Contextual determinants of pain judgments

Transcription

1 Pain 139 (2008) Contextual determinants of pain judgments M.O. Martel a, P. Thibault b, C. Roy c, R. Catchlove d, M.J.L. Sullivan a, * a Department of Psychology, McGill University, 1205 Docteur Penfield, Montréal, Que., Canada H3A 1B1 b Department of Psychology, McGill University, 1205 Docteur Penfield, Montréal, Que., Canada, H3A 1B1 c Department of Psychology, Université de Montréal, C.P Succ Centre Ville, Montréal, Que., H3C 3J7 d Department of Anesthesiology, McGill University, 1205 Docteur Penfield, Montréal, Que., Canada H3A 1B1 Received 17 December 2007; received in revised form 29 May 2008; accepted 5 June 2008 Abstract The objective of this study was to examine the influence of variations in contextual features of a physically demanding lifting task on the judgments of others pain. Healthy undergraduates (n = 98) were asked to estimate the pain experience of chronic pain patients who were filmed while lifting canisters at different distances from their body. Of interest was whether contextual information (i.e., lifting posture) contributed to pain estimates beyond the variance accounted for by pain behavior. Results indicated that the judgments of others pain varied significantly as a function of the contextual features of the pain-eliciting task; observers estimated significantly more pain when watching patients lifting canisters positioned further away from the body than canisters closest from the body. Canister position contributed significant unique variance to the prediction of pain estimates even after controlling for observers use of pain behavior as a basis of pain estimates. Correlational analyses revealed that greater use of the contextual features when judging others pain was related to a lower discrepancy (higher accuracy) between estimated and self-reported pain ratings. Results also indicated that observers level of catastrophizing was associated with more accurate pain estimates. The results of a regression analysis further showed that observers level of catastrophizing contributed to the prediction of the accuracy of pain estimates over and above the variance accounted for by the utilisation of contextual features. Discussion addresses the processes that might underlie the utilisation of contextual features of a pain-eliciting task when estimating others pain. Ó 2008 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. Keywords: Context; Pain judgments; Accuracy; Pain behavior; Catastrophizing 1. Introduction * Corresponding author. Tel.: ; fax: address: michael.sullivan@mcgill.ca (M.J.L. Sullivan). Hadjistavropoulos and Craig [9] have advanced a conceptual model that addresses the processes by which pain is communicated and perceived by others. According to this model, pain experience is communicated through various behavioral displays that are interpreted and decoded by observers. The communications model of pain [9] proposes that sender and observer characteristics as well as contextual factors may affect the manner in which the pain of others is perceived. Research has elucidated several sender and observer characteristics that may influence the estimates of others pain. For example, it has been shown that the sex of the sender as well as the sender s pain behaviors account for significant variance in pain estimates [24,25]. In turn, the sex of the observer [21] and the observers level of catastrophizing [25] have also been shown to influence pain estimates. Little is currently known about the role of contextual information in the process of estimating others pain. In the bulk of studies conducted to date, observers have been asked to make the estimates of others pain /$34.00 Ó 2008 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi: /j.pain

2 M.O. Martel et al. / Pain 139 (2008) based solely on others pain behavior [15,17,19]. Although numerous studies have revealed that observers are sensitive to variations in others pain displays [13 15], findings have also revealed that observers are generally inaccurate in their pain inferences [19,15]. It is possible that the accuracy of pain inferences might be enhanced through the provision of pain-relevant contextual information. Contextual information might function as a heuristic device guiding observers pain inferences. The operation of heuristic-based inferential processes is supported by findings suggesting that observers use gender-related heuristics (e.g., stereotypes) for estimating others emotions [8], including pain [21]. Previous research on heuristic-based