Sex Moderates Reward- and Loss-Related Neural Correlates of Triarchic-Model Traits and Antisocial Behavior

Abstract

Abnormalities in responses to reward and loss are implicated in the etiology of antisocial behavior and psychopathic traits. While there is evidence for sex differences in neural response to reward and loss, it remains unclear how sex differences may moderate links between these neural responses and the phenotypic expression of antisocial behavior and psychopathic traits. This study examined sex differences in associations of neural response to reward and loss with antisocial personality symptoms and psychopathic traits. Functional neuroimaging data were collected during a monetary incentive delay task from 158 participants. Among males, during loss anticipation, activation in the left nucleus accumbens was negatively associated with antisocial behavior. Among females, during loss feedback, activation in the left nucleus accumbens and left amygdala was negatively associated with antisocial behavior. These results suggest that phenotypic sex differences in psychopathic traits and antisocial behavior may in part be attributable to different etiological pathways.

The financial and emotional burden on society generated by individuals diagnosed with antisocial personality disorder (ASPD) or psychopathy is immense, as these individuals are disproportionately responsible for both violent and non-violent crime (Kiehl & Hoffman, 2011). The annual economic cost associated with law enforcement, victim losses, and criminal justice in the United States has been estimated at $1.7 trillion (Anderson, 2012). ASPD is defined by the Diagnostic and Statistical Manual of Mental Disorders, 5^th edition (DSM-5) as a pattern of violating the rights of others beginning in adolescence and persisting into adulthood (APA, 2013). The presence of psychopathic traits can give rise to antisocial behaviors similar to those exhibited by individuals with an ASPD diagnosis and also lead to deficits in interpersonal and affective functioning.

Current conceptions of psychopathy characterize it as a configuration of dimensional tendencies instead of a discrete categorical syndrome (Skeem et al., 2011). Research on the etiologies of ASPD and psychopathic personality implicates deficits in neural and behavioral response to reward and punishment (Arnett, 1997; Byrd et al., 2014). However, as researchers have also found evidence for sex differences in neural response to reward and loss stimuli (Spreckelmeyer et al., 2009), it remains unclear how neural differences between males and females may contribute to sex differences in the phenotypic expression of antisocial behavior and psychopathic traits. Understanding how the neural correlates of psychopathic traits and antisocial behavior may differ by sex is likely to provide clarifying information regarding differential biological underpinnings of these traits across males and females. As ASPD and psychopathy have traditionally been studied in male or majority-male samples (Alegria et al., 2013; Grann, 2000; Jackson et al., 2002; Salekin et al., 1997), further research is needed to assess possible sex differences in neural response to reward and punishment in individuals who exhibit antisocial behavior.

Psychopathic traits, antisocial behavior, and the processing of reward and punishment

Foundational research on the etiology of psychopathic traits and antisocial behavior has highlighted mechanisms related to the processing of reward and punishment, particularly those involved in recognizing and avoiding aversive situations. An influential theoretical framework—the “three arousal” model (Fowles, 1980)—posits that individuals high in psychopathic traits have a weak behavioral inhibition system (BIS; reflecting sensitivity to punishment), in conjunction with an unconstrained behavioral activation system (BAS; reflecting sensitivity to reward), promoting an insensitivity to punishment that results in pervasive patterns of antisocial and rule-breaking behavior. Other contemporary models propose that hypersensitivity to rewarding stimuli in combination with decreased attention to loss or punishment cues may lead to impulsive and disinhibited behaviors (Newman & Lorenz, 2003; Newman & Wallace, 1993).

The triarchic model of psychopathy is a multidimensional framework for reconciling differing conceptions of psychopathy, wherein distinct dispositional traits are theorized to be linked with differing neurobiological processes (Patrick et al., 2009; Patrick & Drislane, 2015). The triarchic model describes psychopathy as encompassing three distinct, but related, phenotypic constructs: boldness, marked by emotional resiliency and social assertiveness, meanness, entailing a lack of empathy and affiliative capacity, and disinhibition, encompassing weak inhibitory control and hostility (Patrick et al., 2009; Patrick & Drislane, 2015). This model posits that differing levels of these traits account for the different expressions of psychopathy that have been described by historic and contemporary writings. These constructs are also informative for understanding the BIS/BAS systems as studies of adult offenders have found that low BIS is strongly associated with high scores on self-reported Boldness and moderately associated with Meanness while high BAS is associated with high levels of Disinhibition (Hall et al., 2014; Sellbom & Phillips, 2013).

