Abstract
Numerous studies have identified differences in the identification of emotional displays between psychopaths and non-psychopaths; however, results have been equivocal regarding the nature of these differences. The present study investigated an alternative approach to examining the association between psychopathy and emotion processing by examining attentional bias to emotional faces; we used a modified dot-probe task to measure attentional bias toward emotional faces in comparison with neutral faces, among a sample of male jail inmates assessed using the Psychopathy Checklist–Revised (PCL-R). Results indicated a positive association between psychopathy and attention toward happy versus neutral faces, and that this association was attributable to Factor 1 of the psychopathy construct.
Introduction
Psychopathy is a personality disorder characterized by affective, behavioral, and interpersonal dysfunction (Cleckley, 1976; Hare, 1991). As suggested by DeLisi (2009), psychopathy is the “unified theory of crime” (p. 265), which has been broadly investigated by the behavioral, social, and criminological researchers to describe crime, delinquency, and antisocial behavior.
Emotional Processing and Psychopathy
Deficits in emotional processing are key components in many prominent theories of psychopathy (e.g., Blair, 2005; Cleckley, 1976; Hare, 1993; Patrick, 1994; Steuerwald & Kosson, 2000) and there is evidence that some of the problems associated with psychopathy such as antisocial and aggressive behaviors may result from deficits in understanding other people’s emotional cues (Blair, 2003; Kirsch & Becker, 2007; Woodworth & Porter, 2002). Previous studies have indicated that psychopaths show deficits in recognizing and processing emotional cues (e.g., Blair et al., 2004; Deeley et al., 2006; Kosson, Suchy, Mayer, & Libby, 2002; Marsh & Blair, 2008). Nevertheless, our understanding of the nature of these differences remains imprecise, in part, due to inconsistent findings and the use of divergent methods across studies (Brook, Brieman, & Kosson, 2013).
The nature of the association between psychopathy and response to emotional cues varies across emotion and presentation modality (i.e., faces vs. verbal stimuli). For example, some previous studies have observed that psychopaths show deficits in emotion discrimination of sad and happy expressions (Habel Egbert, Salloum, Devos, & Schneider, 2002), whereas others found impaired recognition of both sad and fearful facial emotions (Blair, Colledge, Murray, & Mitchell, 2001; Blair et al., 2004), exclusive deficits for sad faces (Dolan & Fullam, 2006), and deficits for both happy and sad emotions (Hastings, Tangney, & Stuewig, 2008). In contrast, Glass and Newman (2006) reported no emotion recognition impairments in psychopaths. Finally, Kosson and colleagues (2002) indicated that psychopaths are deficient at identifying disgust expressions and perform more accurately at recognizing angry facial cues under some conditions. A meta-analysis of 22 studies (Wilson, Juodis, & Porter, 2011) concluded that psychopathy is associated with small recognition deficits for all emotions. These authors suggested that inconsistencies observed in different studies might result from the small effect sizes of said deficits, and that, as a result, deficits may be difficult to detect. In addition, other factors, including socioeconomic status (SES), verbal abilities, IQ, sex, and age might influence facial emotion processing (Herba & Phillips, 2004). In general, the small and somewhat inconsistent effects relating psychopathy to deficits in the recognition of emotions suggest a need for novel approaches to assessing the emotional differences widely proposed to underlie psychopathy.
Attentional Bias, Emotional Recognition, and Psychopathy
The amygdala has been indicated to play an important role in facial emotional processing (Anderson & Phelps, 2000; Stuhrmann, Suslow, & Dannlowski, 2011). It has also been suggested that amygdala dysfunction may partially underlie the antisocial behavior of psychopaths (Blair, 2007). These streams of inquiry converge to provide a rationale for implicating amygdala dysfunction in the affective processing deficits observed in psychopathy. Interestingly, a parallel body of research demonstrates that higher order cognitive processes might moderate amygdala-mediated responses to affective and emotional cues in both healthy samples and in samples of offenders characterized by psychopathic traits (e.g., Newman, Curtin, Bertsch, & Baskin-Sommers, 2010; Pessoa, Padmala, & Morland, 2005). Specifically, psychopaths showed reduced amygdala activation to fear and threat-relevant information only when their attention was engaged in an alternative goal-relevant task prior to presenting threat-relevant task, but showed amygdala activation similar to that seen in non-psychopaths, when explicitly attending to threat (Larson et al., 2013). Other research also points to a potentially important role for attentional processes in the apparent differences in emotional processing that characterize high psychopathy individuals; a study of prison inmates found that attention allocation moderated the relationship between psychopathy and fear-potentiated startle responses (Newman et al., 2010), suggesting that attentional dysfunction might limit psychopathic offenders’ processing of emotion-related cues. Further evidence of differences in attention to affective stimuli can be inferred from a recent study of youth, which associated psychopathy-related deficits in facial affect recognition with reduced attention to the eye region of facial stimuli (Dadds et al., 2006). Taken together, these findings suggest that apparent differences in affective recognition in high psychopathy individuals may reflect, or be moderated by, differences in attention to emotional material.
