Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Facial expressive behaviour of a chronic low back pain population 1985

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
UBC_1986_A1 H92.pdf
UBC_1986_A1 H92.pdf [ 5.52MB ]
UBC_1986_A1 H92.pdf
Metadata
JSON: 1.0097285.json
JSON-LD: 1.0097285+ld.json
RDF/XML (Pretty): 1.0097285.xml
RDF/JSON: 1.0097285+rdf.json
Turtle: 1.0097285+rdf-turtle.txt
N-Triples: 1.0097285+rdf-ntriples.txt
Citation
1.0097285.ris

Full Text

FACIAL EXPRESSIVE BEHAVIOUR OF A CHRONIC LOW BACK PAIN POPULATION I By Susan Ann Hyde B.A. (Hons.) Univers ity of Calgary, 1975 M.A. Univers i ty of B r i t i s h Columbia, 1978 A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR.OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Psychology) We accept th i s thesis as conforming to the required standard THE UN^RSITY OF BRITISH COLUMBIA Ju ly 1985 © Susan Ann Hyde, 1985 In presenting t h i s thesis i n p a r t i a l f u l f i l m e n t of the requirements for an advanced degree at the University of B r i t i s h Columbia, I agree that the Library s h a l l make i t f r e e l y available for reference and study. I further agree that permission for extensive copying of t h i s thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. I t i s understood that copying or publication of t h i s thesis for f i n a n c i a l gain s h a l l not be allowed without my written permission. Department of Psychology The University of B r i t i s h Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date February 28, 1986 - i i - ABSTRACT Assessment of the subjective experience of pain represents an ongoing concern in c l i n i c a l , experimental and natural settings (Melzack, 1983). Previous laboratory studies (Craig & Patrick, 1985; Patrick, Craig & Prkachin, in press) using induced pain have suggested that facial expressive behaviour may provide a useful source of information additional to that offered by self-report in assessing the pain experience. There are, however, problems associated with the use of such behaviour, these being related to the issue of the voluntary and involuntary control that individuals have over their facial behaviour. The present study attempted to extend the findings of the ear l ier analogue research using a c l in i ca l sample of chronic low back pain patients. Self-report of pain was also investigated as were several variables of c l in ica l interest, i . e . , duration of complaint and disabi l i ty status. An attempt was made to overcome the tendency of people to attenuate their facial expressiveness in the presence of others scrutinizing the behaviour. F ina l ly, the ab i l i ty to control facial expression of pain was also investigated by requesting subjects to mask their facial expression during a painful movement and by requesting them to pose an expression of painful distress. Subjects (60 male and 60 female patients at the Shaughnessy Hospital Back Pain C l in ic) underwent a standardized physiotherapy protocol of four movements designed to induce low back or hip jo int pain. Half of the subjects of each sex were given a set of instructions designed to enhance their overall global expressiveness. Al l subjects rated their acute discomfort as well as their more chronic discomfort as experienced on a " typical " basis. Subjects' facial expressions were videotaped during the standardized protocol and subsequently scored by two independent sets of coders using the Facial Action Coding System (Ekman & Friesen, 1978b) and using a global expressiveness rating system. It was hypothesized that i f the Instructional Set manipulation was successful then those subjects receiving the instructions would be rated as more globally expressive than those who did not receive instruction. It was also hypothesized that greater facial act iv i ty would be present in the posed expression of painful distress than in the genuine and masked displays. A positive, albeit modest, relationship was found to exist between facial act iv i ty and self-report. This desynchrony between behaviour and self-report is common in the l i terature (Fordyce et a l . 1984). The manipulation to enhance the global expressiveness of the subjects was unsuccessful. Males and females differed marginally, however, in terms of judges' ratings of global expressiveness with females being rated as more expressive. This difference was also discussed in l ight of the absence of a sex difference on the FACS variables. Subjects were somewhat successful in deliberately attenuating their facial act iv i ty during a painful movement. It was concluded that future research should focus on further investigation of the existence of configurations of facial actions expressive of pain, any one or more of which might be-displayed, rather - iv - than pursuing the existence of a prototypical pain expression. It was suggested that a prototypic expression might be more characteristic of a posed display and further research could investigate this poss ib i l i ty . - V - TABLE OF CONTENTS PAGE ABSTRACT i i TABLE OF CONTENTS v LIST OF TABLES v i i i LIST OF APPENDICES ix ACKNOWLEDGEMENTS x INTRODUCTION 1 LITERATURE REVIEW 5 Use of Self-Report 6 Limitations of Self-Report Measures 7 Use of Nonverbal Expression in Assessment 9 Facial Expressive Behaviour 11 D i f f i cu l t ies with the Use of Facial Expressive Behaviour A. Differential Observer Sensitivity . . . 15 B. Control of Facial Expression: Deception, Dissimulation and Display Rules 16 Use of Behavioural Coding Systems . . . . . . . 19 The Facial Action Coding System 20 Use of FACS in Studies of Facial Behaviour Expressive of Pain 22 Limitations of the Previous Research Using FACS 23 Purpose of the Study 25 METHOD 27 Subjects 27 Experimental Setting 32 Self-Report Measures 32 Videotape Equipment 34 Procedure 34 Experimental Design 38 FACS Data Coding 38 Global Expressiveness Ratings 43 - vi - RESULTS 45 Re l iab i l i t ies 46 Judges Ratings of Global Expressiveness 46 FACS Scoring of Subjects' Facial Expressions 47 Demographic Data 48 Self-Report Variables 49 Intercorrelations Among the Self- Report Measures 53 Global Expressiveness Analysis 55 Analysis of FACS Data 56 Identifying the Principal Pain Related Action Units 56 Group Comparisons 59 Factor Analysis of FACS Data 60 Relationships Among the Variables 64 Relationship of Selected FACS AUs and Global Expressiveness 64 Relationship of Selected FACS AU's and Self-Report 64 Relationship of the "Pain" AUs and the McGill Scales 66 Evaluative Scale 66 Relationship of the "Pain" AUs and the Gracely Scales 66 Painfulness Scale 66 Relationship of Selected Demographic Variables, Self-Report of Chronic Pain and Global Expressiveness . . . . . . 66 DISCUSSION 68 Demographic Variables 68 Self-Report Measures 69 McGill Pain Questionnaire 69 McGll Pain Questionnaire and the Gracely Scales 71 Global Expressiveness 72 FACS Analysis of Facial Expressive Behaviour 74 Genuine Pain Expression 76 Masked Expression 80 Posed Expression 80 - v i i - Genuine Pain Expression Compared to the Masked and Posed Expressions 83 Factor Analysis of the Genuine Pain Segment 85 Relationship between the "Pain" AUs and Global Expressiveness 88 Relationship of the "Pain" AUs and Self-Report 90 Effectivenss of the Painful Stimulus 91 Summary and Conclusions 94 REFERENCES 102 APPENDICES 110 - v i i i - LIST OF TABLES PAGE TABLE 1 Demographic Variables (Males and Females Combined) 28 TABLE 2 Breakdown of Disabi l i ty Status for Males and Females 31 TABLE 3 Group Means and Standard Deviations for Self-Report MANOVA 51 TABLE 4 Means and Standard Deviations for the "Acute" and "Chronic" MPQ and Normative Data from the Shaughnessy Hospital Back Pain Cl in ic 52 TABLE 5a Correlations between "Acute" MPQ and Gracely Scales 54 TABLE 5b Correlations between "Acute" MPQ and "Chronic" MPQ 54 TABLE 6 Facial Expressiveness Rating Frequencies For Males and Females 57 TABLE 7 Action Unit Categories Remaining After Application of Exclusion Criterion 58 TABLE 8 Means and Standard Deviations of Eight AUS Appearing More or Less Frequently in the Pain, Masked and Posed Segments 61 TABLE 9 Post Hoc Comparisons (Tukey Procedure) of Segment Effect 62 TABLE 10 Factor Pattern Matrix 65 TABLE 11 Intercorrelation Matrix of Factors 65 TABLE 12 Correlations Between Judges Ratings of Global Expressiveness and Facial Activity During the Pain Segment 89 TABLE 13 Correlations Between the McGill and Gracely Scales and Facial Activity During the Genuine Pain Segment 92 - ix - LIST OF APPENDICES PAGE APPENDIX 1 Diagnostic Categories 110 APPENDIX 2 McGill Pain Questionnaire Verbal Descriptor Scales I l l APPENDIX 3 Gracely Scales 112 APPENDIX 4 Consent Form 113 APPENDIX 5 Physiotherapist's Instructions to Subject 114 APPENDIX 6 Scoring Units for the Facial Action Coding System 116 APPENDIX 7 Global Coding Manual 117 APPENDIX 8a Summary of Self-Report and Global Expressiveness Data Analysis 120 APENDIX 8b Summary of Analysis of "Acute" versus "Chronic" MPQ 120 APPENDIX 9 Summary FACS Data Analyses: Group Comparison . . . . . . 121 APPENDIX 10 Post Hoc Comparisons of the Segment Effect 122 APPENDIX 11a Relationships Among the Variables: Pain AUs and Global Expressiveness 124 APPENDIX l i b Relationships Among the Variables: Pain AUs and "Acute" MPQ Scales 124 APPENDIX 11c Relationships Among the Variables: Pain AUs and the Gracely Scales 125 - X - ACKNOWLEDGEMENTS I would l ike to thank my supervisor, Dr. K.D. Craig for his time, ideas, encouragement, patience and willingness to part, temporarily at least, with a great deal of valuable reference material. His continued enthusiasm for this project helped to ensure i ts completion. I would also l ike to thank Doug Lee and Chris Patrick for their help, advice and encouragement with a l l phases of this project. I also extend my thanks to Brenda Gerhard and Cara Zaskow for enduring cheerfully the endless hours of FACS coding and to Joe Klancnik and Wendy Ross for doing the global ratings. I am grateful to Dr. R. McMahon and Dr. R. Hakstian for serving on my dissertation committee, and for providing invaluable input during the various stages of this project. I would also l ike to take this opportunity to thank Dr. Peter Wing, Faith Gagnon, the physiotherapists, and the staff of the Shaughnessy Hospital Back Pain Cl in ic without whose support and cooperation this project would not have been possible. F ina l ly , I would l ike to convey my heartfelt thanks to a l l my friends, but especially to Neal, Winnie and Bruno, and to my family for their patience and unflagging support. -1 - INTRODUCTION Assessment of the subjective experience of pain i s a d i f f i c u l t task for both c l i n i c i a n s and nonc l in ic ians . T r ad i t i ona l l y , such assessments have been made on the basis of the pat ient ' s se l f - repor t and on observer judgements of the pat ient ' s behaviour. However, there are problems associated with both these methods of assessment. In the case of the observer, that indiv idual may bring to the assessment a host of factors which w i l l influence his or her judgement, for example, his or her personal and/or professional experience of pain, and his or her expectations about the level of pain to be expected with any given i l l n e s s or injury (Bond, 1979; Teske, Daut & Cleeland, 1983). Se l f - report , while an extremely useful tool (Hi lgard, 1969), i s also subject to several l im i t a t i on s . The indiv idual making the report may not have s u f f i c i en t a b i l i t y , s k i l l or knowledge to be able to communicate the pain experience. In addit ion, the indiv idual may, for a variety of reasons, exaggerate or conceal the amount of pain he or she i s experiencing (Craig & Prkachin, 1983). F i na l l y what people say they do and the i r actual behaviour may d i f f e r , because each type of behaviour i s subject to d i f fe rent condit ioning effects in the natural environment (Fordyce, 1983). It would seem that addit ional sources of information are necessary in order to ensure more accurate assessment of the pain experience (Sanders, 1972; Tan, 1982). I t i s suggested that nonverbal expression, in pa r t i cu la r fac ia l expressive behaviour, may provide such addit ional - 2 - information (Hjortsjo, 1970; Kleck et a l . , 1976). However, there are several d i f f i cu l t ie s associated with the use of facial expressive behaviour. Even when attempts were made to generate systematic observational methods to study facial expression of pain, their use often required acceptance of theoretical assumptions about the underlying meanings of facial behaviour (Craig & Prkachin, 1983). In addition, problems of dif ferentia l observer sensit ivity arise. The various factors which influence observer sensit ivity lead to erroneous judgements about the distress of others, and thus have detrimental effects on the individual being assessed (Jacox, 1980). A major source of d i f f i cu l ty in the use of facial expressive behaviour l ies in the fact that people appear able to exercise both voluntary and involuntary control over their facial expressions. Ekman and Friesen (1971) have suggested that a set of "display rules" governing emotional expression (and presumably expression of painful distress as well) is internalized during the socialization process. These "display rules" can serve to attenuate, neutralize or mask the distress f e l t by an individual. There is some evidence that facial behaviour is indeed influenced by these rules (Ekman, 1977; Kleck et a l . , 1976; Kraut, 1982; Lanzetta, Cartwright-Smith & Kleck, 1976). The use of behavioural observational coding systems, in particular those involving microanalytic studies of filmed or videotaped material, may serve to overcome the d i f f i cu l t i e s posed by differential observer sensit ivity and also describe display rules. Ekman and Friesen's Facial Action Coding System (FACS; Ekman & Friesen, 1976; 1978a; 1978b) has - 3 - been demonstrated to be useful in the study of facial expressive behaviour associated with emotion. Recent studies (Craig & Patrick, 1985; Patrick, Craig & Prkachin, in press; LeResche, 1982) have shown the appl icabi l i ty of FACS to the study of the pain experience. The work of Craig and his colleagues, employing induced pain with college student samples, has suggested overlapping sets of facial actions which seem to characterize the acute phase of noxious stimulation. The present study further defined this configuration of facial actions during c l in ica l pain, using a more demographically heterogeneous sample, including both males and females with chronic low back pain. Use of a c l in ica l sample provided the opportunity to examine relationships among several variables of c l in ica l interest: duration of complaint, d isabi l i ty status, self-report of pain, and facial expressiveness. It was suggested that the tendency to attenuate facial expressive behaviour in the presence of an observer might be influenced by an instructional set designed to encourage the individual to be as expressive of painful distress as possible. Thus half the subjects in the present study were given instructions of this type to examine whether this problem could be overcome. It was also suggested that the methodology of FACS would allow the observers to pick up "microdisplays" of facial behaviour that might remain for those individuals who were less expressive. These attenuated facial actions could also then be used in building the set of facial actions,characteristic of painful distress. A further attempt was also made to investigate the problems posed by the individual 's voluntary and involuntary control of facial behaviour. - 4 - * The individual 's ab i l i ty to attenuate facial behaviour was studied by asking the subjects to deliberately do so during a repetition of the painful movement. Subjects were also requested to deliberately pose an expression of painful distress in order to investigate the relationship that such an expression would have with a genuine facial expression of distress. - 5 - LITERATURE REVIEW Assessment of the subjective experience of pain represents an ongoing concern for c l in ic ians and researchers in the health care f i e ld (Gracely, 1979; Jacox, 1980; McCaffery, 1972; Melzack, 1983; Sternbach, 1974; Wolff, 1978). Tradit ional ly, such assessment involves a judgment by a physician, nurse, physiotherapist or other health care professional about the patient's behaviour based upon evidence of tissue damage, verbal report and appearance. Unfortunately, many additional considerations make the judgment task complex and d i f f i c u l t . While the c l in ic ian attempts to be as objective as possible, the assessment process is influenced nonetheless by the observer's own professional and personal experience of pain, his or her endorsement of traditional socio-cultural beliefs about the level of pain to be expected in any given i l lness , and a variety of other factors such as the observer's personality and occupation (Bond, 1979; Dudley & Holm, 1984; Fordyce et a l . , 1978; Lenburg, Glass & Davett, 1970; Teske, Daut & Cleeland, 1983). In addition, the person experiencing pain has his or her attitudes toward how one should react to and report pain. Other factors such as the person's ethnic background, their response style, and states of anxiety or depression wil l also affect their pain behaviour (Teske, Daut & Cleeland, 1983). Thus, while there is an urgent need to know the nature and severity of the pain patients are suffering, there is a continual risk of misunderstanding in communications between the c l in ic ian and the patient. - 6 - In addition, environmental pressures exist which can lead to purposeful or unwitting dissimulation of the pain experience on the part of the patient. These environmental pressures are generally thought of in terms of secondary gain. Such factors as financial benefits, avoiding an unpleasant work or personal situation, and increased attention from family and friends can al l serve as inducements for the individual to exaggerate or even simulate pain. Other environmental pressures, however, can result in the patient trying to cover up the fact that he or she is in distress. Athletes, for example, may tend to minimize the seriousness of an injury in order to be able to continue to participate. Use of Self-Report Despite al l of the aforementioned variables, c l in ic ians and other individuals such as parents, teachers and employers must use some means to assess pain in both the c l in ica l and natural settings in which i t occurs. A rich variety of categories of expressive responses are available for use in assessment; for example, reflexes, startle response and posture (Melzack & Wall, 1965). However, i f any weight is to be attached to the preponderance of investigations using verbal report in the sc ient i f i c l i terature, then self-report has tradit ional ly been the most heavily used category. There are a number of reasons why this has been the case. It has been stated (Hilgard, 1969) that there are no physiological measures of pain which are as able to discriminate fine differences in stimulus conditions, are as rel iable upon repetition or as responsive to changes in the patient's condition as self-report. Gracely (1983) has noted that - 7 - language allows the separate assessment of the sensory-discriminative aspect of the pain experience and the motivational and emotional aspects. It has also been pointed out (Craig & Prkachin, 1983) that self-report requires minimal training or instrumentation and quantification is readily accomplished. This is apparently the case, at least for simple self-report scales of pain intensity based on unidimensional models of pain. Current formulations of pain, however, are multidimensional and thus necessitate more complex measurement tools (Gracely, McGrath & Dubner, 1978; Melzack, 1975). The McGill Pain Questionnaire (MPQ; Melzack, 1975), for example, allows for the quantitative assessment of several dimensions of pain complaint and permits evaluation of the relative contributions of sensory, affective and evaluative factors to pain communication. Studies using the MPQ have indicated that i t is re l iable, valid and responsive to treatment of somatic or affective disturbances in both acute and chronic pain populations (Kremer & Atkinson, 1983). Limitations of Self-Report Measures Despite numerous advantages, self-report measures do have their limitations (Brena, 1983; Fordyce, 1976). It is possible that the respondent may simply not have suff icient knowledge or sk i l l to be able to interpret or report the pain experience even though he or she may wish to do so. Requests for self-report are also highly obtrusive and sensitize people to situational demand (Craig & Prkachin, 1983). As a consequence, people may intentionally or unintentionally attempt to conceal or exaggerate the amount of pain they are feel ing. In relation - 8 - to this, Fordyce and his colleagues (Fordyce, 1983; Fordyce et a l . , 1984) have noted that what people say they do and their actual behaviour usually encounter quite different consequences in the environment and the two categories of behaviour are therefore subject to different conditioning effects. Thus "say" and "do" actions are free to diverge, without i t being necessary to attribute discrepancies to any loss of the person's candor or truthfulness. A further l imitation to self-report is that young children and certain disabled people are unable to use language as a medium of expression (Craig, 1980). With regard to the relationship between chronic pain and sel f - report, Kremer, Block and Atkinson (1983) have concluded that self - report of pain intensity and other related pain behaviours are distorted by a number of variables. The authors reviewed a number of studies indicating that: (i) depression and the chronicity of the pain complaints were s ignif icantly related to the under reporting of act iv i ty; ( i i ) that the disclosure of pain information is a function of the age, sex and perceived professional stature of the target person; and ( i i i ) that an, as yet, unclear relationship exists between self-report and whether or not the patient is receiving (or hoping to receive) compensation. As a consequence of considering al l the limitations of self-report, i t is clear that additional components of the suffering individual 's expressive repertoire must be used by observers in order to more accurately understand what is happening (Sanders, 1972; Tan, 1982). Indeed, in al l l ikelihood the emphasis on self-report in the l i terature on pain misrepresents the importance of nonverbal expression to - 9 - c l in ic ians and observers in the natural environment. Use of Nonverbal Expression in Assessment Nonverbal expressive behaviour can serve as an additional source of information for those making judgments about someone who is suffering. Indicators of pain may include facial act iv i ty, movements of the limbs including reflexes, postural attitudes, paralinguistic vocalizations such as sighs and groans, and overt signs of autonomic act iv i ty . Al l apparently can be used to enhance these judgments (Craig & Prkachin, 1983). It has been observed (Wall, 1979) that verbal report comes into play late in the sequence of events during a pain episode, whereas other expressive channels play a more immediate