Psychopathic Traits, Inhibition, and Positive and Negative Emotion: Results from an Emotional Go/No-Go Task

Abstract

Difficulty stopping unwanted or inappropriate actions (i.e., inhibitory control) is implicated in antisocial behaviors, which are common in people high in psychopathic traits. Recent research indicates that, for those with antisocial personality, inhibitory control is impaired under negative emotional contexts; however, it is unclear whether this impairment extends to persons with psychopathic traits and to impairments under positive emotional contexts. Identifying some of these distinctions can point to therapeutics that target negative emotion specifically or emotion dysregulation broadly. We sought to identify unique relationships between distinct facets of psychopathy and inhibitory control in the context of positive, negative, and neutral stimuli. Using a community sample (N=117), event-related potentials were recorded during an emotional-linguistic Go/No-Go task. Results indicated distinct cognition-emotion relationships for each psychopathy facet. Higher interpersonal facet scores related to reciprocal interference between cognition and emotion. Higher callous affect facet scores related to reduced inhibitory and emotional processing, except when stimuli were most engaging (emotional No-Go trials). Higher erratic lifestyle facet scores related to increased effort required to process both emotion and inhibition cues. Finally, higher antisocial facet scores related to poorer behavioral inhibition overall. This research challenges theoretical accounts of psychopathy focused on specific deficits in negative emotion, such as fearlessness, while offering some support for theories related to attentional dysfunction. Results also highlight the importance of facet-level theorizing, as results varied by facet. This study may inform efforts to reduce disinhibited behaviors, particularly in emotional contexts, among those high in certain psychopathic traits.

1. INTRODUCTION

Inhibition is a feature of cognitive control central to the ability to stop or prevent an action, and it is theorized that behind many behaviors generally considered maladaptive is inadequate inhibitory control (Anestis et al., 2007; Lubman et al., 2004). Deficits in inhibitory control, particularly in the context of negative emotion, can lead to impulsivity in the form of increased aggression, suicide risk, and antisociality (Barratt et al., 1999; Venables et al., 2015). Although disinhibition is often studied in relation to negative emotion, deficient inhibitory control in the presence of positive emotion may lead to potentially harmful behavior as well, including gambling, risky sexual behavior, and substance abuse (Cyders & Smith, 2008; Zapolski et al., 2009).

Often associated with an increased propensity toward such behaviors, psychopathy is a personality disorder marked by increased disinhibition and involvement in antisocial, aggressive, and substance using behaviors (Smith & Newman, 1990; Verona et al., 2001). For this reason, the study of emotion-inhibition connections in psychopathy is relevant to understanding, and possibly preventing, aggressive and other impulsive behaviors. Prior literature has focused primarily on negative emotional processing disrupting cognition (or vice versa) among individuals high or low on psychopathic traits (e.g., Baskin-Sommers & Newman, 2012; Sadeh & Verona, 2012; Verona et al., 2012). However, it is unclear whether previously observed effects of emotional stimuli on inhibitory processing vary by distinct facets of psychopathy, or are specific to negative emotional contexts in particular (vs. broader emotional arousal). The present study addresses this uncertainty and expands on the growing literature on emotion-cognition interactions in psychopathy by examining the effects of both positive and negative emotional stimuli on inhibitory control during an emotional-linguistic Go/No-Go task. This work also informs whether interventions targeting atypical processing of negative emotion specifically or those targeting emotion broadly may be more useful in reducing or preventing disinhibited behaviors among persons high in psychopathic traits. As such, the current study adds nuance to the understanding of emotion-inhibition relationships among individuals high in psychopathic traits by examining whether any effects of emotion on cognition are specific to negative emotion, and by identifying the ways in which (if any) these effects vary across facets of psychopathic traits.

1.1. Psychopathy facets, emotional processing, and cognition

Individuals high in psychopathic traits exhibit a number of personality characteristics, such as shallow affect, callousness, and deceitfulness (Cleckley, 1976; Hare, 2003). More recent work breaks down psychopathic traits into four facets: interpersonal, affective, lifestyle, and antisocial (Hare, 2003). The interpersonal facet includes features such as superficial charm, lying, and manipulation; whereas the affective facet includes shallowness, callousness, and lack of remorse. In contrast, high scorers on the lifestyle facet exhibit a propensity toward sensation seeking, boredom, and reckless behaviors (e.g., irregular employment, excessive speeding and spending); and the antisocial facet consists of the outputs of this behavioral style, mostly aggressivity, criminality, and juvenile delinquency (Hare, 2003; Williams et al., 2007). Past work suggests that these psychopathic traits are dimensionally distributed in community populations (Marcus et al., 2004; Walters et al., 2007). While prior research in psychopathy has included non-community samples such as undergraduates or incarcerated individuals, these samples tend to represent low and high ends of the spectrum of scores, respectively. Thus, psychopathy research involving community samples is advantageous because it allows for the findings to be generalized to a wider array of the population, thereby increasing the confidence that results are not driven by specific demographic and cultural factors.

Traditionally, individuals high in psychopathic traits, particularly the interpersonal-affective features, have been characterized as fearless, showing diminished sensitivity to emotional stimuli, particularly those which relate to negative emotions (Benning et al., 2005; López et al., 2013; Lykken, 1957; Patrick et al., 1993; Sadeh & Verona, 2012). Individuals high in psychopathy have exhibited, for example, attenuated fear-potentiated startle responses and diminished skin conductance to aversive stimuli (Levenston et al., 2000; López et al., 2013; Patrick et al., 1993). There is evidence, however, that fearlessness may not fully characterize the emotional disturbance accompanying psychopathy, and that this emotional dysfunction is not limited to negative emotion. Individuals high in psychopathy have demonstrated diminished skin conductance to both pleasant and unpleasant sounds (Verona et al., 2004), as well as similar effects of positive and negative emotion on reaction time in decision-making tasks (Williamson et al., 1991). The current study contributes to the literature challenging the focus on fear and other negative emotions by extending its reach to include effects of emotional arousal on inhibitory control.