The affective dot-probe paradigm (MacLeod, Mathews, & Tata, 1986) is a widely used and effective approach to measuring individual differences in attention to affective stimuli (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van IJzendoorn, 2007). Using the affective dot-probe paradigm, previous studies have indicated that individuals with high levels of anxiety and other affective disorders show attentional bias toward threatening cues compared with neutral faces (Mogg & Bradley, 2005; Mogg, Philippot, & Bradley, 2004). Some recent studies have reported anomalies in attentional bias to emotional stimuli in youth with psychopathic features. For example, Kimonis, Frick, Munoz, and Aucoin (2008) reported that youth with callous–unemotional traits who were also characterized by high aggression and high exposure to community violence exhibited reduced attention to distress stimuli relative to youth without this combination of dispositions. These studies suggest that, at least among youth, differences in attention to emotional information may be associated with differences in the constellation of psychopathic traits exhibited.
In sum, several studies have identified an association between psychopathy and differences in emotional recognition. However, the exact nature of these deficits requires further specification. The extent to which these apparent differences in affective recognition reflect, or are accompanied by, differences in attention to affective material represents a potentially valuable avenue for elucidating emotional processing differences related to psychopathy but has not yet been examined in adults. To further understanding of the emotional processing deficits that may underlie psychopathy, we examined the extent to which psychopathic personality was associated with anomalous attention to emotional stimuli in adult criminal offenders. Specifically, we used a facial affect dot-probe task to examine whether psychopathy was associated with differences in attentional biases toward emotional stimuli among male inmates. We further examined the extent to which these attentional biases are related to specific emotions and were attributable to specific subcomponents of psychopathic personality. Based on prior evidence on psychopaths’ emotional and attentional problems (e.g., Dolan & Fullam, 2006; Kimonis et al., 2008), we hypothesized that higher levels of psychopathy would be associated with reduced attention toward both positive and negative affective faces, and that this effect would be most pronounced for negative affective faces. We further hypothesized that these effects would be attributable to the interpersonal and affective traits of psychopathy. This hypothesis was based on the previous studies that indicated attentional bias was related to callous and unemotional traits (Kimonis et al., 2008), which appear to map onto Factor 1 of the PCL measures (analogous to interpersonal and affective traits of psychopathy).
Method
Participants
Inmates were contacted randomly from the jail roster and invited to participate. Those who reported taking psychotropic medication and/or were not able to read English were excluded. Of those invited, approximately 70% agreed to participate. Participants were paid, and provided written informed consent to participate. Data were collected from 33 male inmates in a U.S. county jail. They all gave written informed consent to participate in this study. Participant ages ranged from 19 to 56 (M = 27.70, SD = 11.06). In this sample, 42.4% of offenders described themselves as Caucasians, 48.5% as African Americans, 3.0% as Latino, and 6.1% as “Other.” All participants were literate in English, and the mean years of education was 11.1 years (SD = 1.70). The study was approved by the Rosalind Franklin University of Medicine and Science ethics review committee. All data were used in analyses.
Materials
Psychopathy Checklist–Revised (PCL-R)
The Hare PCL-R is an expert rater measure of psychopathy (Hare, 2003). PCL-R scores were based on a semi-structured interview and a review of institutional files conducted by trained psychologists or graduate students. Past research using samples that partially overlap with the present sample has reported good interrater reliability for PCL-R scores (e.g., Swogger, Walsh, Lejuez, & Kosson, 2010; Walsh, 2013) The PCL-R consists of 20 items that are scored as 0 to 2 based on the degree to which each inmate’s personality and behavior match the characteristics of each item. Individuals were categorized as psychopathic if they received total scores of 30 or higher, non-psychopathic if their total scores were 20 or lower, and middle-scorers if they received scores between 21 and 29. Early studies suggested that two correlated but distinct factors underlie PCL-R total scores (Hare, 1991); Factor 1 describes interpersonal and affective features (e.g., callousness, shallow affect), whereas Factor 2 describes impulsive and antisocial characteristics of psychopathy (e.g., poor behavioral control, impulsivity). More recent studies subdivide Factor 1 and Factor 2 into narrower band four-facet scores. Facet 1 or “Interpersonal” Facet describes superficial charm, cunning, and manipulative features; Facet 2 or “Affective” Facet describes lack of remorse and empathy, and callousness characteristics; Facet 3 or “Lifestyle” Facet includes traits such as irresponsibility, impulsivity, and parasitic lifestyle; and finally, Facet 4 or “Antisocial” Facet describes characteristics such as poor behavioral control, and juvenile delinquency in psychopaths. In our sample, PCL-R scores demonstrated good internal consistency, α = .87. Alpha coefficients for the two factors of PCL-R were .80 for Factor 1 and .75 for Factor 2. Cronbach’s alpha was also calculated for PCL-R Interpersonal (.51), Affective (.78), Lifestyle (.52), and Antisocial (.74) Facets.