judgments suggests that the use of heuristics can either increase or decrease judgmental accuracy [21,5]. The relative weight given to behavioral information and judgmental heuristics in drawing inferences about others pain might be influenced, at least in part, by the characteristics of observers. For example, Sullivan et al. [25] have shown that catastrophic thinking leads to heightened pain estimates through an increased propensity to rely on pain behaviors. Given that clinicians inferences about patients pain level forms the basis of clinical treatment decisions, it is important to investigate further the manner in which contextual features and observer characteristics summate or interact in the process of estimating others pain. In this study, participants were asked to estimate the pain levels of chronic pain patients who were filmed while performing a physically demanding lifting task. The contextual features of the lifting task were varied by manipulating the postural lifting positions of the objects lifted. Variations in postural position were intended to activate the judgmental heuristic that objects further away from the body are more difficult to lift than objects closer to the body. Of interest was whether contextual information (i.e., lifting posture) contributed to pain estimates beyond the variance accounted for by pain behavior. It was also of interest to address whether the utilisation of contextual information influenced the accuracy of pain estimates. Secondary analyses examined the role of observers level of pain catastrophizing as a determinant of pain estimates. 2. Methods 2.1. Participants Ninety-eight (19 men, 79 women) undergraduate psychology students from the University of Montreal volunteered to participate in this study. Participants ranged in age from 18 to 42 years (mean = 21.8; SD = 3.8) Stimuli and Measures Stimuli: Video depictions of patients with back pain Participants viewed videotapes of 20 (12 men, 8 women) patients with persistent back pain lifting a series of canisters. The patients depicted in the video sequences had previously participated in a study where they were asked to rate their pain while lifting canisters at different distances from their body [26]. A set of 80 video sequences of 5 s each in duration were used as stimuli. Participants viewed four different video sequences of the same patient. Patients were shown lifting canisters with their elbow positioned at a 90 degree angle (easy) or with arm fully extended (difficult). Each video sequence consisted of a frontal view of the patient s face, trunk and upper extremities as well as the table and the canisters being lifted by the patients. Each 5-s video sequence began prior to the lifting of a canister and ended after the canister was replaced on the table. Canister position was conceived as a way to address the role of heuristic-based contextual features (e.g., belief that objects further away from the body will be more difficult to lift) Stimuli: pain behavior indices Each video sequence was coded for pain behavior by two independent judges based on a procedure described by Sullivan et al. [26]. Judges were trained to use a pain behavior coding system developed in our laboratory and adapted to the lifting task. For each video sequence, judges recorded the occurrence of the following pain behaviors: (1) facial expressions such as grimacing or wincing and (2) bodily movements such as guarding, holding, touching or rubbing. Judges recorded the duration (in seconds) of pain behaviors and provided intensity ratings on a three-point scale with the following anchors (1) mild, (2) moderate and (3) intense. The means of the two judges ratings for pain behavior intensity and pain behavior duration were used in the analyses. For each video sequence, the intensity of each behavior was multiplied by its duration [18]. These products were summed and provided a composite index of total pain behavior. Percentage agreement for the classification of different pain behaviors was computed for each video sequence. Indices of inter-rater agreement were derived by averaging percentage agreement scores for facial expressions and bodily movements. Percentage agreement for facial expressions was 94% (range %) and 93% (range of %) for bodily movements. Percentage agreement for the ratings of pain behavior intensity was 92% for facial expressions and 96% for bodily movements. Correlations between the two coders ratings of pain behavior duration were.95 and.96 for facial expressions and bodily movements, respectively.