Whereas emotional deficits have been typically associated with interpersonal and affective traits, the antisocial and lifestyle features have been associated with dysfunction in cognition, though this has been studied less often in the context of emotion. For example, event-related brain potential (ERP) studies have identified cognitive deficits in antisocial individuals as indexed by reduced amplitudes and longer latencies of the P3 component across several different tasks (Clark et al., 2019; Gao & Raine, 2009; Pasion et al., 2018). More specifically, although the majority of existing studies have found no P3 amplitude differences associated with total psychopathy scores, investigations at the facet level have suggested that overall P3 amplitude reductions are specific to the antisocial and lifestyle facets (Clark et al., 2019; Gao & Raine, 2009). Additionally, some studies have found positive relationships between overall P3 amplitudes and interpersonal-affective traits (Anderson et al., 2011; Carlson & Thái, 2010; Clark et al., 2019), though the literature on these traits and overall P3 amplitudes is more mixed (Clark et al., 2019; Drislane et al., 2013).

In addition to P3, the N2 component has also been commonly investigated in relation to psychopathy and elicited using Go/No-Go tasks (Clark et al., 2019). Both components typically show larger amplitudes for No-Go (inhibit) versus Go (respond) trials (Falkenstein et al., 1999; Polich, 2007), and are thought to index early inhibition of a response plan (N2) and motor response inhibition (P3; Gajewski & Falkenstein, 2013). Recent work by Waller et al. (2020) addressed the potential confound of stimulus (in)frequency in the interpretation of P3 as an index of motor inhibition. Independent components analysis indicated that the stop-signal P3 and infrequency P3 share a common neural generator and likely relate to motor inhibition, as demonstrated by a strong correlation between amplitude of infrequency P3 and a slowing of motor response to a target following infrequent events in the change-detection task. In other words, whether inhibition is willfully recruited for the purpose of stopping an action or incidentally recruited as part of the detection of infrequent events, both processes appear to be inextricably tied and reflective of the same inhibition process. In support of P3 indexing inhibition, prior work has found that reductions in P3 amplitude predict real-world disinhibition, including self-reported substance use and disinhibited behaviors (Brennan & Baskin-Sommers, 2018; Iacono et al., 2002). Additionally, a recent meta-analysis of N2 and antisociality literature (Pasion et al., 2019) concluded that inhibition was central to explaining reduced N2 amplitudes.

Prior literature has found No-Go P3 to be maximal across different sites, though a frontocentral maximum has been found most consistently (Falkenstein et al., 1999; O’Connell et al., 2009; Polich, 2007). An earlier, frontally maximal P3 may reflect automatic attentional allocation, while a later, parietally maximal P3 may reflect effortful processing (Clark et al., 2019; Polich, 2007). Similarly, No-Go N2 has been primarily frontocentrally maximal, though some studies have found a parietal N2 indexing inhibition as well (Ciesielski et al., 2004; Kiefer et al., 1998).

Despite data showing that No-Go P3 is attenuated in antisocial individuals, but not necessarily in persons high in interpersonal or affective traits (Clark et al., 2019; Gao & Raine, 2009; Pasion et al., 2018), the No-Go N2 and psychopathy literature is more limited and also mixed. Previous findings show that No-Go N2 amplitudes of those higher in psychopathic traits were smaller (Kiehl et al., 2000), larger (Kiehl et al., 2006; Kim & Jung, 2014), or similar (Munro et al., 2007) to those of low-scoring individuals. Further studies on psychopathic traits and the No-Go N2 and P3 are needed, as these components are thought to index somewhat different aspects of inhibition, respectively: early inhibition of a response plan and motor inhibition (or at least the effortful processing of cues for inhibition; Clark et al., 2019; Falkenstein et al., 1999; Polich, 2007; Smith et al., 2008).

1.2. Emotion-cognition interactions in psychopathy

Emotion and cognition exist in dual executive competition, in which emotional stimuli and cognition compete for the same processing resources (Pessoa, 2009). Because of this, emotion can either enhance (when emotional stimuli are task-relevant) or interfere with (when emotional stimuli are task-irrelevant) cognition (Dolcos et al., 2011; Dolcos & Denkova, 2014). In terms of psychopathy, the assumption of the fearlessness hypothesis is that persons high vs. low on psychopathy measures will show less interference of cognition by emotional stimuli, due to their documented insouciance and hyporeactivity to fearful or other negative stimuli (Lykken, 1957; Marsh & Blair, 2008; Patrick, 1994). However, the findings from previous studies are less than conclusive on this point (Anderson & Stanford, 2012; Lorenz & Newman, 2002; Mitchell et al., 2006; Verona et al., 2012). Researchers have reported either less interference by emotional stimuli on cognitive task performance or no difference among individuals high in psychopathy compared to low-scoring controls (Anderson & Stanford, 2012; Lorenz & Newman, 2002; Mitchell et al., 2006).

A separate line of research has examined the influence of cognition on emotion processing. The suggestion in cognitive theories of psychopathy is that attentional deficits account for the reduced emotional processing in psychopathy, refuting the basic tenets of the fearlessness hypothesis. In particular, studies of response modulation (Patterson & Newman, 1993) or attentional bottleneck theories (Newman et al., 2010) have found that reactivity to fearful stimuli among individuals high in psychopathic traits is not deficient when participants are instructed to specifically attend to threat stimuli (Newman & Baskin-Sommers, 2012). However, if the emotional content is peripheral to task demands, emotional deficits are observed in psychopathy. Inhibitory control demands in a Go/No-Go task should supersede processing of peripheral emotional content and thus, according to cognitive theories of psychopathy, inhibitory demands (No-Go vs. Go trials) should diminish emotional processing in psychopathy. The current study therefore has potentially important implications for reconciling existing cognitive and emotional theories of psychopathy. In doing so, this study can inform potential interventions for psychopathy and accompanying disinhibited behaviors, as interventions targeting cognitive-emotional dysfunction among individuals high in psychopathy have thus far been promising (Baskin-Sommers et al., 2015).

