Empathy-related Responses 1

Research supports an “ingroup empathy hypothesis” of higher empathy-related psychophysiological responses towards individuals of the same ethnicity. However, little research has investigated empathy-related responses to non-human targets graded for phylogenetic relatedness. Participants (N = 73) were presented with film stimuli depicting humans, primates, quadruped mammals and birds in victimized circumstances. Phasic skin conductance responses (SCR) and subjective empathyrelated ratings to the film clips increased as phylogenetic similarity to humans increased across animal groups, revealing an empathic bias towards human stimuli. Participants also completed a trait empathy scale. High trait empathy participants gave higher subjective empathy ratings than moderate and low trait empathy participants. Low trait empathy participants showed less corrugator electromyographic activity than moderate and high empathy participants. The moderate trait empathy participants showed higher SCR than the high group. The results confirm an effect of phylogenetic similarity in subjective self-report and psychophysiological measures of empathy-related responses. Additionally, convergence between subjective and objective measures of empathyrelated responses was observed. Empathy-related responses 3 Empathy-related responses to moving film stimuli depicting human and nonhuman animal targets in negative circumstances. The link between inter-human and human-animal violence suggests that a lack of empathy demonstrated by some individuals is a general deficit not simply restricted to its expression towards other human beings (Ascione, 2001; Baldry, 2005; Beirne, 2004). Criminal record studies reveal that offenders with a history of animal abuse are also more likely to have a history of violent behavior (Merz-Perez, Heide, & Silverman, 2001) and concomitant anti-social behaviors, such as drug, public disorder and property offenses (Arluke, Levin, Luke, & Ascione, 1999). Given the growing volume of literature in this field, it appears that the amount of empathy shown towards animals may indicate a more general capacity for empathy and related prosocial behavior (Thompson & Gullone, 2003). Preliminary self-report empathy research currently supports a relationship between human-orientated and animal-orientated empathy (Paul, 2000). However, there is a need for improved objectivity in measures. Importantly, the theoretical link between inter-human empathy and empathy towards non-human animals highlights an opportunity for extended psychological research into empathic processes. Empathy is argued to be an inductive affective and cognitively evaluative process (Hoffman, 2007) affording individuals a vicarious experience of others feeling states that informs and augments subsequent responses. Empathy as a general construct, however, has been difficult to quantify due to variations in definition and measurement (Duan & Hill, 1996; Preston & de Waal, 2002). The distinction between affective and cognitive components of empathy, and the role each plays, poses a continuing problem for researchers and theorists (Chlopan, McCain, Carbonell, & Hagen, 1985; Davis, 1994; Duan & Hill, 1996; Eisenberg & Strayer, 1987; Preston & de Waal, 2002). Despite this challenge, it is accepted that empathy-related emotional responses influence Empathy-related responses 4 outcomes in both autonomic nervous system activity (Bradley & Lang, 2000; Levenson & Ruef, 1992) and overt behavior (Eisenberg et al., 1989) that are subject to affective intensity and cognitive accuracy (Plutchik, 1990). Further, sympathy and pro-social behavior may be facilitated by the meta-processes of empathy as a core motivational component (Eisenberg et al., 1989; Thompson & Gullone, 2003). The Perception-Action Model (PAM) of empathy (for full review, see Preston & de Waal, 2002) provides a broadly inclusive core model that affords a logical approach for the present research. The theoretical construct of “proximal” empathic processes posited by the PAM (considered biologically oriented and automatic in nature) suggest a focus on empathy-related “feeling states” as appropriate for some empathy investigations. Further support has been lent to this theory with the recent discovery of the mirror-neuron system and the role it may play in producing empathy-related responses (Carr, Iacoboni, Dubeau, & Mazziotta, 2003; Iacoboni & Dapretto, 2006). Additionally, the PAM of empathy outlines several other factors that may mediate empathy-related responding in people. One of these factors, “similarity”, appears to be highly relevant when examining empathy directed towards both humans and animals. Similarity influences empathic responding through the tendency of an observer to identify more closely with others who appear to be more similar to themselves in features such as personality (Gruen & Mendelsohn, 1986) and appearance (Brown, Bradley, & Lang, 2006). It may also apply to factors that increase the perception of similarity such as cultural likeness, sentience or social circumstance. Psychophysiological