Dot-probe task
The dot-probe task was first introduced to measure attentional bias to emotional versus neutral stimuli (MacLeod et al., 1986). In the present version of the task (Figure 1), two faces (one emotional face and one neutral face) are presented concurrently in different positions on a computer monitor (top vs. bottom), followed by a probe appearing in the same location as one of the faces. Participants press a response button as soon as they perceive the probe, and the speed of response to probe stimulus is assumed to reflect attention to the stimulus that previously occupied the area containing the probe. The difference score is calculated as the mean response latency to probes appearing in the same location as affective stimuli (congruent trials) minus the mean response latency to probes appears in a different location than affective stimuli (incongruent trials) and is interpreted as indicating biases in the distribution of attention (Chen, Ehlers, Clark, & Mansell, 2002). That is, the degree to which response latencies are faster to congruent than to incongruent trials indicates the degree of attentional bias toward affective stimuli. Conversely, the degree to which response latencies are faster to incongruent than to congruent trials indicates the degree of attentional bias away from affective stimuli. In short, positive difference scores indicate attention directed toward emotional faces, and negative difference scores indicate attention directed away from emotional faces (i.e., toward neutral faces). In the current study, experimental trials were presented in which there were four categories of emotional expression (angry, happy, sad, and fearful). There were six images per category; half of the photographs in each category of emotions were of female faces, and the other half were of male faces. Each trial began with a fixation cross presented for 500 ms at the center of the computer screen. Then the facial stimulus pair was presented for 500 ms with one face above and one below (an emotional and a neutral face) the fixation position. The dot probe appeared immediately after the presentation of the face pair. After 500 ms latency, the next trial began. After participants responded to each probe, the next trial began. Response latencies for congruent and incongruent trials were recorded to calculate attentional bias.

Examples of congruent versus incongruent trials in the modified dot-probe task.
The Hollingshead Index (Hollingshead & Redlich, 1958) was used to measure SES. The Hollingshead Index is an incorporation of pre-incarceration educational and occupational achievements in a single index of SES, and correlates highly with other indicators of SES, which remains as one of the most widely used measures of SES, despite some limitations (Limbers, Ripperger-Suhler, Boutton, Ransom, & Varni, 2011). The Hollingshead Index has been successfully administered to inmates (Walsh & Kosson, 2007). Higher scores on the Hollingshead Index indicate lower SES.
Intelligence was assessed using the revised Shipley Institute of Living Scale (SILS; Zachary, 1986). The SILS demonstrates strong consistency and concurrent validity with Wechsler Adult Intelligence Scale–Revised (WAIS-R) and can be converted to WAIS-R Full Scale IQ estimates using well-validated normative tables (Zachary, 1986).
The Chapman Handedness Scale (Chapman & Chapman, 1987) consisting of 13 questionnaire items was administered to identify which hand participants use most frequently to perform common motor activities. Items were scored as “0” for left hand, “1” for right hand, and “0.5” for either. Men, scoring 8.5 or higher were categorized as right-handed.
The demographic questionnaire was used to assess age, ethnicity, and education level (completed years of education).
Procedure
Participants who provided written consent completed a semi-structured interview and self-report measures to be assessed for psychopathy using the PCL-R and to screen for the general characteristics required for this study. Then, in a separate session, participants were administered the dot-probe task. After receiving instructions, participants were invited to sit about 60 cm from the screen. The dot-probe task was introduced on a 12″ computer monitor. Following successful completion of the trials, participants were thanked for their participation.
