1 INTRODUCTION

Psychopathy is a severe personality condition with distinct emotional, behavioral, and interpersonal features (Cleckley, 1941). While psychopathy historically has been operationalized as a categorical construct, current research suggests that it may be better referred to as a dimensional personality trait cluster rather than a taxon (Edens et al., 2006; Skeem et al., 2011). One of the most striking features of psychopathy is the reduced responsivity to affective information. Specifically altered physiological reactions in terms of a diminished startle blink reflex and galvanic skin conductance (Benning et al., 2005; Patrick et al., 1993) or altered cognitive processes in terms of an impaired emotion recognition (Dawel et al., 2012) have been repeatedly associated with psychopathy. However, less is known about the processes underlying this attenuated affective responsivity in psychopathic individuals. One major approach suggests a fundamental and specific deficiency to accurately encode and process emotionally relevant information (Blair, 2006). Other authors, in contrast, have proposed that individuals with psychopathic traits exhibit (only) difficulties to direct attention to emotional information that is unrelated to the task aim (response modulation hypothesis) (Newman & Lorenz, 2003), thereby conceptualizing psychopathy mainly as a disorder of attention and motivation, but not emotion. Both theoretical concepts have stimulated extensive experimental research in the field including electrophysiological investigations and imaging studies that endeavor to gather information that is more precise on the origin and time course of the abnormal affective responding to emotional information in psychopaths.

In contrast to psychopaths, healthy individuals show a processing priority of emotional information and a large body of research has investigated the electrophysiological underpinnings of this preferential consideration of emotional stimuli. A well-documented correlate of emotional processing is the early posterior negativity (EPN), which occurs as a negative potential at occipital scalp regions between 200 and 300 ms poststimulus. This event-related potential (ERP) is thought to indicate early, automatic, and therefore bottom-up processing of emotional content representing the general mobilization of attentional resources (Hajcak, 2012; Schupp et al., 2006). Further, the late positive potential (LPP), which occurs as positive deflection at centro-patietal areas from as early as 300–800 ms (often referred to as early LPP) to as late as 2,000–3,000 ms poststimulus (i.e., late LPP), has been associated with affective processing. There is evidence to suggest that the LPP reflects nonautomatic or top-down processing and the specific allocation of attentional resources to salient events, such as emotional stimuli (Bradley, 2009; Hajcak, 2012; Hajcak et al., 2010, 2012; Olofsson et al., 2008). Notably, both ERPs have been shown to be modulated by several factors such as attentional focus, that is, indirect as compared to direct processing of emotional distractors (Sand & Wiens, 2011; Wiens et al., 2011, 2012; Wiens & Syrjänen, 2013). However, although the conditions that modulate the amplitude of EPN and LPP have been addressed extensively in previous research, the concrete functional significance of both ERPs remains rather elusive and is an area of ongoing research (Brown et al., 2012; Hajcak et al., 2009; Thiruchselvam et al., 2011).

In a recent literature review, Clark et al. (2019) presented data from studies that investigated early and mid-level potentials (e.g., P1, N1, P2, N2, P3) as well as late-level deflections (e.g., LPP) in psychopathic individuals. While the authors found no evidence to suggest early stage processing abnormalities in subjects who score high versus low in psychopathy, the review points to pronounced differences in late ERPs between both groups, that is, an attenuated LPP in psychopaths. Thus, the specific alterations that have been repeatedly associated with psychopathy may originate relatively late in the processing of emotional information (Clark et al., 2019).

