PSYCHOPATHIC EMOTIONS: FOUNDED ON A N ADDICTION TO AROUSAL by HUGUES F. M . HERVE B.A., Concordia University, 1994 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Department of Psychology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA July 1998 © Hugues F. M . Herve, 1998 in p resen t ing this thesis in part ial fu l f i lment of t h e requ i rements fo r an advanced d e g r e e at t h e Univers i ty o f Brit ish C o l u m b i a , I agree that t h e Library shall make it f reely available f o r re ference and study. I fu r ther agree that permiss ion fo r extensive c o p y i n g o f th is thesis f o r scholar ly pu rposes may b e g ran ted by the head o f my d e p a r t m e n t o r by his o r her representat ives. It is u n d e r s t o o d that c o p y i n g o r pub l i ca t i on o f this thesis fo r f inancial ga in shall n o t be a l l owed w i t h o u t my w r i t t e n permiss ion . D e p a r t m e n t The Univers i ty o f Brit ish C o l u m b i a Vancouver , Canada DE-6 (2/88) Abstract Recent advances in research on affect have shown variations in individuals' affective circumplex, which are related to differences in arousal sensitivity (Blascovich, 1990; 1992; Feldman, 1995a). Given evidence suggesting that psychopaths are hyposensitive to arousal, they should therefore demonstrate an arousal-focus when processing emotional material. The present study was designed to assess this attentional bias. Participants were 41 offenders divided into psychopathic and nonpsychopathic groups according to their scores on the Psychopathy Checklist-Revised (PCL-R; Hare, 1991). Pleasure and arousal ratings and similarity judgments of emotional words and pictures of facial affect were employed to provide both objective and projective measures of the psychopath's interpretation of the material. Results indicated that psychopaths and nonpsychopaths did not differ in either their ratings or similarity judgments of the stimuli. However, when participants were further subdivided into sex offenders and non sex offenders, a pattern emerged that suggested that the psychopath's affective structure is indeed characterized by an arousal-focus, but only when faced with "intriguing" stimulation. Results are discussed in light of a new theory of psychopathy suggesting that psychopaths may be addicted to arousal, as represented in their lifestyles and their affective responses. ii Table of Contents Abstract ii Table of Contents iii List of Tables v List of Figures vii Acknowledgements ix Introduction 1 Emotional Processing 3 Emotional Acquisition 3 Emotional Processing 6 Individual Differences in the Emotional Circumplex 9 Affect, Personality, and Behavior 11 The Role of Arousal 14 Psychopathy and Physiological Arousal 17 Psychopathy and Arousal-Seeking 20 The Behaviors 20 A Proactive Attitude 22 Psychopathy and Affect 24 Psychopathy and Valence-Focus 26 Psychopathy and Arousal-Focus 35 Emotionally Challenged, Not Dull 39 The Psychopath as an Arousal Addict 40 Method 43 Participants 43 iii Stimuli 45 Category Sort 47 Distracter Task 48 Rating task 49 Dependent Variables 50 Similarity Score 50 Global Pleasure and Arousal Score 51 Results 52 Confounding Variables 52 Emotional Words 54 Pictures of Facial Affect 63 Dimensional Analyses 71 Emotional Words 73 Pictures of Facial Affect 77 Dimensional Interpretation 81 Additional MDS Analyses 83 Emotional Words - Part II 84 Dimensional Interpretation - Part II 102 Discussion 108 Conclusion 125 References 126 Appendix 135 iv List of Tables Table 1: Independent t-tests comparing nonpsychopaths and psychopaths on possible confounds 53 Table 2: Pleasure rating rankings for emotional words from most pleasant to least pleasant averaged across subjects from each group 55 Table 3: Arousal rating rankings for emotional words from most arousing to least arousing averaged across subjects from each group 56 Table 4: Paired Sample t-tests comparing the pleasure ratings and arousal ratings of emotional word types for each group 58 Table 5: Pleasure rating rankings for pictures of facial affect from most pleasant to least pleasant averaged across subjects from each group 64 Table 6: Arousal rating rankings for pictures of facial affect from most arousing to least arousing averaged across subjects from each group 65 Table 7: Paired Sample t-tests comparing the pleasure ratings and arousal ratings of pictures of facial affect types 67 Table 8: Correlations between the dimensional coordinates of stimuli resulting from a 2D solution and the pleasure and arousal ratings for both diagnostic groups 82 Table 9: Pleasure rating rankings for emotional words from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending 85 Table 10: Arousal rating rankings for emotional words from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending 86 Table 11: Pleasure rating rankings for pictures of facial affect from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending 87 Table 12: Arousal rating rankings for pictures of facial affect from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending 88 Table 13: Correlations between the dimensional coordinates of stimuli resulting from a 2D solution and the pleasure and arousal ratings for groups classified by psychopathy and sex offending 103 Table 14: Pleasure coordinates for emotional words ranked from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending 104 Table 15: Arousal coordinates for emotional words ranked from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending 105 Table 16: Pleasure coordinates for pictures of facial affect ranked from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending 106 Table 17: Arousal coordinates for pictures of facial affect ranked from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending 107 vi List of Figures Figure 1: The emotional circumplex 6 Figure 2: Variations in the emotional circumplex 10 Figure 3: Pleasure (upper panel) and arousal (lower panel) ratings of the valence-based words for both diagnostic groups 59 Figure 4: Pleasure (upper panel) and arousal (lower panel) ratings of the arousal-based words for both diagnostic groups 62 Figure 5: Pleasure (upper panel) and arousal (lower panel) ratings of valence-based faces 68 Figure 6: Pleasure (upper panel) and arousal (lower panel) ratings of arousal-based faces 70 Figure 7: Stress value by dimensional solutions of the emotional words for both diagnostic groups 74 Figure 8: Nonpsychopath's emotional circumplex for the emotional words 75 Figure 9: Psychopath's emotional circumplex for the emotional words 76 Figure 10: Stress value by dimensional solutions of the pictures of facial affect for both diagnostic groups 78 Figure 11: Nonpsychopath's emotional circumplex for the pictures of facial affect 79 Figure 12: Psychopath's emotional circumplex for the pictures of facial affect 80 Figure 13: Stress value by dimensional solutions of the emotional words for the four diagnostic groups 89 Figure 14: Nonpsychopathic sex offenders' emotional circumplex for the emotional words 91 Figure 15: Psychopathic sex offenders' emotional circumplex for the emotional words 92 vii Figure 16: Nonpsychopathic non sex offenders' emotional circumplex for the emotional words 93 Figure 17: Psychopathic non sex offenders' emotional circumplex for the emotional words 94 Figure 18: Stress value by dimensional solutions of the pictures of facial affect for the four diagnostic groups 95 Figure 19: Nonpsychopathic sex offenders' emotional circumplex for the pictures of facial affect 97 Figure 20: Psychopathic sex offenders' emotional circumplex for the pictures of facial affect 98 Figure 21: Nonpsychopathic non sex offenders' emotional circumplex for the pictures of facial affect 99 Figure 22: Psychopathic non sex offenders' emotional circumplex for the pictures of facial affect 100 Figure 23: Psychopathic non sex offenders' ID solution for the pictures of facial affect 101 Acknowledgements I would like to thank all the people who have contributed, directly or indirectly, to the completion of this thesis. First, I would like to thank my parents, my girlfriend Tracy, all my friends, and my cat for their constant support. Second, I would like to thank Correction Services Canada (CSC) and the staff and inmates from Regional Health Center without whom this research could not have been completed. It should be noted that the reported findings and their interpretation do not necessarily reflect the official policy of the Correctional Service of Canada. Third, I would like to thank my committee members, John Yuille and Wolfgang Linden, for their useful suggestions and advice. Fourth, I would like to thank James Russell and Lawrence Ward for their technical help. Fifth, I would like to thank Andra Smith, Darion D'Angelo, Alicia Spidel, Kylie Erickson, and Kelly Giesbrecht for their help in setting up this study, in data collection, and in data entry. Sixth, I would like to thank the BC Medical Research Foundation and Division 41 of the American Psychological Association for helping fund this project. Finally, and most importantly, I would like to give special thanks to my supervisor, Dr. Robert D. Hare, for his guidance, support, and encouragement since my arrival in his laboratory. Furthermore, I would like to thank him for introducing me and giving me the opportunity to pursue research in such a fascinating area. His own passion for forensics has been a source of inspiration for myself as well as others; a passion that I can only hope to pass on to my fellow students. IX Introduction Clinical experience, theory, and research suggest that the distinctive constellation of affective, interpersonal, and behavioural characteristics that make up the psychopathic profile -irrespective of age, sex, race, or culture (Cooke, 1997; Forth, 1996; Forth, Hart, & Hare, 1990; Kosson, Smith, & Newman, 1990; Rutherford, Alterman, Caciola, & McKay, 1998; Salekin, Rogers, & Sewell, 1997) - creates a guiltless individual capable of repeatedly violating the rights of others. Lacking empathy, guilt, or remorse, the psychopath is able to treat even "significant others" as little more than objects to be used at his/her own convenience. Although only making up about 1% of the general population (Hare, 1993), psychopaths, in all their egocentricity, manipulativeness, deceitfulness and lack of respect for the social and legal rights of society, are nevertheless responsible for a disproportionate amount of antisocial and criminal behaviour, particularly of the violent type (Cornell, Warren, Hawk, Stafford, Oram, & Pine, 1996; Hare, 1995; 1996; Hare & McPherson, 1984). These human predators, representing 15-25% of the criminal population, are the most dangerous and violent of offenders, thus making their study and treatment a high priority for society (Hare, 1995). Unfortunately, although receiving a great deal of attention from researchers and clinicians alike, little is yet known about the etiology and/or developmental course of this socially devastating disorder. Clinical tradition has, however, consistently reflected the importance of affective/interpersonal processes, in addition to poor behavioral controls, in understanding psychopathy (Buss, 1966; Cleckley, 1976; Hare, 1996; Karpman, 1961; Lykken, 1957; McCord & McCord, 1964). This emphasis lead to the development of the Hare Psychopathy Checklist-Revised (PCL-R; Hare, 1991), an instrument that has since been used to operationalize the construct in adult forensic populations (see reviews by Fulero, 1995; Stone, 1995). The PCL-R, taking both personality and behavioral indices into account, contains items that fall into two 1 correlated, yet distinct, clusters, or factors. Factor 1 consists of items that measure the affective and interpersonal features of psychopathy, such as egocentricity, manipulativeness, callousness, and lack of remorse, while Factor 2 consists of items that describe an impulsive, antisocial, and unstable lifestyle. A recent item-response analysis of the PCL-R (Cooke & Michie, 1997), suggests that the interpersonal and affective domains (i.e., Factor 1 items) of psychopathy are more discriminating of the construct than its behavioral manifestation (i.e., Factor 2 items). Thus, although the psychopath's actions are important in their own right, especially in terms of risk assessment and recidivism (Hare, 1996; Harris, Rice, & Cormier, 1991; Hemphill, Templeman, Wong, & Hare, 1998), this finding places further emphasis on the significance of understanding their interpersonal and affective processes. The current investigation aims to provide further clues into the psychopath's emotional deficit by investigating their underlying affective structure. In light of the fundamental role that affect plays in mediating and regulating behavior, it is hoped that studying the psychopath's underlying affective structure will lead to new insights into the development of this socially devastating disorder. Specifically, the present investigation, in line with dimensional theories of affect (e.g., Feldman, 1995a; Russell, 1980), was expected to show that psychopaths rely on the dimension of arousal to interpret emotional events. If so, this would suggest that the psychopath's affective structure developed out of an addiction to arousal resulting from a pathologically under-aroused state. The research described below provides the background to the proposed study, starting with a review of emotional acquisition, processing, and normal variations in these. Thereafter, clues into the psychopath's physiological sensitivity to arousal are discussed, as well as supporting behaviors. This is followed by summaries of research on the psychopath's proficiency with the two fundamental dimensions of affect, valence and arousal. 2 Emotional Processing Emotional Acquisition: Several theories, which are clearly relevant to the current formulation of psychopathy, have been proposed to account for how emotions come to be. Although contesting the importance and sequence of emotional components (the chicken versus the egg debate), the way in which they become activated (mental versus environmental), and their end results (categorical versus dimensional views), most theories of affective development nonetheless agree that emotional experiences depend on two correlated, yet independent mechanisms: one that mediates an arousal response to an emotional event and another that evaluates the meaning/significance of that same event (see Mandler, 1984 for a review). One such theory, proposed by Mandler (1984), views both of these components as essential to the human experience of emotions. Mandler (1984) proposed that emotional experiences are constructed out of autonomic arousal and evaluative cognitions such that their combination results in a subjective emotional state. Arousal, in this scheme, refers to a physiological response produced by the autonomic nervous system (ANS). The arousal, which is nonspecific (i.e., does not differentiate between emotions), solely sets the quantitative specifications for emotional life. In other words, arousal alone does not produce an emotional response (e.g., Schachter, 1971). The jogger is not necessarily emotional, although his ANS is highly active. However, at the physiological level, the autonomic arousal prepares the organism for action, while concurrently signaling the mental organization for attention, alertness, and scanning of the environment. On the other hand, the evaluative cognitions perform a meaning analysis of the emotional event. Mediated by the central nervous system, they ascribe the particular quality (e.g., valence) of the emotion. Although these interpretive-cognitions, or meaning analyses, may be engendered by the arousal (as signal theories of arousal would suggests), they are primarily set by the general situation and 3 current cognitive states of the organism. It is the joint products of both of these systems, arousal and meaning analysis, that construct emotions as we know them. As Mandler (1984) noted, "Arousal provides the intensity of the emotional state, and cognition provides its quality (p. 119)". Emotional events can evoke either mechanism (Mandler, 1984). Arousal can be elicited by an innate automatic release of ANS arousal or by the analysis of the situation. While physical injury can produce an immediate reaction, there are other objects (e.g., a gun) and events (e.g., banking) that are initially neutral in connotation but that become - through classical conditioning - ANS releasers. However, the conditioned response is not the autonomic arousal, but rather the cognitive evaluation, which may in turn produce the attendant arousal. Accordingly, the preprogrammed ANS reactions are much more resistant to change than the cognitively mediated ones. On the other hand, the cognitively based reactions, being rooted in one's autobiographical past, are more likely to be highly idiosyncratic. For example, while the successful bank robber may view guns and banks in a positive light, the victimized bank teller is likely to view these same objects with disgust and fear. Nonetheless, although innate releasers are more resistant to change, they are not immune to it, as reflected in masochistic practices. The end result is a continuum of emotionally evoking events, from the innate to the experiential. In addition to learned differences, there are three other possible sources of individual differences in emotional responsiveness (Mandler, 1984). Firstly, only perceived arousal can have psychological consequences, suggesting different emotional experiences for individuals varying in arousal sensitivity (Blascovich, 1990; 1992). Accordingly, highly sensitive individuals, such as those with relatively low baseline levels of autonomic arousal, may react more strongly to emotionally inducing situations, possibly choosing to avoid such situations altogether. In contrast, those who are insensitive to arousal related cues, such as people whose 4 arousal threshold is higher, may be unfazed by similar emotion-evoking events, probably preferring and seeking circumstances that have intense emotional connotation. Secondly, Mandler (1984) also mentioned the possibility that the arousal and cognitive-interpretive systems may, at times, work in a relatively independent fashion. Thus, while some individuals may be more attentive to their visceral reactions than to their cognitive ones, others may prefer to rely more on their interpretation of an event than on their bodily reaction to that same event. Thirdly, our current state of mind can also influence the way in which we interpret any given situation, suggesting a link between affect and personality. While positive reactions are more likely to occur in scenarios in which we feel in control, the opposite reactions become more likely in situation in which we have no control. For example, a robber, thinking he is in control of the situation, feels excitement and a sense of power. The bank teller, having no control whatsoever, feels scared and victimized - two drastically different emotional reactions to the same situation. Taken together, these points suggests that emotional experiences are, to a large extent, idiosyncratic (Mandler, 1984). Finally, these two systems are believed to interact on the basis of a continuous feedback mechanism (Mandler, 1984). Arousal feeds back into the cognitive system, which registers the state of arousal and, together with a meaning analysis, constructs a subjective emotional state. This emotional state may then alter the autonomic functions, requiring further interpretation. For example, although a loud gun-like sound may induce arousal that is interpreted as fear, which subsequently or concurrently motivates you to scan the environment for possible dangers, the monitoring of the situation may inform you of the neutrality of this sound (e.g., produced by a passing car), thereby reducing the state of arousal and, therefore, the subjective feeling of fear. Consequently, the most important aspect contributing to subjective emotional experiences and their related behaviors is the cognitive-interpretive system. It is crucial in determining what 5 particular emotion will be experienced in any specific situation (Mandler, 1984). Furthermore, events in the external world and the organism's own actions can both contribute to the meaning analyses (as both provide valuable evaluative information), which in turn determine the subjective experience of the emotion. While past experiences guide our current interpretations, our actions provide the current context to which these cognitions can be attributed. Thus, this feedback system serves to modify current emotional experiences and influence future emotional responses. Responses leading to positive affect will be more likely to reoccur, and responses followed by negative emotional states will be less likely to reoccur. Emotional Processing: Congruent with this view of emotional development are dimensional theories of affect. In general, such theories are more concerned with how emotional information is perceived and processed than with how emotions develop. These theories imply that the structure of affective states, being highly related to one another, are best represented as a circumplex - a circular arrangement of affective terms around two correlated, yet independent dimensions (Larsen & Diener, 1992; Russell, 1980; Watson & Tellegen, 1988). Several different models have been suggested (e.g., Larsen & Diener, 1992; Watson & Tellegen, 1988), but these are simply variations of the original pleasure and arousal dimensions (see Figure 1) (Russell, 1980). ^^rousal 'Pleasure Figure 1: Emotional Circumplex (adapted from Russell, 1980) 6 According to this model, emotions vary along these two dimensions in an interrelated and systematic fashion. For example, depression would be viewed as being unpleasant and low in arousal, distress as being unpleasant and high in arousal, excitement as being pleasant and high in arousal, contentment as being pleasant and low in arousal, and so on. Although both components are necessary for an emotional evaluation, neither is sufficient. Being aroused could describe both feelings of excitement and fear, yet only by specifying how they differ in terms of their pleasantness can one differentiate them. Similarly, although excitement and contentment are both pleasurable experiences, they are characterized by their arousal level, with the former being high in arousal and the latter low in arousal. It is important to note that arousal, in this scheme, refers to perceived arousal rather than actual physiological activation (Feldman, 1995b; Mandler, 1984; Russell, 1989), a point elaborated on below. There is now ample evidence suggesting that these two dimensions play a significant role in how people conceptualize their own emotional states (e.g., Russell, 1978; 1979; 1980; Russell & Mehrabian, 1977; Feldman, 1995a; 1995b). For example, research has found that these two dimensions account for most of the variance in commonly used scales of affect (Russell & Mehrabian, 1977; Russell & Steiger, 1982), as well as in self-reports of current emotional states (Russell, 1979; 1980). In addition, pleasure and arousal are also the two most commonly extracted dimensions in studies of emotional processing. When participants are asked to evaluate emotional words (e.g., sad, happy, angry, content, sleepy and aroused) in terms of their perceived similarity or degree of pleasantness and arousal, they consistently do so according to Russell's dimensions, with the end result being a circular ordering of the emotional words (e.g., Bush, 1973; Russell, 1978; 1980; Russell, Lewicka, & Niit, 1989; Russell & Ridgeway, 1983; Feldman, 1995a; 1995b, Whissell, 1982). Similar findings have also been noted with non-verbal material, such as emotional scenes and pictures of facial affect (e.g., Russell & Bullock, 1985; 7 1986; Russell & Lanius, 1984; Russell, et al., 1989), and with various methodologies - verbal self-report, semantic differential ratings, successive-intervals scaling, principle-component analysis, and multidimensional scaling of affective stimuli (e.g., Russell, 1978; 1980). Furthermore, these findings have been consistent across age groups, genders, and cultures (Russell et al., 1989; Russell & Bullock, 1985, 1986). Hence, there exists an impressive amount of evidence, from various sources, that is consistent with a bipolar representation of affect. Nonetheless, not all theorists agree that affect can be represented in a bipolar geometric space (e.g., Ekman & Friesen, 1987; Izard, 1992; Frijda; 1992; Plutchik, 1980; Tomkins, 1984). Still, most support the idea that Russell's dimensions are very important to the human emotional experience (see Mandler, 1984; Feldman, 1995a; Russell, 1980). Of special interest is the congruency between Mandler's emotional acquisition theory and Russell's affective processing model. Mandler (1984) suggested that an emotional event evokes both autonomic arousal and meaning analyses, with the latter giving rise to affective intensity and the former to affective quality. As such, Russell's arousal dimension could be viewed as a reflection of autonomic activity elicited by emotional situations, and the pleasure dimension as a qualitative interpretation of that activity (Russell, 1989). Although used to evaluate actual emotional reactions, these two dimensions are still dependent on the interpretive system, especially at times when one is asked to report, and thus interpret, an emotional state. Only arousal that is perceived (interpreted) as emotional influences one's emotional reaction (Blascovich, 1990; 1992; Mandler, 1984). As such, the valence dimension, relying exclusively on cognitive systems, should account for more variance then the arousal dimension, which itself reflects both cognitive and physiological activation. In support of this, research on the structure of the circumplex to self-reported mood ratings has found that valence accounts for approximately twice as much of the variance as does arousal (e.g., Feldman, 8 1995b; Russell, 1980). A similar reasoning can be applied to the processing of affective material. Specifically, although emotional words and facial expressions do not necessarily give rise to autonomic arousal, such stimuli come to acquire their affective meaning through experience. That is, they become associated, through classical conditioning (see Mandler, 1984), not with autonomic arousal per se, but with a cognitive representation of both the arousal and its interpretation. Thus, unlike self-reported mood, the evaluation of emotional material, relying primarily on cognitive representations, should result in a more equal weighting of the dimensions of affect, which is indeed what research has indicated (e.g., Feldman, 1995b; Russell, 1980). In light of this strong cognitive component, some have suggested that these dimensions represent the semantic components required to interpret and communicate conscious, affective experiences (Feldman, 1995b; Russell, 1980). This suggests that the importance of actual autonomic arousal seems to be context-dependent, while that of valence may be context-independent (Feldman, 1995b). It is also likely that the perception of arousal depends on the intensity of the cues that elicit it. Thus, the findings that arousal accounts for only half of the variance of pleasantness in self-reported mood ratings may be more an artifact than reality. It is likely that laboratory research paradigms, in light of ethical considerations, do not invoke strong levels of arousal, thereby reducing the weight of the arousal dimension. Presumably, more intense situations would result in higher levels of autonomic arousal, which would then play a significant role in people's affective interpretations, as reflected in the bungy jumper's report of the event as being " a real high or rush". Individual Differences in the Emotional Circumplex: In contrast to Mandler's view (1984), Russell's research suggests a high degree of consistency in how people approach