As noted, a major limitation of prior studies is the lack of focus on positive emotional stimuli. Because reward may provide a strong motive for criminal behavior in psychopaths (45% of criminal psychopaths list material gain as a motive; Williamson et al., 1987), clarifying the relationship between inhibitory control and positive stimulus processing may be highly important in understanding and addressing psychopathy-related criminal offending. There are also theoretical reasons to believe that processing of positive material may interact with inhibitory control demands in individuals high in psychopathy. Early conceptualizations of psychopathy describe weak inhibition due to strong reward-seeking and reduced ability to learn from punishment (Cleckley 1976; Fowles, 1988). In agreement with this, neuroimaging studies of reward processing in psychopathy suggest a hypersensitivity to reward (Bjork et al., 2012; Glenn et al., 2010), primarily driven by the impulsive-antisocial factor (Buckholtz et al., 2010). Relatedly, psychopathy has been associated with impaired extinction of previously rewarding responses and improved performance on tasks involving reward (Mitchell et al., 2002; Newman et al., 1987), though none of these studies examined differences between facets of psychopathy. The current study aims to build upon previous work theorizing and supporting increased reward sensitivity among individuals high in psychopathy (Bjork et al., 2012; Buckholtz et al., 2010; Fowles, 1988), and to identify whether emotion-related interference of inhibitory control is specific to negative emotion, or generalizable to broader emotional arousal.

1.3. Present study

Disinhibited behaviors, such as substance use, aggression, and various types of crime, often have high social, individual, and economic burden, particularly when they occur in emotional contexts (Carmichael & Piquero, 2004; Smith & Cyders, 2016). Previous work has indicated that individuals high in psychopathic traits are predisposed to engage in many of these disinhibited behaviors, but that this may vary across facets of psychopathic traits (Hemphill et al., 1994; Verona et al., 2001; Walsh et al., 2007). Facet-level investigations are important because different configurations of psychopathic traits are differentially related to real-world outcomes. For example, elevations on different psychopathic traits are associated with different types of crime (e.g., white collar crime associated with interpersonal traits; Boduszek et al., 2017; Edwards et al., 2017). There are few studies examining cognition-emotion relationships with respect to psychopathic traits on a facet level, however, and even fewer that include positive emotional stimuli (but see Mitchell et al., 2006). It is thus unclear the extent to which previously found cognition-emotion relationships in psychopathy are effects of emotional valence or emotional arousal, a distinction which may be important in refining existing theories of emotion and cognition, including fearlessness and response modulation hypotheses, and for developing potential prevention or intervention measures. The central aim of this study was to address the research need for inclusion of positive emotional stimuli in literature on emotion-cognition relationships in psychopathy, and to improve specificity of this literature by using the four-facet model of psychopathic traits within a community sample.

The first goal of the current study was to determine whether there are differential associations between facets of psychopathy and emotion-related disruptions of inhibitory control. We expected interpersonal and affective facets of psychopathy, particularly the latter, to relate to less interference of negative stimuli on inhibitory control, given theorized blunted fear response associated with these facets (Fowles, 1980; Lykken, 1957). We expected the opposite for lifestyle and antisocial facets, given their associations with increased negative emotionality (e.g., Verona et al., 2001) and negative emotional processing (e.g., Verona et al., 2012). The second goal was to determine whether effects of emotional stimuli on inhibitory control differ by emotional valence, such that positive word stimuli potentially interfere with inhibitory control for individuals high in certain facets of psychopathy. Because the conning/manipulative behaviors associated with the interpersonal facet and the risk-taking behaviors associated with the lifestyle facet may reflect this reward sensitivity, we predicted positive stimuli interference on inhibitory control among individuals high in these facets. The third goal was to identify the reciprocal effects of cognition on emotion, based on previous work on the response modulation/attentional bottleneck hypothesis (Baskin-Sommers et al., 2011; Newman et al., 2010). We expected that increased cognitive demands, associated with having to inhibit a strong response set during No-Go trials, would affect the processing of emotional stimuli, specifically among individuals high in interpersonal/affective psychopathic traits.

2. METHOD

2.1. Participants

Study participants were 138 community members mainly recruited via online ads as part of a larger, grant-funded study on aggression. The focus of the broader study was to investigate interplay between cognition and emotion in relation to aggression. For inclusion in the study, participants had to be between 18 and 40 years old and be able to speak and read English well. Exclusionary criteria consisted of specific medical (e.g., traumatic brain injury, epilepsy) or mental health (i.e., bipolar disorder, schizophrenia) criteria that could contribute to qualitative differences in brain function or disinhibition. Although we recruited from the community, the incidence of externalizing behaviors was relatively high in our sample, with the majority of participants reporting at least one instance of police contact due to crime (59.40%; n = 82).

Of the 138 individuals who completed the study, 21 were excluded because of excessive artifacts in the EEG data (n = 15), corrupted EEG data files (n = 4), and validity concerns including inability to distinguish between Go and No-Go stimuli (n = 1) and noncompliance during the task (n = 1). These exclusions resulted in a final sample size of 117 (see demographic information presented in Table 1). Our final sample had an even gender split, 30% identified as Black and 16% as Hispanic, and was mostly employed with low/middle-income salaries.

Table 1.

Sample characteristics

	Full Sample (n=138)	Task Performance Subsample (n=114)	ERP Subsample (n=117)
Age (M (SD))	29.25(6.36)	29.41 (6.58)	29.35(6.54)
Missing (n(%))	2(1.45)	2(1.75)	2(1.71)
Gender (n(%))
Male	66(47.83)	55(48.25)	57(48.72)
Female	70(50.72)	57(50.00)	58(49.57)
Transgender	2(1.45)	2(1.75)	2(1.71)
Race/Ethnicity (n(%))
Caucasian	75(54.35)	63 (55.26)	65(55.56)
Black/ African American	41(29.71)	34(29.82)	35(29.91)
Asian	9(6.52)	5(4.39)	5(4.27)
American Indian or Alaskan Native	4(2.90)	4(3.51)	4(3.42)
Other	8(5.80)	7(6.14)	7(5.98)
Missing	1(.72)	1(.88)	1(.85)
Ethnicity (n(%Hispanic))	23(16.67)	19(16.67)	19(16.23)
Missing	8(5.80)	7(6.14)	8(6.84)
Employment Status (n(%))
Employed	110(79.71)	88(77.19)	91(77.78)*
Unemployed	20(14.50)	18(15.79)	18(15.38)
Homemaker	6(4.35)	6(5.26)	6(5.13)
Other (e.g., Retired)	1(.72)	1(.88)	1(.85)
Missing	1(.72)	1(.88)	1(.85)
Income (n(%))
<$15,000	25(18.12)	18(15.79)	19(16.24)
$15–30,000	39(28.26)	34(29.82)	35(29.91)
$30–45,000	27(19.57)	23(20.17)	23(19.66)
$45–60,000	21(15.22)	15(13.16)	16(13.68)
$60–75,000	8(5.80)	8(7.02)	8(6.84)
>$75,000	15(10.86)	14(12.28)	14(11.97)
Missing	3(2.22)	2(1. 75)	2(1.71)
Recruitment Source (n(%))
Friend/Relative	14(10.14)	11(9.65)	11(9.40)
Electronic Ads/Flyers	123(89.13)	102(89.47)	105(89.74)
Missing	1(.72)	1(.88)	1(.85)
Psychopathy (M(SD))
Interpersonal Manipulative (IPM)	38.01(9.92)	37.92 (10.09)	37.79(10.00)
Callous Affect (CA)	37.08(8.59)	36.97 (8.58)	36.88(8.52)
Erratic Lifestyle (ELS)	44.30(10.48)	43.79 (10.67)	43.68(10.58)
Antisocial Behavior (ASB)	29.89(10.76)	29.37 (10.94)	29.36(10.91)
Total Psychopathy	149.28(31.26)	148.05 (32.34)	147.71(32.14)

Note.