Data Analyses
All statistical analyses were conducted using SPSS for Windows, v21.0. One-way ANOVA was utilized to compare demographic information and PCL-R score among three groups (psychopathic, non-psychopathic, and middle). As in most dot-probe studies, only correct responses were analyzed and we excluded response latencies greater than 2,000 ms, less than 200 ms, and two standard deviations below or above the mean (see Dewitte, Koster, De Houwer, & Buysse, 2007).
Two steps were used to test the hypothesis of the study: In the first step, bivariate correlation analysis was utilized to examine the correlation between PCL-R total and both Factor and Facet scores and the four attentional bias values (angry, happy, sad, and fearful). In the second step, one-way ANOVA was used to compare three psychopathy groups on their attentional bias scores for the four categories of trials presenting affective faces (angry, happy, sad, and fearful faces).
Results
Table 1 indicates the characteristics of participants categorized as psychopathic (39%, n = 11), non-psychopathic (33%, n = 13), and middle (28%, n = 9) groups and for the total study participants. All participants were right-handed and received estimate IQ scores of 70 or higher. Participants were significantly different in their PCL-R scores, F(2, 29) = 78.68, p< .001, and ages, F(2, 29) = 4.70, p< .05, such that the psychopathic group was significantly older than two other groups. No significant differences were found between groups in SES, years of education, handedness, IQ scores, and SES.
Characteristics of Participants in PCL-R-Defined Groups.
Note. PCL-R = Psychopathy Checklist–Revised (Hare, 2003); Handedness = Self-Report Handedness Measure (Chapman & Chapman, 1987); WAIS-R IQ = Wechsler Adult Intelligence Scale–Revised IQ, as estimated by the Shipley Institute of Living Scale (Zachary, 1986); SES = socioeconomic status.
Groups differ significantly at p< .01, df = 2. **Groups differ significantly at p< .001, df = 2.
Results from the correlation analysis showed that among PCL-R factors, only the relationship between Factor 1 score (callousness and unemotional traits) and attention toward happy faces was significant, r(31) = .39, p< .05 (Table 2). Results from correlation analysis between PCL-R Facet scores and attentional bias values indicated that Facet 2 (Affective) was positively correlated with attention toward happy faces, r(31) = .36, p< .05. Controlling for individual differences in age (Table 3), the relationship between PCL-R Factor 1 score and attention toward happy faces remained significant, partial r(31) = .45, p< .05.
Correlations Between the Four Attentional Bias Values and Psychopathy Score in PCL-R (N = 33).
Note. Factor 1 = score on PCL-R interpersonal and affective characteristics; Factor 2 = score on PCL-R impulsive and antisocial characteristics. PCL-R = Psychopathy Checklist–Revised.
p< .05 (two-tailed).
Partial Correlations Between the Four Attentional Bias Values and Psychopathy Score in PCL-R Controlling for Age (N = 33).
Note. Control variable: Age of participants at the time of interview; Factor 1 = score on PCL-R interpersonal and affective characteristics; Factor 2 = score on PCL-R impulsive and antisocial characteristics. PCL-R = Psychopathy Checklist–Revised.
p< .05.
The positive correlation between PCL-R Factor 1 score and PCL-R Facet 2 with attention toward happy faces could reflect either a greater positive bias or a smaller negative bias. Therefore, to clarify the nature of this correlation, a one-way ANOVA compared the three psychopathy groups on their attentional bias scores for the four categories of trials presenting affective faces (angry, happy, sad, and fearful faces). Attentional bias indices for the four categories of emotional faces are indicated in Table 4. The non-psychopathic group demonstrated attention away from happy faces, whereas the high psychopathy group demonstrated less attentional preference, but still negative difference score for happy faces. The Cohen’s effect size value suggested a moderate to high practical significance between psychopaths and non-psychopaths in attention toward happy faces (d = .7). Furthermore, Cohen’s effect size values between psychopaths and non-psychopaths for other affective faces were less than .3, suggesting a low practical significance. However, no significant differences were found between groups in attention to either happy or any type of negative affective faces in F test at p< .05.
Attentional Bias Indices for the Four Categories of Emotional Faces (N = 33, df = 2).
Note. The Cohen’s effect size value suggested a moderate to large effect between psychopaths and non-psychopaths in attention toward happy faces (d = .7), between psychopaths and inmates in middle group in attention toward fearful faces (d = .7), and between inmates in middle group and non-psychopaths in attention toward happy faces (d = .6), fearful faces (d = .7), and all faces together (d = .7). For F test, results indicated no significant differences between groups in attention to any type of affective faces at p< .05.
