Investigating error-related processing in incarcerated adolescents with self-report psychopathy measures

Abstract

Disparate results have been found in previous reports when incorporating both interview-based and self-report measures of psychopathic traits within the same sample, suggesting such assessments should not be used interchangeably. We previously found Total and Facet 4 scores from Hare’s Psychopathy Checklist: Youth Version (PCL:YV) were negatively related to amplitude of the error-related positivity (Pe) event-related potential (ERP) component. Here, we investigated using the same previously published sample whether scores on four different self-report measures of adolescent psychopathic traits (the Antisocial Process Screening Device [APSD], Child Psychopathy Scale [CPS], Inventory of Callous and Unemotional Traits [ICU], and Youth Psychopathic Traits Inventory [YPI]) were similarly associated with reduced Pe amplitude. Unlike our previous results, adolescent self-report psychopathy scores were not associated with reduced Pe amplitude in multiple regression analyses. Results obtained in the current report support previous research observing incongruent findings when incorporating different assessment types within the same sample.

Psychopathy is a multifaceted personality disorder characterized by interpersonal, affective, and behavioral dysfunction (Hare, 1991, 2003). Currently, the most widely used measure to assess psychopathic traits in adult samples is Hare’s Psychopathy Checklist – Revised (PCL-R) (Hare, 2003), due to its reliability across different samples and its predictive utility, particularly regarding violent recidivism (Bolt, Hare, Vitale, & Newman, 2004; Hemphill, Hare, & Wong, 1998). Despite the number of advantages of the PCL-R, the assessment requires extensive training, lengthy interviews, and access to institutional files to review collateral information. As such, self-report measures were developed to assess psychopathic traits in a more time efficient manner, in community samples without access to collateral information, including the Psychopathic Personality Inventory (PPI) (Lilienfeld & Widdows, 2005), Levenson Self-Report Psychopathy Scale (Levenson, Kiehl, & Fitzpatrick, 1995), Self-Report Psychopathy (SRP) Scale (Paulus, Hemphill, & Hare, 2009), and the Triarchic Psychopathy Measure (TriPM) (Patrick, Fowles, & Krueger, 2009).

In addition to adult samples, researchers have tested the applicability of the psychopathy construct in younger children and adolescent samples. The prevailing view is that intervention efforts targeted towards youth will have a better chance of altering life-course persistent problematic behavior, if such interventions are started early (Caldwell, 2011; Caldwell, McCormick, Umstead, & Van Rybroek, 2007). Similar to assessments used in adult samples, psychopathic traits in youth and adolescent samples are typically measured using either interview-based instruments or self-report measures. Psychopathic traits in youth samples can also be assessed using parent- or caregiver-report. However, antisocial attitudes and behavior in adolescents are often better assessed using self-report compared to parent- or caregiver-report (Jolliffe et al., 2003; Kamphaus & Frick, 2002). Incarcerated youth tend to have less adult supervision and typically come from families where parents have not had enough recent contact with the adolescent to provide current ratings of the child’s characteristics (Fink, Tant, Tremba, & Kiehl, 2012; Loney, Frick, Clements, Ellis, & Kerlin, 2003). Therefore, there is an increased discrepancy between caregiver-reports and self-reports in incarcerated youth (De Los Reyes & Kazdin, 2005), which is often not apparent in other populations, such as youth who meet criteria for attention-deficit/hyperactivity disorder (ADHD) (Bied, Biederman, & Faraone, 2017; McLoughlin, Rijsdijk, Asherson, & Kuntsi, 2011).

It appears as if interview-based and self-report measures of psychopathic traits should not be used interchangeably. For example, self-report measures of adolescent psychopathic traits have shown poor classification agreement across measures (Cauffman, Kimonis, Dmitrieva, & Monahan, 2009; Fink et al., 2012; Lee, Vincent, Hart, & Corrado, 2003; Skeem & Cauffman, 2003). Self-report measures of adolescent psychopathic traits have also shown to be poorer predictors of delinquency and antisocial indices compared to the interview-based Hare Psychopathy Checklist: Youth Version (PCL:YV) (Forth, Kosson, & Hare, 2003), a downward extension of the PCL-R modified for age appropriateness (Boccaccini et al., 2007; Cauffman et al., 2009; Douglas, Epstein, & Poythress, 2008; Fink et al., 2012; Sharp & Kine, 2008; Spain, Douglas, Poythress, & Epstein, 2004). Interestingly, a few studies to date have been performed showing dissimilar executive functioning (Baskin-Sommers et al., 2015) and functional neuroimaging (Harenski, Harenski, & Kiehl, 2014) deficits when assessing psychopathic traits using both interview-based and self-report measures within the same sample. This suggests that self-report measures of adolescent psychopathic traits do not provide a compatible assessment of psychopathic traits measured via interview-based instruments, suggesting further refinement of such tools may be warranted (Fink et al., 2012).

In a recent publication, we found that PCL:YV Total and Facet 4 scores were associated with reduced amplitude of the error-related positivity (Pe) event-related potential (ERP) component (Maurer, Steele, Cope, et al., 2016). Reduced Pe amplitude suggests that youth with elevated psychopathic traits assessed via the PCL:YV can detect when an error has occurred, but exhibit specific dysfunction in post-error related processing (Brazil et al., 2009; Maurer, Steele, Edwards, et al., 2016), including processing the motivational (Ullsperger, Harsay, Wessel, & Ridderinkhof, 2010) or affective (Overbeek, Nieuwenhuis, & Ridderinkhof, 2005) appraisal of such stimuli. Results were most strongly supported through the use of principal component analysis (PCA), which provides a robust decomposition of overlapping variance between and within ERP components. This approach has been incorporated in several reports, providing a more sensitive and predictive measure compared to traditional time-domain ERP analyses (Anderson, Steele, Maurer, Bernat, & Kiehl, 2015; Fink et al., 2016; Maurer, Steele, Edwards, et al., 2016; Steele et al., 2015; Steele et al., 2014; Steele, Maurer, Bernat, Calhoun, & Kiehl, 2016). Dysfunctional Pe amplitude has also been found in adults scoring high on the PCL-R (Brazil et al., 2009; Maurer, Steele, Edwards, et al., 2016; Steele, Maurer, et al., 2016).

