Clarifying fearlessness in psychopathy: An examination of thrill-seeking and physical risk-taking

Abstract

Fearlessness has been described among the traits associated with psychopathy; however, disparities in conceptualizations of fear and how related elements are operationalized among different measures of psychopathic traits have led to some enduring controversy. Here we address a subset of elements characterizing fearlessness represented in thrill-seeking and physical risk-taking among a large sample of incarcerated individuals (n = 688). We examine these relationships utilizing Hare’s Psychopathy Checklist Revised (PCL-R), the Psychopathy Personality Inventory-Short Form (PPI-SF), and the Zuckerman Sensation Seeking Scale (SSS). Among males, the Thrill and Adventure Seeking (TAS) scale of the SSS was not related to features of the PCL-R, but was highly correlated with Fearless Dominance (FD) of the PPI-SF. Among females, TAS was only modestly correlated with PCL-R total score, but neither of the two PCL-R factors. PPI-SF FD remained a strong predictor of TAS among females. We argue for a careful consideration of the boundaries of the term “fearlessness” as it relates to diverse manifestations of behavior and varied conceptualizations of psychopathy. We also reiterate apparent differences between males and females in the presentation of psychopathic traits.

1. Introduction

1.1. Historical Development:

Psychopathy is a severe disorder of personality, behavior, and cognition, characterized by a number of distinctive affective and interpersonal traits. Descriptive features of psychopathy have sometimes included the term fearlessness, but this term has a complex and often misunderstood relationship with psychopathy. This term lacks specificity, and may relate to disparate psychological and physiological features including impaired threat sensitivity, low anxiety, and shallow affective experiences. It has also been extended to encompass the conscious experience of fear along with a predilection physical risk-taking and adventure-seeking behaviors (see Hoppenbrouwers, Bulten, & Brazil, 2016). This under-specificity in conceptual language affects how psychopathy is quantified by some psychological instruments, and arguably contributes to certain disparities in empirical findings when considered in aggregate across varied populations.

Consideration of whether thrill-seeking and physical risk-taking behaviors are useful proxies for core elements of psychopathy requires an appreciation for the historical context of its early development. These roots reach back at least to Philippe Pinel (1745-1826), who described a group of patients afflicted with mania sans délire (insanity without delirium; Pinel, 1801). The term was used to describe individuals who had no conspicuous intellectual problems, but whose behavior was marked by cruelty, antisocial acts, alcohol and drug use, irresponsibility, and immorality. Other early descriptions included moral insanity (Prichard, 1835), which similarly emphasized deficits in emotions and behaviors rather than traditional cognitive impairments or psychosis. By the beginning of the 20^th century, these clinical descriptions were considered among a class of “personality” disorders accompanied by chronic antisocial and criminal behavior (Birnbaum, 1909; 1914).

Given the utility of these and other clinical descriptions, an academic meeting was organized by Karpman and held at St. Elizabeth Hospital in Washington, D.C., in 1923. A goal for this meeting was formalizing a consensus list of traits constitutive of a unified construct of psychopathy (see review in Kiehl, 2014). The list included thirteen items, including mendacity, flat affect, inability to feel love, guiltlessness, aberrant sexual behavior, atypical use of drugs and alcohol, and a strong developmental course. Subsequent lists of psychopathic traits informed by accumulating clinical observations included Cleckley’s now classic delineation of sixteen core characteristics of psychopathy, which overlapped conceptually with the above list and added a lack of nervousness or anxiety (Cleckley, 1941; 1976). This is perhaps the earliest historical origin of fearlessness in psychopathy, and plays a prominent role in our consideration of whether psychopaths are particularly prone to activities involving physical-risks and adventure-seeking behavior.

David Lykken is often credited as the first to study fearlessness in psychopathy in an experimental setting. His and others’ early experiments have anchored our modern neurobiological understanding of psychopathy in the observation of physiological deficits represented by poor acquisition of conditioned fear and low physiological arousal to aversive conditioned stimuli (Hare, 1978; Hare, Frazelle, & Cox, 1978; Lykken, 1957). Lykken (1995) subsequently developed the ‘Low-Fear Hypothesis’ of psychopathy, which has been highly influential in research guiding ongoing conceptual development of psychopathy and its relationship to personality traits, behavior, and psychopathology (see for example Triarchic Model; Patrick, Fowles, & Krueger, 2009).