research focusing on inter-human empathic responding in a normal population has found support for a similarity effect (Brown et al., 2006) due to ingroup versus outgroup identification bias (Tajfel & Turner, 1979). Brown and colleagues (2006) showed African American and European American participants still pictures of Empathy-related responses 5 people of the same, or different, skin color. It was hypothesized that an “ingroup empathetic response” would be reflected in heightened skin conductance, increased startle blink magnitude, and increased corrugator supercilii muscle activity towards unpleasant same ethnicity pictures. Conversely, same ethnicity pleasant pictures should elicit heightened skin conductance and reduced startle blink magnitude and corrugator supercilli muscle activity. Consistent with the ingroup empathy hypothesis, African American participants showed increased corrugator activity to pictures of same ethnicity unpleasant pictures, while European American participants showed elevated skin conductance in response to same ethnicity pleasant and unpleasant pictures. Taken together, the results reported by Brown et al. (2006) provided some support for the conclusion that people show exaggerated affective (i.e., both pleasant and unpleasant) responses to more similar ingroup members than to less similar outgroup members. This finding implies that responses to aversive picture stimuli may be less affected by higher cognitive processing, but instead evoke the more affective proximal mechanisms of empathy-related responding as outlined in the PAM of empathy (Preston & de Waal, 2002). Similarity effects may also be present in psychophysiological measures when non-human animal stimuli are graded along a continuum for phylogenetic similarity and compared with human stimuli. For example, psychophysiological affective responses towards birds, quadruped mammals, primates (non-human bipeds) and humans should increase accordingly as a function of similarity, if the similarity theory holds across species. Hills (1995) investigated the relationship between empathy towards animals and a belief in animal mind (BAM), which refers to the belief that an animal is capable of thinking and feeling to some extent. She hypothesised a positive relationship based on perceived similarity through the process of identification, which is seen to contribute Empathy-related responses 6 to empathic feelings. Six different emotion-eliciting scenarios were administered via printed text to farmers, urban dwellers, and animal rights activists. Evidence emerged of a similarity effect that appeared related to the evolutionary distance between humans and other species. Animals that were closer phylogenetically to humans (mammals) were rated higher in BAM than animals that were phylogenetically more distant (coldblooded animals and invertebrates) (Hills, 1995). Further, perceived similarity contributed most towards empathy in the absence of instrumentality (i.e., the instrumental use of animals for human needs). For example, the farmers’ empathic responses were higher for scenarios involving rhinoceros poaching than they were for those involving chickens or sheep. The similarity hypothesis as applied to different animal species remains untested in a psychophysiological experimental context. The present research examined how indices related to human empathic responding vary across empathy eliciting scenarios depicting different groups of animals. Male and female participants were exposed to real-life film clips depicting victimized circumstances that were distressing, violent or oppressive. Moving film clips were used in order to maximize engagement with the targets and minimize ambiguity in facial expressions (Simons, Detenber, Reiss, & Shults, 2000; Simons, Detenber, Roedema, & Reiss, 1999; Weyers, Muhlberger, Hefele, & Pauli, 2006). The humananimal empathy link was tested using human stimuli alongside three classes of vertebrate non-human animals (birds, quadruped mammals, and primates), representing increasing degrees of similarity. It was hypothesized that a linear pattern of responses across increasing phylogenetic relatedness would be found for each of the subjective and psychophysiological measures of empathic responding, consistent with a similarity hypothesis. In addition, two groups were formed (companion and utilitarian) within the quadruped animal class to expose any possible effect of instrumentality. Dispositional Empathy-related responses 7 emotional empathic tendency was also investigated, consistent with Preston and de Waal’s (2002) interpretation of proximal empathic mechanisms. Individuals classified as high in trait emotional empathy were expected to show generally higher responsiveness on all measures of empathic-related responding than individuals classified as moderate in trait empathy, who in turn, were expected to show greater responses than individuals classified as low in trait empathy. Finally, relationships between subjective and psychophysiological measures were assessed by correlational analyses, as they were expected to provide some