Following up on previously published reports in which disparate results were obtained depending on the specific psychopathy instrument used within the same sample, particularly neuroimaging results (Harenski et al., 2014), in the current report, we sought to investigate whether self-report measures of adolescent psychopathic traits would be negatively related to Pe amplitude. Based on previous evidence that self-report measures do not provide an interchangeable assessment of psychopathic traits with the PCL:YV (Fink et al., 2012), we hypothesized self-report measures of adolescent psychopathic traits would be associated with unique error-related electrophysiological deficits. In the current report, we investigated the following four self-report adolescent psychopathy measures: the Antisocial Process Screening Device (APSD) (Frick & Hare, 2001), Child Psychopathy Scale (CPS) (Lynam, 1997a), Inventory of Callous-Unemotional Traits (ICU) (Essau, Sasagawa, & Frick, 2006), and the Youth Psychopathic Traits Inventory (ICU) (Andershed, Kerr, Stattin, & Levander, 2002).

Factor analyses of these varying self-report assessments yield different factors, ranging in their ability to measure interpersonal/affective traits and lifestyle/antisocial traits. The ICU focuses strictly on affective deficits associated with youth with elevated psychopathic traits, whereas the CPS assesses interpersonal/affective traits, in addition to severe behavioral problems that arise early in life. On the other hand, the APSD and YPI both assess interpersonal/affective and lifestyle traits, but largely ignore early criminogenic behavior. As such, we hypothesized that individual self-report measures would be associated with unique error-related cognitive dysfunction using traditional time-domain ERP and PCA analyses. Specifically, we hypothesized CPS scores would be negatively related to Pe amplitude, as this instrument measures severe behavioral dysfunction like the PCL:YV, whereas APSD and YPI scores should be associated with normal Pe amplitude, as these assessments do not measure behavioral dysfunction arising early in life. However, the APSD and YPI do measure lifestyle traits, including impulsivity. Populations scoring higher on measures of impulsivity typically exhibit reduced error-related negativity (ERN/Ne) amplitude, related to initial, automatic error-detection and action-monitoring processes (Dikman & Allen, 2000; Hall, Bernat, & Patrick, 2007; Heritage & Benning, 2013; Pasion & Barbosa, 2016). Thus, we hypothesized APSD and YPI scores would be negatively related to ERN/Ne amplitude, but unrelated to the Pe. We did not make any a priori hypotheses regarding ICU scores.

Method

Participants

Participants included n = 142 incarcerated adolescent offenders recruited from a maximum security juvenile correctional facility who participated in a larger study (Southwest Advanced Neuroimaging Cohort – Youth (SWANC-Y)). Participants were excluded from analyses for meeting the following criteria: previous history of traumatic brain injury accompanied with a significant loss of consciousness (n = 4), significant movement during data collection, or behavioral performance (i.e., making less than four errors) (n = 16). Reliability analyses suggest that the ERN/Ne and Pe can be quantified in as few as four to six trials (Olvet & Hajcak, 2009; Pontifex et al., 2010; Steele, Anderson, et al., 2016). Participants were also excluded for meeting criteria for mood disorders, including major depression (n = 10) and anxiety disorders including post-traumatic stress disorder (PTSD) (n = 3), due to the important role these disorders play for both the ERN/Ne (Chiu & Deldin, 2007; Olvet & Hajcak, 2008) and Pe (Bridwell, Steele, Maurer, Kiehl, & Calhoun, 2015) amplitude. Finally, female participants (n = 9) were excluded from final analyses, as there were not enough participants to power gender effects. This resulted in a final sample of n = 100 incarcerated male offenders, ranging from 16 to 20 years of age (M = 17.38 years, SD = 0.86) at the time of electroencephalography (EEG) collection. The sample was predominantly right-handed, with 7% of the sample reporting left-hand dominance. Participants largely self-identified as Hispanic/Latino (76%), with the remaining self-identifying as Black/African American (12%), White (10%), or more than one ethnic category (2%). Initial contact was made with potential study participants through announcements made by research staff at the correctional facility. Meetings were scheduled with interested participants and informed consent was obtained. Individuals 18 years of age or older provided written informed consent, and individuals younger than 18 years of age provided written informed assent in conjunction with parent/guardian consent. Participants were informed of their right to terminate participation at any point, the lack of direct institutional benefits resulting from their participation in the study, and that their participation would not affect their facility status or parole. Participants received remuneration at the hourly labor wage of the facility. The University of New Mexico Health Center Human Research Review Committee and the Office of the Human Research Protections approved all procedures. Important to note, the relationship between PCL:YV scores and error-related ERPs has been previously published incorporating this same sample of n = 100 incarcerated male offenders (Maurer, Steele, Cope, et al., 2016). In the current report, we sought to investigate whether self-report measures of adolescent psychopathic traits would be negatively related to Pe amplitude, as PCL:YV Total and Facet 4 scores were in our previous investigation.