1.2. The Neuroscience and Language of Fearlessness:

Extensive psychophysiological research has linked fear processing in the brain with an almond shaped anatomical structure known as the amygdala (Davis, 1992). Other experimental manipulations have confirmed a wide range of amygdala-mediated deficits associated with psychopathic traits, as quantified by various instruments. These findings include abnormal affect-mediated acoustic startle modulation (Anderson & Stanford, 2012; Carmen Pastor, Moltó, Vila, & Lang, 2003; Esteller, Poy, & Moltó, 2016; Patrick, Bradley, & Lang, 1993), and observations of impaired amygdala-mediated conditioning (e.g. Birbaumer et al. 2005; Blair et al. 2004). Paired with the amygdala’s prominent role in fear and anxiety, these findings represent some of the most consistent external correlates of psychopathic traits (broadly characterized), and have reinforced the concept of fearlessness among psychopaths. Support for the Low-Fear Hypothesis is thus anchored in psychophysiological and behavioral evidence that psychopaths often exhibit low anxiety and impaired utilization of emotionally salient information (particularly threat) in a number of contexts. Indeed, such evidence has played an influential role in conceptually differentiating psychopathy from other related diagnostic classes such as antisocial personality disorders (see Hare, 1996; Ogloff, 2006). The tendency to extend these data to encompass more contextually specific representations of fear, and a proclivity for physically dangerous activities is both common and potentially problematic.

Evidence from neuroscience indicates rather specific deficits in the encoding and/or cognitive utility of emotionally-relevant information for higher-order processes among those with psychopathic traits (Anderson & Kiehl, 2012). Under some conditions, psychopaths apparently process threat cues quite normally (Budhani, Richell, & Blair, 2006; Larson et al., 2013; Newman, Curtin, Bertsch, & Baskin-Sommers, 2010). Thus, the primary emotional deficits in psychopathy do not necessarily suggest a broad absence of fear under all circumstances, and these findings do not necessarily support the concept of fearlessness, as most people understand the term. Concordantly, LeDoux and others have advocated for more careful descriptive distinctions between processes involving early-stage, automatic threat processing (more closely associated with salience and attention) and the subjective, contextual experience of fear when describing psychopathic deficits (LeDoux, 2014; Hoppenbrouwers etal, 2016).

Extensions of early psychophysiological data to more general manifestations of fearlessness remain common, and were prevalent in Lykken’s original writings and other early distillations of the Low-Fear Hypothesis. For instance, Lykken had suggested that a number of intrepid heroic figures (typified by Chuck Yeager – the first man to break the sound barrier in an airplane) and some psychopathic murderers are cut from the same cloth, embodying an innate predilection for physical risk-taking (Lykken, 1982; Lynam & Derefinko, 2006). Claims like these have drawn significant critical attention from others in the field who challenge these descriptions of fearlessness in psychopathy (e.g. Neumann, Johansson, & Hare, 2012; Newman & Brinkley, 1997; Hoppenbrouwers et al., 2016; Levenson, 1990).

1.3. The Consequences of Semantics:

There remain prominent points of contention regarding the consequences of conflating elemental neurocognitive abnormalities evident in psychopathy with a higher order lack of fear in macro-behavioral contexts. This controversy plays out in the way we operationalize psychopathy in experimental settings, and further underscores the importance of construct validity between instruments developed to capture psychopathy’s core elements. Early research (e.g. Hare, 1978; Lykken, 1957) lacked a well-validated assessment of psychopathic traits until the development of the Hare Psychopathy Checklist and its revision (PCL-R; Hare, 1991; 2003). This instrument is usually carried out as a semi-structured interview paired with an evaluation of available collateral information allowing trained administers to score individuals on twenty criteria. The Hare PCL-R has become the standard instrument for measuring psychopathic traits in clinical practice and institutional contexts and remains among the most common tools utilized in research settings. Conceptually, the PCL-R has also occasionally been criticized for over-emphasizing formal criminal behavior. Some consider the presentation of antisocial features to be a contextual consequence, rather than a foundation of psychopathic traits (e.g. Skeem & Cooke, 2010, but see also Hare & Neumann, 2010, for counterarguments).

In addition to expert-rated instruments like the PCL-R, several self-report questionnaires have been developed to measure psychopathic traits. For example, Levenson and colleagues developed a 14-item scale that showed strong concordance to classic conceptions of psychopathy (Levenson, Kiehl, & Fitzpatrick, 1995). Lilienfeld and colleagues developed the Psychopathic Personality Inventory (PPI; Lilienfeld & Andrews, 1996) and its revised version (PPI-R) (Lilienfeld & Widows, 2005), motivated in part to assess psychopathic traits in normal-range, non-criminal populations. While intentionally avoiding reliance on formal antisocial behaviors, the PPI instruments also draw directly from behaviors concomitant with sensation-seeking and physical risk-taking: e.g. I might enjoy flying across the Atlantic in a hot-air balloon; I would find the job of a movie stunt person exciting. As a consequence, it tautologically formalizes these elements as constitutive of psychopathic traits (as does the PCL-R with formal antisocial behavior). As such, there are conceptual discrepancies that need to be considered when aggregating research that uses different instruments to quantify psychopathic traits.