Thus, although research suggests intact ‘bottom-up’ but impaired ‘top-down’ modulation of affective responding in psychopathy, less is known about the moderating role of attention processes. Considering that both early and late ERPs are modulated by attentional focus in healthy individuals (Sand & Wiens, 2011; Wiens et al., 2011, 2012; Wiens & Syrjänen, 2013), the response modulation hypothesis would propose that these effects should be even more pronounced in individuals who score high in psychopathy. To date, however, most ERP studies have used passive viewing paradigms (Baskin-Sommers et al., 2013; Medina et al., 2016; Sadeh & Verona, 2012; Venables et al., 2015) or tasks, where the emotional content of the stimuli was not task relevant (Carolan et al., 2014). Only one study implemented two types of experimental design where healthy participants with low versus high psychopathic traits had to respond to emotional stimuli in an oddball task with or without the instruction to differentiate between the displayed emotions (Anderson & Stanford, 2012). Individuals with high psychopathic traits only showed different ERPs for emotional target photographs compared to neutral targets when explicitly attending to the emotional content (and not in the implicit experimental condition). These findings emphasize the role of attention for emotional processing in psychopathy. However, in this oddball task, design emotional stimuli were very rare and task relevant in both conditions (irrespective of the different instructions) and thus, the experimental design used can be considered as a limitation for the implications of this study (Medina et al., 2016).

The present study aimed to replicate and extend the existing literature by using a novel paradigm, which was specifically designed to examine the role of the attentional focus in the processing of emotionally salient information (Schneidt et al., 2018). For this purpose, participants (incarcerated violent offenders with psychopathic traits (VOs) and controls (CTLs)) were presented with rectangles varying in spatial orientation displayed in front of positive, negative, and neutral pictures. In Task 1, all subjects were instructed to rate the rectangles (i.e., to determine orientation) and ignore the emotional stimuli, whereas they had to explicitly rate valence and arousal of these stimuli (and to ignore the rectangles) in Task 2.

Based on the existing literature, we hypothesized the EPN and LPP to be more pronounced in the direct processing condition (Task 2) as compared to the indirect processing condition (Task 1) in the control group. In the offender sample, we expected LPP amplitudes to be equal (as compared to CTLs) in the direct processing condition (Task 2), while amplitudes should be significantly lower in Task 1 (as would be predicted by the response modulation hypothesis of psychopathy). Although, to our knowledge, no study has investigated the EPN in psychopaths, we expected to find the same pattern as in the study by Anderson and Stanford (2012), that investigated the N2 component, with no group differences in Task 2 and an attenuated EPN amplitude in both groups in Task 1. No differences between groups regarding valence and arousal ratings were predicted (Benning et al., 2005; Medina et al., 2016).

2 METHODS 2.1 Participants

Twenty-seven incarcerated male VOs were recruited by facility members (who were not further involved in the study procedure) in cooperating German correctional facilities (Justizvollzugsanstalten Heimsheim and Hohenasperg). Inclusion criteria were a primary conviction for violent crimes (such as first-degree murder, armed robbery, aggravated battery and rape), age between 18 and 65 years, and a sufficient knowledge of the German language. Exclusion criteria were a history of psychotic-spectrum or bipolar disorders (as assessed by clinical interview) and/or a primary conviction for drug-related crime. All assessments were carried out in designated rooms of the facility. VOs were convicted for violent crimes such as first-degree murder, arson, rape, child rape, armed or aggravated robbery, aggravated battery and hostage taking. A control group (CTLs) of 27 healthy individuals was recruited via online advertisements. Inclusion criteria were similar to VOs with the addition that participants were not allowed to have any self-reported convictions or arrests. Both groups were carefully matched in terms of age and education. Participants gave written informed consent and received monetary compensation. The study was approved by the Clinical Ethics Committee at the University Hospital Tübingen and was conducted in accordance with the Helsinki Declaration.

2.2 Clinical measures

Self-reported aggression was assessed with a German version of the Buss–Perry Aggression Questionnaire (BPAQ) (Buss & Perry, 1992). A German version of the Mini International Neuropsychiatric Interview (MINI) (Ackenheil et al., 1999; Sheehan et al., 1998) was administered in order to assess exclusion criteria and comorbid disorders. Psychopathy was assessed using the Hare Psychopathy Checklist-Revised (PCL-R) (Hare, 2003). Independent experts determined PCL-R ratings for the present study as part of the standard forensic diagnostic procedure.