While previous iterations of the DSM have attempted to capture features of psychopathic personality within the diagnosis of ASPD, evidence suggests that only a subset of offenders who meet diagnostic criteria for ASPD demonstrate clinical levels of psychopathic personality (Widiger, 2006). Researchers have found that high levels of disinhibition and meanness are consistently associated with the presence of an ASPD diagnosis, with boldness representing a trait that is uniquely predictive of the presence of psychopathic personality (Venables et al., 2014). Although behavioral and psychophysiological studies have provided evidence that individuals with both ASPD diagnoses and high psychopathic traits display both greater sensitivity to reward and reduced sensitivity to punishment relative to control subjects (Arnett, 1997; Baskin-Sommers et al., 2010; Byrd et al., 2014; Morgan et al., 2014; Ross et al., 2007), it is unclear whether these deficits selectively relate to triarchic model constructs and ASPD symptoms.

Neural response to reward and punishment

Neuroimaging studies aimed at assessing reward and punishment sensitivity in antisocial populations typically utilize functional magnetic resonance imaging (fMRI) paradigms in which participants experience the anticipation and receipt of rewards or losses, such as the monetary incentive delay (MID) task (Knutson et al., 2000). The most consistent finding highlighted by a recent review of this literature in adults (Murray et al., 2018) is an association between antisocial behavior and increased activation in the ventral striatum during reward anticipation (Bjork et al., 2012; Buckholtz et al., 2010; Geurts et al., 2016). Although there are fewer investigations of the receipt of rewards and losses (Murray et al., 2018), individuals with greater levels of psychopathy appear to display decreased activity in the ventral striatum when receiving loss (compared to neutral) feedback (Pujara et al., 2013), but individuals with ASPD have been found to show increased activity in the posterior cingulate cortex and insula during loss receipt (Gregory et al., 2015). Therefore, enhanced striatal responses to reward anticipation appears to be consistently associated with antisocial behavior in adults, and evidence regarding loss feedback appears to be mixed.

Sex differences

The striatum, a subcortical brain region linked to reward evaluation, has been consistently shown to be sensitive to monetary rewards in the MID task in studies of adults (Knutson et al., 2000); however, few studies have directly examined sex differences in striatal responses. Although brain responses to reward and punishment in non-clinical adult samples have been extensively examined (Kujawa et al., 2018; Liu et al., 2011), fMRI investigations into how sex influences these processes have been limited. Behavioral data indicate that such sex differences likely exist, revealing that females are generally more sensitive to reward and punishment than males (Cross et al., 2011; Miettunen et al., 2007; Santesso et al., 2011). In one of the few MID task studies to investigate how males and females differentially process anticipation of monetary reward, males displayed more diffuse brain activation to increasing levels of anticipated monetary rewards, as well as stronger activation than females in the left putamen, a subregion of the striatum, interpreted as greater motivational salience of monetary rewards in males (Spreckelmeyer et al., 2009). A more recent examination of sex differences in a large sample of adolescents displayed similar findings; males demonstrated greater activation in the left putamen, right precuneus, and middle temporal gyrus during reward anticipation (Cao et al., 2019). In a third study (Dumais et al., 2018) examining sex differences in network engagement, there was greater activation of the dorsal attention and default mode networks in females compared to males during exposure to reward and punishment. These findings were taken to suggest that females showed enhanced processing of feedback regarding reward and punishment. Together with behavioral data, these studies suggest that males show enhanced attention and motivation during the anticipation of monetary reward; however, females may be more responsive to feedback regarding the receipt of reward.

Sex differences in the processing of reward and punishment have been particularly understudied in individuals with ASPD or psychopathic traits. It has been established that males demonstrate higher levels of antisocial and externalizing behavior (Cale & Lilienfeld, 2002), and the circulating level of testosterone has been suggested as a potential mechanism that might account for sex differences in such behavior (Portnoy et al., 2014). Studies have found that high circulating level of testosterone, which is associated with approach behavior and high reward sensitivity (i.e., BAS), is also associated with antisocial and aggressive behavior (Glenn et al., 2011; Van Honk et al., 2010). Emerging research suggests that through actions on the mesolimbic dopaminergic system, testosterone may increase reward processing and decrease sensitivity to punishment (van Honk et al., 2004), leading to increased neural response in the ventral striatum and nucleus accumbens (Packard et al., 1997; Welker et al., 2015). As higher levels of testosterone are associated with higher levels of externalizing behavior, it may also account for increased reward-focused behavior. There has been little research conducted examining the association between triarchic model traits and neural response to reward, with only one study examining the association between Disinhibition scores and anticipation of pleasant (compared to unpleasant) images in female college students (Foell et al., 2016). While differences in the task and stimuli do not allow for results to be readily comparable to MID contrasts, the results suggest that for females, higher scores on the Disinhibition scale are associated with decreased activation of the nucleus accumbens (NAcc), another subregion of the striatum, when preparing to view pleasant images. This can be interpreted as suggesting that females who phenotypically present as disinhibited have lower motivational activation when anticipating rewarding stimuli. Additional research is needed to better understand sex differences in relation to receipt of reward and loss and antisocial behavior by directly comparing males’ and females’ neural responses to monetary reward and loss.