3 564 M.O. Martel et al. / Pain 139 (2008) Catastrophizing The pain catastrophizing scale [23] was used as a measure of catastrophic thinking about pain. The PCS contains 13 items describing different thoughts and feelings that individuals may experience when they are in pain. Participants were asked to reflect on past painful experiences, and to indicate the degree to which they experienced each of 13 thoughts or feelings when experiencing pain, on a 5-point scale from (0) not at all to (4) all the time. The PCS has been shown to have high internal consistency (coefficient alpha: total PCS =.87) [23] Estimates of pain in others Participants were provided with a response sheet on which they were asked to estimate the level of pain experienced by the patients presented in the video sequences. Participants estimated the pain of patients depicted in the video sequences on an 11-point scale with the endpoints (0) no pain and (10) extreme pain Procedure Participants were informed that the research was concerned with the accuracy with which others pain can be estimated. At the beginning of the testing session, participants completed the PCS. Participants were then told that they would be watching video sequences of patients with persistent back pain undergoing a physical assessment. The experimenter explained to each participant that his or her task was to estimate as accurately as possible the level of pain experienced by the patient depicted in the video sequence using the 0 10 scale that appeared on the response sheet. The 80 video sequences were presented continuously in a random order on a projector screen (4 5 ft) with the constraint that no two sequences of the same patient appeared in immediate succession. The same random sequence was used for all participants. Following each 5-s video sequence, 4 s of blank screen was inserted to allow the participant sufficient time to estimate the pain of the patient depicted in the video sequence Data reduction and data analytic approach Stimulus-based rating index In order to examine the influence of patients pain behavior on observers pain judgments, within-subject correlations were computed between each observers pain estimates and pain behavior score derived for each of the 80 stimuli. The covariation between observers pain estimates and patients pain behavior score reflects the degree to which observers used stimulus pain behavior as a basis of their pain estimates. The stimulus-based rating index was computed for total pain behavior Accuracy of pain judgment The accuracy of pain estimates was determined by computing the absolute difference between observers pain estimates and the self-reported pain ratings given by the patients depicted in the video. Higher values represent either under or over estimation (lower accuracy) of others pain due to greater discrepancy between estimated and self-reported pain ratings. A position heuristic index was calculated by subtracting observers pain estimates for canisters in armextended position (difficult) from pain estimates made for canisters lifted with the elbow at a 90 angle (easy). Higher values on the position index represent observers greater use of the contextual features of the lifting task to estimate others pain. A sex heuristic index was computed by subtracting mean pain estimates provided for female stimuli from mean pain estimates provided for male stimuli. A within-subjects analysis of variance was used to examine whether observers pain estimates varied as a function of the contextual features of the lifting task (i.e., canister position) and the sex of the patients depicted in the video sequences. A mixed analysis of variance was used to address whether pain estimates also varied as a function of observers sex and level of pain catastrophizing. Analyses of covariance were used to address whether position and sex heuristic indices contributed to pain estimates independent of pain patient s pain behavior. Correlations were computed to examine the relation between the accuracy of pain estimates and the degree to which observers invoked judgmental heuristics (i.e., canister position, sex of patient) to estimate others pain. A hierarchical multiple regression was computed to examine the joint influence of observers catastrophizing and the use of judgmental heuristics on the accuracy of pain estimates. 3. Results 3.1. Stimulus characteristics Table 1 presents the means and standard deviations for patients (i.e., stimuli) demographics and pain condition characteristics. Men and women depicted in the video sequences did not differ significantly in age, t(18) =.07, ns, pain duration (months), t(18) = 1.7, ns, or pain severity, t(18) = 1.5, ns. The self-reported pain ratings of the patients depicted in the video sequences were analysed using a two-way (canister position stimulus sex) mixed ANOVA. Analyses revealed a significant main effect for canister position, F(1,18) = 10.33, p <.01. No main or interaction effects were found for the sex of patient.