2.3 Procedure

Images were presented on a 15.4-inch WXGA wide TFT LCD laptop display at 1,140 × 900 resolution with an approximately viewing distance of 70 cm. Participants were prepared for the electroencephalography (EEG) recording by applying an electrode cap containing 63 electrodes. Subsequently, participants were introduced to the experimental paradigm and completed both Task 1 (indirect processing of emotional stimuli) and Task 2 (direct processing of emotional stimuli) in a fixed order. Both tasks were accompanied by an EEG recording. A detailed description of the paradigm and the used stimuli can be found in the study by Schneidt et al. (2018).

2.3.1 Stimuli

Two sets of rectangles were created based on a yellow square (410 × 410 pixels; without filling and not part of the final stimulus material) with 2-pixel increment/decrement steps, varying between 350 × 410 and 470 × 410 pixels. One set comprised 30 horizontal (e.g., 410 (+2 * 1) × 410 pixels, 410 (+2 * 2) × 410 pixels, …, 410 (+2 * 30) × 410 pixels) and the other set 30 vertically oriented rectangles. In order to manipulate task difficulty in both sets, each set was split into 15 rectangles with low (clear rectangles with vertical or horizontal orientation) and 15 rectangles with high perceptual demand (closer to squares with uncertain orientation). Although there were no specific hypotheses for task difficulty, we decided to keep this manipulation to ensure that participants had to focus on the rectangles. For all analyses or statistical comparisons between Task 1 and Task 2, only low perceptual demand trials were selected since this allowed us to analyze emotional responsiveness under conditions that were very similar with regard to their cognitive demand. Emotional (positive, negative) and neutral images were selected based on their normative scores for arousal and valence (Lang et al., 2008) from the International Affective Picture System (for further details, see Schneidt et al., 2018). The stimulus material excluded images with possible spatial anchors (e.g., black frames). Image complexity and content were controlled. We used the same 30 positive, 30 negative, and 30 neutral images (1,126 × 845 pixels) from the study by Schneidt et al. (2018), which were rated in a pilot study. Each valence category consisted of 15 images with complex scenes and 15 images with a clear figure-ground composition; additionally image content was balanced (images showing people or objects/animals).

2.3.2 Task 1

Participants received instructions to determine the orientation of the rectangles by button press while ignoring the task-irrelevant emotional images. Following each stimulus, participants had to rate the confidence of their decision (on a scale between 1 (very uncertain) and 5 (very certain)). Task 1 consisted of 360 trials separated into two blocks. Task difficulty and spatial orientation of the target stimuli were balanced across the emotional valence classes by randomly assigning each of the 60 rectangles to the emotional images during the first and second block, meaning that each image was presented four times.

Each trial started with a fixation cross (800 ms), which was followed by a stimulus screen with a random emotional image in the background and a centered yellow rectangle (1,000 ms). Subsequently, participants judged the spatial orientation of the rectangle and gave confidence ratings. We jittered the intertrial interval randomly (1,300, 1,400, or 1,500 ms, Figure 1a). Prior to the experiment, there were four practice trials with black-and-white Mondrian images in the background.

Typical trial structure of (a) Task 1, indirect processing of emotional stimuli. (b) Task 2, direct processing of emotional stimuli

2.3.3 Task 2

This task used identical stimuli and temporal trial structure as Task 1 with the exception that participants were instructed to ignore the yellow rectangles and this time to rate the emotional background images (Figure 1b). Participants rated the valence of the pictures between 1 (very positive) and 9 (very negative) in the first block. Next, participants rated the experienced arousal of the pictures between 1 (very low) and 9 (very high) in the second block. Prior to each block, one training trial was administered. This task compromised 180 trials ((30 positive +30 negative +30 neutral images) × 2 blocks). Pictures were presented in random order.