Discussion
In the present study, we investigated the association between psychopathy and attentional bias toward emotional faces among jail inmates, and examined the extent to which the bias was related to any sub-characteristics of psychopathy. Contrary to our expectations, psychopathy was not broadly associated with differential attention to facial emotion stimuli. These results are in contrast with studies that have found a relationship between psychopathy and attention to emotional cues (e.g., Blair et al., 2001; Blair et al., 2004; Dolan & Fullam, 2006; Habel et al., 2002; Hastings et al., 2008). However, the results of current study are in line with some prior findings that have not identified associations between psychopathy and performance on facial affect processing tasks (Glass & Newman, 2006), and have failed to find specific deficits related to psychopathy and recognition of fearful, angry, sad, or happy emotional expressions using both facial (Kosson et al., 2002), and verbal stimuli (Hiatt, Lorenz, & Newman, 2002). As such, our findings may be construed to cast further doubt on the stability of findings of emotional processing deficits among high psychopathic individuals.
Although we did not identify general affective differences associated with psychopathy, disaggregated analyses indicated that the core affective features of psychopathy were associated with attentional bias with regard to happy faces. Interestingly, this effect appeared to reflect greater attentional bias away from happy faces among the group who were lowest in the affective features of psychopathy than among offenders with the highest affective features of psychopathy. Further analysis indicated that the non-psychopathic group demonstrated attention away from happy faces, whereas the high psychopathy group demonstrated less attentional preference. As such, despite previous research that showed different impairments in processing of negative facial emotions in psychopaths, the mechanisms underlying processing of facial expressions of happiness may be functionally intact (Deeley et al., 2006). However, the lower (but still negative) difference score for psychopaths indicates that they are less likely to take their attention away from happy faces. These results may reflect higher rate of depression among correctional compared with community samples (Butler et al., 2006). Attenuated response to happy stimuli and attenuated attentional bias to positive cues have been reported in depressed individuals (Thomas et al., 2011), as has attentional bias to negative stimuli (Joormann & Gotlib, 2007). Therefore, the attention away from happy faces that particularly characterized the lower psychopathy participants in our sample may reflect higher negative affect among these participants. Indeed, several studies have reported that psychopathy inversely correlates with depressive symptoms (e.g., Stinson, Becker, & Tromp, 2005; Willemsen, Vanheule, & Verhaeghe, 2011), particularly the higher score on affective traits that we found to be related positively to attention toward happy faces.
The present study has several limitations, including the small sample size and the utilization of only male inmate participants. Therefore, we cannot generalize the results of the current study to female inmates and non-inmates. In addition, we did not control for some confounding variables, such as the length of incarceration that may affect the results. In general, the novelty of the current findings, the diverse outcomes of prior research, the relatively small sample, and large number of analyses highlight the need for cautious interpretation. Nevertheless, the moderate to large effect size values detected among three groups in attention toward some emotional faces (Table 4) suggest that such relationships might become significant in a larger sample.
In conclusion, results of this study provide further evidence regarding emotional processing deficits associated with psychopathy, and suggest that these deficits may not be generalized across emotions. This study further highlights the ambiguity surrounding the centrality of emotional processing differences to the psychopathy construct. Putative emotional deficits occupy a central role in the conceptualization of clinical and forensic responses to high psychopathy individuals, and apparent empirical equivocation regarding the extent and nature of these difficulties might caution such conceptualizations. It appears that we do not know, with any certainty, how different the emotions of the highly psychopathic are, and as such further research is required to support approaching these individuals as effectively “emotionless.” In light of the poor specificity and inconsistent results of investigations of emotional deficits, further psychopathy research might benefit from de-emphasizing the quest for emotional deficits, in favor of a broader examination of cognitive and behavioral differences, which may be more stable and potentially responsive to intervention.
Our study is the first investigation of differences in attention to affective materials in adult psychopaths. Therefore, we consider the result of this study as a primary step in this area and other studies are needed to assess the attention to affective materials using various methods and in larger samples of psychopathic individuals. Future studies are required in this area to elucidate this potentially valuable avenue and explore the exact nature of these interactions in psychopaths.
Footnotes
Acknowledgements
We thank Patrick Firman, Jennifer Witherspoon, Frank Kuzmickus, and the staff of the Lake County Jail and Division of Adult Probation for their consistent cooperation and support during the conduct of this research. We thank Marc Swogger for assistance with data collection.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research and preparation of this article were supported, in part, by National Institutes of Health Grant MH57714 to David S. Kosson, and by the Social Sciences and Humanities Research Council of Canada to Zach Walsh.