^*participants with excluded ERP data were significantly more likely to be employed compared to the full sample.

2.2. Procedures

Prior to participation, all participants provided written informed consent. Data collection was completed over two sessions as part of a larger study, one session involving interviews and a shock threat attention task (not analyzed in this paper), and a second session 2–3 days later involving self-report questionnaires (including the psychopathy measure) and the emotional-linguistic Go/No-Go task. The latter session is the focus of this paper. Participants received monetary compensation for their participation.

2.3. Psychopathy measure

Psychopathic traits were assessed via the Self-Report Psychopathy Scale-III (SRP-III; Paulhus et al., 2009). The SRP-III consists of 64 items answered on a 5-point Likert-type scale (1 = disagree strongly to 5 = agree strongly) and is comprised of four subscales mirroring the four factor structure of the Psychopathy Checklist-Revised (PCL-R; Hare, 2003), including 1) Interpersonal Manipulation (IPM), which captures characteristics such as pathological lying and manipulation, 2) Callous-Affect (CA), which reflects low empathy and general lack of concern for others, 3) Erratic-Lifestyle (ELS), which includes items relating to recklessness and impulsivity, and 4) Antisocial Behavior (ASB), which measures overt antisociality including criminal acts and aggression. Compared to previous studies using the SRP-III, mean psychopathy scores in our sample were similar to previous community samples (Gordts et al., 2017; Watt & Brooks, 2012) and lower than incarcerated samples (Sandvik et al., 2012). Reliability of SRP-III facets in our sample ranged from good to acceptable (IPM: α = .83, CA: α = .75, ELS: α = .81, ASB: α = .81).

2.4. Emotional-linguistic Go/No-Go task

To measure inhibitory control in the context of emotional stimuli, participants completed an emotional-linguistic Go/No-Go task (Goldstein et al., 2007) modified for use in ERP studies (Verona et al., 2012). Task stimuli were 96 words selected from the Affective Norms for English Words (ANEW; Bradley & Lang, 1999; see Supplementary Table 1 for a list of words used), organized into 3 blocks: positive (POS; e.g., terrific, reward), neutral (NEU; e.g., basket, lawn), and negative (NEG; e.g., assault, cruel) emotion words. Words belonging to negative blocks were selected in prior research to be salient to those with histories of aggression and behavioral dysregulation (e.g., “hostile”, “trauma”; see Sprague et al., 2010 for validation). The positive blocks had not been used previously but were matched to negative words on arousal, based on normative ratings from ANEW (Bradley & Lang, 1999). All words were matched on word length and frequency of use.

Participants were instructed to respond quickly but accurately to features (i.e., normal vs. italic font) of the presented word by pressing a button with their dominant hand for words presented in normal font (Go trial), or by inhibiting this response to words presented in italicized font (No-Go trial). They were not given instructions regarding word valence (positive vs. neutral vs. negative). Participants completed 6 blocks for each of the 3 word categories for a total of 18 blocks. Word order was randomized within each block. Block order and order of emotional blocks were counterbalanced across participants. Each category of emotional word block contained fewer No-Go trials (9 per block, 144 total) than Go trials (23 per block, 432 total), to establish a strong response set. Each word was presented for 1400 ms followed by a random (750–1000 ms) inter-trial interval. Participants were allowed to respond until presentation of the next word. Each block followed a short rest period. Prior to the task, 20 practice trials with neutral words not used elsewhere in the task were administered to ensure participants understood the instructions. Average task completion time was about 20 minutes.

2.5. Physiological data acquisition and analysis

2.5.1. ERP data collection

ERPs were collected using Electrical Geodesics system hydrocel 64-channel sensor nets and amplifiers (EGI, Eugene, OR). Electrodes on our EGI nets (see Supplementary Figure 1 for sensor layout) can be mapped onto a standard 10–10 placement (Chatrian et al., 1985). Impedances were kept below 50 kΩ. Data were continuously recorded using Net Station 5.3.0.1 (Eugene, OR) at a sampling rate of 250 Hz and referenced to the vertex during online recording. Analog signals were amplified online using Net Amps 400 amplifiers.

2.5.2. ERP data processing

Data reduction was completed offline using Net Station 5.4.2. Data were band-pass filtered at .1 to 30 Hz (rolloff: 7.53 dB/octave) and segmented into 1000 ms epochs by condition (6 total conditions; trial type: Go, No-Go; word category: positive, neutral, negative). Only correct trials were examined. Epochs were segmented 200 ms before and 800 ms after stimulus onset. Artifacts were automatically detected and manually verified for exclusion from additional analysis (bad channel >200 μV, eye blinks >140 μV, and eye movement >55 μV). No ocular artifact correction was applied. Bad channels (channel with 20% or more segments with fluctuations over 200 μV) were spherical spline interpolated from nearby electrodes. A segment was marked bad if it contained more than 10 bad channels. An average of 83.14% of trials were retained for Go trials (M = 361.45, SD = 44.20) and 83.36% were retained for No-Go trials (M = 119.34, SD = 16.83). Participants with fewer than 20 total possible trials per each of the 6 conditions (n = 15) were excluded from ERP analyses to ensure the minimum number of trials needed for statistically stable (e.g., internally consistent) measurement. Following these procedures, 117 subjects were retained. Data were baseline adjusted using a 200 ms pre-stimulus period and then re-referenced from vertex to an average reference of all 64 channels. Previous research has found that average references are appropriate for dense arrays (i.e. ⩾ 64 channels; Hu et al., 2018).

