تفاوت های فردی در همدلی: نقش شناخت حالت چهره

Abstract We investigated individual differences in empathy and emotion recognition. Emotion recognition was operationalized as accuracy on the six emotions composing Ekman’s Pictures of Facial Affect. Facets of empathy were measured with the Empathy Quotient (EQ) and the empathic concern factor of the Interpersonal Reactivity Scale. Two aspects of emotion recognition were investigated: exposure length (50 ms and 2000 ms) and emotion type (fear). Both empathic concern and the EQ were related to accuracy at the brief exposure, however empathic concern accounted for the EQ findings. The EQ was connected to accuracy at the long duration, especially the “social skills” factor. Empathic concern was unrelated to fear recognition, whereas the EQ was highly related. These findings extend previous research by showing that empathy subtype predispositions are differentially related to expression recognition.

1. Introduction Accurate recognition of emotional facial expressions is an initial step to empathic responding. The capacity to experience empathy differs amongst individuals: what advantages do more empathic people have in facial expression recognition? Can they better identify emotions from fleeting facial expressions? Are they more adept at detecting certain emotional states, particularly distressing ones that elicit concern? This study examined facets of empathy and their relationship to expression presentation speed and emotion type. Distinguishing empathy subtypes and linking them to theoretically driven facial expression tasks may enrich understanding of empathy processes. Typically, the empathy construct has been separated into two types: cognitive and emotional (Davis, 1983 and Decety and Jackson, 2006). Cognitive empathy refers to imaginatively understanding another person’s thoughts, feelings and actions. Emotional empathy is feeling the emotion of another person, but maintaining a compassionate, other-focused perspective. It is characterized by visceral, automatic reactivity (Mehrabian & Epstein, 1972). Social skills have sometimes been included as a component of empathy; inappropriate social responses hinder empathic interactions (Lawrence et al., 2004 and Riggio et al., 1989). These dimensions are considered independent; for example, deficient cognitive empathy can coexist with elevated emotional empathy (Baron-Cohen and Wheelwright, 2004 and Blair, 2005). These independent dimensions may differentially relate to facial expression recognition. Few studies have directly examined self-reported empathy and facial expression recognition ability. Preliminary research has found positive relationships between self-reported emotional empathy and expression recognition (Martin et al., 1996, Riggio et al., 1989 and Gery et al., 2009). In addition, empathy is part of trait emotional intelligence (EI) and EI is positively associated with facial expression recognition (Austin, 2004, Ciarrochi et al., 2001 and Petrides and Furnham, 2003). Less is known about cognitive empathy’s connection to facial expression recognition. Evidently, self-ratings of cognitive empathy ability are unrelated to actual empathic accuracy, as measured by accuracy in determining the thoughts and feelings of another person after a brief interaction (e.g. Davis and Kraus, 1997 and Ickes et al., 2000). The Empathy Quotient (EQ; Lawrence et al., 2004) is a self-report scale that measures empathy multi-dimensionally, with an emphasis on cognitive empathy and social understanding, rather than emotional empathy ( Lawrence et al., 2004 and Muncer and Ling, 2006). This scale has been linked to better performance on social perception tasks, such as reading mental states from only the eyes ( Baron-Cohen, Wheelwright, Hill, Raste, & Plumb, 2001), and gender judgements based on animated facial movement ( Penton-Voak, Allen, Morrison, Gralewski, & Campbell, 2007). Its relationship to recognition of the basic emotional facial expressions has not yet been examined. Research suggests varying presentation duration speed and targeting specific types of emotions, such as fear, may produce connections between empathy and facial expression recognition (see Clark et al., 2008, Marsh et al., 2007, Marsh and Blair, 2008 and Martin et al., 1996). These approaches are discussed in more detail below. 