anger or more anger than sadness. Those who reported feeling equal levels of anger and sadness were not eligible to participate in the study.
Although data was collected from a total of 366 participants, 199 of these cases were excluded due to the following: (a) not completing all tasks of the intervention (n = 76), (b) reporting a mix of both anger and sadness in response to the interpersonal difficulty (n = 84), (c) not completing the intervention in a single sitting– as indicated by a completion time greater than 2 h (n = 35), (d) missing data on a key variable (i.e., Visual Analogue Scale; n = 1), and not following instructions for the intervention (e.g., repeating the same response or writing about unrelated topics; n = 3). The final sample size consisted of N = 167 participants recruited from four recruitment methods. A total of n = 144 participants were recruited from a university Psychology participant pool. The remaining participants were recruited through advertisements sent through email to the University of Windsor student body (n = 14), Amazon Turk (n = 6), and social media Facebook account created for the study (n = 3). Participants were compensated through either course credit, a modest renumeration, or a chance to win a gift card. Within the total sample (N = 167), over 83% of participants identified as women (n = 139), 16% were men, and the remaining participant was gender non-binary. Age ranged from 17 to 66 (M = 22.57, SD = 7 years). Most participants (61%) identified as white/Caucasian, 8% identified as East Asian, 8% as Arab/Middle Eastern, 5% described their ethnicity as black/African American/African Canadian, 4% were Latinx/Hispanic, 4% (n = 7) identified as South Asian, and 10% identified as multiracial or another race/ethnicity. With respect to sexual orientation, 78% of the sample identified as heterosexual, 3% identified as homosexual, 7% identified as bisexual, 12% identified as other sexualities or preferred not to answer. Most participants (73%) were single, while 25% were partnered/married/common-law or 2% separated/divorced. The demographics within each experimenter-identified group (i.e., angry and sad) are presented in Table 1.
Table 1
Demographics for angry and sad group
Demographic characteristic | Experimenter-identified angry group (n = 70) | Experimenter-identified sad group (n = 97) | |
|---|---|---|---|
Age (M, SD) | 22.17 years (5.51) | 22.86 years (7.93) | |
Gender (%, n) | Men | 22.86% (16) | 10.34% (11) |
Women | 75.71% (53) | 87.63% (86) | |
Non-binary | 1.43% (1) | 0% (0) | |
Sexual Orientation (%, n) | Heterosexual | 75.71% (53) | 79.38% (77) |
Homosexual/Lesbian/Gay | 3.59% (6) | 6.19% (6) | |
Bisexual | 4.29% (3) | 9.28% (9) | |
Other Sexual Orientations | 7.14% (5) | 5.15% (5) | |
Did not report sexual orientation | 4.29% (3) | % (0) | |
Race (%, n) | Caucasian | 62.86% (44) | 60.82% (59) |
Black/African American/African Canadian | 8.57% (6) | 2.61% (2) | |
Arab/Middle Eastern | 10% (7) | 6.19% (6) | |
Latin/Hispanic | 1.43% (1) | 5.15% (5) | |
East Asian | 4.3% (3) | 12.37% (12) | |
South Asian | 5.7% (4) | 3.09% (3) | |
Multiracial | 1.4% (1) | 6.19% (6) | |
Other | 5.71% (4) | 4.12% (4) | |
Relationship Status (%, n) | Single | 74.3% (52) | 72.2% (70) |
Partnered | 18.6% (13) | 16.5% (16) | |
Common Law | 0% (0) | 2.1% (2) | |
Separated | 1.4% (1) | 0% (0) | |
Married | 4.3% (3) | 7.2% (7) | |
Divorced | 1.4% (1) | 2.1% (2) | |
Measures
Demographics Measure
A demographics questionnaire assessed participant gender, sexual orientation, age, year of study, employment status, marital status, and race/ethnicity.