To characterize the internal reliability of the ERP components within our sample, dependability estimates were computed using the ERP Reliability Analysis Toolbox v 0.4.8 (Clayson & Miller, 2017), which uses formulas based on generalizability theory (Baldwin et al., 2015). Results suggested our ERP components had good internal reliability (N2 = .85-.93; P3 = .87-.93).

All processed, artifact-free segments were then averaged together for each condition and participant to produce an individual average waveform. Additionally, all conditions were then averaged across all participants to produce a single grand average. This grand average was used for visual inspection of the waveform and scalp topography to aid in selection of N2 and P3 component time-windows and examine where on the scalp effects were maximal, respectively. Electrode sites were chosen based on scalp distributions of the present data and on previous literature (Rietdijk et al., 2014) indicating that inhibitory control effects are maximal over frontocentral, central, or parietal sites. See Figure 1 for sensor layouts.

Figure 1.

ERP Waveforms and Topographies by Condition at the Parietal Site

Adaptive means were computed separately for the N2 and P3, which identified individual measurement windows for each participant to account for subtle differences in waveform morphology across participants while reducing the biasing impact of background EEG noise (Clayson et al., 2013). The N2 component was defined as the minimum mean-peak amplitude (average amplitude across +/− 15 ms around the peak) detected between 250 to 350 ms post-stimulus. The P3 component was defined as the maximum mean-peak amplitude (average amplitude across +/− 25 ms around the peak) detected between 360 to 600 ms post stimulus. The adaptive mean for the N2 and P3 for each subject was then exported for statistical analysis to be completed in SPSS Version 24. Visual inspection of the scalp topography and grand average waveform revealed that both the N2 and P3 effects were maximal and most apparent at the parietal site (see supplementary material). Thus, we decided to focus on parietal N2 and P3. Of note, because the P3 was also visible during our time-window of interest at frontocentral and central sites, general linear model (GLM) analyses were conducted to examine whether the pattern of inhibitory control processing differed by site. Our results suggested no significant differences in inhibitory control across P3 sites. Further, because the No-Go N2 effect is typically observed frontocentrally, we conducted supplementary analyses to explore the extent to which our results were impacted by the decision to use the parietal rather than frontocentral site. Results of these analyses suggested that our main findings were largely similar regardless of whether the frontrocentral or parietal site was used (see supplementary material for more information).

2.5.3. Task performance data

Task performance indicators of inhibitory control, including commission errors (number of erroneous “Go” responses during No-Go trials) and reaction time (RT; average RT across Go trials) were also analyzed. Performance data were calculated separately for each condition (see Table 2). Of the 117 participants who completed the task and had valid EEG data, 3 were excluded from performance data analyses due to missing (n = 1) and corrupted (n = 2) data files. Therefore, 114 individuals were retained for task performance analyses.

Table 2.

Descriptive Summary of the ERP and Task Performance Data from the Emotional Go/No-Go Task

	Trial Type M(SD)		Emotion Word Category M(SD)
	Go	No-Go	Positive (POS)	Negative (NEG)	Neutral (NEU)	Emotion (POS/NEG)
N2	−.39 (1.80)	−.84 (2.07)	−.66 (1.86)	−.51 (1.95)	−.67 (1.92)	−.59 (1.86)
P3	2.01 (1.96)	3.33 (2.78)	2.58 (2.24)	2.79 (2.40)	2.65 (2.19)	2.68 (2.28)
RT (ms)	479.90 (90.25)	-	478.31 (89.99)	484.92 (93.64)	476.49 (90.25)	481.61 (90.95)
Commission Errors	-	12.06% (8.64%)	12.63% (10.81%)	11.28% (10.26%)	12.30% (10.65%)	11.94% (10.22%)

2.6. Statistical analysis

Data from the emotional Go/No-Go task (N2, P3, RT, commission errors) were analyzed using separate mixed-model repeated measures GLMs with trial type (Go, No-Go) and emotion word category (negative, neutral, positive) as within-subjects variables, and continuous SRP-III facet scores (IPM, CA, ELS, ASB) as between-subject variables. We also included interactions involving psychopathy scores, trial type, and emotional word category. Emotion word category effects were parsed into two orthogonal a-priori planned contrasts 1) emotional valence (negative vs. positive) and 2) emotional arousal (i.e., negative/positive vs. neutral). Only within-subjects contrasts were interpreted, not multivariate effects. Using planned contrasts helped control the false-positive rate and was also theoretically meaningful, allowing us to separately examine the overall effects of emotional arousal and emotional valence on inhibitory control.

Significant interaction effects were decomposed and interpreted within levels of the categorical variables (e.g., effect of emotion word category within each trial type). To provide the best representation of our effects, interactions were also decomposed using partial correlations to examine relationships between the psychopathy facet of interest and ERP/performance variables within each condition, controlling for other psychopathy facets. In addition to p-values, effect sizes were reported for GLM results using partial eta squared (η_p²; 0.01 = small effect, 0.06 = medium effect, 0.14+ = large effect; Cohen, 1988).

3. RESULTS

3.1. Overall task results

Grand average waveforms for each condition are illustrated in Figure 1 and a descriptive summary of the Emotional Go/No-Go task data can be found in Table 2. Also, see supplementary materials for correlations between task performance and ERP variables as well as summaries of GLM results.

Separate repeated measures trial type x emotion word category GLMs were conducted on the parietal N2 and P3. These analyses yielded the expected, large main effect of trial type (Go vs. No-Go), such that the N2 (F(1, 116) = 16.28, p <.001, η_p² = 0.12) and P3 (F(1, 116) = 59.35, p < .001, η_p² = 0.34) amplitudes were enhanced for No-Go vs. Go trials. Next, though we observed no effects of emotional arousal (NEG/POS vs. NEU) on the N2 (p = .253) or P3 (p = .720), we did observe that the pattern of cognitive processing differed by emotional valence (NEG vs. POS). Specifically, analyses of the P3 revealed a medium effect of emotional valence (F(1, 116) = 7.57, p = .007, η_p² = 0.06), such that the P3 was larger during negative vs. positive word categories. Similarly, there was a small (but non-significant) effect of emotional valence for the N2, F(1, 116) = 3.43, p =.067, η_p² = 0.03, such that N2 was greater (more negative) during positive than during negative emotional blocks.