Self-Report Adolescent Psychopathy Measures

Antisocial Process Screening Device (APSD)

The APSD (Frick & Hare, 2001) is a 20-item self-report measure designed to assess behavior similar to the adult construct of psychopathy as measured via the PCL-R (Hare, 2003). Each item is scored either a 0 (not at all), 1 (sometimes true), or 2 (definitely true). Factor analysis of the APSD reveals three dimensions: Narcissism (Factor 1), Impulsivity (Factor 2), and Callous-Unemotional (Factor 3) traits (Frick, Bodin, & Barry, 2000). The mean APSD score for the sample was 16.14 (SD = 5.80); APSD scores were unavailable for n = 3 participants. The Cronbach’s alpha for the APSD (all items) was .68, .72 for the Narcissism subscale, .61 for the Impulsivity subscale, and .43 for the Callous-Unemotional subscale. The Chronbach’s alpha for the APSD Total score in the current sample is lower than those previously reported (typically ranging from .72 to .82) (Lee et al., 2003; Murrie & Cornell, 2002; Pardini, Lochman, & Frick, 2003), while falling in similar ranges for the subscales (.56 - .72 for the Narcissism subscale, .44 to .60 for the impulsivity subscale, and .36 - .56 for the Callous-Unemotional subscale) (see (Douglas et al., 2008).

Child Psychopathy Scale (CPS)

The CPS (Lynam, 1997a) is a 50-item self-report measure adapted for youth to assess 12 of the 20 items measured by the PCL-R. Each item is scored either a 0 (No) or 1 (Yes). Factor analysis of the CPS reveals two subscales: Callous-Unemotional traits (Factor 1) and Antisocial Behavior (Factor 2). (Lynam et al., 2005). The mean CPS Total Score for the sample was 18.84 (SD = 7.38). The Cronbach’s alpha for the CPS (all items) was .55, .54 for the Callous-Unemotional subscale, and .50 for the Antisocial Behavior subscale, which is lower than what has been reported in other studies (alpha for Total Scores previously ranged from .84 to .91) (Fink et al., 2012; Lynam, 1997a).

Inventory of Callous and Unemotional Traits (ICU)

The ICU (Essau et al., 2006) is a 24-item self-report measure developed from the Callous-Unemotional subscale of the APSD. The ICU was developed to overcome some of the limitations associated with the APSD, including moderate internal consistency due to the smaller number of items (Munoz & Frick, 2007). Each item is scored on a four-point Likert scale from 0 (not at all true) to 3 (definitely true). Factor analysis of the ICU reveals three factors: Callousness (Factor 1), Uncaring (Factor 2), and Unemotional (Factor 3) traits (Kimonis et al., 2008). The mean ICU Total Score for the sample was 28.90 (SD = 7.79). The Cronbach’s alpha for the ICU (all items) was .64, .85 for the Callousness subscale, .78 for the Uncaring subscale, and .36 for the Unemotional subscale. Cronbach’s alpha values typically range from .77 - .93 for ICU total score, .59 - .88 for the Callousness scale, .55 - .87 for the Unemotional scale, and .47 - .87 for the Unemotional subscale (Byrd, Kahn, & Pardini, 2013; Ciucci, Baroncelli, Franchi, Golmaryami, & Frick, 2014; Essau et al., 2006; Ezpeleta, de la Osa, Granero, Penelo, & Domenech, 2014; Fanti, Frick, & Georgiou, 2009; Houghton, Hunter, & Crow, 2013; Kimonis et al., 2008; Pechorro, Ray, Barroso, Maraco, & Gonçalves, 2016; Waller et al., 2015).

Youth Psychopathic Traits Inventory (YPI)

The YPI (Andershed et al., 2002) is a 50-item self-report measure designed to measure core features of psychopathy in youth. Each item is scored on a four-point Likert scale ranging from 0 (does not apply at all) to 3 (applies very well). The fifty items of the YPI form ten subscales, with factor analyses showing these ten subscales represent three separate factors: Grandiose-Manipulative (Factor 1), Callous-Unemotional (Factor 2), and Impulsive-Irresponsible (Factor 3) traits (Andershed et al., 2002). The mean YPI Total Score for the sample was 117.03 (SD = 20.94); YPI scores were unavailable for n = 4 participants retained in analyses. The Cronbach’s alpha for the YPI (all items) was .93, .92 for the Grandiose-Manipulative subscale, .76 for the Callous-Unemotional subscale, and .87 for the Impulsive-Irresponsible subscale, which is consistent with previously published reports (Fink et al., 2016; Pechorro, Dibeiro da Silva, Rijo, Gonçalves, & Andershed, 2017). See Table 1 for the remainder of descriptive statistics for self-report measures of adolescent psychopathic traits.

Table 1.

Descriptive statistics for all participants (n = 100)

Variable	n	Mean	SD	Range
APSD Total	97	16.14	5.80	1–36
APSD Factor 1	97	4.39	2.62	0–13
APSD Factor 2	97	5.14	1.92	0–10
APSD Factor 3	97	5.62	1.69	3–9
CPS Total	100	18.84	7.38	2–42
CPS Factor 1	100	10.87	4.29	3–23
CPS Factor 2	100	8.62	4.04	0–19
ICU Total	100	28.90	7.79	15–54
ICU Factor 1	100	9.66	3.89	3–27
ICU Factor 2	100	11.41	3.52	2–21
ICU Factor 3	100	7.47	2.44	2–14
YPI Total	96	117.03	20.94	60–175
YPI Factor 1	96	41.61	10.64	20–68
YPI Factor 2	96	33.98	6.29	18–52
YPI Factor 3	96	41.44	7.52	17–58

Note. Assessments: APSD Total, Factor 1 (Narcissism), Factor 2 (Impulsivity), and Factor 3 (Callous-Unemotional) scores are derived from the Antisocial Process Screening Device (APSD) (Frick & Hare, 2001); CPS Total and Factor 1 (Callous-Unemotional) and Factor 2 (Antisocial Behavior) scores are derived from the Child Psychopathy Scale (CPS) (Lynam, 1997); ICU Total and Factor 1 (Callousness), Factor 2 (Uncaring), and Factor 3 (Unemotional) scores are derived from the Inventory of Callous and Unemotional Traits (ICU) (Frick et al., 2003); YPI Total and Factor 1 (Grandiose-Manipulative), Factor 2 (Callous-Unemotional), and Factor 3 (Impulsive-Irresponsible) scores are derived from the Youth Psychopathic Traits Inventory (YPI) (Andershed et al., 2002).