4 M.O. Martel et al. / Pain 139 (2008) Table 1 Stimulus characteristics: pain patients depicted in the video sequences Men (n = 12) Women (n =8) p Age (years) 41.2 (8.8) 40.9 (10.3) ns Injury site 100% back 100% back ns Pain duration (years) 9.9 (7.8) 4.5 (6.0) ns MPQ-PRI 19.6 (11.0) (15.2) ns Note: Numbers in parentheses are standard deviations. MPQ-PRI, McGill Pain Questionnaire Pain-Rating Index Judging others pain: The influence of contextual factors A two-way (canister position stimulus sex) repeated analysis of variance (ANOVA) was conducted on observers pain estimates. Means and standard deviations are presented in Table 2. The analysis revealed a significant main effect for canister position, F(1, 97) = , p <.001, indicating that higher pain estimates were made when observing patients lifting canisters further away from the body (arm extended). Measure of effect size for canister position was calculated using partial eta square value (N p 2 ) and was.75. Consistent with previous research [21], the analysis revealed a significant main effect of stimulus sex, F(1, 97) = 16.22, p <.001, where observers rated the pain of women to be more intense than the pain of men. A significant canister position stimulus sex interaction was also found, F(1, 97) = 31.75, p <.001, indicating that observers rated the pain of women to be significantly more intense than the pain of men when observing patients lifting canisters that were further away from the body (arm extended). Analysis of covariance was used to address whether the influence of sex of the patient and the contextual features of the lifting task (i.e., canister position) on observers pain estimates could be accounted for by variations in pain behavior. An ANCOVA was performed on observers pain estimates, using the stimulus-based rating index as the covariate. ANCOVA results indicated that the main effect of canister position remained significant even when controlling for the stimulus-based rating index, F(1, 96) = 21.23, p <.001. The main effect of stimulus sex also remained significant even when controlling for the stimulus-based rating index, F(1, 96) = Table 2 Observers pain estimates as a function of canister position and stimulus sex Stimulus sex Easy Difficult Men 2 (1.2) 2.8 (1.2) Women 3.5 (1.3) 4.9 (1.6) Note: Numbers in parentheses are standard deviations. Easy refers to the canister position closest to the patient s body. Difficult refers to the canister position furthest away from the patient s body , p <.005. Partial eta square values were.18 and.10 for canister position and stimulus sex main effects, respectively. These results indicate that 18% of unique variance in pain estimates was accounted for by canister position, and 10% of unique variance in pain estimates was accounted for by the sex of the patients. In other words, invoking judgmental heuristics about the difficulty of the lifting task and the sex of the patients contributed to observers pain estimates, independent of the variance accounted for by patients pain behavior Perception of others pain: the role of observer characteristics A three-way (observers level of catastrophizing observers sex canister position) mixed analysis of variance (ANOVA) was performed on pain estimates. The results of this analysis revealed a significant main effect for observers level of catastrophizing, F(1, 94) = 6.25, p <.05, where high catastrophizers estimated more intense pain in others (M = 3.6, SD = 1.3) than low catastrophizers (M = 2.8, SD = 1.8) (see Table 3). There were no interaction effects involving observers level of catastrophizing. The sex of observer did not reach significance in any main effect or interaction. Partial eta squared value for the main effect of catastrophizing was.06, indicating that 6% of the variance in pain estimates was accounted for by observers level of catastrophizing. Correlational analyses revealed that the level of catastrophizing was not significantly correlated with the stimulus-based rating index, r =.01, ns. In other words, higher levels of catastrophizing were not related to the degree to which observers relied on pain behavior to make their estimates Accuracy of pain judgments The accuracy of pain judgments was assessed by computing the absolute difference between observers pain estimates and the self-reported pain ratings given by Table 3 Pain estimates as a function of observers level of catastrophizing, observers sex and canister position Easy Difficult Low catastrophizers Men 2.0 (1.1) 2.9 (1.4) Women 2.5 (1.0) 3.4 (1.2) High catastrophizers Men 2.9 (0.4) 3.9 (0.6) Women 3.2 (1.4) 4.0 (1.4) Note: Numbers in parentheses are standard deviations. Easy refers to the canister position closest to the patient s body. Difficult refers to the canister position furthest away from the patient s body.