Behavioral measures of performance including RT and commission errors were analyzed as a function of emotion word category (see Table 2). Overall, our task was easy to moderate in difficulty (mean commission errors = 12.06%). Analysis of commission errors revealed a medium-sized effect of emotional valence (NEG vs. POS), F(1, 113) = 7.77, p = .006, η_p² = 0.06, such that participants made more commission errors during positive (vs. negative) word conditions. RT analyses yielded medium-sized effects of emotional valence (F(1, 113) = 7.76, p = .006, η_p² = 0.06) and emotional arousal (F(1, 113) = 7.74, p = .006, η_p² = .06), such that RTs were slower during emotional (vs. neutral conditions) blocks, but especially slow during negative (vs. positive) blocks.

Together, our results suggest that the task elicited large effects of trial type (inhibitory demands) and small to medium effects of emotional valence (negative vs. positive) across the N2 and P3. Further, RT and commission error results indicated greater behavioral interference by emotion. This indicates that both manipulations (of trial type and emotion) worked as expected. The main question is whether emotion interferes with inhibitory control, or inhibitory demands interfere with emotion, among persons with certain levels of psychopathic traits.

3.2. Psychopathy results

3.2.1. ERP data

Mixed model GLMs that included the SRP-III facets, along with emotion word category and trial type, were conducted on the ERP data. Although all 4 facets were entered together in analysis, the results are summarized for each facet in turn.

Interpersonal Manipulation (IPM)

Analysis of N2 revealed no effects involving the IPM facet (ps > .05). P3 analyses revealed a small three-way interaction of IPM x emotional arousal (NEG/POS vs. NEU) x trial type (Go vs. No-Go), F(1, 112) = 5.05, p =.027, η_p² = 0.04. To examine the effects of emotion on inhibitory control, these interactions were decomposed by examining the IPM x trial type interaction separately within emotional (average of POS/ NEG) and neutral word categories. This revealed that the IPM x trial type interaction was stronger during emotional (F(1, 112) = 3.03, p = .085, η_p² = 0.03) than during neutral (F(1, 112) < 0.01, p = .948, η_p² < 0.01) blocks. Follow-up partial correlations revealed IPM was associated with smaller No-Go P3 index (No-Go minus Go) during emotional (r = −.16) than during neutral blocks (r = −.01). The No-Go P3 index in emotional blocks was correlated .89 with the No-Go P3 index in neutral blocks.

To examine the effects of cognition on emotion, we decomposed the IPM x emotional arousal interaction separately within Go and No-Go trials (the latter associated with greater inhibitory control demands), which revealed that the IPM x Emotional Arousal interaction was stronger during No-Go trials (F(1, 112) = 2.85, p = .094, η_p² = 0.03), than during Go trials (F(1, 112) = 1.44, p = .227, η_p² = 0.01). Follow-up partial correlational analyses revealed that higher levels of IPM traits were associated with smaller amplitude differences between emotional vs. neutral blocks in No-Go (r = −.16), and with slightly bigger amplitude differences during Go (r = .11) trials. The emotional arousal P3 index (emotional – neutral) in Go trials was correlated .72 with this index in No-Go trials. Results of effects of cognition on emotion and vice versa for IPM are graphed in Figure 2. In sum, emotion decreased inhibitory processing, and inhibitory control demands decreased emotional processing, among those high vs. low on IPM. Though of note, because the follow-up analyses of this IPM effect were small and non-significant, the meaningfulness of this finding should be interpreted with caution.

Figure 2.

Interaction between IPM, emotional arousal, and trial type on parietal P3, controlling for CA, ELS, and ASB facets

Callous Affect (CA)

Analyses of N2 revealed that CA traits were associated with a small overall effect of a larger (more negative) N2 amplitude, F(1, 112) = 4.92, p = .029, η_p² = 0.04 (partial r = −.21). Meanwhile, P3 analyses revealed a small, three-way CA x emotional arousal x trial type interaction, F(1, 112) = 4.55, p = .035, η_p² = 0.04. To evaluate emotion word effects on inhibitory processing, the CA x trial type interaction was decomposed separately within emotional and neutral word categories. This revealed that the CA x trial type interaction was non-significant during both emotional (F(1, 112) = .012, p = .913, η_p² < 0.001) and neutral (F(1, 112) = 2.80, p = .097, η_p² = 0.02) blocks. Follow-up partial correlations within each emotion word category indicated that higher CA scores were associated with somewhat decreased No-Go P3 during neutral (r = −.16) compared to emotional blocks (r = −.01); as mentioned above, No-Go P3 was highly correlated between the emotional and neutral conditions, r = .89. Though again, these effects were not significant and quite small in size.

To examine effects of inhibitory control demands on emotional processing, analyses were conducted within each trial type. Analyses revealed that CA was associated with less P3 emotion processing (emotion-neutral) in Go trials (F(1, 112) = 4.95, p = .028, η_p² = 0.04; partial r = −.21), than No-Go trials (F(1, 112) = 0.70, p = .405, η_p² = 0.01; partial r = .08). As mentioned above, the emotional arousal P3 index (emotional – neutral) was correlated .72 between Go and No-Go trials. Effects of CA for P3 are graphed in Figure 3. In sum, P3 results fail to indicate that inhibitory control demands interfere with emotional processing for those higher in CA traits (and may indicate the opposite).

Figure 3.

Interaction between CA, emotional arousal, and trial type on parietal P3, controlling for IPM, ELS, and ASB

Erratic Lifestyle (ELS)

N2 analyses revealed no effects involving the ELS facet. As with IPM, analyses of P3 revealed a small three-way ELS x emotional arousal x trial type interaction, F(1, 112) = 6.29, p = .014, η_p² = 0.05. Follow-up analyses revealed that the ELS x trial type interaction was stronger during emotional blocks (F(1, 112) = 5.56, p = .020, η_p² = 0.05) than neutral blocks (F(1, 112) = 0.23, p = .630, η_p² < 0.01). Specifically, follow-up partial correlations within each emotion word category indicated that higher ELS scores were associated with higher No-Go P3 during emotional (r = .22) than during neutral (r = .05) blocks (see above for No-Go P3 correlations between emotional and neutral blocks).