role in communicating the experience. The importance of nonverbal signs of subjective distress was highlighted in a study (Jacox, 1980) in which nurses reported preferring and relying more heavily on physiological signs, body movements and facial expressions than on the patients' verbal complaints of pain. DiMatteo, Friedman and Taranta (1979) have reported that patients tend to prefer physicians who can understand their feelings even when the patient is unable or unwilling to express such feelings verbally. De Paulo et a l . (1978) found that people attach greater cred ib i l i ty to nonverbal expression than to self-report when the two conf l ict (when making judgments of emotional expression). F inal ly, Kraut (1978) reported that the accuracy of judgments of others' attempts at deception is enhanced when the observers are provided with nonverbal as well as verbal expressive information. Another advantage of using nonverbal expressive behaviour is that i t - 10 - seems to be less amenable to conscious distortion than self-reports of subjective states (Ekman & Friesen, 1969; 1974). It would seem that people do not monitor their nonverbal expressions, particularly bodily cues, to as great a degree as they do their verbal behaviour. This suggests that in some cases unobtrusive measurement of the former might provide more accurate information about subjective states, including pain. Despite the importance of nonverbal behaviour as an index of the pain experience, "the systematic study of these phenomena has been sparse, occasionally erroneous and without substantial impact on the knowledge base in the f i e ld " (Craig & Prkachin, 1983). For example, Johnson, Kirchoff and Endress (1975) in a study involving children undergoing cast removal, investigated the frequency of a number of nonverbal behaviours. Behaviour was categorized as major (e.g., kicking, screaming) or as minor (e.g., grimace, frown), however, no explanation was provided as to why these particular behaviours were selected. Kendall et a l . (1979) had judges rate a number of nonverbal behaviours (e.g., clenching f i s t s , shaking or moving around "unnecessarily") while assessing the adjustment of patients during a cardiac catherization procedure. Again no explanation was given as to why these behaviours were selected. Nonverbal expression is not as easily measured as verbal behaviour and i t is only recently that comprehensive, rel iable measurement strategies for the former have begun to appear. Keefe and Block (1982) developed an observation method of assessing pain behaviour in chronic low back pain patients. A variety of motor - 11 - patterns such as bracing, rubbing, grimacing and sighing were demonstrated to be reliably observed and were correlated to patient's ratings of pain. Moreover, the frequency of these behaviours tended to decrease with treatment and these changes in behaviour also correlated with changes in pain ratings. Facial Expressive Behaviour One area that has seen rapid progress over the past few years is the study of facial expressions as indicators of emotional experience. In summing up the f ie ld of nonverbal communication, Harper, Weins and Matarazzo (1979) stated that the face, and facial expression in particular, may be the most important area of the body in nonverbal communication. The authors maintain that this is the case because of the amount of information the face can convey (especially in a short period of time) and the types of information conveyed, for example, emotional and att i tudinal . The authors c i te numerous findings in support of this claim (e.g., Dittman, 1972; Ekman, Friesen & Ellsworth, 1972; Weitz, 1974). Although this f ie ld has a lengthy history, systematic attempts to measure facial expressive act iv i ty are a more recent development. Darwin's (1872) work on emotional expression represents one early attempt to study such act iv i ty. However, he focussed on the adaptive, evolutionary role of expressive behaviour and actually concluded "that in the case of the chief expressive actions they are not learned but are present from the earl iest days and throughout l i f e are quite beyond our contro l . . . " (p. 352), an assertion that would be debated today. Darwin - 12 - argued for the universality of the "chief expressions" and provided detailed descriptions of expressions of happiness, anger, sadness, distress and so forth. Ekman, Friesen and Ellsworth (1982) in a review of the early laboratory studies using s t i l l photographs (e.g., Plutchik, 1962; Tomkins and McCarter, 1964; Woodworth, 1938) concluded that a minimum of seven categories of emotion could be discriminated by untrained observers on the basis of impressionistic judgments of facial features. The categories the authors identif ied were happiness, surprise, fear, anger, sadness, disgust/contempt, and interest. Ski l led c l inic ians have also demonstrated f a c i l i t y in identifying important patterns of nonverbal expression of painful distress. Ambeau (1982) suggested that pain disorders with greater contributions from psychological than organic determinants would be characterized by greater expressiveness in general, more verbal complaint and report, acute reactions when changing bodily position, groaning as opposed to sucking in the breath during pangs of distress, and tendencies to interrupt act iv i t ies by stopping talking and averting the eyes when in particular distress. However, the observations were anecdotal and unsystematic and need to be validated before any conclusions can be drawn. There is also evidence that observers, whether naive or trained, can have considerable success in using nonverbal cues in making judgments about the pain of others based upon observation of behaviours. Kleck et a l . (1976) and Lanzetta, Cartwright-Smith and Kleck (1976) both reported that untrained judges could rel iably assess the' amount of distress expressed by subjects exposed to painful e lectr ic shocks. Prkachin, - 13 - C u m e and Craig (1983) found that naive judges were able to discriminate facial expressions for low, medium and high e lectr ic shock intensit ies delivered to volunteer subjects. Other research efforts have been aimed at identifying specific behaviour patterns signifying pain to observers. Hjortsjo (1970) described facial changes brought about by the actions of the obicularis occuli and the masseter muscles that he believed represented a particular pain expression. Leventhal and Sharp (1965) developed a system for coding the morphology of facial changes induced by chi ldbirth labour and described facial correlates of apparent reductions in comfort and increases in distress which were related to the extent of cervical d i lat ion. However, i t has been pointed out (LeResche & Dworkin, 1984) that i t was not possible to derive descriptions of facial expression of pain from this study as i t was not clearly defined what was meant by distress; that is pain, fear, anxiety or some combination of these states. In addition, a methodological flaw resulted in uncertainty as to whether the observed facial behaviours occurred sequentially or in some configuration with one another. Izard et a l . (1980) recorded the facial expressions of 1 to 9 month old infants during innoculations or the taking of blood samples. They identif ied a facial pattern which, they thought, was a discrete expression of pain that was relat ively specific to infants. This was described as a lowering of the brows, broadening of the nasal root, an angular squarish mouth and tightly closed eyes. (There is some dispute - 14 - about the behavioural specifics however, as H. Oster (personal communication to K.D. Craig, 1982) has suggested the Izard coding system may be flawed as i t is based upon s imi lar i t ies between infant's faces and the investigators' conception of adult's pain expressions. The Izard system apparently fa i l s to take into account differences between infant 's and adult's faces). Other evidence involves developmental changes which have been observed in pain related expression. Izard et a l . (1981) reported that facial expression of painful distress during innoculations decreased with age whereas anger expressions increased with age. These reports provide some evidence that pain relevant, possibly pain specif ic, information is encoded in expressive behaviour. A broad range of behavioural differences in pain response associated with age in infants was observed by Craig, McMahon, Morrison, and Zaskow (1984). Coders used a time-sampling behavioural observation system to distinguish among vocal expressive categories (language, crying, screaming, etc.) and nonvocal expressive categories (activity in the face, torso and limbs) in expressive reactions to routine immunization injections in the f i r s t two years of l i f e . The findings indicated that the reactions of children under one year were more spontaneous, global and linked to the tissue insult of the injection, whereas the children between 13 and 24 months displayed more anticipatory distress, used descriptive language during the session and engaged in self protective voluntary movements. Thus, expressions of pain were seen to change systematically during the f i r s t two years of l i f e as infants accumulate - 15 - experiences and acquire motor sk i l l s and the capacity to recognize and influence others. D i f f i cu l t ies With the Use of Facial Expressive Behaviour A. Differential Observer Sensit ivity The use of facial expression is not without i t s own d i f f i cu l t i e s . It seems (von Baeyer, Johnson & McMillan, 1982, unpublished manuscript) that the impact of nonverbal behaviour is mediated by individual differences in observer sensit ivity to the nonverbal expression of pain, von Baeyer (1982) found that sensitizers assigned higher pain ratings than repressors when rating slides portraying low levels of non-verbal pain expression. In another study (von Baeyer et a l . , 1982, unpublished manuscript) involving an analogue patient/cl inical relationship, the results indicated that the nurturance levels (as measured by the Adjective Checklist) of the raters interacted with the degree of nonverbal expressiveness exhibited by the "patients". Prkachin, Currie and Craig (1983) and Patrick et a l . (in press) found that observers' judgments of the distress of others could be influenced by instructional set by affecting their willingness to attribute pain at varying levels of severity. In natural settings, people may be relatively sensitive or insensitive to, or unwilling to respond to cues when making judgments about the amount of stress another person is experiencing. Parents of children suffering recurrent abdominal pain without known organic origin have been characterized as over-anxious, over-protective and fearfu l . These concerns appeared to make then unduly sensitive and possibly over- - 16 - reactive to minimal signs of physical distress (Craig, 1980). On the other hand, a nurse who expects patients experiencing pain to clearly communicate i t verbally or nonverbally, will incorrectly assess a quiet patient as not being in any distress (Jacox, 1980). Such erroneous judgments can have detrimental effects on the patient or individual being judged. In the former case pain behaviour will be reinforced, the chi ld is l ike ly to continue to suffer, and i t is improbable that any change wil l be effected until the behaviour of the parents is altered. In the latter case the individual often endures needless suffering because the prescribed analgesics are not administered. Development of systematic, objective and sensitive measures of the nonverbal behaviour demonstrated to be associated with painful distress would aid in the more accurate assessment of that distress to the benefit of the patient and those around him or her. B. Control of Facial Expression: Deception, Dissimulation and Display Rules. One of the proposed advantages of nonverbal measures of pain is that they are less subject to conscious and unconscious control than are verbalizations. However, the various types of nonverbal expression of pain are familiar to most people and to some degree do seem to be subject to voluntary control. Children play at being i l l and actors can express pain in varying degrees, as do certain individuals seeking the secondary gains of being i l l . At a minimum, there are cultural conventions concerning stereotypic displays of pain that enable people to enact them - 17 - with ease. (Whether these cultural stereotypes correspond to prototypical pain expressions or not is another matter.) The process by which emotional expression comes to be under voluntary control has been the subject of several investigations. Ekman and Friesen (1971) noted that emotional experience comes to be under the control of "display rules" through the social ization process. Socially learned (culture specific) display rules can modify expression in a number of ways; for example, intensify, deintensify, neutralize and mask. Such rules are thought to be learned in childhood and become habits; that i s , learned to the point of being automatic. It should be noted, however, that there may be limitations in generalizing from the l i terature on emotional expression to the expression of painful distress. It has been suggested (LeResche & Dworkin, 1984) that i t remains unclear whether the facial expressions that accompany pain are indications of emotional response (e.g., s tart le, fear, anger) to the sensory aspect of the pain experience, or whether they represent one or more dist inct pain states. In addition, different types of pain can be distinguished by different patterns of sensory, affective and evaluative verbal response (Melzack, 1975) and these pain types might conceivably be associated with different facial expressive behaviours. There is some evidence that facial behaviour is indeed influenced by these display rules. Ekman (1977), for example, described a study in which subjects viewed a stressful fi lm either alone or in the presence of an observer. Slow motion films of the subjects in the latter condition, revealed the presence of i n i t i a l facial movements indicative of distress - 18 - which were rapidly suppressed, and replaced by neutral or positive expressions such as smiles. Kleck et a l . (1976) found that subjects undergoing e lectr ic shock in the presence of an age peer observer were judged as being less expressive or distressed than when alone. In addition the subjects were less physiologically responsive (as measured by skin conductance levels) and their self reported distress was lower when under observation. A further study (Lanzetta, Cartwright-Smith & Kleck, 1976) found that when instructed to deceive, subjects were successful in convincing judges that they had received a more or less intense shock than had in fact been delivered. Kraut (1982) found that subjects smelling pleasant and disgusting odors were less successful communicators of the experience when aware that someone was in the room with them (despite the fact that the other person could not see the subject). When requested to hide their expressions they were similarly less successful when the same "unobserving" other was present; that is they "leaked" their evaluations more than when completely alone. Ekman and Friesen (1969) have suggested that such leakage may indeed occur when subjects are deliberately trying to deceive an observer by attenuating facial behaviour. However, they also state that expert observers viewing facial behaviour on videotape may well be able to detect such leakage. While the authors do not specify the mechanisms of detection, one possibi l i ty may relate to their discussion of "micro affect displays". Micro displays are those facial behaviours which may be "fragments of a squelched, neutralized or masked display" (p. 98) or - 19 - they may be extremely rapid versions of the macro display. When such micro displays are shown in slow motion they do convey emotional information which can be recorded by observers (Ekman & Friesen, 1969). In the case of faking painful distress, careful study of posed versus genuine displays of facial expressive behaviour associated with pain similar to that carried out by Ekman and Friesen (1982) with smiles, may provide the necessary clues to detect such behaviour. Use of Behavioural Coding Systems Craig and Prkachin (1983) have emphasized the need for the development of methodologies for decoding the behavioural repertoire of expressions of pain. Many of the studies on emotional expression and indeed, of pain expression, have suffered from the disadvantage of relying upon observational systems that require acceptance of strong theoretical assumptions about the underlying meaning of expressive acts. Judges appear to rapidly impose these assumptions on nonverbal expression that i s , by nature, subtle and ambiguous (e.g., Johnson, Kirchoff & Endress, 1975; Kendall, Williams, Pechacek, Shisslak & Herzoff, 1979; Langer & Janis, 1975). However, there are several recently devised behavioural observational coding systems that seem to offer some promise when there is insistence upon operational definitions of the target behaviours being observed and interobserver r e l i a b i l i t y . Unfortunately while these systems y ie ld valuable information, they generally involve the use of real time observations and the speed and complexity of changes in facial expression can exceed the observer's capacity to encode them (Cataldo et - 20 - a l . , 1979; Ekman, Friesen & Ellsworth, 1982; Melamed & Siegel, 1975). Microanalytic studies of filmed or videotaped material presented to judges in slow motion or on a frame by frame basis seem to be necessary to capture the complexity of visual displays. This has been demonstrated to be the case in studies of emotional expression. The Facial Action Coding System Perhaps the most sophisticated development to date in the measurement of facial expressive behaviour is the Facial Action Coding System (FACS; Ekman & Friesen, 1976, 1978a, 1978b). The system was designed to measure al l v i s ib le facial behaviour, not just that presumably involved in emotional expression. It is based on an analysis of the anatomical basis of facial movement and an examination of how each muscle of the face acts to create a vis ible appearance change. FACS identif ies 44 separate facial Action Units (discrete movements in the forehead, eye, cheek, nose, mouth, chin and neck regions) along with 20 additional Action Descriptors (e.g., changes in the orientation of the head). Any facial expression can be described in terms of the action unift (AU) or the combination of AUs that produced i t . FACS can also be used to code the duration and intensity of facial movements. The system is very comprehensive and equally complex, requiring considerable time and practice to master. The major advantage of FACS is that i t is objective, rel iable, and atheoretical since the trained observers are able to consistently use expl ic i t definitions of the specific components of facial expression. Therefore, there is l i t t l e opportunity to impose subjective judgments - 21 - regarding supposed underlying psychological states. This nicely sidesteps the previously discussed problem of observer bias. Using FACS, the observer does not make inferences about what the facial behaviour means; rather the specifics of the behaviour present are simply recorded. The facial behaviours which are identif ied can be related to cr i ter ion variables of interest, for example self-report of an emotional state, psychophysiological measures, or other nonverbal indicators, and relationships with l i f e history and environmental events can be explored. Studies u t i l i z ing the FACS system have recently begun to be carried out by various investigators. For example, Ancoli (1979) and Ancoli, Kamiyua and Ekman (1980), compared female subjects' responses to pleasant and unpleasant films and found that a pattern of autonomic nervous system changes was found only when the face showed what FACS identif ied as positive or negative emotions. Ekman, Friesen and Ancoli (1980) found that FACS hypotheses about a number of aspects of emotional experience predicted the subject's report on multidimensional scales immediately after viewing pleasant and unpleasant fi lms. Facial action (as measured by FACS) was found to provide accurate information as to which of two enjoyable experiences was enjoyed most. FACS also provided accurate information as to whether a subject was unhappy and the extent of the negative emotion. In addition, some evidence was found to differentiate unhappiness from disgust on the basis of the FACS data. These and other studies (e.g., Ekman, Friesen & Simons, 1982; Ekman, Hager & Friesen, 1981) provide the basis for considering FACS to be an important measurement tool for facial - 22 - expressive action in association with a wide variety of. emotional and even nonemotional states. Use of FACS in Studies of Facial Expressive Behaviour of Pain Recently, Craig and his colleagues have applied FACS to the study of facial expressive behaviour of pain. While pain is not considered to be an emotion, the pain experience embodies a signif icant emotional component (Craig, 1984) and there is reason to bel'ieve that specific facial movements may be associated with the experience of pain (cf. Hjortsjo, 1969). In a study of female facial expressive movement in response to the cold pressor experience, Craig and Patrick (1985) identif ied a number of specific facial actions associated with painful distress: narrowing of the eye aperture from below, raising the upper l i p , pulling the l ip corners, parting of the l ips or dropping the jaw, and eyes closing or frequently blinking. These actions were strongly associated with the subject's report of subjective distress at the onset of the noxious stimulation, although they declined over time. Patrick et a l . (in press) obtained results consistent with the previous study when investigating facial expressive behaviour of females provoked by a range of painful and nonpainful e lectr ic shocks. In this study, narrowing of the eye aperture from below, raising the upper l i p , and blinking were al l demonstrated. However, several action units identif ied previously were not seen and one additional AU, brow lowering, was observed in this shock study. The authors attribute these differences to the different nature of the shock and cold pressor experiences; that is the former is a brief, noxious stimulus - 23 - while the latter is a relat ively more enduring, aching pain. Limitations of the Previous Research Using FACS The research of Craig and his colleagues is innovative and has yielded valuable and interesting results; however, i t does suffer from several l imitations. The f i r s t is that only female subjects were studied. This was largely due to the fact that, at least in studies of emotional expression, females have generally been found to be more accurate encoders or senders of information (Buck, Mi l ler & Caul, 1974; Schwartz, Brown & Ahern, 1980). As i t is thus unclear whether or not males would exhibit a similar set of facial behaviours when in pain, the results can only be generalized to other females. Another, perhaps less serious l imitat ion, is the homogenous nature of the two samples studied. Both samples were