Initial factor analyses of these instruments and their items revealed ostensibly similar structures. The Hare PCL-R scores are commonly separated into emotional-interpersonal characteristics (Factor 1) and lifestyle-antisocial characteristics (Factor 2) (Hare, 2003; Hare & Neumann, 2006). Analyses of the PPI-R also indicated two main factors accounting for emotional/interpersonal elements and impulsive/antisocial traits (Benning et al., 2003). Harkening back to the Low-Fear Hypothesis, the first PPI-R factor was dubbed “Fearless Dominance;” the second was named “Self-Centered Impulsivity.” Unfortunately, the confluence of these factor elements across measures has not been well supported, and the Fearless Dominance factor, specifically, has shown significant divergence from Factor 1 of the PCL-R (Miller & Lynam, 2012). It has been demonstrated that Fearless Dominance conceptually overlaps with features such as Boldness, an important element of the Triarchic Model of Psychopathy (Murphy et al. 2016). Nevertheless, experimental findings attributable to Fearless Dominance do not always align with findings attributable to PCL-R Factor 1. Divergences like these, driven by variation in the operationalization in psychopathy, can thus lead to failures in replication efforts, difficulties reconciling experimental outcomes, and problematic departures in theoretical interpretations drawn from different measures (Baskin-Sommers, et al. 2015; Maurer, Steele, Fink, Vincent, Calhoun, & Kiehl, 2018).

Considering various representations of fear, anxiety, and emotional limitations in psychopathy, Levenson and colleagues were the first to examine differences between psychopathic individuals and others who are prone to risk-taking, high-arousal, and adventurous activities (Levenson, 1990; Levenson, Kiehl, & Fitzpatrick, 1995). These studies found that thrill seeking (i.e., physical risk taking) was not significantly correlated with other self-reported psychopathic traits measured among a non-clinical, community sample. Others have also explored relationships between thrill-seeking and psychopathy as measured by different instruments. Factor 2 of the PCL-R positively correlated with the Thrill/Adventure Seeking (TAS) subscale of the Zuckerman Sensation Seeking Scales (SSS; Zuckerman, Eysenck, & Eysenck, 1978), but Factor 1 scores were unrelated to TAS (Harpur, Hare, & Hakstian, 1989). The PPI/PPI-R on the other hand, has previously shown significant correlations between thrill seeking (and other sensation-seeking elements) and both of its two factors in samples of university students (Benning, Patrick, Blonigen, Hicks, & Iacono, 2005) and inmates (Kastner, Sellbom, & Lilienfeld, 2012).

1.4. The Current Study:

These prior studies of risk-taking behaviors have typically measured psychopathic traits using only one instrument, and few studies have examined these relationships in samples representing clinical ranges on psychopathy. None that we are aware of have specifically examined gender differences in this conceptual intersection. The current study sought to compare two measures of psychopathy and their relationships with endorsement of physical risk taking behaviors in a large, incarcerated sample (n = 688), and to examine gender differences in these relationships. We examined concordance between elements of the PCL-R and the PPI (short form; PPI-SF), comparing relationships with the Thrill and Adventure Seeking (TAS) and other subscales of the SSS. Each measure examines related elements (factors) accounting for emotional and interpersonal features of psychopathy and behavioral styles accompanying those features. We hypothesized that sensation-seeking measures would be positively associated with both measures of psychopathy. However, we anticipated that thrill seeking behavior would be significantly positively correlated with Factor 1 (Fearless Dominance) of the PPI-SF, but that it would not be significantly correlated with features of psychopathy measured by the PCL-R.

2. Methods

2.1. Sample

Participants included n = 688 incarcerated adult offenders (405 males and 283 females) housed at three different local state correctional facilities. The participants’ age range was between 19 and 63 years of age (Mean 34.7, SD 8.5). Ethnicity and race were self-reported with 50% identifying as Hispanic and 50% identifying as non-Hispanic, while racial classification included 47% identifying as White, 1% as other/mixed, 8% American Indian, 9% African American, and 34% that did not wish to respond to that question (in accordance with the National Institutes of Health established categories of race and ethnicity). Exclusionary criteria for the study included traumatic brain injury accompanied with a significant loss of consciousness, personal or familial history of psychosis, or IQ scores under 70. All participants completed the PCL-R and SSS while 617 participants (278 female) completed the PPI-SF. Where data was unavailable for pairwise comparisons, participants were excluded analysis by analysis (pairwise).

2.2. Measures

The Psychopathy Checklist-Revised (PCL-R)

The PCL-R (Hare, 1991; 2003) is the most widely used and well-validated assessment for measuring psychopathy among incarcerated and clinical populations. The PCL-R was developed for use in forensic settings, comprises 20 items, each scored on a two-point scale, and has a recommended cutoff score of 30 (out of 40) for designation of psychopathy in forensic or clinical settings. The PCL-R evaluates different psychopathic traits by collecting information from criminal and medical files in conjunction with a semi-structured interview with the participant. These twenty items are commonly broken down into a two factor system.

