“Gradually it was disclosed to me that the line separating good and evil passes not through states, nor between classes, nor between political parties either—but right through every human heart—and through all human hearts … And even in the best of all hearts, there remains … an unuprooted small corner of evil”

— Aleksandr Solzhenitsyn, The Gulag Archipelago Volume II, p. 615

“This story of ordinary men is not the story of all men… those who killed cannot be absolved by the notion that anyone in the same situation would have done as they did. For even among them, some refused to kill and others stopped killing. Human responsibility is ultimately an individual matter”

— Christopher Browning, Ordinary Men, p. 188

The Problem of Evil

What is true evil for a consequentialist? To put it another way, what tendencies in an agent would we expect to result in highly negative outcomes?

Drawing from historical examples and personal experience, one may come up with a number of traits. These may include “cruel”, “manipulative”, “egoistic”, “violent”, and “greedy”. But even if we constrained ourselves to just these traits, it’s not entirely clear which are the most consequential. For example, while highly violent individuals may inflict greater direct physical harm on others, they may ultimately tend to be relatively inconsequential, since they may also tend to be incarcerated or killed by others. We should therefore think carefully about which traits are most important to consider, at what level they are the most dangerous, and how they interact with other traits.

It should also be noted that current measurement instruments appear to have a symptom-focused approach: they identify the most saliently harmful individuals, but not necessarily the most harmful ones. For example, many measures of Psychopathy have high impulsivity as a key feature; however, Psychopaths with low impulsivity may actually do more harm in the medium- and long-term (Althaus & Baumann, 2022; Katz et al., 2022). This is a reason not to use established measurement instruments naively.

One of the purposes of this article is to sketch out what an “evil” personality profile might look like. In doing this, I draw inspiration from Latent Profile Analysis (LPA; Spurk et al., 2020). In the case of personality, LPA may be used to identify subgroups within a population. For example, a “comic relief guy” personality profile might be defined by low conscientiousness, high extraversion, high agreeableness, and medium neuroticism. While a trait-based approach may overemphasise extreme levels of traits, a LPA-like approach allows for the significance of relatively moderate levels of traits, which may be important for our purposes. A LPA-like approach is also useful for creating less arbitrary cutoffs, which in turn is useful for practical applications of psychometric instruments.

A Note on Terminology

It has been suggested that less loaded terms should be used when discussing this subject. There are a couple of major reasons for this. First, traits are adaptive and maladaptive to varying degrees, and a loaded term such as malevolence implies a lack of fitness. Second, loaded terms seem to lack compassion or understanding for those they refer to. Currently, the terms used tend to be “dark” or “malevolent”, but it has been suggested that more neutral terms be found or constructed.

Though I am sympathetic to this impulse, I do not agree with this line of thinking. The word “malevolent” appears to be etymologically accurate to its current cultural meaning, coming from the Latin male-, meaning bad, and volens, meaning to will or to wish (according to the Merriam-Webster dictionary); essentially, malevolence is ill-intent (to others). This is what traits such as spite, callousness, and Sadism seem to capture. To say someone is malevolent—that they have ill-will—is not itself necessarily a judgement; rather, it is we who infer the judgement.  

Furthermore, it does not seem to be the word that makes individuals feel uncomfortable, but what it represents. Consider the word “Psychopath”. This doubtless conjures up quite a negative impression. However, the word “Psychopath” comes from the German psychopastiche, meaning “suffering soul” (Kiehl & Hoffman, 2011, Section I part B). The true meaning of the word thus differs wildly from its current cultural meaning. Still, the word has acquired a highly negative connotation because of the concept it has been associated with. This suggests that the underlying meaning of a construct like malevolence corrupts any word which is assigned to represent it. We should therefore stick to the word which is not only popular but appropriate for describing what we seek to understand: malevolence.

Malevolence

Perhaps the most popular conception of malevolence is the Dark Triad, which is composed of Psychopathy, Machiavellianism, and Narcissism (Paulhus & Williams, 2002) . In this framework, the Dark Triad traits can be shortly described as such:

• Psychopathy: a blend of callousness and disinhibition.
• Machiavellianism: self-interested manipulativeness
• Narcissism: a blend of grandiosity and entitlement/rivalry

More recently, the trait of Sadism (the enjoyment of others’ suffering) has been added to this constellation to form the Dark Tetrad.

The Dark Triad/Tetrad has often been implicitly used as a model which represents all of dark personality. However, to my knowledge, there has never been an empirical test which has evaluated the representativeness of the Dark Triad/Tetrad (Kowalski et al., 2021, see “What Defines a Dark Trait?”). It is therefore possible that the Dark Triad/Tetrad framework does not capture substantial aspects of dark personality.