emotional situations. Yet, consistency does not negate the 9 possibility of individual differences. As Mandler (1984) noted, although the underlying framework may be similar, experience, personality, and physiology can all lead, either alone or in combination, to idiosyncratic emotional systems. In light of this, Feldman (1995a) investigated the possibility of individual differences in the affective circumplex. She had participants rate their affective states over a period ranging from 62 to 91 consecutive days using a mood questionnaire constructed to represent the entire emotional circumplex. In addition, she had participants rate the degree of similarity between mood-related words. As expected, she found variations in the circumplex. Some individuals placed more weight on pleasantness when evaluating their daily emotional experiences, while others placed more weight on arousal. Similarly, she found that some of her participants relied more on pleasantness, while others on degree of arousal, when judging the similarity between affective terms. She referred to the former individuals as valence-focused and to the latter as arousal-focused (see Figure 2). These two indices were also found to be significantly and negatively correlated to one another. Thus, an arousal-focus seemed to preclude a valence-focus, and vice versa. Valence-Focus Arousal-Focus Figure 2: Variations in the emotional circumplex (adapted from Feldman, 1995a) However, although focusing more on one component, her participants nevertheless relied on both dimensions when asked to evaluate emotional events. On the one hand, while valence-10 focused individuals tended to rely much less on arousal-based information, they still extracted some information from it. On the other, the arousal-focused individuals, extracting more arousal-related information, still depended on valence-related information to quantify their emotional reactions. That is, although the valence-focused individual placed significantly more weight on valence than on arousal, the arousal-focused person only showed a slightly greater weighing of valence as compared to arousal. These findings support Mandler's (1984) position that the cognitive-interpretation system, which presumably gives rise to a valence judgment, is the most important aspect of subjective emotional experiences (Feldman 1995a; 1995b). In addition, these idiosyncratic emotional patterns correlated differently with other affective constructs, thereby further validating her findings (Feldman, 1995a). While participants who were arousal-focused were better able to differentiate between arousal categories, those who were valence-focused were more successful at separating valence categories. Feldman (1995a) therefore proposed that her results are indicative of an attentional bias such that those who are arousal-focused pay more attention to the arousal components of their affective states and to arousal-related information of their environment, while those who are valence-focused demonstrate the opposite attentional bias. However, it is important to note that no matter the type of attentional bias, this bias is not automatic but rather mediated by the cognitive-interpretative system, which then instructs the organism to pay attention to one component over the other (Mandler, 1984). Unfortunately, although having potentially important implications, little is yet known about how these individual differences in affective focus are related to personality and how, if at all, they influence behavior. Affect. Personality, and Behavior; Considering the importance of affect in the human experience and the extensive l l variations between human affective repertoires, one can assume that such differences do exist. As proposed by Mandler (1984), there likely are individuals who vary in their sensitivity to arousal, and individuals who focus more strongly on one component, either on their autonomic arousal or their cognitive-interpretation, than on the other. Therefore, one could argue that such fundamental differences would, at the very least, influence how people interact with their environment. The arousal-sensitive individual, getting enough stimulation in his/her day-to-day activities, is not likely to be the first person in line at a bungy jump, although his/her arousal-insensitive counterpart, anticipating the rush, is likely to have camped there over night. Indeed, there is mounting evidence to suggest that different affective structures should lead to different action tendencies. For example, recent research has shown that different emotions lead to different actions, action tendencies, and goals, showing that affect has motivational properties (Roseman, Wiest, & Swartz, 1994). Thus, it should be no surprise that Feldman (1995a), linking her findings to Blascovich's (1990; 1992) biopsychosocial model of affect labeling and regulation, made a similar statement. Like Mandler (1984), Blascovich (1990; 1992) proposed that emotional reactions are evaluated in terms of their intensity and quality. As did others before him, he emphasized the importance of defining arousal as that which is "perceived" as emotionally important, rather than simply resulting from any physiological activation (Feldman, 1995b; Mandler, 1984; Russell, 1989). As such, the labeling of affect would first require the perception of arousal and, thus, depend on one's sensitivity to this component. Blascovich's (1990; 1992) theory postulates that individuals varying in degrees of arousal-sensitivity focus on different aspects of emotional events when qualifying their affect. Mandler (1984) suggested that either internal or external signals can alert the individual to an emotional experience and that there are likely some who attend to one aspect over the 12 other. While the hypersensitive individual is likely to attend to internal (somesthetic) over external (environmental) cues when labeling his/her affect, the hyposensitive person is more likely to rely on external rather than internal ones (Blascovich, 1990; 1992). That is, the hyposensitive place more weight on their interpretation of the emotion-evoking event, and the hypersensitive place more weight on their interpretation of the perceived arousal (Mandler, 1984). Nonetheless, one's ability to focus on one's arousal sensations is likely to improve as one experiences higher degrees of arousal. Thus, while the arousal hypersensitive individual is likely to perceive all gradations in arousal, the arousal hyposensitive person is likely to only react to his/her internal cues in states of high arousal. As such, arousal will be important to affective labeling in conditions of high arousal, irrespective of one's sensitivity to this emotional dimension (Blascovich, 1990; 1992; Feldman, 1995a). However, this sensitivity will have different consequences for hypersensitive and hyposensitive individuals. While the hyposensitive individual is likely to enjoy high-arousal situations, the hypersensitive one is likely to become extremely uncomfortable and thus disturbed by such emotional intensity. For example, the former is likely to describe his/her bungy jumping experience as a positive experience (e.g., a real rush), while the latter is likely to have felt panicked and terrified by the experience (e.g., a near death experience). But what happens to the hyposensitive individual in states of low arousal? If such individuals rely only on external cues to alert them to emotional events, then there should be a restricted range of events which they would deem emotional (i.e., highly arousing). Yet, they are still able to make sense of such events. Note that an emotional event is first and foremost an event and, as such, it will evoke some cognitive interpretation, presumably based on one's past experiences (Mandler, 1984). In more innocuous situations, hyposensitive individuals probably revert back to some cognitive interpretation, which may be based on past experiences, the 13 environmental context, and/or reflecting some type of social desirable response. As such, their interpretations of relatively unarousing emotional events, being primarily cognitive in nature, are likely to be less informative than those more sensitive to arousal, whose interpretations include the important information that affect conveys. Thus, an individual's attentional capacity, at least as it pertains to affect, seems split between internal and external cues (Blascovich, 1990; 1992), implicating a negative relationship between these two types of attentional biases. As noted previously, such a negative correlation exists between valence- and arousal-focus. Moreover, research has found that valence interpretations partially reflect social desirability (e.g., Feldman-Barrett, 1996), a component that is definitely based on external cue. Accordingly, Feldman (1995a) proposed that a valence-focus, as induced in the hyposensitive, could be seen as related to allocating attention to the environmental context in which emotions arise, while, on the other hand, arousal-focus, as seen in the hypersensitive, would be related to an attentional bias towards somesthetic stimuli. However, this interpretation stands in contrast to optimal level of stimulation or arousal theories (see Ellis, 1987; Zuckerman, 1979). The Role of Arousal: As reflected in the theories of Mandler (1984) and Blascovich (1990; 1992), arousal theory suggests that there exist considerable individual differences in people's baseline levels of arousal. However, to function adequately, organisms do require some optimal level of arousal -one that is neither too low nor too high (Ellis, 1987). It follows that the intensity/novelty of stimuli required to keep one's state of arousal at an optimal level may be different for different people (see Ellis, 1987; Zuckerman, 1979). In general terms, one could think of three different types of people - high, medium, and low in baseline arousal - with different behavioral and 14 motivational presentations (Ellis, 1987). The highly aroused individual, who is already in a state of tension (e.g., the anxious, neurotic, and introverted), is likely to avoid further environmental stimuli; the one whose baseline levels are in the average range (e.g., the well-adjusted) is likely to interact with his/her environment in a predictable and stable fashion; and the low aroused person (e.g., the extroverted and sensation seekers), in trying to reach optimal levels of arousal, is likely to actively seek out stimulation in his/her environment. In relation to Blascovich's (1990; 1992) model, highly aroused individuals, having a low arousal threshold, should be very alert to their internal state of arousal, while those with low resting states of arousal should be very alert to arousal-related signs in their environment. In this scheme, the highly aroused and the under aroused refer to hypersensitive and to hyposensitive individuals, respectively. Therefore, unlike Feldman's (1995a) proposal, being hypersensitive to arousal would engage a valence-focus rather than an arousal-focus, while being hyposensitive to arousal would engage an arousal-focus, not a valence-focus. Although this may seem counterintuitive, it actually expands on Feldman's (1995a) findings more so than her original hypothesis. The affective structure, as postulated by Mandler (1984), works through a continuous feedback mechanism between arousal and the cognitive-interpretative system. This process is triggered either through internal or external cues: the hypersensitive by internal cues and the hyposensitive by external cues (Blascovich, 1990; 1992). The cue, irrespective if internal or external, will then trigger a meaning analysis, which is essential for the subjective experience of affect, that will attribute a qualitative interpretation to the emotional event (Mandler, 1984). Thus, the interpretive system of the hypersensitive individual will be focused on internal states and that of the hyposensitive will be focused on external cues. The hypersensitive, being almost obsessive about his/her internal state, will therefore be very alert to the "goodness or badness" of a wide range of arousing situations, 15 thereby explaining the greater weight placed on the valence dimension of his/her emotional experience by the meaning analysis. On the other hand, the hyposensitive individual, desperately requiring stimulation, will be very sensitive to the arousing aspects of his/her environment, explaining the greater importance that they place on the arousal dimension. However, the hyposensitive, interpreting highly arousing events as pleasurable and low-to-moderate ones as neutral in connotation, still relies on the valence dimension. That is, while the hypersensitive individual's affective circumplex is more elliptic, that of the hyposensitive person is more circular, as reflected in Feldman's (1995a) findings. Arousal sensitivity may restrict the usefulness that an arousal interpretation has for one's subjective emotional state. In summary, theories of emotional acquisition, processing, and labeling all suggest that the subjective emotional experience is based on the integration of two systems: one that quantifies (i.e., mediates arousal) and another that qualifies (i.e., interprets) emotional events (Blascovich, 1990; 1992; Mandler, 1984). As such, arousal and pleasure (the interpretation) have been deemed as central to the expression and evaluation of emotional events (Feldman, 1995a; 1995b; Russell, 1980). However, although these constitute the foundation of emotional experiences, they are nonetheless relied upon to differing degrees by different people (Blascovich, 1990; 1992; Feldman, 1995a; Mandler, 1984). On the one hand, some individuals seem to place more importance on valence (i.e., valence-focus) when interpreting emotional events, while, on the other hand, some place more weight on arousal (i.e., arousal-focus) when evaluating emotional situations. This distinction is believed to be mediated by one's sensitivity to somesthetic sensations such that being hypersensitive or hyposensitive to arousal results, respectively, in a valence-focus or an arousal-focus (Blascovich, 1990; 1992; Ellis, 1987; Mandler, 1984; Zuckerman, 1979). As such, a valence-focus serves to evaluate the goodness or badness of all perceived arousal and an arousal-focus serves to alert an individual to possibly 16 arousing events, helping him/her replenish his/her normally under-aroused state (Blascovich, 1990; 1992; Feldman, 1995a). Furthermore, there is a strong theoretical and empirical basis to suggest that these attentional biases, in turn, result in systematic personality and behavioral presentations (Ellis, 1987; Feldman, 1995a; Zuckerman, 1979). Although the hypersensitive, valence-focused individual is likely to avoid further autonomic stimulation, the hyposensitive, arousal-focused person is likely to seek such stimulation in his/her environment. One could thus speculate that the valence-focused individual is more likely to be introverted and neurotic, while the arousal-focused person is more likely to be extroverted and sensation-seeking (Zuckerman, 1979), and - if pathologically under-aroused - to be a psychopath. Psychopathy and Physiological Arousal As noted above, there are differences in the amount of stimulation that people require in order to function (Zuckerman, 1979). While those with high baseline levels of physiological arousal will tend to avoid further stimulation, those with low baseline levels will actively seek it out. Following in these theoretical formulations, several investigators have proposed that since psychopaths actively seek out stimulation, then they must suffer from an abnormally low level of arousal (see Hare, 1978 for a review). As highlighted by Quay (1965), "the level and variation of sensory inputs which are necessary for the maintenance of pleasant affect are much greater for the psychopath than for the ordinary individual (p. 181)". Consequently, in light of a growing body of research showing cardiovascular, electrodermal, and electrocortical abnormalities in psychopaths, it was further proposed that a chronic autonomic and cortical hypo-arousal was at the foundation of the psychopathic personality disorder (see reviews by Hare, 1970, 1978). Indeed, Hare (1970) remarked that many of the traits of the psychopath (e.g., apparent lack of anxiety, guilt, or empathy) have autonomic components that reflect a deficiency in this area. 17 Unfortunately, much of this earlier work produced inconsistent findings, probably due to differences in the selection of participants (i.e., assessment procedures), the relative insensitivity of the technology of the time, the types of responses measured, and procedures used between laboratories. These inconsistencies indicated that the "cortical arousal theory" was relatively unsubstantiated and, therefore, too weak to further our understanding of this disorder (Syndulko, 1978). These differences notwithstanding, some consistencies emerged suggesting that psychopaths, assessed according to Cleckley's (1976) formulation of the disorder, are indeed hyposensitive to arousal. Firstly, during monotonous, repetitious, and boring (i.e., low arousal) tasks, the electrodermal response of psychopaths tend to decrease more rapidly than that of nonpsychopaths (Hare, 1968, 1978). However, in tasks of higher arousal, such as those requiring more effort on the part of participants, these group differences decreased, sometimes to the point of nullifying the differences altogether (Hare, 1978, 1980). Thus, as predicted by arousal theory (Ellis, 1987; Zuckerman, 1979), psychopaths seem to require more stimulation to perform at optimal levels. Secondly, psychopaths have also been found to consistently show lower electrodermal conditioning to fearful events (see Hare, 1978; Newman & Wallace, 1993 for reviews). This suggests a deficient fear reaction, which would be expected in those who prefer a lifestyle marked by thrill seeking activities (see Ellis, 1987; Zuckerman, 1979). Thirdly, psychopaths have also been found to show a unique electrodermal, cardiac, and cortical pattern in anticipation of a punishment (e.g., shock, loud noise). Although nonpsychopaths show the expected increases in skin conductance, larger cortical responses, and moderate decreases in heart rate, psychopaths consistently show relatively small increases in skin conductance and cortical responses, and large increases in heart rate in anticipation of a punishment (Hare, 1978; Larbig, Veit, Rau, Schlottke, & Birbaumer, 1992; Ogloff & Wong, 1990). Based on research on 18 the autonomic components of the orienting and defensive responses, Hare (1978; 1998) proposed that this idiosyncratic response reflects an active coping mechanism such that psychopaths are able to "tune out" the impending punishment, thereby reducing the fear associated with it. Accordingly, psychopaths, insensitive to impending punishment, would be able to engage in riskier behaviors, thereby further helping them achieve optimal arousal levels (Ellis, 1987). In addition, recent research suggests that relatively emotionally innocuous stimuli may not impact psychopaths to the same degree as nonpsychopaths. Several studies have found that emotional processing, of both linguistic and non-linguistic stimuli, is less lateralized to the right hemisphere in psychopaths (Day & Wong, 1996; Forth, 1992; Intrator, Hare, Stritzke, Brichtswein, Dorfman, Harpur, Bernstein, Handelsman, Schaefer, Keilp, Rosen, & Machac, 1997; Mills, 1995). For example, Day and Wong (1996) found that processing of negative emotional words - which are relatively innocuous as compared to real life events, is less lateralized to the right hemisphere in psychopaths. Since modulation of ANS responses that result in actual felt emotional reactions seems mediated by the right hemisphere (see Gainotti, Caltagirone, & Zoccolotti, 1993 for a review), suggesting that low arousing, yet emotional stimuli may be less salient to the psychopath (Day & Wong, 1996) - further implying that psychopaths are arousal hyposensitive. Thus, although the "cortical arousal theory" has fallen from favor over the years, one can not deny the usefulness of "arousal theory" in trying to understand this disorder. Not only is there autonomic evidence supporting this view (as cited above), but the psychopath is often referred to as the prototypical sensation-seeker (Zuckerman, 1979). Indeed, a diagnostic feature of the disorder is a need for stimulation and an inability to tolerate boredom (e.g., PCL-R items 3, 10, 11, and 20; Hare, 1991). Moreover, Ellis (1987), reviewing the literature on psychopathy, found eight behavioral correlates of psychopathy relating to under-arousal: resistance to 19 punishment, poor academic performance or performance below intellectual potential, impulsiveness or minimal persistence at tasks assigned and coordinated by others, childhood hyperactivity, risk taking and sensation seeking, recreational drug use, preference for active social interactions, preference for broad-ranging sexual experiences and unstable pair bonding tendencies. In summary, there exists some evidence linking psychopathy to deficiencies in autonomic arousal (e.g., Day & Wong, 1996; Forth, 1992; Hare, 1970; 1978; Intrator et al., 1997; Mills, 1995; Newman & Wallace, 1993; Ogloff & Wong, 1990). Accordingly, arousal theory seem to hold some promise in terms of furthering our understanding of this disorder (Ellis, 1987; Zuckerman, 1979). Being generally under-aroused and hyposensitive to arousal, psychopaths should require much higher levels of stimulation in order to function in their day-to-day lives. That is, as predicted by Mandler (1984), Blascovich (1990; 1992), Zuckerman (1979), and Ellis (1987), this need for stimulation should be reflected in the psychopath's behaviors. Moreover, these behaviors, consistent with a pathologically under-aroused state, should be much more extreme and actively engaged in than what would be seen by those with more normal variations in arousal sensitivity (Feldman, 1995a). Psychopathy and Arousal-Seeking The Behaviors: In their pathologically under-aroused state, psychopaths should demonstrate a unique behavioral pattern: one solely aimed at acquiring arousal. Indeed, in their search for stimulation, psychopaths tend to be nomadic individuals, indiscriminately moving from town to town, job to job, and (sexual) partner to partner. This need for arousal is further reflected in their criminal patterns. Psychopathic criminals, as compared to their nonpsychopathic counterparts, tend to 20 have a much more versatile criminal history (Hare, 1991). For further excitement, psychopaths often take unnecessary risks (e.g., unprovoked violence), which do not necessarily lead to material rewards. For example, a recent study by the Federal Bureau of Investigation (1992) found that of those offenders who killed on-duty officers (one of the most intense criminal acts), the personality features of a large proportion of them resembled that of the psychopath. Presumably, such individuals habituate quickly to the intensity of their crimes and move on to more exciting things. Thus, add a lack of empathy for others and an extremely egocentric view of the world ,and you get a sensation seeker with few or no boundaries, one who is able to commit acts of unbelievable cruelty for nothing more than the sheer pleasure of it (Zuckerman, 1979). Indeed, this aspect of psychopathy is so central to the disorder that some have suggested that psychopaths are actually addicted to violence (e.g., Hodge, 1992). However, although engaging in an inordinate amount of violence, psychopaths nonetheless engage in other, more socially acceptable types of thrill seeking behavior, suggesting an addiction to excitement rather than to the more specific concept of violence. The psychopath, much like a predator seeking its prey or the addict seeking his/her fix, perpetually seeks his/her rush. Accordingly, the psychopath's pathological need for stimulation has serious behavioral consequences - affecting society and, more specifically, their victims in unimaginable ways. Several investigations have found that psychopathy is one of the best predictors of crime, especially that which is extremely violent and, thus, highly arousing. Psychopathic offenders, as compared to their nonpsychopathic counterparts, commit more crimes (Hare & Jutai, 1983); engage in more violent and aggressive behaviors (Hare & McPherson, 1984); are more violent within institutions (Hare, 1981; Hare & McPherson, 1984); sexually offend indiscriminately and do so in an opportunistic and angry fashion (Brown & Forth, 1997; Porter, Drugge, Fairweather, & Herve, in preparation); and recidivate upon release at a higher and faster rate and do so more 21 violently (see Hemphill, Templeman, Wong, & Hare, 1998). Accordingly, their apparent lack of concern for others, fueled by their need for arousal, enables them to act in ways that most others could not fathom. Psychopaths, as exemplified by their post-incarceration pattern of re-offending, actively seek such situations, presumably to compensate for their normally under-aroused state (Ellis, 1987; Zuckerman, 1979). Furthermore, they are not likely to overlook an arousing event, as seen in their opportunistic sexual histories. A Proactive Attitude: More importantly, the psychopath's violent and aggressive behaviors are more goal directed than that of other offenders. Psychopaths seem to "seek out" violence for their arousing attributes, while the nonpsychopaths, presumably having a lower arousal threshold, are more likely to become violent "in reaction to" emotional arousal. Williamson, Hare, and Wong (1987) found that a large proportion of the nonpsychopaths in their study, and none of the psychopaths, committed murders during a domestic dispute or under periods of extreme emotional arousal. Thus, while nonpsychopathic offenders murder in reaction to some emotionally evoking situations, psychopathic offenders commit their violent crimes purely to fulfill personal needs (e.g., material gains, revenge, or retribution). Interestingly, this suggests that psychopaths do not get upset over interpersonal disputes. Indeed, recent research with wife assaulters has found this to be the case. Jacobson and Gottman (1998), for example, found that approximately 20% of the males in his sample showed a peculiar heart rate response when engaged in a heated debate with their wives. These males, who apparently fit the personality profile of the psychopath, appeared angry and aggressive but showed a decrease in heart rate during the debate, rather than the expected increase seen in their nonpsychopathic counterparts. This decrease in heart rate suggested that the psychopathic husbands were actually interested by the evolving conflict, 22 becoming internally calm despite their emotionally aggressive behaviors. Showing an interest in the argument, psychopaths would therefore be less likely to experience the intense and negative emotional arousal that leads nonpsychopaths to kill their spouse. Consequently, psychopaths, seeking further excitement, are more likely to be violent towards male strangers, who would give them a better fight, while nonpsychopaths, who are presumably easily aroused, are more likely to aggress female acquaintances that "upset" them (Williamson, et al., 1987). In addition, as noted above, it is likely that the psychopath's affective structure, founded upon a system that is hyposensitive to arousal, is systematically different from that of the nonpsychopath (see below), causing him/her to be oblivious to this aspect of human experiences. Intrigued by this finding, Cornell and his colleagues (1996) investigated the type of aggression that psychopaths are likely to engage in. Research in this area suggests that there exist two types of aggressive acts: one motivated by an affective response, and the other by some external goal or reward. The former is termed reactive aggression, which occurs in response to feelings of