restricted to a university student population whose ages ranged from 17 to 28. Whether the results can be generalized to other or more heterogeneous populations is again unclear. A further l imitation is the fact that the previous research has attempted to define a facial topography of pain using a r t i f i c i a l l y induced pain rather than studying c l in ica l pain or pain induced in the natural environment. Analogue research is extremely valuable as i t allows for the precise and systematic control of the painful stimulus. In addition, there are fewer log ist ica l constraints in a laboratory setting, enabling the experimenter, for example, to administer rating scales concurrently with the stimulus or to take physiological measures. However, despite such advantages, there is always concern as to - 24 - the generalizabil ity of results outside the laboratory. Laboratory induced pain presumably has less emotional weight attached to i t and is therefore different from c l in i ca l pain (Beecher, 1959). Moreover, i t is always an acute experience whereas c l in ica l pain can be either acute, such as the pain experienced at the time of tissue injury or during a medical diagnostic procedure, or chronic in the sense that i t persists or is recurrent over a long period of time, for example low back pain. As several prominent researchers have noted (Bonica, 1980; Fordyce, 1983), the physiological, emotional and behavioural responses to chronic pain can be quite different from those to acute pain. It may be that the facial behaviour of chronic pain patients undergoing an acute pain experience may be different from that of a nonclinical population. Studying a c l in ica l population has additional advantages beyond ecological va l id i ty . It allows for an exploration of the relationship between important c l in ica l variables, such as chronicity of pain and behaviour. Another variable of interest in a c l in i ca l population is d isabi l i ty status. It is generally thought that i f people are compensated for i l lness behaviour, the likelihood of repetition of that behaviour wil l increase. Thus, many behaviourally oriented pain programmes wil l not treat patients involved in l i t i gat ion or with pending Workers' Compensation claims (e.g., Fordyce, 1976). In a review of the l i terature on the relationship between compensation status, self-report and pain behaviour, Kremer, Block and Atkinson (1983) concluded that at present the results are confl ict ing as to whether compensation - 25 - perpetuates i l lness behaviour. Thus, i t would be of interest to explore the relationship of d isabi l i ty status, facial behaviour and self-report of pain. Purpose of the Study In the present study, an attempt was made to define a facial topography of pain in a sample of chronic low back pain patients. The results were compared with those of previous research (Craig & Patrick, 1985; Patrick et a l . , in press) which employed nonclinical subjects in analogue situations. In addition, both males and females were studied in the present research to determine i f differences existed in their facial expressive behaviour or self-report of painful distress. As the c l in ica l sample used was more heterogeneous in terms of demographic characteristics such as age, education and ethnic group than samples in the previous research, there was an opportunity to determine i f the ear l ier findings could be cross-validated with a more diverse sample. Another opportunity provided by use of a c l in i ca l sample was that of examining the relationship of several variables of c l in ica l interest (for example, duration of complaint, d i sabi l i ty status) with self-report of pain and overall or global facial expressive behaviour. An attempt was made to address the problems arising from the potential control that individuals can exert over their facial expressive behaviour. Two types of individuals present d i f f i cu l ty with regard to the accurate assessment of painful distress. The f i r s t is the one who attempts to maintain a stoical expression; the second, the individual who, for a variety of reasons, exaggerates or even fakes behaviour - 26 - indicative of painful distress. It the case of the former, the FACS procedure, which allows for repeated viewings in real time, slow motion and frame by frame, should make i t possible to pick up any of the "micro displays" discussed ear l ier. An attempt was also made to determine i f instructing subjects to be as expressive as possible of their distress would circumvent any display rules whose effect would be to attentuate facial behaviour expressive of pain. In order to judge the overall effect of this manipulation, non FACS trained coders were asked to make judgments as to the degree of global facial expressiveness of each subject. To further investigate the control individuals may be able to exert over their facial behaviour, a l l subjects were requested to deliberately attentuate such behaviour during a repetition of the previously identif ied most painful movement. To address the problem of the individual who may exaggerate or even fake facial behaviour indicative of painful distress, a l l subjects were asked to deliberately pose an expression presumably associated with such distress. FACS scoring was used f i r s t to identify the action units (AUs) associated with a painful movement and second, to identify the AUs associated with the masked and posed facial behaviour associated with pain. It was hypothesized that greater facial act iv i ty, in terms of the number and/or frequency of AUs, would be present in the posed display than in the genuine one. It was also hypothesized that fewer and/or less frequent AUs would be present in the masked display as compared to the genuine and posed displays. - 27 - METHOD Subjects Subjects were 60 males and 60 females selected from among the patients undergoing assessment at the Shaughnessy Hospital Back Pain C l in ic . The Back Pain Cl in ic is an assessment unit for chronic back pain patients; that i s , those individuals with a history of back pain of 6 months duration or longer. Patients are referred to the Cl inic by their family physician, by special ists (orthopedic, neurological, rheumatological and general surgery), by the Workers' Compensation Board and by the Insurance Corporation of Brit ish Columbia. Each patient is assessed by a special ist (orthopedic surgeon or rheumatologist), a general pract ic it ioner, a physiotherapist and a psychologist. Only those patients whose referral information indicated a history of pain in the lower back and/or hips were approached to be in the study. A number of selection c r i te r i a were applied to those individuals approached for participation in the study; 1) the patient had to be experiencing pain at the time of the assessment; 2) the patient had to be able to understand written and spoken English; and 3) the patient could not be discernably under the influence of alcohol. Use of a prescribed medication was not an exclusion cr iterion as this is the case with the majority of chronic pain patients. Descriptive information about the patients appears in Table 1. The mean age of the subjects was 42.7 years with a range of 17 to 78 years. The majority of the subjects (58.6%) had at least a high school diploma. TABLE 1 - 2 8 Demographic Variables (Male and Females Combined) Variable Name SD Percentage Range n 1 Age (in years) Education level Grade school/junior high High school diploma Post secondary education Technical school or sk i l led trade Ethnic group Caucasian-born in Canada or U.S., no recent ethnic influence All other ethnic groups Duration of complaint Chronic (in years) Duration of compl ain't Current (in years) Locus of complaint Low back pain Back and leg pain Back and hip or hip only pain 42.7 2.2 13.71 8.42 10.33 1.52 66.7 25.0 8.3 41.4 28.7 16.1 13.8 77.12 22.88 17-78 117 81 118 .58-55.0 120 .25-8.0 120 120 - 29 - Variable Name M SD Percentage Range n* Prior Back Surgery (38 subjects) — - - 32 - - 120 One operation 25 Two operations or more 6.6 Disposition 120 Further investigation 6.9 and/or invasive procedures Noninvasive procedures 93.1 or treatment recommended f Disabi l i ty Status 115 Receiving or hoping to 46.1 receive benefits No benefits 53.9 1. This figure represents the total number of subjects for whom the information on a particular variable was available - 30 - with the remainder having at least a grade school education. The ethnic background of the subjects was predominantly North American and Caucasian (75.8%) with the remainder representing a wide variety of nationalit ies or ethnic backgrounds. The mean duration of complaint was 8.42 years. However as the standard deviation was large (JSD = 10.33 years) the median, 4 years, is perhaps a more representative figure. For 35 of the subjects the current complaint differed from the chronic complaint. For example, an individual with a chronic, intermittent 10 years history of back pain had a current complaint of 8 months duration. The mean duration of the current complaint was 2.2 years with a standard deviation of 1.52 years. The locus of complaint was primarily the lower back (66.7%). A smaller proportion of subjects presented with back and leg pain (25%) while an even smaller group presented with lower back and hip pain or hip pain alone (8.3%). Thirty-two percent of the subjects had had at least one prior back operation. Of these, seven had had at least two operations and one unfortunate individual had had twelve operations. Chi-square analysis for d isabi l i ty status (2 x 2X^) indicated that the distribution for sex over the categories of d isabi l i ty status was unequal "^- 2( 1) = 9 , 4 1 » p ^ • u u 5 « M o r e wales than females were receiving or expecting to receive some form of financial compensation for their continuing pain problem. Conversely, almost twice as many women as men were not receiving or expecting to receive any benefits (see Table 2). Each individual received at least one diagnosis, 69 received two and - 31 - TABLE 2 2 x 2 Contingency Table for Disabi l i ty Status (Benefits or No Benefits) by Sex1 Receiving or Hoping No Benefits to Receive Benefits Male 34 22 Female 19 40 X 2 9.41, p < .005 1. (missing data on five subjects) - 32 - 17 received three. Of the 206 diagnoses or opinions given, 77 were based on hard physical signs such as evidence on an X-Ray or other diagnostic test. Twenty-two were attributed to "mechanical" problems and 20 were attributed to chronic benign pain syndrome. Twenty-six were attributed to l i f e style problems such as obesity and the remaining diagnoses were too various to categorize meaningfully (see Appendix 1). Each patient received at least one recommendation for disposition, 87 individuals received two, 32 received three, f ive received four and two individuals received five recommendations. Of the total of 242 recommendations for treatment only 35 (14.5%) were for further investigation or invasive procedures. The remainder were for noninvasive forms of treatment (e.g., physiotherapy, back education). Experimental Setting The physiotherapy assessments al l took place in the same examination room in the Back Pain C l in ic in Shaughnessy Hospital. A video-camera was mounted on a bracket in the upper l e f t corner of the room directly over the head of the examination table. An adjacent room contained the videotape recording equipment. A wall with a window-sized, curtained, sl iding glass door partition and a door separated the two rooms. A microphone was placed behind the curtain out of view of the subject. Self-Report Measures The McGill Pain Questionnaire (MPQ; Melzack, 1975) verbal descriptor scales consist of 20 groups of words which y ie ld three measures of the affective, sensory and evaluative dimensions of pain (see Appendix 2). A fourth measure, number of words chosen (NWC) is also derived from the - 33 - questionnaire. The MPQ has been shown to be a sensitive instrument in the assessment of pain control methods (e.g., Melzack, 1975; Melzack & Perry, 1975), has shown considerable diagnostic power (Dubuisson & Melzack, 1976) and is a sensitive descriptor of differences in pain experience (e.g., Agnew & Mersky, 1976; Brightman, 1977; Reading & Newton, 1977). The MPQ was administered to each subject under two sets of instruc- tions: 1) acute pain and 2) chronic pain. The "acute" pain instructions asked the subject to f i l l out the MPQ according to his or her recall of the pain experienced during the most painful movement of the experimental protocol. The "chronic" pain instructions asked the subject to f i l l out the MPQ according to his or her recall of the pain experienced on a typical or average day. The Gracely (Gracely, 1979; Gracely, McGrath & Dubner, 1978, 1979) verbal descriptor scales consist of two scales of 15 items each and one scale of 12 items. The three scales assess sensory intensity, unplea- santness and painfulness (see Appendix 3). The scales were developed through the use of cross-modality matching and scaling procedures. They are re l iab le, objective and the sensory and unpleasantness scales have been demonstrated to be d i f ferent ia l ly sensitive to placebo, narcotic and tranqui l l iz ing drugs (Gracely, et a l . , 1978, 1979). While some contro- versy exists concerning their psychometric properties (Hall, 1981; Gracely & Dubner, 1981), the scales are considered to be rel iable for use in describing naturally occurring acute and chronic pain. Rel iabi l i ty coefficients between groups of similar subjects have been reported as - 34 - 0.96 for sensory intensity and 0.89 for unpleasantness. No re l i ab i l i t y data has been reported for the painfulness scale. Videotape Equipment Subject facial expressions were recorded on Scotch T-120 video cassettes, using an RCA 2011/N high sensit ivity black and white TV camera and a Panasonic 1/2 inch, VHS video-cassette recorder. An RCA video time/date generator, Model TC-1440-B, was connected to the video cassette recorder. It provided the videotapes with a digital time display (minutes, seconds, 60ths of a second) so they could subsequently be divided into segments for FACS coding. A Sony portable microphone was used to provide pickup for the audio portion of the videotaped record. For coding purposes, the tapes were played back on an RCA model JD-975 VW 19-inch television monitor. Procedure Subjects were approached by the experimenter to serve as volunteers for the study while they were waiting for their physiotherapy examination. The experimenter introduced herself, provided background information and had each person read the informed consent form which indicated they were to be videotaped (see Appendix 4). The study was described as a review of physiotherapy procedures and no mention was made of a particular interest in facial expression. After questions were answered, each subject was asked i f he or she would be wil l ing to participate in the study. Of those approached, a small proportion (10.4%) refused to do so. Of those consenting, a few were wi l l ing in theory, but were unsure i f they would be in too much discomfort by the - 35 - time the study began. In such cases the physiotherapist was specifically requested to enquire as to the amount of distress the patient was experiencing before proceeding. In addition, i f the physiotherapist thought that the patient was in distress, despite the lack of any verbal indication, she would enquire as to whether the patient would prefer not to continue. Only two patients, both females, indicated that they would prefer not to continue because they were in too much pain. The physiotherapist subsequently agreed with both patient's assessments. All subjects were assessed in the same examination room. The major component of the physiotheraphy exam involved determining the patient's ability to carry out a variety of range of motion tests. These tests were conducted while the patient was standing, sitting and lying down. When the last section was completed, that is, when the patient was lying in a supine position, the physiotherapist announced the commencement of the study procedure. The physiotherapist then consulted with the experimenter (who was seated in the adjacent room, near the glass partition) as to the correct positioning of the patient's head as shown on the video monitor. When this was accomplished satisfactorily, videotaping began. Half of the subjects, selected randomly, were first read a set of instructions designed to enhance the nonverbal expressiveness of the subject. The importance to the assessment of a variety of nonverbal cues was emphasized and subjects were requested to be as expressive as possible of their distress (see Appendix 5). The other half of the subjects received no such instructions. Other than this initial - 36 - procedural difference there were no further ones in the subsequent procedures for the two groups. The physiotherapist then commenced with the f i r s t of a series of four act iv i t ies involving both active and passive movements of the legs (see Appendix 5). At least one of these movements was l ike ly to induce either low back or hip joint discomfort. These movements were presented in the same order for a l l subjects. After completing the standardized protocol the physiotherapist asked the subject to identify which of the four movements had been the most distressing. The subject was told that this movement would be repeated under two sets of instructions: 1) genuine and 2) masked. The former involved a repeat of the painful movement and the latter another repeat of the movement with the additional instruction that the subject was to try to express as l i t t l e discomfort as possible. In addition, each subject was asked to pose an expression of pain. During this condition, the physiotherapist repeated a movement which had not caused the subject any discomfort, in order for the subject to have an event to which to relate the posed expression (see Appendix 5 ) . Al l subjects began with the repeat of the most painful movement, but the order of the masked and posed instructions was counterbalanced randomly across subjects. After the subjects had completed the procedure they were asked to get dressed and to meet with the experimenter. The subjects then viewed the portion of the videotape involving the presentation of the standard protocol and were requested to indicate the point at which they had fe l t the most pain. This frequently involved several reviews of the section to pinpoint accurately the precise moment. Once this was accomplished - 37 - the experimenter copied down the time/date generator (TDG) number sequence which corresponded to this moment. The subject was then requested to f i l l out the MPQ and the Gracely scales to characterize the acute distress experienced during the most painful movement. The subject then f i l l e d out the MPQ scales again to characterize the chronic pain they typical ly experienced on a daily basis. The experimenter sat with the subjects as they f i l l e d out the questionnaires and answered any questions the subjects had regarding specific items. After the questionnaires were completed the subjects were thanked for their participation and any further questions regarding the study were answered. In addition the subjects were informed as to the actual purpose of the study, that i s , a study of the facial expressiveness of pain rather than a response to physiotherapy assessment per se. When the physiotherapist had completed her assessment on the patients (up to three) that she was able to see on that particular occasion she viewed the appropriate portion of the videotape of the patients who had participated in the study. After this i n i t i a l viewing, the physiotherapist was then asked to indicate the point where she recalled having judged the subject as being in the most distress. In order to f ac i l i t a te reca l l , the physiotherapists were asked to make note of this point immediately after completing the standardized protocol with each subject. Thus, the physiotherapist would note, for example, that the patient seemed to experience the most distress during the l e f t straight leg raise. The portion of the tape involving that particular movement would then be reviewed and the physiotherapist would attempt to - 38 - pinpoint the exact moment of the most severe distress. The judgment therefore was based on both the physiotherapist's recall of the subject's verbal and nonverbal behaviour and the information recorded on the tape. Once the precise moment was identif ied the experimenter noted the corre- sponding TDG time. Experimental Design There were four groups. Within each of the two Instructional Set Conditions half the subjects were female and half were male. Thus, there were 60 subjects in each instructional set condition, 30 males and 30 females. Al l subjects were requested to provide a genuine, a masked and a posed expression of pain. The design was therefore a 2 x 2 x 3 factor- i a l : Type of Instruction (nonverbal expressiveness enhancing, no nonver- bal expressiveness enhancing) x Sex x Type of Facial Expression (genuine, posed, masked). FACS Data Coding Eight 6-second segments from each subject's videotape were selected for scoring. The 6 seconds represented a "window" around a particular point in time; i . e . , 3 seconds preceding and 3 second succeeding a par t i - cular event as indicated on the videotape by a TDG number. The f i r s t of these (Segment 1) was a "neutral" segment, in which the subject's face was expressionless and he or she was at rest. The segment was taken from the beginning section of the session when the subject was waiting for the commencement of the standardized protocol. If the subject's face was not neutral during this period, a search was made throughout the videotape until a 6 second segment which met the same - 39 - criteria could be found. The neutral segment served as a reference for the coders when coding other segments and provided information about individual facial structure, lines, wrinkles, etc. which might otherwise influence the rater's judgments of the presence or absence of particular action units. The second segment (Segment 2) was a baseline segment taken from the first 6 consecutive seconds which were scorable of each subject's videotape. (It was possible for the baseline and neutral segments to overlap.) This segment was selected to provide a record of spontaneous facial expression (if any) which would be compared with facial behaviour during other segments. The third segment (Segment 3) corresponded to the 6 second "window" around the instant identified as marking the point the subject had identified as the peak of the discomfort he or she had felt during the most painful movement of the standardized protocol. This segment was chosen so that i t could be compared with the experimenter's and the physiotherapist's judgment of the same moment. The fourth segment (Segment 4) corresponded to the 6 second "window" around the moment which marked the point the physiotherapist had identified as the peak (in her judgment) of the discomfort experienced by the patient during the most painful event. This segment was chosen to be compared with the experimenter's and the subject's judgment of the same moment. The fifth segment (Segment 5) corresponded to the 6 second "window" around