Another issue with this framework is that measures with similar names are often conflated when they appear to measure importantly different aspects. For example, in the popular conception, a Psychopath might be considered as being an individual who is cold and manipulative, superficially charming, calculating, and imperturbable (think Hannibal Lecter); this is close to what is often termed “Primary Psychopathy”. However, there is also a “Secondary Psychopathy” which captures impulsivity, sensation-seeking, and criminality. These two Psychopathies may share significant overlap in uncaringness towards others, but appear to pull almost in opposite directions in some other respects (Sellbom & Drislane, 2021, p. 3). For example, “Primary Psychopaths” are characterised as being cold and emotionally stable, while “Secondary Psychopaths” are characterised as being impulsive, antisocial, and emotionally unstable (p. 4). Still, there is a tendency to conflate these two because they both refer to “Psychopathy”. This kind of issue appears to be most salient in Psychopathy (Sellbom & Drislane, 2021) and Narcissism (Ackerman et al., 2011), but affects Machiavellianism and Sadism as well. 

There are also concerns that measures which are supposed to capture different traits are actually measuring practically the same thing (Muris et al., 2017, pp. 193–195). In particular, there has been a longstanding debate on whether or not Machiavellianism can be distinguished from Psychopathy (e.g., McHoskey et al., 1998). More recent evidence suggests that attempts to disentangle the new addition of Sadism from Psychopathy have been mostly unsuccessful (Blötner & Beisemann, 2022, pp. 3–5).

This suggests that the Dark Triad/Tetrad framework is not an optimal model of dark personality. Looking closer sheds additional light on why this might be the case. Each of the Dark Tetrad traits has a number of aspects/subtraits, and many of these appear to be quite similar to one another (Bader et al., 2023, Tables 2 and 3). In particular, analyses of the Dark Triad have tended to find that Psychopathy is central to the Dark Triad and can explain the variance of substantial portions of the other Dark Triad traits (Glenn & Sellbom, 2015, pp. 193–195; Muris et al., 2017). Further analyses have found that (sub)traits including callousness, interpersonal manipulativeness, Machiavellianism, Narcissistic rivalry, and spitefulness are quite central and highly related to one another (Dinić et al., 2023; Marcus et al., 2018; Trahair et al., 2020). This suggests that these traits are part of a “core of darkness” which may express itself to varying degrees in a wide variety of traits which are considered to be interpersonally toxic.

This is generally the philosophy that a new framework of malevolence—the Dark Factor—has championed. The Dark Factor has been defined as “individual differences in the tendency to maximize one’s individual utility—disregarding, accepting, or malevolently provoking disutility for others—accompanied by beliefs that serve as justifications” (Moshagen et al., 2018, p. 656, 657). It is suggested that D can be broken down into five facets (Bader et al., 2021, Figure 1): 

• Callousness: a lack of empathy but emphasising indifference towards others’ welfare
• Sadism: provoking disutility for others for one’s own pleasure, often with feelings of power/dominance as motivators
• Vindictiveness: resentfulness and strong vengefulness
• Deceitfulness: tendency towards deceptive and/or rule-breaking behaviors
• Narcissistic Entitlement: a feeling that one is more deserving than others

There are differences in the extent that specific malevolent traits (and even their subtraits) are captured by D. Traits which have been recognised as uniformly malevolent—such as Psychopathy and Sadism—appear to be almost completely represented by D (Moshagen et al., 2020, Table 2), while traits which are more neutral or composed of aversive and non-aversive components—such as Narcissism—are substantially but only partially represented (Moshagen et al., 2020, Table 2; Neumann et al., 2022, Table 2).

I suggest that willingness to cause disutility to others should be the primary focus of D. Many malevolent characteristics, such as callousness, do not entail self-utility maximisation; some traits, such as spitefulness, even index a willingness to harm oneself in order to harm a targeted other. D also appears to be especially good at capturing willingness to cause disutility to others as opposed to capturing self-utility maximisation traits such as greed (Horsten et al., 2021, see “General Discussion”).

Moshagen and colleagues hypothesise that D is analogous to general mental ability (g or GMA) in that both are fluid or reflective constructs (Moshagen et al., 2018, p. 658). In the case of general intelligence, correlations between performance by a variety of seemingly unrelated metrics—such as general job performance, reading comprehension, and cognitive tests—can be traced to a common (causal) root factor: one’s ability to deal with cognitive complexity/complex information processing (Gottfredson, 1997, pp. 92–93). Each metric reflects this general mental ability in its own way: most jobs value quick learning and problem-solving skills, cognitive tests may examine pattern recognition and/or working memory, and reading comprehension measures ability to quickly process/synthesise written information. Similarly, all malevolent traits seem to reflect (i.e., are caused by) a willingness to tolerate or cause disutility to others.