5 566 M.O. Martel et al. / Pain 139 (2008) the patients depicted in the video sequences. Correlational analyses were then computed to examine the relation between the accuracy of pain estimates and the degree to which observers invoked judgmental heuristics (i.e., position heuristic index, sex heuristic index) to estimate others pain. Results indicated a significant association between the accuracy of pain estimates and the position heuristic index, r =.30, p < In other words, greater use of canister position when judging others pain was related to a lower discrepancy (higher accuracy) between estimated and self-reported pain ratings. The correlation between the sex heuristic index and the accuracy of pain estimates was not significant, r =.11, ns. Higher levels of pain catastrophizing were associated with a lower discrepancy (higher accuracy) between estimated and self-reported pain ratings, r =.26, p < As in the previous research, high catastrophizers increased accuracy was due to their tendency to underestimate others pain less than low catastrophizers [25]. Observers level of catastrophizing was not associated with the position heuristic index, r =.05, ns, nor with the sex heuristic index, r =.06, ns. A hierarchical multiple regression was computed to examine the joint influence of observers catastrophizing and the use of judgmental heuristics on the accuracy of pain estimates (see Table 4). The sex heuristic index was not considered in this analysis given that it was not significantly correlated with inferential accuracy. The position heuristic index was entered in the first step of the analysis and contributed significant variance in the prediction of the accuracy of pain estimates, R =.30, F(1, 96) = 9.2, p <.01. The PCS total score was entered in the final step of the analysis and contributed significant unique variance in the prediction of accuracy of pain estimates, R 2 change =.06, F(1,95) = 6.5, p <.05. These findings suggest that observers level of catastrophizing still explained 6% of the variance in the accuracy of pain estimates, even when controlling for the use of contextual features. Table 4 Regression analysis examining the influence of observers catastrophizing and the use of judgmental heuristics on the accuracy of pain estimates Dependent variable = accuracy of pain estimates Variable B R r Step 1 Position heuristic index.28 ** ** Step 2 PCS total.24 * * Note: B, standardized beta weight; R, multiple R; r, zero order correlation. Beta weights are from the final regression equation. * p <.05, ** p < Discussion The primary objective of this study was to investigate the effects of variations in contextual features of a lifting task on judgments of others pain. Consistent with previous research, the findings revealed that pain estimates were influenced by the sex of the person exhibiting pain behavior [21,24,25] and by the level of catastrophizing of the observer [25]. The results of the present research extend previous findings showing that pain estimates are influenced by variations in the contextual features of a pain-eliciting task; observers estimated significantly more pain when watching patients lifting canisters positioned further away from the body than canisters closest from the body. The findings also revealed that the use of judgmental heuristics (e.g., canister position, sex of patient) for estimating pain is independent of patients pain behavior, and independent of the observers sex or level of catastrophizing. In other domains of research on social perception, it has been suggested that observers attempt to simplify the burden of inferential processing by relying on judgmental heuristics [1,5]. Gender stereotype is one such heuristic that has been previously examined [8,11]. For example, gender stereotypes have been shown to influence inferences of others emotional experience, as women are expected to feel emotions more intensely than men [8,11]. The results of this study show that the sex of the patient contributes to pain estimates even when controlling for patients pain behavior. In others words, gender stereotypes (e.g., women feel more pain than men) appear to operate independently from behavioral information displayed by the patients. Observers also appear to have invoked a position heuristic for estimating others pain. Previous research suggests that observers a priori beliefs concerning the painfulness of a pain-eliciting task might influence pain judgments [2]. Here, beliefs that objects further away from the body will be more difficult to lift than objects closest from the body led to estimates of more intense pain, even though variations in canister position were not necessarily reflected in pain behavior. These results suggest that over and above a person s behavior, information about postural position influences pain estimates. It is interesting to speculate that even without pain behavior displays, a physical task involving postural positions likely to exacerbate back pain might be enough to elicit responses in observers such as empathy, support or physical assistance [3,7]. Neurophysiological approaches to social perception also provide useful insight into the mechanisms by which individuals might interpret physical movements executed by others [10]. It has been proposed that a mirror neuron system enables observers to understand the actions of others, as well the perceivable consequences associated with these actions [6,20]. In the