frustration, provocation, or perceived threat. The latter is termed instrumental aggression, which denotes a self-initiated, goal-directed, purposeful and predatory type of violence. As predicted, although psychopaths and nonpsychopaths were as likely to have engaged in some reactive aggression, psychopaths were found to commit significantly more crimes involving instrumental aggression (Cornell, et al., 1996). Recent research has also shown that nonpsychopaths are more likely to engage in instrumental violence due to some external/environmental factor, while psychopaths are more likely to do so for some personal reward or satisfaction (Herve, Petitclerc, & Hare, in preparation). That is, instrumental violence is more egosyntonic for the psychopath (e.g., to show off to peers or to get away with their crime), while that of the nonpsychopath is more egodystonic (e.g., in relation to an addiction or to pay off debts). Accordingly, if stimulation does 23 not find the psychopath, then the psychopath will find stimulation. As predicted by Blascovich (1990; 1992), psychopaths - being hyposensitive to arousal - will look towards their environment for arousal-related cues and, being extremely impulsive, will immediately act upon such information. Furthermore, a tendency towards violence seems related to an offender's emotional sensitivity. Offenders with a high degree of emotional insensitivity, as reflected by higher scores on Factor 1 of the PCL-R, are more likely to engage in violent acts, while the offender's history of impulsive and antisocial behaviors, as reflected in Factor 2 of the PCL-R, seems to be more related to the manner in which the violence is conducted (Dempster, Lyon, Sullivan, Hart, Smiley, & Mulloy, 1996), further suggesting that psychopaths' arousal seeking is related to their affective structure. In summary, as postulated by arousal and emotional theories (Blascovich, 1990; 1992; Ellis, 1987; Mandler, 1984; Zuckerman, 1979), psychopaths, being pathologically under-aroused, seem to be characterized by a need for stimulation that motivates them to seek the most arousing aspects of life, from the socially acceptable to the outright despicable (see Hare, 1996, 1997 for reviews). Accordingly, if psychopaths are relatively unresponsive to low levels of arousal (i.e., are hyposensitive), which is a necessary condition for normal emotional growth (Mandler, 1984), then their affective structure should be different that that of individuals who's arousal system is unimpaired (Blascovich, 1990; 1992). More specifically, the psychopath's affective structure should have developed an attentional bias for arousal-related information, which - if stimulating enough - they can act upon to help them replenish their normally under-aroused state (Blascovich, 1990; 1992; Feldman, 1995a; Roseman et al., 1994). Psychopathy and Affect Advances in the field of emotions indicate that individuals varying in their arousal 24 sensitivity should show different personality, behavioral, and affective patterns. As reviewed above, there exists ample evidence that the psychopath is indeed inflicted with an under-aroused autonomic system, one reflecting a pathological hyposensitivity to arousal. Consequently, to achieve optimal arousal levels psychopaths seek out emotionally intense situations - usually resulting in great costs to society. If psychopaths are indeed pathologically hyposensitive to arousal, then they should be highly focused on arousal-related information in their environment. That is, their affective structure should be different than that of their nonpsychopathic counterparts, which coincides with both clinical and empirical evidence. Cleckley (1976) proposed that psychopaths suffer from a "semantic aphasia": a deep-seated semantic disorder, presumably caused by some brain dysfunction, in which the affective and semantic components of language are dissociated. Similar observations have also been noted by others. For example, Johns and Quay (1962) remarked that psychopaths "know the words but not the music" (p.217); Grant (1977) considered psychopaths to know "only the book meaning of words" (p.50); and Gillstrom and Hare (1988) suggested that emotions are "like a second language to the psychopath". Such observations suggest that although psychopaths may understand the literal (denotative) meaning of affective events, they seem unable to appreciate, in its entirety, their more abstract, emotional (connotative) significance. In support of this long-standing clinical tradition, several investigations have found that the language, cognitions, and experiences of psychopaths appear to lack depth and affective meaning (for a review see Hare, 1998). Unfortunately, most of the research in this area has employed valence-related stimuli. Thus, although much is known about the psychopath's deficiency in processing valence-based information, little knowledge exists concerning their ability to grasp the other component of affect - the arousal dimension. Nonetheless, clues to the psychopath's affective structure have 25 surfaced. One could argue that, at the very least, psychopaths are not valence-focused individuals. As shown by Feldman (1995a), the valence-focused are quite able to differentiate discrete aspects of valence-related information, suggesting that the psychopath, being insensitive to this aspect of affect (see below), is not valence-focused. This leaves us with two possibilities: psychopaths can either have a normal affective structure or one that is arousal-focused. Obviously, one can not possibly suggest that the psychopath's affective structure is anywhere close to being normal. As such, since the lack of a valence-focus suggests an arousal-focus (Blascovich, 1990; 1992; Feldman, 1995a), psychopaths must be arousal-focused individuals. With this point in mind, we now turn to research on the psychopath's emotional deficit. Psychopathy and Valence-Focus: There is considerable evidence indicating that psychopaths suffer from a profound emotional deficit. Unfortunately, in contrast to advancements in the field of emotional processing, research in this area has taken a unidimensional view of affect, almost exclusively focusing on the valence dimension. Nonetheless, if the psychopath's affective structure does demonstrate an arousal-focus, as predicted by his hyposensitivity to arousal, then certain predictions can be made concerning how he should process valence-related information. As demonstrated by Feldman (1995a), arousal-focused individuals are less proficient at distinguishing between valence categories than their valence-focused counterparts. As such, psychopaths should be less sensitive to differences in material varying in emotional valence. That is, they should not differentiate pleasant from unpleasant stimuli to the same extent as nonpsychopaths. Indeed, research has shown this to be the case. For example, Williamson, Harpur, and Hare (1991) found that, at the cortical level, psychopaths did not differentiate between pleasant and unpleasant words. However, the strongest evidence for this comes from 26 Patrick and his colleagues in their research on the psychopaths' emotional startle reflex modulation (see Patrick, 1994 for a review). In such studies, participants are presented with emotionally evocative and neutral slides to induce desired affective states. During some of the inter-slide intervals, an acoustic startle probe (i.e., sudden noise bursts) occurs at which time the eye-blink component of the startle response is measured. This component is believed to reflect a valence response to stimuli such that blink reactions are augmented during the exposure to unpleasant slides (e.g., mutilations, punishment) and reduced during exposure to pleasant slides (e.g., happy, erotic) as compared to neutral slides (e.g., household objects, non-expressive faces). Thus, if psychopaths, scanning for arousal-related information, are less sensitive to valence, then they should not show eye-blink differences between negative and positive slides. Accordingly, Patrick and his colleagues (1993) subjected 54 incarcerated sex offenders to the above paradigm while they recorded the blink startle reactions (BSR) to the noise probes and the skin conductance (SCR) and heart rate (HR) responses to each slide. Pleasant and unpleasant slides were matched for arousal, which exceeded that of the neutral slides. Although nonpsychopaths showed a significant linear trend across slide valence categories, the psychopathic sex offenders displayed a significant quadratic pattern (Patrick, Bradley, & Lang, 1993). That is, while the nonpsychopathic and mixed individuals showed the expected augmentation and reduction in BSR to unpleasant and pleasant slides, respectively, as compared to the neutral ones, the psychopaths BSR was reduced to both pleasant and unpleasant slides. This finding was further found to be related to subjects' overall emotional sensitivity. Within those sex offenders who received high Factor 1 scores on the PCL-R, it was found that higher scores predicted a diminished or reversed startle reflex to the aversive stimuli, placing further importance on the role of affect for understanding psychopathy. However, no autonomic group differences were observed, nor was the self-reported pleasantness or degree of arousal of the 27 slides for psychopaths any different than that of the nonpsychopaths. These authors concluded that this pattern of results lends further support to the idea that psychopathy is characterized by an abnormal processing of emotional stimuli and that this effect is best accounted for by the interpersonal/affective characteristics of the disorder (Patrick et al., 1993). More specifically, since the startle response is strictly a measure of emotional valence (see Patrick, 1994 for a review), they argued that the psychopath's emotional deficit is a failure to appreciate the significance of negative, aversive and potentially dangerous stimuli. Psychopaths do not seem to show the expected defensive, avoidant reaction to fearful situations, although showing an appropriate reaction to pleasant ones. However, the fact that psychopaths did show the normal autonomic differentiation between emotional and neutral stimuli suggests that they are nonetheless capable of identifying affective events, a finding also confirmed by language-based investigations (e.g., Williamson, Harpur, & Hare, 1990). Therefore, psychopaths' reactivity to fearful stimuli without a defensive response suggests that such individuals may easily approach dangerous situations in their search for arousal (see Ellis, 1987). Being insensitive to valence-related information, then, psychopaths should rely less on this dimension when evaluating emotional material, probably preferring to depend on something a little more tangible to them. Language can be used to express both denotative and connotative information. The former refers to the more concrete, referential dictionary meaning of words, while the latter refers to the more abstract, less explicit associative and emotional overtones of word meanings. Thus, while the word "cold" denotes "the absence of heat", it also connotes "meanness" or "ruthlessness". As such, one could expect psychopaths to use denotative over connotative information when having to evaluate the pleasantness of stimuli. To test this idea, Williamson, Harpur, and Hare (1990) used a word triad task where they had psychopaths and nonpsychopaths choose the pair of words, presented in series of three, that 28 were most similar in meaning. Unbeknownst to their subjects, the words (warm, cold, deep, shallow, loving, hateful, wise, and foolish) were chosen to allow relationship-based groupings into six possible categories. Of interest to the current discussion, is that two of these categories represented denotative relationships (antonym and domain) and that two represented connotative ones (metaphor and polarity). Psychopaths were found to use polarity (pleasant-unpleasant) relationships less often and use metaphor, domain, and antonym pairings as often as nonpsychopaths, supporting the hypothesis that psychopaths make less use of valence-related information when evaluating affective material. This difference notwithstanding, they did rely on some aspect of connotation, most notably metaphorical relationships, when performing this task, suggesting that they are not completely insensitive to this aspect of language (Williamson et al., 1990). However, their metaphorical use could have been based on a denotative instead of a more connotative, emotional understanding. As pointed by the authors, "affective similarity may be seen as one component of metaphorical similarity, and an absence of (or inability to judge) the former need not preclude an understanding of the latter" (Williamson et al., 1990). Psychopaths, not perceiving any arousal, are thus likely to skip over valence-based information choosing, rather, to simply denote what they see. Accordingly, this should result in a relatively lack of expertise in dealing with such information. The more that someone works at something the better they get at performing that task. For example, trying to learn how to speak French would be better accomplished by immersing oneself in a francophone community than by simply relying on part-time use of some text book. The in vivo experience is likely to result in a deeper understanding of the situation than the more detached book approached, with the latter resulting in greater confusion over the more subtle aspects of the language that the former. Thus, one could speculate that valence, to the inexperienced psychopath, is likely to have been learned under the more detached approach, 29 resulting in a lower comprehension of such material. Although psychopaths may have the ability to identify emotional material, if they are unskilled with the identification of valence, then they should perform poorly on tasks requiring knowledge of this dimension. To make decisions about polarity, one must be able to comprehend bipolar relationships. This was tested by Williamson and her colleagues (1990) in a second experiment in which they had 63 male criminals match phrases on the basis of inferred emotional information. Two types of errors could be made: polarity errors, in which participants matched the target to one sharing descriptive characteristics but of opposite emotional polarity; and non-polarity errors, in which participants matched the target to one that was neutrally toned and either sharing or not sharing descriptive characteristics. Although participants in general made few errors, the pattern of errors revealed that psychopaths made significantly more polarity errors and fewer non-polarity errors than nonpsychopaths, a finding consistent with that of others (e.g., Hayes, 1995). As noted above, emotional metaphors depict both connotative and denotative information. Although psychopaths are likely to grasp the more concrete meaning of emotional metaphors, they should show greater errors when specifically asked to evaluate their pleasantness. Interestingly, this is exactly what Hayes (1995) found when he had 35 offenders perform an emotional metaphor Q-sort. In this task, participants were required to hierarchically sort 60 metaphors into six piles, three positive and three negative in connotation, such that the extreme two piles contained two, the less extreme piles contained nine, and the remaining tow middle piles each contained 19 metaphors. This procedure results in a normal distribution of metaphorical valence. Although psychopaths and nonpsychopaths did not differ in their interpretation of the denotative meaning of the metaphors, psychopaths were found to make more mistakes then nonpsychopaths when using emotional valence as a sorting criteria. Moreover, 30 their mistakes were more likely to involve sorting errors that identified metaphors as being extreme members of the opposite valence category. This confusion of emotional polarity was also highlighted in the above startle-reflex experiment (Patrick et al., 1993). Psychopaths' identical SCR and BSR reactions to both pleasant and unpleasant slides suggests that they are confused by this dimension, possibly interpreting both pleasantness and unpleasantness as one and the same. This may help explain the clinical observations that although psychopaths enjoy the use of metaphors in their day-to-day interactions, they sometimes do so erroneously. Nonetheless, these findings do suggests that, although confused by the dimension of pleasantness, psychopaths do not seem to be completely oblivious to affective content. Rather, they are likely to be evaluating such material in a particular, unique fashion. In addition, psychopaths, being less proficient at processing valence-based material, should require more effort when having to process such information. That is, valence processing should require more cognitive effort on the part of such individuals. A recent single photon emission computerized tomography (SPECT) study by Intrator and her colleagues (1997) seems to suggest that this is indeed the case. They investigated differences in relative cerebral blood flow (rCBF) between psychopathic and nonpsychopathic individuals, recruited from a non-forensic substance treatment program, while they performed a lexical decision task using emotional (pleasant and unpleasant) and neutral words. Although psychopaths did not differ from nonpsychopaths in their behavioral responses (decision times, accuracy, and speed-accuracy trade off), psychopaths nevertheless demonstrated greater subcortical rCBF activity while processing the emotional words as compared to nonpsychopaths, suggesting that emotional processing - based on a unidimensional view of affect - demands more of the psychopath's brain than it does of the nonpsychopathic individual (Intrator, et al., 1997). Valence-related information seems like a second language to the psychopath, a language which requires a 31 considerable amount of effort and mental processes to grasp, (Gillstrom & Hare, 1988). This of course makes sense if one adopts the view that psychopaths are indeed arousal-focused. Not only does practice make perfect, but it also leads to a more efficient performance in terms of mental requirements. If valence is not necessarily seen as an affective component to the psychopath, then s/he should not evaluate it as such. While nonpsychopaths should rely on an emotionally-based cognitive-interpretative system when faced with any affective event, psychopaths, in all their hyposensitivity, should, when faced with relatively innocuous stimuli (e.g., that denoted by valence), revert to a cognitive interpretative system free from affective connotation (see Mandler, 1984). The fact that psychopaths rely more so on denotative than on connotative understanding when evaluating valence-based affective material highlights this quite well. However, if truly relying on some emotionally-free interpretive system, then psychopaths should extract less information from affective (valence-based) events, as affect is believed to add significantly to the information load of such material. Although research with normal individuals has shown that the extra informational load that affect adds to language is reflected in electrocortical activity, research with psychopaths has consistently demonstrated that this is not the case for them (Kiehl, Hare, McDonald, & Brink, in press; Williamson, et al, 1991). For example, in a study employing a lexical decision task (is it a word or not?) psychopaths, as compared to nonpsychopaths, failed to show the expected electrocortical differentiation, in the form of larger event-related brain potential (ERP) amplitudes to emotional versus neutral words (Williamson, et al., 1991). Moreover, this insensitivity to affect was observed irrespective of word valence (i.e., positive or negative) and did not result in any accuracy differences between the two groups. Since psychopaths have not been found to process more simple, non-affective words any differently that nonpsychopaths 32 (Hare, 1979), such an effect could not have been due to some perceptual or general language deficit. Accordingly, psychopaths do not necessarily evaluate words, defined in terms of valence, as being emotional in connotation. Finally, if indeed insensitive to valence-based information, then psychopaths should not be influenced by such information within their environment. That is, since psychopaths are presumably scanning their world for arousal-related cues, they should be relatively unaffected by the pleasantness or unpleasant of events. A well defined effect in the eye-witness literature is the narrowing of attention to the more salient, central aspects of a negative, emotional event (e.g., a crime scene), at the detriment of more peripheral, irrelevant aspects of that same scene. This finding is believed to be related to the emotional intensity, or negative arousal, that such scenes evoke for the witness. Much like the "tunnel-vision" that one suffers when trying to avoid a car accident, individuals witnessing a highly arousing crime also experience a kind of "tunnel-vision"; an effect that could be described as a strong empathetic response. This narrowing of attention has since been empirically validated such that individuals, faced with a negative emotional scene (e.g., a hit-and-run), remember the central detail better than the peripheral ones. This parallels real life situations in which, for example, a witnesses to a crime can often give detailed reports of the victim and/or the offender, while drawing a blank when asked if other witnesses were at the scene. Based on this, Christianson and his colleagues (1996) postulated that psychopaths, being emotionally deficient, would not show this narrowing of attention when faced with a negative emotional event. To test this, they had a group of offenders, varying in psychopathy, recall the central and peripheral details of emotionally negative (e.g., a women hit by a car as she was riding her bicycle, with another car in the background) and neutral (e.g., a women riding her bicycle in front of a car, with another car in background) scenes. As predicted, their were no 33 group differences in the recall of the details of the neutral slides, with both groups recalling as many of the central as the peripheral details. In addition, the nonpsychopaths demonstrated the expected narrowing of attention for the emotional slide, with more central details remembered as compared to peripheral ones. In contrast, psychopaths' recall performance of the details of the emotional slide paralleled that of the neutral slide. That is, they recalled as many of the central as the peripheral details, regardless of the slides valence (Christianson, Forth, Hare, Strachan, Lidberg, & Thorell, 1996). These results further suggest that psychopaths are affected differently by the emotional experiences that the world offers them. To summarize, research has demonstrated that psychopathy is not characterized by a valence-focus. Psychopaths, as compared to nonpsychopaths, are less sensitive to differences in emotional valence (e.g., Patrick et al., 1993; Williamson, et al., 1990); rely less on valence to make sense of emotional material (e.g., Williamson, et al., 1990); make more mistakes when forced to make valence-based decisions (e.g., Hayes, 1995; Patrick et al., 1993; Williamson, et al., 1990); require greater cognitive demands when having to process such information (e.g., Gillstrom & Hare, 1988; Intrator et al., 1997); do not necessarily perceive valence as an emotional component and thus extract less information from valence-based material (e.g., Kiehl et al., in press; Williamson et al., 1991); and consequently are not affected by events depicting valence-related information (e.g., Christianson et al., 1996). However, although insensitive to valence, psychopaths are nonetheless able to identify emotional events (e.g., Patrick et al., 1993; Williamson, et al., 1990). As such, this suggests a variation in their affective structure rather than a lack of one altogether. Therefore, since an arousal-focus precludes a valence-focus (Feldman, 1995 a), one could assume that psychopaths are arousal-focused, thereby enabling them to identify and prey upon the arousing aspects of their environment. 34 Psychopathy and Arousal-Focus: Research has clearly demonstrated that although psychopaths may be emotionally deficient as compared to nonpsychopaths, they are not oblivious to this aspect of human experiences. It seems as if psychopaths simply do not have an understanding of emotional valence. If psychopaths are truly arousal-focused, than they should be able to differentiate between events varying in degrees of arousal (see Feldman, 1995a). This seemed to have been the case in Patrick's (1993) research on the psychopaths' emotional startle reflex response. Although psychopaths demonstrated deficient BSR responses to unpleasant slides, they where nonetheless no different than the nonpsychopaths in their autonomic reactions. As compared to neutral slides, SCRs were increased to both types of emotional slides, which were matched for arousal and exceeded that of the neutral slides (Patrick et al., 1993). Since research on SCR suggests that electrodermal activity is a reliable and valid (i.e., not confounded by valence) measure of emotional arousal (see Patrick, 1994 for a review), this suggests that psychopaths are able to correctly differentiate between arousal categories. Psychopaths, in this study, also showed a quadratic pattern across slide valence categories. Since psychopaths suffer from a hyper-focused attention, whereby once engaged they are unable to switch their attention irrespective of the consequences (Newman & Wallace, 1993), such findings suggest that, once perceived, psychopaths evaluate arousal, irrespective of its valence, as pleasant. This may explain why psychopaths have such difficulty evaluating emotional polarity. The psychopath, evaluating the arousing qualities of life, may not be able to pay attention to other incoming emotional information, no matter the pleasantness or unpleasantness of such information. This finding makes sense given their hyposensitive state: Like the starved individual who will devour whatever food available, the psychopaths will feed on whatever arousal present. 