Remaining Assessments

Psychopathology

Psychopathology, including major depression and anxiety disorders, was assessed using the Kiddie Schedule for Affective Disorders and Schizophrenia (K-SADS) (Kaufman, Birmaher, & Brent, 1997) by trained research staff. IQ was estimated using the Vocabulary and Matrix Reasoning sub-tests of the Wechsler Adult Intelligence Scale – Third Version (WAIS-III) (Wechsler, 1997) for participants 16 years of age or older and from the Wechsler Intelligence Scale for Children – Fourth Version (WISC-IV) (Wechsler, 2003) for participants younger than 16 years of age.

Stimuli and Task

EEG was collected in a small room separate from the general population housing at the correctional facility. After placement of electrodes, participants were seated in a comfortable chair 60 cm away from a computer monitor on which the task stimuli were presented and were instructed to refrain from excessive blinking and movement during data collection. Participants then performed a response inhibition Go/NoGo task (Kiehl, Liddle, & Hopfinger, 2000). Stimuli were presented to participants using the Neurobehavioral Systems Inc. visual software package, Presentation (www.neurobs.com). Each stimulus appeared for 250 ms in white text within a continuously displayed rectangular fixation box against a black background. Participants were instructed to respond as quickly and accurately as possible with their right index finger every time the target Go stimulus (a white “X”) appeared, and to withhold a response whenever the distracter NoGo stimulus (a white “K”) appeared. Targets appeared with higher frequency (84%, 412 trials, with 206 on each run) than distracters (16%, 78 trials, with 39 on each run) to establish a strong stimulus-response mapping on Go trials. NoGo stimuli were never presented sequentially. The inter-stimulus interval was pseudo-randomly jittered (1–3 seconds stimulus onset asynchrony [SOA] averaging 1.5 seconds). The SOA between Go stimuli varied to the constraint that three Go stimuli were presented within each 6 second period. The NoGo stimuli were interspersed among the Go stimuli in a pseudo-random manner subject to two constraints: the minimum SOA between Go and NoGo stimuli was 1000 ms and the SOA between successive NoGo stimuli was in the range of 8 to 14 seconds. Hits were defined as successful responses to Go stimuli, whereas False Alarms (FA’s) were defined as incorrect responses to NoGo stimuli. Prior to recording, each participant performed a block of ten practice trials to ensure task instructions were clearly understood.

EEG Recordings

EEG data were collected using two computers and a 64-channel BioSemi amplifier. The first computer used Presentation software to deliver the stimuli, accept responses, and send digital triggers to the EEG acquisition computer when a stimulus or response occurred. The second computer acquired EEG data using BioSemi software and amplifiers. All signals collected with the BioSemi software were low-pass filtered using a fifth-order sinc filter with a half-power cutoff of 204.8 Hz, then digitized to 1024 Hz during data collection. EEG activity was recorded using sintered Ag-AgCl active electrodes placed in accordance with the 10-20 International System (Jasper, 1958). The participant’s nose was used as a reference. Six electrodes were placed on the participant’s face above, below, and lateral to the canthus of each eye to measure electrooculogram. All offsets were kept below 10 kΩ.

Analytic Strategy

Pre-processing included down-sampling to 512 Hz, bad channel detection and replacement, epoching, eye-blink removal, and low-pass filtering to 15 Hz. Bad channels were identified as having activity four standard deviations away from the mean of the surrounding electrodes. ERP epochs were defined relative to the response, from 1000 ms pre- to 2000 ms post-response. An independent component analysis (ICA) eye-blink removal protocol was also performed. The ICA utility in EEGlab software (Delorme & Makeig, 2004) was used to derive components; then, using an in-house template matching algorithm (Jung, Makeig, Westerfield, Courschesne, & Sejnowski, 2000), blink components were identified and removed from the data.

Classic time-domain response-locked ERP components relatives to FA’s were extracted: the ERN/Ne, the mean amplitude of the negative deflection occurring 0 – 100 ms and the Pe, the mean amplitude of the positive deflection occurring 94 – 500 ms post-response. Response-locked components were baseline corrected with a −200 to −50 ms window relative to FA’s. Within each trial, individual electrodes with activity ± 100 μV were omitted from analyses. An additional data reduction method, PCA, was also performed on the response-locked data (Chapman & McCarry, 1995). Temporal PCA with varimax rotation was carried out on the covariance matrix from all scale electrodes to define a four component response-locked solution (PC1, PC2, PC3, and PC4) for FA trials accounting for 94.97% of the variance. A subset of nine electrodes representing maximal time-domain component activation was selected for both the ERN/Ne and Pe (AF3, AFz, AF4, F3, Fz, F4, FC3, FCz, and FC4) and was used in subsequent analyses.

To see which time-domain ERP components the individual principal components best represented, multiple regression analyses were performed. The time-domain ERP component (i.e., either the ERN/Ne or Pe) was entered as the dependent measure, and the four principal components were entered as predictor variables. The time-domain ERN/Ne ERP component was best predicted by the third principal component (PC3) (β = .91, p < .001), and the time-domain Pe ERP component was best predicted by the first principal component (PC1) (β = .29, p < .001), the second principal component (PC2) (β = .48, p < .001), the third principal component (PC3) (β = .12, p = .001), and the fourth principal component (PC4) (β = .27, p < .001). Although PC3 shared variance with both the ERN/Ne and Pe time-domain components, this subcomponent temporally reflects the ERN/Ne. The three remaining PCA-derived subcomponents reflect individual subcomponents underlying the time-domain Pe, with PC1 reflecting an early subcomponent of the Pe, PC2 reflecting a middle subcomponent of the Pe, and PC4 reflecting a late subcomponent of the Pe. These three subcomponents are thought to reflect distinct patterns of cognitive processing (Edwards, Calhoun, & Kiehl, 2012).

