پیش بینی مقررات خود انگیخته ابراز هیجانی منفی و عملکرد شناختی بعدی با کنترل واگ قلبی

Abstract The present research investigated whether cardiac vagal control (as measured by respiratory sinus arrhythmia, RSA) predicts an individual's predisposition to suppress negative emotional expressions. One hundred thirty-six participants watched either a negative film or a neutral film. Facial expressions were recorded during the film and subjective emotional responses were assessed afterwards. Participants performed verbal and spatial working memory tasks both before and after the film clips. We found that resting RSA modulated the degree of coherence between facial expressions of emotion and subjective emotional experience in the negative film condition. Specifically, participants with higher resting RSA expressed less but reported feeling just as much negative emotion as those with lower resting RSA. Moreover, higher resting RSA predicted smaller pre-film to post-film improvements in spatial working memory performance in the negative film condition, suggesting that expressive suppression among high RSA participants temporarily undermined the operation of working memory. In the neutral film condition, resting RSA did not relate to expressive or subjective responses or subsequent working memory performance. These results support the notion that cardiac vagal control reflects an internal marker of self-regulatory tendencies and suggest that spontaneous self-regulation associated with individual differences in resting RSA may temporarily deplete self-regulatory resources.

. Results Given the strong directional nature of our predictions based on prior evidence, all statistics related to hypotheses testing are reported as one-tailed (except the second regression analysis for testing of Hypothesis 1). Please see Table 1 for demographic, physiological, self-report, facial valence, and cognitive data. Table 1. Demographic, physiological, self-report, facial valence, and cognitive data by film. Neutral film (N = 68) Negative film (N = 68 a) Mean SD Mean SD # Men/women 32/36 34/34 Age 18.76 1.12 19.04 1.78 Resting RSAb (ms2) 6.31 1.39 6.22 1.08 Self-reported valencec 3.29 (3.35/3.24) 0.62 (0.60/0.65) 2.03 (2.00/2.06) 0.88 (0.89/0.89) Facial valencec 3.93 (3.97/3.88) 0.28 (0.12/0.37) 3.47 (3.72/3.21) 0.82 (0.62/0.92) Reaction time (ms) Pre-film Verbal 952.17 284.91 1000.33 275.87 Spatial 895.07 258.54 953.17 251.74 Post-film Verbal 830.15 248.98 907.08 237.96 Spatial 806.56 252.59 816.62 198.88 Difference scored Verbal 122.02 143.54 93.87 141.45 Spatial 88.50 168.57 135.93 146.69 Error rate (%) Pre-film Verbal 12.78 8.95 13.15 7.49 Spatial 12.79 10.09 10.24 6.43 Post-film Verbal 9.54 7.72 9.88 6.89 Spatial 11.49 9.09 7.29 5.52 Difference scored Verbal 3.24 6.70 3.35 6.59 Spatial 1.31 6.89 2.85 5.39 a For the negative film group, the sample size for most analyses is 68. However, for analyses involving facial valence data, the sample size is 67 because one participant in this group did not consent to be videotaped while watching the film clip. b RSA, Respiratory sinus arrhythmia. c Scales: Self-report valence = 1 to 3 to 5, negative to neutral to positive; Facial valence = 1 to 4 to 7, negative to neutral to positive. Numbers in the parentheses are values for the high and low resting RSA subgroups respectively (high resting RSA subgroup/low resting RSA subgroup). Participants were parsed into the two subgroups based on their resting RSA relative to the median (6.22 ms2 for the negative film group, 6.11 for the neutral film group). d Difference score is calculated by subtracting post-film data from pre-film data. Table options 2.1. Sample characteristics Data from the 136 participants were included in the analyses below.1 Half the participants watched the negative film clip (N = 68, 34 female, mean age = 19.04 years, SD = 1.78 years) and half watched the neutral film clip (N = 68, 36 female, mean age = 18.76 years, SD = 1.12 years). Chi-square test and independent sample t-tests revealed no significant differences between the two film groups with regard to gender, χ2(1) = 0.12, p = .73, age, t(134) = −1.09, p = .28, or resting RSA level, t(134) = 0.41, p = .69 (for neutral film group, mean resting RSA = 6.31 ms2, SD = 1.39 ms2; for the negative film group, mean resting RSA = 6.22 ms2, SD = 1.08 ms2). 2.2. Manipulation checks Both self-reported emotional experience (from the SAM) and facial valence (averaged across two raters) were submitted to independent sample t-tests using film clip type as the independent variable. As expected, significant effects of film clip on both self-reported affective experience, t(134) = 9.66, p < .001, and facial valence, t(133) = 4.36, p < .001, were found. Specifically, participants who watched the negative film clip showed more negative facial expression (M = 3.47, SD = 0.82) and reported greater negative affect (M = 2.03, SD = 0.88) compared to participants who watched the neutral film clip (facial valence M = 3.93, SD = 0.28; self-reported valence M = 3.29, SD = 0.62). 