35 Yet, one would have expected that psychopaths, being pathologically under-aroused, would have scanned their environment for the most arousing stimulation. That is, they should have showed some attentional bias for the negative slides. As eluded to previously, HR deceleration is often interpreted as an orienting response suggesting increased attention to environmental events (Hare, 1978; 1998). In this study, the greatest cardiac deceleration occurred to unpleasant slides (Patrick et al., 1993), suggesting that psychopaths, like their nonpsychopathic counterparts, focused more so on unpleasant slides than on other ones. Taken together, these findings indicate that psychopaths are appropriately aroused by affective events, and, although evaluating all arousal as pleasant, nevertheless pay the greatest attention to the most arousing aspect of their environment. Furthermore, psychopaths also seem relatively insensitive to fearful events, thereby enabling them to seek out the most arousing events imaginable. Thus, like the indebted gambler, the psychopath, being pathologically under-aroused, will go after the biggest pay off, irrespective if it is socially acceptable or not. An arousal-focus in psychopaths may also explain their apparently normal behavioral responses to emotional stimuli. The extra informational load that affect adds to language has been found to facilitate the processing of semantic information, irrespective if the individual is psychopathic or not (Forth, 1992; Kiehl, et al., in press; Intrator et al, 1997; Mills, 1995; Pham & Rime, 1994). Psychopaths, like their nonpsychopathic counterparts, show quicker reaction times to emotional words as compared to neutral ones (see Williamson et al., 1991 for a failure to replicate). However, measures of emotional valence in these same studies found psychopaths to be emotionally impaired as compared to nonpsychopaths. As such, it may be that the psychopaths behavioral responses reflect their utilization of affective arousal, thereby helping them perform as if normal. Nonetheless, this systematic difference in their affective structure should lead psycopaths 36 to evaluate emotional events differently than nonpsychopaths (see Blascovich, 1990; 1992; Mandler, 1984). One method of testing this hypothesis would be to ask psychopaths to comment on the affective response of others in various emotion evoking situations. In one such study, psychopathic and nonpsychopathic offenders were presented with short stories about an individual engaging in behaviors that would result in feelings of happiness, sadness, embarrassment, or guilt (Blair, Sellars, Strickland, Clark, Williams, Smith, & Jones, 1995). When asked to attribute emotions to the story protagonist, psychopaths, as compared to nonpsychopathic offenders, did not differ in their evaluations of the vignettes depicting happiness, sadness, and/or embarrassment. On the other hand, the two groups of offenders were found to significantly differ in their affective attribution to the guilt inducing story. While nonpsychopaths correctly identified the feeling of guilt resulting from the protagonist's behavior (i.e., resulting in harm to another), the psychopaths' dominant attribution to such events was one of happiness or indifference, further indicating that psychopaths have problems in labeling affect, especially when negatively toned. To explain this, these authors suggested that psychopaths lack what they term a Violence Inhibition Mechanism (VIM). The VIM is proposed to be a schema activated by non-verbal communication of distress. This other-related distress cue is then believed to evoke arousal in the individual that, when one possesses a functional VIM, initiates a cognitive appraisal, such as feelings of guilt, remorse, and/or sympathy, that ultimately can result in a withdrawal response. Accordingly, Blair took the psychopath's lack of guilt attribution as a reflection of a deficient VIM. However, another explanation may be that they simply enjoyed, or "empathized", with the guilt inducing protagonist. That is, psychopaths may have simply seen the behavior as arousing and/or congruent with their own behavior, which could explain why some of them attributed feelings of happiness where one would expect guilt. Interestingly, this also demonstrates an attentional and cognitive bias towards arousal-related themes within the 37 environment, as expected in hyposensitives (see Blascovich, 1990; 1992). Accordingly, unlike the hypersensitive that looks within for emotional experiences, the hyposensitive psychopath, looking outwards, should be relatively uninfluenced by internally produced affective content (Blascovich, 1990; 1992). That is, they should not be as affected by thoughts or images depicting emotional events. Indeed, despite similar affective ratings of imagined experiences and similar heart rate, skin conductance, corrugator EMG responses to neutral sentences, psychopaths displayed smaller physiological responses during the imagery of fearful material than did nonpsychopathic sex offenders (Patrick, Cuthbert, & Lang, 1994). Thus, the cognitions of psychopaths, in the absence of environmental stimulation, do seem devoid of affective connotations. This also implies that while the nonpsychopathic individual's affective states are relatively context independent (i.e., triggered from the inside), the psychopath's emotional experiences are context dependent (i.e., triggered from the outside). As such, to have an emotional experience the psychopaths needs to seek the appropriate context - one that is high in arousal. Finally, if arousal is what truly matters to psychopaths, they should quickly dismiss non-arousing stimuli. This may help explain their relatively little use of higher cortical functions in processing linguistic emotional material. Although, psychopaths show the expected early ERP responses to emotional (pleasant and unpleasant) and non-emotional words, they do lack the later ERP differentiation seen in nonpsychopaths (Kiehl, et al., in press; Williamson, et al., 1991). Similarly, psychopaths show significantly more rCBF in the occipital lobe and significantly less rCBF in the central-parietal areas when processing linguistic information (Intrator et al., 1997). Accordingly, it seems as if the psychopaths rely more so on earlier, almost perceptual, functions to differentiate between valence-based words and nonwords, while nonpsychopathic individuals process these same stimuli to a greater extent, extracting important semantic and affective 38 information. Valence-related material, as demonstrated above, is relatively innocuous to the psychopath and not likely to parallel real life stimulation. Thus, it seems that psychopaths, having an attentional bias for the arousing, focus their attention to arousal-related information and quickly dismiss that which is of no interest. Furthermore, the fact that this occurs relatively early in the processing stream suggests that psychopaths have developed an efficient arousal searching strategy, enabling them to focus their energy on the task at hand - refueling on arousal. In summary, psychopaths, like nonpsychopaths, can evaluate the intensity of the arousal in affective events (e.g., Patrick et al., 1993), are sensitive to varying gradations in arousal (e.g., Patrick et al., 1993), seem to evaluate all arousal as pleasant irrespective of valence (e.g., Blair et al., 1995; Patrick et al., 1993), focus on the most arousing situations (e.g., Patrick et al., 1993), may depend on arousal rather than valence to perform emotionally-based paradigms (e.g., Forth, 1992; Kiehl, et a l , in press; Intrator et al, 1997; Mills, 1995; Pham & Rime, 1994), do not look inward for affective experiences (e.g., Patrick et al., 1990), and quickly dismiss non-arousing stimulation (e.g., Kiehl, et al., in press; Intrator et al., 1997; Williamson, et al., 1991), suggesting that they have developed an efficient strategy to locate stimulation. Emotionally Challenged. Not Dull: ; Although psychopaths are characterized by a unique affective structure, both research and clinical evidence suggest that they have adapted to it quite well. For example, the rating of emotional material by psychopaths on emotion-related dimensions (e.g., good-bad, pleasant-unpleasant, arousing-unarousing, etc.) has consistently been found to approximate that of nonpsychopathic individuals (e.g., Forth, 1992; Kiehl, et al., in press; Patrick et al, 1993; Williamson et al., 1991). Since such studies nonetheless find affective processing differences between such groups with more sensitive measures (e.g., physiological), it can be inferred that 39 there is a dissociation between what psychopaths say and how they truly feel. In addition, although there exists ample evidence indicating that they do suffer from some affective deficit, several studies have nonetheless failed to show affective differences between psychopaths and nonpsychopaths. For example, although non-language based emotional deficits have been noted (e.g., Christianson et al, 1996; Forth, 1992; Mills, 1995, Patrick et al, 1994; 1993), others have failed to show such differences using different methodologies (e.g., Day & Wong, 1996; Forth, 1992; Williamson et al, 1990). Their ability to mask their valence-deficient affective circumplex may explain why psychopaths are so easily able to con and manipulate others into thinking that they are "outstanding citizens", a fact highlighted by the number of prison staff that they are able to manipulate in doing them "favors". This suggests that they have adapted quite nicely to their deficit, possibly basing their affective decision on some socially derived information (see Feldman-Barrett, 1996). As Cleckley so eloquently put it, "We are dealing here not with a complete man at all but with something that can mimic the human personality perfectly...So perfect is this reproduction of a whole and normal man that no one who examined the psychopath in a clinical setting can point in scientific or objective terms why, or how, he is not real (p.229)." The Psychopath as an Arousal Addict The literature on emotions suggests that there exist normal variations in arousal sensitivity which result in systematic differences in affective structures (Blascovich, 1990; 1992; Feldman, 1995a; Mandler, 1984). These differences are further believed to lead to sensitivity-specific personality and behavioral characteristics (Blascovich, 1990; Ellis, 1987; Feldman, 1995a; Mandler, 1984; Zuckerman, 1979). While the hypersensitive, neurotic individual would tend to avoid further stimulation, the hyposensitive, sensation-seeker is more likely to seek some out. If these are indeed normal variations, then one could expect clinically significant differences in arousal sensitivity to lead to more exaggerated personality and behavioral patterns (Feldman, 1995a). As such, the psychopathic personality disorder could be characterized by a pathological hyposensitivity to arousal. Not only is there physiological evidence supporting this assertion (see Hare, 1970, 1978, 1998 for reviews), but their behaviors speak for themselves. These individuals are responsible for a disproportionate amount of crime and, more importantly, are the most violent, versatile, opportunistic, and driven of offenders (see Hare, 1996, 1998 for reviews), which definitely suggests a pathological rather than a normal variation in arousal sensitivity. Indeed, even the psychopath's own self-reports of their lives indicates an intense need for arousal: A sex offender, asked to describe how he felt when he raped for the first time replied, "Pumped up, a real high. Yeah, I know it was wrong, but thinking about it still gives me a buzz." About the death of his mother after a long illness his most salient comment was, "The funeral was a real drag. I went to sleep." Further examination clearly indicated that this man described and evaluated events and experiences solely in terms of their ability to arouse or stimulate him. "If something gives me a rush, gets the adrenaline flowing, that's good. If it doesn't, that's bad. End of story." The offender was a psychopath! (personal communication, Hare, 1997) The pathological nature of this aspect of psychopathy is further highlighted by the fact that such individuals engage in these highly arousing activities, irrespective of the consequences (i.e., social disapproval and punishment) (see Hemphill et al., 1998; Newman & Wallace, 1993 for reviews), suggesting an addiction to stimulation rather than a simple indulgence. Like the addict, the psychopath is willing to do whatever it takes to get aroused. Accordingly, being hyper-hyposensitives and having an extremely rigid attentional system, psychopaths should demonstrate an affective structure characterized by an arousal-focus that is significantly more pronounced than that seen in normal, non-criminal individuals. Unlike normal, arousal-focused individuals whose emotional circumplex nevertheless reflects some use 41 of valence-related information (Feldman, 1995a), the emotional circumplex of the arousal-focused psychopath should reflect little or no use of such information. This strong attentional bias, paralleling that seen in addicts (see Childress, et al., 1993 for a review), would presumably enable them to quickly and efficiently identify stimulating situations; situations that they can subsequently take advantage off to refuel their normally under-aroused state. Unfortunately, the unidimensional affective formulation taken by past investigations in this area provides only limited clues into this aspect of this disorder. Nonetheless, although never directly tested, there is mounting evidence pointing to an arousal-focus in psychopaths. The present study is an initial step in the systematic investigation of an arousal addiction in psychopaths. As such, the main purpose of this investigation was to examine whether psychopaths are indeed arousal-focused. To address this, the affective structure of psychopathic and nonpsychopathic offenders was assessed by comparing their similarity judgments, using a category sort procedure, with their self-reported pleasantness and arousal ratings to a group of emotional words and pictures of facial affect. Consistent with previous findings in non-criminal populations, it was predicted that the similarity ratings would result in variations in the emotional circumplex, with that of psychopaths yielding an arousal-focus. Furthermore, while a two factor solution should account for most of the variance in the similarity ratings of nonpsychopaths, with valence accounting for more of the variance than arousal, a one factor solution should emerge out of the psychopaths' similarity ratings, with arousal accounting for most, if not all, of the variance. That is, psychopaths should show an attentional bias that is pathologically arousal-focused. In addition, if psychopaths are truly addicted to arousal, then they should perceive such arousal irrespective of the stimuli employed. That is, unlike the specificity of the attentional bias seen in drug addicts (Childress, Hole, Ehrman, Robbins, McLellan, & O'Brien, 1993), the 42 psychopath, being addicted to the far more reaching concept of arousal, should consistently demonstrate an arousal-focus, irrespective of the event being evaluated. Accordingly, psychopaths are expected to show an arousal-focus when processing both emotional words and facial expressions. Finally, given evidence that psychopaths perform relatively normally on objective tasks, a third hypothesis was that psychopaths and nonpsychopaths should not differ in the more objective self-report part of this study. That is, while the projective nature of the similarity task should bring out the psychopath's attentional bias, the more objective self-report task, which can be solely performed according to denotative, socially desirable information, should facilitate the emergence of the psychopath's mask of sanity. Method Participants: Participants were 41 adult male inmates from the Regional Health Center (RHC), a federal maximum security forensic psychiatric facility in Abbottsford, BC. They were volunteers in a violent offender or sex-offender treatment program who had expressed a willingness to participate in research projects during their detention. Each inmate volunteered for the project, giving his informed consent to access his institutional files, to conduct an interview, and to be tested. Participants were required to have normal or corrected-to-normal vision, have at least elementary reading ability, have an average or above average IQ, be free from any psychosis, and have English as a first language. Participants were paid $10 for their participation. Relying on both a semi-structured interview and file information, each participant was assessed for psychopathy via the PCL-R (Hare, 1991). The PCL-R is a highly reliable instrument that has been extensively validated and used to operationalize the construct in forensic 43 populations (see Hare, 1998 for a review). It consists of 20 items that measure the personality traits (interpersonal/affective) and behavioral (antisocial lifestyle) characteristics of the psychopath. Individual items, scored on a 3-point scale according to how representative they are of the inmate, are summed to yield a total score. The total score can range from 0 to 40 and represents the degree to which an individual resembles the prototypical psychopath. The interrater reliability and alpha coefficients for total scale scores are usually above .80 (Hare, 1991). A taped interview was completed by a trained and experienced graduate student. To assess interrater reliability, two trained graduate students independently completed the PCL-R. One participant had to be excluded because a valid PCL-R could not be conducted due to insufficient file information (see Hare, 1991). The Spearman-Brown intraclass correlation coefficient of reliability was computed for a single rating and for the average of two independent ratings. These were .95 and .97, respectively. The mean PCL-R score for the total sample was 28.18 (SD = 6.33), a value higher than that normally obtained with male inmates (M « 24; S_D « 8; Hare, 1991). Inmates with a score of 30 or greater on the PCL-R (n = 21, M = 32.80, £D = 2.49) and those with scores less than 30 (n = 19, M = 23.06, SD = 5.24) were defined as psychopaths and nonpsychopaths, respectively. This cutoff was selected based on research criteria and to reflect the clinical use of the instrument (Hare, 1991). The kappa coefficient of interrater reliability for these classifications was .90. The groups did not differ in age (Mtotai = 34.43, S_Dtotai = 9.75) or level of education, with all participants having at least completed grade eight. Although IQ and reading ability were not directly measured, the screening process for acceptance into RHC required volunteers to be of average to above average intelligence and to be able to read and write at the secondary school level (minimum 8 years of education). Moreover, research has consistently failed to find IQ differences between PCL-R defined psychopaths and nonpsychopaths (Hare, Frazel, Bus, & 44 Jutai, 1980; Gretton, 1998; Harpur, Hare, & Hakstian, 1989; Newman & Kosson, 1986). Stimuli: Emotional terms (Russell, 1980) and pictures of facial affect (Russell & Bullock, 1985), extensively validated by Russell and his colleagues, were employed in this study (see Appendix). These 28 words and 9 pictures were selected to be appropriate for use in young-adult populations and to be representative of the entire affective space, irrespective if one adopts a categorical or dimensional view of emotions (Ekman & Friesen, 1987; Russell, 1980; 1985). Al l pictures were of a white adult woman and were presented individually. For the sorting task, words were individually printed in bold black capital letters (Font 28) on white cards (width = 14 cm, height = 9 cm), which were then laminated. Similarly, the black and white photographs were scanned with a Hewlett Packard ScanJet 4c (settings: 300 dpi, brightness = 137, contrast = 146) and professionally printed on white card paper (width =14 cm, height = 9 cm), and then laminated (border = 0.5 cm). A pilot study, employing a very liberal significance criterion (family wise error rate; FWER = 0.9), comparing the scanned images to their original photographic format in twenty students at the University of British Columbia found no significant differences in their pleasantness and arousal ratings (p > .05). Two randomly generated orders (A and B) were created for the words and faces, respectively, for use in the rating task. The words and faces in this task were professionally printed, using the same settings as above, at the top of a page containing rating scales printed in a random order. Procedure: Participants were approached in order to get consent to review their institutional file and to conduct an interview in order to complete the PCL-R. Thereafter, two trained graduate 45 students assessed participants on the PCL-R and assigned them their stimuli order of presentation (words or faces) and stimuli list (A or B), which were then provided to the experimenter. This resulted in a counterbalanced presentation of the material across participants and ensured that the experimenter remained blind to psychopathy. The experimenter then contacted the participant to set up a time for the experimental session. The experimental material was given to all participants in the following order: a demographic questionnaire, two mood measures, a category sort task, a distracter task, a free-recall memory test of the experimental words, a rating task, a measure of desirable responding, and a mood questionnaire. Thereafter, participants were asked to complete several other questionnaires given as part of a larger research project. The category sort and rating tasks were not counterbalanced in order to avoid contaminating participants' emotional processing by providing the dimensions of interests. Participants were seated at a desk throughout the study. They had the procedures explained to them and then read and signed the experimental consent form. Participants were informed, both orally and in writing, that they may withdraw from participation at any time. Afterwards, they completed a demographic questionnaire, required to rule out possible confounds; the brief measure of Positive and Negative Affect (PANAS; Watson et al., 1988); and the State Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, Lushene, 1970). The PANAS [referred to as PANAS(l)] was used to measure participants' current affective state in terms of both positive affect (PA) and negative affect (NA). PA reflects the extent to which a person feels enthusiastic, active, and alert, while NA is a general measure of subjective distress and unpleasurable engagement (Watson et al., 1988). The STAI was used to measure both the state and trait anxiety of offenders (Spielberger, et al., 1970). Together, these mood measures were required to rule out possible confounds related to how participants approached the experimental 46 tasks (i.e., to evaluate initial mood-related cognitive sets). If participants had to be excluded at this point (refer to above participant selection criteria), the reason(s) were explained to them. Participants were then given a chance to familiarize themselves with the stimuli before proceeding with the tasks. They were given the pile of cards (words then faces), presented in the same order for all participants, and given a chance to ask questions about stimuli that were not understood. They were only allowed to go through the set once and were restricted, at this time, to asking stimuli-related questions. Definitions were provided for misunderstood words. If the participant needed more information, synonyms (not used in the study) were provided sequentially until an understanding of the term was acquired. Category Sort fWard. 1977V The category sort is a projective-like technique, as participants must rely on their own understanding of the stimuli in order to complete this task. This task, which yields similarity appraisals between all word- and face-pairs, was chosen to control for the possibility that participants, especially the psychopaths, may try to perform according to how they feel others would respond (i.e., give socially desirable responses), rather than according to their own understanding of the material. The participant's task was to group emotionally similar stimuli into a specified number of piles (two, four, and six piles and four, seven, ten, and thirteen piles for the faces and the words, receptively), which were chosen as to render a range of normal distributed similarity scores. The order of presentation of the stimuli (faces or words) was counterbalanced across participants. Participants were provided with standard, written instructions. To obtain an unbiased representation of the dimensions inmates use to categorize emotional stimuli, participants were not instructed how to sort the stimuli (e.g., emotional/non-emotional, pleasant/unpleasant, etc.) but rather told to rely on their subjective understanding of the material and their relationship to one another. In addition, to ensure that they employed psychologically-relevant dimensions in their similarity judgments, participants were instructed not to group stimuli according to their physical characteristics (e.g., word length, smile, etc.). To familiarize the participants with this procedure, they were asked to group 10 cards containing words not used in the study (see Appendix) and asked to sort these into 3 piles. This was repeated until the participant showed a good grasp of the task at hand. Participants were instructed that questions concerning this task could only be answered at this time. Thereafter, they were asked to divide the experimental stimuli into a specified number of piles, with the order of trials randomized across participants. Prior to each trial, the cards were shuffled (i.e., each participant received the cards in a random order) so that the participant could not hierarchically subdivide previous divisions of the stimuli. It was assumed that nonpsychopaths, like non-criminals, would be affected by the emotionality of the material, both in terms of its valence and arousal, and thus take this into account during their similarity appraisal, while psychopaths, being arousal-focused, would group these stimuli according to their arousal-relatedness. This section took approximately 40 minutes and involved no deception or coercion. Distracter Task: Since stimuli order in the category sort task was counterbalanced, some participants concluded the task with the faces while others with the words. As such, a distracter task was required to ensure that all participants had some time elapse between the word sorting and the memory task. Participants were required to complete the Sensation Seeking Scale-Form V (SSS; Zuckerman, 1979), a measure collected as part of a larger research project. The SSS took approximately ten minutes to complete. 48 Free-recall memory task: Following the administration of the SSS, participants were asked to write down as many of the experimental words as they could remember. No time limit was imposed for this task, which was used to compare groups on attention and motivations during the category sort. Rating task (Mehrabian & Russell. 