Current Study

Evidence suggests that a motivational imbalance (heightened BAS, or sensitivity to reward, coupled with lowered BIS, or sensitivity to punishment) may contribute to the presence and persistence of antisocial behavior and psychopathic personality traits. However, much of this work has focused on male or primarily male samples, despite the evidence for sex differences in neural response to reward and punishment (Cao et al., 2019; Dumais et al., 2018; Spreckelmeyer et al., 2009). Therefore, the current study sought to examine the moderating effects of sex on associations between neural response to reward and loss and facets of psychopathy and antisocial behavior. As the MID task has previously been used extensively to examine neural response to monetary reward and loss, we identified a priori regions of interest (ROIs) previously found to be relevant to reward and loss processing on this task (Oldham et al., 2018). Eight ROIs were identified for the anticipation phase of the MID task, consisting of bilateral limbic regions (bilateral NAcc, amygdala, thalamus, and anterior insula) and eight ROIs were identified as relevant to the feedback phase of the task, comprised of limbic and prefrontal regions (bilateral NAcc, amygdala, thalamus, left orbitofrontal cortex, right posterior cingulate cortex), consistent with regions identified in meta-analytic work on the MID task (Oldham et al., 2018).

To examine these moderating effects of sex, we first investigated main effects. We predicted that higher levels of antisocial behavior would be associated with increased neural response to reward anticipation and feedback (a positive association with BOLD [blood oxygenation level dependent] activation in one or more of the eight ROIs for reward anticipation and feedback) and decreased neural response to punishment anticipation and feedback (a negative association with BOLD activation in one or more ROIs when anticipating loss and when receiving feedback of loss and reward). Only one previous study has examined the role of Disinhibition in response to rewarding/non-rewarding stimuli and due to differences in the task and stimuli it is difficult to make direct comparisons to the current study (Foell et al., 2016). To our knowledge, this is the first study to examine how the triarchic model constructs of boldness, meanness, and disinhibition relate to neural response to monetary reward and punishment. Therefore, analyses involving triarchic model scales were exploratory. Main effects of sex were also examined. Based on previous research (van Honk et al., 2004; Welker et al., 2015) we predicted that males would show greater neural response to reward anticipation (i.e., M>F activation of one or more of the eight ROIs during the reward anticipation condition) and less neural response to loss anticipation and reward and loss feedback than females (i.e., F>M activation of one or more of the eight ROIs during the loss anticipation and reward and loss feedback conditions).

While we did hypothesize that, in males only, antisocial behavior would be positively associated with BOLD activation in one or more of the eight ROIs during reward anticipation and feedback and negatively associated with BOLD activation in one or more of the ROIs during loss anticipation and feedback, analyses examining moderation by sex were treated as exploratory. As there has been no previous research on the association between neural response to monetary reward and loss and antisocial behavior in females, we had no specific predictions about what the brain-behavior relationships would be unique to females. Similarly, effects of sex on response to reward and loss in relation to triarchic model scales were treated as exploratory.

Materials and Methods

Participants

Participants were 158 right-handed subjects aged 18 to 24 years (M age at date of scan = 19.8, SD = 1.40, 41.8% female) selected from an ongoing fMRI study of individuals recruited from the Michigan Longitudinal Study (MLS). Racial and ethnic composition of the sample was representative of the MLS dataset with the majority of participants identifying as white (88.6%) then African American (6.3%), Hispanic (2.5%), and biracial (2.5%). At the time of the scan 87% of the sample had completed High School and 54% of the sample had completed at least one year of college. The MLS is a community-based, longitudinal study of families with individuals at high risk for developing a substance use disorder (parental history of alcohol use disorder [AUD]) and lower-risk individuals without a family history of AUD who were living in the same neighborhoods (Zucker et al., 1996, Zucker et al, 2000). Sixty-one percent of the current study’s sample had one or both parents meet criteria for an AUD diagnosis during the participants’ lifetime, and 12% had a parent meet criteria for an ASPD diagnosis. Males and females did not differ on parental AUD diagnosis (X² (1, 157) = 1.11, p = .29) or parental ASPD diagnosis (X² (1, 140) = .49, p = .48).

Exclusionary criteria for the neuroimaging subsample of the MLS were: any neurological or chronic medical illness, any treatment with psychotropic medications currently or within the past 6 months, a history of psychosis or schizophrenia in first-degree relatives, presence of an Axis I psychiatric or developmental disorder except for conduct disorder, attention-deficit/hyperactivity disorder (ADHD), or substance use disorder. Using t-scored data from the Adult Self Report (ASR; Achenbach & Rescorla, 2003), 3.1% of the sample reported ADHD symptoms in the clinical range, 1.8% of the sample reported ASPD symptoms in the clinical range, and 4.1% reported substance use problems in the clinical range. Participants were instructed to abstain from ADHD medication and substance use for 48 hours before scanning and were given a multi-drug 5-panel urine drug screen prior to scanning to confirm this abstinence. The study protocol was approved by the University of Michigan Institutional Review Board, and all participants provided informed consent.