Interpersonal Event Questionnaire (Pascual-Leone & Sawashima 2018)
The Interpersonal Event Questionnaire is an eight-item measure that examines the quality of the distressing interpersonal events that participants selected for the study. In the present study, it was used to assess the amount of time since the event, the amount of distress caused by the event, and any psychotherapy or psychiatric medications used in response to event.
Anger-Sadness Comparison item (developed for use in the present study)
This is a single item self-report measure of the relative intensity of participants’ state anger and sadness. Through the prompt, “When I think about this interaction, I feel…”, participants are asked to compare their current feelings of anger and sadness on a 9-point scale, ranging from “Only angry, not at all sad” to “Only sad, not at all angry.” Typically, measures of emotional arousal assess the intensity of each emotion state individually. In the present study, when evaluating the intensity of each emotion state, it was important for participants to directly compare the intensity of their anger to the intensity of their sadness and indicate the relative strength of these emotions. For this reason, this item was developed for use in the present study.
Visual Analogue Scale (VAS)
The Visual Analogue Scale is a single-item self-report measure that is used to assess the intensity of a specific emotion state. In the present study, a written version of the Visual Analogue Scale was used to assess the intensity of state anger and sadness at three points during the experiment. A verbally-administered 10-point version of the Visual Analogue Scale has convergent validity in the assessment of emotional arousal of anger and sadness (Rochman & Diamond, 2008).
Study Design
Participant Groups
In the current study, two experimenter-identified groups were created, consisting of participants who presented as predominantly angry (i.e., angry group) or predominantly sad (i.e., sad group). The experimenter-identified groups were intended to ensure that participants’ group reflected their emotional state after the mood induction, rather than emotional state at the time of the pre-screen questionnaire, which typically took place several weeks prior. The angry group (n = 70) includes participants who reported feeling more angry than sad on both measures of emotional state (i.e., Anger-Sadness Comparison item and Visual Analogue Scale). The sad group (n = 97) includes participants who reported feeling more sad than angry on both measures of emotional state (i.e., Anger-Sadness Comparison item and Visual Analogue Scale).
Experimental Conditions
Participants in each group were randomly assigned to one of two mood induction conditions (i.e., anger-before-sadness condition or sadness-before-anger condition). In different sequences, each condition included a written anger intervention, which was intended to activate the target emotion of anger, and a written sadness intervention, which was intended to activate the target emotion of sadness. In the anger-before-sadness condition, participants completed the anger intervention first and the sadness intervention second. In the sadness-before-anger condition, participants completed the sadness intervention first and the anger intervention second.
Procedure
The following protocol was approved by the institution Ethics Board at University of Windsor (Windsor, Canada) Approval Code 18–178. A schematic diagram of the design is presented in Fig. 1. Participants were first asked to complete demographic and baseline measures. Then, participants completed two 15-minute written mood induction task (for a total of 30 minutes) in which they were asked to describe the interpersonal interaction that they selected for the present study. As a manipulation checks, participants were also asked to complete self-report measures (i.e., Anger-Sadness Comparison item and the Visual Analogue Scale).
Fig. 1
Experimental protocol: two naturally occurring groups, each randomized to two conditions
×
![]()
Participants completed the experimental intervention, during which they completed written interventions to facilitate anger and sadness, with the sequence differing by condition (anger-before-sadness; sadness-before-anger). Within each written emotion intervention, participants were asked to complete a sequence of five brief writing tasks intended to activate the target emotion (i.e., either anger or sadness), and then guide participants through the process of feeling the emotion in question. The selected tasks were adapted from written tasks informed by the Sequential Model of Emotional Processing (Pascual-Leone, 2010; Pascual-Leone & Greenberg, 2007). Some of these tasks have also been used in previous experiments to induce target emotions.