6 M.O. Martel et al. / Pain 139 (2008) context of this study, the latter explanation would suggest that observers were equipped to understand that patients execution of specific movements (i.e., flexion to reach canisters) might be accompanied by increases in pain levels, thus leading to higher pain intensity judgments. The activation of a mirror neuron system remains a speculative process that might, in part, underlie judgments of others pain when observers are exposed to variations in others movements (i.e., different postural lifting positions). At present, it is not clear whether heuristic-based processes and mirror neuron-related processes influence pain estimates through similar or distinct pathways. In line with recent work [25], observers with high levels of catastrophizing estimated significantly more intense pain in others than observers with low levels of catastrophizing. However, observers level of catastrophizing did not interact with the heuristic processes (i.e., contextual features of the lifting task or the sex of the patient), and was not associated with an increased tendency to rely on patients pain behaviors to estimate pain. While high catastrophizers appear to see more pain in others, they do not appear to be more sensitive to subtle variations in others pain behavior. As shown in previous studies, sender s pain behavior accounted for significant variance in observers pain estimates [24,25]. Pain behavior displayed by the patients depicted in the video sequences explained 9% of the variance in observers pain estimates. In previous research, however, pain behavior has been shown to explain up to 36% of the variance in observers pain judgments [25]. It is possible that the availability of contextual information might have reduced observers reliance on pain behavior as the basis of their pain estimates. When asked to estimate others pain, participants might have used the contextual features of the lifting task to re-interpret behavioral information displayed by the patients [4,12]. Contextual factors might also have led some observers to discount the information value of pain behavior. For example, high pain behavior in low physical demands conditions might have led individuals to consider pain behavior as an unreliable index of actual pain experience. The results of this study are consistent with previous research showing marked inaccuracy in observers pain judgments [19,22]. Analyses examined the degree to which the utilisation of judgmental heuristics (i.e., position heuristic, sex heuristic) could increase the accuracy of pain estimates. Overall, greater use of the position heuristic was associated with less discrepancy between estimated and self-reported pain ratings. The use of the sex heuristic was not associated with more accurate pain estimates. In others words, invoking gender stereotypes as a basis for judging pain will not necessarily yield more accurate pain estimates. However, the results suggest that using the contextual features of the pain-eliciting task for judging pain might represent a judgmental strategy that increases the accuracy of pain judgments. Interestingly, higher levels of catastrophic thinking were significantly correlated with the accuracy of pain estimates, but not with a greater use of the contextual features of the lifting task. In addition, the results of a regression analysis indicated that observers catastrophizing accounted for significant unique variance in the accuracy of pain estimates, even after controlling for observers use of the contextual features to estimate pain. The present results support recent work [25] suggesting that high catastrophizers greater accuracy when estimating others pain might not be due to greater integration of the pain task s contextual features into their pain judgment, but rather due to a systematic tendency to underestimate others pain less than observers with low levels of pain catastrophizing. Catastrophizing might be associated with higher levels of empathy and/ or by greater reactivity to pain stimuli (even others pain) [25], leading to higher pain estimates and reducing the underestimation bias. There are limitations to the current study that must be considered in the interpretation of the findings. The high number of video sequences used in the study (80 5-s) might have reduced the degree of attention that participants devoted to the stimuli in making their pain estimates. Previous research has shown that judgments of pain in others can be altered by repeated exposure to pain stimuli [18]. In addition, the use of a student sample and the low representation of men in the study sample limit the generalizabillity of the findings. Despite the above limitations, the present research provides insights into the manner in which judgmental heuristics might influence pain estimates. As highlighted by Hadjistavropoulos and Craig [9], the task of estimating others pain is complex, involving consideration of pain behavior, the context within which pain behavior is expressed, the consistency of pain behavior, the pain stimulus, and various characteristics of the sender and the observer. The findings of this study suggest that certain stimulus features might activate judgmental heuristics that will be brought to bear when inferring another person s pain. These judgmental heuristics appear to operate independent of the pain behavior expressed by the patient. Some judgmental heuristics appear to increase accuracy of pain estimates (e.g., postural position), while other heuristics (e.g., sex of patient) do not contribute to greater accuracy. Observer characteristics (e.g., catastrophizing) also influence pain estimates, but again, in a manner unrelated to sensitivity to variations in pain behavior, or the tendency to invoke judgmental heuristics. Future research on the different judgmental heuristics brought to bear in pain estimation will be required to bring greater specificity to conceptual models of pain communication. Although pain communication models