Measures

The NEO-Triarchic (NEO-Tri; Drislane et al., 2018) scales were developed to index the triarchic model constructs using items from the NEO Personality Inventory Revised (NEO-PI-R; Costa & Mac Crae, 1992). The NEO-PI-R administered closest to the date of the scan was used in analyses (M age = 19.72, SD = 1.41) of MLS data collection. There is strong evidence that scores on the NEO-PI-R have high test-retest reliability (Costa & McCrae, 2008; McCrae et al., 2011) and therefore while the measure was not administered at the time of the scan, there is evidence to suggest that the scores would be consistent. Each item was rated on a 5-point scale (1= strongly disagree to 5 = strongly agree) and displayed acceptable reliability (Table 1).

Table 1.

Mean, standard deviation, and reliability statistics.

	Full Sampl (N = 158)		Males (n =92)		Females (n = 66)
	M (SD)	α	M (SD)	α	M (SD)	α
Adult Self Report
Antisocial Personality Problems	54.14 (5.91)	.81	54.70 (6.64)	.84	53.36 (4.66)	.71
NEO-Tri Scale
Boldness	3.28 (.46)	.84	3.38 (.47)	.87	3.14 (.42)	.81
Meanness	2.41 (.49)	.79	2.58 (.45)	.77	2.18 (.43)	.79
Disinhibition	2.75 (.52)	.88	2.78 (.46)	.91	2.70 (.60)	.87

Note. M = mean, SD = standard deviation, α = Chronbach’s alpha, scores for the Antisocial Personality Problems scale are T-scored.

The Adult Self Report (ASR; Achenbach & Rescorla, 2003) assessment closest to each participant’s fMRI scan was used to measure emotional and behavioral problems (M age = 19.66, SD = 0.98). Again, while this self-report measure was not administered at the time of the scan, there is substantial evidence for the long-term stability of ASR scores (Achenbach & Rescorla, 2003). The ASR DSM-oriented Antisocial Personality Problems scale was used to assess symptoms of ASPD. The Mean Substance Use scale was used to determine sensitivity of effects. When sex was not included in analyses, T-scores for both scales were used.

fMRI task

Brain response during anticipation and receipt of incentive stimuli was probed using a modified version of the MID task (Heitzeg et al., 2014; Knutson et al., 2001; Yau et al., 2012). Each trial consisted of four stimuli presented consecutively: 1) 2000 ms incentive cue (“win $5.00”, “win $0.20”, “don’t lose $5.00”, “don’t lose $0.20”, or “no money at stake”), 2) 2000 ms anticipation fixation cross, 3) variable-duration (typically 200–300 ms) response target (i.e., a solid black shape cuing a button press), and 4) a variable-duration (typically 1700–1800 ms) feedback (e.g., “Correct response! You earn $5.00!”; “Incorrect response! No money at stake!”). There was no inter-trial interval; however, the trial order was pseudo-randomized. Therefore, due to the variability in the presentation order, the regressors for the different trials are still independent. The duration of the response target was calculated based on the individual subject’s reaction time during a practice session before scanning. The allotted duration was calibrated such that the overall success rate was approximately 60%. Two runs of the task were performed, each lasting 5 minutes and including 20 reward, 20 loss, and 10 neutral (i.e., no money at stake) trials. Participants were informed before the scan that they would receive any money that they won during the task (M = $39.93).

fMRI data acquisition and analysis

Whole-brain BOLD images were acquired on a 3.0 Tesla GE Signa scanner (Milwaukee, WI, USA) using a T2*-weighted single-shot combined spiral in-out sequence (Glover & Law, 2001) with the following parameters: 29 axial slices; repetition time: 2000 ms; time to echo = 30 ms; flip angle = 90°; field of view = 200 mm; 64×64 matrix; in-plane resolution = 3.12×3.12 mm; and slice thickness = 4 mm. A high-resolution anatomical T1 scan was obtained for spatial normalization (three-dimensional spoiled gradient-recalled echo, repetition time = 25 ms; minimum time to echo; field of view = 25 cm; 256×256 matrix; slice thickness = 1.4 mm). Participant head motion was minimized using foam pads placed around the head along with a forehead strap.

fMRI preprocessing

An iterative algorithm was used to reconstruct functional images (Noll et al., 2005; Sutton et al., 2003). Subject head motion was corrected using FSL 5.2.2 (Analysis Group, FMRIB, Oxford, UK; Jenkinson, Bannister, Brady, & Smith, 2002). Analysis of estimated motion parameters confirmed that overall head motion within each run did not exceed 3 mm translation or 3° rotation in any direction. All remaining imaging processing (including slice timing correction) and statistical analysis were completed using Statistical Parametric Mapping (SPM8; Wellcome Institute of Cognitive Neurology, London, UK). Functional images were spatially normalized to a standard stereotaxic space as defined by the Montreal Neurological Institute. A 6-mm full-width half-maximum Gaussian spatial smoothing kernel was applied to improve signal-to-noise ration and to account for differences in anatomy.