Specifically, participants were given a series of incomplete sentence “stems” related to their anger or sadness about the interaction and asked to use them in writing about what seemed most significant, meaningful, or true for what they felt. The five tasks pertained to clarifying: (1) emotional meaning (e.g., anger, “What I resent is…”; sadness, “What I miss is…”); (2) somatic sensations (e.g., anger, clenched jaw; sadness, heaviness); (3) action tendencies (e.g., “what does the anger/sadness make you want to do?”; (4) identifying unmet need; and finally (5) writing a letter to the person expressing these emotions. The intervention for the first emotion was followed by the Anger-Sadness Comparison item and Visual Analogue Scale as a manipulation check. These steps were then repeated for the second emotion intervention. At the conclusion, participants were asked to speculate about the aim of the study, to evaluate the degree to which they were blind to study hypotheses.
Results
Manipulation Checks To Confirm Protocol Validity
Manipulation Checks for Baseline Mood Inductions and Group Membership
A set of paired-sampled and independent sample t-test analyses were conducted to confirm that the initial mood induction (see step 5 in Fig. 1) appropriately elicited predominate feelings of sadness or anger and that this corresponded to their membership in experimenter-identified groups. After the mood induction, participants in the angry group felt significantly more angry (M = 61.27; SD = 26.01) than sad (M = 29.40; SD = 24.15; t(69) = 10.70, p <.001). As expected, their anger was significantly greater than anger reported by those in the sad group (M = 25.79; SD = 21.55; t(165) = 9.62, p <.001). Similarly, participants in the sad group felt significantly more sad (M = 71.25; SD = 23.14) than angry (M = 25.79; SD = 21.55) after the mood induction (t(96) = -17.41, p <.001). As expected, their sadness was significantly greater than sadness reported by participants in the angry group (M = 29.40; SD = 24.15; t(165) = -11.32, p <.001).
Manipulation Checks of Emotion Interventions
A set of paired-sample t-tests were conducted to confirm the activations of anger and sadness after each of the corresponding writing interventions, irrespective of condition and identified group (see steps 8 and 11 in Fig. 1). For all participants, anger intensity significantly increased from before (M = 39.90; SD = 29.95) to after (M = 47.05; SD = 31.30) the anger intervention (t(166) = -3.75, p <. 001). Similarly, sadness intensity significantly increased from before (M = 52.53; SD = 31.55) to after (M = 57.13; SD = 31.89) the sadness intervention (t(166) = 2.59, p =.01).
Another set of paired samples t-tests confirmed these emotional changes were not attributable to general arousal but that each intervention discriminated by emotion. Participants reported feeling significantly angrier after the anger intervention (M = 47.05; SD = 31.30) than after the sadness intervention (M = 35.78; SD = 30.51; t(166) = 6.07; p <.001). After the sadness intervention, participants felt significantly more sad (M = 57.13; SD = 31.89) than angry (M = 35.78; SD = 30.51; t(166) = -6.05; p <.001).
Taken together, manipulation checks corroborate the experimenter-identified groupings of participants who were predominantly troubled by anger vs. those predominantly troubled by sadness and indicate that participants experienced anger and sadness while engaging in each of the interventions in this study. Accordingly, the results of the present study are presumed to be a valid reflection of engaging in the prescribed emotional sequences that were being tested.
Ruling Out the Individual Effects of Anger or Sadness
Data from manipulation checks confirmed the impact of individual mood inductions and the impact of interventions, but they also provided a step-by-step temporal account of emotional arousal. This timeline allowed us to also examine the intensity of emotion at each ordered step using within group comparisons, thereby clarifying the effect of single mood inductions. As such we might rule out the potential impact of expressing either anger or only sadness in isolation. If the effect of individual mood inductions could be ruled out, overall changes to emotional intensity would be attributable to the sequential process of ordered emotions (i.e., expressing anger-before-sadness, or sadness-before-anger). In turn, that would justify testing hypotheses about the sequences of two emotions effectively reducing the target concern (whether that be anger or sadness).