D vs. Psychopathy

In numerous analyses of the Dark Triad/Tetrad, Psychopathy seems to be the most consistently “central” trait. Moreover, Psychopathy appears to have the key characteristics that define D, such as callousness and antagonism. So why not just use Psychopathy?

The main advantage of D appears to be theoretical. Practically speaking, I know of no evidence that suggests the “gold-standard” measures of Psychopathy and D substantially differ in usefulness. However, traditional measures of Psychopathy (e.g., Paulhus et al., 2009) include aspects of high boldness, low inhibition, and (sometimes) high erratic emotionality, all of which may obscure or distract from the callous affect facet. D also predicts how various malevolent traits relate to one another (i.e., through a shared dark core) and thus appears to be a more global model of dark personality. Thus, D seems to have theoretical advantages over Psychopathy which may make it more useful for formulating subsequent predictions regarding dark personality. However, since Psychopathy is an older construct which has been more widely tested and used than D, I may invoke it to make predictions about malevolence where there are insufficient data to use D. 

On Uncaring

Why We Should Expect High Levels of Malevolence to be Robustly Bad

In practically any system of morality that is taken seriously in the world today, individuals with high levels of malevolence are bad. From a utilitarian perspective, they are highly willing to cause great disutility for arbitrary or selfish reasons. From a rights-based perspective, they have no respect for the rights or intrinsic worth of others. From a virtue-ethics perspective, they are unwilling to form the correct motivations to live a moral life. From a contractarian perspective, they have no respect for promises made to others. Thus, they appear uniquely robustly bad.

The one socially positive outcome of malevolence might be in the service of a “greater good”. However, I think even here the costs tend to outweigh the benefits (particularly for extremely malevolent individuals). Individuals high in malevolent traits will tend to be deceptive, overselling their own usefulness and altruism while hiding their true intentions and negative qualities. These intentions may include taking over an organisation to gain resources/power for themselves. The primary uses for such traits—such as disempowering/eliminating a rival group, forcefully/duplicitously acquiring resources, and forcefully establishing order—themselves tend to be harmful activities, such that even “successful” use of malevolent traits has substantially harmful effects. Evil that is done in the name of a greater good is no less evil; it is at best only a tragic necessity.

While the Dark Core of personality has been defined—and can be thought of as a singular construct—it may be instructive to further explore the structure of malevolence, as there seem to be some subtle differences which might have important implications. Below, I list a few distinctions which are based on my review of dark personality and Psychopathy.

Callousness vs. Antagonism

Unfolding of the factor structure of D appears to reveal that Narcissistic Entitlement and Deceitfulness tend to be the most distinct from the other traits (we will not focus on these facets), while Callousness, Sadism, and Vindictiveness are more closely related (see Table S3 and Table S4; both from Bader et al., 2021). However, across both analyses, Sadism and Vindictiveness both share relatively high similarities with Antagonism compared to Callousness (which has low or even slightly negative similarity with the other two!). My intuition is that the difference is this: Antagonism is the “positive” drive to cause others harm, while Callousness is the “negative” lack of caring about (or ability to disregard) others’ well-being.

One might compare callousness to congenital insensitivity to pain and anhydrosis (CIPA; the congenital inability to feel physical pain). Individuals who are highly callous are not necessarily malevolent just as those with CIPA are not sado-masochistic: rather, much of the harm results from a lack of automatic and affect-laden awareness that results from certain actions (Blair, 2005, pp. 874–878). Additionally, like how individuals with CIPA do not conventionally learn not to bite their tongues or scald their hands, highly callous individuals are most likely impaired in learning to avoid harming others. This is likely because their automatic, affective recognition of having done harm (i.e., a sad/fearful/hurt expression directed towards them) is impaired. Where individuals with CIPA would need to develop compensatory habits to avoid potentially life-threatening self-injury, individuals high in callousness would need to develop analogously compensatory habits to avoid falling into chronic antisociality.

I think grasping callousness is less intuitive than grasping antagonism. Though even the kindest people have probably had the urge to hurt someone in anger, I think most people are held back by compassion or feel regret after the fact. Even when we think of villains—both real and imagined—I think we think of rage, sadism, instability, cynicism, and Narcissism, but forget callousness. 

Though callousness is quiet and unassuming, one may think of it as the “midwife of malevolence”, as it removes the guardrails which generally hold us back from harming others. I hypothesise that this extends to all types of harm, including uncaring harm (e.g., cheating), reactive harm (e.g., vengefulness) and instrumental harm (i.e., exploitation). 

Additionally, someone with affective feelings such as guilt, empathy, and compassion is in some sense precommitted to feeling an attraction towards the good. Theoretically, non-callous individuals are those who can be morally reasoned with to a significant extent: that is, even if they are in a position to exploit others and can benefit (by all utilitarian calculations) from doing so, they may be persuaded against such exploitation by appeals to compassion. By contrast, a completely callous individual can only be restrained by ignorance or (external) force: either they must not care enough to exploit, or they must be credibly persuaded that exploitation would not be to their personal benefit.