Individual analysis was completed using a general linear model (GLM). Two events were modeled: 1) the incentive cue and anticipation fixation, and 2) the response target and feedback. For the incentive cue and anticipation fixation event, five cues were modeled separately that each represented neural activation related to the participants’ anticipation of the amount they could win or lose on that trial, depending on whether their performance was successful: large loss, small loss, large reward, small reward, neutral [i.e., no money at stake]. The response and feedback event was modeled separately for each cue and feedback type: positive (i.e., hit) and negative (i.e., miss) for each of the large loss, small loss, large reward, and small reward conditions. The regressors were convolved with the canonical hemodynamic response function, with event durations of 4 s from stimulus presentation. Motion parameters were included as nuisance regressors to remove residual motion artifacts.

The main contrasts of interest in the current study during the anticipation phase were : 1) large loss anticipation vs. neutral (i.e., don’t lose $5.00 vs. no money at stake), 2) large reward anticipation vs. neutral (i.e., win $5.00 vs. no money at stake). During the feedback phase, we were interested in the following contrasts: 3) large loss positive feedback (i.e., during a $5 loss trial, a participant receiving positive feedback that they had responded quickly enough that there was no loss) vs. negative feedback (i.e., participant receiving feedback that they had responded too slowly and lost $5.00), herein referred to as No Loss vs. Loss, and 4) large reward positive feedback (i.e., during a $5 reward trial, a participant receiving feedback that they had responded quickly enough and won $5.00) vs. negative feedback (i.e., participant receiving feedback that they had responded too slowly and not won any money), herein referred to as Reward vs. No Reward. Large loss and reward contrasts were used as these have been found to provide a more robust BOLD signal in comparison to the small loss and reward contrasts (Bjork et al., 2010). Contrasts were calculated by linearly combining parameter estimates over both runs of the task.

Analytic plan

Based on a priori hypotheses, regions-of-interest (ROI) were defined in the limbic and prefrontal regions implicated in the anticipation and receipt of reward and loss (Oldham et al., 2018). The majority of ROIs were defined structurally, using the Wake Forest Pick Atlas template (bilateral nucleus accumbens, bilateral amygdala, bilateral thalamus, left orbiotofrontal cortex [OFC]). Additional, functionally relevant regions were defined by drawing two 10-mm-radius spheres for the left and right anterior insula and a 6-mm-radius sphere for the right posterior cingulate cortex centered around coordinates reported in recent meta-analytic work (Oldham et al., 2018). Average beta weights from 10 regions were extracted using MarsBaR (Brett et al., 2002) and imported into IBM SPSS Statistics v25 (IBM Corporation, 2018) for further analyses. Correlations were conducted with beta weights in ROIs, NEO-Tri scales, and ASR scales. Associations between ROIs and ASR scales were framed as disjunction tests (Rubin, 2021), and therefore the alpha level was adjusted to correct for multiple comparisons (p < .05/ 8 ROIs for each contrast = p < .00625). Exploratory analyses were put forth in the current study as individual tests (Rubin, 2021), and therefore an uncorrected alpha of p < .05 was used to determine significance for these analyses. As neural response to reward and punishment has already been found to be associated with substance use (Conrod & Nikolaou, 2016), the ASR Substance Use scale was used as a covariate in follow up analyses to determine if bivariate associations were unique to ASPD and/or triarchic model scales. Multivariate GLMs were run to examine the effect of sex and interaction between sex and self-report measures on brain response.

Results

Associations between MID activation, triarchic model scales, and ASPD symptoms

Correlations were used to examine the associations between brain response during loss and reward anticipation and feedback, NEO-Tri scales, and ASPD symptoms. Consistent with previous work describing associations between self-report measures, Boldness was not significantly associated with Disinhibition or Antisocial Personality Problems; however, it was negatively associated with Meanness (r = −.22, p = .006). Meanness was moderately associated with both Disinhibition (r = .43, p < .001) and Antisocial Personality Problems (r = .44, p < .001). In addition, Disinhibition was highly correlated with Antisocial Personality Problems (r = .56, p < .001).

When examining bivariate associations with brain response, inconsistent with our hypotheses, Antisocial Personality Problems was not significantly associated with neural response to reward anticipation, reward feedback, loss anticipation, or loss feedback. Boldness was positively associated with BOLD activation in the right anterior insula during anticipation of loss, with a similar, although non-significant association in the left anterior insula (Table 2; Figure 1). Meanness and Disinhibition scales were not associated with activation during the loss or reward anticipation contrasts. During feedback, Meanness was significantly associated with decreased bilateral amygdala activation during the No Loss vs. Loss condition and decreased left orbital frontal cortex activation during the Reward vs. No Reward condition (Table 2; Figure 1). There were no other significant associations between NEO-Tri scales and neural response during feedback.

Table 2.