We then examined the ELS x emotional arousal interaction separately for Go and No-Go trials, which indicated ELS was associated with increased emotion processing (emotion vs. neutral P3) during No-Go (F(1, 112) = 6.06, p = .015, η_p² = 0.05; partial r = .227) compared to Go (F(1, 112) = 0.27, p = .607, η_p² < 0.01; partial r = −.05) trials. Results for P3 involving ELS are graphed in Figure 4 and indicate that inhibitory control demands were related to increased attention to emotion among persons scoring high vs. low on ELS.

Figure 4.

Interaction between ELS, emotional block, and trial type on parietal P3, controlling for IPM, CA, and ASB

Antisocial Behavior (ASB)

Analyses revealed no effects involving ASB traits.

3.2.2. Task performance data

All psychopathy facets were entered together as continuous between-subjects variables, with emotion word category as the within-subject variable, in repeated measures ANOVAs conducted on RT and commission errors. Analyses revealed a medium-sized effect of IPM and a small but non-significant effect of ASB on overall RT (IPM: F(1, 109) = 7.84, p = .006, η_p² = 0.07; ASB: F(1, 109) = 3.75, p = .055, η_p² = 0.03), such that increased IPM scores were associated with faster RT (partial r = −.26) and increased ASB scores were associated with slower RT (partial r = .18). ASB was also associated with commission errors, at a small effect size, F(1, 109) = 5.05, p = .027, η_p² = 0.04 (partial r = .21). Analyses yielded nonsignificant small effects involving CA or ELS on performance measures. A marginally significant ELS x emotional valence (NEG vs. POS) interaction on RT, F(1, 109) = 3.41, p = .067, η_p² = 0.03, indicated that ELS scores were associated with slightly faster RT during positive (r = - .05, p = .572) than during negative blocks (r < .01, p = .980). Overall, there was little evidence of relationships between psychopathy facets and emotional interference on behavioral task performance.

4. DISCUSSION

Disinhibited behaviors, which are common among individuals high in some facets of psychopathic traits, can result in a variety of negative consequences (Anestis, 2007; Cyders & Smith, 2008; Hemphill et al., 1994; Verona et al., 2001). To further the literature that informs the treatment, management, and prevention of disinhibited behaviors and their consequences, the current study addressed two major questions pertaining to inhibition-emotion relationships among individuals high in psychopathic traits. First, this study used an emotional-linguistic Go/No-Go task containing both positive and negative emotional word stimuli to clarify whether cognition-emotion dysfunction in psychopathy is related to emotion broadly, or specific to negative emotion. The inclusion of positive emotional stimuli in this study was also relevant to early theoretical accounts linking psychopathy with reward hypersensitivity (Fowles, 1988; Gorenstein & Newman, 1980). Second, the present study added specificity to the current literature on cognition-emotion relationships in psychopathy by identifying the impact of emotional stimuli on inhibitory control for each of the four facets of psychopathic traits.

Results supported differential effects of emotional versus neutral stimuli on inhibitory processing among individuals high in distinct facets of psychopathy, with little evidence that positive and negative emotional words impacted inhibitory processing differently. These findings suggest that persons high in certain facets of psychopathy show interference of inhibitory control by emotional arousal, rather than just negative emotion (i.e., fear; Lykken, 1957; Patrick et al., 1993) or just positive emotion (i.e., reward; Cleckley, 1964; Fowles, 1988). Findings for some facets also indicated interference of emotional processing by cognitive demands, thus lending support to cognitive models of psychopathy such as the response modulation hypothesis.

4.1. Overall task results

Overall task results indicated the expected large effects of trial type, with increased amplitudes for No-Go P3 and N2, both at parietal sites. Although prior literature has more consistently identified No-Go P3 and N2 as frontocentrally maximal, some studies with Go/No-Go tasks have found these components maximal at parietal sites as well (Gajewski & Falkenstein, 2013). Further, supplementary analyses indicated that results at frontocentral sites were largely the same as those at parietal sites (see supplementary materials). Overall results also suggested a medium effect of emotional valence on P3 amplitudes, such that amplitudes were larger during negative relative to positive emotional words. These results should be replicated, as they diverge somewhat from prior work finding enhanced P3 amplitudes for emotional arousal, or emotion more broadly, though this divergence may be due to differential responding to words rather than pictures (Bradley et al., 2012; Keil et al., 2002).

4.2. Psychopathy facets, inhibition and emotional processing

One key finding from this study is that persons scoring higher vs. lower on the interpersonal manipulation (IPM) features of psychopathy (e.g., grandiosity, manipulation, deceit) showed a pattern of results different from those scoring higher on other facets. Specifically, P3 results suggested that individuals high in IPM traits exhibited reciprocal cognition-emotion interference. That is, IPM traits were associated with decreased inhibitory processing during emotional blocks, and decreased emotion processing during No-Go trials. This indicates that IPM scores are associated with decreased processing of dual features of a stimulus (i.e., italicized font and emotional content), or alternatively, that persons high in IPM make less of an attempt to differentiate these features (as indicated by their quicker overall RT) than those low in IPM. Though this does not support our first hypothesis (IPM associated with less disruption of cognition by emotion), it supports research on response modulation models indicating that persons high in these traits have difficulty incorporating peripheral information when focused on salient aspects of a task (Baskin-Sommers et al., 2011; Baskin-Sommers et al., 2012). Thus, the focus on inhibition during No-Go trials (which were infrequent and most task-relevant) resulted in potentially less differential processing of emotional and neutral stimuli. During Go trials, P3 was larger for emotional than neutral stimuli, indicating that the removal of inhibition demands allowed for adequate emotion differentiation among persons high in IPM. Results of follow-up correlations for IPM traits were small and non-significant, however, and therefore should be interpreted with caution and require replication.