Results

Behavioral Results

Response times (RTs) and frequency for Hits and FA’s were analyzed and have been previously reported (Maurer, Steele, Cope, et al., 2016). As expected, participants responded faster to NoGo stimuli (M = 381 ms, SD = 43 ms) than Go stimuli (M = 419 ms, SD = 51 ms), t(99) = 7.74, p < .001. Participants made significantly more errors (i.e., FA’s) to NoGo stimuli (M = 23.76, SD = 11.85) compared to Go stimuli (M = 12.61, SD = 14.65), t(99) = 17.89, p < .001. For the full sample, there was a main effect of post-error slowing (PES) (M = 28 ms, SD = 73 ms) (Rabbit, 1981). Participants responded more slowly after error trials (M = 384 ms, SD = 84 ms) than after correct trials (M = 356 ms, SD = 33 ms), t(99) = 3.86, p < .001.

Correlation Analyses

Self-report adolescent psychopathy scores (both Total and Factor scores) were significantly correlated with PCL:YV scores previously reported (Maurer, Steele, Cope et al., 2016) (see Table 2). However, scores on the four self-report psychopathy assessments were not significantly correlated with time-domain ERN/Ne or Pe mean amplitude, or PCA subcomponents reflecting these ERP components (p-values ranging from .20 to .93) (see Table 3). In addition, the time-domain ERN/Ne was significantly correlated with the time-domain Pe (r = .53, p < .001), as well as the individual subcomponents underlying the Pe in PC1 (r = .31, p = .002), PC2 (r = .24, p = .017), and PC4 (r = .39, p < .001). As expected, the time-domain ERN/Ne was significantly correlated with PC3 (r = .87, p < .001). PES did not significantly correlate with adolescent psychopathy scores or time-domain ERP or PCA measures reflecting ERN/Ne or Pe mean amplitude. In addition, adolescent psychopathy variables were not significantly correlated with response times or error rates.

Table 2.

Correlations between Psychopathy Measures

	PCL:YV Total	PCL:YV Factor 1	PCL:YV Factor 2	PCL:YV Facet 1	PCL:YV Facet 2	PCL:YV Facet 3	PCL:YV Facet 4
APSD Total	.509**	.416**	.472**	.258*	.445**	.463**	.368**
APSD Factor 1	.388**	.369**	.316**	.258*	.364**	.363**	.174
APSD Factor 2	.393**	.295**	.412**	.251*	.247*	.376**	.363**
APSD Factor 3	.380**	.313**	.306**	.183	.346**	.296**	.277**
CPS Total	.384**	.281**	.401**	.138	.338**	.396**	.312**
CPS Factor 1	.345**	.301**	.302**	.177	.332**	.318**	.217**
CPS Factor 2	.324**	.175	.410**	.056	.240*	.382**	.341**
ICU Total	.310**	.283**	.275**	.139	.341**	.237*	.247*
ICU Factor 1	.118	.137	.082	.065	.167	.119	.022
ICU Factor 2	.355**	.289**	.352**	.081	.409**	.299**	.302**
ICU Factor 3	.065	.123	−.009	.216*	−.011	−.067	.065
YPI Total	.281**	.253*	.255*	.217*	.203*	.181	.265**
YPI Factor 1	.187	.194	.124	.192	.130	.119	.088
YPI Factor 2	.205*	.217*	.187	.201	.158	.075	.240*
YPI Factor 3	.347**	.249*	.379**	.164	.249*	.273**	414**

Note. Assessments: PCL:YV scores have been reported in our previous manuscript (Maurer et al., 2016), and are reported here to compare to self-report measures of adolescent psychopathic traits. PCL:YV Total, Factor 1 (Interpersonal/Affective), Factor 2 (Lifestyle/Behavioral), and Facet 1 (Interpersonal), Facet 2 (Affective), Facet 3 (Lifestyle), and Facet 4 (Antisocial) scores are derived from the Hare Psychopathy Checklist: Youth Version (PCL:YV) (Forth et al., 1983); APSD Total, Factor 1 (Narcissism), Factor 2 (Impulsivity), and Factor 3 (Callous-Unemotional) scores are derived from the Antisocial Process Screening Device (APSD) (Frick & Hare, 2001); CPS Total and Factor 1 (Callous-Unemotional) and Factor 2 (Antisocial Behavior) scores are derived from the Child Psychopathy Scale (CPS) (Lynam, 1997); ICU Total and Factor 1 (Callousness), Factor 2 (Uncaring), and Factor 3 (Unemotional) scores are derived from the Inventory of Callous and Unemotional Traits (ICU) (Frick et al., 2003); YPI Total and Factor 1 (Grandiose-Manipulative), Factor 2 (Callous-Unemotional), and Factor 3 (Impulsive-Irresponsible) scores are derived from the Youth Psychopathic Traits Inventory (YPI) (Andershed et al., 2002)

^**p < .01

^*p < .05

Table 3.

Correlations between Psychopathy Measures and Time-Domain and Principal Component Variables