Single Mood Induction Effects on Angry Group
A paired-sample t-test was conducted using the anger-before-sadness condition to parcel out the individual effects of the first mood induction from the synergistic effect created by adding the second mood induction. For those in the anger-before-sadness condition, angry mood induction alone did not decrease anger intensity t(35) = 1.78, p =.08, (baseline, M = 61.69, SD = 25.66; post anger induction; M = 56.14, SD = 30.76). Similarly, those in the sadness-before-anger-condition did not experience a decrease in anger after the first mood induction (i.e., sadness; t(33) = 0.35, p =.73; baseline, M = 60.82, SD = 19.90; post sadness induction; M = 59.38, SD = 30.67). Changes in means and their associated standard deviations are summarized in Fig. 2. These findings indicate any observed decline in anger would be attributable to the two-step sequence of emotion, as opposed to either mood induction alone. That sequential effect is tested in Hypothesis 1 (below).
Fig. 2
On average, for participants whose distress was primarily related to feeling angry, there was no significant change following a single mood induction, whether that was anger or sadness. However, one of the specific sequences of two ordered emotions did have an impact. That is, there was a greater decrease in the intensity of anger for angry individuals when they expressed anger-before-sadness as compared to when the same emotions were expressed in the reverse order. Moreover, the anger-before-sadness condition reduced angry feelings in this group by 27.96% from pre-intervention
×
![]()
Single Mood Induction Effects on Sad Group
A paired-sample t-test was conducted using the sadness-before-anger condition to parcel out the effect of the first mood induction from the synergistic effect created by adding the second mood induction. For those in the sadness-before-anger condition, a mood induction of sadness alone did not decrease sadness intensity t(47) = 1.35, p =.19, baseline, M = 76.46, SD = 18.25; post-sadness induction, M = 72.60, SD = 23.86). Likewise, those in the anger-before-sadness-condition did not decrease in their anger after the first mood induction (i.e., anger; t(48) = 1.72, p =.09; baseline, M = 66.14, SD = 29.29; post anger induction; M = 60.02, SD = 30.76). Changes in means and their associated standard deviations are summarized in Fig. 3. Once again, findings confirm a decline in sadness intensity could be uniquely related to the ordered sequence of feeling sadness followed by anger, as opposed to experiencing isolated mood inductions. That effect is tested in Hypothesis 2 (below).
Fig. 3
On average, for participants whose distress was primarily related to feeling sad, there was no significant change following a single mood induction, whether that was anger or sadness. However, one of the specific sequences of two ordered emotions did have an impact. That is, there was a greater decrease in the intensity of sadness for sad individuals when they expressed sadness-before-anger as compared to when the same emotions were expressed in the reverse order. Moreover, the sadness-before-anger condition reduced angry feelings in this group by 25.65% from pre-intervention
×
![]()
Hypothesis Testing
Each hypothesis was first separately evaluated through three-step hierarchical regression analyses. In step 1, five demographic variables were entered to control for their potential covariate effects (i.e., age, gender, ethnicity, sexual orientation, and relationship status). In step 2, pre-intervention visual analogue scale scores were entered to control for differences in initial levels of emotion intensity. In step 3, a dummy-coded variable representing the interaction of the group (i.e., angry, or sad) and experimental condition (i.e., anger-before-sadness and sadness-before-anger) was entered as a predictor. Notably, demographic variables did not significantly predict intensity of anger (R2 = 0.05, R2Adjusted = 0.03; F(5, 62) = 0.58, p =.716) or sadness (R2 = 0.07, R2Adjusted = 0.032; F(5, 90) = 1.41, p =.227). In both analyses, only pre-intervention scores and the interaction between groups and condition significantly predicted post-intervention changes in emotion intensity. As such, the models were re-analyzed using a two-step hierarchical multiple regression. Findings are presented below.
Hypothesis I: Changes in Anger Intensity
The first two-step hierarchical multiple regression analysis was conducted to evaluate whether participant group and condition predicted changes in anger intensity (see Table 2). Results showed a final model that significantly explained 48% of the variance in post-intervention ratings of anger intensity on the Visual Analogue Scale (R2Adjusted = 0.48; F(2, 67) = 33.16, p <.001).