Reinhard Heydrich: A Case Study

Reinhard Heydrich, described by Hitler as “the man with an iron heart”, is perhaps one of the more anomalous members of the mass murderer club. I know of no anecdotes which indicate that he had the passion of Hitler, Himmler, or Goebbels; nor did it appear that he had the personal malice or paranoia of someone like Stalin. It may not be possible to determine whether or not this lack of anecdotes is due to confounds such as Heydrich not being as well-known. Still, it is possible that Heydrich is an archetype of highly callous and relatively non-antagonistic individuals who have the potential to quietly and effectively cause mass harm.

There appears to be little question about Heydrich’s coldness. Heydrich had few friends, and maintained strictly professional relations with his colleagues in the SS. His vision for the mass murder of the Jews in Nazi Germany and its satellites was quite clinical:

Under appropriate direction the Jews are to be utilized for work in the East in an expedient manner in the course of the final solution. In large (labor) columns, with the sexes separated, Jews capable of work will be moved into these areas as they build roads, during which a large proportion will no doubt drop out through natural reduction. The remnant that eventually remains will require suitable treatment; because it will without doubt represent the most [physically] resistant part, it consists of a natural selection that could, on its release, become the germ-cell of a new Jewish revival. (see the notes of the Wannsee Conference)

Heydrich was not entirely free of ideology. Though he was relatively uninterested in politics in his early years, it appears his immersion in the Nazi sociopolitical circle did deeply affect him. He came to believe in the cause of murdering Jews and other ethnicities and minorities targeted by the Nazis. Perhaps here, Heydrich’s “ruthless determination and burning ambition” (Jasch & Kreutzmüller p. 60) partially explains this shift. He desired success, and by the metric of the time and place, success was measured by one’s position in the Nazi hierarchy and one’s commitment to the Nazi ideology; and he was willing to do whatever it took to be successful.

On the other hand, he does not seem to have been particularly sadistic, vengeful, or otherwise passionate about harming others. Unlike many of his colleagues, he did not join the Nazi Party out of passion or outrage at the aftereffects of WWI; rather, he joined because his fiancée (an ardent Nazi) persuaded him it was a good opportunity and because he had few other options (pp. 58–62). And though he suffered a number of significant slights—including a dishonorable discharge from the Navy, a humiliating investigation into his (rumored to be Jewish) heritage, and his eventual assassination—there is no note in any of these instances that he attempted to take retributive action, or even harbored substantial resentment towards involved parties.

Thus, Heydrich appears to represent a cluster of individuals who may go underappreciated because they do not display the most flashy signs of malevolence—such as antagonism or entitlement—but who may nevertheless be instrumental in the initiation or perpetuation of harm because they just don’t care.

Reactive vs. Instrumental Antagonism

In the literature on Psychopathy from a neuroscientific perspective, there is an important distinction between reactive vs. instrumental (or proactive) aggression (Blair, 2005, pp. 867–868; Raine, 2006). Reactive aggression is aggression which is triggered by fear or frustration and is associated with a perceived transgression or threat. Instrumental aggression, by contrast, is not motivated by threat but by premeditated, goal-directed anticipation of some reward; a salient example would be aggressing with the aim of obtaining another’s resources (hence “instrumental”). Instrumental aggression—as well as the lack of empathy/guilt associated with it—is a specific defining feature of Psychopathy; Blair (2005) notes no other psychopathology demonstrates significant instrumental aggression (p. 868).

Interestingly, there appears to be a mapping from this distinction to the distinction of the two descendants of Antagonism: Vindictiveness and Sadism. Vindictiveness can be conceived of as an extreme of reactive aggression: it is antagonism motivated by frustration (resentment) and/or in response to threat (vengefulness). Sadism can be conceived of as an extreme of instrumental aggression, as it is the inflicting of pain (other-disutility) to derive pleasure (self-utility).

While I don’t doubt that a host of negative outcomes can result (and have resulted) from reactive antagonism alone, I do think that the game-theoretic landscape might be made substantially less hostile with the removal or restraint of instrumentally antagonistic actors. I would also guess that a disproportionate number of conflicts (between individuals and groups) are started by individuals who are instrumentally antagonistic and then continued by reactively antagonistic individuals. However, this is merely speculation from my intuition, and more rigorous and externally valid models would be needed to confirm or falsify this hypothesis.