the time which represented the experimenter's judgment of the - 40 - moment of greatest facial activity. In order to make this judgment the experimenter reviewed the section of each subject's videotape involving the standardized protocol. After ascertaining which diagnostic movement the subject had identified as being the most painful or uncomfortable, the complete section of the tape involving that particular movement was viewed until the experimenter could identify the point at which the greatest amount of global facial activity was taking place. This segment was chosen to be compared to the subject's and the physiotherapist's judgments of the same moment. The purpose of ascertaining the moments that the subject and the physiotherapist would identify was to validate the experimenter's choice. Thus, i t was assumed that the three segments would overlap substantially or be identical. Visual inspection of the data revealed that this seemed to be the case. Preliminary analyses were planned to determine i f , nonetheless, any differences in AU type or frequency existed among the three segments. The sixth segment (Segment 6) corresponded to the 6 second "window" around the time which represented the exprimenter's judgment of the moment of greatest facial activity during the repeat of the movement previously identified by the subject as being the most painful. This segment was chosen to be compared to the subject's, physiotherapist's and experimenter's judgments of the initial presentation of the most painful movement. The purpose was to determine if repetition of the painful stimulus would result in any habituation effect as measured by reduced facial activity. The seventh segment (Segment 7) corresponded to the masked - 41 - expression of the subject. As almost a l l subjects were quite successful in substantially attenuating or even eliminating most facial act iv ity under the masked instructions, i t was d i f f i c u l t , i f not impossible to determine where the peak of the painful experience occurred. In order to produce a 6 second segment which would be most l ike ly to encompass this peak the following procedure was followed: As the movement involved was completed the physiotherapist would always say "okay". The tape was therefore stopped as close to the beginning of the verbalization of "okay" as possible and the accompanying TDG number sequence noted. Six seconds counted back from this number sequence provided the segment. Reviewing the entire section revealed that this 6 seconds included most of the movement sequence for a l l subjects. The masked segment was compared with the baseline, pain and posed segments. The eighth segment (Segment 8) corresponded to the posed expression of the subject. The 6 second window was placed around the moment that represented the experimenter's judgment of the greatest amount of global facial act iv ity during the pose. This segment was chosen for comparison to the baseline, pain and masked segments. Segments 1 through 7 were then reordered randomly for each subject for coding purposes. The FACS data coders were blind as to the nature of each segment (other than the neutral and baseline segments) as well as to the group membership ( i .e . , Instructional Set) of each subject. The data coders were a ful l -t ime research assistant who coded 40% of the subjects, a part-time undergraduate researcher who coded 60% of the subjects and a graduate research assistant who did the re l i ab i l i t y - 42 - coding. Al l three individuals were thoroughly trained and experienced in the use of FACS, and had successfully met the re l i ab i l i t y c r i te r i a for scoring required by the authors (Ekman and Friesen, 1978a) for cert i f icat ion as proficient FACS coders. Each segment of the videotape ( i .e . , 2-8) was scored for al l 44 facial action units (AUs) and one action descriptor (AD) specified by the FACS system. For each AU or the AD scored in a particular segment, two types of information were derived. F i r s t , frequency or number of occurrences of each AU/AD in a segment was recorded. Second, the onset and offset of each AU/AD appearance was scored so that the duration of a facial action in a segment could be calculated. In order to do this, a procedure known as apex scoring was used. According to Ekman and Friesen (1978a), the apex of an AU/AD refers to "the period during which the movement was held at the highest intensity that i t reached" (p. 145). There are several reasons why scoring the onset/offset of the apex is preferable to scoring the f i r s t indication of appearance/complete decay: F i r s t , r e l i ab i l i t y is l ike ly to be better with apex scoring. Second, Ekman and Friesen suggest that apices may be more c r i t i c a l for defining emotional expression than absolute onset/offset. For example, they suggest that two or more AUs may be regarded as elements of a unitary expression when their apices are observed to overlap in time. The duration scoring was used in the calculation of the r e l i ab i l i t y of the coders. A complete l i s t of the facial action units and the action descriptor specified by the FACS system and used in the present study may be found in Appendix 6. - 43 - Global Expressiveness Ratings The primary rater was a male, part-time research assistant who rated 100% of the subjects. The secondary rater was a female, part-time research assistant who did the re l i ab i l i t y coding, rating 50% of the sub- jects. Both coders were well trained, having spent 10 hours in training sessions and a further 8 hours in practice rating and were able to reach a high level of r e l i a b i l i t y . (Neither coder had any famil iar ity with FACS.) In making the ratings, the raters were informed that a l l the sub- jects to be judged had, indeed, experienced pain and that the task of the rater was to judge the degree to which each subject expressed that pain. The raters were blind as to the group membership ( i .e . , Instructional Set) of the subjects. The ratings were made on a 3 point scale (inexpressive, somewhat expressive, very expressive) according to the presence and intensity of particular signs (see Appendix 7). The rating was to be based on the subject's facial expression during the most painful movement of the standardized protocol. The TDG number sequence for segment 5 (the experimenter identif ied pain segment) was provided for each subject so that the raters could identify the peak of the expression. In order for the raters to have a more general context in which to place each subject's expression of pain, the TDG number sequences identifying the beginning and the end of the entire standardized protocol were also provided. In those instances where a decision between adjacent points on the scale was particularly d i f f i c u l t , - 44 - the raters were instructed to view the section of the tape involving the repeat of the painful movement. - 45 - RESULTS The results of the study are presented in five separate sections. F i r s t , the re l i ab i l i t y figures are given for the global judgments of facial expressiveness and the FACS scoring of subject facial expressions. Second, descriptive stat i s t ics are given for the demographic variables. The third section outlines an analysis of the effects of the independent variables (Sex, Instructional Set) on the various self-report measures of pain sensit iv ity and on the ratings of Global Expressiveness. The purpose of these analyses was to determine the effect of sex on self-report and of sex and instructional set on the ratings judges made of overall facial expressiveness. The fourth section presents an analysis of the FACS data. An analysis was performed to determine which segment from among the subject, physiotherapist and experimenter identif ied segments would be designated as the "pain" segment and whether any habituation had occurred during the repeat of the painful movement segment. A subsequent analysis examined the effects of sex, instructional set, and segment type on the occurrence of AUs during the pain, masked, posed and baseline segments. A factor analysis was conducted to determine i f any meaningful combinations of AU's existed. The f i f t h section presents the regression analyses performed to examine the relationships among selected demographic variables, the self- report variables and selected FACS variables. Due to the large number of analyses performed in the study the issue - 46 - of making Type I errors arose. While an experimentwise error rate of .01 was not used, this more stringent criterion was applied to certain of the more complex analyses in an attempt to reduce the likelihood of making such errors. Thus the MANOVA involving the demographic, self-report and global expressiveness variables; the FACS analyses (excluding the factor analysis); and the multiple regressions all required that the .01 level of significance be reached before rejecting the null hypothesis. Reliabilities Judges' Ratings of Global Expressiveness A reliability check of the ratings of global expressiveness was conducted. The reliability coder scored a random sample (subject to the constraint that equal numbers of each sex and each instructional set were represented) of 50% of the data set scored by the primary coder. In order to determine interrater reliability two types of correlations were used: Pearson's "r" and an intraclass correlation procedure (Haggard, 1958). Haggard suggests that the intraclass correlation statistic, R, is the appropriate measure when only one variable is involved, such as judges ratings of some factor, whereas r, the Pearson correlation coefficient, should be used when two variables (e.g., height and weight) are to be correlated. Thus R is well suited to be used in determining reliability, where the purpose is to determine the similarity of scores, each of the same kind, from a number of individuals. However, as r. is the more commonly used statistic, i t was calculated as well. The intraclass correlation was calculated according to the following formula: - 47 - Between Class Mean Square - Within Class Mean Square R = Between Class Mean Square + (k-1) Within Class Mean Square The resultant intraclass correlation figure was .92 which was well within acceptable limits. The resultant Pearson r_ figure was .87, again well within acceptable 1imits. FACS Scoring of Subject's Facial Expressions Reliability scoring was performed on a random sample of 14% of the data coded by the primary coder, subject to the constraint that all subjects were represented in the reliability check. To meet this criterion, an independent "blind" coder scored one segment chosen at random from the seven videotape segments obtained from each of the 120 subjects. Percent Agreement was calculated according to the formula recommended by Ekman and Friesen (1978a): No. of Agreements x 2 Percent Agreement = Total no. of AU's scored (This form of percent agreement was used rather than one which utilized both occurrence and nonoccurrence agreement. Had the latter been used the reliability would have been inflated as the nonoccurrence of many of the 44 AUs was often far more frequent than the occurrence; - 48 - cf. House, House & Campbell, 1981.) An agreement was scored only i f two coders reported the AU/AD as occurring at the same point in a segment as defined below. For the one AD scored, and the majority of AU's, an agreement was scored i f the time during which an AU/AD was said to have occurred overlapped with that reported by the other scorer. For AU 45 (blink), which is, by definition, brief in duration, the two scorers had to concur within .2 seconds on the time of occurrence for an agreement to be scored. Under this requirement, the overall percent agreement figure was 76%. This is within the range of the figures reported in most other FACS papers. Ekman and Friesen (1978a) reported a frequency reliability figure of 76% as did Ekman, Friesen and Ancoli (1980). The figure reported by Ekman, Hager and Friesen (1981) was high (87%) but this figure was based on very limited set of deliberately performed actions which probably showed less ambiguity. Kappa was lower, i.e., .63. Craig and Patrick (1984) reported an overall percent aggreement figure of 71% and Patrick et al. (in press) reported a figure of 74%. Thus i t can be seen that the reliability figure obtained in the present study lies well within the acceptable range. Demographic Data In order to determine whether any differences existed between males and females on the uncontrolled subject variables (see Table 1), the sex distributions on the variables were examined. A two-way (Sex x Instructional Set) multivariate analysis of variance (MANOVA), which included selected demographic variables, showed that the two groups did ' not differ significantly in terms of age, duration of chronic and/or - 49 - acute complaint or the number of back operations (p < .05) (see Appendix 8a). Chi-square analyses of sex and education level (10 x 2 X 2 ) ; ethnic group (18 x 2X. 2); locus of complaint (4 x 2 X 2 ) ; disposi- tion (16 x 2X 2) or diagnosis (19 x 2*X2) did not yield any signi- ficant results. However, the chi-square analysis for disability status ( 2 x 2 X 2) did show an uneven distribution over the two groups; X 2 ( i ) = 9.41, p < .005 as previously noted. The males were receiv- ing or hoping to receive financial benefits more often than the females. Self-Report Variables Subject responses to the McGill Pain Questionnaire subscales for both acute and chronic pain and to the Gracely pain descriptor scales for induced, acute pain were included as dependent measures in a 2 x 2 (Sex x Instructional Set) Multivariate Analysis of Variance (MANOVA). Judge's ratings of global expressiveness were also included as dependent variables in this analysis. However, the results will be discussed in a separate section. The subscales on the McGill questionnaire consist of the number of words chosen and the sensory, affective and evaluative scales. The Gracely consists of three scales, measuring sensory intensity, unpleasantness and painfulness. The four scores from the "acute pain" administration of the MPQ and the three from the Gracely verbal descriptor scales corresponded to the pain which the patient indicated he or she felt during "the most painful movement" of the physiotherapy protocol. These scores were based on the subject's recall of the pain about twenty minutes after the actual event. The four scores from the "chronic pain" administration of the MPQ corresponded to the - 50 - subject's report of the amount of pain he or she experienced during an average or typical day. The multiple omnibus F_ indicated a marginally significant difference for the main effect of Sex, F ( 1 6 1 Q 1 ) = 2.098, p < .05. To determine which of the dependent measures were responsible for the overall Sex effect, individual multiple comparisons were conducted at the ^ = .01 level using the Bonferroni (BON) procedure on the univariate analyses of each of the dependent variables for the sex effect. (That is, each univariate analysis was tested at the .01/16 or .0006 level for significance.) Results of the BON tests showed that the two sexes differed only on judges' ratings of global expressiveness, F_( 1,116) = 13.53, £ •< .0001) The two sexes did not differ significantly on any of the other self-report variables included in the analysis. The overall tests for the Instructional Set main effect, F _ Q 6 = -590, £ > .50 and the Sex x Instructional interaction effect, f.Qg ̂ Q J ) = .943, £ > .50 were not significant. This analysis is summarized in Appendix 8a, and group means for the dependent measures appear in Table 3. As there were no significant differences between males and females on the self-report data the scores of the two groups were combined in all further analyses. The combined scores of the "chronic" MPQ were compared to the normative data on the MPQ provided by the Shaughnessy Hospital Back Pain Clinic ( F . Gagnon, personal communication, 1984). Single sample t-tests were used to make the comparisons as the combined study scores were not completely independent of the clinic data. The scores TABLE 3 • - 52 Group Means and Standard Deviation (in Parentheses) for Self-Report Measures Measure No Non Verbal Enhancing Instructions Non Verbal Enhancing Instructions Males Females Mai es Females M SD M SD SD SD Acute MPQ 10.57 (4.33) 9.23 (4.45) NWC Acute MPQ 14.77 (6.09) 14.50 (7.02) Sensory Scale Acute MPQ 2.00 (2.24) 1 .60 (1 .96) Affective Scale Acute MPQ 2.60 (1.65) 2.77 (1.56) Evaluative Scale Gracely Sensory 11.33 (2.93) 11.23 (2.70) Scale Gracely Un- 8.73 (4.40) 6.97 (4.05) pleasantness Scale 9.10 (4.35) 7.90 (4.27) 13.57 (7.21 ) 12.27 (7.68) 1.67 (2.12) 1.40 (2.47) 2.37 (1 .56) 2.07 (1 .48) 10.90 (2.94) 11.00 (2.89) 7.37 (4.30) 7.37 (4.23) Gracely Painfulness Scale 8.13 (3.43) 7.20 (3.03) 7.57 (2.65) 7.93 (2.78) Chronic MPQ 12.03 (3.89) 11.43 (4.41) NWC Chronic MPQ 15.53 (6.88) 17.47 (8.12) Sensory Chronic MPQ 3.10 (2.52) 2.57 (2.22) Affective Scale Chronic MPQ 2.70 (1.26) 2.57 (1.38) Evaluative Scale 10.17 (4.38) 10.43 (4.77) 15.77 (7.46) 14.00 (7.11) 2.50 (2.13) 2.23 (2.37) 2.60 (1.04) 2.17 (1 .29) - 52 - TABLE 4 Means and Standard Deviations (in parentheses) for the "Acute" and "Chronic" MPQ scale scores and for the Normative Data from the Shaughnessy Hospital Back Pain Clinic Measure Acute Chronic Back Pain Clinic (n = 134) M SD M SD M SD Number of words chosen 0 (NWC)1^ 9.2 (4.40) 11.17 (4.36) 12.37 (4.03) 1 ? Sensory ' 13.78 (7.00) 15.54 (7.42) 17.20 (7.50) Affective 1 , 2 1.67 (2.19) 2.60 (2.31) 4.07 (4.30) 2 Evaluative 2.45 (1.57) 2.51 (1.24) 3.20 (2.00) 1. The difference between the "acute" and "chronic" administation of this scale was significant at the £ < .005 level. 2. The difference between the "chronic" score and the Back Pain Clinic normative data of thhis scale was significant at the £ .01 level. - 53 - for all four scales of the MPQ were significantly lower for the combined study sample than for the clinic normative data. (NWC 1(1,119) = -7.86, £ .01; Sensory 1(1,119) = -5.33, £ .01; Affective 1(1,119) = -11.97, £ < .01; Evaluative 1(1,119) = 9.78, £ * C .01). An Hotel ling 1"2 analysis was conducted to determine i f any significant differences existed between the scores on the "acute pain" MPQ and the "chronic pain" MPQ. The overall test was significant F_(4,116) = 13.15, £ -<C .0001 (see Appendix 8b). Multiple comparisons at the .05 level using the BON procedure indicated that for three of the dependent variables, the NWC, Sensory and Affective scales, the scores were significantly lower for the "acute" MPQ as compared to the "chronic" MPQ. There was no significant difference between the two types of MPQ administration for the Evaluative Scale. These results are presented in Table 4. Intercorrelations Among the Self-Report Measures As the test for homogeneity of variance was not significant (Box's M, £(198,28955) = 1.03, £ < .10) the self-report data was collapsed across sex and instructional set in order to construct a correlation matrix designed to investigate the relationships among the scores on the "acute pain" MPQ, the "chronic pain" MPQ and the Gracely. (See Table 5). The correlations among the scales of the two self-report measures of the subject's acute pain felt during the physiotherapy protocol (i.e., the "acute MPQ" and the Gracely) all showed significant positive correlations with one another. However, the two sensory scales correlated more highly with the other scales measuring affective distress and subjective overall I - 54 - TABLE 5 a. Correlations Between the Acute MPQ and the Gracely Gracely Sensory Unpleasantness Painfulness NWC .40* .46* .36* Acute Sensory .29* .37* .31* MPQ Affective .45* .50* .46* Evaluative .58* .59* .57* *£< .001 b. Correlations Between Acute MPQ and the Chronic MPQ Chronic MPQ NWC Sensory Affective Evaluative NWC .70** .59** .53** -.04(ns) Acute Sensory .65** .63** .40** -.03(ns) MPQ Affect- ive .50** .40** .58** -.01(ns) Evalu- ative .21(ns) .17(ns) .27** .23* * £< .05 **£< .005 / \ \ - 55 - intensity than they did with one another. This suggests that each of the two scales is measuring a rather different aspect of the sensory experience than the other. The McGill items have been analyzed to yield as many as five factors, for example, Crockett, Prkachin & Craig, 1977. In such analyses the sensory scale often loads on two factors. Thus, i t is possible that had a four or five factor solution been employed in the present study, the correlation between the Gracely sensory scale and one or other of the refactored McGill sensory scales would have been higher. The correlations among the scales of the "acute pain" MPQ and the "chronic pain" MPQ showed, for the most part, significant positive correlations. The two evaluative scales correlated only modestly with one another in comparison to the relatively substantial correlations between the two sensory scales, the two affective scales and the number of words chosen (NWC) on each administration of the questionnaire. Global Expressiveness Analysis As previously noted, the main effect for Sex was marginally significant in the 2 x 2 (Sex x Instructional Set) MANOVA which included the variable of global expressiveness. The multiple comparisons for the Sex effect revealed that i t was this variable which was statistically significant F(l, 116) = 13.53, £ < .0001. The mean rating for males was 1.93 whereas that for females was 2.40. An analysis of proportions (Glass and Stanley, 1970) clarified the picture. More males than females, 1 = 3.47, £ < .01 were rated as inexpressive and more females than males, 7. = -2.48, £ < .05 were rated - 56 - as very expressive. There was no significance in the numbers of males and females rated as somewhat expressive (see Table 6). Analysis of FACS Data Identifying the Principal Pain Related Action Units As mentioned previously, each of the seven video segments was coded for the 44 AUs and one of the 20 ADs specified by the FACS system. Before proceeding with the statistical analysis of these data i t was necessary to identify within this large quantity of dependent measures those specific to the phenomena of major interest. First the one AD (talking) was eliminated because i t was of no interest in the study. Then the infrequently occurring AUs were removed from further consideration. The inclusion criterion for AU frequency was an average of more than six occurences of an AU averaged over all seven segments for all 120 subjects (i.e., more than 6/840). An AU was eliminated i f i t showed a lower average frequency than this when occurences for all 120 subjects were summed and then averaged over all segments. Of the 44 scorable actions specified by the FACS system, 14 remained after these exclusion criteria were applied (see Table 7). As noted, seven videotape components were identified for each subject: baseline, subject identified pain, physiotherapist identified pain, experimenter identified pain, repeat of the painful movement, masked pain and posed pain. Two forms of agreement were calculated for the subject, physiotherapist and experimenter identified pain segments. The first was agreement on the most painful movement and the second was agreement on the moment of greatest distress within the chosen movement - 57 - TABLE 6 Facial Expressiveness Rating Frequencies Inexpressive Somewhat Expressive