1974): Participants were required to independently rate the pleasantness and arousability of the previous stimuli on Mehrabian and Russell's (1974) six-item scales of pleasure-displeasure (i.e., happy-unhappy, pleased-annoyed, satisfied-unsatisfied, contented-melancholic, hopeful-despairing, relaxed-bored) and degree of arousal (i.e., stimulated-relaxed, excited-calm, frenzied-sluggish, jittery-dull, wide-awake-sleepy, aroused-unaroused). These scales employed a semantic differential format with a numerical scale ranging from -3 to +3 for each dimension. Standard, written instructions on how to use semantic differential scales were provided to the participant. Groups of stimuli (words/faces) were rated separately and each stimulus within a group was rated on all scales prior to moving to the next stimulus. The scale order was randomized across stimuli and three scales within each factor were reversed (pleased-annoyed, contented-melancholic, relaxed-bored, frenzied-sluggish, jittery-dull, and aroused-unaroused). To avoid other order and/or sequence related confounds, the word and face ratings, as well as the stimuli lists (A and B), were counterbalanced across participants. Participants were instructed that questions concerning this task would only be answered at this time. Given research findings that psychopaths and nonpsychopaths do not differ from one another on such tasks (e.g., Forth, 1992; Patrick et. al., 1993), no group differences were expected. This section took approximately 40 minutes and involved no deception or coercement. Once participants had finished the ratings, they were asked to complete the Balanced 49 Inventory of Desirable Responding (BIDR; Paulhus, 1994). Research with offenders has found that the BIDR results in three independent indices reflecting different response styles (Kroner & Weekes, 1996): Impression management (IM) is characterized by situational variations in responding that are guided by a desire to create a favorable impression on others; denial of the negative (DN) reflects a tendency to repudiate one's negative attitudes; and over-confident rigidity (OCR) reflects a self-perception of infallibility and inflexibility. As such, this questionnaire was required to rule out possible confounds in the rating task. Thereafter, participants were asked once more to complete the PANAS [referred to as PANAS(2)]. However, unlike previously, this administration was used to assess how they felt during the experimental procedure, evaluating any group differences on affect, attention, and motivation related variables. Participants were then required to complete several questionnaires as part of a larger research project. Note that two strategically placed typos were included in these as to assess participants' task-oriented attention. The first (typo 1) consisted of a wrong letter (i.e., Q instead of O) representing a response choice (i.e., occasionally) and the second (typo 2) was a nonsense sentence (i.e., "I lose my temper very quote). As such, although the results of these questionnaires are not included in the current project, the attention-related variable will be assessed. Finally, each participant was then rated for his perceived level of attention, motivation, and cooperation throughout the study by the experimenter using a 7-point Likert scale. Following the experiment, participants were debriefed and given a chance to ask questions. Dependent Variables: Similarity Score: Rendered from the category sort, this variable reflects the relative relationship between pairs of stimuli (pairs of words or pairs of faces). It was dependent on (a) the number of trials in 50 which a participant placed two stimuli together in the same pile, such that the more often stimuli were paired, the higher the similarity score between them; and (b) the number of categories required, in that the groupings became more homogeneous as the number of categories increased (see Ward, 1977). Stimuli placed within the same category were therefore given a similarity score that reflected the number of categories used in that trial. For example, two faces placed in the same pile during the "4-pile trial" were given a similarity score of "4". Similarly, two words placed in the same pile during the "13-pile trial" were given a similarity score of "13". This resulted in a "trial similarity score" for each pair of stimuli. To get the "total similarity score" for each pairing, "trial similarity scores" were added. A score of 1 was also added to each pair (minimum similarity score) as one would have to place them in the same category in a degenerate "sort" with only one category. Accordingly, if two words were placed in the same pile in all trials, then they were given a similarity score of 35 (1+4+7+10+13). Similarly, if two faces were placed in the same pile in the 2 and 6 category trials, then they were given a score of 9 (1+2+6). The range of scores was from 1-13 and 1-35 for the faces and the words, respectively. This procedure resulted in a 3 dimensional matrix of similarities (participant x stimuli x stimuli). To avoid scoring errors, similarity scores were independently calculated by two research assistants and then compared. Thereafter, the similarity data from the category sort task were re-coded into dissimilarity data (i.e., by reversing values) and then subjected to the multidimensional scaling (MDS) analyses. Global Pleasure and Arousal Score: Participants' responses to each stimulus were averaged across the six pleasure-related dimensions and the six arousal-related dimensions - yielding a global pleasure and a global arousal score for each stimulus per participant (see Mehrabian & Russell, 1974). This method 51 was chosen to yield scores that were more sensitive than the single scale scores used by others (e.g., Forth, 1992; Patrick et al., 1993) Results The analyses of the data proceeded in a step-wise fashion. Firstly, possible confounds that may have biased the results were investigated. Secondly, psychopaths and nonpsychopaths were compared on their global pleasure and arousal ratings of the words and faces, respectively, as these ratings were required for dimensional interpretations. Thirdly, the underlying psychological dimensions employed by participants to complete the word and face category sort, respectively, were extracted and interpreted. Fourthly, based on the previous analysis, the affective structure of psychopaths and nonpsychopaths were investigated and compared. Finally, additional analyses were conducted to clarify previous findings, which were completed in the same step-wise fashion. Confounding Variables: Responses to each of the measures aimed at identifying possible confounds were averaged across groups and compared via 15 separate independent t-tests (see Table 1). A liberal significance criterion (FWER = .75) was employed to ensure that groups did not differ on these variables. Levene's homogeneity of variance test was performed for each analysis and found to be nonsignificant (p > .05). None of the t-tests were found to be significant (p > .05). Accordingly, it was assumed that subsequent between-group analyses were not confounded by such variables as initial affective state, attention, motivation, cooperation and/or response style. Al l participants were found to be appropriately motivated for the task (M = 5.98, S_D = .42), cooperative throughout the experiment (M - 6.48, S_D = .55), and seemed to pay attention and 52 Table 1 Independent t-tests comparing nonpsychopaths and psychopaths on possible confounds Confounds df t-value p PANAS (1): PA 38 -0.90 .38 PANAS (1): NA 38 1.83 .08 STAI: State 38 1.54 .13 STAI: Trait 38 0.42 .68 Memory Test 38 1.82 .08 BIDR: DN 38 0.22 .83 BIDR: IM 38 0.70 .49 BIDR: OCR 38 -0.11 .91 PANAS (2): PA 38 -0.26 .80 PANAS (2): NA 38 0.88 .38 Attention 38 0.85 .40 Cooperation 38 1.13 .26 Motivation 38 1.11 .28 Typo 1 38 1.27 .21 Typo 2 38 0.95 .35 Note: FWER = .75, * p < .05 concentrate on the task at hand (M = 6.15, SD = .43). Rating Analyses: Emotional Words: Valence and arousal means for each word were calculated for the entire sample. Words were then ranked from most to least pleasant and from most to least arousing, separately. The word valence rankings for the entire sample were then used to divide the 28 words into pleasant (n = 9), neutral (n = 10), and unpleasant (n = 9) categories. Similarly, the word arousal rankings were used to divide the 28 words into high (n = 9), medium (n = 10), and low (n = 9) arousing categories. Group means for the valence and arousal ratings for each word were then calculated. Table 2 presents the words, ranked from most pleasant to least pleasant, for psychopaths and nonpsychopaths. Table 3 presents the words, ranked from most arousing to least arousing, for the same two samples. The valence and arousal rankings, respectively, for each group were then correlated using Spearman's rank correlation coefficient, revealing that psychopaths and nonpsychopaths did not differ in their relative valence (rs = .99, p < 0001) and arousal (rs = .97, p < .0001) rankings of the words. Although having equal rankings, psychopaths and nonpsychopaths could still differ in the absolute degree to which they find these words pleasant or arousing. Moreover, if psychopaths and nonpsychopaths have different patterns of valence and arousal sensitivity, then these differences should be mediated by the affective characteristics of the words. Accordingly, the emotional word ratings of the groups where compared across valence and arousal categories. To investigate whether psychopaths and nonpsychopaths differed in their reported sensitivity to the word valence dimension, the pleasure and arousal ratings for words within each valence category were averaged and then analyzed via two 2 (group: psychopaths vs. 54 Table 2 Pleasure rating rankings for emotional words from most pleasant to least pleasant averaged across subjects from each group Words1 P NP Happy 1 1 Pleased 2 3 Satisfied 5 2 Delighted 3 4 Serene 4 5 Glad 6 7 Content 7 6 At ease 9 8 Relaxed 8 9 Calm 10 10 Excited 11 11 Aroused 13 12 Sleepy 14 13 Astonished 12 14 Tired 15 15 Alarmed 17 16 Tense 18 14 Droopy 16 19 Afraid 19 18 Annoyed 20 20 Distressed 22 21 Frustrated 21 22.5 Bored 23 22.5 Sad 24 24 Angry 25 26 Gloomy 26 25 Miserable 27 27 Depressed 28 28 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample Table 3 Arousal rating rankings for emotional words from most arousing to least arousing averaged across subjects from each group Words1 P NP Excited 1 1 Alarmed 2 2 Angry 4 4 Aroused 3 6 Afraid 5 3 Astonished 6.5 5 Tense 8 7 Delighted 9 8 Frustrated 6.5 9 Happy 10 10 Annoyed 13 11 Distressed 12 13 Glad 11 14 Pleased 14 12 Satisfied 15 15 Miserable 16 16 At ease 17 22 Serene 19 17 Content 18 ' 18 Sad 20 19 Calm 22 21 Depressed 23 20 Bored 21 25 Relaxed 25 23 Gloomy 24 24 Droopy 26 26 Sleepy 27 28 Tired 28 27 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample psychopaths) x 3 (word type: pleasant vs. neutral vs. unpleasant) analyses of variance (ANOVA), with group as a between-subject factor and slide type as a within-subject repeated measure factor. In terms of the pleasure ratings, the data were found to violate the assumptions of homogeneity of covariance [Boxs M = 48.51: %2 (6) = 44.34, p < .0001] and sphericity [Mauchley's test, W = .35: %2 (2, N. = 40 ) = 38.32, g < .0001]. Since this indicated a liberal condition (see Glass & Hopkins, 1996), the group effect, which was found to be non-significant [F(l, 38) = 3.71, p > .05], was deemed accurate, suggesting that psychopaths (M = -39, SD = 1.42) and nonpsychopaths (M = -09, SD =1.29) did not differ in their overall pleasure ratings of the valence-based words. In light of the above homogeneity of covariance assumption violation, the within subject factor (i.e., valence category) was analyzed for psychopaths and nonpsychopaths via two separate repeated measures ANOVAs (see Glass & Hopkins, 1996). In each case, sphericity was tested with the use of Mauchley's test. For both nonpsychopaths and psychopaths, the assumption of sphericity was violated [W_ = .24: %2 (2, N. = 40) = 24.46, p < .0001; W = .36: %2 (2, N = 40) = 19.59, p < .0001, respectively], suggesting the use of conservative tests [i.e., epsilon (s) corrected degree of freedoms] in interpreting the results. Valence-based words were found to differ in terms of their pleasure ratings for both nonpsychopaths [e = .5: F(l , 18) = 194.33, p < .0001] and psychopaths [s = .5: F(l , 18) = 83.96, p < .0001]. Follow-up analyses were conducted via paired sample t-tests (see Table 4: Panel A), which were Bonferroni-corrected to control for type 1 errors (FWER3 contrasts = -05). Figure 3 (upper panel) depicts the mean pleasure ratings for valence-based words for both psychopaths and nonpsychopaths. For both groups, a significant linear relationship emerged, such that pleasant words were rated as significantly greater in pleasure than neutral words, which were themselves rated as significantly more pleasurable than unpleasant words. Finally, although univariate homogeneity of variance tests between groups for the unpleasant and neutral word 57 Table 4 Paired Sample t-tests comparing the pleasure ratings and arousal ratings of emotional word types for each group Comparison df t-value E A. Pleasure Ratings of Valence NP: Pleasant-Neutral 18 -11.25 .0001 *** NP: Pleasant-Unpleasant 18 14.58 .0001 *** NP: Neutral-Unpleasant 18 16.13 .0001 *** P: Pleasant-Neutral 20 -8.67 .0001 *** P: Pleasant-Unpleasant 20 9.67 .0001 *** P: Neutral-Unpleasant 20 8.16 .0001 *** B. Arousal Ratings of Valence NP: Pleasant-Neutral 18 4.32 .0001 *** NP: Pleasant-Unpleasant 18 0.32 .7500 NP: Neutral-Unpleasant 18 6.38 .0001 *** P: Pleasant-Neutral 20 3.55 .0100* P: Pleasant-Unpleasant 20 -0.02 .9900 P: Neutral-Unpleasant 20 3.93 .0010 ** C. Pleasure Ratings of Arousal NP: High-Medium 18 -11.35 .0001 *** NP: High-Low 18 4.27 .0001 *** NP: Medium-Low 18 -10.69 .0001 *** P: High-Medium 20 -6.97 oooi *** P: High-Low 20 3.10 .0100* P: Medium-Low 20 -9.68 .0001 *** D. Arousal Ratings of Arousal NP: High-Medium 18 9.69 .0001 *** NP: High-Low 18 11.33 .0001 *** NP: Medium-Low 18 -10.68 .0001 *** P: High-Medium 20 8.03 .0001 *** P: High-Low 20 9.74 .0001 *** P: Medium-Low 20 -9.03 .0001 *** Note: P = psychopaths; NP = nonpsychopaths (NP) * p < .01; ** p < .001; *** p < .0001 • Psychopaths • Nonpsychopaths Unpleasant Neutral Pleasant Unpleasant Neutral Word Valence Category Pleasant Figure 3: Pleasure (upper panel) and arousal (lower panel) ratings of the valence-based words for both diagnostic groups categories were found to be significant [Bartlett-Box: F(l , 4300) = 13.25, p < .0001; F(l , 4300) = 16.39, p < .0001, respectively], the interaction effect was found to be nonsignificant through both the apparent [e = 1: F(2, 76) = .55, p > .50] and conservative [s = .5: F(l , 38) = .55, p > .25] tests, thereby attesting to its nonsignificance. The homogeneity of covariance matrices [Boxs M = 16.00: %2 (6, N. = 40) = 14.62, p < .05] and sphericity [W_ = .80: %2 (2, M = 40) = 8.25, p < .05] assumptions were also found to be violated for the analysis of the arousal ratings. The group effect was not found to be significant [F(l, 38) = .03, p > .50], even under this liberal condition, indicating that psychopaths did not differ from nonpsychopaths in their overall arousal ratings of valence-based emotional words (M = .12, S_D = .44; M = -13, SJJ = .29, respectively). Two separate repeated measures ANOVAs were then conducted, given the above assumption violation, to investigate the effect of valence on the arousal ratings of both psychopaths and nonpsychopaths. While sphericity was not violated for nonpsychopaths [W. = .87: Y 2 (2, N. = 40) = 2.33, p > .25], the same can not be said about psychopaths [W = .63: x 2 (2, N = 40) = 8.77, p < .05]. For both nonpsychopaths and psychopaths, a significant within-subject effect emerged [e = 1: F(2, 36) = 16.97, p < .0001; s = .5: F(l , 18) = 6.47, p < .05, respectively]. Subsequent analyses (FWER = .05) revealed a significant quadratic effect for both groups (see Table 4: Panel B). Neutral words were found to be rated significantly more arousing than both pleasant and unpleasant words, which were themselves not significantly different from each other [see Figure 3 (lower panel)]. As above, although the univariate homogeneity of variance assumption was violated for both unpleasant and neutral categories [F(l, 4300) = 7.03, p < .01; F(l , 4300) = 5.74, p < .05, respectively], the interaction effect was not significant, irrespective if tested via the apparent [e = 1: F(2, 76) = .45, p > .50] or conservative [s = .5: F(l , 38) = .45, p > .50] tests. To investigate whether psychopaths and nonpsychopaths differed in their reported 60 sensitivity to the word arousal dimension, the valence and arousal ratings for words within each valence category for each group were averaged and then analyzed via two 2 (group: psychopaths vs. nonpsychopaths) x 3 (word type: high vs. medium vs. low arousal) repeated measures ANOVAs. The pleasure and arousal rating data were found to violate the homogeneity of covariance assumption [Boxs M = 43.41: %2 (6, N = 40) = 39.69, p < .0001; Boxs M = 19.26: %2 (6, N = 40) = 17.61,p<.01, respectively]. In addition, the arousal ratings were also found to violate the assumption of sphericity [W = .32: %2 (2, N. = 40) = 42.44, p < .0001]. As above, differences between psychopaths and nonpsychopaths were not observed in either pleasure (M = .39, SD = .84; M = .06, SD = .56, respectively) or arousal (M = .12, SD = 1.10; M = .14. SD = 1.08, respectively) ratings of arousal-based emotional words under this liberal condition [F(l, 38) = 3.78, p > .05; F(l , 38) = .05, p > .50, respectively]. Given the above homogeneity of covariance violations, the within-subjects factors were tested for sphericity and analyzed via separate repeated measures ANOVAs. For the pleasure ratings, sphericity was violated for nonpsychopaths [W. = .57: %2 (2, N. = 40) = 9.44, p < .01] but not for psychopaths [W = 1.00: % (2, N = 40) = .05, p > .50]. Nonetheless, the within-subject effect was found to be significant for both groups [s = .5: F(l , 18) = 101.95, p < .0001; s = 1: F(2, 38) = 51.78, p < .0001, for nonpsychopaths and psychopaths, respectively], and followed-up with Bonferroni-corrected paired sampled t-tests (see Table 4: Panel C). A significant linear effect was observed for both groups, with medium arousing words being rated significantly more pleasing than high arousing words, which were themselves significantly more pleasing than low arousing words [see Figure 4 (upper panel)]. Similarly, the arousal ratings for nonpsychopaths and psychopaths were found to violate the assumption of sphericity [W_ = -28: %2 (2, N = 40) = 21.82, p < .0001; W = .34: %2 (2, N = 40) = 20.44, p < .0001, respectively], and thus tested via conservative tests (s = .5). Once more, the word type (i.e., arousal) factor was found to be 61 • Psychopaths • Nonpsychopaths 2 -, 1.5 -1 --1 -1.5 --2 . . Low Medium High Low Medium Word Arousal Category High Figure 4: Pleasure (upper panel) and arousal (lower panel) ratings of the arousal-based words for both diagnostic groups 62 significant for both nonpsychopaths and psychopaths [F(l, 38) = 117.89, p < .0001; F(l , 38) = 85.85, p < .0001, respectively]. Further analyses revealed a significant linear trend (see Table 4: Panel D). High arousing words were rated as significantly more arousing than medium arousing words, and both were rated significantly more arousing than low arousing ones [Figure 4 (lower panel)]. Finally, the univariate homogeneity of variance assumption was violated for the pleasure ratings across all three arousal categories [F(l, 4300) = 21.85, p < .0001; F(l , 4300) = 11.74, p < .001; and F(l , 4300) = 15.34, p < .0001 for high, medium, and low arousing words, respectively] and for the arousal rating of the moderately arousing words [F(l, 4300) = 4.71, p < .05]. Nonetheless, the group x valence interaction effects for both the pleasure and arousal ratings were not found to be significant when tested by either the apparent [s = 1: F(2, 76) = .81, p > .25; s = 1: F(2, 76) = .26, p > .50, respectively] or conservative [s = .5: F(l , 38) = .81, p > .25; s = .50: F(l , 38) = .26, p > .50, respectively] tests. To establish whether or not these two dimensions were independent of one another (see Russell, 1980), a Pearson correlation coefficient between the total sample mean valence and arousal ratings was computed. This correlation was found to be very low (r = .03, p >.75), indicating that these two dimensions are indeed orthogonal to one another. Furthermore, similar results were obtained with the ratings of psychopaths (r = .04, p > .75) and nonpsychopaths (r = .01, p > .75), further attesting to the independence of these two dimensions. Pictures of Facial Affect: The same procedures as above were employed with the pictures of facial affect. Table 5 presents the face valence rankings for the psychopaths and nonpsychopaths, while Table 6 depicts the face arousal rankings for these same groups. As above, the total sample rankings were used to divide the 9 faces into valence and arousal categories, resulting in 3 pleasant, 3 63 Table 5 Pleasure rating rankings for pictures of facial affect from most pleasant to least pleasant averaged across subjects from each group Faces1 P NP Excitement 1 1 Contentment 3 2 Pleasure 2 3 Sleepiness 4 4 Neutral 5 5 Arousal 6 6 Distress 7 7 Misery 8 9 Depression 9 8 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample Table 6 Arousal rating rankings for pictures of facial affect from most arousing to least arousing averaged across subjects from each group Faces1 P NP Excitement 1 1 Misery 2 2 Distress 3 3 Arousal 4 4 Pleasure 5 5 Contentment 6 6 Depression 7 7 Neutral 8 8 Sleepiness 9 9 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample neutral, and 3 unpleasant faces and 3 high, 3 medium, and 3 low arousing faces, respectively. Between-group correlations of the valence and arousal ratings, separately, revealed that psychopaths and nonpsychopaths did not differ in their valence (rs = 1.00, p < 0001) and arousal (rs = .97, p < 0001) rankings of these stimuli. To investigate possible group differences on facial valence sensitivity two 2 (group) x 3 (valence category ) ANOVAs were conducted, with valence and arousal ratings as separate dependent variables, group as the between-subject factor and category as the within-subjects repeated measure factor. For the pleasure rating, neither the homogeneity of covariance nor the univariate homogeneity of variance assumptions were violated. Group differences were not found to.be significant [F(l, 38) = .01, p > .75], indicating that psychopaths and nonpsychopaths did not vary in their pleasure ratings of valence-based pictures of facial affect (M = -14, SD = 1.34; M = -13, S_D = 1.31, respectively). Conservative tests were used to interpret the within factor and interaction effects as the sphericity assumption was violated [W_ = .71: %2 (2, N = 40) = 12.84, p < .01]. A significant effect was observed for the within-factor (valence category) effect [s = .5: F(l , 38) = 214.93, p < .0001], while the interaction effect was found to be nonsignificant [e = .5: F(l , 38) = .60, p > .25]. The within-subject effect was followed-up with Bonferroni-corrected paired sample t-tests. A significant linear trend was observed (see Table 7: Panel A), such that pleasant faces were rated as significantly more pleasing than neutral faces, and both were significantly more pleasing than unpleasant ones [see Figure 5 (upper panel)]. No assumption violations were found with the arousal ratings of the valence categories. The group and interaction effects were not found to be significant [F(l, 38) = .31, p > .50; F(2, 76) = .31, p > .50, respectively]. Thus, both groups rated the arousal of valence-based words equally (M = .47, SD = .87; M = .41, SD = .78 for psychopaths and nonpsychopaths, respectively). As above, a significant valence category effect was observed [F(2, 76) = 82.95, p < 66 Table 7 Paired Sample t-tests comparing the pleasure ratings and arousal ratings of pictures of facial affect types Comparison df t-value p A. Pleasure Ratings of Valence Pleasant-Neutral 39 -13.02 .0001 *** Pleasant-Unpleasant 39 16.92 .0001 *** Neutral-Unpleasant 39 11.14 .0001 *** B. Arousal Ratings of Valence Pleasant-Neutral 39 -12.58 .0001 *** Pleasant-Unpleasant 39 1.44 .1600 Neutral-Unpleasant 39 -9.22 .0001 *** C. Pleasure Ratings of Arousal High-Medium 39 -10.60 .0001 *** High-Low 39 1.88 .0700 Medium-Low 39 -13.47 .0001 *** D. Arousal Ratings of Arousal High-Medium 39 7.35 .0001 *** High-Low 39 13.76 .0001 *** Medium-Low 39 -12.14 .0001 *** Note: P = psychopaths; NP = nonpsychopaths (NP) * p < .01; ** p < .001; *** p < .0001 1.5 1 O) •E 0 . 5 - I (0 O 0 3 ra -0 .5 Q. -1 -1.5 -2 Unpleasant Neutral Pleasant 2 1.5 1 (0 U) .E 0.5 ra ? 0 ra g -0 .5 < -, -1.5 Unpleasant Neutral Face Valence Category Pleasant Figure 5: Pleasure (upper panel) and arousal (lower panel) ratings of valence-based faces 68 .0001]. Follow-up analyses revealed a quadratic effect (see Table 7: Panel B), where both pleasant and unpleasant faces, which were themselves not significantly different in rated arousal, were significantly more arousing than neutral faces [see Figure 5 (lower panel)]. Two 2 (group) x 3 (arousal category) ANOVAs were conducted to investigate group differences in facial arousal sensitivity with valence and arousal ratings as the dependent variables, respectively. The pleasure ratings were not found to violate any of the assumptions. Group and interaction effects were found to be nonsignificant [F(l, 38) = .01, p > .75; F(2, 76) = 1.21, p > .25, respectively], negating pleasure rating differences of arousal-based emotional faces between psychopaths and nonpsychopaths (M = -14, S_D = .72; M = .13, S_D = .73, respectively). Once more, a significant effect was found for the arousal category factor [F(2, 76) = 101.09, p < .0001]. Follow-up analyses (see Table 7: Panel C) indicated that offenders viewed the medium arousing faces as the most pleasing, followed by the high and low arousing faces, respectively [see Figure 6 (upper panel)]. For the arousal ratings, no group differences were observed [F(l, 38) = .31, p > .50], indicating that psychopaths did not differ from nonpsychopaths in their arousal ratings to faces classified according to their level of arousal (M = -47, £D = 1.10; M = .41, S_D = 1.00, respectively). Since arousal ratings were found to violate the assumptions of sphericity [W = .62: X 2 (2, N = 40) = 17.69, p < .0001], conservative tests were used to evaluate the within-subject and interaction effects. Arousal categories were found to be significantly different [e = .5: F(l , 38) = 151.19, p < .0001]. A significant linear effect was observed (see Table 7: Panel D) such that arousal ratings increased as the faces progressed from low to medium to high levels of arousal [see Figure 6 (lower panel). The interaction effect was not found to be significant [s = .5: F(l,38) = .54,p>.25] Finally, correlations between valence and arousal ratings for the entire sample, the 69 1.5 1 0.5 0 -0.5 -1 -1.5 Low Medium High 2 1.5 1 0.5 0 -0.5 -1 -J -1.5 Low Medium Face Arousal Category High Figure 6: Pleasure (upper panel) and arousal (lower panel) ratings of arousal-based faces psychopaths, and the nonpsychopaths were conducted to further support the independence of these two dimensions. As with the emotional words, these dimensions were found to be independent of one another for all three samples (r = 06, p > .75; r = .13, p > .50; and r = -.02, p > .75 for the total sample, psychopaths, and nonpsychopaths, respectively). Dimensional Analyses: Similarity scores were converted into dissimilarity scores for multidimensional scaling (MDS) procedures. MDS relies entirely on the individual's knowledge of emotion and emotional material and thus provides the methodology needed to uncover the psychopath's elusive affective structure. This procedure begins with simple behaviors (e.g., grouping similar objects into piles), from which properties of the cognitive processes yielding those behaviors can be deduced. Distance (in this case, based on similarity judgments) is then used to represent the relationships between emotion categories. The result is a geometric space of specified dimensionality where more similar categories are placed closer together and dissimilar categories are placed farther apart (Romney, Shepard, & Nerlove, 1972; Ward, 1977; Ward & Russell, 1981). The resulting dimensions are interpreted as the fundamental attributes characterizing the class of stimuli (Carroll & Chang, 1970; Feldman, 1995a; Takane, Young, & Leeuw, 1977; Young & Hamer, 1987). To identify the underlying cognitive-affective processes used by psychopaths and nonpsychopaths in this study, the (dis)similarity data for these two groups were analyzed separately via nonmetric classical multidimensional scaling (ALSCAL: Takane et al., 1977) procedures. ALSCAL relies on an unweighted Euclidean model to uncover the geometric space common to all participants (i.e., the emotional circumplex) (Takane, et al., 1977; Young & Hamer, 1987). An iterative least-squares solution is employed to minimize distortions of the 71 original data (Takane, et al., 1977; Young & Hamer, 1987). This serves to maximize the fit between the judgments computed from the weighted dimensions and the transformed (scalar-products) original judgments for all participants (Takane, et al., 1977), thereby maximizing the reliability of the resulting dimensional solution. In addition, unlike other scaling procedures, ALSCAL can be used with various types of data. It allows for data based on any measurement level (Takane, et al., 1977), which was essential given that the similarity data in the current study was ordinal in nature. It also allows for data that are either conditional or unconditional on some aspect of the experiment (Takane, et al., 1977; Young & Hamer, 1987). Since the similarity data in this study were conditional on the interpretative system of individual participants (i.e., one can not assume that participants sorted the stimuli in exactly the same fashion), the procedure chosen had to account for such a bias. Furthermore, the ALSCAL solution results in unrotatable and easily interpretable