convergent validity. Experiment 1 Method Participants The sample of 73 participants (M = 22.5 years, SD = 6.9 years) consisted of 36 male and 37 female first year psychology and health science students over the age of 18 years who participated in return for partial course credit. All participants reported English as their first language and met pre-screening criteria for below moderate levels of depression, anxiety and stress according to the abbreviated version of the Depression Anxiety Stress Scale (DASS-21: Lovibond & Lovibond, 1995). Demographic characteristics reflected a relatively homogeneous sample with most participants reporting that they grew up in a semi-rural (30.1%) or suburban (49.3%) locale in contrast to a rural (15.1%) or urban (5.5%) locale. Similarly, most participants reported companion animal experience, either during childhood (27.4%) or adulthood only (5.5%), or throughout both childhood and adulthood (63%). Apparatus Participants completed the experiment individually in a sound attenuated room (Neumann & Waters, 2006). A Dell Optiplex Model GX270 computer presented the Empathy-related responses 8 task stimuli through a Panasonic Model PT-L557E LCD projector and recorded the subjective empathy ratings. Participants used the computer keyboard to enter subjective empathy ratings. A second Dell Optiplex Model GX270 computer interfaced with the first recorded physiological responses using Powerlab Model 4/20 (ADInstruments, Sydney) hardware and software and a 1000 Hz sampling rate. Skin conductance responses, respiration, and corrugator muscle activity were recorded continuously throughout the experiment. Skin conductance responses were measured by attaching ADInstruments MLT116F electrodes to the distal phalanges of the first and second fingers of participants’ non-preferred hand. The electrodes were connected to an ADInstruments Model ML116 GSR Amp. Respiratory influences on skin conductance were monitored via an ADInstruments Model MLT1132 Piezo Respiratory Belt Transducer placed around the participants’ lower chest. Corrugator responses were recorded using a pair of 4-mm-diameter Ag/AgCl domed electrodes filled with Surgicon E10 electrolyte paste. The electrodes were placed over the corrugator supercilii (for corrugator) muscle in accordance with established guidelines for psychophysiological research (Blumenthal et al., 2005; Fridlund & Cacioppo, 1986; Stern, Ray, & Quigley, 2001). The set of electrodes was connected to an ADInstruments Bioamplifier and readings were acquired using a band pass setting of 500 to 30 Hz. Trait empathy was measured using the Balanced Emotional Empathy Scale (BEES: Mehrabian, 1996) that utilized a nine point scale ranging from +4 to -4 (where 4 = “very strong agreement” and -4 = “very strong disagreement”) to assess participants’ level of agreement with statements of situations that are likely to elicit empathic emotions. Sample items from the scale include: “I cannot feel much sorrow for those who are responsible for their own misery” and “Unhappy movie endings haunt me for hours afterwards”. Empathy-related responses 9 The empathy-eliciting stimuli consisted of 10 s video vignettes of five animal groups (humans, primates, companion mammals, utilitarian mammals, and chickens) in distressing situations. Four clips for each of the five target groups, depicting as closely analogous situations as possible resulted in 20 clips. Each target group contained a scene showing a close up of a face (in pain), a scene of injury, a scene of confinement, and a scene of a victim being struck or roughly handled. These 60 cm high by 70 cm wide clips were projected onto a screen 2 m in front of participants. The clips were sourced from the internet and documentary video. From the same sources as the edited video clips, five color stills were produced and printed as 5 x 7 inch glossy photographs for use in the informed consent procedure. The images used as stills depicted analogous, but different, scenes to those in the main experimental stimuli. Five pleasant color pictures (1463, 1670, 811, 2091, and 1710) were sourced from the International Affective Picture System (IAPS: Center for the Study of Emotion and Attention (CSEA MIMH), 2001) and used as a visual debriefing tool in order to help ameliorate any lingering effects of viewing the potentially distressing images during the experiment. Procedure After providing informed consent, participants completed the DASS-21 and the BEES scales. Physiological recording sensors for corrugator EMG, skin conductance, and respiration were attached. After completing a three minute rest period, a definition of empathic ability was displayed as follows: “What is Empathy? It is the capacity to which one is able to objectively enter into another’s feelings and situation. This allows a vicarious experience and understanding of another’s predicament.” Participants were visually cued on the screen with the written instruction “Please make rating now” after each film clip presentation to rate their level of empathic feeling using a scale of 0 to 9 Empathy-related responses 10 (where 