• Interpersonal/Affective (Factor 1): including items such as glibness and lack of empathy

• Social Deviance (Factor 2): including items such as early behavioral problems and impulsivity

Psychometric properties of the PCL-R have been comprehensively vetted in prior literature (Hare, 2003; Hare & Neumann, 2006). The PCL-R is internally reliable across male offenders, male forensic psychiatric patients, and female offenders with ICC ranging from .86 to .97, alpha ranging from .81 to .85, and mean inter-item correlation ranging from .19 to .23 across samples (Hare and Neumann, 2006). For this study, all interviews were conducted by research staff who have undergone extensive training and supervision. All interviews were videotaped, and a portion of these are double-rated to ensure consistency and avoid rater drift. Interrater reliability for our trained staff is 0.96 (Intraclass Coefficient). Internal reliability coefficients (alpha) for the present sample were acceptable (.85 males; .79 females).

Psychopathic Personality Inventory-Short Form

Psychopathic traits were also assessed using the PPI-SF (Lilienfeld & Hess, 2001; Vaughn, Howard, & DeLisi, 2008). The PPI-SF was developed as an abbreviated (56-item) version of the longer original 187 item; (Lilienfeld & Andrews, 1996). The PPI is designed to identify psychopathic traits by self-report in a non-forensic setting, but has been validated in both institutionalized and community samples (Patrick, Edens, Poythress, Lilienfeld, & Benning, 2006). A total score for global psychopathy and eight subscales that culminate into two main factor scores are computed.

• Fearless Dominance (Factor 1): Social Potency, Fearlessness, and Stress Immunity

• Impulsive Antisociality (Factor 2): Machiavellian Egocentricity, Blame Externalization, Impulsive Nonconformity, and Carefree Nonplanfulness

The PPI-SF total score and the PPI total score have been shown to be significantly correlated with one another (r = .90) (Lilienfeld and Andrews, 1996; Lilienfeld & Hess, 2001). More recent, updated editions of the PPI-SF have subsequently been made available (Tonnaer, Cima, Sijtsma, Uzieblo, & Lilienfeld, 2013). Internal reliability (alphas) for the present sample were acceptable (.70 males; .73 females).

Zuckerman Sensation Seeking Scale

The Zuckerman Sensation Seeking Scale (SSS) form V (Zuckerman et al., 1978) is a self-report assessment comprising 40 questions that measure an assortment of risk taking behavior that loads on to four factors or sub-scales.

• Thrill and Adventure Seeking (TAS): desire to engage in sports or activities involving physical risks, speed, and danger

• Experience Seeking (ES): desire for experience through the mind and senses, travel, and a non-conforming lifestyle

• Disinhibition (Dis): desire for social and sexual disinhibition

• Boredom Susceptibility (BS): aversion to repetition, routine, and dull people.

The form V is a revision of Zuckerman et al.’s (1964) original measure. Because the elements of the SSS are presumed to align with attitudes about behavior, it is used both as a measure of the personality and an indicator of risk preferences. Reliability coefficients (alphas) in this sample were in acceptable ranges for both males (.78) and females (.82).

2.3. Control Measures

To assess IQ, trained researchers administered the Vocabulary and Matrix Reasoning subtests of the Wechsler Adult Intelligence Scale (Wechsler, 1997; Ryan, Lopez, & Werth, 1999). IQ scores ranged from 72 to 134 (Mean: 96, SD: 12.1) in the current sample. To assess traumatic brain injury, trained researchers administered a post-head injury symptoms questionnaire (adapted from King, Crawford, Wenden, Moss, & Wade, 1995) that assesses history, number, and duration of traumatic brain injuries, in addition to related symptoms. To assess psychosis, trained researchers administered the Structured Clinical Interview for Diagnostic and Statistical Manual of mental disorders, 4th edition (DSM-IV) for Axis I disorders (First, Spitzer, Gibbon, & Williams, 1997).

2.4. Data Analysis

Using SPSS (version 20) software we examined bivariate correlations comparing PCL-R total scores and factor scores with PPI-SF total scores and factor scores, as well as with Zuckerman’s SSS total scores and its sub-scales. A total of 18 pairwise comparisons were examined for males and females separately. In order to account for inflated risk of type-I errors, we give preferential attention to effects with probabilities of p ≤ .003 (.05/18 = .003). However, given the large sample (and consequently small p-values overall), attention to effect sizes across analyses is paramount.

Regression analyses were performed, examining the relative influence of factors from the PCL-R and PPI-R as predictors of scores on the TAS scale. Analyses examined the contributions of all four factors (from the two measures) together, followed by direct comparisons of PCL-R Factor 1 and PPI-SF Factor 1, in separate models.