	ERN/Ne	Pe	PC1	PC2	PC3	PC4
PCL:YV Total	.066	−.144	−.054	−.228*	.090	−.110
PCL:YV Factor 1	.095	−.015	.114	−.112	.116	−.057
PCL:YV Factor 2	.047	−.169	−.113	−.239*	.056	−.099
PCL:YV Facet 1	.108	.078	.176	−.032	.129	.048
PCL:YV Facet 2	.052	−.106	.014	−.158	.066	−.147
PCL:YV Facet 3	.050	−.113	−.063	−.178	.083	−.020
PCL:YV Facet 4	.007	−.243*	−.202*	−.293**	−.003	−.197*
APSD Total	−.047	−.051	.027	−.104	−.050	.011
APSD Factor 1	−.013	.006	.116	−.042	.012	.034
APSD Factor 2	−.122	−.131	−.076	−.167	−.121	−.045
APSD Factor 3	−.040	−.049	−.070	−.104	−.063	.048
CPS Total	−.118	−.115	−.052	−.113	−.130	−.071
CPS Factor 1	−.151	−.057	.007	−.040	−.164	−.025
CPS Factor 2	−.051	−.144	−.085	−.161	−.050	−.095
ICU Total	.047	.056	.009	.020	−.020	.059
ICU Factor 1	.021	.092	.093	.087	−.010	.052
ICU Factor 2	.017	.058	.073	.014	−.013	.089
ICU Factor 3	.103	.006	−.078	−.046	.080	.056
YPI Total	−.046	−.019	−.012	−.013	−.044	.037
YPI Factor 1	−.017	.028	.060	.022	−.021	.043
YPI Factor 2	−.052	.007	−.011	.006	−.042	.075
YPI Factor 3	−.061	−.100	−.109	−.074	−.058	−.020

Note. PCL:YV scores have been reported in our previous manuscript (Maurer et al., 2016), and are reported here to compare to self-report measures of adolescent psychopathic traits. ERN and Pe reflect the ERN/Ne and Pe time-domain ERP components; PC1 reflects the early subcomponent of the Pe ERP component; PC2 reflects the bottom subcomponent of the Pe ERP component; PC3 reflects the subcomponent reflecting the ERN/Ne ERP component; PC4 reflects the late subcomponent of the Pe ERP component. Assessments: PCL:YV Total, Factor 1 (Interpersonal/Affective), Factor 2 (Lifestyle/Behavioral), and Facet 1 (Interpersonal), Facet 2 (Affective), Facet 3 (Lifestyle), and Facet 4 (Antisocial) scores are derived from the Hare Psychopathy Checklist: Youth Version (PCL:YV) (Forth et al., 2003); APSD Total and Factor scores are derived from the Antisocial Process Screening Device (APSD) Total Score and Narcissism, Impulsivity, and Callous-Unemotional factors (Frick & Hare, 2001); CPS Total and Factor scores are derived from the Child Psychopathy Scale (CPS) Total Score and Callous-Unemotional and Antisocial Behavior factors (Lynam, 1997); ICU Total and Factor scores are derived from the Inventory of Callous and Unemotional Traits (ICU) Total Score and Callousness, Uncaring, and Unemotional Factors (Frick et al., 2003); YPI Total and Factor scores are derived from the Youth Psychopathic Traits Inventory (YPI) Total Score and Grandiose-Manipulative, Callous-Unemotional, and Impulsive-Irresponsible factors (Andershed et al., 2002).

^**p < .01

^*p < .05

Time-Domain ERP Multiple Regression Analyses

Here, we performed multiple regression analyses incorporating our full sample (n = 100) to see if self-report adolescent psychopathy scores (both Total and Factor scores) were significant predictors of either the time-domain ERN/Ne or Pe ERP components. Each of the multiple regression analyses performed had an ERP component as the dependent measure (i.e., either the time-domain ERN/Ne or Pe). Self-report adolescent psychopathy scores (either Total or Factor scores for each measure) were entered as simultaneous predictor variables.

In total, sixteen separate multiple regression analyses were performed (eight for each of the two time-domain ERP components). Regressions 1 & 2 included APSD Total or Factor (Narcissism, Impulsivity, and Callous-Unemotional) scores, Regressions 3 & 4 included CPS Total or Factor (Callous-Unemotional and Antisocial Behavior) scores, Regressions 5 & 6 included ICU Total or Factor (Callousness, Uncaring, and Unemotional) scores, and Regressions 7 & 8 included YPI Total or Factor (Grandiose-Manipulative, Callous-Unemotional, and Impulsive-Irresponsible) scores. After implementation of a Simes-Hochberg multiple comparison procedure (Hochberg, 1988; Simes, 1986), none of the self-report adolescent psychopathy scores were significant predictors of either the time-domain ERN/Ne or Pe mean amplitude in the sixteen multiple regression analyses performed. The Simes-Hochberg multiple comparison procedure is designed to maintain an acceptable family wise error rate. This correction is in the class of sequential Bonferroni correction methods, which consists of arranging the obtained p-values within a family of tests from largest to smallest, and excluding tests on a sequential basis on whether they are associated with a p-value that is less than a previously adjusted alpha level.

PCA Multiple Regression Analyses

Additional multiple regression analyses were performed to assess the amount of variance in four PCA-derived subcomponents (one component reflecting the time-domain ERN/Ne [PC3] and three components reflecting early, middle, and late subcomponents underlying the Pe [PC1, PC2, and PC4, respectively]) explained by self-report adolescent psychopathy scores. Similar to the time-domain multiple regression analyses, self-report psychopathy scores were entered as simultaneous predictor variables in 32 separate regression analyses (8 separate regressions for each of the four self-report measures) for each of the four PCA components. After implementation of the Simes-Hochberg multiple comparison procedure, self-report psychopathy scores were not significant predictors of PC3 mean amplitude, reflecting the time-domain ERN/Ne, or PC1, PC2, or PC4 mean amplitude, reflecting early, middle, and late subcomponents underlying the time-domain Pe. However, APSD Factor 1 (Narcissism) scores were a moderate predictor of increased PC1 amplitude, reflecting an early subcomponent of the Pe (β = .221, p = .067). See Figures 1–4 for non-significant group differences in respect to the ERN/Ne and Pe amplitude, and the supplemental material for full statistics for all multiple regression analyses performed.

Figure 1.

Response-locked event-related potential (ERP) analyses. (Left) Representative ERP waveform plotted at Fz for each group with negative voltage plotted up. Youth scoring high on the APSD (APSD Total score ≥ 20, n = 25) (red) and youth scoring low on the APSD (APSD Total score ≤ 13, n = 25) (blue) are plotted. ERP components of interest (the error-related negativity [ERN/Ne] and error-related positivity [Pe]) are identified. (Right) Statistical (black and white) maps are plotted for each component window highlighting non-significant group differences between high- and low-scoring groups.