Table 2
Regression coefficients for best predictor of changes in target emotion intensity in angry group
b | SE b | β | T | R | R2 | R2Adjusted | |
|---|---|---|---|---|---|---|---|
Step 1 | 0.65* | 0.42* | 0.41* | ||||
Pre-intervention anger intensity | 0.76* | 0.11 | 0.65 | 6.97* | |||
Step 2 | 0.71* | 0.50* | 0.48* | ||||
Constant | 14.92 | 7.15 | 2.09 | ||||
Pre-intervention anger intensity | 0.75* | 0.10 | 0.65 | 7.51* | |||
Condition | − 17.01* | 5.19 | − 0.28 | − 3.28* |
In step 1, pre-intervention ratings of anger intensity contributed significantly to the overall regression model, (R2 =0.42, R2Adjusted = 0.41; F (1, 68) 48.61, p <.001), accounting for 41% of the variance in post-intervention ratings of anger intensity. An interpretation of the unstandardized coefficients revealed that within the angry group, those with higher ratings of anger intensity at baseline, reported a greater decrease in anger, by 0.75 units on average, at post- intervention (b = 0.75, t = 7.51, p <.001).
In step 2, condition (i.e., anger-before-sadness, or sadness-before-anger) was found to predict a significant medium effect (d = − 0.56) that explained an additional variance of 7% over and above pre-intervention ratings of anger intensity (ΔR2Adjusted = 0.07). Among those in the angry group, participants in the anger-before-sadness reported a significantly greater decrease in anger intensity, by 17.01 units on average, than those in the sadness-before-anger condition (b = -17.01, t = -3.28, p <.002). The relative unstandardized and standardized coefficients are summarized in Table 2. Figure 2 presents the means and standard deviations for pre-to-post intervention ratings of anger intensity for each experimental condition.
Hypothesis 2: Changes in Sadness Intensity
The second hierarchical multiple regression analysis was conducted to evaluate whether participant group and condition predicted changes in sadness intensity (see Table 3). Results showed a final model that significantly explained 42% of the variance in post-intervention ratings of anger intensity on the Visual Analogue Scale (R2 = 0.43; R2Adjusted = 0.42; F(2, 94) = 35.08, p <.001).
Table 3
Regression coefficients for best predictor of changes in Target-Emotion intensity in sad group
b | SE b | β | t | R | R2 | R2Adjusted | |
|---|---|---|---|---|---|---|---|
Step 1 | 0.60* | 0.36* | 0.35* | ||||
Pre-intervention anger intensity | 0.76* | 0.11 | 0.60 | 7.23* | |||
Step 2 | 0.65* | 0.43* | 0.42* | ||||
Constant | 9.05 | 7.45 | 1.21 | ||||
Pre-intervention intensity | 0.84 | 0.10 | 0.66 | 8.21* | |||
Condition | − 16.23* | 4.70 | − 0.28 | − 3.46* |
In step 1, pre-intervention ratings of sadness intensity contributed significantly to the overall regression model, (R2 = 0.36; R2Adjusted = 0.35; F (1, 95) 52.20, p <.001), accounting for 35% of the variance in post-intervention ratings of sadness intensity. Within the sad group, participants with higher ratings of sadness intensity at baseline reported significantly greater reduction in intensity of sadness, by 0.84 units on average, during post-intervention (b = 0.84, t = 8.21, p <.001).
In step 2, the condition (i.e., anger-before-sadness, sadness-before-anger) was found to predict a significant small effect (d = − 0.26) that explained an additional variance of 7.2% over and above pre-intervention ratings of sadness intensity, and significantly predicted post-intervention ratings of sadness intensity (ΔR2Adjusted = 0.7). In the sad group, those in the sadness-before-anger reported a significant 16.23 unit decrease in sadness intensity on average, relative to those in the anger-before-sadness condition (b = -16.23, t = 3.46, p <.001). The relative unstandardized and standardized coefficients are summarized in Table 3. Figure 3 presents means and standard deviations for pre-to-post intervention ratings of sadness for each experimental condition.