1.1. Rapid recognition and empathy Recognition of briefly presented facial expressions requires rapid judgment with little conscious effort. Researchers largely agree there are two main levels for interpreting emotional stimuli with different brain pathways: one generates meaning automatically, with minimal cortical input; the other is deliberate and conscious (e.g. Buck and Ginsburg, 1997 and LeDoux, 1996). Testing accuracy to briefly presented expressions presumably isolates an important early component of the empathy process, accessing a more instinctive, biological level of emotion processing. Research has linked accurate recognition of the valence of briefly presented facial expressions to self-reported emotional empathy (Martin et al., 1996). Other research has linked recognition of briefly presented facial expressions to trait emotional intelligence (Austin, 2004 and Petrides and Furnham, 2003) and to disorders with cognitive empathy deficits (Clark et al., 2008). The apparent link between rapid recognition of facial expressions and empathy requires clarification: which aspects of empathy are most related? Experiencing emotional empathy often happens automatically and involuntarily; emotional empathy may therefore utilize automatic emotional processing levels more than other empathy subtypes. 1.2. Fearful expression and empathy Types of emotional expressions are processed in different parts of the brain, indicating they have distinct functions. The fearful expression is a distress cue and the amygdala plays a central role in its interpretation (Adolphs et al., 1998, Phillips et al., 1997 and Sato et al., 2002). Compassion for distress cues defines emotional empathy; a link between fear recognition and emotional empathy is therefore plausible. Empathy’s association with fear recognition has not been directly investigated, however a recent meta-analysis links disorders with deficient empathy, such as antisocial personality disorder and psychopathy, with impaired fear recognition (Marsh & Blair, 2008). Similarly, pro-social behaviour is associated with superior fear recognition (Marsh et al., 2007). If fear recognition and empathy are linked, does the more difficult brief exposure duration accentuate this link? What type of empathy is most related? A predisposition to have compassion for those in distress (emotional empathy) may coincide with better recognition of distress cues, such as fear. 1.3. The present study Facial expressions were presented to participants at brief (50 ms) and long durations (2000 ms). Empathy was measured with the EQ and the empathic concern subscale of the Interpersonal Reactivity Scale ( Davis, 1983). Relationships between the EQ and its subscales, empathic concern, exposure length and fearful emotion recognition were explored. We hypothesized the EQ would be associated with deficits at both the brief and long exposure. However, we hypothesized empathic concern would be a better predictor of recognition at brief exposures. We also hypothesized that empathic concern would show a relationship to impaired fear recognition, particularly at the brief exposure.

Results 3.1. Descriptive statistics and preliminary analyses Descriptive statistics for all variables are reported in Table 1. Mean accuracy rates were higher for 2000 ms (81%) than for 50 ms (66%). Bivariate correlations between the variables are reported in Table 2. The control task was unrelated to IRS-empathic concern, r = .03; the EQ, r = −.04; 50 ms r = .14; and 2000 ms, r = .07. Gender was unrelated to 50 ms r = −.08; 2000 ms r = −.13; and empathic concern, r = −.17; but was significantly related to the EQ, −.26, p = .002. Neither the 50 ms control task nor gender were significantly related to the emotion recognition tasks. Their inclusion did not change the main results, and therefore were not included as control variables. Table 1. Means and standard deviations for variables. Variables Means SD EQ total scores 31.39 9.31 EQ social skills 3.90 2.13 EQ emotional reactivity 4.6 2.00 EQ cognitive empathy 4.03 2.02 IRS-empathic concern 26.87 3.18 50 ms 27.77 4.24 2000 ms 34.00 3.75 Fear 50 3.24 1.47 Fear 2000 4.23 1.49 Acculturation 97.10 19.34 Table options Table 2. Bivariate correlations between study variables. 1 2 3 4 5 6 7 8 9 10 1. EQ total scores α = .83 .55 .64 .71 .19 .21 .17 .28 .05 .44 2. EQ social skills α = .62 .14 .26 .17 .30 .21 .25 .17 .12 3. EQ emotional reactivity α = .50 .36 .13 .19 .04 .24 .11 .58 4. EQ cognitive empathy α = .74 .14 .07 .12 .16 -.03 .17 5. 50 ms α = .68 .39 .54 .33 .27 .22 6. 