Considering possible conceptual discrepancies between measures of psychopathy, Cohen’s Kappa’s were calculated for individual group membership in low, medium, or high categories of the PPI-SF and PCL-R. Groups were designated as occupying a range below .5 standard deviation from the mean (low), between −.5 and +.5 standard deviation from the mean (medium), and above .5 standard deviation from the mean (high), for each measure. Kappas indicate proportion of categorical agreement over and above chance (expected cell counts), and reflect the degree to which individuals are reliably designated as low, medium, and high scorers on each measure of psychopathy.

3. Results

Descriptive statistics across all measures are provided in Tables 1 and 2. Total scores for the PCL-R, SSS, and PPI-SF were all significantly correlated with each other. We also confirmed the expectation that PCL-R Factor 1 (interpersonal/affective) and PPI-SF Factor 1 (Fearless Dominance) scores were positively correlated with each other (Tables 3, 4), although they have a small effect size (r = .121 males; r = .173 females). We also found, as predicted, that the TAS subscale of the Sensation Seeking Scale was significantly correlated with PPI-SF Fearless Dominance scores in both males and females, but was not significantly correlated with PCL-R Factor 1 scores. Nor was TAS significantly correlated with PCL-R Factor 2. There were trends showing modest positive associations between TAS and PCL-R total and F2 scores among females only; however these effects were small (rs < .13) and did not achieve thresholds established for multiple comparisons.

Table 1.

Descriptive Statistics, Males

	Mean (SD)	Range
Psychopathy Checklist-Revised (PCL-R) Total	20.97 (7.4)	3.2 – 37.6
PCL-R Factor 1	6.26 (3.6)	0 - 15
PCL-R Factor 2	12.56 (4.0)	2 - 20
Psychopathic Personality Inventory Short Form (PPI-SF) Total	135.02 (12.9)	98 - 179
PPI-SF Factor 1	57.94 (8.68)	27 - 84
PPI-SF Factor 2	61.48 (10.2)	32 - 103
Sensation Seeking Scale (SSS) Total	20.86 (5.9)	5 - 38
SSS Trill and Adventure Seeking (TAS)	6.92 (2.8)	0 - 10
SSS Experience Seeking (ES)	5.72 (1.9)	1 - 10
SSS Disinhibition (DIS)	5.55 (2.5)	0 - 10
SSS Boredom Susceptibility (BS)	2.66 (1.9)	0 - 9

Table 2.

Descriptive Statistics, Females

	Mean (SD)	Range
Psychopathy Checklist-Revised (PCL-R) Total	18.54 (6.3)	2.2 – 34
PCL-R Factor 1	4.48 (2.7)	0 - 13
PCL-R Factor 2	11.98 (3.9)	0 - 20
Psychopathic Personality Inventory Short Form (PPI-SF) Total	130.96 (13.7)	93 - 173
PPI-SF Factor 1	55.87 (9.3)	31 - 83
PPI-SF Factor 2	60.37 (9.9)	32 - 89
Sensation Seeking Scale (SSS) Total	18.89	4 - 35
SSS Trill and Adventure Seeking (TAS)	6.02 (2.9)	0 - 10
SSS Experience Seeking (ES)	5.90 (1.9)	1 - 10
SSS Disinhibition (DIS)	4.75 (2.4)	0 - 10
SSS Boredom Susceptibility (BS)	2.20 (1.8)	0 - 9

Table 3.

Pearson Correlations between Psychopathy Measures and Factor Scores, Males

	PCL-R F1	PCL-R F2	PPI-SF Total	PPI-SF F1	PPI-SF F2	SSS Total
PCL-R Total	.848** n=405	.880** n=401	.372** n=339	.092 n=339	.360** n=339	.267** n=405
PCL-R F1		.534** n=401	.299** n=339	.121* n=339	.248** n=339	.186** n=405
PCL-R F2			.333** n=335	.036 n=335	.369** n=335	.284** n=401
PPI-SF Total				.591** n=339	.652** n=339	.487** n=339
PPI-SF F1					−.151** n=339	.361** n=339
PPI-SF F2						.285** n=339

^*p<.01

^**p<.001

Table 4.