0 = none and 9 = maximal response) using a computer keyboard. Participants were instructed to separate any feelings of disgust from their ratings. Film clips were presented in two blocks with each block containing one set of the 20 clips (i.e., 4 different vignettes of each of the 5 stimulus groups). All stimuli were presented in randomized order within each block, with the nature of the animal type depicted in the first clip in each block counterbalanced across participants. After viewing and entering an empathy rating response, a 20 to 25 s intertrial-interval elapsed, in which the screen was blank. Upon experiment completion, participants were visually debriefed by asking them to spend a few minutes viewing the IAPS pictures. Data scoring Subjective empathy ratings were averaged across the four film clips presented in each block, separately for each animal stimulus group. An unanticipated programming error resulted in a systematic pattern of missing data in that some participants did not receive all trial presentations. In most cases, this resulted in some participants only receiving one presentation of one or more film clips. No participant was missing more than 8 of the 40 clips presentations. Due to the error, extra participants were recruited, and analyses were performed to assess whether the ratings differed between the first and second presentations of each film clip in participants with complete data sets. Repeated measures t-tests with Bonferroni corrected α = .0025 revealed no differences. This result is consistent with the research literature into the effects of repeated presentations of aversive/unpleasant picture stimuli (Bradley, Kolchakian, Cuthbert, & Lang, 1997). On this basis, missing ratings were replaced with those given for the same clips on a subsequent or previous presentation. This allowed means to be calculated that were weighted for each specific film clip within each animal type grouping while preserving individual integrity of participants’ scores. The same process was applied for the Empathy-related responses 11 psychophysiological data. In the final sample, 67% of participants had complete data sets. The remaining portion of participants had data sets with some replaced data, the majority (80.3%) of which had one (10.7%), two (34.8%), or four (34.8%) replaced values from the total of 40 presentations. Only seven participants (24.9%) had between six and eight values replaced. Each set of data was then collapsed into a mean score for that animal group. Corrugator EMG responses were scored as the difference between mean rectified corrugator supercilii signals present during the 10 s presentation of the film clip and the mean rectified signal level in the 1 s prior to film clip onset for non-probed vignettes. A positive change score indicates that corrugator activity was greater during the film clips than during the baseline period. A logarithmic transformation was applied [log10 (corrugator EMG + 3)] to normalize the positively skewed distributions prior to further analysis. Equipment failure resulted in missing corrugator EMG data for seven participants, thus, these participants were excluded from EMG analyses. The phasic skin conductance (SCR) response was scored as the magnitude of the distance between the trough and the peak of the curve to responses that began within 1 to 4 s after nonprobed film clip presentations. Skin conductance responses were subjected to a square root transformation [SQRT (SCR + 1)] to normalize the distributions before statistical analyses (Venables & Christie, 1980). Respiration artifacts were examined during the scoring of skin conductance to control for spurious effects of throat clearing and coughing that may interfere with data integrity. Skin conductance was scored as missing if gross artifacts were present (e.g., Neumann, Lipp, & McHugh, 2004). According to this criterion, missing scores were negligible (.1 %). Empathy-related responses 12 Design This experiment utilized a 3 x 5 (Empathy Group x Animal Type) mixed quasiexperimental factorial design. The between groups independent variable was Empathy Group (low: M = 10.75, SD = 7.13; moderate: M = 39.42, SD = 7.13; or high: M = 67.32, SD = 12.17) formed post hoc using a tertile division of BEES scores (M = 39.55, SD = 25.89, Min-Max = -20 – 89). The within groups independent variable was Animal Type (bird, utilitarian mammal, companion mammal, primate, and human). Dependent variables were subjective ratings of empathic feelings experienced, corrugator EMG level of change and phasic SCR. Preliminary comparisons to check for any effect of instrumentality on each dependent measure revealed no differences between companion mammals and utilitarian mammals (all p’s > .077). Thus, these groups were collapsed into one new variable, “quadruped mammals”, to simplify the statistical analyses. To determine the effects of empathy level (low, moderate, high) and type of animal stimulus (bird, quadruped mammal, primate or human) on the dependent measures, a series of 3 x 4 (Empathy Group x Animal Type) mixed factorial ANOVA’s were