Callous Affect (CA; e.g., lack of remorse, shallow affect) was the only facet with an effect on N2. More specifically, higher scores on CA were associated with enhanced (more negative) overall N2 amplitudes. Because N2 measured during inhibitory control tasks is generally thought to reflect earlier inhibition of a response plan (rather than motor inhibition, which is more often associated with P3; Gajewski & Falkenstein, 2013), this result may reflect increased early cortical engagement, perhaps in agreement with literature finding these traits to be associated with heightened early selective attention (Baskin-Sommers et al., 2011). However, this engagement may not be related to improved task performance or sustained engagement with the task. Indeed, P3 analyses revealed that individuals high in CA showed overall decreased inhibitory processing (i.e., reduced No-Go P3), but only during neutral blocks. During emotional blocks, CA scores were unassociated with inhibitory processing, suggesting regulatory, not disruptive, effects of emotional stimuli in persons high in CA. However, CA scores were also associated with smaller P3 emotional differentiation during Go but not No-Go trials, indicating that the dampened salience of emotion among persons high in CA was only found when inhibitory control demands were low. Given the low cortical and physiological arousal associated with CA traits (McDonald & Verona, 2019), this combination of results suggests that persons scoring highly on this facet require especially engaging stimuli, represented by the emotional No-Go trials in particular, to elicit sufficient cortical engagement with the task. Given these effects were not hypothesized and follow-up correlations were small and non-significant, replication is required to confirm these interpretations.

Unlike CA, which was associated with reduced inhibitory processing during neutral but not emotional blocks, scores on a measure of erratic lifestyle (ELS; e.g., sensation seeking, recklessness) were associated with increased no-go P3 during emotional compared to neutral blocks. This may mean that either emotion facilitated inhibitory processing among persons high in ELS traits or that participants higher in ELS had to exert greater effort in processing No-Go stimuli during emotional than during neutral conditions, potentially because they were distracted by the emotional content. The results are consistent with Verona et al. (2012), who found that antisocial personality disorder was associated with heightened processing of negative emotional words, especially during No-Go trials. Given previous findings, and other work supporting normal or enhanced emotional processing associated with ELS (Cigna et al., 2017; Verona et al., 2004), we interpret our results as indicating that individuals high in ELS prioritize processing of emotional stimuli even when task-irrelevant, and in turn, this focus on emotion requires increased effort to process inhibitory cues. This interpretation remains in line with prior work indicating that emotional contexts can disrupt response inhibition among persons high in behavioral dysregulation (Sprague & Verona, 2010; Verona & Bresin, 2015). This difficulty manifests in increased risk for engaging in disinhibited behaviors in the presence of emotion, such as substance use or aggression, both of which have been associated with these traits (Miller et al., 2015; Sellbom et al., 2017).

Finally, persons high on a measure of antisocial behavior traits (ASB; aggressivity, criminality, juvenile delinquency) showed overall poorer task performance compared to individuals low in ASB, consistent with work showing poor behavioral inhibition associated with antisocial traits and antisocial personality disorder (Gao & Raine, 2009; Pasion et al., 2018; Verona et al., 2012), although this was not reflected in ERP data. This conflicts with previous work finding blunted No-Go P3 amplitudes among individuals high in antisocial traits (Gao & Raine, 2009), and thus warrants further investigation. It may be that individuals with high ASB scores have deficits in response inhibition, but that these deficits are related to different cognitive processes than those indexed by the ERP components investigated in this study (N2/P3).

4.3. Positive emotional stimuli

Our results do not support differentiation between positive and negative emotional stimuli in terms of their impact on cognitive control among individuals high in distinct facets of psychopathic traits. The disinhibition often accompanying psychopathic traits was not influenced uniquely by exposure to positive or reward cues in our study, as may be expected from early accounts of psychopathy-related reward sensitivity (Fowles, 1988; Gorenstein & Newman, 1980). Importantly, although the positive word stimuli in the present study were relevant to reward, they were not rewarding in and of themselves. Previous studies investigating reward sensitivity in psychopathy have found increased sensitivity to anticipation of reward (Bjork et al., 2012; Buckholtz et al., 2010) and reward-based learning (Arnett et al., 1993; Schmauk, 1970), processes not reflected in the task utilized in this study. Our results should instead be considered evidence that implicit, reward-related stimuli do not impact cognitive processing differentially among individuals high in psychopathy. Results were also not specific to negative emotion for any facet of psychopathic traits, suggesting that cognition-emotion dysfunction in psychopathy is related to atypical processing of emotion broadly.

4.4. Limitations and contributions

This study has several limitations. First, the task used simulates the processing of emotional content peripheral to an individual’s goals and tasks, such as emotional cues present in facial expressions and body language. Our results should be considered in relation to these situations, rather than those eliciting strong emotional reactions or requiring deeper comprehension of emotional content, such as empathetic responding. Second, future studies may benefit from employing an inhibitory control task with greater difficulty and ecological validity (i.e., gambling tasks; Bechara et al., 1994), which could reveal the ways emotional contexts affect goal-directed behavior in real life. Third, although psychopathy scores in our sample were similar to those found in other community samples (e.g., Gordts et al., 2017), similar work among samples with more extreme levels of psychopathy (i.e., offender samples) may extend understanding of these relationships across the full spectrum of psychopathic traits. Fourth, effects involving psychopathic traits in the current study were rather small and therefore conclusions and interpretations drawn from these effects are tentative. The size of the effects reported in this study also may be too small to warrant use in future clinical or intervention applications. Future studies replicating these effects are required before firmer conclusions can be drawn. Finally, it is possible that the N2 and P3 in our data may be driven by a single phenomenon, rather than representing separable, distinct processes. Though component overlap between N2 and P3 is not uncommon given their adjacent time windows (e.g., Dimoska et al., 2006; Harper et al., 2014, 2016), it is worth noting that our results may be reflective of a single inhibitory process.

The current study offers novel insights into cognition-emotion interactions in psychopathy, and supports the importance of facet-level distinctions and a conceptualization of psychopathy as configural and multidimensional (Lilienfeld, 2018). For any of the facets, there were no specific effects for emotional valence, which is at odds with assumptions that atypical emotional processing in psychopathy is limited to negative emotion. The main distinction found was for the interpersonal psychopathic traits, which were associated with difficulty simultaneously processing emotional stimuli and inhibition cues, somewhat in line with attention and information processing deficit accounts of psychopathy (e.g., Baskin-Sommers et al., 2011). The other facets, however, were not associated with emotion-cognition interference. Higher callous affect traits were associated with lower inhibition and emotion processing, except during the most engaging trials (emotion + No-Go cues). We interpreted the pattern of results involving the erratic lifestyle traits as indicating that persons high in these traits required greater effort to process inhibitory cues in emotional contexts than those low in these traits. This interpretation is consonant with potential heightened attention to emotional stimuli, even when not task relevant, among persons high in erratic lifestyle traits. The current literature would benefit from future work attempting to replicate our findings and better understand the interplay between cognition and emotion among individuals high in distinct facets of psychopathy.