Compassion-focused reappraisal, benefit-focused reappraisal, and rumination after an interpersonal offense: Emotion-regulation implications for subjective emotion, linguistic responses, and physiology

Abstract

This repeated measures psychophysiology experiment studied three responses to a past interpersonal offense (38 females and 33 males). We compared rumination with two offense reappraisal strategies. Compassion-focused reappraisal emphasized the offender's humanity, and interpreted the transgression as evidence of the offender's need for positive transformation. Benefit-focused reappraisal emphasized insights gained or strengths shown in facing the offense. Supporting the manipulations, compassion-focused reappraisal stimulated the most empathy and forgiveness, whereas benefit-focused reappraisal prompted the most benefit language and gratitude. Both reappraisals decreased aroused, negative emotion, and related facial muscle tension at the brow (corrugator). Both reappraisals increased happiness and positive emotion in ratings and linguistic analyses. Compassion stimulated the greatest social language, calmed tension under the eye (orbicularis oculi), and slowed heart beats (R–R intervals). A focus on benefits prompted the greatest joy, stimulated smiling (zygomatic) activity, and buffered the parasympathetic nervous system against rumination's adverse effects on heart rate variability (HRV).

Keywords:

• benefit-finding
• compassion
• forgiveness
• gratitude
• rumination
• facial EMG
• heart rate variability (HRV)
• spectral analysis

Acknowledgments

We gratefully acknowledge the support offered through a grant to C.V. Witvliet from the Fetzer Institute (grant no. 2393), and consultation from Heath Demaree and Christopher Barney on cardiovascular variables; Terry Blumenthal on facial EMG; Al Dueck on LSA; Mike Alexander, Timothy Brandt, Lindsey Lawrence, Nora Slenk, Allison Smith, and Ronna Zeluff for assistance with data collection and computation; John Shaughnessy and Scott VanderStoep for consultation on analyses; and David G. Myers, Mike McCullough, Sidney Callahan, and Philip Reynolds for their specific critiques of an earlier draft. This work contributes to an interdisciplinary project on The Pursuit of Happiness established by the Center for the Study of Law and Religion at Emory University and supported by a grant from the John Templeton Foundation.

Notes

1. Emotion-regulation strategies are categorized into antecedent-focused strategies, which are employed to influence responses prior to the full experience of an emotional event, and response-focused strategies, which are used after experience of an emotion to down- or up-regulate its effect (Gross, 1998). Such strategies may be employed consciously or automatically. Studies focusing on emotion-regulation timing features typically present novel emotion-eliciting sensory stimuli in the laboratory (Gross, 1998; Gross & Levenson, 1997). In assessing an emotional response to a real-life interpersonal offense, however, the preexistence of the emotional event may not allow for the categorization of emotion-regulation strategies using time unless it is calculated in relationship to the onset of memory activation.

2. LSA (http://lsa.colorado.edu) was used to investigate the similarity of participants’ language in these paragraphs compared to affective texts within the context of a repertoire of Western writing. LSA does not count words, but rather simulates representations of human knowledge. LSA uses a semantic corpus based on a large repertoire of Western writing from the third grade level through the first year of college and then applies a technique similar to factor analysis (singular value decomposition). Within the semantic space, LSA determines the similarity of two texts by calculating a cosine value. Here, we compared the narrative a participant produced in an experimental condition to a comparison positive emotion text, and then to a negative emotion text. Thus, for each emotion word probe (positive and negative), each participant has a cosine for each experimental condition (compassion-focused reappraisal, benefit-focused reappraisal, and rumination).