Discussion
The present study examined whether resolution of lingering anger and sadness depends on the sequence in which anger and sadness are experienced. To date, the literature has primarily examined emotional sequences of sadness and anger in analogue experimental manipulations in a lab, while the few studies that have examined individuals’ lived experiences have each tested one sequence in isolation. In addressing these limitations, the findings of the present study are important for understanding how the purposeful activation of opposing emotions in specific sequences (i.e., anger-then-sadness or sadness-then-anger) may be used to alleviate lingering sadness and anger related to interpersonal conflict. The current findings may inform the use of these emotion sequences in psychotherapy.
In line with both hypotheses, results of the present study suggest that the best sequence of emotions depends on the presenting concern. Participants with lingering anger reported a greater reduction in anger (d = − 0.56) when they experienced anger-before-sadness (instead of sadness-then-anger). Likewise, participants who presented with lingering sadness reported a greater reduction in sadness (d = − 0.26) when they experienced sadness-before-anger (rather than the reverse sequence).
Activate the Target Emotion First and Opposing Emotion Second
In essence, findings suggest that individuals benefit from being asked to express and explore their presenting emotion first, before activating an opposing emotion second. These findings align with the general pattern of working with distressing emotions seen in psychotherapy, in which the client’s most distressing complaints often need to be acknowledged and validated before efforts are made to help regulate associated emotional distress or other distressing complaints are explored in detail. For example, in cognitive-behavioral therapy, thought records and behavioral experiments begin by targeting the thought or emotion that clients rate as most distressing, before beginning to form more balanced thoughts to attenuate distress and addressing other distressing thoughts and emotions (Greenberger & Padesky, 2016). Similarly, emotion-focused therapy interventions typically begin with validating and expressing presenting emotion states before shifting to more adaptive states (Greenberg, 2017). As follows, participants may have benefited from being able to first acknowledge and express their presenting emotional concern (i.e., their sadness or anger), especially given that those concerns had been lingering for approximately 6 months.
In addition, it may be necessary to activate a target emotion to create carry-over effects. In other words, anger or sadness might need to be explicitly experienced in the present moment for the subsequent successful activation of an opposing emotion and to produce the overall synergistic effect observed in our study. Consistent with the literature on emotional processing, important facets of emotional information (e.g., thoughts, beliefs, unmet needs) that underly a distressing emotion are more readily accessed when the emotion is activated and experienced in the present moment, which is necessary for emotion regulation (Ask & Granhag, 2007; Greenberg & Pascual-Leone, 2006). In the current study, expressing the opposing emotion first wasnot as effective, perhaps because the target emotion had not yet beenactivated and explored in detail. In contrast, participants who were guided to feel their presenting (target) emotion first may have been better able to use the subsequent opposing emotion to then further elaborate facets of emotional information through the lens of a new and integrated experience of emotion. Ultimately, it is the experience of two emotions in an ordered combination that helped attenuate a participant’s lingering problem state, whether that was anger or sadness.
Anger-Before-Sadness Softens Lingering Anger
For individuals with lingering anger, expressing anger before sadness appears more useful than the inverse order. This finding is consistent with some experimental research (Zhan et al., 2017a) and parallels research on opposite action training (Rizvi et al., 2011). Initially, this finding appears inconsistent with Lutz and Krahé’s (2018) finding that sadness induction reduced aggressive behaviors, regardless of the order in which anger and sadness were induced. However, an important contrast with that study’s approach is it did not conceptualize emotion itself as a potential meaning-making experience but rather framed it as a symptom to be quelled, which informed certain methodological decisions. For instance, Lutz and Krahé (2018) studied anger induced by instructing participants to complete challenging numerical problems, as opposed to anger that was related to a previous interpersonal interaction. In addition, the study assessed anger through the frequency of aggressive behavior, rather than the degree of someone’s perceived anger intensity or the meaning it entailed.