dimensions, thereby reducing the need for subjective judgments in the analyses (Carroll & Chang, 1970; Feldman, 1995a; Takane, et al., 1977; Young & Hamer, 1987). As such, when based on a set of a priori theoretical dimensions (in this case, valence and arousal), the recovered (unrotated) dimensions should correspond to them in a one to one fashion (Carroll & Chang, 1970), thereby further reducing the need for "guess" work. As such, ALSCAL provided the methodology needed to investigate systematic differences in the emotional circumplex of psychopaths and nonpsychopaths. ALSCAL yields two measures of fit for each dimensional solution specified by the user (Takane, et al., 1977). The first is a stress value, ranging from 1 (worst possible fit) to 0 (perfect fit), that indicates the extent of the model's departure (i.e., the distances) from the (transformed) observed data. The second measure of fit is a squared correlation (RSQ) between the data and the distances. As such, RSQ, which ranges from 0 (worst possible fit) to 1 (perfect fit), represents the proportion of variance the scaling solution accounts for in the distances (i.e., similarity) between 72 terms. To decide upon the actual number of dimensions characterizing the class of stimuli employed, a stress x dimension plot is created. When the addition of dimensions to the solution no longer significantly improves stress, the last (significant) dimensional solution is interpreted as being correct. Ultimately, the appearance of a clear elbow in the plot suggests that stress is no longer being significantly improved (Davison, 1983; Takane, et al., 1977; Young & Hamer, 1987). However, in the absence of a clear elbow, a stress value less than .10 can be used as the decision point (Davison, 1983; Kruskal & Wish, 1978). Emotional Words: Participants' word dissimilarity matrices were averaged across groups and subjected to the ALSCAL procedure, yielding MDS solutions ranging from one to six dimensions. Figure 7 represents the plot of the stress values by the number of dimensions for both diagnostic groups. As can be seen, a clear elbow emerged at the two-dimensional (2D) solution for both psychopaths (stress = .09) and nonpsychopaths (stress = .11), suggesting its suitability. In addition, although the one-dimensional (ID) solution for both psychopaths and nonpsychopaths accounted for a substantial amount of the fit between the original data and the distances (stress = .36; stress = .37, respectively), with stress being virtually the same for both groups, a one-dimensional solution was not viewed as sufficient to explain the pattern of results (i.e., stress was not less than .10). The RSQ values for the 2D solution of psychopaths and nonpsychopaths were .96 and .94, respectively. Taken together, these measures of fit indicate that a 2D solution best represents the underlying characteristics of the emotional words, irrespective of psychopathy. For both groups, the words fell in a circular arrangement around the two dimensions (see Figure 8 and 9 for nonpsychopaths and psychopaths, respectively), further suggesting that the participants weighed each dimension equally (see Feldman, 1995a). An inspection of the —•— Nonpsychopaths ---»--• Psychopaths 0.45 —i 0.4 J D1 D2 D3 D4 D5 D6 Dimensional Solutions Figure 7: Stress value by dimensional solutions of the emotional words for both diagnostic groups 1.5 IT" 1 0.5 H 1 1 1 h H 1 1 1 1 — I 1 1 1 1 1—I—0- H — I 1 1 1—I 1 1 1 — I 1 — I h 0.5 1 ^ — I — i — i — i — h -1.5 -V i . i -1 -0.5 -0.5 + -1 -1.5 Valence Figure 8: Nonpsychopath's emotional circumplex for the emotional words 1.5 1 1 J-0.5 -L 00 (0 3 o H 1 1 1 1 1 1 1 1 1 1 1 1 — 0 --1.S -1 -0.5 H 1 1 1 1 1 1 1 1 1 1 1 1 h j — | 1 0 0.5 1 M.5 -0.5 1 -1 i -t:5~ _=2_ Valence Figure 9: Psychopath's emotional circumplex for the emotional words 76 arrangement of the words along each dimension suggested that the horizontal dimension is one of valence and that the perpendicular dimension is one of arousal, as predicted by the emotional circumplex theory of affect (Russell, 1980). As such, unlike predicted in the current study, psychopaths did not seem to show an arousal focus when processing emotional words. Pictures of Facial Affect: A similar procedure was employed to extract the underlying dimensions used by psychopaths and nonpsychopaths when interpreting pictures of facial affect. However, due to the smaller number of facial stimuli (9 versus the 28 words), the maximum number of dimensions extractable was four (see Takane, et al., 1977). As seen in Figure 10, a clear elbow emerged at the 2D solution for psychopaths (stress = .04), while for nonpsychopaths the elbow emerged at the three-dimensional (3D) solution (stress = .003). However, the nonpsychopaths' stress value was not found to significantly improve with the addition of a third dimension as the 2D stress value (.06) was already below the decision cutoff of .10 (Davison, 1983; Kruskal & Wish, 1978), thereby also suggesting a 2D solution for nonpsychopaths. In addition, as above, the stress values of psychopaths and nonpsychopaths for the ID solution (stress = .17;stress = .21, respectively), although highly similar, were above the decision cutoff, further attesting to the appropriateness of a 2D solution. Note, however, that unlike the ID solution of the words, the ID solution of the faces accounted for a large amount of the stress, thereby suggesting that offenders placed more weight on the first dimension than on the second. Nevertheless, the 2D solution for both psychopaths and nonpsychopaths was found to account for a large proportion of the variance in the distances (i.e., similarity) between faces (RSQ = .99; RSQ = .98, respectively). As such, a 2D solution for both diagnostic groups seemed to best characterize the pictures of facial affect. The faces fell in an elliptic arrangement around the two dimensions (see Figure 11 and 12 77 —•— Nonpsychopaths ---»--• Psychopaths 0.45 , 0.4 J 0.35 J 0.3 -o = 0 .25 -ro > in D1 D2 D3 D4 Dimensional Solutions Figure 10: Stress value by dimensional solutions of the pictures of facial affect for both diagnostic groups 78 2 1.5 -L 11 0.5 CO 3 o < - H — i — i — i — i — I — i — i — i — i — | — i — i — i — Q — H 1 1 1 1 1 1 1 1 1 1 1 1 h -1.5 -1 -0.5 © 0.5 1 • 1i -0.5 -1 I •1.5 1 -2 Valence Figure 11: Nonpsychopath's emotional circumplex for the pictures of facial affect 79 1.5 0.5 ro 3 O H — i — i — i — I — i — i — i — i — I — i — i — i — O - H — i — i — i — I — i — i — i — i — | — i — i — i — i — h H — I — I — I --1.5 -1 -0.5 0 0.5 14 -0.5 -1.5 . Valence Figure 12: Psychopath's emotional circumplex for the pictures of facial affect 80 for nonpsychopaths and psychopaths, respectively), which visual inspection revealed to be valence and arousal, respectively. Like the words, however, these findings do not seem to support the original hypothesis that psychopaths, as compared to nonpsychopaths, are arousal-focused. In fact, the elliptic formation and the ID stress values actually suggests a valence-focus on the part of the offenders. Dimensional Interpretation: Although the MDS arrangement of the words and faces suggested that these dimensions represent valence and arousal, several researchers have criticized this subjective mode of interpretation (e.g., Ortony, Clore, & Collins, 1988; Smith & Ellsworth, 1985). As such, the stimulus coordinates of each stimulus (resulting from the 2D solution) for each group of stimuli (i.e., words and faces) were correlated with their respective valence and arousal ratings. Table 8 presents the absolute values of the correlations for both psychopaths and nonpsychopaths. Absolute values are presented since MDS result in arbitrary unidimensional arrangements. For the emotional words, dimension 1 and dimension 2 for both diagnostic groups were found to correlate significantly with valence and arousal ratings, respectively. That is, both groups processed the emotional words as predicted by the emotional circumplex. For the faces, a similar pattern emerged for psychopaths, while correlations of nonpsychopaths were found to be significant only between the second dimension and the arousal ratings. This suggested that dimension one of the nonpsychopaths is not one representing valence. This is an odd finding, in that, one would have expected nonpsychopaths to have behaved like normals and, thus, reproduce the emotional circumplex. If taken at face value, this would indicate that nonpsychopaths are the ones with an emotional deficit, not psychopaths - the first 81 Table 8 Correlations between the dimensional coordinates of stimuli resulting from a 2D solution and the pleasure and arousal ratings for both diagnostic groups Nonpsychopaths Psychopaths Pleasure Arousal Pleasure Arousal Words: dl 97 *** .06 97 *** .08 d2 .07 97 *** .09 9g *** Faces: dl .58 .13 97 *** .04 d2 .04 .84* .02 91 ** Note: dl = dimension 1; d2 = dimension 2 * p < .01; ** p < .001; *** p < .0001 r = absolute value such finding in the literature. However, in light of several factors, this interpretation was not deemed likely, thereby prompting further analyses. Additional MDS Analyses: As stated above, several factors suggested that further analyses were indeed warranted. Firstly, it was deemed unlikely that nonpsychopaths in this study, and in no other, were found to have an emotional deficit. Secondly, preliminary results based on a subset of offenders (n = 32) indicated that nonpsychopaths did rely on the emotional circumplex, as defined by Russell (1980), to make their similarity judgments (Herve & Hare, 1998). Thirdly, if nonpsychopaths were indeed characterized by a multidimensional emotional deficit, then they should have shown a similar pattern when processing the emotional words. Taken together, this suggested that the nonpsychopaths' nonsignificant coordinate x rating correlation may have been due to some group x stimulus interaction. As such, a possible link between the pictures of facial affect and some subset of the current offender group was sought after. Given the fact that some offenders were chosen from a treatment program for sex offenders, that most of the participants excluded in the preliminary analyses were sex offenders (see Herve & Hare, 1998), and that the pictures employed were of a woman, groups were further subdivided by sex offenders, yielding four groups: nonpsychopathic non sex offenders (NPNSO: n = 10); nonpsychopathic sex offenders (NPSO: n = 9), psychopathic non sex offenders (PNSO: n = 15); and psychopathic sex offenders (PSO: n = 6). Although the MDS solutions did not indicate a need to reanalyze the data for the emotional words, these were included for comparison purposes. However, note that while MDS group differences for the emotional faces are expected to emerge from this new classification of participants, no such differences are expected for the emotional words. The mean PCL-R scores for NPNSOs, NPSOs, PNSOs, and PSOs were 25.05 (SD = 3.11), 20.86 (SD = 6.36), 33.08 (SD = 2.67) and 32.13 (£D = 2.04), respectively. Groups were not found to differ in age or in years of education. These groups were also compared to identify possible confounds and rating differences, but results remained unchanged. Table 9 and 10 represent valence and arousal rankings resulting from the rating of the words and Table 11 and 12 represent the valence and arousal rankings, respectively, for the faces for all groups. These were included for comparison purposes. Emotional Words - Part II: The same procedure was used as above, except that, in this case, the subjects' word dissimilarity matrices were averaged across the four groups of offenders. The stress by dimension plot (Figure 13) for all four diagnostic groups, which was highly similar to that observed in the original MDS analyses, showed a clear elbow at the 2D solutions for all diagnostic groups, further suggesting its suitability (stress = .11, .14, .09, and .14 for NPNSOs, NPSOs, PNSOs, and PSOs, respectively). The stress of the ID solution, irrespective of the group, was not found to be low enough to suggest that this solution was sufficient to account for the similarity judgments of participants. However, the stress accounted for by the ID solution of the non sex offenders (average stress = .34) was found to be greater than that of the sex offenders (average stress = .43), indicating that more variance in the data is accounted for by one dimension for non sex offenders than for sex offenders. As such, this suggests that the addition of a second dimension is more important to sex offenders than it is to non sex offenders. That is, sex offenders seem to place more weight on the second dimension than non sex offenders. The RSQ value for the 2D solution of the NPSOs, PSOs, NPNSOs, and PNSOs was .90, .89, .93, .96, respectively. Accordingly, overall a 2D solution seemed to best account for the cognitive-affective processes used by offenders in making similarity judgments between emotional words. Table 9 Pleasure rating rankings for emotional words from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending Words1 NPNSO NPSO PNSO PSO Happy 1.5 1 1 3 Pleased 3 4 3 2 Satisfied 1.5 2 5 6 Delighted 4 3 4 4 Serene 5 6 2 8 Glad 9 5 7 1 Content 6 8 6 10 At ease 10 7 10 9 Relaxed 8 9 8 11 Calm 7 11 9 12.5 Excited 11 13 11 5 Aroused 13 10 13 12.5 Sleepy 13 12 12 16 Astonished 13 15 14 7 Tired 15 14 15 14 Alarmed 16 17 16 17 Tense 19 16 18 18 Droopy 17 20 17 15 Afraid 18 18 19 20 Annoyed 20 21 21 23.5 Distressed 24 19 22 22 Frustrated 21 23 23 21 Bored 23 22 20 28 Sad 22 26 26 19 Angry 26 24.5 24 23.5 Gloomy 25 24.5 25 25 Miserable 27 28 27 26 Depressed 28 27 28 27 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample Table 10 Arousal rating rankings for emotional words from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending Words1 NPNSO NPSO PNSO PSO Excited 1 2 1 1 Alarmed 2 1 2 4 Angry 4 6 3.5 3 Aroused 5 6 3.5 2 Afraid 3 3 5 7.5 Astonished 6 4 9 5 Tense 9 6 6 10.5 Delighted 8 8 8 12 Frustrated 7 10.5 7 7.5 Happy 11 9 10 10.5 Annoyed 10 13 12 13 Distressed 13.5 12 11 14 Glad 13.5 14 13 6 Pleased 12 10.5 14 9 Satisfied 15 15 15 16 Miserable 16 17 17 18 At ease 20 22.5 18 15 Serene 19 16 22 18 Content 17.5 19 19 20 Sad 17.5 10.5 20 22 Calm 21 20.5 21 25 Depressed 22 18 25.5 21 Bored 24 25 23 18 Relaxed 23 22.5 24 26 Gloomy 25 24 25.5 23.5 Droopy 26 26 16 23.5 Sleepy 27 28 27 27 Tired 28 27 28 28 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample Table 11 Pleasure rating rankings for pictures of facial affect from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending Faces1 NPNSO NPSO PNSO PSO Excitement 1 2 1 1 Contentment 2 1 2 3 Pleasure 3 3 3 2 Sleepiness 4 4 4 4 Neutral 5 5 5 5 Arousal 6 6 6 6 Distress 7 7 8 7 Misery 9 9 9 9 Depression 8 8 7 8 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample Table 12 Arousal rating rankings for pictures of facial affect from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending Faces1 NPNSO NPSO PNSO PSO Excitement 1 1 1 1 Misery 2 3 2 4 Distress 3 2 3 3 Arousal 4 5 4 2 Pleasure 5 4 5 5 Contentment 6 6 6 6 Depression 7 7 7 8 Neutral 8 8 8 7 Sleepiness 9 9 9 9 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample Sex Nonpsychopaths * Sex Psychopaths «— Nonsex Nonpsychopaths -••»--• Nonsex Psychopaths 0.45 D1 D2 D3 D4 D5 D6 Dimensional Solutions Figure 13: Stress value by dimensional solutions of the emotional words for the four diagnostic groups The words fell in a more-or-less circular arrangement around what seemed to be the dimensions of valence and arousal for all groups (see Figure 14, 15, 16, and 17 for NPSOs, PSOs, NPNSOs and PNSOs, respectively), as predicted by the emotional circumplex. In addition, these figures revealed that sex-offenders placed more weight on the arousal dimension than non sex offenders, in that, the former's arousal dimension range seemed slightly longer than that of the latter. Note that a box depicting the dimensional ranges was added to each figure for illustrative purposes. As such, these findings do not seem to support the original hypothesis that psychopaths, as compared to nonpsychopaths, are arousal-focused. Rather, these seem to suggest that sex offenders placed more weight on the dimension of arousal than non sex offenders, suggesting that the former pays more attention to the arousal dimension than the latter. Pictures of Facial Affect - Part II: Figure 18 represents the stress by dimension plot for all four groups resulting from the MDS analyses of the emotional faces. A clear elbow was found for the NPSOs, PSOs, and PNSOs at the 2D solution, but not for the NPNSOs. However, since the 2D stress value for all groups was below the decision cutoff (stress = ..09; .06; .04; and .08 for NPNSOs, NPSOs, PNSOs, and PSOs, respectively), the addition of a third dimension could not be justified, even though the NPNSOs' stress by dimension plot demonstrated a clear elbow at the three dimensional solution (Davison, 1983). In terms of the ID solutions, clear differences emerged between the four groups (see Figure 18), suggesting differential weighing of the dimensions. As with the words, although more pronounced, sex offenders' ID stress value (average stress = .29) was greater than that of non sex offenders' (average stress = .15), indicating that the former benefited more from the addition of a second dimension than the latter. Moreover, the additional subdivision by psychopathy seemed to amplify the distinction, such that, the fit of a ID solution 90 1.5 ro 3 O < 0.5 H 1 1 1 1 1 1 1 1 1 1 1 h --1 -0.5 H 1 1 1 1 1 1 1 1 1 1 1 h --1.5 0.5 1.5 -0.5 -L -1 -1.5 1 Valence Figure 14: Nonpsychopathic sex offenders' emotional circumplex for the emotional words 91 2. 1.5tL 1 1 0.5 H 1 1 1 1 1 1 1 1 1 1 1 1 — I 1 1 1—I—0--1! 5 # • -1 -0.5 ffl -0.5 -1 -1.5 H—i—i—i—|—i—i—i—i—|—i—i—i—i—H—i—i—i—i-0.5 1 Valence Figure 15: Psychopathic sex offenders' emotional circumplex for the emotional words 2. 1.5 1 .,,,), I [ I j • * 1 • 0.5 -» I 1 i 1 1 1 1 i I . 1 •... 1. 1 1 (l) • • • • • • • • i i i i i i i > i < i i 2 -1 ^ 1 1 1 | 1 1 1 1 | t * 1 r ^ -1 -0.5 ( -o.5: • • - 1 . -1.5 . 1 1 1 1 1 1 1 1 I | 1 1 1 I 0.5 1 • • • • • • • m. I—I I — I — I l — 1.5 ; - ? I Valence Figure 16: Nonpsychopathic non sex offenders' emotional circumplex for the emotional words 2. 1.5 1 1 0.5 03 (/> 3 O H 1 1 1 h H 1 1 1 1 1 1 1 — I 1 1 1 1 --1.S» -1 -0.5 $ H — I 1 1 1—I 1 — I 1—I 1 — I 1 — H 1 1 1 — I H 0.5 1 I 1.5 -0.5 -1 ,4,5.. Valence Figure 17: Psychopathic non sex offenders' emotional circumplex for the emotional words 94 —*— Sex Nonpsychopaths * Sex Psychopaths —•— Nonsex Nonpsychopaths ---»--• Nonsex Psychopaths 0.45 -, 0.4 -0.35 4 D1 D2 D3 D4 Dimensional Solutions Figure 18: Stress value by dimensional solutions of the pictures of facial affect for the four diagnostic groups became greater as one progressed from NPSOs (stress = .33), to PSOs (stress = .25), to NPNSOs (stress = .18), and to PNSOs (stress = .11), with that of the latter suggesting that a ID solution may be most appropriate for this group (PNSOs' stress « .10 = decision cutoff). NPSOs tended to benefit the most from the addition of a second dimension, followed by PSOs, and then by NPNSOs, while PNSOs barely benefited from an additional dimension. For the 2D solutions, the RSQ values were .97, .96, .96, and .99 for NPSOs, PSOs, NPNSOs, and PNSOs, respectively, while the RSQ of the PNSOs' ID solution was .96. Note the similarity between the 2D and ID RSQ values of the PNSOs; a similarity that further attests to the significance of the ID solution for this group. Thus, while a 2D solution seems to best account for the cognitive-affective processes used by NPSOs, PSOs, and NPNSOs in making similarity judgments between emotional faces, a ID solution seems to be most appropriate for PNSOs. Nonetheless, the 2D solution of PNSOs will be included in subsequent analyses, as will their ID solution, for comparison purposes. Based on the 2D solutions, the faces fell in a circular orientation around the two dimensions for NPSOs (Figure 19), PSOs (Figure 20), NPNSOs (Figure 21), and PNSOs (Figure 22), with dimension one and dimension two reflecting valence and arousal, respectively. Note that even though the best solution for PNSOs was not a 2D one, they were nonetheless able to extract some information from the second dimension, as reflected in their apparently appropriate arrangement of the faces along the arousal dimension. Inspection of these figures indicated that these groups weighed the arousal dimension differently. For illustrative purposes, a box depicting the dimensional ranges was added to each figure. As can be seen by the height of the boxes, the NPSOs placed the greatest weight on arousal, followed, in decreasing order, by PSOs, NPNSOs, and PNSOs. Visual inspection of the ID solution of PNSOs (Figure 23), suggested that it is one of valence. Thus, these findings indicated that NPSO and, to a lesser degree, PSO, 96 2 1.5 0.5 1 jH—i—i—i—i—I—i—i—i—i—I—l—l—l—0-41.5 -1 -0.5 $ 0.5 1 H 1 1 1 1 1 1 1 1 1 1 1 1 1 H—I 1 — I H i.b -0.5 1 •m-:5--2 Valence Figure 19: Nonpsychopathic sex offenders' emotional circumplex for the pictures of facial affect 2, • 0.5 -» • i • i i i i i i i i i l i l • • i • • i i i i i i i i i i 1 1 1 1 1 1— 2 -1.5 1 1 l 1 1 1 1 1 | 1 1 1 *r -1 -0.5 ( -0.5 -• -1 . 1 1 1 1 | 1 1 1 1 | 1 1 1 1 ) 0.5 1 • • • | I 1 I 1 1 1.5 : -1.5 4-_ = 2 J Valence Figure 20: Psychopathic sex offenders' emotional circumplex for the pictures of facial affect 2 . 1.5 1 1 0.5 I ra 3 o < H 1 1 1 1 H -|2 -1.5 H 1 i 1 1 1 1 1 1 1 1 h - H 1 1 1 1 1 1 1 1 1 1 1 1 h4l 1 1 1 r --0.5 0 0.5 .1-5 -0.5 -1.5 1 Valence Figure 21: Nonpsychopathic non sex offenders' emotional circumplex for the pictures of facial affect 99 1.5 4 1 4 0.5 4 ro I ' l ' l V ^ ^ V 1 | 1 1 1 -1.5 -1 1 | W I r -0.5 • 1 i i 0 0.5 H 1 1—••• 1—(——|——| 1 r — 1 1.5 : Valence Figure 23: Psychopathic non sex offenders' ID solution for the pictures of facial affect 101 may be arousal-focused; that NPNSO presumably, like normals, simply replicate the original emotional circumplex; and that PNSO, contrary to the original hypothesis, may in fact be valence-focused when processing emotional faces characterized by a white adult woman. Dimensional Interpretation - Part II: To avoid possible biases resulting from a subjective interpretation of the dimensions, the resulting 2D MDS coordinates of each stimulus for each group were correlated with their respective valence and arousal ratings. These correlations are depicted in Table 13. For both words and faces, dimension one was significantly correlated with the valence ratings and dimension two with the arousal ratings for all groups of offenders. As such, this indicated that the additional offender subdivision by sex offending was indeed warranted. Similarly, the coordinates for the emotional faces resulting from the ID solution of the PNSO group was found to correlate significantly with their respective valence ratings (r = .93, p < .0001). Finally, to simplify comparisons between the rating and dimensional data, the ranked 2D coordinates of the words and faces were compared to the ranked ratings of the total sample. Table 14 and 15 represent the valence and arousal rankings, respectively, for the words and Table 16 and 17 represent the valence and arousal rankings, respectively, for the faces for all groups. 102 Table 13 Correlations between the dimensional coordinates of stimuli resulting from a 2D solution and the pleasure and arousal ratings for groups classified by psychopathy and sex offending Non Sex Offender Sex Offender Pleasure Arousal Pleasure Arousal Nonpsychopath: Words: dl 9 7 * * * .04 9 4 * * * .14 d2 .01 9 7 * * * .17 92 *** Faces: dl 96 *** .05 9 5 * * * .00 d2 .10 92 *** .06 .83 * ychopaths: Words: dl 9 7 * * * .06 91 *** .07 d2 .06 92 *** .18 g 7 * * * Faces: dl 92 *** .02 9 7 * * * .02 d2 .05 g 9 * * * .07 .85 * Note: dl = dimension 1; d2 = dimension 2 * p < .01; ** p < .001; *** p < .0001 r = absolute value 103 Table 14 Pleasure coordinates for emotional words ranked from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending Words1 NPNSO NPSO PNSO PSO Happy 6 7 5 7 Pleased 1 4 2 4 Satisfied 5 3 4 3 Delighted 4 8 3 10 Serene 9 10 7 9 Glad 2 2 1 5 Content 3 1 6 1 At ease 10 9 8 6 Relaxed 8 5 10 8 Calm 7 6 9 2 Excited 11 11 11 13 Aroused 12 12 13 11 Sleepy 14 13 14 15 Astonished 13 14 12 12 Tired 15 15 15 14 Alarmed 17 19 18 17 Tense 20 25 20 21 Droopy 18 17 16 18 Afraid 19 22 19 20 Annoyed 24 21 25 24 Distressed 25 28 24 28 Frustrated 26 26 26 25 Bored 16 16 17 16 Sad 21 20 23 23 Angry 23 24 21 26 Gloomy 22 18 22 19 Miserable 27 27 28 27 Depressed 28 23 27 22 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample pleasure ratings Table 15 Arousal coordinates for emotional words ranked from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending Words1 NPNSO NPSO PNSO PSO Excited 3 3 5 3 Alarmed 1 4 1 4 Angry 6 6 6 8 Aroused 4 2 3 2 Afraid 5 8 4 5 Astonished 2 1 2 1 Tense 11 13 7 6 Delighted 10 7 9 10 Frustrated 7 10 8 7 Happy 13 12 13 15 Annoyed 8 5 10 9 Distressed 9 14 14 11 Glad 12 11 12 12 Pleased 14 9 11 17 Satisfied 16 15 15 14 Miserable 17 17 16 18 At ease 23 18 19 20 Serene 19 22 18 16 Content 15 16 17 13 Sad 22 23 22 21 Calm 20 21 23 19 Depressed 18 20 21 23 Bored 25 27 25 24 Relaxed 21 19 20 22 Gloomy 24 24 24 25 Droopy 26 25 26 28 Sleepy 28 28 28 26 Tired 27 26 27 27 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample arousal ratings Table 16 Pleasure coordinates for pictures of facial affect ranked from most pleasant to least pleasant averaged across subjects from each group classified by psychopathy and sex offending Faces1 NPNSO NPSO PNSO PSO Excitement 2 2 2 2 Contentment 3 3 3 3 Pleasure 1 1 1 1 Sleepiness 5 5 5 4 Neutral 4 4 4 5 Arousal 6 6 6 6 Distress 8 7 7 7 Misery 9 9 9 9 Depression 7 8 8 8 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample pleasure ratings 106 Table 17 Arousal coordinates for pictures of facial affect ranked from most arousing to least arousing averaged across subjects from each group classified by psychopathy and sex offending Faces1 NPNSO NPSO PNSO PSO Excitement 4 3 4 3 Misery 3 5 3 5 Distress 2 2 2 2 Arousal 1 1 1 1 Pleasure 5 4 5 4 Contentment 6 6 6 6 Depression 7 7 7 7 Neutral 8 8 8 8 Sleepiness 9 9 9 9 Note: P = psychopaths; NP = nonpsychopaths (NP) 1 = Rank order from total sample arousal ratings Discussion There has been a considerable amount of research aimed at investigating the psychopath's emotional deficit. Although much has been learned from these investigations, most researchers have focused on one aspect of emotional processing, valence, thereby excluding an important emotional component - arousal. The present study aimed to investigate emotions in psychopaths in terms of the circumplex theory of affect (Russell, 1980). It was suggested that psychopaths, as compared to nonpsychopaths, are addicted to arousal and, as such, would over-focus on the arousal dimension when interpreting emotional material, thereby resulting in a valence-based deficit. To investigate the scope of the disorder, both emotional words and pictures of facial affect were used, as well as objective and projective measures. Multidimensional scaling techniques were employed to extract the affective dimensions used by offenders when processing emotionally toned material. Psychopaths were not found to differ in their pleasure and arousal ratings of emotional words and pictures of facial affect. Previous studies have also found psychopaths to mask their emotional deficit on such objective measures (e.g., Forth, 1992; Patrick et al., 1993). Differences in emotional processing between psychopaths and nonpsychopaths are usually revealed with either more sensitive (e.g., Intrator et al., 1997; Patrick et al., 1993; Williamson et al., 1991) or more projective (e.g., Blair et al., 1995; Christianson et al , 1996) techniques. Therefore, in addition to the self-report measure, the present study employed Ward's (1977) category sort task; a projective task requiring participants to evaluate the "perceived" similarity between stimuli. It was expected that psychopaths, not given any clues on how to perform this task (e.g., instructed to sort according to pleasantness), would revert to their own, subjective interpretation system, which was hypothesized to be characterized by an arousal-hyper-focus. However, at first glance, no such effect emerged. 