2000 ms α = .64 .25 .56 .15 .13 7. Fear 50 α = .49 .52 .21 .10 8. Fear 2000 α = .59 .18 .07 9. Acculturation α = .81 -.02 10. IRS-empathic concern α = .61 Note. Correlations equal to or greater than r = .17 are significant at p < .05, using two-tailed tests. Table options Most variables were comprised of few items (5–7) and therefore had low internal reliabilities. Internal reliabilities were also somewhat low for overall expression recognition scores, likely because emotion types differ in recognition difficulty. 3.2. Analyses of main hypotheses The first hypothesis predicted the IRS-empathic concern subscale would be more related to recognition accuracy at 50 ms than the EQ scale. As shown in Table 3, this hypothesis was confirmed: EQ was significant at 50 ms, t(132) = 2.17, β = .18, p = .03, but this finding disappeared when empathic concern was entered into the equation and empathic concern became significant, t(131) = 2.04, β = .19, p = .04. A different pattern emerged at 2000 ms: EQ total scores were significant at 2000 ms, t(132) = 2.44, β = .21, p = .02, and empathic concern was not. EQ total scores remained significant when empathic concern was entered, t(131) = 1.95, β = .18, p = .05. Acculturation as a control variable was also significantly related at 50 ms, t(126) = 3.05, β = .27, p = .003, but was not at 2000 ms. Follow-up regression analyses of EQ factors showed that social skills were highly related at 2000 ms, t(130) = 3.17, β = .28, p = .002 and emotional reactivity was marginally related, t(130) = 1.89, β = .17, p = .06 (see Table 4). Table 3. Hierarchical regression analyses predicting overall accuracy to expressions at 50 ms and 2000 ms, with IRS and EQ as predictors. Predictors Overall accuracy 50 ms Overall accuracy 2000 ms β t p R2 β t p R2 Step 1 .11 .06 EQ total scores .18 2.17 .03 .21 2.44 .02 Step 2 .13 .06 EQ total scores .09 1.06 .29 .18 1.95 .05 IRS-empathic concern .19 2.04 .04 .05 .51 .61 Note. The control variable acculturation was entered at the first step, followed by the EQ. Coefficients are values at last entry. Table options Table 4. Hierarchical regression analyses predicting overall accuracy and accuracy to fear at 50 ms and 2000 ms, with EQ factors and empathic concern as predictors. Predictors Overall accuracy 50 ms Overall accuracy 2000 ms Fear accuracy 50 ms Fear accuracy 2000 ms β t p R2 β t p R2 β t p R2 β t p R2 EQ social skills .10 11.09 .28 .11 .28 33.17 .002 .12 .16 1.84 .07 .08 .19 2.24 .03 .12 EQ emotional reactivity .05 .51 .61 .17 11.89 .06 −.04 −.48 .63 .19 2.10 .04 EQ cognitive empathy .11 11.16 .25 −.07 −.72 .48 .10 1.01 .28 .05 .50 .62 Note. The control variable acculturation was entered at first, followed by the EQ factors. Coefficients are values at last entry. Table options The second hypothesis predicted empathic concern would be related to fear deficits, but the EQ would not be. Contrary to the hypothesis, the EQ was related and empathic concern was not. A median split was used to derive high and low EQ scores. A 2 (EQ scores) × 14 (emotions) MANCOVA was performed, controlling for acculturation. Emotion type was significantly related to EQ scores (Wilks’ Lambda = .78, F(14, 119) = 2.34, p = .007, partial η2=.22). Univariate analyses of emotion type showed that several emotions were significantly related to EQ scores, however, after adjusting the alpha level with a Bonferroni correction, only fear at 2000 ms remained significant, F(1, 132) = 15.90, p = .0001, partial η2 = .11 The next most significant emotions were disgust at 2000 ms F(1, 132) = 7.60, p = .007, partial η2 = .05 and anger at 50 ms, F(1, 132) = 5.26, p = .02, partial η2 = .04. As noted in Table 5, fear was not the most difficult expression for the sample to recognize, either at 50 ms or at 2000 ms. Table 5. Recognition accuracy means for expression type at 50 ms and 2000 ms. Exposure duration Anger Sad Disgust Fear Neutral Surprise Happy Brief 5.73 3.50 2.77 3.24 5.10 2.93 4.45 82% (0.63) 50% (1.32) 40% (1.36) 46% (1.47) 73% (1.00) 42% (1.50) 64% (1.30) Long 5.64 4.62 4.86 4.24 5.34 4.13 5.15 81% (0.58) 66% (1.19) 69% (1.09) 61% (1.50) 76% (.924) 59% (1.29) 74% (1.05) Note. Recognition accuracy means and percentages for emotion are the results for the entire sample. Mean values are number correct out of 7. Standard deviations are in parentheses. Table options A second 2 × 14 MANCOVA compared high and low IRS-empathic concern scores. The overall F test was not significant, Wilks’ Lambda = .90, F(14, 119) = .93, p = .529, partial η2 = .10. Of note, low internal reliabilities may have diminished the strength of the main findings: connections between 2000 ms and social skills; 2000 ms and emotional reactivity; 50 ms and empathic concern; and fear recognition and the EQ, may actually be stronger than this study indicates.