2.3. Hypotheses testing Hypothesis 1. Individuals with higher resting RSA will exhibit less coherence between emotional experience and facial expression in response to the negative film clip. On the basis of previous work (Gross and John, 2003 and Gross et al., 2000), we calculated two measures of the coherence between emotional experience and emotional expression to test Hypothesis 1. One measure was the difference score between each participant's emotional experience and facial expression. The other measure was the correlation between the emotional experience and facial expression (the smaller the correlation, the less the coherence). The first coherence measure was based on the difference between experience and expression. We standardized self-reported valence (sValence) and facial valence (fValence) and then calculated a difference score for each participant by subtracting standardized facial valence from standardized self-reported valence. Because smaller values on both variables indicate more negative valence, a positive difference score indicates that a participant expressed more negative emotion than they reported feeling, whereas a negative difference score indicates that a participant expressed less negative emotion than they reported feeling. To assess Hypothesis 1, we examined the extent to which resting RSA predicted the coherence (difference) score within the negative and neutral film groups separately. As expected, in the negative film condition, resting RSA was inversely related to the coherence score, r = −.21, p = .05. That is, consistent with the view that these participants suppressed their facial expression of negative emotion, participants with higher resting RSA exhibited less coherence between self-reported emotional experience and facial expression of emotion. The relationship between resting RSA and the coherence score was not significant in the neutral film clip condition, r = −.02, p = .44. Fig. 1 depicts these two correlations. These different patterns between the two film groups suggest an interaction between resting RSA and film type in terms of their influence on the coherence between participants’ self-reported valence and facial valence. To test this interaction effect, we performed a regression analysis using the difference score as dependent variable and standardized resting RSA, film clip, and their interaction (RSA × Film) as predictor variables. As expected, the interaction term was significant, B = −0.33, SE = 0.18, t = −1.80, p < .05. Moreover, the main effect of film type was also significant (B = −0.56, SE = 0.17, t = −3.23, p < .05), whereas the main effect of resting RSA was not (B = 0.33, SE = 0.26, t = −1.27, p > .10). For participants in the negative film group only, there was a negative ... Fig. 1. For participants in the negative film group only, there was a negative correlation between resting RSA and the difference score between self-reported valence and facial valence (self-reported experience minus facial expression). Specifically, the higher RSA participants had at baseline, the less negative emotion participants showed compared to what they actually felt. Self-reported experience and facial expression data were standardized (z-score) before being entered into the analysis. Note: A positive difference score indicates expressing more negative emotion than one feels and a negative difference score indicates expressing less negative emotion than one feels (i.e., automatic suppression of facial expression). Figure options As a second way to test Hypothesis 1, we examined the correlation between emotional experience and emotional expression and tested the extent to which resting RSA moderated the relationship. This analysis included data from the negative film group only. In a regression model, facial valence (fValence) was the criterion variable and self-reported valence (sValence), resting RSA, and their interaction term (i.e., RSA × sValence) were the predictor variables. We expected to find that lower (i.e., more negative) self-reported valence predicted more negative facial expressions, unless participants had relatively high resting RSA. In order to minimize both Type I and Type II errors when testing the interaction term, two quadratic terms – the square of resting RSA (RSA2) and the square of self-reported valence (sValence2) – were also included in the regression equation as predictor variables. 2 All predictor variables were standardized before being entered into the regression analysis. Consistent with response coherence theory ( Ekman, 1992, Lazarus, 1991 and Levenson, 1994) and previous research relating high resting RSA and expressive suppression (e.g., Demaree et al., 2004 and Demaree et al., 2006a), results showed that both self-reported valence (B = 0.29, SE = 0.10, t = 2.85, p = .01) and resting RSA (B = 0.25, SE = 0.10, t = 2.40, p = .02) significantly predicted facial valence. Specifically, greater negative self-reported affect was associated with greater negative facial expression whereas higher resting RSA was associated with less negative facial expression. Neither the quadratic terms of resting RSA (B = −0.03, SE = 0.06, t = −0.62, p = .54) nor