Associations between BOLD activation in regions of interest during the monetary incentive delay task and psychopathic traits and symptoms of ASPD

Region	NEO-Tri Scales			ASR
	Boldness	Meanness	Disinhibition	ASPD
Anticipation: Loss vs. Neutral
L Nucleus Accumbens	.06	−.09	−.04	−.11
R Nucleus Accumbens	.06	−.15	−.13	−.07
L Amygdala	.11	−.14	−.02	−.05
R Amygdala	.02	−.02	.06	.03
L Thalamus	.11	−.03	−.04	−.04
R Thalamus	.08	−.04	−.01	−.01
L Anterior Insula	.15	−.11	−.11	−.23
R Anterior Insula	.18*	−.05	−.04	−.05
Anticipation: Reward vs. Neutral
L Nucleus Accumbens	.02	−.03	.04	.02
R Nucleus Accumbens	.05	−.09	−.04	−.01
L Amygdala	.04	.05	−.02	.06
R Amygdala	−.01	.07	−.02	−.03
L Thalamus	.11	.11	.04	.02
R Thalamus	.12	.07	−.01	−.01
L Anterior Insula	.12	.01	.06	.02
R Anterior Insula	.15	.05	.05	.08
Feedback: No Loss vs. Loss
L Nucleus Accumbens	.03	−.06	−.02	.04
R Nucleus Accumbens	.04	.00	−.08	.03
L Amygdala	.00	−.17*	−.07	−.03
R Amygdala	.12	−.18*	−.13	−.09
L Thalamus	−.01	.10	.10	.13
R Thalamus	.03	.08	.01	.07
L Orbitofrontal Cortex	−.06	−.02	.12	.13
R Posterior Cingulate Cortex	.04	−.01	.04	.10
Feedback: Reward vs. No Reward
L Nucleus Accumbens	−.05	.00	.06	.15
R Nucleus Accumbens	−.01	.01	−.07	.00
L Amygdala	.03	−.04	−.13	−.10
R Amygdala	.03	−.12	−.14	−.08
L Thalamus	.14	−.03	−.03	.03
R Thalamus	.09	−.04	−.01	.01
L Orbitofrontal Cortex	.05	−.19*	−.08	.07
R Posterior Cingulate Cortex	−.09	−.14	−.02	−.02

Notes. N= 158. L= left; R= right; ASR = Adult Self Report; ASPD= Antisocial Personality Problems scale from Adult Self Report (T-score). Significance threshold for associations between MID ROIs and NEO-Tri Scales is p < .05; significance threshold for associations between MID ROIs and ASPD is p < .00625;

^*denotes significant association.

Figure 1.

Scatterplots of significant bivariate associations between BOLD response during the MID task and psychopathic traits or ASPD symptoms. Associations in the Anticipation: Loss vs. Neutral contrast between a) BOLD activation in the right anterior insula and NEO-Boldness. b) Bilateral anterior insula activation during the Anticipation: Loss contrast. Association between c) BOLD activation in the right amygdala during the Feedback: No Loss vs. Loss condition and the NEO-Meanness scale and d) corresponding right amygdala activation during the Feedback: No Loss vs. Loss condition. Association between e) BOLD activation in the left orbital frontal cortex during the Feedback: Reward vs. No Reward contrast and NEO-Meanness and f) corresponding left orbital frontal cortex activation during the Feedback: Reward vs. no Reward contrast. * p < .05.

Sex Differences

Males scored significantly higher on self-report scales of Boldness (t₍₁₅₆₎ = −3.3, p = .001), Meanness (t₍₁₅₆₎ = −5.56, p < .001), and Antisocial Personality Problems (t_(155.7) = −2.93, p = .004). Separate multivariate GLMs for each contrast were used to examine the effect of sex on activation. Consistent with the individual hypotheses, during reward anticipation, males had significantly greater activation in the left amygdala (F_{(8, 149)} = 5.16, p = .02, partial eta² = .03), left and right thalamus (F_{(8, 149)} = 8.07, p = .005, partial eta² = .05; F_{(8, 149)} = 7.01, p = .009, partial eta² = .04, respectively), and left anterior insula (F_{(8, 149)} = 4.94, p = .03, partial eta² = .03). We did not find evidence for the hypothesized sex differences during the loss anticipation, reward feedback, or loss feedback conditions.

Multivariate GLMs were also used to probe significant sex by trait and sex by symptom count interactions associated with brain response during the MID task. There was a significant sex by Antisocial Personality Problems effect associated with left nucleus accumbens activation in the loss anticipation contrast (F_(8,147) = 5.24, p = .02, partial eta² = .03; Figure 2a). This effect was driven by a significant negative association between Antisocial Personality Problems scale scores and NAcc activation during loss anticipation in males (r_male = −.24, p = .02). In females, this association was nonsignificant (r_female = .16, p = .19.). In addition, there was a significant sex by Boldness interaction associated with left nucleus accumbens activation during the loss anticipation contrast (F_(8,147) = 7.50, p = .007, partial eta² = .05; Figure 2b). This interaction was driven by an opposing significant, positive association between Boldness and left nucleus accumbens activation in males (r_male = .24, p = .02) and nonsignificant association in females trending in the opposite direction (r_female = −.21, p = .09).