3. An LIWC dictionary of benefit words was created for the current study based on words present in participants’ written paragraphs. Because analyses with our benefit dictionary yielded the same results as using the one developed by McCullough et al. (2006), we report the results using their dictionary. Our LIWC forgiveness word dictionary was altruism, amend*, compassion*, empath*, forgave, forgiv*, love*, loves*, loving*, merciful*, mercy*, sympath*, and appreciativ*. Our gratitude word dictionary for LIWC was blessed, glad, gladness, grateful*, gratitude, and thank*. Two raters – blind to condition – developed mutually exclusive forgiveness and gratitude categories from a random order of participant responses. Raters separately determined whether to accept or reject each word based on goodness-of-fit in its designated category. Inter-rater reliability was 100% for gratitude and 97% for forgiveness, with consensus used to discard two forgiveness words and reach 100% agreement.

4. To serve as a ground, we attached skin conductance level pregelled Biopac EL507 snap electrodes fitted to LEAD110A electrode leads placed on the index and middle fingers of the left hand. Data were sampled at 62.5 Hz and amplified by a Biopac GSR100C electrodermal response amplifier set for a gain of 5 mho/V. As in other imagery studies, only habituation was found. Facial EMG activity was measured (Biopac EMG 100C units) on a second-to-second basis for the zygomaticus (cheek) muscle, orbicularis oculi (under eye) muscle, and corrugator supercilii (brow) muscle regions using two 4 mm EL258RT Biopac Ag–AgCl electrodes placed at each site on the left side of the face. Skin was first prepared with an alcohol pad and Biopac Gel 100. Each electrode was fitted with a Biopac ADD204 adhesive collar and filled with gel. EMG was sampled at 2000 Hz amplified by Biopac EMG100C amplifiers set for a gain of 1000 and using 10 Hz high-pass and 5 kHz low-pass filters. EMG data was first digitally filtered using the Comb band stop filter to select the line frequency at 60 Hz and overharmonics selecting all up to the Nyquist frequency. Data were filtered using the FIR bandpass option to select the Bartlett window with a low-frequency cutoff fixed at 28, HF cutoff fixed at 500, and Q-coefficients set to 286. Next, the EMG data were rectified and integrated by averaging over 10 samples and taking the root mean square of the entire wave form. ECG data were measured by placing one Biopac pregelled El503 snap electrode fitted to a Lead110S on the left rib and one on the right clavicle. Rubbing alcohol was used to clean each electrode placement site. Heart rate data were sampled at 1000 Hz and amplified by 1000 Hz using the Biopac ECG100C ECG amplifier. Continuous R–R intervals were calculated in seconds for each condition using ECG data. The HRV specialized analysis function of Acqknowledge used methods and produced values that were not consistent with guidelines and expected ranges based on the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (1996) paper. Using the standards published by the Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology (1996), Paul DeYoung wrote a software program that followed the specifications published for calculating and using the HF component of the power spectrum to determine the parasympathetic contribution to the cardiac cycle. The 120 s trial R–R data were interpolated with cubic splines and then 1024 uniformly spaced values were calculated. A Welch periodogram estimate of the power spectrum density (PSD) was calculated from the Fast Fourier Transform of de-trended subintervals of the 120 s period (7 segments with a 50% overlap). Each subinterval was multiplied by a Hamming window. Results were cross validated with two other programs (HRV Analysis Software 1.1 from the Biomedical Signal Analysis Group, Department of Applied Physics, University of Kuopio, Finland; Mindware HRV 2.51). We also calculated correlations with values produced using the root mean squared successive differences (RMSSD) method (all rs ≥ 0.8), and we report the RMSSD results below.

5. As a comparison to spectral analysis, we used the time domain method of calculating the square root of the mean of the sum of the squares of the differences between consecutive R–R intervals (RMSSD). RMSSD is sensitive to the HF indicators of parasympathetic activation, but it also includes some lower frequency fluctuations indicative of sympathetic contributions (Berntson, Lozano, & Chen, 2005). The benefit-focused reappraisal effect on HRV was more reliable for the HF than for the RMSSD method. Only the benefit-focused RMSSD was marginally higher than the relevant offense RMSSD, F(1, 61) = 3.95, p = 0.051, partial η ² = 0.06, consistent with Berntson et al.'s (2005) characterization of HF as preferable for repeated measures analyses of HRV.