108 Psychopaths did not sort the words any differently from nonpsychopaths. The similarity judgments of both groups were found to be characterized by a two dimensional solution. As predicted by the circumplex theory of affect (Russell, 1980), the words fell in a circular arrangement around these two dimensions, which were revealed to be valence and arousal through both visual inspection and correlations with pleasure and arousal ratings. As that found with non-criminal samples (e.g., Russell, 1980; Feldman, 1995b), offenders weighed each dimension equally. As such, these findings suggested that psychopaths, like nonpsychopaths, are not arousal-hyper-focused, or even arousal-focused, but rather rely on both valence- and arousal-related information to interpret emotional material. This contrasts with previous findings indicating that psychopaths have a profound deficit in processing valence-based stimuli (e.g., Christianson et al., 1996; Intrator et al., 1997; Patrick et al., 1993; Williamson et al., 1991). A two-dimensional solution also emerged out of the similarity judgments of the pictures of facial affect. As with the words, visual inspection of the arrangements of the faces around the dimensions indicated these to be valence and arousal. Unlike the words, participants seemed to weigh the valence dimension more so than the arousal dimension (i.e., demonstrated a valence-focus), thereby producing an elliptic arrangement. This is unlike that predicted by the literature (see Feldman, 1995b; Russell, 1989), in that, a circular arrangement was expected. However, this may have been due to any one of a number of factors. Firstly, the smaller number of stimuli used, as compared with the words, may have resulted in a less sensitive task, thereby reducing the amount of variance attributable to a second dimension (Hull, 1980). Secondly, the more abstract nature of faces may have confused participants, motivating them to base their similarity judgments on what they deemed to be most socially desirable, thus resulting in seemingly valence-based judgments (see Feldman-Barrett, 1996). Thirdly, these faces may simply not have been intriguing enough to participants as to engage their attention to the same extent as the 109 words, which would also result in some socially acceptable response pattern (see Feldman-Barrett, 1996). Nonetheless, these results did suggest that offenders seem to demonstrate a valence-focus when interpreting pictures of facial affect, a result opposite to that which was predicted. However, while correlational analyses validated the visual interpretation of the psychopath's circumplex, they did not support that of the nonpsychopath. Correlations for this group were found to be significant only for the arousal dimension, indicating that nonpsychopaths relied on arousal and some dimension other than valence when interpreting emotional faces. Thus, at first glance, the current findings did not seem to support the original hypothesis. In fact, nonpsychopaths, not psychopaths, seemed characterized by an emotional deficit - the first such finding of its kind. Given the wealth of research showing that nonpsychopathic offenders are relatively normal compared to psychopathic ones, this interpretation was, however, deemed unlikely and prompted further analyses. Several factors, grounded in both theory and research, pointed to the possibility that sex offenders in this sample were behaving differently than non sex offenders. As such, participants were further subdivided into sex and non sex offenders, yielding four groups composed of psychopathic and nonpsychopathic sex offenders and non sex offenders. This lead to important differences between groups. As above, ratings of the emotionality of the material were not found to distinguish between groups, which further highlights the fact that objective tests are not sensitive enough to reliably measure the psychopath's affective deficit. Psychopaths seem to understand, at an intellectual level, the characteristics of affect; a fact that has frustrated many clinicians. Remorse, for example, is a highly motivating emotion that can be employed in the therapeutic endeavor, especially in forensic settings. Unfortunately, such an affective state rarely, if ever, strikes the 110 psychopathic offender. Although such individuals, usually seeking treatment to impress the parole board (Ogloff, Wong, & Greenwood, 1990; Rice, Harris, & Cormier, 1992), admit to feeling remorse over their actions, their descriptions, although correct at some level, too often seem devoid of any real feeling. Consequently, therapists cannot capitalize on this important therapeutic tool to engage their psychopathic clients. Given that even a skilled therapist has difficulty tapping the psychopath's emotions, it is not surprising that objective measures also fail at this task. Such measures, providing the participant with the dependent variable of interest (e.g., pleasure and arousal ratings), can only truly provide researchers with the psychopath's intellectual understanding of emotions, not of their subjective feelings. As such, the present findings further indicate that psychopaths, like other offenders, understand emotions, irrespective if these are truly felt. Although the ratings were not found to be useful in differentiating between groups, the same cannot be said of the sortings, a task which provided no clues about the dependent variable of interest. Like the results seen in the original analyses, offenders, irrespective of psychopathy, were found to rely on the dimension of valence and arousal to interpret emotional words. However, in this case, sex offenders were found to place relatively more weight on the arousal dimension than non sex offenders. As such, these findings could be viewed as either suggesting that sex offenders are relatively arousal-focused or that non sex offenders are relatively valence-focused. Since a valence-focus produces a clear elliptical formation (see Feldman, 1995a), which was not observed, these findings were viewed to be in line with the former assumption, that sex-offenders are arousal-focused as compared to non sex offenders. Although this could indicate that sex offenders are relatively insensitive to physiological arousal compared to non sex offenders (see Blascovich, 1990; 1992; Mandler, 1984), there is no such evidence in the literature (or none read by the present author) indicating this to be the case. i l l Indeed, sex offenders, who - in the classical sense - are rarely psychopathic (Porter et al., in preparation), often seem anxious (Prentky & Knight, 1991), thereby suggesting an arousal hypersensitivity, not an arousal hyposensitivity. Rather, it is suggested that this effect may have been due to some aspect of their disorder, whereby, through experience (see Mandler, 1984), such individuals have come to view emotional events as more arousing than do non sex offenders - an assumption supported by their interpretations of the pictures of facial affect. When processing emotional faces, participants were found to reproduce the emotional circumplex as defined by Russell (1980). Furthermore, this structure was validated through both visual inspection and correlational analyses, thereby indicating that the additional offender subdivision was indeed warranted. Sex offenders, in general, were again found to place more weight on the arousal dimension as compared to non sex offenders, but to a much greater degree than that seen with the emotional words. In light of the fact that an arousal-focus occurs when cued by some environmental stimuli (Blascovich, 1990; 1992), this would indicate that the faces engaged the sex offenders' attention to a greater degree than the emotional words, thereby highlighting their arousal-focus and, thus, increasing the differences between these two groups. That is, the emotional faces, depicting an adult woman, seemed more salient to the sex offenders than to non sex offenders. Much like the drug addict who has come to associate needles, through classical conditioning (Childress et al., 1993; Mandler, 1984), as a way to fulfill his/her needs, the sex offender may have come to associate women as the objects of his needs. Moreover, while a needle without the drug may do little for the addict, a picture of a woman may in itself be adequate stimulation for the sex offender, especially when incarcerated in an all male environment. Indeed, current theories suggest that sex offenders are actually addicted to sex, especially of the deviant type (McCulloch, Snowden, Wood, & Mills, 1983; Pitchers, 1990). Therefore, any sex-related stimuli, such as a picture of an adult female, should engage their 112 attention and, more specifically, their arousal-focus. Accordingly, the pictures employed in the current study served as a stimulating cue for the sex offenders, not because they are necessarily inflicted with an arousal hyposensitive system, but rather because such stimuli, in light of their addiction, have previously been associated with arousal (Mandler, 1984). In a similar way, emotional language may have also come to reflect a greater degree of arousal for non sex offenders than sex offenders. Such individuals, presumably starting off with a normal affective structure, may have used emotional terms to describe their sexual acts, which are themselves characterized by a high degree of arousal. For example, the term "love", for the sex offenders, may have been repeatedly associated with arousing events (e.g., I love to rape), while the use of the term by non sex offenders may have been in relation to more mundane experiences (e.g., I love mom). Thus, although both types of individuals would have had the chance to associate both valence and arousal to their linguistic structures, a greater degree of arousal is likely to have been (classically) conditioned to that of sex offenders (see Mandler, 1984), thereby explaining the above findings. However, this interpretation should be viewed as tentative until validated by future research. Such investigations may employ both sex-related and sex-unrelated stimuli, thereby clarifying the impact that sex offenders' addiction has had on their affective structure. In addition to furthering our understanding of sex offenders, these results also revealed a sex offender x psychopathy interaction that was highly congruent with the theory that psychopaths, being pathologically under-aroused, are indeed addicted to arousal. As noted above, sex offenders, irrespective of psychopathy were found to be arousal-focused. However, psychopathic sex offenders seemed less aroused by the female pictures than their nonpsychopathic counterparts, as suggested by their relatively less pronounced arousal-focus. Interestingly, this finding closely parallels current views on psychopathic sex offenders (Brown 113 & Forth, 1997; Porter et al., in preparation). Nonpsychopathic sex offenders, being addicted to some deviant sexual act (e.g., rape, pedophilia, etc.), have a long standing history of sexual offending (Prentky & Knight, 1986; 1991), often ranging into hundreds of offenses, which are usually committed against only one type of victim (e.g., just women or just boys or just girls). In contrast, psychopathic sex offenders, although engaging in crime from a very early age, are not often long term sexual offenders, nor are their sexual offenses limited to only one type of victim. Instead, presumably addicted to arousal more so than sex, psychopaths are opportunistic sexual offenders (Brown & Forth, 1997), who, given the chance, will assault men, women, and/or children (Porter et al., in preparation). Indeed, this opportunistic and indiscriminant type of sexual behavior is often highlighted by their own rationalizations, as exemplified by one such offender who, when asked why he raped a young girl, replied "a pair of hips was all I saw". Thus, given the much greater amount of sexual crimes committed by nonpsychopathic sex offenders than by psychopathic ones, such stimuli are likely to have had many more opportunities to have been paired with arousal in the former than in the latter. Therefore, the current findings may reflect the view that psychopathic sex offenders are not "true" sex offenders and, as such, the pictures were not as salient to them as to their nonpsychopathic counterparts. In addition, these pictures, which were relatively mundane compared, for example, to pornographic pictures, were probably more in line with the nonpsychopathic sex offender's addiction than with that of the psychopathic ones. That is, for the nonpsychopathic sex offender, these pictures were likely perceived as actual sexual stimulation, while psychopaths perceived only a possible source of arousal, thereby explaining their different degrees of arousal-focus. Interestingly, this effect occurred even though the majority (if not all) of the participants in the current study described the woman in the pictures in very derogatory terms. As such, this 114 suggests that even though they may not necessarily like what a person looks like, they may nonetheless perceive him/her as a source of arousal, that is, as a victim to be preyed upon. While the sex offenders all showed an arousal-focus to the pictures of facial affect, the non sex offenders, psychopathic or not, seemed to place more weight on the dimension of valence when interpreting these stimuli, as demonstrated by their more elliptic emotional circumplex. For comparison purposes, the nonpsychopathic non sex offenders were assumed to have completed the sorting in a relatively normal fashion. Indeed, their emotional circumplex reflected a good balance between valence and arousal (see Russell, 1980; Feldman, 1995a). However, because these were nonetheless individuals incarcerated for violent crimes, the addition of a non-criminal control would have helped to strengthen this assumption. In terms of the psychopathic non sex offenders, results indicated that they almost exclusively relied on valence when interpreting the pictures of facial affect. That is, they demonstrated a strong valence-focus - a finding in direct contrast to the original hypothesis. It was predicted that psychopaths, being pathologically under-aroused, would show an arousal-hyper-focus; one that resulted in a valence processing deficit. Yet, none of the psychopaths, sex offenders or not, were found to show such a deficit. In retrospect, this finding, in light of the pattern demonstrated by the psychopathic sex offenders, actually provides more support for the arousal addiction theory than does the original arousal-hyper-focused hypothesis. Research with drug addicts has shown that drug-related cues come to elicit, through classical conditioning, a drug-congruent reaction in addicts (Childress et al., 1993). While relatively mundane cues, such as a "joint" for the heroine addict, does not grab the addict's attention, cues depicting his/her drug of choice (e.g., a needle) not only engage his/her attentional system, but actually elicit a state of craving; a reaction believed to be at the root of their relapse. Similarly, while the female pictures in the current study may have been salient to the 115 psychopathic sex offenders, these may have been too mundane for the psychopathic non sex offenders. These pictures seem to have been a possible source of stimulation for the psychopathic sex offenders, thereby triggering their attentional bias (i.e., arousal-focus). On the other hand, for psychopathic non sex offenders, who had no reason to view these as arousing given that sex offending had yet to be included in their criminal repertoires, these pictures probably elicited a non-affective interpretation strategy; one uncharacterized by an arousal-focus. Indeed, if pathologically under-aroused, then only stimuli that they deem arousing would elicit an affective response (Blascovich, 1990; 1992; Ellis, 1987; Mandler, 1984), while unarousing events would be quickly dismissed by some cognitive interpretive system devoid of any affective meaning (Mandler, 1984). Thus, these results support the proposed theory that psychopaths are addicted to arousal. Like drug addicts, psychopaths only become intrigued - and thus only show an arousal-focus - when presented with stimuli that are actually related to their addiction (addiction-congruent stimuli). Although in line with an arousal addiction, the fact that psychopaths in this study did not show a valence-based deficit suggests that they are not arousal-hyper-focused as predicted, but rather simply arousal-focused. As such, this would suggest that psychopaths are not completely insensitive to valence-based information, even when engaged by some arousing event. However, since their interpretation of what is pleasurable or not is based on an idiosyncratic affective structure, they are likely to have a unique understanding of valence. For example, psychopaths are likely to view all highly arousing events as pleasing, moderate ones as neutral, and low arousing ones as negative. The lack of a valence-based deficit also contrasts previous research in the area (e.g., Christianson et al., 1996; Williamson et al., 1990; 1991; Patrick et al., 1993). There is substantial research showing that psychopaths do indeed process valence-based emotional material 116 differently than do nonpsychopaths. However, although called a deficit by most, these researchers have usually found that psychopaths, although processing affective material differently, are nonetheless able to perform the task at hand with no more errors or processing speed requirements than their nonpsychopathic counterparts (Forth, 1992; Kiehl et al., in press; Mills, 1995; Patrick et al., 1993; Pham & Rime, 1994). Therefore, the current findings could be viewed as one more example of the psychopath's mask of sanity (Cleckley, 1976). Not only were psychopaths found to rate the emotionally-toned material no differently than nonpsychopaths, but, although they rely on a different cognitive structure than that of other offenders, psychopaths were able to mimic the pattern shown by their nonpsychopathic counterparts when interpreting emotional faces. This would suggest that while relying much less on the arousal dimension as compared to other offenders, psychopathic non sex offenders were still able to extract some arousal-based information from these pictures. Yet, if they do not rely on an affective meaning analysis, how were the psychopathic non sex offenders able to mimic the pattern demonstrated by other offenders? Given the high degree of overlap between the valence dimension and one representing social desirability (Feldman-Barrett, 1996), it is suggested that these individuals responded in what they believed to be the most socially desirable response; a strategy most likely learned through experience (see Mandler, 1984). They were able to produce a dimensional solution that resembled one of valence, further attesting to their chameleon-like abilities (Cleckley, 1976). However, this hypothesis was not directly tested and, thus, awaits validation. Therefore, future research in this area would benefit from adding some measure reflecting the socially desirable responses elicited from stimuli being employed (see Feldman-Barrett, 1996). Nevertheless, these findings highlight the importance of selecting appropriate stimuli when investigating the psychopaths affective structure. Like drug addicts who only focus on cues 117 reflecting their drug of choice, psychopaths will only engage their attention when stimulated by something they deem arousing. Indeed, this may explain past findings that have indicated psychopaths not showing an affective response where one was expected. For example, the lack of a fear imagery reaction demonstrated by psychopaths in the study by Patrick and his colleagues (1994) may in fact reflect that the task at hand was simply too mundane, rather than suggesting that psychopaths are unaffected by fearful imagery. Presumably, some other, more salient, imaginary stimulation would have triggered some reaction in these individuals. Therefore, researchers interested in tapping the psychopath's affective structure will need to select appropriately arousing material, or run the chance of simply investigating the psychopath's cognitive interpretation system. Finally, a question remains to be answered: Why were individual differences in attentional foci seen with the pictures of facial affect and not with the emotional words? It is suggested that, much like the faces being too mundane for some offenders, the words were in fact not salient enough to psychopaths as to engage their arousal-focus. That is, psychopaths relied on some non-emotional interpretive system when processing these words. Indeed, this conclusion is supported by research employing highly similar stimuli (Intrator et al., 1997; Kiehl et al., in press; Mills, 1995). In these investigations, at the behavioral level, psychopaths were consistently found to perform much like their nonpsychopathic counterparts; nevertheless, at the cortical and subcortical level, psychopaths demonstrated profound differences in the way these stimuli are actually processed. However, this does not explain why psychopathic non sex offenders were found to have a valence-focus when processing emotional faces, while demonstrating no such focus when interpreting the emotional words. It is suggested that this difference may be due to varying amounts of "corrective feedback" that psychopaths have received about the use of such 118 information. Psychopaths, being highly sociable and verbose, may have received ample feedback about the correct use of emotional words throughout their development. In contrast, in light of their lack of concerns for others, they may never have had the chance to get such feedback concerning their evaluations of facial expressions. While they may have acquired the linguistic knowledge required to mimic the emotional circumplex of others when processing emotional words, their knowledge of emotional facial reactions may be much more limited, thereby resulting in a more unidimensional understanding. Indeed, when psychopaths have been presented with material that is low in feedback frequency (e.g., emotional metaphors), they have not performed as well. Their facade did not hold up (e.g., Hayes, 1995). If this effect is indeed due to some corrective feedback mechanism, then adolescent psychopaths, who have had less opportunities to learn the corrective use of emotional language, should show more of a socially desirable valence-focus as compared to their adult counterparts - an assumption awaiting further research. The current findings suggest that psychopaths, when faced with salient stimulation, demonstrate an arousal-focus. Moreover, the addiction-congruent nature of this attentional bias, which parallels that seen in drug addicts, suggests that psychopaths are indeed addicted to arousal; an addiction proposed to have evolved out of a hyposensitivity to physiological arousal (see Blascovich, 1990; 1992; Ellis, 1987; Hare, 1978). As such, they may have learned to compensate for their normally under-aroused state by focusing their attention on the stimulating aspects of life, thereby helping them refuel on arousal. When faced with an emotional situation, such as the prospect of a sexual liaison, psychopaths are likely to become focused on its arousing potential and, given their lack of inhibition, quickly act upon it. However, when faced with a relatively neutral situation, such as that imposed by treatment, psychopaths may never get the chance to show this pattern, enabling them to engage their mask of sanity. It is thus likely that 119 the psychopath's true colors will only emerge under states of intense arousal. Indeed, this is often what is reported by their victims: "When we met, he was so nice and charming, but once we got back to my place.. .he seemed transformed.. .like a predator looking at its kill . . .that's when I knew I was in trouble". Furthermore, psychopaths, requiring environmentally produced arousal to function at optimal levels (see Ellis, 1987), are not likely to discriminate against different types of arousal. That is, the arousal addicted psychopaths are likely to view all types of arousal as a positive experience, irrespective of its presumed valance. Indeed, while winning at Blackjack would be both a positive and arousing experience for both psychopaths and nonpsychopaths, only the psychopaths is likely to get the same emotional response to the decapitation of an innocent victim. Their need for arousal may also explain their calmness in the face of traumatizing events. As Mandler (1984) noted, those who feel in control of a situation (e.g., the offender) are likely to experience positive affect, while those who perceive no control over this same situation (e.g., the victim) are likely to feel negative affect. Since the psychopath is more often than not the instigator, as reflected in their tendency to engage in instrumental aggression (Cornell et al., 1996; Herve et al., in preparation), s/he is likely to perceive a sense of control over the situation, resulting in a positive experience. This would suggest that the psychopath's grandiosity and egocentricity may be a result of his/her unique affective structure which has "taught" him/her to seek (or perceive) control over situations to maximize their chances for arousal. Even more disturbing is the fact that psychopaths are likely to continue to engage in their socially devastating behaviors as these are consistently followed by a state of positive affect, that is, one characterized by arousal (see Mandler, 1984). Only by pairing their behaviors with negative affective states can we expect to control these same behaviors. Taking an arousal addiction view of psychopathy may also help to explain why 120 psychopaths, as compared to nonpsychopaths, are responsible for a much greater amount of (violent) crime upon being released from prison, or even, from forensic treatment programs (Hart, Kropp, & Hare, 1988; Hemphill, et al., 1998; Ogloff, et al., 1990; Rice, et al., 1992). Like the drug addict who feels an unbearable craving when faced with drug-related cues, the arousal addicted psychopath may similarly feel such a craving when faced with arousal-related cues, thereby re-engaging his/her criminal cycle. A similar type of reasoning could also be applied to sex offenders, who are presumably addicted to sex. Like the drug addict who must fight off cravings for most of his/her life, the sex offender, when faced with the "right" victim, may also have to ward off such a craving; one compelling him to rape, sodomize, and terrorize his victims. Obviously, this can only have extremely aversive consequences for the victims. Not only does the psychopath rely on his/her own actions for arousal, but certain types of individuals (e.g., the lonely woman) may actually cue the psychopath's addiction, setting a craving into motion which they are not likely to be able to control. In addition, the victims' own reactions are likely to create an even more arousing situation, thereby only adding to the psychopaths' pleasure. Their sense of helplessness, their tears, their fears, and their constant pleading are all likely to validate the psychopath's sense of control, as well as increase the intensity of the act, which can only serve to reinforce the psychopath's actions. As such, s/he is likely to seek the same reaction in future victims, even if it means relying on torture to achieve it. Accordingly, only by gaining a better understanding of the motives of psychopaths can we expect to one day control the impact that such individuals have on society. The present study seems to be a step in the right direction. By demonstrating that psychopaths are