Figure 2.

Scatterplots of significant sex by ASPD or sex by trait interactions. a) Sex by ASPD effect associated with left nucleus accumbens BOLD activation during the loss anticipation contrast; b) sex by Boldness associated with left nucleus accumbens BOLD activation during the loss anticipation contrast; c) sex by ASPD effect associated with left nucleus accumbens BOLD activation during the loss receipt contrast; d) sex by ASPD symptoms effect associated with left amygdala BOLD activation during the loss receipt contrast. * p < .05. Male triangles and lines denoted in red, female dots and lines denoted in blue.

During the Feedback: No Loss vs. Loss contrast there was a bilateral sex by Antisocial Personality Problems effect for nucleus accumbens activation (left: F_(8,147) = 8.69, p = .004, partial eta² = .05; right: F_(8,147) = 7.18, p = .008, partial eta² = .05; Figure 2c). Females who scored high on Antisocial Personality Problems showed greater nucleus accumbens activation when receiving feedback that they had lost money in comparison to not losing money (left: r_female = −.30, p = .02; right: r_female = −.27, p = .03), whereas males demonstrated positive, although non-significant associations between Antisocial Personality Problems scores and nucleus accumbens response in this contrast (left: r_male = .20, p = .05; right: r_male = .17, p = .11).

There was also a significant sex by Antisocial Personality Problems interaction effect on left amygdala reactivity during the Feedback: No Loss vs. Loss contrast (F_(8,147) = 4.33, p = .04, partial eta² = .03; Figure 2d). Again, in females, higher Antisocial Personality Problems scores were associated with greater BOLD response in the left amygdala during loss feedback (r_female = −.27, p = .03) in comparison to feedback that they had not lost, while, inconsistent with hypotheses, Antisocial Personality Problems in males were not associated with left amygdala response during this condition (r_male = .09, p = .41.). We found no evidence to support our hypotheses that in males, antisocial behavior would be positively associated with BOLD activation during reward anticipation and reward feedback.

Discussion

The current study aimed to examine the possible moderating effect of sex on associations between neural response during the MID task and triarchic model traits as well as symptoms of ASPD. We found evidence for sex by ASPD symptoms and sex by Boldness interactions associated with neural response during loss anticipation and feedback regarding loss.

We did not find evidence for main effects of Antisocial Problems scores on reward and loss processing, inconsistent with our hypotheses. Previous studies have found ASPD symptoms to be associated with increased reactivity in the VS during reward anticipation (Bjork et al., 2012; Buckholtz et al., 2010; Geurts et al., 2016), while in the current study this effect was not replicated. However, as these previous studies utilized small samples (ns ranging from 24–34), such associations may have been spurious (Demidenko et al., 2021).

This is the first study to examine associations between triarchic model scales and neural activation in the MID task, and as such, analyses were exploratory. Nonetheless, significant associations at the bivariate level were detected that can contribute to our understanding of the interface between triarchic model constructs and neural processing of reward and loss. Higher levels of Boldness were associated with increased activation in the right anterior insula during the loss anticipation condition, suggesting that loss anticipation was perceived as more salient to bolder individuals. This finding is somewhat at odds with previous research that suggests that individuals high in Boldness demonstrate a weak BIS system (Fowles, 1980) and decreased sensitivity to punishment cues (Patrick et al., 2009; Ribes-Guardiola et al., 2020). However, as the current study did not directly probe BIS, it is unclear if this association is actually predictive of behavioral differences in the BIS system.

Results also showed that individual differences in Meanness were associated with neural response during the feedback conditions. Higher Meanness was associated with decreased bilateral amygdala reactivity during the Feedback: No Loss vs. Loss condition. As the amygdala responds to stimulus salience (Haber & Knutson, 2010), individuals high in Meanness may have found the monetary loss stimulus to be more arousing as opposed to the positive “no loss” feedback. Previous studies reported that adolescents with persistent disruptive behavior disorders demonstrated increased bilateral amygdala reactivity to loss in comparison to healthy controls and adolescents who had desisted from disruptive behavior (Cohn et al., 2015), although in that study this decreased association was not specific to those high in meanness (measured as callousness in adolescents). Here, Meanness was also associated with decreased left OFC activity during the Feedback: Reward vs. No Reward condition. Across studies, the OFC has been found to show increased response to reward receipt (Haber & Knutson, 2010). These findings suggest that individuals high in Meanness show normative brain response during the anticipation of loss and reward but demonstrate decreased reactivity to feedback involving positive cues (either a reward received or a loss averted). Lastly, Disinhibition was unrelated to neural responses. This suggests that deficits related to reward and loss processing in individuals who exhibit antisocial behavior may arise separately from disinhibitory tendencies and the deficits in inhibitory control the traits are thought to reflect.