Very Expressive Total Sex Male 19 26 15 60 Female 4 28 28 60 Total 23 54 43 120 TABLE 7 " ~ Action Unit (AU) Categories Remaining After Application of Exclusion Cr i ter ia ; Frequency Means and Standard Deviations (in Parentheses) per Subject for Baseline, Pain, Masked and Posed Segments AU Baseline Pain Masked Posed M SD M SD M SD M SD 1 (inner brow raise) .108 (.338) .183 (.41) .05 (.219) .225 (.493) 2(outer brow raise) .117 (.371) .192 (.436) .05 (.219) .225 (.476) 4(brow lowerer)l .067 (.250) .292 (.509) .167 (.417) .592 (.642) 6(cheek r a i s e)l .075 (.295) .233 (.463) .050 (.254) .475 (.621) 7(1 ids t i ghtU .008 (.091) .142 (.416) .10 (.328) .158 (.389) 10(upper l ip r a i s e)l 0.0 (0.0) .108 (.384) .008 (.091) .15 (.496) 12(1ip corner pul1 )1 .133 (.429) .258 (.572) .117 (.371) .725 (.84) 17(chin raise) .042 (.239) .083 (.18) .042 (.239) .83 (.278) 18(1ip pucker) .025 (.203) .075 (.295) .017 (.129) .05 (.314) 20(1ip stretch) .050 (.219) .083 (.278) .008 (.091) .125 (.401) 25(1 ips partU .342 (.667) .575 (.729) .208 (.428) .567 (.753) 26(jaw drop)l .292 (.640) .35 (.617) .25 (.53) .35 (.544) 43(eyes closedH .100 (.363) .358 (.547) .142 (.373) .475 (.733) 45(blink)l 4.242 (2.948) 4.033 (3.07) 3.008 (2.627) 2.95 (2.863) lConsistent with Craig and Patrick's (1985) and/or Patrick, et a l . ' s (in press) findings. - 59 - section of the tape. Agreement in the latter case was considered to exist i f the 6 second segments overlapped. At least two of the three individuals (subject, physiotherapist and experimenter) agreed 96 percent of the time on which movement was the most painful. At least two of the three individuals agreed on the 6 second segment 95 percent of the time. A one-way MANOVA (subject, physiotherapist, experimenter identified segments, repeat of the painful movement segment) using, as dependent variables, the 14 AUs remaining after the exclusion criterion was performed. The overall F test was not significant indicating that there were no significant differences among the four segments. Segment 3 (experimenter identified pain segment) was chosen for use in the subsequent analysis. Group Comparisons A 2 x 2 x 4 (Sex x Instructional Set x Segment) repeated measures MANOVA was performed using, as dependent variables, the 14 AUs remaining after the exclusion criteria were applied. The first two factors in the analyses were the male/female and instruction/no instruction dimensions. The third factor was the repeated measures factor: the MANOVA compared facial activity of the different groups during the four segments; pain, masked, posed and baseline expressions. The only significant main effect was the segment effect, F_̂ 2 944) = 6.83, £ .0001; none of the interactions was significant (see Appendix 9). BON comparisons for the segment effect .01/14 level) revealed a significant difference on nine of the dependent - 60 - measures: AUs 2, 4, 6, 7, 10, 12, 25, 43, and 45. (See Table 8) Tukey comparisons were then carried out for each of the dependent measures at the .0007 level (i.e., .01/14). (A linear interpolation based on tables by Harter (1960) was used to generate the values for the .0007 level.) Only two AUs (4 and 43) occurred significantly more frequently in the pain segment than in the baseline segment. No AUs occurred more frequently during the masked segment than in the baseline segment but one AU (45) occurred significantly less frequently. Four AUs (4, 6, 12 and 43) occurred significantly more frequently and two AUs (7 and 10) were marginally more frequent in the posed segment than in the baseline segment. One AU (45) occurred significantly less frequently during the posed segment. One AU (25) occurred significantly more frequently in the pain than in the masked segment. In addition two AUs (43 and 45) occurred marginally more frequently in the pain segment. Three AUs (4, 6 and 12) occurred significantly less frequently in the pain than in the posed segment. One AU (45) occurred significantly more frequently in the pain segment (See Table 9 and Appendix 10). Factor Analysis of FACS Data The 14 AUs remaining after the application of the frequency criterion (i.e., those which occurred an average of greater than six times for all subjects over all seven segments) were used as dependent variables in the factor analysis. The purpose of the analysis was to explore the possibility that some underlying pattern of relationships might exist for the AUs. In particular, i t was of interest to JABkLJ. i ' b i " AU's Appearing More Frequently During Painful Movement, Masked and/or Posed Segments: Frequency Means and Standard Deviations (in Parentheses) for Baseline, Masked and Posed Segments AU Baseline Pain Masked Posed M SD M SD M SD M SD 2(outer brow raise) .117 (.371) — — — — — — 4(brow lower) .067 (.250) .292 ,(..509) . ___ — .592 (.642) 6(cheek raise) .075 (.295) — .475 (.621) 7(1 ids tight) .008 (.091) — — --- .158 (.389)2 10(upperlip raise) 0.0 (0.0) — — - .150 (.496)2 12(1 ip corner pull) .133 (.429) — — .725 (.84) 25(1ip part) .342 (.667) — — — — — — 43(eyes closed) .100 (.353) .358 (.547) .475 (.733) 45(blink) £ 4.242 (2.948) — — 3.008 (2.627) 2.95 (2.863) 1. AU 45 occurred less frequently in the masked and posed segments than during the baseline segment. 2. This AU occurred marginally more frequently as compared to baseline. TABLE 9 _ 62 - Post Hoc Comparisons (Tukey procedure) of the Segment Effect Dependent Measures: AUs 2, 4, 6, 7, 10, 12, 25, 43 and 45 Compari son AU Segment 5bvs 2a Segment 7 C vs 2 Segment 8 d vs 2 Segment 5 vs 7 Segment 5 vs. 8 2 9(4,347) =2.34 9(4,347)=-2.09 9(4,347) =3.38 9(4,347) =4.44 9(4,347)= -1.03 4 9(4,347); =6.08* 9(4,347)=2.70 9(4,347): =14.19* ' 9(4,347) =3.38 9(4,347)= -8.11* 6 9(4,347) =4.27 9(4,347)=0.68 9(4,347) =10.81* 9(4,347) =4.95 9(4,347)= -6.54* 7 9(4,347-) =4.79 9(4,347)=3.29 9(4,347): =5.36** 9(4,347) =1.50 9(4,347)= -0.57 10 9(4,347) =3.72 9(4,347)=0.28 9(4,347): =5.17** 9(4,347) =3.45 9(4,347)= -1.45 12 9(4,347)^ =2.50 9(4,347)=0.32 9(4,347): =11.84* 9(4,347) =2.82 9(4,347)= -9.34* 25 9(4,347) =4.48 9(4,347)=2.58 9(4,347) =4.33 9(4,347) =7.06* 9(4,347)= -0.15 43 9(4,347): =6.0* ' 9(4,347)=0.98 9(4,347): =8.72* 9(4,347): =5.02** 9,4,347)= -2.72 45 9(4,347) =0.21 9(4,347)=-6.47* • 9(4,347): =-6.34* 9(4,347) =5.37** 9(4,347)= 5.68* a = Baseline Segment * £ > .0007 b = Pain Segment **= marginal c = Masked Segment d = Posed Segment - 63 - determine i f the AUs found to be of interest in previous research (Craig & Patrick, in press; Patrick et al., in press) would group together in any meaningful way. The correlation matrix of the 14 AUs was subjected to a principal - component analysis and to a maximum-likelihood common factor analysis. The purpose of the component analysis was to obtain the latent roots of the correlation matrix for use in deciding on the correct number of factors to retain. The Kaiser-Guttman rule of retaining as many factors as there are latent roots of the correlation matrix which exceed 1.0 suggested that six factors be retained. Cattell's (1966) scree test on the latent roots gave no clear indication as to the number of factors to be retained. The results of the likelihood ratio tests associated with the maximum-likelihood common factor analysis yielded the following results: For three factors, X ^ (52) = 76.32, £ < .05; for four factors, X 2 (41) = 49.90, £ <£ .10. From these results i t was decided to test hypotheses that either three or four factors might provide the most meaningful interpretation of the data. An unweighted least-squares common-factor solution was obtained for each of the two previous solutions. Each solution was rotated (using the Kaiser-Harris procedure) to the oblique simple structure which allowed the clearest interpretation. The four factor solution transformed using the Kaiser-Harris procedure (with the power parameter, £, set to .50) yielded the optimal results. This solution is presented in Table 10 and the intercorrelations among the four - 64 - factors are shown in Table 11. The four factors were tentatively labelled according to the AUs loading most heavily on them or on the basis of a subjective impression of the underlying expression. The labels are as follows: Factor I - eye brow raise; Factor II - smile; Factor III - pain; Factor IV - lower face actions probably related to speech. Relationships Among the Variables Relationship of Selected FACS AUs and Global Expressiveness A full stepwise multiple regression was performed to examine the relationship between judges' ratings of global expressiveness and facial activity in the pain segment. The predictor variables were the six AU frequency variables which showed the greatest difference in the expected direction between baseline and the genuine pain segment: AUs 4, 6, 7, 10, 25 and 43. The criterion was the judge's rating of global expressiveness assigned to each subject. As males and females differed only marginally on the criterion variable of global expressiveness a combined regression analyses was performed for both groups. The regression coefficient was significant up to the second "step" of the analysis, F(2, 117) = 21.16, p < .0001, at which stage AUs 43 and 6 had entered the equation (see Appendix 11a). Judge's ratings of global expressiveness were significantly predicted by the frequency of AU 43. AU 6 significantly improved the prediction, but AUs 4, 7, 10 and 25 were irrelevant. Relationship of Selected FACS AUs and Self-Report A series of full stepwise multiple regressions was performed to - 65 - TABLE 10 Oblique Primary Factor Pattern Matrix at Optimal Position of Simple Structure AU C-l C-2 C-3 C-4 01 0.985 0.104 0.048 0.036 02 0.914 0.064 -0.009 -0.030 04 0.371 0.095 0.685 -0.097 06 -0.092 -0.968 0.103 0.136 07 0.140 -0.060 0.160 -0.031 10 -0.150 0.011; 0.563 -0.040 12 0.141 -0.613 0.022 -0.041 17 0.053 -0.013 0.109 0.278 18 -0.001 0.013 0.097 0.683 20 -0.054 -0.018 -0.080 0.307 25 -0.069 -0.060 0.291 0.324 26 0.022 0.085 -0.050 0.227 43 0.141 0.024 0.193 0.000 45 0.082 0.135 -0.098 -0.050 TABLE 11 Intercorrelation Matrix Among the Primary Factors 1 1.000 2 0.101 1.000 3 -0.004 0.130 1.000 4 -0.027 0.058 0.039 1.000 - 66 - examine the relationship between the self-report variables and facial activity in the pain segment. The predictor variables were again the six "pain" AUs. The criterion variables were the four McGill scales (acute instructions) and the three Gracely verbal descriptor scales. As there were no significant differences between males and females on these variables the analyses were performed on the combined data for the two groups. Only those analyses which resulted in at least one significant regression coefficient will be reported. Relationship of the "Pain" AUs and the McGill Scales Evaluative Scale The regression coefficient was significant only at the first "step" of the analysis, FJ 1,118) = 15.16, £ < .0005, at which stage only AU 6 had entered the equation (see Appendix lib). Relationship of the "Pain" AUs and the Gracely Scales Painfulness Scale The regression coefficient was significant only at the first "step" of the analysis, F_( 1,118) = 11.72, £ < .001, at which stage only AU 6 had entered the equation (see Appendix 11c). Relationship of Selected Demographic Variables, Self-Report of Chronic Pain and Global Expressiveness Two further full stepwise multiple regressions were performed to examine whether the judges' ratings of global expressiveness were related to two demographic variables and the self-report of chronic pain. The predictor variables were disability status, duration of complaint, and the four scales of the MPQ (chronic pain). The - 67 - criterion was the judge's rating of global expressiveness assigned to each subject. As noted previously, males and females differed only marginally significantly on the criterion variables of global expressiveness. However, as they differed significantly on the predictor variable of disability status separate regression analyses were performed for each group. There was no relationship between the demographic variables, self-report and global expressiveness for either males or females. The inclusion limits (£ < .01) were reached at the first step of the analysis for each group. - 68 - DISCUSSION The demographic data are discussed first followed by a discussion of the findings for the self-report variables. The results of the global expressiveness ratings are discussed next, together with a discussion of the relationship between these ratings and several demographic variables. This is followed by a discussion of the FACS analysis of the facial behaviour in the genuine pain, masked and posed segments and a discussion of the factor analysis of the FACS data from the genuine pain segment. A discussion of the relationships of the six AUs, which showed the greatest difference between the genuine pain segment and baseline, and global expressiveness and between the same AUs and the self report variables follows. Finally, a summary and an analysis of the implications for future research completes the discussion. Demographic Variables The characteristics of the sample used in the present study were similar to those used in several other studies of low back pain populations (e.g., Garron & Leavitt, 1979; Keefe & Block, 1982) in terms of age and education level. There were no sex differences on the demographic and clinical variables with the exception of disability status. The difference between males and females in terms of who was receiving or hoping to receive financial benefits was not unexpected as i t reflected the situation of the general Back Pain Clinic patient population. The proportions were somewhat different as only approximately twice as many males as females were being or hoping to be - 69 - financially compensated in the present study, as compared to almost three times as many males as females in the general clinic population (F. Gagnon, personal communication, 1984). Self-Report Measures McGill Pain Questionnaire There were no differences between males and females on either the "acute" or the "chronic" administration of the MPQ verbal descriptor scale (Sensory, Affective and Evaluative scales and the Number of Words Chosen scale). It was difficult to find published data with which to compare the MPQ scale scores obtained in the present research. Most of the published studies of low back pain have focussed on refactoring the individual items on the questionnaire and have not reported the scores on the three factors suggested by Melzack (1975). Melzack (1975) does report data on a very small (n = 14) sample of low back pain patients and the scores were similar to those of the present study on the "chronic" administration of the MPQ. It was possible to compare the scores of the "chronic" MPQ to the data collected by the Back Pain Clinic (F. Gagnon, personal communication, 1984). The scores provided by the clinic were significantly higher than those obtained in the present study. A possible source of this discrepancy was the differing demand characteristics of the two administrations of the MPQ, that is, the clinic's versus the present study's. In f i l l i n g out the MPQ for the clinic assessment i t seemed the individual would likely wish to emphasize his or her discomfort as part of an overall strategy to obtain help. - 70 - i When f i l l i n g out the MPQ for research purposes there would be no additional hope of potential treatment, thus the patient would tend to present a more candid picture of their discomfort. In partial support of this explanation, Dworkin (1970), in a study of absolute pain thresholds and pain tolerance, found that measures of these three variables were accompanied by significantly lower stimulus intensity measures in a dental setting compared to the laboratory. More studies need to be done oh the impact of demand on reports of pain in clinical settings. Another possible source of discrepancy was the method of administration. In the clinic the patient either sat down at a computer terminal and typed in the responses or filled out a form. In either case there was no immediate help available to answer questions or to emphasize the instructions that words need not be chosen from each of the 20 groups. In the present study the experimenter sat with the subject as he or she fil l e d out the questionnaires, clarified the instructions, answered questions about the meaning of certain words, etc. It was possible that this latter procedure resulted in subjects choosing only those words they were sure described their discomfort. The clinic procedure may have resulted in additional words being chosen because the patient had a "better be safe than sorry" approach to f i l l i n g out the questionnaire. The scores on the "acute" and the "chronic" administrations of the MPQ differed significantly. The Number of Words Chosen, the Sensory and the Affective scale scores were all lower for the acute administration. - 71 - One possible explanation was that the pain suffered during the most painful movement was indeed less than that usually experienced on a day to day basis. As regards the Affective scale score i t seemed likely that the pain experienced during a movement designed specifically to induce low back pain discomfort (if that was the problem area) may have had quite a different emotional impact from that experienced during more mundane activities. The patient may not have been too surprised that certain movements in the clinic assessment elicited pain and emotional factors may not have been too important. However, when an action that is frequently performed causes pain, for example, the forward extension of leaning over the sink, or counter, i t is probably more distressing to the patient because i t interferes with the ability to carry out normal activities. McGill Pain Questionnaire and the Gracely Scales The relationship between the "acute" administration of the MPQ (Melzack, 1975) and the Gracely scales (Gracely et al., 1978; 1979) was also examined. This comparison was of interest as the two were developed using different procedures. The MPQ was developed using category scaling procedures whereas the Gracely scales were developed using cross modality matching procedures. All the scales correlated positively and significantly with one another (see Table 5), but close inspection reveals that the correlations ranged only from .29 to .59. The lowest correlation was between the two scales measuring the sensory aspect of the pain experience and the highest was between the MPQ evaluative scale and the Gracely unpleasantness scale (a measure of affective distress). - 72 - The MPQ and Gracely measures of affective distress correlated at the .50 level and the MPQ Evaluative scale and Gracely Painfulness Scale correlated at the .57 level. While there is some overlap between the MPQ and the Gracely Scales there is some indication that they are also measuring different facets of each aspect of the pain experience. Global Expressiveness The manipulation to influence overall or global facial expressiveness of painful distress was not successful; that is, the instructions to try to be as expressive as possible of painful distress did not influence measured expressiveness. Previous research (Ekman, 1977; Kleck et al., 1976; Kraut, 1982) has shown that people tend to attenuate their facial expressiveness in the presence of an observer due to display rules or other factors. It was thought that by emphasizing the desirability of being as nonverbally expressive of painful distress as possible, the attenuation problem might be circumvented. There were several possible explanations as to why this manipulation was ineffective. First, i t was possible that the particular instructions used were simply not powerful enough to overcome the effect of the display rules. Alternatively, i t may have been that the demand characteristics of the situation were such that all the subjects were already as expressive as possible. However, this seemed unlikely given that 23 subjects were rated as inexpressive. Had the manipulation been a powerful one, presumably the individuals who received the nonverbal enhancing instructions would have been rated as being at least somewhat - 73 - expressive. It may have been that the display rules which govern facial expressiveness of pain are so overlearned that the brief instructions employed here failed to circumvent them. However, only further research utilizing a more powerful manipulation than that employed in the present study would be able to address the question. The present study found that judges rated women as marginally more expressive of painful distress than the men. Almost twice as many women were rated as being very expressive whereas almost five times as many men were rated as being very inexpressive. These results were consistent with the findings of previous research which found that judges correctly categorized slides of emotional expression exhibited by females more often than those of males (Buck, Miller & Caul, 1974) and that women evidence more facial electromyographic activity than men while imagining several different emotion eliciting situations (Schwartz, Brown & Ahern, 1980). The relationship of global expressiveness to self-report of pain ("chronic" MPQ), duration of complaint and disability status was also examined. The results indicated that no consistent relationship existed for either the males or the females. It is of particular interest that no consistent relationship between the predictor variables and the criterion variable was found for the males. An examination of the correlation matrix revealed that all six predictor variables correlated at a low level with the criterion variable. The lack of a relationship between disability status (that is, compensation) and global expressiveness was noteworthy because slightly - 74 - over half of the males were receiving or expecting to receive financial compensation. Receiving or hoping to receive financial benefits may be a factor which serves as a reinforcer of pain behaviour although the evi- dence regarding the effect of compensation on pain behaviour is conflict- ing (Kremer, Block & Atkinson, 1983). It would seem that, for this par- ticular group of males at least, reinforcement for pain behaviour (in the form of global facial expressiveness) does not bear any relation to whether compensation is involved thus adding to the aforementioned conflict in the literature. FACS Analysis of Facial Expressive Behaviour The 14 AUs remaining after the exclusion criteria were applied were used as dependent measures in a global analysis. The purpose of the ana- lysis was to determine whether facial expressions were systematically influenced by the two independent variables in the study (Sex, Instruc- tional Set) or differed during the four segments of interest (baseline, genuine, masked and posed). The two independent factors (Sex, Instructional Set) were found to have no effect on the occurrence of the 14 AUs, either singly or inter- actively. Only the repeated measures variable "Segment", had any effect, with nine of the 14 AUs being significant. It was not expected that the instructional set manipulation would have a particular effect on the FACS variables. The manipulation was aimed, at enhancing overall or global expressiveness and i t was not known whether this might be achieved by means of an increase in the absolute amount of facial activity, by increased