2.4 Electrophysiological recording methods and apparatus

EEG data were recorded with a BrainAmp Recorder (BrainProducts, Munich, Germany) from 63 electrodes according to the extended International 10–20 system. We used the BrainVision Analyzer Version 2.1.2.327 software (BrainProducts, Munich, Germany) to analyze signals offline. To register vertical eye movements, one additional electrode was placed below the left eye. The ground electrode was placed at Fpz and the recording reference was placed at FCz. The sampling rate was 500 Hz and electrode impedances were kept below 10 kΩ. A phase shift-free Butterworth IIR filter from 0.1 to 30 Hz was used to filter continuous EEG data. Further data were re-referenced to the average across all scalp channels. Ocular artifact detection and correction were applied to the raw EEG using independent component analysis. Time epochs lasting from −200 before to 1,000 ms after stimulus presentation was extracted. A baseline correction was done with the 200 ms before stimulus presentation. Difference curves were calculated between the mean ERPs of negative/positive and neutral images for the examination of the EPN und LPP. Based on the current literature (e.g., Hajcak, 2012) and visual inspection of our data (Figure 2a–d), EPN mean amplitudes were computed at Oz electrode between 200 and 280 ms. We calculated mean amplitudes for the LPP at the Pz electrode between 400 and 1,000 ms.

Electroencephalography results: (a) and (b) Early posterior negativity (EPN) mean amplitudes of Task 1 (respectively Task 2) computed at electrode position Oz between 200 and 280 ms (indicated by gray area), split by valence (negative, positive, and neutral) and group (VO and control [CTL]); (c) and (d) Late positive potential (LPP) mean amplitudes of Task 1 (respectively Task 2) computed at electrode position Pz between 400 and 1,000 ms (indicated by gray area), split by valence (negative, positive, and neutral) and group (VO and CTL)

2.5 Statistical analysis

All statistical analyses were performed using SPSS version 24 for Windows (IBM SPSS Statistics, IBM Corporation, Armonk, NY). Demographic and clinical variables were compared by t tests. Accuracy and reaction times of Task 1 were examined for outliers in order to ensure that all participants understood the instructions. In order to do so, outliers with data above or below 2 standard deviations were excluded from further analysis. Image valence and arousal rating scores in Task 2 were analyzed with repeated-measures analyses of variance (ANOVAs) with the within-subject factor Emotion (negative vs. positive vs. neutral) and the between subject factor Group (VO vs. CTL). To investigate the impact of attentional focus on ERPs, difference amplitudes of the EPN and LPP (positive minus neutral and negative minus neutral) were analyzed using repeated-measures ANOVAs with the within-subject factors Emotion (difference waves for the positive condition vs. negative condition) and Task, and the between subject factor Group. We only compared the easy version of Task 1 with Task 2 since this allowed us to analyze emotional distraction under similar conditions with regard to their cognitive load. Therefore, the only difference was the attentional focus (attended vs. not attended). Post hoc analyses were conducted with two-sided t tests. Pairwise comparisons were Bonferroni-corrected. The Greenhouse–Geisser method to correct for violations of sphericity in ANOVAs was applied if necessary. For all analyses, p-values lower than 0.05 were considered significant. Effect sizes are presented as adj. partial η2.

3 RESULTS 3.1 Participants

Two VOs and one CTL were excluded from analysis due to excessive EEG artifacts based on visual inspection. To examine the behavioral data for correct responses and RT, outliers (cutoffs were set by data inspection), defined as overall correct responses or RT deviating more than 2SDs, were excluded from analysis. One VO and one CTL had extreme low accuracy rates and another two CTLs had extreme high reactions times, exclusion of these participants resulted in a final sample of 24 VOs and 23 CTLs.

A demographic and clinical sample description for both VOs and CTLs is displayed in Table 1. The groups did not differ in terms of age and years of education. VOs showed higher aggression scores as indicated by BPAQ subscales “Physical aggression” and “Anger” as well as the total score. Twenty-one PCL-R ratings were available for the VO group with a mean score of 21.81 (SD 9.54, Min 4, Max 36). Three PCL-R ratings were not fully determined as part of the standard forensic diagnostic procedure and are therefore not available for the current study.