Emotion-focused theory has characterized anger as a tendency to reject other viewpoints or assert oneself (Pascual-Leone et al., 2013), which could make one less willing to explore or engage opposing emotions, at least at first. However, attempts to activate and explore an opposing emotion may be more effective after feelings of anger have already been fully expressed. When anger is expressed first and sadness is expressed second, anger seems to become amenable to transformation through feelings of sadness, at the tail end of the intervention sequence. In this way, the presenting anger is effectively “softened” by subsequent sadness.
From a cognitive behavioral perspective, expressing sadness may allow angry individuals to adopt a new perspective and challenge the beliefs that were contributing to their anger. Day et al. (2008) reviewed the importance of perspective-taking in cognitive behavioral interventions for anger. They found that for participants presenting with lingering anger, expressing their anger may have had a ruminative function as they rehearsed the beliefs that contribute to their anger (e.g., blaming others), whereas the expression of sadness may allow them to express other beliefs about the situation (e.g., feeling sympathetic for others/oneself).
In contrast, for participants in our study who started out angry, ending the intervention with the expression of that anger seems to have had almost zero impact on their final anger intensity. This is remarkable, and as if the entire sequence had been an inert control condition. Among people with anger problems, research has pointed to the self-rewarding nature of anger (Pascual-Leone et al., 2013; Peters et al., 2018). Participants in the angry group had largely self-identified as having unresolved anger, so prompting them to end the intervention in anger may have reignited those initial feelings, possibly neutralizing any impact the sadness expression might have had.
Sadness-Before-Anger Counteracts Sadness
For individuals who present with lingering sadness, the expression of sadness first and anger second appears to be more effective than the inverse sequence of emotions, which is consistent with previous research conducted in those with depression (Choi et al., 2016). One potential explanation for the finding may be the interaction between the different meaning-making elicited by each emotion state. Sadness is often associated with recognizing the losses and/or woundedness that occur from interpersonal conflict (e.g., opportunities lost), whereas anger is associated with righteous assertion of the injustices or unfairness one experienced (e.g., personal boundary being violated; Greenberg & Goldman, 2017; Pascual-Leone, 2018). Expressing sadness first may have allowed for such losses and woundedness to be explicitly acknowledged and explored. However, the subsequent expression of anger seems to have facilitated an interaction, where the lingering sadness state was then combined with a new meaning-making state (i.e., assertive anger). In this way, those in the sadness-before-anger condition may have been better able to recognize how the interpersonal conflict negatively impacted them, attenuating their presenting sadness.
Alternatively, the findings may be explained by those with lingering sadness as having presented with global distress, which is generally unproductive. The literature has suggested that those with lingering sadness may be more likely than those with lingering anger to be in a state of global distress, which is commonly described as a sense of “hopelessness” or vague “helplessness” and may manifest as sadness (Pascual-Leone & Greenberg, 2007; Pascual-Leone, 2018). If people initially presenting with lingering sadness are in fact feeling global distress, then they will have relatively less differentiation in the meaning of their emotional state. As such, the subsequent expression of anger may have been helpful because it directs individuals to a more differentiated state of meaning making.
This finding is also contextualized by research demonstrating the association between low levels of assertiveness and greater depression symptoms, as well as research demonstrating that assertiveness training may reduce depression symptoms (e.g., Speed et al., 2017; Sanchez et al., 1980). Assertiveness can be conceptualized as a form of healthy anger, whereby one asserts their needs to others (e.g., Pascual-Leone, 2018). Perhaps, by ending the intervention with the expression of anger, participants with lingering sadness may feel more motivated to assert themselves. In contrast, when individuals initially felt sad, ending the intervention by expressing that sadness may instead focus them back on depressive rumination, which has been implicated in the maintenance of depressive symptoms (Lyubomirsky & Tkach, 2004).