intensity or by both. As the - 75 - manipulation was unsuccessful in terms of altering global facial expressiveness, the absence of results for the specific FACS variables was not surprising. It was interesting that sex had no effect on the frequency of the AUs, particularly as males and females differed (at least marginally) in judges' ratings of global expressiveness. It would seem, therefore, that the judges' ratings of males and females were based on some factor other than a difference in frequency of the 14 AUs as investigated here . One possible explanation would be that the global judgments were based on the occurrence of different combinations of the AUs exhibited by the two groups. For example, some AU combinations might conceivably be perceived as "more expressive" and these might be displayed more often by females. Alternatively, any combination of AUs might have been perceived as more expressive than any single AU and perhaps females exhibit more combina- tions than males. Another possibility may be differences in the inten- sity of the AUs with females exhibiting AUs of greater intensity than males. Unfortunately intensity ratings have been unreliable due to difficulties with the scoring criteria, so an analysis was not under- taken. Another possibility was that the duration of single AUs or combi- nations of AUs may have differed between males and females. There may also have been a systematic sex bias for the judges. Regardless of the sex of the judge, males may have been evaluated differ- ently than females even when the same scoring criteria were used. While the judges were not informed by the experimenter that women have been shown to be better encoders of emotion than men (e.g., Buck et al., 1974), they may have held this expectation. If so, any display of facial - 76 - activity on the part of a female may have led them to assign a rating of "somewhat" or "very expressive" whereas a male would have had to demonstrate more activity to obtain such ratings. A further possible explanation related to the issue of individual differences. Subjects varied widely in the manner in which AUs were exhibited and how often a particular AU was displayed when i t was present. The majority of the AUs were eliminated as being too infrequent to be of interest. It was possible that females might have shown a greater variety of these more infrequently occuring AUs, albeit unsystematically and, hence, were rated as more expressive. It was likely that judges' global ratings were based on a number of the aforementioned factors and future research may be able to clarify the situation. The segment factor was significant as noted. Post hoc comparisons were made using as dependent variables the nine AUs which reached significance in the MANOVA. The following comparisons will be discussed: Genuine versus Baseline; Masked versus Baseline; Posed versus Baseline; Genuine versus Masked; and Genuine versus Posed. Genuine Pain Expression This comparison was aimed at identifying the AUs associated with the experience of pain (those AUs significantly more prominant during the administration of the most painful movement as compared to baseline). Only two AUs were found to occur more frequently during the most painful movement: AU 4 (brow lower) and AU 43 (eyes closed). Even these two AUs were not displayed universally by all 120 subjects during the pain - 77 - segment. Only 45 percent of the subjects exhibited one or both of these two AUs. It was surprising that so few AUs were found to be prominent during the pain segment (relative to baseline) given the results of the previous studies. Craig and Patrick (1985) found AUs 6/7, 10, 12, 25, 26/27 and 43/45 to be associated with cold pressor pain, and Patrick et al. (in press) found AUs 4, 6, 10, 45 and 7 (the later only marginally significant) to be associated with pain induced by a brief electric shock. Even LeResche (1982), in a study utilizing FACS to score a series of candid photographs that were less than optimal quality for rating purposes, found evidence of AUs 4, 6, 7, 11, 20, 25, 26, 27 and 43. A possible explanation for the relative lack of results in the present study would be that they were due to differences in the nature of the samples used. Factors such as the use of both sexes versus females only, the wide range of ages and education levels versus more narrow ones, the use of chronic pain patients versus college students, the use of different types of stimuli to induce distress and the different settings used might all account for the difference in the results of the present study. Another possible explanation for the relative lack of results is that the large number of analyses performed in the present study necessitated the adoption of a more stringent criterion of statistical significance to guard against Type I errors. The earlier studies involved a limited number of analyses and thus were able to employ a less stringent level of alpha. A methodological difference made comparison with the first study in - 78 - this series difficult. Craig and Patrick (1985) used combined AU "categories" in an attempt to reduce the overall number of AUs for the purposes of statistical analyses. For example, AU 6 (cheek raise) and AU 7 (lids tight) were combined into one category. The use of combined categories was not necessary in the present study because of the substantial sample size. The most notable difference between the results of the present study and the earlier ones was the failure of AU 45 (blink) to distinguish between the pain segment and baseline in the present study. Despite AU 45 being the most frequently occurring of all the AUs (93 percent of the subjects exhibited i t , with a range of occurrence per subject of one to 21 times), i t occurred less frequently during painful stimulation, albeit nonsignificantly, than during baseline. Craig and Patrick (1985) have postulated that AU 45 may "represent a form of reflexive coping behaviour which appears with increased frequency during any ordeal, independent of other facial activity." Thus, blinking would signal forms of distress other than pain. If the assumption is made that the administration of the most painful movement qualified as "an ordeal", then i t can only be speculated as to why AU 45 did not occur with a greater frequency during this segment. Awaiting the series of range of motion tests could be construed as an ordeal that provoked blinking. Alternatively, a cognitive-based explanation has some plausibility. In the shock study, the subject waited for the shock but had nothing in particular to focus on. In the present study the subject often watched the physiotherapist (despite - 79 - instructions to look at the ceiling) as she guided the limb through a movement. If this focussing of attention were to be considered a cognitive activity then i t might account for the reduced frequency of AU 45, during the painful movement, making i t comparable to the baseline frequency. That is, individuals tend to blink less frequently during cognitive activities than when not actively engaged in cognition (Holland & Tarlow, 1972, 1975). Another plausible explanation has to do with the presence of AU 43 (eyes closed). If the eyes of the subjects were closed more often during exposure to the current painful stimulus than during the previous studies, this would account for the reduced number of AU 45s (AU 43 and AU 45 cannot be scored simultaneously). AU 43 was exhibited by 40 subjects a total of 44 times. The average duration of AU 43 was 1.61 seconds although the range was large (.58 to 6) seconds. It was possible that having one third of the subjects holding their eyes closed an average of one quarter of the time during the pain segment served to decrease the overall number of AU 45s. Another possible explanation related to the nature of the noxious stimulus in the previous studies as compared to the present study. The cold pressor experience appears to be one of a progressive mounting of pain until the subject signals that i t can no longer be tolerated. The electric shock experience is one of brief duration and the intensity can be precisely determined. In the Patrick et al. (in press) study five different precisely controlled current intensities were used. In the present study, the intensity of the noxious stimulus could not be - 80 - measured independently of self-report and the duration of the stimulus probably varied from subject to subject. For some subjects (judging from both their verbal and nonverbal behaviour) the pain appeared to have a very abrupt onset whereas for others i t appeared to onset more gradually. Perhaps these differences in intensity and duration resulted in the "most painful movement" experience being qualitatively different from the cold pressor and shock experiences such that the "reflexive coping" function of AU 45 was not elicited. Masked Expression An attempt was made to determine whether any AUs would be associated with the subject's endeavour to attenuate his or her facial expression of pain as compared to the subject's baseline behaviour. None of the AUs occurred more or less frequently than during baseline, with the exception of AU 45 which occurred significantly less frequently. The reduced frequency of AU 45 might be explained with reference to the cognitive explanation put forth in discussing the similar finding in the genuine pain versus baseline comparison. If the conscious effort involved in attempting to control one's facial expression is conceived of as a cognitive activity then i t would be predicted that the frequency of blinking would be reduced (Holland & Tarlow, 1972, 1975). Generally, i t appeared that subjects were quite successful in suppressing facial activity (as compared to baseline) while experiencing a movement previously identified as being the most painful of the four making up the protocol. Posed Expression Subjects were also requested to pose an expression of painful - 81 - distress. No model or instructions were provided, so the resulting facial expressions presumably represented either the subjects' impressions of what they looked like when in pain or their impression of how other people look when in pain. Four AUs were found to occur more frequently in the posed segment as compared to baseline: AU 4 (brow lower), AU 6 (cheek raise), AU 12 (lip corner pull) and AU 43 (eyes closed). Two AUs occurred marginally more frequently: AU 7 (lids tight) and AU 10 (upper lip raise). AU 45 again occurred less frequently than during baseline; perhaps again due to the increased frequency of AU 43, or due to cognitive activity related to thinking about a pain face. The presence of AU 12 may be related to the fact that a number of subjects appeared to be rather self-conscious about posing an expression or expressed amusement after completing the pose. AU 12 plus AU 6 or AU 7 is the combination associated with "felt" smiles (Ekman & Friesen, 1982). When AU 12 appears in the absence of AU 6 or 7 i t is considered to be a "false" smile. Apparently false smiles do, however, occasionally occur in the presence of AU 6 or 7. When this is the case, the cue then used to determine if the smile is indeed false, is the duration of AU 12, that is, when the duration is less than two thirds of a second or more than 4 seconds, the smile is considered to be false. Ekman and Friesen (1982) suggested that these false smiles fall into two categories: the "phony" smile where nothing much is felt but the individual endeavors to appear as if a positive emotion is felt; and the "masking" smile in which a strong negative emotion is felt and an attempt is made to conceal such a feeling by seeming to feel positively. - 82 - In the present study, according to the aforementioned criteria for felt and false smiles, 35.5 percent of the smiles were felt and 65.5 percent were false. It seemed most likely that the felt smiles were ones of amusement exhibited by subjects after completing the pose. The false smiles seemed to f i t the category of the masking smile, that is, those covering up a negative emotion. In the present instance, the most pro- bable negative emotion being covered up would seem to be embarrassment. Craig and Patrick (1985) reported that AU 12 was one of the set of AUs which differentiated between the cold pressor exposure segments and baseline. This was not the case in the present study or in the results reported by Patrick et al. (in press). In the Craig and Patrick study, AU 12 was most prominent during the first 10 seconds of cold pressor exposure. It would be of interest to determine i f these action units represented felt smiles, perhaps associated with laughter at the initial shock of the temperature of the water, or false smiles associated with an attempt to mask some experience of distress. Alternatively, these smiles may fall into a final category labelled by Ekman and Friesen (1982) as "miserable" smiles. The criteria for such smiles were not as well delineated, but apparently appear to be "superimposed" on a clear negative expression rather than appearing as an attempt to mask such an expression. Up to this point, the assumption has been made that i t is valid to consider that hypotheses about action units such as AU 12, made in the context of studies of emotional expression were applicable to studies of expressions of painful distress. While the pain experience has a - 83 - significant emotional component, i t is not an emotion per se, having sensory and motivational components as well. The exact nature of the relationship between pain and emotion is unclear and any discussion as to the meaning of an action unit such as AU 12 must be considered as tentative. Genuine Pain Expression Compared to the Masked and Posed Expressions The first comparison to be discussed will be that of the Genuine Pain versus the Masked expression. Of the nine AUs serving as dependent variables, only one, AU 25, occurred significantly less frequently during the masked than in the genuine pain segment. Two AUs, 43 and 45, were marginally less frequent in the masked segment. Thus i t would seem that the subjects were only somewhat successful in attenuating their facial expression (relative to the pain segment) when requested to do so. LeResche and Dworkin (1984) have noted the importance of ascertaining whether individuals can conceal facial behaviour. The results of the present study seem to indicate that at least partial concealment is possible. Perhaps a more complete concealment could be achieved i f the individual had more to gain by doing so than merely complying successfully with a set of instructions. With regard to the comparison between the genuine and the posed seg- ments, three AUs occurred significantly more often in the posed segment: AUs 4, 6 and 12. AU 45 occurred significantly less often in the posed segment. The presence of AU 12 in the posed segment relative to baseline has been discussed and the same conclusions are relevant here. Examination of the raw data revealed that AU 6 occurred with AU 12, - 84 - 63 percent of the time in the posed segment. Thus, the majority of AU 6s occurred with a smile of some type and, since smiles were evident only in segment 6, this may well have accounted for the increased frequency of AU 6 as compared to the genuine pain segment. A possible explanation for the greater frequency of AU 4 in the posed segment may be that i t is very salient when present in a naturally occurring expression of painful distress. Hence when asked to pose "a pain face" i t may be one of the main facial behaviours that comes to mind and is thus exhibited more often even though the individual might not exhibit i t in their own spontaneous facial behaviour. The decreased frequency of AU 45 in the posed segment might again be reasonably attributed to cognitive activity. In the posed expression the subject had to visualize a "pain face" and then consciously execute i t , all of which would seem to plausibly constitute a cognitive process. In general, i t would seem that the attempts to mask and pose expressions of painful distress were at least partially successful. Even when considering all the AUs exhibited in each segment (that is those AUs other than AU 45, but including those that were eliminated as infrequent or failing to discriminate between a particular segment and baseline) the pattern remained. In examining the raw frequency data i t was evident that over twice as many AUs were exhibited in the genuine pain segment than in the masked segment. The overall difference between the genuine and posed segments was less notable, but even so almost half again as many AUs were seen in the posed expression. Posed pain faces displayed a more caricatured pattern than the genuine expression. Including AU 45, - 85 - the most frequently occuring AU, blurs the pattern, but even here one and a half times more AUs were evidenced in the genuine expression than were seen in the masked. Factor Analysis of the Genuine Pain Segment The factor analysis was based on the data from the genuine pain segment. The fourteen AUs remaining after the exclusion criteria were applied to the total data base, served as the variables in the analysis. Four factors emerged from the analysis and were labelled either according to the type of AUs loading most heavily on each or else by some apparent underlying construct. Factor one was labelled "eye brow raise" as AUs 1 and 2, whose joint action results in the raising of the inner and outer corners of the brows, loaded most heavily on this factor. Examination of the raw data reveals that the two AUs co-occurred virtually 100 percent of the time. Given that AUs 1 and 2 were not pain-related this factor would not seem to characterize pain expressions. Factor two was labelled "smile" due to the fact that AUs 6 and 12 showed the greatest loadings on the factor. AU 6 actually co-occurred with AU 12 sixteen of the 28 times (or 57 percent) i t appeared during the pain segment. The prominence of this expression during both the genuine and posed pain expressions emphasizes the interpersonal qualities of pain behaviour and the sensitivity to others that people display even when in pain. Factor three was labelled "pain" as two of the AUs which were most prominent on the factor (AU 4 and AU 10) were among those postulated by Patrick et al. (in press) to represent a possible prototypical pain face. - 86 - AU 43 also loaded on this factor although not to as great a degree as AUs 4 and 10. Patrick et al. (in press) have suggested, on the basis of the intercorrelation among a more limited set of AUs, that AUs 4 and 10 in combination with AU 6 or AU 7 might represent "a common reaction to acute noxious stimulation". An examination of the factor pattern matrix (see Table 10) shows that AU 6 loads very lightly on Factor three as does AU 7. AU 6 is more closely associated, as noted, with AU 12 and, as discussed previously, i t is questionable whether AU 12 represents a facial movement expressive of pain. It should be noted that the pattern of correlations revealed by the factor analysis is not an index of co-occurrence, that is the extent to which the occurrences of two or more AUs overlap in time. The factors simply reflect the extent to which these different facial movements occurred together within the same 6 second segment. Thus, i t is of interest to examine the actual co-occurrences of the AUs which load on the different factors. With examination of the raw data i t can be seen that AUs 4, 10 and 43 occurred in some combination with one another a total of 22 times in the pain segment. (Note that AU 4 appeared on 35 occasions, AU 10 on 13 and AU 43 on forty three). These co-occurrences were limited to 18 of the total of 120 subjects; thus once again the lack of universality is underscored. While AUs 6 and 7 did not load in any significant way on Factor three, an examination of their co-occurrences with AUs 4, 10 and 43 does present a somewhat altered picture. AU 6 co-occurred in combination with one or more of AUs 4, 10 and 43 (and in the absence of - 87 - AU 12) 36 percent of the time i t appeared in the pain segment. AU 7 co-occured in combination with one or more of these AUs 59 percent of the time i t appeared. If the two AUs were considered as a category (bearing in mind that they could not co-occur with one another) they appeared in combination with one or more of AUs 4, 10 and 43 forty-four percent of the time. Thus while AU 6 in particular appears more often in combination with AU 12, there remains some support for Craig and Patrick's suggestion that the combination of AUs 4, 10 and 6 or 7 may represent a configuration associated with pain. AU 25 loaded most heavily on Factor four which was labelled "lower face AUs associated with speech". However, i t also loaded to a lesser degree on Factor three . An examination of the raw data showed that AU 25 occurred either immediately before or after AU 50 (mouth movements specifically associated with speech) on approximately 32 percent of occasions i t appeared. It co-occurred with one or some combination of AUs 4, 10 and 43 (in the absence of AU 50) 28 percent of the time i t appeared. Thus i t could be seen that in the case of AU 25, the factor loadings are reflected in the actual co-occurrences. While the factor analysis and the examination of the co-occurences of the various combinations of AUs provided information relevant to the consideration of a possible "prototypical" pain face, i t must be pointed out again that individual differences rather than communalities seemed to be the most prevalent. While 88 subjects exhibited one or more of AUs 4, 6, 7, 10, 25 and 43, only 54 of those subjects exhibited at least one of the three AUs most prominent on Factor three (AUs 4, 10, and 43). In - 88 - fact only 18 subjects actually exhibited two or more of AUs 4, 10 and 43 in combination with one another. Thus i t seemed that this attempt to define a prototypical expression was a failure. Rather, delineating a constellation of AUs seems to be the more useful endeavour. Relationship between the "Pain" AUs and Global Expressiveness In order to examine the relationship between the specific facial expressive behavior obtained from the FACS analysis and judges' ratings of general or global facial behaviour associated with pain a stepwise multiple regression was performed. The predictor variables were the six AUs which showed the greatest difference in the expected direction between baseline and the genuine pain segment: AUs 4, 6, 7, 10, 25 and 43. The criterion variable was the rating of overall or global expressiveness of pain assigned to each subject by a judge. The results are not discussed separately for males and females as the groups differed only marginally in judges' ratings of global expressiveness. A positive significant relationship was found between some of the predictor variables and the judge's ratings of global expressiveness. The magnitude of the multiple R was .52 with two variables participating in the equation. These were, in descending order of weight AUs 43 and 6. The correlation between each of the predictor variables and the criterion was positive (see Table 12). The other AUs were missing from the equation, possibly due to their relatively low correlations with the criterion. In considering the results of the regression analysis, i t seems noteworthy that a relationship was indicated between even