those of self-reported valence (B = 0.003, SE = 0.09, t = 0.03, p = .98) significantly predicted facial valence. Of greatest importance is the significant RSA × sValence interaction term, B = −0.23, SE = 0.11, t = −2.03, p = .05. The negative regression coefficient of this interaction term indicates that higher values of resting RSA predict less coherence (i.e., a weaker correlation) between self-reported emotional experience and facial expressions of emotion. This is precisely what we had expected. Please refer to Fig. 2, which depicts the correlation between emotional expression and emotional experience for individuals high versus low in resting RSA. For participants in the negative film group only, correlation between ... Fig. 2. For participants in the negative film group only, correlation between self-reported valence and facial valence was moderated by resting RSA. For the low RSA subgroup, the correlation between self-reported and facial valence was .54 (p < .001). For the high RSA subgroup, the correlation between the two variables was only .17 (p = .17). The difference between the correlations was significant, z = 1.69, p = .05). RSA, Respiratory sinus arrhythmia. Figure options In the neutral film condition, because expressive suppression is unlikely to occur while not feeling any emotion per se, we did not expect the correlation between self-reported emotional experience and facial expression to be influenced by participant's resting RSA. To test this hypothesis, we repeated the regression analysis using data from the neutral film group. As expected, the regression coefficient for the RSA × sValence interaction term was not significant, B = 0.01, SE = 0.06, t = 0.22, p = .83. Resting RSA, self-reported affect, and their quadratic terms (RSA2 and sValence2) were all nonsignificant (for resting RSA and its quadratic term, B = 0.02, SE = 0.04, t = 0.60, p = .55; B = −0.01, SE = 0.03, t = −0.39, p = .70, respectively; for self-reported affect and its quadratic term, B = 0.08, SE = 0.06, t = 1.49, p = .14; B = −0.03, SE = 0.03, t = −1.18, p = .24, respectively). Please see Table 2 and Table 3 for full reports of the results of these two regression analyses. Hypothesis 2. Individuals with higher resting RSA will exhibit smaller improvements from pre-clip to post-clip measurements of working memory performance, but only in the negative film clip condition. Table 2. Correlation matrix of regression analyses for Hypothesis 1. fValence RSA sValence RSA × sValence RSA2 RSA .17 (.08) – – – – .12 (.16) – – – – sValence .34 (.002) −.07 (.29) – – – .17 (.08) .10 (.22) – – – RSA × sValence −.15 (.11) .44 (<.001) −.10 (.21) – – −.13 (.14) −.55 (<.001) −.28 (.01) – – RSA2 .06 (.31) −.15 (.11) .23 (.03) −.38 (<.001) – −.11 (.18) .09 (.24) −.33 (.003) .09 (.23) – sValence2 .11 (.18) −.10 (.20) .30 (.01) −.14 (.13) −.005 (.49) −.02 (.42) −.13 (.15) .65 (<.001) .16 (.10) −.12 (.16) Note: Numbers in the parenthesis are p values for correlation coefficients. Numbers at the top of each cell are data for the negative film group while those at the bottom are data for the neutral film group. fValence, Facial valence; RSA, baseline respiratory sinus arrhythmia; sValence, self-reported valence; RSA × sValence, the product of resting RSA and self-reported valence; RSA2, the square of resting RSA, sValence2; the square of self-reported valence. Table options Table 3. Results of regression analyses for Hypothesis 1. DV IVs B SE t p (B) View the MathML sourcerpart2 (%) Tolerance R2 (%) p (R2) Negative film fValence RSA 0.25 0.10 2.40 .02 7.50 .81 20.86 .01 sValence 0.29 0.10 2.85 .01 10.53 .86 RSA × sValence −0.23 0.11 −2.03 .05 5.37 .70 RSA2 −0.03 0.06 −0.62 .54 0.50 .81 sValence2 0.003 0.09 0.03 .98 0.001 .88 Neutral film fValence RSA 0.02 0.04 0.60 .55 0.53 .68 6.72 .49 sValence 0.08 0.06 1.49 .14 3.34 .38 RSA × sValence 0.01 0.06 0.22 .83 0.07 .54 RSA2 −0.01 0.03 −0.39 .70 0.23 .85 sValence2 −0.03 0.03 −1.18 .24 2.08 .44 Note: fValence, Facial valence; RSA, baseline respiratory sinus arrhythmia; sValence, self-reported valence; RSA × sValence, the product of resting RSA and self-reported valence; RSA2, the square of resting RSA; sValence2, the square of self-reported valence; DV, dependent variable; IVs, independent variables; B , unstandardized regression coefficient; SE, standard error of the regression coefficient; View the MathML sourcerpart2, the square of part (or semipartial) correlation coefficient; R2, the square of the multiple correlation coefficient. All the independent variables are standardized before being entered into the regression analyses. Regression coefficients (B) that are significant are in bold. A Tolerance score that is smaller than 0.05 indicates serious multicollinearity problem, which is not the case for all the independent variables for both of the two regression analyses. Table options Paired sample t-tests showed that, for both the spatial and verbal 2-back tasks, participants’ performance improved from before the film clip to after the film clip regardless of which film clip participants viewed. That is, for participants in both film groups (negative and neutral), reaction time and error rate on both tasks significantly decreased following the film (please refer to Table 1), all ps < .001. The only exception was that the decreased error rate on the spatial 2-back task in the neutral film group was only marginally significant, p = .06. To test whether higher resting RSA was associated with lesser improvements in working memory performance, correlations between resting RSA and reaction time difference scores were calculated for the spatial and verbal 2-back tasks, respectively (please see Table 4). As expected, resting RSA significantly predicted the reaction time difference score for the spatial task, r = −.29, p = .01. We also observed a trend whereby resting RSA predicted the reaction time difference score on the verbal task, r = −.17, p = .08. To determine whether these decreased reaction time difference scores found among high resting RSA people was caused by a trade-off between speed and accuracy, correlations between resting RSA and error rate difference scores on both tasks were also calculated. Results revealed that resting RSA did not predict the difference score of error rate for either task (for spatial tasks, r = .01, p = .48; for verbal tasks, r = .12, p = .17). These results indicate that individuals with higher RSA levels (i.e., those who tend to regulate their facial expression to negative stimuli) had relatively impaired working memory functioning as compared to lower RSA individuals—that is, they were slower without improved accuracy. Table 4. Correlation between resting RSA and difference score of cognitive performance. Resting RSA Negative Neutral ΔReaction time (spatial) −.29 (.01) .05 (.34) ΔReaction time (verbal) −.17 (.08) .09 (.23) ΔError rate (spatial) .01 (.48) −.13 (.14) ΔError rate (verbal) .12 (.17) .02 (.43) Note: Numbers in the parentheses are the p values (one-tailed) for the correlation coefficients. Correlation coefficients that are significant (or marginally significant) are in bold. Resting RSA, baseline respiratory sinus arrhythmia; ΔReaction time (spatial), difference score of reaction time on spatial 2-back task; ΔReaction time (verbal), difference score of reaction time on verbal 2-back task; ΔError rate (spatial), difference score of error rate on spatial 2-back task; ΔError rate (verbal), difference score of error rate on verbal 2-back task. Table options In the neutral film group, cognitive performance following the film was not anticipated to correlate with resting RSA levels. Correlations using data from the neutral film group indicated that, as expected, resting RSA predicted none of the four difference scores (please see Fig. 3). For the difference score of reaction time on the spatial working memory task, r = .05, p = .34, for that of error rate, r = −.13, p = .14. For the difference score of reaction time on the verbal task, r = .09, p = .23, for that of the error rate, r = .02, p = .43. Correlation between resting RSA and the difference score between performance ... Fig. 3. Correlation between resting RSA and the difference score between performance (both reaction time and error rate) on verbal and spatial 2-back tasks for both neutral film and negative film groups. For participants in the negative film group only, resting RSA was negatively correlated with the decrease in reaction time on the spatial 2-back task, r = −.29, p = .01. Resting RSA also negatively correlated with the decrease in reaction time on the verbal 2-back task in the negative film group, but this correlation was marginally significant, r = −.17, p = .08. The correlations between resting RSA and the decrease in error rate on both tasks were nonsignificant. Thus, the relationship between resting RSA and working memory reaction time was not confounded by a trade-off between speed and accuracy. The correlation between resting RSA and the difference scores in cognitive performance (both reaction time and accuracy scores) were not significant in the neutral film condition. Figure options It is possible that the smaller differences in pre-film to post-film performance among high RSA individuals reflects a ceiling effect, such that high RSA participants performed exceedingly well on the initial assessment and hence had limited room to improve their performance on the post-film assessment. To rule out this possibility, we conducted regression analyses using post-film performance as the criterion variable and both pre-film performance and resting RSA as predictor variables. Results were similar to the difference score analyses reported above. Resting RSA predicted post-film reaction times (but not error rates, ps > .10) on the spatial working memory task in the negative film condition, β = .12, SE = 13.31, t = 1.71, p < .05, but did not predict post-film performance in the neutral film group (ps > .10). The only divergence from the difference score analysis is that resting RSA failed to predict reaction time on the verbal working memory task following the negative film clip, when controlling for pre-clip performance on the verbal task, β = .08, SE = 13.87, t = 1.20, p > .10.