Conclusions, Limitations, and Future Research
Findings point to the conclusion that the order of emotions matters, but the optimal order seems to depend on one’s presenting concern (i.e., presenting anger or sadness). For individuals who were angry about an interpersonal problem, expressing their anger and then their sadness was more productive than if they had expressed the same emotions in reverse order. However, when individuals presented predominantly with sadness over what happened, then exploring and expressing their sadness before moving on to feelings of anger was more productive, as compared to if the same emotions were experienced but reversed. Overall, ending an intervention with the “new” (i.e., opposing) feeling seems to be most helpful.
The present study was limited by a somewhat low sample size, particularly for participants presenting predominantly with feelings of anger. Future studies should focus recruitment on identifying potential participants who are feeling angry about an interpersonal interaction. This study was also limited by its exclusive use of one self-report measure. However, measure selection was limited due to the online nature of the experiment. Future studies should consider replicating the current findings using an in-person intervention, which may allow for the use of additional measures to better gauge changes in intensity of anger and sadness (e.g., physiological measures, observer-rater measures of expressive arousal).
We did not follow up whether changes in intensity of the target (i.e., most problematic) emotion persisted beyond the conclusion of the intervention. Future studies should consider examining whether more sustained effects can be achieved when emotion sequences are repetitively activated over time and with help from a therapist. Research on emotion-focused therapy has demonstrated that even if moment-by-moment emotional changes are not sustained at the end of a session, repeated processing of lingering and opposing emotions can build over the course of treatment and help consolidate enduring change (Pascual-Leone et al., 2017).
It is worth noting that the current design, as well as others (e.g., Rizvi et al., 2011) used a computer-based interface to prompt sequences, which suggests working with “the flow of emotion” is not necessarily contingent on the presence of a responsive therapist. This process was mechanized through a careful order of step-by-step direct instructions, which opens the possibility of developing self-administered clinical tools. Still, we believe the intervention effects would be larger when delivered with the support of an empathically attuned therapist who could offer validation of the different experiences. Whatever the case, further directions in cognitive and behavioral research should explore this notion from emotion-focused therapy (Greenberg, 2017), that one emotion might be used as a means to undo another.
Finally, approximately half of the participants in the present sample were Caucasian and over three quarters identified as women. More research is needed to examine whether results generalize to individuals of diverse cultural backgrounds and other genders.
Clinical Implications
Our study showed that lingering feelings of anger and sadness experienced in response to interpersonal conflicts can be reduced by expressing the opposing emotion. In short, one emotion may be used as an antidote to another, depending on the presenting concern. The idea that emotions themselves could serve as processes to leverage change (Pascual-Leone, 2018), goes beyond traditional cognitive and behavioral theory, which had framed emotion as a symptom to be attenuated. Considering emotion-focused theory offers an alternative perspective on the mechanisms of therapeutic change (Greenberg, 2017). The study also demonstrates that a single-session of computer-mediated writing can be used to reduce the intensity with which individuals experience unresolved anger and sadness associated with a past interpersonal conflict. This is consistent with prior studies using guided emotional sequences, which have demonstrated that therapeutic emotional change can occur in a single-session exercise (Narkiss-Guez et al., 2015; Pascual-Leone & Baher, 2023; Zhan et al., 2017a).
Guided, online writing interventions, such as the intervention in the current study, could be explored as an adjunct treatment to therapy or a helpful exercise for individuals who are experiencing lingering distress about a interpersonal interaction.
Our results also serve as a starting point to understand the sequences of emotion that therapists might optimally guide clients towards during psychotherapy. Generally, the findings suggest that activation of a presenting emotion may be a necessary first step to effectively make use of subsequent emotional states as processes for leveraging moment-by-moment change. More specifically, individuals who present with lingering sadness about an interpersonal conflict seem to benefit from expressing their sadness first and then following through with an exploration of their anger. However, people presenting with anger seem to benefit more from starting with the expression of that anger before then moving on to explore their sadness. A critical issue is that an ordered pair of emotions seems to create a synergistic impact that is not better explained by the activation of any single state. Furthermore, the best order of these emotional sequences depends on the individual’s presenting concern. This supports the idea that sequential transformations of emotion may be a unique form of emotional processing, where one emotion can be used to undo and change another.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.