TABLE 1. Participant characteristics and clinical sample description VO (N = 24) CTL (N = 23) Statistics Demographics Age 43.79 (10.89) 43.96 (10.90) t(45) = −0.52, p = 0.959 Education (years) 9.75 (1.54) 9.57 (0.51) t(45) = 0.56, p = 0.582 BPAQ Physical aggression 22.00 (8.87) 15.83 (4.40) t(45) = 3.04, p = 0.004 Verbal aggression 14.92 (2.90) 14.30 (2.29) t(45) = 0.80, p = 0.427 Anger 14.58 (4.85) 11.35 (3.43) t(45) = 2.63, p = 0.012 Hostility 18.88 (6.31) 16.57 (5.20) t(45) = 1.37, p = 0.179 Total score 70.38 (18.52) 58.04 (11.25) t(45) = 2.77, p = 0.009 Note The data presented in the table refer to means and standard deviations for each measure (in parentheses). Abbreviations: BPAQ, Buss–Perry Aggression Questionnaire; CTL, control group; VO, violent offender group. 3.2 Behavioral analysis

The analysis of valence scores (Task 2) yielded a significant main effect of Emotion (F(1.28,57.38) = 530.19, p < 0.001, adj. = 0.92), indicating the expected rating pattern, that is, positive > neutral > negative images. The analysis of arousal scores yielded a significant main effect of Emotion (F(1.58,71.26) = 117.69, p < 0.001, adj. = 0.72), which was due to negative > positive > neutral arousal ratings. The interaction between group and emotion failed to reach statistical significance (F(1.58,71.26) = 2.73, p = 0.084, adj. = 0.06). Post hoc t tests indicated equal arousal ratings for negative and positive emotions in both groups (ps > 0.30) and a trend for higher arousal for neutral emotions in CTLs (mean ± SE, 2.99 ± 0.24) as compared to VOs (mean ± SE, 2.39 ± 0.22; t(45) = 1.84, p = 0.073). No other main effects or interactions reached significance (all ps > 0.10).

3.3 EEG analysis 3.3.1 EPN

With regard to the impact of attentional focus, the analysis of the EPN difference curves yielded a significant main effect of Task (F(1,45) = 7.32, p = 0.010, adj. = 0.14), indicating that Task 2 led to enhanced negativity of the EPN difference curves compared to Task 1 under low task difficulty (all other ps > 0.10). This demonstrates that focusing on the emotional background images has led to an enhanced early processing of emotional information in both groups (Figure 3).

Electroencephalography results of the early posterior negativity (EPN): Mean amplitude differences for each emotional valence category (negative minus neutral and positive minus neutral), task (Task 1 and Task 2), and group (VO and CTL). Analysis of the EPN difference curves yielded a significant main effect of Task (F(1,45) = 7.32, p = 0.010,  = 0.14), indicating that Task 2 led to enhanced negativity of the EPN difference curves compared to Task 1 3.3.2 LPP

With regard to the impact of attentional focus on the LPP, significant main effects of Emotion (F(1,45) = 5.74, p = 0.021,  = 0.11) and Task (F(1,45) = 7.17, p = 0.010,  = 0.14) were observed and were further qualified by a Group × Task interaction which, however, failed to reach the formal level of statistical significance (F(1,45) = 3.27, p = 0.077, adj. = 0.07). All other effects were not significant (ps > 0.10). The most relevant Group × Task interaction was followed up by post hoc t test analyses both between and within groups. While we found similar LPP amplitudes for both tasks in VOs (p > 0.50), a significantly increased LPP in Task 2 as compared to Task 1 (t(22) = 3.09, p = 0.005) was observed in CTLs. Post hoc t tests between groups revealed no differences in Task 1 (p > 0.60) and numerical but nonsignificant differences with higher LPP amplitude for CTLs as compared to VOs in Task 2 (t(45) = 1.88, p = 0.066, see Figure 4).