Pearson Correlations between Psychopathy Measures and Factor Scores, Females

	PCL-R F1	PCL-R F2	PPI-SF Total	PPI-SF F1	PPI-SF F2	SSS Total
PCL-R Total	.753** n=283	.885** n=283	.359** n=278	.136* n=278	.362** n=278	.369** n=283
PCL-R F1		.433** n=283	.316** n=278	.173* n=278	.230** n=278	.218** n=283
PCL-R F2			.300** n=278	.047 n=278	.370** n=278	.360** n=283
PPI-SF Total				.663** n=278	.648** n=278	.586** n=278
PPI-SF F1					−.086 n=278	.456** n=278
PPI-SF F2						.386** n=278

^*p<.01

^**p<.001

In males, the relationship between TAS and PPI-SF Factor 1 was primarily driven by the PPI-SF F1 subfactor Fearlessness, which was strongly correlated with TAS, r(339) = .651, p < 10⁻⁶. Other PPI-SF F1 subfactors, Social Potency (r(339) = .168, p = .002) and Stress Immunity (r(339) = .165, p = .002) were also strongly correlated with TAS. Incidentally, Impulsive Nonconformity, an element of PPI-SF Factor 2, positively correlated with TAS, r(339) = .152, p = .005. No other PPI-SF subfactors were significantly correlated with TAS. Similarly in females, the relationship between TAS and PPI-SF Factor 1 was primarily driven by the PPI subfactor Fearlessness, r(278) = .701, p < 10⁻⁶. Social Potency (r(278) = .138, p = .021) and Stress Immunity (r(278) = .158, p = .008) were also correlated with TAS in females. The PPI-SF F2 subfactor Impulsive Nonconformity, positively correlated with TAS, r(278) = .283, p < 10⁻⁵ in females as well. No other PPI-SF subfactors were significantly correlated with TAS.

Regression analyses confirmed, when all four factors from both psychopathy measures were included, PPI-SF Fearless Dominance was the strongest simultaneous predictor of TAS, with no other predictors achieving significance levels (see Tables 7, 8). Considering only PPI-SF Factor 1 and PCL-R Factor 1, Fearless Dominance was the only significant predictor of TAS scores (Tables 9, 10).

Table 7.

Regression Analysis: Four Combined PCL-R and PPI-SF Factors Predicting TAS Scores, Males

Predictors	B	SE B	t	β	Sig.
PPI-SF Factor 1	.162	.016	10.381	.507	< .001
PPI-SF Factor 2	.011	.014	0.801	.042	.424
PCL-R Factor 1	−.107	.045	−2.413	−.138	.016
PCL-R Factor 2	.071	.040	1.781	.105	.076

R² = .253, R = .503, F(1,334) = 28.005, p < .001.

Table 8.

Regression Analysis: Four Combined PCL-R and PPI-SF Factors Predicting TAS Scores, Females

Predictors	B	SE B	t	β	Sig.
PPI-SF Factor 1	.167	.017	9.988	.526	< .001
PPI-SF Factor 2	.025	.017	1.519	.085	.130
PCL-R Factor 1	−.068	.062	−1.084	−.063	.279
PCL-R Factor 2	.066	.045	1.470	.088	.143

R² = .279, R = .528, F(1,277) = 26.389, p < .001.

Table 9.

Regression Analysis: PPI-SF Factor 1 and PCL-R Factor 1 Predicting TAS Scores, Males

Predictors	B	SE B	t	β	Sig.
PPI Factor 1	.161	.015	10.514	.501	< .001
PCL-R Factor 1	−.055	.037	−1.479	−.070	.140

R² = .248, R = .498, F(1,338) = 55.299, p < .001.

Table 10.

Regression Analysis: PPI-SF Factor 1 and PCL-R Factor 1 Predicting TAS Scores, Females

Predictors	B	SE B	t	β	Sig.
PPI Factor 1	.163	.017	9.737	.512	< .001
PCL-R Factor 1	−.003	.057	−0.060	−.003	.952

R² = .262, R = .512, F(1,277) = 48.769, p < .001.

Finally, agreement for categorical assignment into low, medium, and high groups based on total scores on the PPI-SF and the PCL-R were evaluated using Cohen’s Kappas. Kappas indicated poor agreement on categorizations. For males κ = .227, and for females κ = .206, indicating minimal agreement (<15%) on group assignment per individual. Generally acceptable levels of agreement are indicated by Kappa values over .6 (McHugh, 2012).

4. Discussion

Consistent with hypotheses, fearlessness as characterized by physical risk taking and thrill-seeking behavior measured by TAS of the Zuckerman Sensation Seeking Scale was not significantly related to Hare PCL-R Factor 1, or 2 in males or females. TAS exhibited robust positive associations with the Fearless Dominance factor of the self-report PPI-SF in both males and females. Indeed, this latter relationship among females (r = .512) was the strongest effect observed between TAS and all other variables studied, suggesting significant construct overlap between Fearless Dominance and TAS. Divergent relationships between TAS, the PCL-R, and PPI-SF, along with poor agreement in categorical assignments of individuals into ordinal groups, support prior suggestions that these measures of psychopathic traits are quantifying conceptually different constructs (Miller & Lynam, 2012).