two of the - 89 - TABLE 12 Correlations between Judges Ratings of Global Expressiveness and Facial Activity During the Pain Segment AU Rating 4 0.37 6 0.41* 7 0.09 10 0.34 25 0.06 43 0.44* * £ < .0001 - 90 - predictor variables and the criterion given the relatively stringent level of alpha (.01) used as an inclusion criterion in the analysis. In addition the magnitude of the relationship was striking. It should be emphasized that the ratings of global expressiveness were based on the subject's facial behaviour during the entire standardized physiotherapy protocol (albeit with particular attention to be paid to the genuine pain segment) whereas the six predictor variables were obtained from only the 6 second genuine pain segment. Even if only the genuine pain segment had been considered, the judges' ratings were to have been based on more than just the six AUs found to discriminate between the segment and baseline, thus allowing for more consideration of individual differences. Despite these differences however, two of the FACS variables were able to predict to a significant degree the more global ratings. Relationship of the "Pain" AUs and Self-Report To examine the relationship between subjective distress and facial expression a series of stepwise multiple regressions were performed. The predictor variables were again the six AUs whose frequencies which showed the greatest difference in the expected direction between baseline and the genuine pain segment: AUs 4, 6, 7, 10, 25 and 43. The criterion variables were the magnitude of subject self-report associated with the most painful movement, that is the scores on the three scales of the MPQ "acute" as well as the total number of words chosen, and the scores on the three Gracely scales. The results are discussed in terms of the measures of sensory intensity (MPQ and Gracely Sensory Scales), affective distress (MPQ Affective and Gracely Unpleasantness scales), subjective - 91 - overall intensity (MPQ Evaluative scale) and the Gracely Painfulness Scale (a measure intended to approximate the information usually solicited by a clinician). No significant relationship was found between either of the measures of sensory intensity and facial expressive behaviour or for either measure of affective distress and the criterion. A positive, significant relationship was found between the MPQ Evaluative scale and one of the predictors and between the Gracely Painfulness scale and one of the predictor variables. The magnitude of the multiple R for the MPQ Evaluative scale was .34 with only one variable (AU 6) participating in the equation. The correlations were positive with the exception of that for AU 25 which was negative (see Table 13). The magnitude of the Multiple R for the Gracely Painfulness scale was .30 with only one variable (AU 6) participating in the equation. The "correlations were all positive with the exception of AU 7 which was negative (see Table 13). Thus a significant, albeit very limited, positive relationship existed between the subjective report of the overall intensity of the pain experience (as measured by the MPQ evaluative scale) and facial activity and between the Gracely Painfulness scale and facial activity. Effectiveness of the Painful Stimulus An issue relevant to the differences in the AUs found in the present study as compared to previous research relates to the question of whether the subjects in the present study were actually in pain. Unlike electric shock, the intensity and duration of which can be precisely controlled, - 92 - TABLE 13 Correlations between the McGill and Gracely Scales and Facial Activity During the Genuine Pain Segment AU MPQ Evaluative Gracely Painfulness Scale Scale 4 .20 .14 6 .34* .30* 7 .08 -.002 10 .06 .16 25 .02 .08 43 .01 .07 * p < .0005 - 93 - and the cold pressor experience where duration can be controlled and at least the initial discomfort, is well documented, there was no control over the duration and intensity of the painful stimulus in the present research. Moreover, as the pain could be elicited through any of the four movements (depending on the locus of the patient's complaint) the nature of the pain could vary from individual to individual. However, the four movements used in the protocol were chosen because the physiotherapists judged they would produce low back or hip (sacroilliac joint) pain (and would f u l f i l l the requirement that subjects be lying down while they were performed). Despite all the steps taken to ensure that the subjects in the study were, indeed, those individuals who experienced pain, the possibility does remain that some individuals exaggerated their distress. Although it was repeatedly emphasized that the protocol (including the Subsequent masked and posed instructions), was part of a study, the procedure was embedded in the clinic's physiotherapy assessment, and some subjects may have exaggerated their distress as part of their overall attempt to obtain help. If this were the case, i t might provide an alternate explanation for the similarity between the genuine and the posed expression. It might also account for the relatively modest relationship between the six most frequent AUs from the genuine pain segment and the self-report variables (although discontinuities between behaviour and self-report are quite common, for example, Fordyce et al., 1984). It is rare for a patient to be referred to the Back Pain Clinic as a suspected malingerer (C. Solyom, personal communication, 1983) and, even more - 94 - rarely, does the clinical team which assesses each patient come to the conclusion that a patient is malingering. Any individuals who might have been suspected of malingering according to the referral information were screened out and only one subject was considered to be a possible candidate after the team assessment. Summary and Conclusions The demographic characteristics of the sample were similar to those reported in other studies of low back pain patients. The only variable which differentiated between males and females was disability status, that is, receiving or hoping to receive financial compensation. The results of the self-report analysis failed to find any differences between the two sexes, the two instructional groups or any interactive effects. The chronic MPQ scores were significantly lower in comparison to the Back Pain Clinic norms, perhaps due to the subjects being more objective in their assessment of their pain for research purposes. The acute and chronic MPQ scores differed significantly and i t was hypothesized that this might be due either to the chronic pain indeed being more intense, or being evaluated as such because of the greater impact i t had for the patient due to its being elicited by mundane movements. The manipulation to influence global expressiveness was unsuccessful. The most likely explanation was that the instructional set used was not powerful enough to overcome the implicit display rules in effect for the assessment. An instructional set emphasizing more clearly the importance of being expressive of distress with closer questioning of - 95 - the subject to confirm understanding of the instructions might provide a better test of whether the inhibiting effect of having observers present could be circumvented. Women were rated as being marginally more globally expressive than men consistent with previous research (Buck, et al., 1974; Schwartz, et al., 1980). The relationship of a number of self-report and demographic variables was examined with the result that no relationship was found to exist for either sex. It was somewhat surprising that no significant relationship was found for disability status for the men, given the large proportion of men receiving or hoping to receive financial compensation. This contributes to the conflicting results in the literature regarding the role financial benefits play in reinforcing pain behaviours (Kremer, Block and Atkinson, 1983). The instructional set manipulation failed to influence the frequencies of the FACS variables. Sex also did not affect the type or frequency of AUs observed. The latter finding was of particular significance as the two sexes did differ marginally in terms of the ratings of global expressiveness. A number of possible explanations for this discrepancy between the FACS data and the judges' global ratings were put forth. Among these was the possibility that differences between males and females in terms of the configuration or intensity of the AUs might account for the judges' ratings, or the possibility that females exhibited a greater variety of AUs (including those eliminated as being too infrequent for analyses). Further research investigating a possible systematic sex bias on the part of the judges might clarify the results. - 96 - The repeated factor (segment) was significant and post hoc comparisons elucidated the differences. Analysis of the genuine pain segment (as compared to baseline) resulted in only two AUs being identified as being more frequent in the former: AUs 4 and 43. It was important to point out that these AUs did not occur universally, either alone or in combination with each other. That so few AUs were found to distinguish between the pain and baseline segments was in contrast to the results of the previous research (Craig & Patrick, 1985; Patrick et al., in press). Several possible explanations were discussed including that of differences between the present study and the earlier ones and the use of a more stringent criterion for rejection of the null hypothesis. The other major difference was the failure of AU 45 (blink) to discriminate between the pain and baseline segments in the present study. Several explanations were discussed, the most likely being that the greater . frequency of AU 43 (eyes closed) in the pain segment might account for the reduced number of 45s as the two cannot be scored simultaneously. The relative paucity of results in the present study is of interest given the query (e.g., LeResche & Dworkin, 1984) as to whether chronic and acute pain populations would differ in terms of facial expressive behaviour. Chronic pain patients have been shown (e.g., Fordyce, 1976) to evidence a frequency and/or variety of behaviours not generally seen in acute patients. The results of the present study seem to contradict this evidence as less facial activity (as measured by FACS) was observed than in the previous studies which used an acute pain stimulus with a nonchronic sample. - 97 - Analysis of the masked expression as compared to baseline revealed that the subjects were successful in attenuating their facial expressive behaviour when instructed to do so. In fact, one action unit (AU 45) occurred significantly less frequently than during baseline. Analysis of the posed expression as compared to baseline revealed a constellation of AUs including those found in the pain segment. The presence of one of these, AU 12 (lip corner pull), was tentatively explained by reference to Ekman and Friesen's (1982) research on "false" and "felt" smiles. It was suggested that the presence of AU 12 in the Craig and Patrick (1985) study might be explained in the same way. One AU (25) occurred significantly less frequently and two marginally less frequently in the masked segment as compared to the genuine pain segment. These results provided limited support for the position that subjects could attenuate facial activity on a voluntary basis. Three AUs (4, 6 and 12) occurred significantly more frequently and one AU (45) significantly less frequently during the posed segment as compared to the genuine segment. The increased frequency of AU 6 appeared to be related to the presence of AU 12 in the posed segment. It was suggested that the increased frequency of AU 4 was possibly due to its being particularly salient to an observer and thus more easily remembered when posing an expression. This would suggest that a posed expression might be more stereotypic or caricatured than a genuine, "felt" expression. Further research, perhaps contrasting each subject's personal genuine "pain face" with his or her posed "pain face" might - 98 - substantiate this possibility. If so i t would provide a clue as to whether an individual is simulating distress or not. The decreased frequency of AU 45 was again attributed to the cognitive activity presumably involved in producing a facial expression indicative of pain. This decreased frequency, in combination with the increased frequency of other AUs might serve as an additional clue in the assessment of possible deception. The factor analysis of the AUs remaining after the initial exclusion criterion was applied resulted in a four factor solution. Factor three was of the greatest interest as two of the AUs which loaded heavily on i t had been suggested (Patrick et al., in press) as forming part of a "prototypical pain expression". Analysis of the actual co-occurrences of the AUs in the pain segment did lend some support to this hypothesis. However, further examination of the co-occurrences of the Factor three AUs revealed that they were limited to a small number of subjects. Little evidence was found to suggest that the pain AUs combined with any regularity into a unitary, prototypic pain expression. Rather, the evidence suggests the existence of a constellation of AUs, any one of which might be displayed by an individual in pain. The relationship between the six AUs which showed the greatest difference in the expected direction between the pain and the baseline segments and the judges' ratings of global expressiveness was examined. A positive relationship was found between two of the AUs and the criterion. The magnitude of the relationship between the predictor variables and the criterion was striking considering that the global - 99 - ratings were made on a wider range of facial actions over a longer period of time. The relationship of the pain AUs to the self-report measures was also discussed. Only the relationships between the two measures of subjective overall intensity and one of the six pain AUs were significant. The magnitudes of both relationships were modest, further underlining the tendency, noted in the literature (e.g., Fordyce, 1984) for pain behaviour and self-report to be discontinuous. Patrick, et al., (in press) found a similar discrepancy between facial behaviour and self- report. They suggested that the two measures might represent different aspects of a "common reaction pattern" at least in the case of acute noxious stimulation and the results of the present study seem to support this. The final issue discussed was whether or not the subjects in the study were actually in pain. It was concluded that sufficient precautions had been taken to ensure that they were. However, i t was also suggested that the possibility remained that some subjects might have exaggerated their facial behaviour as part of an overall strategy to obtain help. If this were the case i t would provide an alternate explanation for the similarity between the posed and genuine pain segments. But then this need to present oneself as in need of help may characterize most situations in which pain is observable by others (Craig & Prkachin, 1980) and thus may not be unique to the present study. It is difficult to assess whether or not the case which has been made by previous studies (Craig & Patrick, 1985; LeResche & Dworkin, - 100 - 1984; Patrick et al., in press) for using facial expressive behaviour as a tool in assessing the painful distress of others, has been strengthened by the results of the present study. A very limited, albeit congruent, set of AUs was found to distinguish the pain and baseline segments, compared to the more extensive sets found in the earlier studies. More research with chronic pain patients will be needed in order to determine if FACS is as useful a tool with clinical as i t has been shown to be with nonclinical populations. It was hypothesized that use of the FACS procedure would provide clues to identify distress in the stoical individual because of the system's ability to identify subtle facial behaviours. The subjects were somewhat successful in voluntarily attenuating their facial behaviour (which provides indirect support for the notion that display rules can serve the same function). The results of the FACS analysis also supported the hypothesis that more frequent and/or varied facial activity would be evident in the posed expression than in the genuine expression, again providing possible clues in assessing the genuineness of an individual's distress. Given the results of the present study and the work of Craig and his colleagues, i t would appear that future research might continue to investigate the existence of a constellation of pain AUs and its variability among different populations rather than pursuing the search for a prototypic pain expression. In fact, the results of the analysis of the posed expression suggest that a prototypic or stereotypic expression ought to be more common when people are simulating painful - 101 - distress rather than when they are genuinely in such distress. When acceptable reliability is achieved for the FACs intensity scoring, a number of other questions could also be addressed. Among these would be that of possible sex differences in the intensity of the "pain" AUs and the existence of differences in intensity among the AUs differentiating posed, masked and genuine pain faces. In the case of the posed expression in particular, informal viewing of the videotapes suggests that there may be a considerable difference in the intensity scores of the AUs as compared to the genuine pain segment. Such a finding would provide additional information which could be used when attempting to assess the genuineness of a pain complaint. With regard to the relationship between self-report and facial behaviour, i t would be of interest to determine i f the relationship would be stronger if more immediate assessment of self-report took place. While the memory for pain, as measured by the MPQ at least, has been shown to be fairly reliable (Hunter, Phillips, & Rachman, 1979) i t is s t i l l possible that the delay in assessing the subject's report of his or her distress resulted in an attenuated relationship. In general, the results of the present study provide further support that FACS represents a tool with which to utilize the rich variety of facial expressive behaviour. With further research the system may well provide information which will reliably increase the accuracy of judgments about painful distress with both clinical and nonclinical populations. - 102 - REFERENCES Agnew, D.C. & Mersky, M. (1976). Words of chronic pain. Pain, 2, 73-81. Ambeau, J. (1982). Recognizing psychological overlay in back pain patients. Unpublished manuscript. Ancoli, S. (1979). Psychological response patterns to stress (Doctoral dissertation, University of California, San Francisco, 1979). Dissertation Abstracts International, 40, 2887B. Ancoli, S., Kamiya, J., & Ekman, P. (1980). P sychophysi ologi cal differentiation of positive and negative affect. Paper presented at the Annual Meeting of the Biofeedback Association of America, Colorado Springs. Beecher, H.K. (1959). Measurement of subjective responses. New York: Oxford University Press. Bond, M.R. (1979). Pain: Its nature, analysis and treatment. London: Churchill Livingston. Bonica, J.J. (1980) Pain Research and Therapy: Past and current status and future needs. (In) L.K.Y. Ng and J.J. Bonica (Eds.) Pain, discomfort and humanitarian care, (pp. 1-46) Amsterdam: Elsevier. Brena, S.F. (1983). The medical diagnostic process. In S.F. Brena and S.L. Chapman (Eds.) Management of patients with chronic pain, (pp. 101-110) New York: Spectrum. Brightman, V. (1977). The quality of oro-facial pain sensations. Journal of Dental Research, 56A, 167. Buck, R., Miller, R., & Caul, W. (1974). Sex, personality, and physiological variables in the communication of affect via facial expression, Journal of Personality and Social Psychology, 30, 587-596. ~ Cataldo, M., Bessman, C, Parker, L., Pearson, J. & Rogers, M. (1979). Behavioral assessment for pediatric intensive care units. Journal of Applied Behavior Analysis, 12, 83-97. Cattel, R.B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1_, 245-276. Craig, K. (1984a). A social learning perspective on pain. In D.J. Oborne, M.M. Gruneberg & J.R. Eisert (Eds.), Psychology and medicine, (pp. 27-34) New York: Academic Press. - 103 - Craig, K.D. (1984b). Emotional aspects of pain. In Melzack, R., & Wall, P. (Eds.) Textbook of pain (pp. 153-161). Edinburgh: Churchill Livingston. Craig, K., McMahon, R., Morrison, J., & Zaskow, C. (1984). Developmental changes in infant pain expression during immunization injections, Social Science and Medicine, 1£, 1331-1337. Craig, K.D., & Patrick, C. (1985). Facial expression during induced pain. Journal of Personality and Social Psychology, 48, 1080-1091. Craig, K.D., & Prkachin, K. (1983). Nonverbal measures of pain. In R. Melzack (Ed.) Pain measurement and assessment (pp. 173-182). New York: Raven Press. Crockett, D., Prkachin, K., & Craig, K. (1977). Factors of the language of pain in patient and volunteer groups. Pain, 4_, 175-182. Darwin, C. (1872). The expression of emotion in man and animals. New York: Philosophical Library. De Paulo, B.M., Rosenthal, R., Eisenstat, R.A., Rogers, P.C., & Finkelstein, S. (1978). Decoding discrepant nonverbal cues. Journal of Personality and Social Psychology, 36, 313-323. DiMatteo, M.R., Friedman, H.S., & Taranta, A. (1979). Sensitivity to bodily nonverbal communication as a factor in practitioner-patient rapport. Journal of Nonverbal Behavior, 4_, 18-26. Dittman, A.T. (1972). Interpersonal messages of emotion. New York: Springer. Dubuisson, D., & Melzack, R. (1976). Classification of clinical pain descriptions by multiple group discriminant analysis. Experimental Neurology, 51_, (pp. 480-487). Dudley, S.R., & Holm, K. (1984). Assessment of the pain experience in relation to selected nurse characteristics. Pain, 18, 179- 186. Dworkin, S.F. (1970). Pain responsivity and its relationship to situational context and body-concept (Doctoral dissertation, New York University, 1969). Dissertation Abstracts International, 31/03-B, 1533B. Ekman, P. (1977). Biological and Cultural contributions to body and facial movement. In: J. Blacking (Ed.), Anthropology of the body, London: Academic Press. Ekman, P., & Friesen, W. (1969). Nonverbal leakage and clues to - 104 - deception. Psychiatry, 32, 88-105. Ekman, P., & Friesen, W. (1971). Constants across cultures in the face and emotion. Journal of Personality and Social Psychology, 17, 124-129. ~ Ekman, P., & Friesen, W. (1974). Detecting deception from the body or face. Journal of Personality and Social Psychology, 29, 288-298. Ekman, P., & Friesen, W. (1976). Measuring facial movement. Environmental Psychology and Nonverbal Behaviour, 1_, 56-75. Ekman, P., & Friesen, W. (1978a). Investigator's guide to the facial action coding system. Palo Alto, California: Consulting Psychologists Press. Ekman, P., & Friesen, W. (1978b). Manual for the facial action coding system. Palo Alto, California: Consulting Psychologists Press. Ekman, P., & Friesen, W. (1982). Felt, false and miserable smiles. Journal of Nonverbal Behavior, 6_, 238-249. Ekman, P., Friesen, W., & Ancoli, S. (1980). Facial signs of emotional experience. Journal