Note that this division may be useful in allowing us to selectively decrease a relatively maladaptive form of antagonism (instrumental) while retaining (to a certain degree) a less maladaptive form of antagonism (reactive). One worry of reducing antagonistic traits is that there may be a danger of inviting exploitation or destruction; however, the aforementioned division suggests that a more careful process would allow for the reduction of instrumental antagonism while still maintaining defenses against exploitation.

It’s also worth noting that reactive antagonism also has its dangers, as an agent who is extremely retributive will, in certain circumstances, be driven to the creation of massive disvalue, perhaps even at their own expense.

Putting It Together

How do callousness, instrumental antagonism, and reactive antagonism fit with one another?

I think callousness most likely encompasses instrumental antagonism. Perhaps some individuals are callous, but do not care much even for their own utility, and thus tend not to be instrumentally antagonistic towards others. However, deferring to Blair (2005), I think at most a trivial proportion of instrumentally antagonistic individuals will not be highly callous.

On the other hand, I think callousness and instrumental antagonism substantially (but not entirely) overlap with reactive antagonism. What’s interesting here is that reactive antagonism appears to be associated with greater reactivity to fear/threatening stimuli, while callousness/instrumental antagonism appears to be associated with lesser reactivity. This would suggest a negative association between the two. However, callousness may still be positively associated with reactive antagonism: perhaps a callous individual is less easily perturbable, but at the same time, even a comparatively small perturbation to him could result in disproportionate retaliation because of a lack of any countervailing force.

Given this, we might expect their relations to look something like this:

Malevolence in the Real World

Overall mean = 2.62971

Overall standard deviation = 0.6809548

5%ile = 1.5625; 95%ile = 3.8

Data and visualisation are courtesy of Dr. Morten Moshagen and colleagues at darkfactor.org

The General Population

This is a graph which represents the rough distribution (N = 1,982,574; ~57% female) of Dark Core levels in the general population; presumably, anyone who has completed the scale and given consent for it to be used is represented. This graph suggests a bell-curve distribution that is somewhat positively skewed (the 50th percentile is 2.614286, while the scale midpoint is 3). This also suggests that it would be theoretically feasible to identify and exclude exceptionally malevolent individuals from important positions (one might have the cutoff at scores higher than 3.8, for example) without affecting a large subset of the overall population (in this case, ~5% of the population would be affected).

Pareto Strikes Again

Though I am not aware of any statistics regarding the Dark Factor, those regarding Psychopathy paint an expected picture: though only 1% or less of those in the North American general population qualify as clinical psychopaths, an estimated 15% to 30% of those in the North American incarcerated population are Psychopaths (Kiehl & Hoffman, 2011, Section I Part A; van Dongen, 2020, “Psychopathy: An Overview”). One study estimated that a soft lower bound for the societal costs of Psychopathy was 460 billion USD per year, greater than the costs of alcohol-substance abuse, obesity, smoking, and schizophrenia (Section B). A more recent study estimated the annual cost of Psychopathy in the United States to be between 245.5 billion and 1.59 trillion USD (Gatner et al., 2023). Note that to be considered a clinical Psychopath, one must be scored >= 30 on the PCL-R; this means that many individuals who have substantial scores are still considered non-Psychopaths in this analysis. 

Given these data, and given the apparent positive skew of the dark personality distribution, we should naively expect that individuals with the highest levels of malevolent traits to cause a hugely disproportionate amount of harm in a more general sense. That is, perhaps individuals in the >99%ile range for malevolent traits tend to cause order(s) of magnitude more harm than average. This appears good for tractability, as removing only a tiny minority of the population from positions of power can have disproportionately beneficial results.

I say naively because there are undoubtedly confounding factors which I am not accounting for. I am unsure which way, if any, they would shift this conclusion. Some factors, such as systemic/institutional constraints on leaders, might reduce the expected proportion of harm from highly malevolent individuals. Some factors, such as the potential for lasting cultural/institutional damage, appear to increase it. Until contrary evidence is produced, however, I think we should assume that a greatly disproportionate amount of general harm will be driven by a small subset of actors.

How is Malevolence Adaptive?

If highly malevolent individuals are so generally bad, then why do they exist? Conversely, if they out-exploit everyone, then why isn’t everyone malevolent?

Glenn, Kurzban, & Raine (2011) reviewed available evidence on the evolutionary basis of Psychopathy and identified a few potential explanations for the preservation of phenotypic diversity. These include differences in environmental pressures, frequency-dependent selection, and polygenic mutation-selection balance.