This study also examined potential sex differences at the intersection of psychopathy and sensation-seeking. While noting largely similar effects across sex, PCL-R total scores were more closely associated with TAS scores among women than in men, but these differences did not carry over to the factor scores of the PCL-R. It is therefore possible that items which do not load on either factor are responsible for this association, namely promiscuous sexual behavior and many short-term marital relationships. Indeed, when these items are removed from PCL-R total score, the relationship is no longer significant (r = .10; p = .09). Gender differences in psychopathy have been noted elsewhere, particularly with respect to presentation of antisocial and borderline traits. Females score lower than males on psychopathy items relating to life-course persistent antisocial behavior (Bolt et al., 2004), and exhibit less interpersonal violence (but more self-directed violence) than men (Strand & Belfrage, 2005; Verona, Sprague, & Javdani, 2012). Antisocial items on the PCL-R accounting for violence and criminal offending may not be representative of female psychopathy to the same degree as male psychopathy. It has also been suggested that elevated borderline traits (e.g. self-destructive behaviors, unstable relationships) may be more reflective of a female-specific phenotypic expression of psychopathy (Sprague et al., 2012). In general, more research needs to be done to investigate gender differences in psychopathy, including how these traits present as external correlates, e.g. TAS and sensation-seeking.

Consistent with the present findings, previous literature has shown that the PPI-R and PCL-R show moderate correlations between their Total Scores as well as their respective Factor 2 (behavioral features) scores but a weaker correlation between their respective Factor 1 (interpersonal/affective) scores (Poythress et al, 2010). Miller and Lynam reported that Fearless Dominance more precisely measures stable extraversion and is “largely assessing psychological adjustment and adaptive functioning in a manner that is distinct from how the psychopathy construct has been conceived of and studied over the past 30 years,” (Miller & Lynam, 2012, p. 320). This lack of convergence between measures appears to be quite common in the field of psychopathy (see also Fink, Tant, Tremba, & Kiehl, 2012), but it is something that is not discussed often. The PPI and its derivatives have been developed under a different conceptual framework than the PCL-R and other self-report assessments of psychopathy (i.e., Levenson self-report scale; Levenson et al., 1993), appealing to, in part, more normative and adaptive representations of these traits (Lilienfeld et al., 2012). Indeed, it appears that, when it comes to fearlessness, the PPI (and its derivatives) is not assessing the same construct, as it intentionally characterizes these elements in a fundamentally different way, explicitly referencing thrill-seeking type behaviors at the item level. While this conceptualization aligns closely with the Triarchic Model of psychopathy (Murphy et al., 2016), the issue of divergence between measures remains an issue that affects empirical generalizability and replication efforts in diverse populations.

A recent meta-analysis examined the historical timeline of clinical descriptions of psychopathy and noted that regarding the concept of “fearlessness,” with the exception of Lykken’s Low Fear Hypothesis and the Triarchic Model of psychopathy, there has been no clinical theory of psychopathy that includes fear as an integral characteristic of the psychopathic personality (Hoppenbrouwers, Bulten, and Brazil, 2016). This is particularly puzzling in light of how the concept of fearlessness (or indifference to danger) has become entrenched in contemporary conceptualizations of psychopathy in both scientific and non-scientific circles. Hoppenbrouwers et al (2016) consider that “fear deficits” are more accurately attributed to impairments in threat detection and processing which does not require fear, per se, to be present. They also argue that fear should be differentiated between autonomic responses and higher order conscious operations, the former involving basic threat detection and the latter encapsulated by a cognitive recognition of being in a state of fear. After disentangling conscious from autonomic processes, they conclude that, although psychopaths exhibit context-specific deficiencies in threat detection and response in both behavioral and neurobiological measures, there is far less empirical evidence for any reduction in the ability to experience fear as a subjective cognitive state.

Indeed, there have been a number of theories that challenge certain fundamental assumptions in the Low Fear Hypothesis. The Paralimbic Model suggests that psychopaths exhibit neurocognitive difficulties brought on by protracted dysfunction in wider ranging brain areas than those specific to fear (Kiehl, 2006; Anderson & Kiehl, 2012). Likewise, the Integrated Emotion Systems (IES) model describes disrupted function across multiple intersecting neural systems important for emotional processing and cognitive control (Blair, 2005). The Response Modulation Hypothesis, and a number of closely related theories, suggest that features of attention are more fundamentally disrupted in psychopathy (Newman et al., 2010; Newman & Lorenz, 2003; Hamilton, Hiatt Racer, & Newman, 2015). These models suggest that rather than a global lack of fear (and other emotions), psychopathic individuals can demonstrate normal-range processing of affective information in many contexts, but they may have difficulty incorporating these signals into higher-order cognitive functions and modifying their behavior under certain complex demands as a result (Newman, Curtin, Bertsch, & Baskin-Sommers, 2010). Attention-based models can arguably account better for some variable findings in emotion-related studies of psychopathy and further embrace a more contemporary, integrative-neural network perspective for understanding cognitive-emotional dysfunction (Hamilton et al., 2015). Still, without careful attention to subtle construct dissimilarity between various measures of psychopathy, outcomes in ongoing research will likely continue to dilute consistency of empirical findings aimed at clarifying these models even further, especially across diverse samples.