of Personality and Social Psychology, 39, 1125-1134. Ekman, P., Friesen, W., & Ellsworth (1972). Emotion in the human face: Guidelines for research and an integration of the findings. New York: Pergamon Press. Ekman, P., Friesen, W., & Ellsworth, P. (1982). What emotion categories or dimensions can observers judge from facial behavior? In P. Ekman (Ed.), Emotion and the human face (pp. 39-55). London: Cambridge University Press. Ekman, P., Friesen, W., & Simons, R. (1983). The boundary between emotion and reflex: An examination of the startle. Manuscript submitted for publication. Ekman, P., Hager, J., & Friesen, W. (1981). The symmetry of emotional and deliberate facial actions. Psychophysiology, 18, 101- 106. Fordyce, W.E. (1976). Behavioral methods for chronic pain and illness. St. Louis: CV. MosByT Fordyce, W. (1983). The validity of pain behavior measurement. In R. - 105 - Melzack (Ed.), Pain measurement and assessment (pp. 145-154). New York: Raven Press. Fordyce, W., Brena, S.F., Holcomb, R.J., DeLateur, B.J., & Loeser, J.D. (1978). Relationship of patient semantic pain descriptions to physician diagnostic judgments, activity level measures and MMPI. Pain, 5_, 293-303. Fordyce, W.E., Lansky, D., Calsyn, D., Shelton, J., Stolov, W., & Rock, D. (1984). Pain measurement and pain behavior. Pain, 18, 53- 69. Garron, D.C, & Leavitt, F. (1979). Demographic and affective covariates of pain. Psychosomatic Medicine, 41, 525-534. Glass, G., & Stanley, J. (1970). Statistical methods in education and psychology. Englewood Cliffs: Prentice Hall. Gracely, R.H. (1979). Psychophysical Assessment of Human Pain. In J. Bonica (Ed.) Advances in pain research and therapy, 2» (PP* 805-824) (Ed.) New York": Raven Press. Gracely, R. (1983). Pain language and ideal pain assessment. In R. Melzack (Ed.), Pain measurement and assessment (pp. 71-78). New York: Raven Press. Gracely, R., McGrath, P., & Dubner, R. (1978). Validity and sensitivity of ratio scales of sensory and affective verbal pain descriptors: Manipulation of affect by diazepam. Pain, 5_, 19- 29. Gracely, R., McGrath, P., & Dubner, R. (1979). Narcotic analgesia: Fentanyl reduces the intensity but not the unpleasantness of painful pulp sensations. Science, 203, 1261-1263. Gracely, R.H., & Dubner, R. (1981). Pain assessment in humans - a reply to Hall. Pain, JM, 109-120. Haggard, E.A. (1958). Intraclass correlation and the analysis of variance. New York: Dryden Press, Inc. Hall, W. (1981). On "ratio scales of sensory and affective verbal pain descriptors." Pain, 11_, 101-108. Harper, R.G., Wiens, A.N., & Matarazzo, J.D. (1978). Nonverbal communication, New York: Wiley. Harter, H.L. (1960). Tables of range and studentized range. Annals of - 106 - Mathematical Statistics, 31, 1122-1147. Hilgard, E. (1969). Pain as a puzzle for psychology and physiology. American Psychologist, 103-113. Hjortsjo, Carl Herman. (1969). Man's face and mimic language. Lund: Studentlitteratur. Holland, M.K., & Tarlow, G. (1972). Blinking and mental load. Psychological Reports, _32, 119-127. Holland, M.K., & Tarlow, G. (1975). Blinking and thinking. Perceptual and Motor Skills, _41_, 403-406. House, A.E., House, B.J. & Campbell, M.B. Measures of interobserver agreement: Calculating formulas and distribution effects. Journal of Behavioral Assessment, 3_, 37-57. Hunter, M., Phillips, C, & Rachman, S. (1979). Memory for pain. Pain, 6, 35-46. Izard, C, Dougherly, L.M., Coss, C.L. & Hembree, E.A. (1981). Age- related and individual differences in infant's facial expressions following acute pain. Unpublished manuscript. Izard, C, Heubner, R.R., Resser, D., McGinness, G.C. & Dougherly, L.M. (1980). The infant's ability to produce discrete emotion expressions. Developmental Psychology, 16, 132-140. Jacox, Ada K. (1980). The Assessment of Pain, in L. Smith, H. Mersky, S. Grass (Eds.) Pain: Meaning and management, New York: SP. Johnson, J., Kirchoff, K. & Endress, M.P. (1975). Altering children's distress behaviour during orthopedic cast removal. Nursing Research, 824, (6) 404-410. Keefe, F.J., & Block, A.R. (1982). Development of an observation method for assessing pain behavior in chronic low back pain patients. Behavior Therapy, 363-375. Kendall, P., Williams, L., Pechacek, T., Shisslak, C, & Herzoff, N. (1979). Cognitive-behavioral and patient education interventions in cardiac catheterzation procedures: The Palo-Alto medical psychology project. Journal of Consulting and Clinical Psychology, 47 (1), 49-58. Kleck, R.E., Vaughan, R.C., Cartwright-Smith, J., Vaughan, K.B., Colby, CF. & Lanzetta, J.T. (1976). Effects of being observed on expressive, subjective and physiological responses to painful - 107 - stimuli. Journal of Personality and Social Psychology, 34, 1211-1218. Kraut, R.E. (1978). Verbal and nonverbal cues in the perception of lying. Journal of Personality and Social Psychology, 36, 380-391. Kraut, R.E. (1982). Social presence, facial feedback, and emotion. Journal of Personality and Social Psychology, 42, 853-863. Kremer, E.F., & Atkinson, J.H. (1983). Pain language as a measure of affect in chronic pain patients. In R. Melzack (Ed.), Pain measurement and assessment (119-128). New York: Raven Press. Kremer, E.F., Block, A., & Atkinson, J.H. (1983). Assessment of pain behavior: Factors that distort self-report. In R. Melzack (Ed.) Pain measurement and assessment (pp. 165-172). New York: Raven Press. Langer, E.J., Jam's, I.L., & Wolfer, J.A. (1975). Reduction of psychological stress in surgical patients. Journal of Experimental and Social Psychology, 11 , 155-165. Lanzetta, J., Cartwright-Smith, J., & Kleck, R. (1976). Effects of nonverbal dissimulation on emotion. Journal of Personality and Social Psychology, 33, 354-370. Leavitt, F., Garron, D., Whisler, W., & Sheinkop, M. (1978). Affective and sensory dimensions of back pain. Pain, 1_, 273-281 . Lenburg, C.B., Glass, H.P., & Davitz, L.J. (1970). Inferences of physical pain and psychological distress in relation to the stage of the patients illness and occupation of the perceiver. Nursing Research, J_9, 392-398. LeResche, L. (1982). Facial expression in pain: A study of candid photographs. Journal of Nonverbal Behavior, 7,.464. LeResche, L., & Dworkin, S. (1984). Facial expression accompanying pain. Social Science and Medicine, 19(12) 1325-1330. Leventhal, H., & Sharp, E. (1965). Facial expressions as indicators of distress: In S.S. Tompkins and C.E. Izard (Eds.). Affect, Cognition and Personality, (pp. 296-318) New York: Springer. McCaffery, M. (1972). Nursing Management of the Patient with Pain. Philadelphia: Lippincott. - 108 - Melamed, B., & Siegel, L. (1975). Reduction of anxiety in children facing hospitalization and surgery by use of filmed modeling. Journal of Consulting and Clinical Psychology, 43, 511-521. Melzack, R. (1975). The McGill Pain Questionnaire: Major properties and scoring methods. Pain, 1_, 277-299. Melzack, R. (1983). The McGill Pain Questionnaire. In R. Melzack (Ed.), Pain measurement and assessment, (pp. 41-43). New York: Raven Press. Melzack, R., & Perry, C. (1975). Self regulation of pain: The use of alpha feedback and hypnotic training for the control of chronic Pain. Experimental Neurology, 46, 452-469. Melzack, R., & Togerson, W. (1971). On the language of pain. Anesthesiology, 34, 50-59. Melzack, R., & Wall, P. (1965). Pain mechanisms: A new theory. Science, 150, (3699), 971-979. Patrick, C, Craig, K.D., & Prkachin, K. (in press). Facial expression determinants of observer judgements of acute pain. Journal of Personality and Social Psychology. Plutchik, R. (1962). The emotions: facts, theories and a new model. New York: Random House. Prkachin, K., & Craig, K. (1985). Influencing non-verbal expressions of pain: Signal detection analysis, Pain, 21, 399-409. Prkachin, K.M., Currie, A.N., & Craig, K.D. (1983). Judging nonverbal expressions of pain. Canadian Journal of Behavioural Science, 15, 4 0 9 _ 4 2 1 < Reading, E., & Newton, J. A comparison of primary dysmenorrhea and intrauterine device related pain. Pain, 1977, 3_, 265-276. Rinn, W.R. (1984). The neuropsychology of facial expression: A review of the neurological and psychological mechanisms for producing facial expressions. Psychological Bulletin, 95, 52-77. Sanders, S. (1979). Behavioral assessment and treatment of clinical pain: Appraisal of current status. In Progress in Behavior Modification, Vol. 8. New York: Academic Press, 1979. Schwartz, G., Ahern, G., & Brown, S. (1979). Lateralized facial muscle response to positive versus negative emotional stimuli. Psychophysiology, 16, 561-571 . - 109 - Sternbach, R. (1974). Pain patients: Traits and treatment. New York: . Academic Press. Tan, Siang-Yang, (1982). Cognitive and cognitive-behavioral methods for pain control: A selective review. Pain, 12, 201-228. Teske, K., Daut, R.L., and Cleeland, C.S. (1983). Relationships between nurses' observations and patients' self-report of pain, Pain, 16, 289-296. Tompkins, S.S., & McCarter, R. (1964). What and where are the primary affects? Some evidence for a theory. Perceptual and Motor Skills, 18, 119-158. von Baeyer, C. (1982). Repression-sensitization, stress and perception of pain in others. Perceptual and Motor Skills, 55, 315-320. von Bayer, C, Johnson, M., & McMillan, M. (1982). Consequences of nonverbal expression of pain: Patient distress and observer concern. Unpublished manuscript. Wall, P.D. (1979). On the relation of injury to pain. The John J. Bonica lecture. Pain, 6., 253-264. Weitz, S. (Ed.) (1974). Nonverbal communication: readings with commentary. New York: Oxford University Press, 1974. Wolff, B.B. (1978). Behavioral measurement of human pain. In R.A. Sternbach (Ed.) The psychology of pain, (pp. 129-168). New York: Raven Press, 1978. Woodworth, R.S. (1938). Experimental psychology. New York: Henry Holt. - 110 - APPENDIX 1 Diagnostic Categories Category Frequency Disc degeneration 22 Disc space narrowing 3 Disc protrusion 4 Discogenic low back pain 7 Nerve root irritation 12 Facet joint syndrome 5 Spinal stenosis 2 Spondylolithesis 5 Post operative pain 7 Soft tissue injury/strain 10 Mechanical low back pain/ Postural problems 22 Fibrocytis 3 Chronic benign pain syndrome 20 Inactivity 8 Obesity 13 Drug dependency/Alcoholism 5 Secondary pain/Psychological overlay 15 Other 43 - I l l - APPENDIX 2 McGill Pain Questionnaire Verbal Descriptor Scales What Does Your Pain Feel Like? Circle ONLY those words that best describe the type of pain you have been asked to evaluate. Leave out any category that is not suitable. Use only a single word in each appropriate category - the one that applies best. 1 2 3 4 5 6 1 Flickering Quivering Pulsing Throbbing Beating Pounding 1 Jumping 2 Flashing 3 Shooting 1 2 3 4 5 Pricking Bori ng Drilling Stabbing Lancinating 1 Sharp 2 Cutting 3 Lacerating 8 Pinching Pressing Gnawing Cramping 1 2 3 Tagging Pulling Wrenching 10 1 Hot 2 Burning 3 Scalding 4 Searing 11 Tingling Itchy Smarting Stinging 12 1 Dull 2 Sore 3 Hurting 4 Aching 5 Heavy 13 1 2 3 4 Tender Taut Rasping Splitting 14 1 Tiring 2 Exhausting 15 1 Sickening 2 Suffocating 16 1 Fearful 2 Frightful 3 Terrifying 17 1 2 3 4 5 Punishing Gruel 1 ing Cruel Vicious Killing 18 1 Wretched 2 Blinding 19 Annoying Troublesome Miserable Intense Unbearable 20 Spreading Radiating Penetrating 4 Piercing 1 Tight 2 3 4 5 Numb Drawing Squeezing Tearing 1 Cool 2 Cold 3 Freezing Nagging Nauseating Agonizing Dreadful Torturing - 112 - APPENDIX 3 GRACELY SCALES One word or word-pair on each of the scales below describes the most severe pain you experienced during the examination. Circle the word or word-pair on each scale that best describes i t . Use only a single word or word-pair on each scale. Sensory 2. Unpleasantness 3. Painfulness Extended Intense Excruciating Not Painful Very Intense Unbearable Faintly Painful Intense Intolerable Mildly Painful Strong Agonizing Somewhat Painful Slightly Intense Horrible Slightly Painful Clear-Cut Dreadful Moderately Painful Barely Strong Frightful Rather Painful Moderate Miserable Quite Painful Slightly Moderate Distressing Decidedly Painful Mild Upsetting Pretty Painful Very Mild Irritating Very Painful Weak Unpleasant Unusually Painful Very Weak Uncomfortable Extremely Painful Faint Annoying Extremely Weak Distracting - 113 - APPENDIX 4 CONSENT FORM We are interested in studying how physiotherapists carry out their various procedures and how a variety of clients respond to them. For our study we are requesting your permission to videotape a short segment of the physiotherapy assessment you are about to undergo. In addition we will be asking you to repeat certain movements under different instructions and to f i l l out several brief questionnaires related to the pain (if any) you experience during those movements and more generally on a day-to-day basis. All of this information will be confidential, and you will be identified only by number to ensure your anonymity. We would greatly appreciate your cooperation; however, you are free to refuse and doing so will in no way affect your assessment at the Back Pain Clinic. Thank you. Susan Hyde, M.A. Kenneth D. Craig, Ph.D. Department of Psychology University of British Columbia I agree to participate in this study. Date Signature - 114 - APPENDIX 5 Nonverbal Expressions Enhancing Instructions It is important for you to know that in trying to accurately assess your pain we rely heavily on nonverbal information as well as on what you tell us. By nonverbal information I mean the expression on your face, any sounds you make like "ouch" or groaning and movements of your body. Sometimes people feel they ought to cover up such indications that they are in pain. We are asking you to try to avoid doing this. Instead we want you to try to express as much as you can how you are really feeling. Do you understand what I'm asking you? ... Fine. Physiotherapy Protocol Okay. First I'm going to try to pull your feet towards me. I'd like you to try to prevent me from doing that by trying to pull your feet towards you. Fine, now I'm going to bend your right knee, Okay, now I'm going to move your knee inward towards the middle of the table, tell me when to stop. Now I'm going to repeat that movement with your left leg. Okay, now I'd like you to bend your right knee again, lock your hands around i t , and pull i t up towards your chest as far as you can. Okay, now I'd like you to do the same thing with your left knee. Okay, now keeping your legs straight, I'm going to li f e your right leg straight up until you tell me to stop. Okay, now I'm going to do the same thing with your left leg. - 115 - Appendix 5 (cont'd) Now I'd like you to tell me which of the movements we just went through created the most discomfort for you. Okay, we're going to repeat that movement but this time I'm going to give you some additional instructions. Instructions for Generating Genuine, Masked, and Posed Facial Expressions Genuine I am going to/I'd like you to repeat movement X. Masked Again I am going to/I'd like you to repeat movement X. This time I would like you to try to pretend that i t doesn't bother you at a l l , no matter how much i t does. Try to cover up the fact that you find i t painful or distressing. Posed Use a Nonpainful movement. Now, I'm going to repeat movement Y which you said did not bother you. This time I want you to act (behave, pretend) as i f movement Y is really causing you a lot of pain. Let me know by looking at you that i t is painful. Make a "pain face". - 116 - APPENDIX 6. Scoring Units for the Facial Action Coding System (Ekman & Friesen, 1978b) Upper Face Lower Face AU Name AU Name 1 Inner Brown Raise 9 Nose Wrinkle 2 Outer Brow Raise 10 Upper Lip Raise 4 Brow Lower 11 Nasolabial Deepen 5 Upper Lid Raise 12 Lip Corner Pull 6 Cheek Raise 13 Cheek Puff 7 Lids Tight 14 Dimpler 41 Lids Droop 15 Lip Corner Depres: 42 Slit 16 Lower Lip Depress 43 Closed 17 Chin Raise 44 Squint 18 Lip Pucker 45 Blink 20 Lip Stretch 46 Wink 22 Lip Funnel 23 Lip Tight 24 Lip Press 25 Lips Part 26 Jaw Drop 27 Mouth Stretch 28 Lip Suck Miscellaneous Actions AU Name AU Name 8 Lips Toward 33 Blow 19 Tongue Show 34 Puff 21 Neck Tighten 35 Cheek Suck 29 Jaw Thrust 36 Tongue Bulge 30 Jaw to Sideways 37 Lip Wipe 31 Jaw Clench 38 Nostril Dilate 32 Bite 39 Nostril Compress Other AD Name 50 Mouth Movement Note: AU = Action Unit; AD - Action Descriptor - 117 - APPENDIX 7 GLOBAL CODING MANUAL Look for These Signs: Eyes 1. Narrowed or tightly closed eyes 2. Eyebrow raise, i.e. * • 3. Frown, i.e., eyebrows lowered with a bulge or crease between them Nose 1. Wrinkled Mouth 1. Tightened lips or upper lip raise 2. One or both corners of the mouth pulled back 3. Jaw drop or stretch (as i f gasping or saying ow!) Rate on the _3 Point Scale Rule of Thumb . If you are having trouble deciding between a 1 or 2 or a 2 or 3, keep viewing the tape to observe how the individual reacts to the repeat of the painful movement. - 118 - Appendix 7 (cont'd) Global Coding (cont'd) 1 2 3 very inexpressive somewhat expressive very expressive DEFINITIONS INEXPRESSIVE: . the face remains with a neutral expression throughout even though there may be evidence of jaw and throat tension. E.g., a sign occurs to a mild degree but appears fairly frequently - 2-4 times OR a, sign occurs to a moderate degree but occurs only once or twice. . the face can be observed moving from neutral to expressive. . there may be signs of jaw and throat tension SOMEWHAT one or more of the signs are present in a mild to EXPRESSIVE a somewhat moderate form. VERY EXPRESSIVE one of the signs is present in an extreme, i.e., close to maximum form. two or more signs are present in a moderate, i.e., - 119 - Appendix 7 Global Coding (cont'd) obvious but not maximum form and occur frequently, i.e., more than once or twice . there may be evidence of jaw and throat tension . the face can be observed moving from neutral to expressive. E.g., a person whose face is fairly expressionless, apart perhaps from some jaw and/or throat tension, until the movement(s) of pain when a definite moderate to maximum expression is visible OR . an individual whose face is constantly in motion (unrelated to talking) with frequent moderate or maximum grimaces. - 120 - APPENDIX 8 a) Summary of Demographic, Self Report and Global Expressiveness Data Analysis" Dependent Measures: Age, Duration of current complaint; Duration of chronic complaint; Number of prior operations; McGill Pain Questionnaire (MPQ) - acute pain instructions: Number of words chosen, sensory, affective, evaluative scales; MPQ-chronic pain instructions: number of words chosen, sensory, affective, evaluative scales; Gracely verbal descriptor scales: sensory, unpleasantness, painfulness; Global Expressiveness ratings. Wilkes Summary Table Source Wilkes Lambda Approx. F df Prob. Sex 0.7506 2.0976 16,101 .014 Instructional Set 0.9146 0.5895 16,101 .885 S x I 0.8701 0.9425 16,101 .524 b) Summary of the Analysis of the "Acute" versus the "Chronic" MPQ, Hotellings T2 - Dependent measures: The four scales of the McGill Pain Questionnaire (MPQ) Wilkes Summary Table Source Wilkes Lambda df Approx. F df Prob. Subjects 5.5519E-03 4,119,119 2.6369 476,444 0.00 Type of Administra- 6.8797E-01 4,1,119 13.1530 4,116 .000 tion - 121 - APPENDIX 9 SUMMARY FACS DATA ANALYSIS Group Comparison The following analysis was a 2 x 2 x 3 (Instructional Set x Sex x Segment) MANOVA with repeated measures on one factor (Segment) using as dependent measures the 14 AUs which remained after the initial exclusion criteria were applied. Wilkes Summary Table Source Wilkes Lambda df Approx. F df Prob. Instructional 9.0630E-01 14,1,117 0.7680 14,104 0.7008 Set Sex 8.8894E-01 14,1,117 0.9281 14,104 0.5317 Is x Sex 9.3590E-01 14,1,117 0.5088 14,104 0.9234 Segment 4.7251E-01 14,3,347 6.8253 42,944 0.0000 Is x Segment 8.7764E-01 14,3,347 1.0621 42,944 0.3669 Sex x Segment 8.9524E-01 14,3,347 0.8973 42,944 0.6587 Is x S x Seg. 8.8813-01 14,3,347 0.9633 42 ,944 0.5391 Subjects 0.0000+00 14,117,0 0.0000 0,0 0.0000 Subjects x Seg. 0 .0000+00 14,3,347 0.0000 0,0 0.0000 - 122 - APPENDIX 10 Post Hoc Comparisons of Segment Effect Using Tukey Procedure for Repeated Measures on One Factor J_. Dependent measure: AU 2 Summary Table Source SS df MS F Prob. Segment 2.21 3 7.361E-01 5.95 0.0007 Seg x Sub 42.91 347 1.236E-01 Z_. Dependent measure: AU4 Summary Table Source SS df MS F Prob. Segment 18.68 3 6.225 38.77 0.0000 Seg x Sub 55.71 347 1.601E-01 3. Dependent measure: AU6 Summary Table Source SS df MS F Prob. Segment 13.75 3 4.58 28.46 0.0000 Seg x Sub 55.88 347 1.610E-01 4_. Dependent measure: AU7 Source SS df MS F Prob. Segment 1.62 3 5.410 5.93 0.0007 Seg x Sub 31.65 347 9.122E-02 - 123 - Appendix 10_ (cont'd) 5_. Dependent measure: AU10 Summary Table Source SS &f_ MS F Prob. Segment 1.98 3 6.611E-01 6.70 0.0003 Seg x Sub 34.23 347 9.866-02 6_. Dependent measure: AU12 Summary Table Source SS df MS F Prob. Segment 29.22 3 9.75 32.09 0.0000 Sex x Sub 105.31 347 3.035E-01 7_. Dependent measure: AU25 Summary Table Source SS df MS F Prob. Segment 11.57 3 3.86 11.86 0.0000 Sex x Sub 112.87 347 3.253E-01 .8. Dependent measure: AU43 Summary Table Source SS df MS F Prob. Segment 11.43 3 3.81 17.37 0.0000 Seg x Sub 76.08 347 2.192E-01 9_. Dependent measure: AU45 Summary Table Source SS df MS \ F Prob. Segment 163.82 3 54.61 12.42 0.0000 Seg x 1525.47 347 4.40 Sub. - 124 - APPENDIX 11 Relationships Among the Variables a. Relationship Between "Pain" AUs and Global Expressiveness Each of the following was a stepwise multiple regression analysis examining the relationship between judges' ratings of global expressiveness and the frequency of "pain" AUs, (i.e., the six AUs which showed the greatest difference in the expected direction between baseline and the genuine pain segment). Variables in the Equation F test for Mult. R Variable Mult. R. RSQ AU 43 0.4411 AU 6 0.5154 AU 4, 7, 10, 25 0.1946 0.2656 RSQ Change B 0.1946 0.447 0.0710 0.449 Beta df F £ 0.337 1,118 28.51 .0000 0.286 2,117 21.56 .0000 Variables not in the Equation b. Relationship Between the "Pain" AUs and the "Acute" MPQ Evaluative Scale The following was a stepwise multiple regression analysis examining the relationship between the subjects' scores on an "acute" MPQ scale and the frequency of the "pain" AUs (i.e., the six AUs which showed the greatest difference in the expected direction between baseline and the genuine pain segment). Variables in the Equation F test for Mult. R. Variable Mult. R. RSQ RSQ Change _B Beta df F £ AU 6 0.3374 0.1138 031138 3.179 0.337 1,118 15.16 .0002 Variables not in the Equation AU 4, 7, 10, 25, 43 - 125 - c. Relationship between "Pain" AUs and the Gracely Painfullness Scale The following was a stepwise multiple regression analysis examining the relationship between the subjects' scores on the Gracely Painfulness scale and the frequency of the "pain" AUs (i.e., the six AUs which showed the greatest difference in the expected direction between baseline and the genuine pain segment). Variables in the Equation F test for Mult R. Variable Multi R. RSQ RSQ Change B Beta df F_ £ AU 6 0.3006 0.0903 0.0903 1.9306 0.301 1,118 11.72 .0000 Variables not in the Equation AU 4, 7, 10, 25, 43

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
China 24 1
Canada 1 0
United States 1 0
City Views Downloads
Beijing 24 0
Unknown 1 13
Ashburn 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}

Share

Share to:

Comment

Related Items