Figure 4.

Response-locked event-related potential (ERP) analyses. (Left) Representative ERP waveform plotted at Fz for each group with negative voltage plotted up. Youth scoring high on the YPI (YPI Total score ≥ 130, n = 25) (red) and youth scoring low on the YPI (YPI Total score ≤ 104, n = 25) (blue) are plotted. ERP components of interest (the error-related negativity [ERN/Ne] and error-related positivity [Pe]) are identified. (Right) Statistical (black and white) maps are plotted for each component window highlighting non-significant group differences between high- and low-scoring groups.

Discussion

Psychopathic traits in youth samples are typically measured using either interview-based or self-report measures. Evidence suggests however, that these assessments should not be used interchangeably, as evidenced by poor classification agreement between measures (Cauffman et al., 2009; Fink et al., 2012; Lee et al., 2003; Skeem & Cauffman, 2003). Additionally, self-report measures of adolescent psychopathic traits show poor predictive utility of delinquency and antisocial indices compared to the interview-based PCL:YV, likely due to self-report measures typically excluding items indexing antisocial behavior in order to reduce potential criterion contamination (Boccaccini et al., 2007; Cauffman et al., 2009; Douglas et al., 2008; Fink et al., 2012; Sharp & Kine, 2008; Spain et al., 2004). However, little research to date has incorporated the use of both assessment modalities within the same sample, to examine whether similar types of dysfunction are observed between measures. To date, only a few studies have been performed, showing dissimilar executive functioning (Baskin-Sommers et al., 2015) and functional neuroimaging (Harenski et al., 2014) deficits between self-report and interview-based measures within the same sample. To date, no study has investigated whether both assessment types are associated with dysfunction measured with event-related potentials within the same sample.

In a recent publication, we found that PCL:YV Total and Facet 4 scores were negatively related to amplitude of the Pe ERP component (Maurer, Steele, Cope, et al., 2016). In the current report, we incorporated the use of four different self-report measures of adolescent psychopathic traits to examine whether scores on these assessments would similarly be negatively associated with Pe amplitude. However, in the current report, self-report measures of adolescent psychopathic traits (assessed via total and factor scores from the APSD, CPS, ICU, and YPI) were not significantly associated with dysfunction with either the ERN/Ne or Pe using traditional time-domain ERP or PCA analyses. The findings obtained in the current investigation support a growing body of literature suggesting interview-based and self-report measures of psychopathic traits are not associated with similar forms of cognitive dysfunction within the same sample (Baskin-Sommers et al., 2015; Harenski et al., 2014).

Results obtained in the current report support previous research showing poor classification agreement between self-report measures of adolescent psychopathic traits and the interview-based PCL:YV (Cauffman et al., 2009; Fink et al., 2012; Lee et al., 2003; Skeem & Cauffman, 2003). Youth scoring high on the PCL:YV are not necessarily the same individuals who score high on self-report measures of adolescent psychopathic traits (Fink et al., 2012), which may partially explain the disparate results obtained in the current investigation and our previous report (Maurer, Steele, Cope, et al., 2016). In fact, in the current report, there was very little overlap regarding participants who scored high on both the PCL:YV and four different self-report measures of adolescent psychopathic traits. In our previous publication, n = 21 participants scored ≥ 30 on the PCL:YV, reflecting the top quartile of scorers (Maurer, Steele, Cope et al., 2016). In the current report, there was an overlap of n = 7 who scored in the top quartile of the APSD and PCL:YV (33% overlap), n = 9 who scored in the top quartile of the CPS and PCL:YV (43% overlap), n = 9 who scored in the top quartile of the ICU and PCL:YV (43% overlap), and n = 7 who scored in the top quartile of the YPI and PCL:YV (33% overlap). Observing dissimilar results between interview-based and self-report measures of adolescent psychopathic traits with respect to the Pe mean amplitude appears to be strongly related to the fact that participants who scored high on the PCL:YV are not necessarily the same individuals who scored high on the APSD, CPS, ICU, and YPI.

In our previous report, PCL:YV Total and Facet 4 scores were negatively related to Pe mean amplitude (Maurer, Steele, Cope, et al., 2016). PCL:YV Facet 4 includes traits that are largely underemphasized in self-report measures of adolescent psychopathic traits. These assessments tend to be largely used in non-incarcerated community samples, where criminogenic behavioral traits occur at substantially lower levels compared to incarcerated settings (Neumann & Hare, 2008). This antisocial facet of the PCL:YV includes traits directly tapping into the early onset of psychopathic traits, including early behavioral problems before the age of 12 (Forth, Hart, & Hare, 1990; Neumann, Wampler, Taylor, Blonigen, & lacono, 2011). The emergence of early antisocial behavior in childhood is one important factor for the future development of psychopathy in adulthood (Frick et al., 2003). Some remain apprehensive over the inclusion of this antisocial facet within the superordinate construct of psychopathy, believing criminogenic and antisocial behavior to be a consequence, rather than a foundation of psychopathic traits (Cooke & Michie, 2001; Skeem & Cooke, 2010). Others argue eliminating this antisocial facet may result in a considerable narrowing of the psychopathy construct, particularly in regards to the developmental course of psychopathic traits (Hare & Neumann, 2010; Lynam, 1997a; Neumann et al., 2011). Thus, it is interesting to note that the discounting of these criminogenic behavioral items appears to be another potential reason for the incongruent results obtained in the current study and our previously published study (Maurer, Steele, Cope, et al., 2016).