conducted. Huynh-Feldt corrections were applied where sphericity violations were present. All hypothesis tests were evaluated against an α-level of .05. Results Data Screening Correlations were calculated between BEES scores, DASS-21 subscale scores and absolute baseline skin conductivity in order to screen for any effect of sub-clinical influences of depression, anxiety, stress or physiological arousal, which may confound the interpretation of subsequent analyses. BEES scores were unrelated to levels of depression, anxiety, stress, or to baseline skin conductivity, all r’s < ± .14, all p’s > .23. Empathy-related responses 13 Effects of Empathy Group Membership and Animal Stimulus Type Subjective empathy ratings. As shown in Figure 1, a main effect was present for animal stimulus type on subjective empathy ratings, F(3, 182) = 25.62, p < .0005, η2P = .27, and consistent with prediction, an increasing linear trend in subjective empathy ratings as animal groups became more similar to humans was supported, F (1, 70) = 29.49, p < .0005, η2P = .3. Subjective empathy ratings also differed between the low, moderate, and high empathy groups, as confirmed by a significant main effect for empathy group, F(2, 70) = 13.94, p < .0005, η2P = .29. Inspection of the 95% confidence intervals (CI) indicated that the high empathy group (95% CI: 6.19 < M = 6.7 < 7.26) made significantly higher ratings than both the moderate (95% CI: 4.99 < M = 5.52 < 6.06) and low groups (95% CI: 4.2 < M = 4.74 < 5.28), which did not significantly differ. ---------------------------Insert Figure 1 ----------------------------------Corrugator EMG activity. Figure 2 shows that corrugator EMG activity significantly differed across the three empathy groups, F(2, 64) = 3.46, p = .04, η2P = .1. Subsequent inspection of 95% CI’s revealed that the low empathy group (95% CI: .55 < M = .62 < .69) exhibited significantly less corrugator activity than the high group (95% CI: .67 < M = .74 < .81), but not the moderate group (95% CI: .65 < M = .72 < .8). Although an Empathy Group x Animal Type interaction was found, F(5, 147) = 2.8, p = .02, η2P = .08, no main effect of animal type was evident in corrugator activity, F(2, 147) = 2.54, p = .07, η2P = 0.04. As suggested graphically, simple effects analysis confirmed that the only animal group upon which no differences were found between the empathy groups was the primate group, F(2, 64) = 2.34, p = .1. ---------------------------Insert Figure 2 ----------------------------------Empathy-related responses 14 Phasic skin conductance response. As can be seen in Figure 3, there was a main effect for animal stimulus type on SCRs, F(3, 187) = 11.88, p < .0005, η2P = .15. Consistent with the hypothesis for a linear trend, higher SCRs were evident as phylogenetic similarity to humans increased, F (1, 70) = 19.05, p < .0005, η2P = .21. However, an unexpected pattern emerged with this measure in that responses tended to be similar in the low and high empathy groups and largest in the moderate empathy group. A main effect of empathy group, F(2, 70) = 4.6, p = .01, η2P = .12, confirmed this impression. The Games-Howell (Games & Howell, 1976) post hoc comparisons for groups with unequal variance indicated that no significant difference existed between the high and low groups (95% CI(diff): -.14 < M(diff) = .04 < .06) or the moderate and low groups (95% CI(diff): -.02 < M(diff) = .01 < .21). However, the moderate group showed significantly larger responses than the high empathy group (95% CI(diff): .02 < M(diff) = .13 < .25). ---------------------------Insert Figure 3 ----------------------------------Relationships Between Measures. BEES scores were positively correlated with overall subjective empathy ratings (r = .6, p < .0005) and corrugator activity (r = .35, p = .004) as expected, however no relationship was evident with SCR (r = -.19, p = .11). Additionally, subjective empathy ratings were positively correlated with corrugator activity (r = .41, p = .001). BEES scores, subjective empathy ratings and corrugator activity thus converged meaningfully. Discussion The present study examined subjective self-reported empathy ratings, corrugator EMG activity and phasic SCRs during film clips depicting humans and non-humans in oppressive circumstances. Significant linear trends indicating higher subjective empathy ratings of empathy and larger SCRs were found as the stimuli became closer in Empathy-related responses 15 phylogenetic relatedness to humans. The effect of animal type interacted with the trait empathy group for corrugator EMG activity. Taken together, the results suggest that empathic responding towards humans is generalized to other species. The greater the similarity of the species towards humans, the larger is the empathic response. The findings support the notion that there is a relationship between human empathy directed towards other humans and human empathy directed towards non-human animals (Paul, 2000). In the subjective empathy measure, it remains a possibility that demand characteristics influenced the ratings that were given and that these demand