Perhaps a decent illustration of the effects of differential environment is the case of sickle-cell anemia in malaria-endemic environments. Despite the potential disbenefit this causes the carrier and/or their offspring, moderate levels of sickle-cell work well enough as a deterrent against malaria for sickle-cell traits to survive and propagate themselves. Similarly, certain amounts of coldness, selfishness, vengefulness, and conniving appear to have been adaptive throughout history in many societies/environments. For example, one theory regarding cultures of honor states that the retributivism found in honor cultures is adaptive largely because it acts as a deterrent against future threats/aggressive agents (Nowak et al., 2016). The analysis conducted by Nowak et al. also suggests that aggressive behavior is adaptive to the extent that it is not retaliated against (either by law or by other individuals; see “What happens to the population when honor agents are not present?”). This appears to be in line with theories which conceive of the Dark Triad (Jonason, Koenig, & Tost, 2010), and specifically antisociality, aggression, and Machiavellianism (McDonald, Donnellan, & Navarette, 2012), as arising from a “fast” life history strategy which emphasises immediate survival over future reproduction/parenting. This suggests that the continued existence of malevolent traits is due to factors at both the individual and group levels.

On top of this, malevolent traits may be frequency-dependent, as the proportion of highly malevolent individuals in the population has an influence on the fitness of malevolence. Malevolent individuals may flourish in an environment where there are few other malevolent individuals because they can exploit others without fear of retribution/running into other malevolent individuals. However, malevolent individuals may tend to do worse in environments with many other malevolent individuals, both because others may have developed mechanisms to avoid exploitation and because the chance of running into another malevolent individual is higher. The roughly 1% incidence of (clinical) Psychopathy observed in the general population may be attributable to frequency-dependent selection.

In contrast to Sickle-Cell, which arises due to the mutation of a single gene, malevolent traits may arise due to the cumulative effect of many genes (polygenicity). Glenn et al. (2011) lay out a potential pathway by which many small harmful mutations may cumulatively result in a highly unadaptive phenotype (Section 3). Because the harmful mutations are small and spread across many genes, they may persist in the population for a relatively long time before being weeded out.

I suspect that all these potential causal factors are compatible with one another and do substantially contribute to the persistence of malevolence in the population. The polygenic nature of malevolence would allow for a relatively continuous distribution of malevolence in the population. Individuals who are too malevolent or not malevolent enough would tend to have lower fitness than individuals who are moderately malevolent, meaning that individuals who possess “dark” alleles for some but not all relevant genes will tend to be selected for. And within society, highly malevolent individuals would have a niche they could fill, but their exploitative tendencies would undermine their own success, resulting in a relatively stable proportion of highly malevolent individuals in a society.

When is Malevolence Adaptive?

It appears that there are also particular environments which favor highly malevolent individuals. The results of the BBC Prison Study (Reicher & Haslam, 2006) suggest that in situations of weak/illegitimate authority and tumultuous circumstances, more authoritarian individuals tend to become more vocal and insistent, while less authoritarian individuals become more acquiescent. This pattern seems to hold true in the real world as well. Hitler, Stalin, and Mao, along with many other eventual dictators, came to power in importantly similar environments. The countries in which they came to power were ravaged by years of war and humiliated by foreign incursions. The incumbent governments were weak and lacked legitimacy at the time of their overthrow. In Germany’s case, the turbulent period after the dissolution of the monarchy ended with the Weimar Republic. The Russian monarchy was similarly overthrown after years of brutal warfare and replaced with a Provisional Government. And the Nationalists in China had borne the brunt of brutal assaults from the Imperial Japanese for nearly nine years and had held only nominal authority over a warlord-riven China for a decade before that. In all these cases, existing central authority was weak and perceived as illegitimate, and very real and salient threats were or had been arrayed against the nation and its people.

I suggest a few reasons why a subset of highly malevolent individuals are particularly adapted to these environments. First, there seems to be a smaller chance of retaliation against violence, deceptiveness, and other antisocial behaviors, particularly from established authorities. Second, malevolent individuals—particularly paranoid ones—will tend to be better at surviving and progressing in a violent, fast-moving, and lawless political-military hierarchy. Callous and Machiavellian individuals will be uniquely naturally inclined towards preemptive strikes against others for instrumental reasons, giving them an edge over rivals. Third, they will be willing to eliminate internal and external threats, which is something that will win the support of many sectors of the population. Fourth, their malevolence represents a threat that any resistance or plots against them will be met with reprisal; this threat may be particularly effective if the individual is highly sadistic/vindictive. 

Malevolent individuals appear to have a “double-advantage” in chaotic, violent environments. On the individual level, they are more equipped with the dispositions that would allow for survival and progression. On the group level, they are favored for their abilities to hold onto power and retaliate against potential threats. Therefore, fostering peace, cooperation, and lawfulness may be effective in reducing risks from malevolent actors.