Despite the number of advantages of assessing psychopathic traits via interview-based measures such as the PCL:YV, these measures suffer from the need for extensive training, lengthy interviews, and access to institutional files to review collateral information. Self-report measures of adolescent psychopathic traits were developed to assess such traits in a more time efficient manner, typically in non-incarcerated community settings. However, the use of these self-report measures to assess psychopathic traits has been criticized for a number of reasons, including that several features central to the construct of psychopathy, including grandiose sense of self-worth and shallow affect, are likely to result in inaccurate descriptions of one’s own behavior (Miller & Lynam, 2011). This is made more difficult in youth samples, as many psychopathic traits, including a lack of realistic, long-term goals, need for stimulation, and impulsivity are normative features of typical adolescent development (Bongers, Koot, van der Ende, & Verhulst, 2003). In general community samples, where the vast majority of individuals score low on measures of psychopathic traits (Neumann & Hare, 2008), self-report measures of adolescent psychopathic traits appear to be more appropriate measures of psychopathic traits than when used in incarcerated settings. The use of self-report measures of adolescent psychopathic traits in community samples have identified many abnormalities consistent with incarcerated adult psychopathic offenders (Budhani & Blair, 2005; Lockwood et al., 2013; Viding et al., 2012; White et al., 2012).

Limitations

Several limitations of this study should be considered in evaluating the generalizability of these findings. First, the current report did not incorporate the use of caregiver-report measures, instead solely relying on self-report measures of adolescent psychopathic traits. The response rates on caregiver-report measures in the current report were too low to be included. Assessing psychopathic traits via caregiver-report in incarcerated settings is often an unrealistic method give the level of disruption associated with these families. In addition, incarcerated youth tend to have less adult supervision and typically come from families where parents have not had enough recent contact with the adolescent to provide current ratings of the child’s characteristics (Fink et al., 2012; Loney et al., 2003). As such, there is often an increased discrepancy between caregiver-reports and self-report (De Los Reyes & Kazdin, 2005). Whether our current results extend to the use of caregiver-report measures remains to be seen.

Second, our study recruited participants from a maximum-security youth correctional facility. Compared to youth recruited from community samples, incarcerated youth tend to differ on several variables, most notably, substance use history, general intelligence, and trait anxiety (Foley, 2001; Wasserman, McReynolds, Lucas, Fisher, & Santos, 2002). In addition, self-report measures of adolescent psychopathic traits appear to be better measures of psychopathic traits in general community samples, observing similar neurocognitive deficits as adult psychopathic offenders (Budhani & Blair, 2005; Lockwood et al., 2013; Viding et al., 2012; White et al., 2012). Perhaps then, self-report measures of adolescent psychopathic traits may be associated with dysfunctional electrophysiological measures of cognitive control, including error-related processing deficits, in general community samples.

Finally, in line with our second limitation, some of the self-report adolescent psychopathy measures used in the current report had lower Cronbach alpha values than those previously reported in other studies. Self-report measures of adolescent psychopathic traits were designed to measure such traits in non-incarcerated samples, without having to rely on reviews of collateral information, and focus primarily on personality-oriented deficits, rather than indexing criminogenic behavior. As such, measurement of psychopathic traits via self-report measures in incarcerated settings may inappropriately inflate the presence of psychopathic traits in youth with otherwise normative development or representative conduct problems (Edens, Skeem, Cruise, & Cauffman, 2001).

Conclusions

In sum, scores on measures of adolescent psychopathic traits (i.e., total and factor scores from the APSD, CPS, ICU, and YPI) were not associated with ERN/Ne or Pe dysfunction, measured with traditional time-domain ERP analyses or PCA. Such results are inconsistent with a previous report from our laboratory, whereby PCL:YV Total and Facet 4 scores were negatively related to Pe amplitude within the same sample of n = 100 incarcerated male adolescent offenders (Maurer, Steele, Cope, et al., 2016). The disparate results between studies is likely due to the fact that those individuals who scored high on the PCL:YV were not necessarily the same as those who scored high on the four self-report measures of adolescent psychopathic traits. In addition, the four self-report measures incorporated in the current study tend to underemphasize early criminogenic and antisocial behavior that arises early in life, unlike the interview-based PCL:YV. Consistent with previous studies performed (Baskin-Sommers et al., 2015; Harenski et al., 2014), the current study shows interview-based and self-report measures used within the same sample do not identify similar forms of cognitive dysfunction.

Figure 2.

Response-locked event-related potential (ERP) analyses. (Left) Representative ERP waveform plotted at Fz for each group with negative voltage plotted up. Youth scoring high on the CPS (CPS Total score ≥ 24, n = 25) (red) and youth scoring low on the CPS (CPS Total score ≤ 14, n = 25) (blue) are plotted. ERP components of interest (the error-related negativity [ERN/Ne] and error-related positivity [Pe]) are identified. (Right) Statistical (black and white) maps are plotted for each component window highlighting non-significant group differences between high- and low-scoring groups.

Figure 3.

Response-locked event-related potential (ERP) analyses. (Left) Representative ERP waveform plotted at Fz for each group with negative voltage plotted up. Youth scoring high on the ICU (ICU Total score ≥ 34, n = 25) (red) and youth scoring low on the ICU (ICU Total score ≤ 24, n = 25) (blue) are plotted. ERP components of interest (the error-related negativity [ERN/Ne] and error-related positivity [Pe]) are identified. (Right) Statistical (black and white) maps are plotted for each component window highlighting non-significant group differences between high- and low-scoring groups.

Highlights.

• Self-report psychopathy scores (including the Antisocial Process Screening Device, Child Psychopathy Scale, Inventory of Callous-Unemotional Traits, and Youth Psychopathic Traits Inventory) were not associated with error-related negativity (ERN/Ne) or error-related positivity (Pe) dysfunction.

• Self-report psychopathy scores were not associated with ERN/Ne or Pe dysfunction measured with traditional time-domain event-related potentials (ERPs) and Principal Component Analysis (PCA).

• Results support previous research incorporating both assessment types within the same sample, whereby dissimilar cognitive dysfunction is observed between assessment types.