arousal-focused, a new clue into their motivational system has emerged. Psychopaths seemed motivated to seek arousal, irrespective of the cost to society or to themselves. This information may enable clinicians to devise better treatment programs - programs specifically aimed at treating the 121 psychopathic offender. Only by knowing what motivates the psychopath's thoughts and behaviors can we ever expect to be able to control these same thoughts and behaviors. Rather than continuing to treat such offenders with techniques developed for their nonpsychopathic counterparts, efforts should concentrate on controlling their addiction, or replacing the source of their "fix". Indeed, this was highlighted by Brink (1997): "Teaching psychopaths how to feel is an exercise in futility. What one needs to do is identify their tools and teach them how to use them.. .For us, emotions are potent "alarm" bells, bells that we use to monitor our impulses, while for the psychopath, these are, at best, only faint, muffled "sounds" heard at a distance." In addition to having important theoretical and therapeutic implications, this new theory may help us understand something about their criminal patterns. As addicts, psychopaths are likely to habituate fairly quickly to any one stimulation "dose", thereby requiring either more exciting events or a quicker re-offending pattern in order to return their systems to optimal arousal levels. As such, this information could have valuable practical applications. Indeed, the homicide investigator, in his/her search for a serial murderer, should be alerted to the possibility that s/he is dealing with a psychopathic individual if the murders are becoming more intense (i.e., more gruesome) or are occurring at a much faster rate. Conversely, at the beginning of an investigation, if the suspect is known to be a psychopath, then law enforcement agencies should expect the individual's crimes to become increasingly "intense" and/or closer in succession. Finally, though important for our understanding of psychopathy, these findings also point to the validity of dimensional views of affect (e.g., Mandler, 1984; Russell, 1980; 1989). While no differences emerged when words and faces were categorized in terms of their valence or arousal levels, important differences were identified when adopting a dimensional view -differences that may help to clarify many aspects of the disorder. Previous investigations, relying on a categorical definition of emotions, are therefore likely to have misinterpreted some of their 122 findings. Indeed, as mentioned earlier, several of the proposed psychopathic deficits are likely to have been due to emotional variations between participants, only explainable as such when adopting a dimensional view. For example, the lack of a guilt attribution in psychopaths may, rather than being due to a lack of some Violence Inhibition Mechanism (Blair, et al.1995), have reflected that psychopaths, instead of seeing guilt, may have perceived arousal, thereby "empathizing" with the guilt inducing protagonist. Furthermore, these findings also support previous research suggesting individual variations in people's affective circumplex (Feldman, 1995a; 1995b), and that these are likely to have important behavioral consequences (Blascovich, 1990; 1992; Mandler, 1984). Thus, research on psychopathy is not only likely to have important forensic implications, but also help clarify current debates in the field of emotions. Nevertheless, future research needs to be conducted before adopting an arousal addiction view of psychopathy, as the current investigation could be criticized due to several methodological considerations (some of which have already been mentioned and thus will not be repeated). Firstly, it should be noted that the separation between the two groups of participants in terms of psychopathy was not as sharp as it has been in other studies, in that, the mean PCL-R score for the current sample was several points higher than that observed in previous investigations. Consequently, the nonpsychopathic groups were not as nonpsychopathic as they might otherwise have been, therefore attesting to the strength of the current findings. Secondly, the present findings were based on a subdivision of offenders that was not part of the original hypotheses. In addition, the post-hoc nature of this subdivision resulted in fewer participants than normally recommended (i.e., n = 10) for group based MDS analyses (Hull, 1980). Although this may suggest that the current results may not be as reliable as assumed, the fact that they were in line with the proposed theory indicates otherwise. Thirdly, although the present analyses demonstrated a clear effect between groups 123 defined in terms of psychopathy and sex offending, the results were limited to a relatively descriptive interpretation, rather than the preferred statistical one. An argument could thus be made that an experimenter bias could have affected the interpretations due to the more subjective nature of the procedure employed. Thus, further research is needed to validate the current findings, research that should seek more statistically sound procedures (see Feldman, 1995a). Fourthly, as alluded to above, a greater number of pictures of facial affect would have likely produced a more sensitive measure (see Hull, 1980). However, given that these were chosen on theoretical grounds (i.e., to represent the emotional circumplex) there is no reason to believe that the addition of stimuli would have significantly altered the current pattern of results (see Carroll & Chang, 1970). Instead, it is suggested that such an addition may enable researchers to gain a greater understanding of the subtle affective differences between different types of offenders. In a similar vein, the addition of non-relevant stimuli (e.g., non-emotional) may have helped to cloak the true nature of the current investigation, thereby decreasing the chance that offenders will respond according to some socially desirable ideation rather than to their own. This is an important point to keep in mind when conducting research with psychopaths as these individuals are expert manipulators and, when it suits them, awesome people pleasers. Indeed, a large proportion of the offenders in the current project were able to guess that they were performing an emotional task. As such, future tasks developed for the purpose of unmasking the psychopath's affective structure should make great efforts to try to hide the variable of interest. In addition to these limitations, future investigations should take into account that different types of stimuli have different effects for various offenders. Accordingly, it is strongly suggested that investigations in this area should include stimuli that vary in both valence and arousal, thereby tapping both the psychopath's cognitive and affective interpretive systems. 124 Furthermore, in addition to the use of projective measures to elicit the psychopath's true affective structure, researchers interested in expanding the current findings should include physiological measures (e.g., electrodermal and blink startle measures); measures that could validate the manipulation of the independent variable employed. Conclusion The present investigation was an initial step in the systematic investigation of a possible arousal addiction in psychopaths. It was suggested that if psychopaths do indeed suffer from such an addiction, then they should demonstrate an arousal-focus when processing affective material. Although this did not seem to be the case at first glance, the additional subdivision of participants into sex offenders and non sex offenders was congruent with the proposed theory. Not only was an arousal-focus demonstrated, but this attentional bias, like that seen in drug addicts, was found to be addiction sensitive, in that, only stimuli that were related to the offenders addictive cycle were found to elicit an arousal-focus. When not intrigued, psychopaths simply relied on what they knew best, that is, an interpretation system geared towards presenting themselves in a favorable light. As such, it was suggested that the psychopath's unique affective structure has caused them to develop an opportunistic world view in which people and objects are evaluated solely in terms of how potentially beneficial they may be. This research may thus help us to understand something about the processes that allow them to treat others as little more than objects.. .objects needed to satisfy their "hunger" for arousing stimulation. 125 References Blair, R. J. R., Sellars, C , Strickland, L, Clark, F., Williams, A. O., Smith, M. , & Jones, L. (1995). Emotional attributions in the psychopath. Personality and Individual Differences. 19, 431-437. Blascovich, J. (1990). Individual differences in physiological arousal and perceptions of arousal: Missing links between Jamesian notions of arousal-based behaviors. Personality and Social Psychology Bulletin. 16. 665-675. Blascovich, J. (1992). A biopsychosocial approach to arousal regulation. Journal of Social and Clinical Psychology. 11. 213-237. Brown, S. L. & Forth, A. E. (1997). Psychopathy and sexual assault: Static risk factors, emotional precursors, and rapist subtypes. Journal of Consulting and Clinical Psychology. 65(5). 848-857. Bush, L. E. (1973). Individual differences multidimensional scaling of adjectives denoting feelings. Journal of Personality and Social Psychology. 25. 50-57. Buss, A. H. (1966). Psychopathology. New York, NY: Wiley. Carroll, J. D., & Chang, J. J. (1970). Analysis of individual differences in multidimensional scaling via an n-way generalization of "Eckart-Young" Decomposition. Psychometrika. 35(31 283-319. Childress, A. R., Hole, A. V., Ehrman, R. N. , Robbins, S. J., McLellan, A. T., & O'Brien, C. P. (1993). Cue reactivity and cue reactivity interventions in drug dependence. NIDA Research Monographs. 137. 73-95. Christianson, S., Forth, A. E., Hare, R. D., Strachan, C , Lidberg, L., & Thorell, L. (1996). Emotion and memory in criminal psychopaths. Personality and Individual Differences. 20, 437-443. Cleckley, H. (1976). The mask of sanitv. St. Louis, MO: Mosby. Cooke, D. J. (1997). Psychopaths: Oversexed, overplayed but not over here? Criminal Behaviour and Mental Health. 7. 3-11. Cooke, D. J., & Michie, C. (1997). An Item Response Theory analysis of the Hare Psychopathy Checklist. Psychological Assessment. 9. 3-13. Cornell, D., Warren, J., Hawk, G., Stafford, E., Oram, G , & Pine, D. (1996). Psychopathy in instrumental and reactive violent offenders. Journal of Consulting and Clinical Psychology. 64. 783-790. Davison, M . L., (1983). Multidimensional scaling. Toronto, Ontario: Wiley. 126 Day, R., & Wong, S. (1996). Anomalous perceptual asymmetries for negative emotional stimuli in the psychopath. Journal of Consulting and Clinical Psychology. 105. 648-652. Dempster, R. J., Lyon, D. R., Sullivan, L. E., Hart, S. D., Smiley, W. C , & Mulloy, R. (1996, August). Psychopathy and instrumental aggression in violent offenders. Paper presented at the Annual Meeting of the American Psychological Association, Toronto, Ontario. Ekman, P., & Friesen, W. V. (1987). Universals and cultural differences in the judgments of facial expressions of emotions. Journal of Personality and Social Psychology. 53. 712-717. Ellis, L. (1987). Relationship of criminality and psychopathy with eight other apparent behavioral manifestations of sub-optimal arousal. Personality and Individual Differences. 8(6). 905-925. Federal Bureau of Investigation, (1992). Killed in the Line of Duty. Washington, DC: United States Department of Justice. Feldman, L. A. (1995a). Valence focus and arousal focus: Individual differences in the structure of affective experience. Journal of Personality and Social Psychology. 69(1). 153-166. Feldman, L. A. (1995b). Variation in the circumplex structure of mood. Personality and Social Psychology Bulletin. 21(S\ 806-817. Feldman-Barrett, L. (1996). Hedonic tone, perceived arousal, and item desirability: Three components of Self-reported mood. Cognition and Emotion. 10(1). 47-68. Forth, A. E. (1992). Emotion and psychopathy: A three-component analysis. Unpublished doctoral dissertation, University of British Columbia, Vancouver, British Columbia. Forth, A. E. (1996). Psychopathy in adolescent offenders: Assessment, family background, and violence. In D. J. Cooke, A. E. Forth, J. P. Newman, & R. D. Hare (Eds.), Issues in Criminological and Legal Psychology: No. 24. International perspectives on psychopathy (pp. 42-44). Leicester, UK: British Psychological Society. Forth, A. E., Hart, S. D., & Hare, R. D. (1990). Assessment of psychopathy in male young offenders. Psychological Assessment: A Journal of Consulting and Clinical Psychology. 2, 342-344. Frijda, N . H. (1992). The empirical status of the laws of emotions. Cognition and Emotion. 6(6). 467-447. Fulero, S. M . (1995). Review of the Hare Psychopathy Checklist-Revised. In J. C. Conoley & J. C. Impara (Eds.), Twelfth mental measurements yearbook (pp. 453-454). Lincoln, NE: Buros Institute Gainotti, G., Caltagirone, C , & Zoccolotti, P. (1993). Left/right and cortical/subcortical dichotomies in the neuropsychological study of human emotions. Cognition and Emotion. 7(1). 71-93. 127 Gillstrom, B. J., & Hare, R. D. (1988). Language-related hand gestures in psychopaths. Journal of Personality Disorders. 2. 21-27. Glass, G. V., & Hopkins, K. D. (1996). Statistical Methods in Education and Psychology (3 rd ed.). Boston, MA: Allyn and Bacon. Grant, V. (1977). The menacing stranger. New York, NY: Dover. Gretton, H. M . (1998). Psychopathy and recidivism in adolescence: A ten-year retrospective follow-up. Unpublished doctoral dissertation, University of British Columbia, Vancouver, British Columbia. Hare, R. D. (1968). Psychopathy, autonomic functioning and the orienting response. Journal of Abnormal Psychology Monographs 73 ( 3, Pt. 2), 1-24. Hare, R. D. (1970). Psychopathy: Theory and Research. New York, NY: John Wiley & Sons. Hare. R. D., (1978). Electrodermal and cardiovascular correlates of psychopathy. In R. D. Hare & D. Schalling (Eds)., Psychopathic Behavior: Approach to Research (pp. 107-144). New York, NY: John Wiley & Sons. Hare. R. D., (1979). Psychopathy and laterality of cerebral function. Journal of Abnormal Psychology. 88. 605-619. Hare, R. D. (1981). Psychopathy and violence. In J. R. Hays, T. K. Roberts, & K. S. Soloway (Eds.), Violence and the violent individual (pp. 53-74). Jamaica, NY: Spectrum. Hare, R. D. (1991). The Hare Psychopathy Checklist - Revised. Toronto, Ontario: Multi-Health Systems. Hare, R. D. (1993). Without conscience: The disturbing world of the psychopaths among us. New York: Simon & Schuster. Hare, R. D. (1995). Psychopaths and their victims. Harvard Mental Health Letter. 12. 4-5 Hare, R. D. (1996). Psychopathy: A clinical construct whose time has come. Criminal Justice and Behavior. 23. 25-54. Hare, R. D. (1998). Psychopathy, affect, and behavior. In D. Cooke, A. Forth, & R. Hare (Eds.). Psychopathy: Theory. Research, and Implications for Society (pp. 105-137). Dordrecht, The Netherlands: Kluwer. Hare, R. D., Frazel, J., Bus, J., & Jutai, J. (1980). Psychopathy and structure of primary mental abilities. Journal of Behavioral Assessment. 2. 77-88. Hare, R. D., & Jutai, J. W. (1983). Criminal history of the male psychopath. In K. R. Van 128 Dusen, & S. A. Mednick (Eds.), Prospective studies of crime and delinquency. Boston, MA: Kluwer-Nijhoff. Hare, R. D., & McPherson, L .M. (1984). Violent and aggressive behavior by criminal psychopaths. International Journal of Law and Psychiatry. 7. 35-50. Harpur, T. J., Hare, R. D., & Hakstian, R. (1989). A two-factor conceptualization of psychopathy: Construct validity and implications for assessment. Psychological Assessment: A Journal of Consulting and Clinical Psychology. 1. 6-17. Harris, G. T., Rice, M . E., & Cormier, C. A. (1991). Psychopathy and violent recidivism. Law and Human Behavior. 15. 625-637. Hart, S. D., Kropp, P. R., & Hare, R. D. (1988). Performance of psychopaths following conditional release from prison. Journal of Consulting and Clinical Psychology. 56. 227-232. Hayes, P. J. (1995). Psychopathy and sensitivity to emotional metaphors. Unpublished doctoral dissertation, University of British Columbia, Vancouver, British Columbia. Hemphill, J. F., Templeman, R., Wong, S., & Hare, R. D. (1998). Psychopathy and crime: Recidivism and criminal careers. In D. Cooke, A. Forth, & R. Hare (Eds.), Psychopathy: Theory. Research, and Implications for Society (pp. 375-398). Dordrecht, The Netherlands: Kluwer. Herve & Hare. (1998, March). Psychopathy: A deficit in multidimensional emotional processing. Poster presented at the Biennial Meeting of the American Psychology-Law Society (APA Div. 41), Los Angeles, California. Herve, Petitclerc, & Hare. Instrumental aggression in psychopath: Egosyntonic versus egodystonic aggression, (in preparation). Hodge, J. E. (1992). Addicted to violence: A new model of psychopathy. Criminal Behavior and Mental Health. 2. 212-223. Hull, C. H. (1980). Session X V multidimensional scaling: A symposium on design considerations for multidimensional scaling. Behavior Research Methods and Instrumentation. 12(2). 274-280. Intrator, J., Hare, R., Stritzke, P., Brichtswein, K., Dorfman, D., Harpur, T., Bernstein, D., Handelsman, L., Schaefer, C , Keilp, J., Rosen, J., & Machac, J. (1997). Brain imaging (SPECT) study of semantic and affective processing in psychopaths. Biological Psychiatry. 42. 96-103. Izard, C. E. (1992). Basic emotions, relations among emotions, and emotion-cognition relations. Psychological Review. 99. 561-565. Jacobson, N. , & Gottman, J. (1998). When Men Batter Women. New York, NY: Simon & Schuster. 129 Johns, J. H., & Quay, H. C. (1962). The effect of social reward on verbal conditioning in psychopaths and neurotic military offenders. Journal of Consulting Psychology. 26. 217-220. Karpman, B. (1961). The structure of neurosis: With special differentials between neurosis, psychosis, homosexuality, alcoholism, psychopathy, and criminality. Archives of Criminal Psychodynamics. 4. 599-646. Kiehl, K. A., Hare, R. D., McDonald, J. A., & Brink, J. (in press). Semantic and affective processing in psychopaths: An event-related potential (ERP) study. Psychophysiology. Kosson, D. S., Smith, S. S., & Newman, J. P. (1990). Evaluation of the construct validity of psychopathy in Black and White male inmates: Three preliminary studies. Journal of Abnormal Psychology. 99. 250-259. Kroner, D. G., & Weekes, J. R. (1996). Balanced Inventory of Desirable Responding: Factor structure, reliability, and validity with an offender sample, Personality and Individual Differences. 212(3). 323-333. Kruskal, J. B., & Wish, M . (1978). Multidimensional Scaling. Beverly Hills: CA; Sage. Larbig, W., Veit, R., Rau, H , Schlottke, P., & Birbaumer, N. (1992, October). Cerebral and peripheral correlates in psychopathy during anticipation of aversive stimulation. Paper presented at the Annual Meeting of the Society for Psychophysiological Research, San Diego. Larsen, R. J., & Diener, E. (1992). Affect intensity as an individual difference characteristic: A review. Journal of Research in Personality. 21, 1-39. Lykken, D. T. (1957). A study of anxiety in the sociopathic personality. Journal of Abnormal and Social Psychology. 55. 6-10. Mandler, (1984). Mind and Body. New York: Norton. McCord, W., & McCord, J. (1964). The psychopath: An essay on the criminal mind. Princeton, NJ: Van Nostrand. McCulloch, M . J., Snowden, P. R., Wood, P. J. W., & Mills, H. E. (1983). Sadistic fantasies: Sadistic behaviors and offending. British Journal of Psychiatry. 143. 20-29, Mehrabian, A. & Russell, J. A. (1974). An Approach to Environmental Psychology. Cambridge, Massachusetts: M . I. T. Press. Mills, B. (1995). Cerebral asymmetry in psychopaths: A behavioral and electrocortical investigation. Unpublished doctoral dissertation. University of British Columbia, Vancouver, Canada. Newman, J. P., & Kosson, D. S. (1986). Passive avoidance learning in psychopathic and nonpsychopathic offenders. Journal of Abnormal Psychology. 95. 252-256. 130 Newman, J. P., & Wallace, J. F. (1993). Psychopathy and cognition. In P. Kendall & K. Dobson (Eds.), Psychopathology and Cognition (pp. 293-349). New York: Academic Press. Ogloff, J. R., & Wong, S. (1990). Electrodermal and cardiovascular evidence of a coping response in psychopaths. Criminal Justice and Behavior. 17. 231-245. Ogloff, J., Wong, S., & Greenwood, A. (1990). Treating criminal psychopaths in a therapeutic community program. Behavioral Sciences and the Law. 8. 81-90. Ortony, A., Clore, G. L., & Collins, A. (1988 ). The Cognitive Structure of Emotions. New York, NY: Cambridge University Press. Patrick, C. J. (1994). Emotion and psychopathy: Startling new insights. Psychophysiology. 31. 319-330. Patrick, C. J., Bradley, M . M. , & Lang, P. J. (1993). Emotion in the criminal psychopath: Startle reflex modulation. Journal of Abnormal Psychology. 102. 82-92. Patrick, C. J., Cuthbert, B. N. , & Lang, P. J. (1994). Emotion in the criminal psychopath: Fear image processing. Journal of Abnormal Psychology. 103. 523-534. Paulhus, D. L. (1994). Two-component models of socially desirable responding. Journal of Personality and Social Psychology, 46(3). 598-609. Pham, T. H., & Rime, B. (1994, June). Psychopathy and emotion words: Response latencies and categorization. Paper presented at the Annual Meeting of the Belgian Psychological Society, Liege, Belgium. Pitchers, W. D. (1990). Relapse prevention with sexual aggressors: a method for maintaining therapeutic gain and enhancing external supervision. In W. L. Marshall, D. R. Laws, & H. E. Barbaree (Eds), Handbook of Sexual Assault: Issues. Theories and Treatment of the Offender (pp. 343-361). New York, NY: Plenum Press. Plutchik, R. (1980). Emotion: A psychoevolutionary synthesis. New York, NY: Harpur & Row. Porter, S., Drugge, D., Fairweather, H., & Herve, H. F. (in preparation). Psychopathy and sexual offending: Implications for recidivism and treatment. Prentky, R. A. & Knight, R. A. (1986). Impulsivity in the lifestyle and criminal behavior of sexual offenders. Criminal Justice and Behavior. 13(2). 141-164. Prentky, R. A. & Knight, R. A. (1991). Identifying critical dimensions for discriminating among rapists. Journal of Consulting and Clinical Psychology. 59(5). 643-661. Quay, H. (1965). Psychopathic personality as pathological stimulation-seeking. American Journal of Psychiatry. 122. 180-183. 131 Rice, M . E., Harris, G. T., & Cormier, C. A. (1992). An evaluation of a maximum security therapeutic community for psychopaths and other mentally disordered offenders. Law and Human Behavior. 16. 399-412. Romney, A. K., Shepard, R. N. , & Nerlove, S. B. (1972). Multidimensional Scaling. Volume II: Applications. New York, NY: Seminar Press. Roseman, I. J., Wiest, C , & Swartz, T. S., (1994). Phenomenology, behaviors, and goals differentiate discrete emotions. Journal of Personality and Social Psychology. 67 (2). 206-221. Russell, J. A. (1978). Evidence of convergent validity on the dimensions of affect. Journal of Personality and Social Psychology. 36. 1152-1168. Russell, J. A. (1979). Affective space is bipolar. Journal of Personality and Social Psychology. 37. 345-356. Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology. 39(6\ 1161-1178. Russell, J. A. (1989). Measures of emotion. In R. Plutchik & H. Kellerman (Eds.), Emotion: Theory. Research, and Experience (vol. 4., pp. 83-111). Toronto, Ontario: Academic Press. Russell, J. A., & Bullock, M . (1985). Multidimensional scaling of emotional facial expressions: Similarity from preschoolers to adults. Journal of Personality and Social Psychology. 38. 1290-1298. Russell, J. A., & Bullock, M . (1986). Fussy concepts and the perception of emotion in facial expressions. Social Cognition. 4. 309-341. Russell, J. A., & Lanius, U. F. (1984). Adaptation level and the affective appraisal of environments. Journal of Environmental Psychology. 4. 119-135. Russell, J. A., Lewicka, M. , & Niit, T. (1989). A cross-cultural study of a circumplex model of affect. Journal of Personality and Social Psychology. 57. 848-856. Russell, J. A., & Mehrabian, A. (1974). An Approach to Environmental Psychology. Cambridge, Massachusetts: MIT Press. Russell, J. A., & Mehrabian, A. (1977). Evidence for a three-factor theory of emotions. Journal of Research in Personality. 11. 273-294. Russell, J. A., & Ridgeway, D. (1983). Dimensions underlying children's emotion concepts. Developmental Psychology. 19. 795-804. Russell, J. A., & Steiger, J.H. (1982). The structure in persons' implicit taxonomy of emotions. Journal of Research in Personality. 16. 447-469. 132 Rutherford, M . J., Alterman, A. I., Caciola, J. S., & McKay, J. R. (1998). Gender differences in the relationship of antisocial personality disorder criteria to Psychopathy Checklist-Revised scores. Journal of Personality Disorders. 12(1). 69-76. Salekin, R. T., Rogers, R., & Sewell, K. W. (1997). Construct validity of psychopathy in a female offender sample: A multitrait- multimethod evaluation. Journal of Abnormal Psychology. 106(4). 576-585. Schachter, S. (1971). Emotion. Obesity, and Crime. New York, NY: Academic Press. Smith, C. A., & Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. Journal of Personality and Social Psychology. 48. 813-838. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). The State-Trait Anxiety Inventory (Self-Evaluation Questionnaire). Palo Alto, CA: Consulting Psychologists Press. Stone, G. L. (1995). Review of the Hare Psychopathy Checklist-Revised. In J. C. Conoley & J. C. Impara (Eds.), Twelfth mental measurements yearbook (pp. 454-455). Lincoln, NE: Buros Institute. Syndulko, K. (1978). Electrocortical investigations of sociopathy. In R. D. Hare & D. Schalling (Eds), Psychopathic Behavior: Approach to Research (pp. 145-156). New York, NY: John Wiley & Sons. Takane, Y., Young, F. W., & Leeuw, J. (1977). Nonmetric individual differences multidimensional scaling: An alternating least squares method with optimal scaling features. Psychometrika. 42(1). 7-67. Tomkins, S. (1984). Affect theory. In K. Scherer & P. Ekman (Eds.), Approaches to Emotions (pp. 163-195). Hillsdale, NJ: Lawrence Erlbaum. Ward, L. (1977). Multidimensional scaling of the molar physical environment. Multivariate Behavioral Research. 12. 23-42. Ward, L. M. , & Russell, J. A. (1981). The psychological representation of molar physical environments. Journal of Experimental Psychology: General. 110 (2). 121-162. Watson, D.C., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology. 54, 1063-1070. Whissell, C. W. (1982). Dimensions of affect as predictors of free classification of affective words. Perceptual and Motor Skills. 54. 821-822. Williamson, S. E., Hare, R. D., & Wong, S. (1987). Violence: Criminal psychopaths and their victims. Canadian Journal of Behavioral Science. 19. 454-462. 133 Williamson, S. E., Harpur, T. J., & Hare, R. D. (1990, August). Sensitivity to emotional polarity in psychopaths. Paper presented at the meeting of the American Psychological Association, Boston. Williamson, S. E., Harpur, T. J., & Hare, R. D. (1991). Abnormal processing of affective words by psychopaths. Psychophysiology. 28. 260-273. Young, F. W., & Hamer, R. M . (1987). Multidimensional Scaling: History. Theory, and Applications. Hillsdale, NJ: Lawrence Erlbaum Associates. Zuckerman, M . (1979). Sensation Seeking: Beyond the Optimal Level of Arousal. Hillsdale, NJ: Erlbaum 134 Appendix Emotional Words: Practice Words: 1. Bike 2. Cat 3. Dog Experimental Words: 1. Afraid 2. Alarmed 3. Angry 4. Annoyed 5. Aroused 6. Astonished 7. At ease Pictures of Facial Affect: 4. Fish 5. Flower 6. Garage 8. Bored 9. Calm 10. Content 11. Delighted 12. Depressed 13. Distressed 14. Droopy 7. House 8. Plant 9. Tent 15. Excited 16. Frustrated 17. Glad 18. Gloomy 19. Happy 20. Miserable 21. Pleased 10. Tree 22. Relaxed 23. Sad 24. Satisfied 25. Serene 26. Sleepy 27. Tense 28. Tired 135