Augmentative Factors

While D may be the primary psychological factor which contributes to agent-driven astronomical suffering, other factors may augment the effect of malevolence. These include:

Admiration

Admiration can be defined as “the tendency to approach social admiration by means of self-promotion (assertive self-enhancement)” (Back et al., 2013, p. 1015). This arises from the desire to be special, admired, and great. Individuals high in Admiration will tend to seek out positions of greater power, status, and wealth, so it follows that malevolent and admiration-seeking individuals will tend to be more dangerous because they will tend to seek and obtain more power than similarly malevolent but less admiration-oriented individuals. A number of studies have linked the Dark Triad traits to valuing power (Kajonius et al., 2015, Table 2; Kaufman et al., 2019, Table 7) and dominance-seeking (Davis & Vaillancourt, 2023, Figure 2). 

Admiration is conceptually derived mainly from components of Narcissism but also from some aspects of Machiavellianism (Bader et al., 2023; see Additional Materials 5 from OSF). This is in line with results from Chen et al. (2021), who found that Narcissism was consistently associated with general interest in politics. Narcissism was also associated with political participation; which includes active political participation (e.g., protesting, joining a political party), active social participation (e.g., boycotting a product, signing a petition), and donating to charity (including donating to a political/societal cause). I note that a drive towards Admiration does not necessarily preclude a drive towards Rivalry; in fact, Agentic Extraversion (which is similar to Admiration) is moderately positively correlated (r = .41) with Self-Centered Antagonism (Bader et al., 2023, Table 2). An agent high in Admiration and D might—particularly under certain circumstances—be both motivated to and effective in gaining and holding onto power.

A drive towards Admiration may help to explain why some highly malevolent individuals (such as Hitler and Stalin) seemingly dedicate themselves to a group cause. Such individuals may also have a grandiose desire to be the supreme authority, to be greeted by adoring crowds, and to be remembered in the history books—to be admired. Indeed, Davis & Vaillancourt (2023) found that Narcissism is uniquely and relatively strongly associated with prestige-seeking (Table 2). But for these things to become realities, there must be people that submit to that authority, form adoring crowds, and write and read history books. These malevolent individuals may not care to truly connect with others or join a collective endeavor; rather, they may simply attach themselves to a cause where they can gain notoriety or authority.

Boldness

Boldness can be defined as “a capacity to remain calm and focused in situations involving pressure or threat, an ability to recover quickly from stressful events, high self-assurance and social efficacy, and a tolerance for unfamiliarity and danger” (Patrick et al., 2009, p. 926). The capacity for resilience to and swift recovery from stressors seems to be generally individually adaptive. Furthermore, such a disposition seems to be particularly instrumental in achieving positions of high power. Whether one seeks to become an influential CEO or the next infamous dictator, one must face many highly stressful, quickly developing, and potentially ruinous events. The situations they face often are actually high-risk and full of uncertainty, and doubtless many—perhaps even the majority—of individuals in such situations fail (i.e., are ruined or killed). However, on the group level, one may find that positions of high power are filled almost exclusively by particularly bold individuals, because they are overwhelmingly the individuals who are willing to take such potentially ruinous or deadly risks in the first place. I also think the high self-assurance and social efficacy may lend a competitive edge to such bold individuals by causing others to support them. A salient example is the sheer charisma and daringness of Napoleon which made his restoration and the subsequent 100 Days campaign possible.

Disinhibition/Planfulness

Planfulness is composed of deliberation and order (Collison et al., 2018, p. 9). It has a strong negative correlation (r = –.68) with disinhibition (Bader et al., 2023, Table 3); essentially, individuals high in planfulness tend to be less impulsive and more self-controlled. Planfulness has almost no correlation with Boldness and only a weak negative correlation with Agentic Extraversion. This means that there are likely many individuals who are high in Admiration and Boldness and have the capacity to be self-controlled, forward-thinking, and orderly. The abilities to feign respect or submission, to outmaneuver opponents, to organise parties or coalitions, and to make plans years in advance are not only generally useful for an individual but may have contributed to the success of many (authoritarian) leaders.

This may be a case where—assuming Planfulness is a relatively stable trait—a moderate amount will tend to be the most successful. Even in the most chaotic of times, the ability to minimally restrain oneself from the most obviously medium- to long-term bad decisions will no doubt be necessary for surviving and attaining power. However, greater Planfulness requires greater time- and resource-devotion to planning, so too high a level of Planfulness may sap time and resources without producing sufficient returns.

Conscientiousness

Conscientiousness is one of the “standard” personality traits, being included in both the Big Five and HEXACO models of personality. Conscientiousness may be defined as the propensity to be “self-controlled, responsible to others, hardworking, orderly, and rule abiding” (Roberts et al., 2014, p. 1). It is most commonly thought to include the facets of orderliness and industriousness, but may also include impulsivity and responsibility.