characteristics were not limited to the human stimuli. The finding of a similar linear trend across animal groups with the objective psychophysiological measure of SCRs, however, renders this explanation unlikely. The subjective and psychophysiological measures during the film stimuli were also related to the level of trait empathy obtained from the BEES (Mehrabian, 1996). Subjective empathy ratings were higher and corrugator activity was greater for individuals classified as high in trait empathy than in individuals classified as low. In both cases, individuals with moderate levels of trait empathy yielded responses that were in between the high and low groups. The results are consistent with prior research findings showing increased corrugator activation to unpleasant stimuli (Dimberg, 1982) and with the complex nature of empathy related responding to unpleasant stimuli (Brown et al., 2006). Taken together with the subjective trait measures of empathy, the results suggest that the high empathy group was consistently more responsive to the film clips than the low empathy group. Since human film clips only accounted for one fifth of the clips in the present study, it is difficult to attribute the difference between empathy groups as reflecting only the human portion of stimuli. The relationship Empathy-related responses 16 between trait empathy and subjective and corrugator EMG responses across the stimuli further suggests that the human empathic response is not exclusively anthropocentric. An exception to the general pattern of higher corrugator EMG activity for the high trait empathy group was found for the primate film stimuli, for which the low, moderate and high empathy groups exhibited a marked similarity in corrugator EMG responses. This suggests the presence of some qualitative difference in the primate stimuli that elicited unexpected and uniform responses across all participants. Frowning is a facial activity that is associated with responses to unpleasant stimuli and outwardly expressed disgust, anger, distress and concerned attention (Bradley, Codispoti, Cuthbert, & Lang, 2001; Dimberg, 1990; Eisenberg et al., 1994). It could be the case that the corrugator EMG activity during the primate stimuli was reflecting one of these psychological states that shows some independence from empathy. The primate film clips consisted largely of scenes from vivisection experiments, as these were the only stimuli that could be ethically sourced. Comments and questions by participants during debriefing suggested some degree of curiosity or uncertainty associated with the primate stimuli. For instance, some participants asked questions such as “What was that on the monkey’s head?” If the primate stimuli were less clear or more confusing than the other stimuli, they may have elicited a corrugator EMG response that was uniform across trait empathy groups due to frowning, rather than an emotional response related to empathy. Moreover, low corrugator activity for the most part in the low empathy group may represent a categorical difference in affective valence in that people low in trait empathy may not experience stimuli in the same way as those who are moderate or high in empathy. Thus, if the low empathy group construe stimuli that is distressing to others as simply neutral instead, then the same activation of corrugator activity would not be likely to occur. While EMG data are continuous by their very nature, further Empathy-related responses 17 interpretation may include corrugator as an indicator of a shift in valence (in this case from negative to neutral) (e.g., Bradley et al., 2001; Davis et al., 1995; Lang, Greenwald, Bradley, & Hamm, 1993). Video-recordings of participant facial responses will help clarify the nature of the corrugator response in future research, although subtle expressions may be harder to detect with the naked eye than with electrophysiological recordings. Skin conductance responses reflect sympathetic arousal which is generally interpreted as an orienting response to salient or novel stimuli (Dawson, Schell, & Filion, 2000) and can indicate emotional arousal, independent of stimulus valence (Lang et al., 1993). The SCR responses were greater in the moderate empathy group, than in the high empathy groups. Moreover, the difference in skin conductance activity between the moderate and low empathy groups approached significance. The pattern across empathy groups is unexpected because those higher in emotional trait empathy would be expected to be more affectively reactive to, and oriented towards, empathy eliciting scenes. However, it cannot be ruled out that those in the high empathy group may have diverted their gaze away from the film clip to avoid distress, resulting in the expected corrugator activity, but not the anticipated skin conductance reactivity. Future research could employ video recordings to rule out this possibility. Alternatively, the recording of heart rate may be used to determine whether the film clip elicited an acceleratory response that is characteristic of a defensive reaction (Ekman, Levenson, & Friesen,