Electroencephalography results of the late positive potential (LPP): Mean amplitude differences for each emotional valence category (negative minus neutral and positive minus neutral), task (Task 1 and Task 2), and group (VO and CTL). Significant main effects of Emotion (F(1,45) = 5.74, p = 0.021,  = 0.11) and Task (F(1,45) = 7.17, p = 0.010,  = 0.14) were observed and were further qualified by a Group × Task interaction (F(1,45) = 3.27, p = 0.077,  = 0.07) 4 DISCUSSION

The present study investigated whether attentional focus (direct vs. indirect) would modulate affective responsiveness in an incarcerated offender sample with psychopathic traits as compared to CTLs. Early (EPN) and late (LPP) electrophysiological potentials were measured in response to emotional (positive and negative) and neutral stimuli. Participants were instructed to either ignore the emotional images and respond to rectangles with different shapes (Task 1) or to focus on the emotional images by rating their valence and arousal (Task 2). The main results of the current study can be summarized as follows: First, no differences in early stage processing could be observed between groups. Both groups exhibited a more pronounced (i.e., more negative) EPN amplitude when focusing on the emotional content of the stimulus (Task 2) as compared to the indirect processing condition (Task 1). Second, we observed group differences with regard to the late processing stage. CTLs showed increased (i.e., more positive) LPP amplitudes in Task 2 as compared to Task 1, indicating that task demands (i.e., attentional focus) had an effect on the processing of the emotional stimuli. In contrast, LPP amplitudes in the VO group were largely unaffected by task demands, thereby suggesting late alterations in the neural processing of emotional stimuli. In sum, this study provides new evidence for intact early “bottom-up” but deficient late “top-down” processing of affective information in violent offenders with psychopathic traits.

With respect to the early processing of emotional stimuli, our data suggest no abnormal responses in individuals with psychopathic traits. To our knowledge, no study has examined associations between the EPN amplitude and psychopathic traits. Anderson and Stanford (2012), however, used an oddball paradigm to investigate the N2, an ERP which largely overlaps with the EPN (Hajcak, 2012). The authors found the N2 to be sensitive to attentional focus in terms of increased ERP differentiation between emotional and neutral cues when attention was explicitly directed toward these features. These previous findings stand in contrast to the present data and might—at least partly—be explained by different samples (community participants vs. inmates) and task characteristics. While participants in our study were confronted with emotional images in every single trial of the experiment, Anderson and Stanford (2012) used an oddball paradigm, where emotional stimuli were presented rarely. As the N2 is indicative of changing features in the stimulus environment (Luck & Hillyard, 1994a, 1994b), their findings may likely reflect the changing stimuli rather than the emotional content or affective responsiveness of the participants. In line with this argument, a recent review reported that a vast majority of studies also found no associations between psychopathic traits and deviant early ERPs in various experimental designs (Clark et al., 2019). Taken together, these findings argue against early neural processing deficits of (emotional) information in individuals with psychopathy. In addition, the present work provides evidence suggesting that attentional focus modulates these early processes in violent offenders with psychopathic traits in a similar way as in healthy individuals.

With regard to late-stage processing of emotional information, we found that CTLs showed a more pronounced LPP in the direct processing condition (Task 2) as compared to indirect processing in Task 1, which was in line with expectations. In contrast to our hypotheses, our data showed no evidence for a modulatory effect of attentional focus on the LPP in VOs. They exhibited a LPP amplitude which was largely independent of task condition (i.e., attentional focus). This finding again stands in contrast to the results of the study by Anderson and Stanford (2012). While they found indirect processing of emotional stimuli to attenuate LPP amplitudes in high psychopathic individuals, they observed direct processing to increase LPP amplitudes. This effect was not present in low psychopathic controls. Possible explanations for these diverging results can be attributed to the abovementioned sample and task characteristics. In support of our findings, a majority of studies found the LPP also to be generally impaired in individuals with psychopathic traits (Clark et al., 2019).