We conclude by emphasizing that, contrary to descriptions by Lykken and others, primary responses to threat and neurocognitive abnormalities in the integration of affect/emotion into higher order decision-making do not necessarily lead to the kinds of physical risk-taking typified in thrill-seeking, adventure sports, and physically risky activities (Levenson, 1990). Nor is endorsement of such behavior particularly distinguishing of psychopathy (see also Blackburn, 1969; Hansen & Breivik, 2001). This assumption has likely been adopted by some to conceptualize psychopathic traits apart from formal antisocial behavior. It is perhaps reasonable to expect that low sensitivity to threat (a more universally accepted manifestation of psychopathic traits) may sometimes affect motivations to participate in physically risky activities. Conversely, rather than representing the kind of recklessness or impaired decision-making commonly typified in psychopathic traits, regular involvement in these activities requires keen insight into one’s own skill-level and experience and requires careful risk-management decisions. Indeed, investigations into the motivations of thrill-seekers and extreme sports enthusiasts often reveal aims of self-improvement and facing ones fears, rather than having none (Allman et al., 2009; Brymer & Oades, 2009; Lupton & Tulloch, 2002). As such, endorsement of these kinds of activities may just as easily be counter-indicative of psychopathy (see also Gridley, 1990), suggesting little overall discriminative utility in these items for identifying psychopathic traits. Ultimately, contexts under which low cognitive and physiological sensitivity to threat contributes to these downstream attitudes and behaviors requires further research.

Important limitations for this study include reliance on an incarcerated sample. While this sample allows examination of effects among those with relatively elevated psychopathy scores, this sample also differs from community samples in many other ways concomitant with behaviors and traits associated with incarceration. Further, this study was designed to examine only a narrow aspect of fearlessness in order to address a specific behavioral manifestation sometimes attributed to psychopaths. Additional studies should continue to examine other fear-related traits and behaviors with discriminating specificity as an explicit goal. In light of the present findings and numerous other studies examining various manifestations of fear and psychopathy, it remains apparent that there is something short of a cogent consensus for a precise definition of fear and fearlessness across studies (see also Hoppenpenbrowers et al. 2016; LeDoux, 2014). Here we demonstrate that one expression of fearlessness as represented by physical risk taking and thrill-seeking behaviors is not uniformly represented by two prominent measures of psychopathic traits; and further, these instruments have generally poor agreement on ordinal placement of those scoring high and low on each measure. It is beyond the scope of this research to claim what core elements should be constituents of the overall construct of psychopathy. We only emphasize that the only ground truth that may be appealed to is consensus among experts and agreement between instruments designed to operationalize this construct. The two instruments examined in this report fail to achieve this agreement with respect to behaviors and attitudes that some consider concomitant with fearlessness. We suggest that this demonstrates a source of potential construct shift, and inconsistencies in certain empirical findings, particularly across diverse samples. Finally, we advocate for more careful denotation of concepts related to threat processing, anxiety, emotional salience, and the experience of fear, especially as they are operationalized in neurophysiological experimental paradigms. Continued attention to these distinctions will promote more consistency and replicability across future studies.

Table 5.

Pearson Correlations (r) between Psychopathy Measures and Zuckerman SSS scales, Males

	TAS	ES	DIS	BS
PCL-R Total	.040 n=405	.131** n=405	.267** n=405	.288** n=405
PCL-R F1	.016 n=405	.098* n=405	.157** n=405	.248** n=405
PCL-R F2	.052 n=401	.132** n=401	.309** n=401	.267** n=401
PPI-SF Total	.283** n=339	.234** n=339	.383** n=339	.337** n=339
PPI-SF FD	.493** n=339	.173** n=339	.126** n=339	.049 n=339
PPI-SF SCI	−.029 n=339	.148** n=339	.342** n=339	.313** n=339

^*p<.01

^**P<.001

Table 6.

Pearson Correlations (r) between Psychopathy Measures and Zuckerman SSS scales, Females

	TAS	ES	DIS	BS
PCL-R Total	.125* n=283	.180** n=283	.396** n=283	.366** n=283
PCL-R F1	.082 n=283	.088 n=283	.224** n=283	.236** n=283
PCL-R F2	.115 n=283	.177** n=283	.400** n=283	.348** n=283
PPI-SF Total	.388** n=278	.359** n=278	.438** n=278	.448** n=278
PPI-SF FD	.512** n=278	.298** n=278	.165** n=278	.221** n=278
PPI-SF SCI	−.058 n=278	.236** n=278	.471** n=278	.371** n=278

^*p<.01

^**P<.001