7 568 M.O. Martel et al. / Pain 139 (2008) have been proposed [9,16], the manner in which different sender, observer and situation characteristics summate or interact has not been specified. Advances in this domain will inform not only psychological models of pain inference, but might also have bearing on the elaboration of models that address links between psychological and physiological processes in responses to others pain [7]. Beyond contributing to theoretical refinement, the ultimate goal of this research is to contribute to more effective management of the challenges facing patients with pain-related disorders. Since assessment of back pain in clinical settings is often performed through observational method using standardized protocols similar to those used in this study [27], these findings could have important clinical implications if replicated with a sample of health professionals. Given that the accuracy of pain estimates is critical to the effective treatment of pain conditions, more research is warranted on the nature and manner in which heuristics are brought to bear in making estimates about others pain. Acknowledgements The authors thank André Savard and Dorothée Ialongo-Lambin for their assistance in data collection. This research was supported by grants from the Canadian Institutes for Health Research and the Social Sciences and Humanities Research Council of Canada. None of the authors has any financial interests related to the material discussed in this paper. References [1] Beike DR, Sherman SJ. Social inference: inductions, deductions, and analogies. In: Wyer RS, Srull TK, editors. Handbook of social cognition. Hillsdale: Erlbaum; p [2] Breau LM, McGrath PJ, Stevens B, Beyene J, Camfield CS, Finley GA, et al. Healthcare professionals perceptions of pain in infants at risk for neurological impairment. BMC Pediatr 2004;4:23. [3] Deyo KS, Prkachin KM, Mercer SR. Development of sensitivity to facial expression of pain. Pain 2004;107: [4] Ekman P, O Sullivan M. The role of context in interpreting facial expression: comment on Russell and Fehr J Exp Psychol Gen 1988;117:86 8. [5] Fiske ST, Taylor SE. Social cognition. New York: Random House; [6] Gallese V, Fadiga L, Fogassi L, Rizzolatti G. Action recognition in the premotor cortex. Brain 1996;119: [7] Goubert L, Craig KD, Vervoort T, Morley S, Sullivan MJL, Williams AC, et al. Facing others in pain: the effects of empathy. Pain 2005;118: [8] Grossman M, Wood W. Sex differences in intensity of emotional expression: a social role interpretation. J Pers Soc Psychol 1993;65: [9] Hadjistavropoulos T, Craig KD. A theoretical framework for understanding self-report and observational measures of pain: a communications model. Behav Res Ther 2002;40: [10] Jackson PL, Decety J. Motor cognition: a new paradigm to study self-other interactions. Curr Opin Neurobiol 2004;14: [11] Johnson JT, Shulman GA. More alike than meets the eye. Perceived gender differences in subjective experience and its display. Sex Roles 1988;19: [12] Kappesser J, Williams AC. Pain and negative emotions in the face: judgements by health care professionals. Pain 2002;99: [13] Poole GD, Craig KD. Judgments of genuine, suppressed and faked facial expressions of pain. J Pers Soc Psychol 1992;63: [14] Prkachin KM, Currie NA, Craig KD. Judging nonverbal expressions of pain. Can J Behav Sci 1983;15: [15] Prkachin KM, Berzins S, Mercer SR. Encoding and decoding of pain expressions: a judgement study. Pain 1994;58: [16] Prkachin KM, Craig KD. Expressing pain: the communication and interpretation of facial pain signals. J Nonverbal Behav 1995;19: [17] Prkachin KM, Solomon P, Hwang T, Mercer SR. Does experience affect judgements of pain behavior? Evidence from relatives of pain patients and health-care providers. Pain Res Manag 2001;6: [18] Prkachin KM, Mass H, Mercer SR. Effects of exposure on perception of pain expression. Pain 2004;111:8 12. [19] Prkachin KM, Solomon PE, Ross J. Underestimation of pain by health-care providers: towards a model of the process of inferring pain in others. Can J Nurs Res 2007;39: [20] Rizzolatti G, Fogassi L, Gallese V. Neurophysiological mechanisms underlying the understanding and imitation of action. Nat Rev Neurosci 2001;2: [21] Robinson ME, Wise EA. Gender bias in the observation of experimental pain. Pain 2003;104: [22] Solomon P. Congruence between health professionals and patients pain ratings: a review of the literature. Scand J Caring Sci 2001;15: [23] Sullivan MJL, Bishop SR, Pivik J. The pain catastrophizing scale: development and validation. Psychol Assess 1995;7: [24] Sullivan MJL, Martel MO, Tripp D, Savard A, Crombez G. The relation between catastrophizing and the communication of pain experience. Pain 2006;122: [25] Sullivan MJL, Martel MO, Tripp DA, Savard A, Crombez G. Catastrophic thinking and heightened perception of pain in others. Pain 2006;123: [26] Sullivan MJL, Thibault P, Savard A, Catchlove R, Kozey J, Stanish WD. The influence of communication goals and physical demands on different dimensions of pain behavior. Pain 2006;125: [27] Waddell G, Turk DC. Clinical assessment of low back pain. In: Turk DC, Melzack R, editors. Handbook of pain assessment. New York: Guilford Press; p