The most interesting finding of our study was the complete absence of a more pronounced LPP amplitude for VOs in the direct processing condition (Task 2). Several explanations might account for this observation: First, VOs show a generally attenuated late response to emotional stimuli. This means that these individuals as compared to healthy controls appear to quickly terminate the processing of emotional information (Clark et al., 2019). Second, emotional images might have no intrinsic motivational significance to the VOs. Previous research could show that this feature modulates the positivity of the LPP (Schupp et al., 2000). Thus, our findings may reflect a general lack of motivational relevance of the stimuli for offenders with psychopathic traits. The latter explanation is, however, ruled out by the fact that arousal ratings in our study did not differ between groups. Third, and more specific, the lack of self-reference of the stimuli used in the present experiment might have accounted for the reduced LPP amplitudes observed in VOs. This possible explanation receives further support from experimental findings suggesting that psychopaths show reduced neural activity to images depicting pain of others, whereas they show normal activation after requesting them to imagine that they were the ones in pain (Decety et al., 2013; Meffert et al., 2013). Therefore, the results obtained give grounds to assume that deviant affective processing in criminal individuals with psychopathic traits might reflect motivational deficits rather than fundamental impaired abilities.

As with other empirical studies, the present study has some limitations. A clear limitation is that the sample consisted of exclusively male violent offenders with psychopathic traits. Whether the reported results would hold for female participants needs to be determined by future replications. Further, PCL-R scores of most VOs were under the predefined cutoff score of 30 (Skeem et al., 2011). While there is evidence for a dimensional rather than a taxonomic trait structure of the concept psychopathy (Edens et al., 2006; Skeem et al., 2011), it nevertheless would be important to replicate our study with high psychopathic individuals. On a related note, we did not assess psychopathy levels in controls, and therefore cannot convincingly conclude that it is psychopathy that drives the differences between groups. Thus, there might exist group differences between incarcerated offenders and controls other than criminal psychopathy that might at least partly explain our results (e.g., effects of incarceration, socioeconomic status). Consequently, future studies should carefully account for these factors by comparing offender groups with high versus low psychopathic traits. Another concern is that the used emotional images were only manipulated in terms of valence, thus allowing only for conclusions about broad affective categories. There is evidence for biologically relevant content (e.g., pictures of mutilation or erotic stimuli) to result in more pronounced EPN or LPP amplitudes than other stimulus classes (Briggs & Martin, 2009; Weinberg & Hajcak, 2010). Psychopathic traits might have a stronger association to these biologically relevant stimuli and therefore, future studies should investigate the role of stimulus content in psychopathy. Further, the fixed ordering of tasks in the present study design presents an array of potential confounds related to order effects, particularly concerning interaction effects involving repeated presentations of the same stimuli and group membership.

To summarize, we investigated early and late processing of emotional stimuli in violent offenders with psychopathic traits while manipulating the attentional focus. While we found early processes to be modulated by direct and indirect processing, late processes were generally attenuated, independent from attentional focus. With respect to the existing literature, our findings most likely reflect a specific motivational deficit or lack of self-reference in late attentional processes rather than a general processing deficit.

DECLARATION OF TRANSPARENCY

The authors, reviewers and editors affirm that in accordance to the policies set by the Journal of Neuroscience Research, this manuscript presents an accurate and transparent account of the study being reported and that all critical details describing the methods and results are present.

ACKNOWLEDGMENTS

The authors thank Annika Boss, Clara Held, Franziska von Helmolt, and Marie Heßlinger for their support in data collection. They also thank the staff of the correctional facilities Hohenasperg and Heimsheim for their support in recruitment of participants and organization of data collection.

CONFLICT OF INTEREST

The authors have no conflict of interest to declare.

AUTHOR CONTRIBUTIONS

All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Conceptualization, M.S. and A.S.; Methodology, A.S. and J.S.; Software, A.S., Investigation, J.S.; Formal Analysis, J.S., A.S., and M.S.; Resources M.S.; Writing – Original Draft, J.S. and M.S.; Writing – Review & Editing, M.S. and J.S.; Visualization, J.S.; Supervision, A.S. and M.S.; Funding Acquisition, M.S.