Regarding effects of sex, our hypotheses were partially supported. Males had significantly greater neural activation during the reward anticipation contrast in limbic regions, including the left amygdala, bilateral thalamus, and left anterior insula, consistent with research implicating the role of testosterone in reward processing, although previously reported sex differences in the left striatum (Cao et al., 2019; Spreckelmeyer et al., 2009) were notably absent. When examining sex by ASPD symptom interactions, there was a significant effect for activation in the left nucleus accumbens during the loss anticipation contrast. When this interaction was probed, it was determined that for males, antisocial behavior was negatively associated with neural response to loss anticipation (Figure 2). These findings are consistent with foundational hypotheses about the role of punishment sensitivity in males with antisocial behavior (Fowles, 1980) and fit into the existing literature which suggests that high levels of testosterone decrease sensitivity to punishment (van Honk et al., 2004). In addition, there were significant sex by ASPD symptom interactions in BOLD response in the left nucleus accumbens and left amygdala during the Feedback: No Loss vs. Loss condition. When associations were examined by sex, for females, antisocial behavior was negatively associated with BOLD activity in both regions (Figure 2), suggesting a greater sensitivity to loss in comparison to the “no loss” feedback in females with more antisocial traits. These results suggest that females with high levels of ASPD symptoms appear to maintain their sensitivity to punishment, demonstrating increased reactivity in limbic regions to monetary loss in comparison to feedback that they had not lost.

When examining the interaction between sex and NEO-Tri scale scores, we found that a sex by Boldness interaction was associated with left nucleus accumbens response during the loss anticipation condition. This interaction was driven by a pattern in which higher levels of Boldness in males was associated with increased BOLD response while higher levels of Boldness in females was associated with decreased BOLD response in anticipation of monetary loss (Figure 2). This finding is somewhat inconsistent with previous research that found individuals high in Boldness demonstrate behavioral and psychophysiological insensitivity to punishment cues (Ribes-Guardiola et al., 2020; Vaidyanathan et al., 2009). The findings reported in the current study do parallel reports from one recent study (Ribes-Guardiola et al., 2020), which found that females high in Boldness demonstrated behavioral insensitivity to punishment feedback while in males, Boldness was not associated with a change in behavioral response. Also of note, Boldness was not associated with neural response to feedback regarding loss, suggesting that while there is increased left nucleus accumbens activation in anticipation of loss in males high in Boldness, their response when receiving feedback of loss in comparison to no loss is unrelated. In sum, findings suggest that there may be sex differences in how individuals high in Boldness respond to loss cues; however, Boldness does not then relate to neural response when receiving feedback, and it is unclear how this differentiation in neural response manifests behaviorally. Of the triarchic model constructs, Boldness is the least related to ASPD and is a trait that is uniquely predictive of psychopathic personality, when in the presence of Meanness or Disinhibition (Venables et al., 2014). The sex effects in the left nucleus accumbens during loss anticipation are a point of divergence between ASPD symptoms and Boldness (Figure 2) and highlight the importance of parsing heterogeneity when examining these constructs in future studies. However, as we did not have a priori hypotheses regarding this sex by Boldness interaction, these results should be viewed as exploratory, and further work is needed to replicate these findings.

Findings in the current study must be considered in light of certain limitations. First, this was not a forensic or clinical sample and therefore scores on both the Antisocial Personality Problems scale and NEO-Tri scales are somewhat restricted, potentially attenuating some effects. Second, the bivariate associations between MID activation and NEO-Tri scale scores were exploratory. Therefore, our results need to be replicated in larger samples with a wider representation of NEO-Tri scale scores and ASPD symptoms. Third, while study hypotheses were largely based on research that examined the influence of testosterone on neural response to reward, data on testosterone levels was not available in the current sample. In addition, a smaller subset of studies also suggests that fluctuations in levels of estrogen and progesterone in females may lead to enhanced neural response to reward (Sakaki & Mather, 2012). Therefore, future studies that collect biospecimens to examine how hormone levels relate to neural response to reward and loss as well as antisocial behavior are needed to further understand the etiology of these processes. In addition, the associations between triarchic model scales and MID activation were treated as independent tests and the alpha was not corrected for multiple comparisons (Rubin, 2021). Therefore, replication of these findings in an independent sample is a crucial next step towards generalizing these findings. Lastly, the analyses of the current study focused on specific regions of interest known to be relevant to reward and loss processing during the MID task. Future work is needed to determine if there are main and interactive effects of sex, antisocial behavior, and triarchic traits in other areas of the brain.

In sum, the current work highlights the importance of considering sex differences when examining reward and loss processing in the context of antisocial behavior and triarchic model traits. These findings also suggest that Boldness and Meanness show differential associations with neural response during the MID task, highlighting the utility of the triarchic model framework in research to understand the different etiological pathways contributing to psychopathic personality.