Conscientiousness appears to relate to Admiration and Planfulness. Planfulness appears to map on quite well to the facets of orderliness and impulsivity. Empirical evidence is in line with this assumption, as disinhibition has a fairly strong (r = –.45) correlation with Conscientiousness (Stanley, Wygant, & Sellbom, 2013, Table 2). Admiration, on the other hand, does not appear to map on well to any part of Conscientiousness (Back et al., 2013, Table 5). However, I think Admiration is something like the steering wheel to the engine of Conscientiousness. Admiration points the individual in the direction of wanting power, but doesn’t itself drive them to act. Conscientiousness drives an individual to work hard and be orderly, but itself gives no direction. Admiration, while not mapping onto Conscientiousness, relates to it by giving it a consistent direction.

Conscientiousness appears to be a quite robust predictor of general success in many life domains, including general academic success and job performance (Wilmot & Ones, 2019). I think that the most harmful individuals will tend to be those who are highly conscientious, as this will help them attain power and achieve their goals, whatever those may be. At the very least, a minimal level of Conscientiousness seems necessary for an agent to attain the power needed to inflict great harm.

General Intelligence

While one may not need to be a genius to cause a massive amount of disvalue, I think that there is a certain threshold of intelligence below which a person does not pose a risk, no matter how malevolent they are. Furthermore, it appears that general intelligence is a powerful predictor of measures of success (Gottfredson, 1997), including socioeconomic and occupational status (Lo, 2017). Finally, I hypothesise that general intelligence would tend to assist malevolent agents to be more reasonable in goal-setting and more effective in goal achievement. Therefore, it may be true that—at least to a certain extent—greater general intelligence would play a role in increasing the magnitude of expected value that could be created by a malevolent agent. Note, for example, that 17 of the 21 defendants at the Nuremberg Trials had IQ scores at or above the 95th percentile (Coolidge, Davis, & Segal, 2007, pp. 33–34)

IQ appears to be practically unrelated to malevolent traits; a meta-analysis of the relations between IQ and the Dark Triad yielded very small correlations (all |r| < .05) and no appreciable shared variance (O’Boyle et al., 2013, Table 2). There may be a small moderate correlation between Psychopathy and intelligence, but this may be attributed to the fact that Psychopathy captures impulsive and aggressive tendencies, which are themselves linked to lower intelligence (Michels, 2021, Tables 1 and 2). This would suggest that higher general intelligence is not associated with better or worse outcomes, but only with more consequential ones.

Psychoticism

Psychoticism may be roughly defined as one’s ability or propensity to (accurately) reality-test. Psychotic individuals will tend to make spurious associations, seeing patterns which aren’t there and making unfounded inferences (more than the average person does). While malevolence may preclude morally reasoning, Psychoticism may preclude rational reasoning. The combination of the two may thus make it practically impossible to reason with an individual in any respect.

Psychoticism may result in worse outcomes due to its association with paranoia and because it may generally impair reasoning and decision-making. Agents who are relatively psychotic may not be reliably swayed by threats against them, as they may disbelieve in the threats or believe they can overcome them. In general, they may be more swayed towards extremes, such as the promise of a utopia or heavenly afterlife or the inevitability of a hell on earth. I hypothesise that a degree of psychoticism is a key aspect in the ideological fanaticism of many authoritarian leaders and their followers. As an example, an aggregation of 5 expert ratings found substantial evidence for Hitler having highly elevated levels of psychosis and paranoia (Coolidge, Davis, & Segal, 2007). This suggests that Psychoticism (as well as related concepts such as paranoia-proneness/general mistrust; Hilbig et al., 2021, see “Introduction”) could be important in augmenting malevolence. 

I expect that a moderate amount of Psychoticism will tend to be the most dangerous. A low amount of Psychoticism would entail a good ability to reality-test, which would correspond to a greater awareness of the consequences of one’s actions and less paranoia. A high amount of Psychoticism would lead to too much dysfunction, such that it would be unlikely that an individual could attain or keep power. Just the right amount of Psychoticism would leave an individual sufficiently functional, but would also correspond to less situational awareness and greater paranoia.

Recommendations

What to Measure and When

I’ve listed a number of augmentative traits which would contribute towards an agent’s ability to do harm. While I think it is useful to keep these traits in mind, it may not always be useful to explicitly measure them. One case in which I think most of these traits do not need to be measured is when screening individuals who are in contention for or have already attained positions of significant power. The fact that an individual has done so indicates that they possess the necessary traits to seek and attain power, which is what most of the augmentative factors are really screening for. The exception here is Psychoticism, since Psychoticism does not reflect one’s capacity or drive for attaining/seeking power. Thus, one’s attainment of power may reflect little about one’s level of Psychoticism, only signaling that one is not clinically psychotic.

However, in cases where one is studying individuals who have not had the opportunity to demonstrate power-seeking capacity/drive, screening for the augmentative traits may be useful. For example, if one is looking to evaluate AIs for potential to cause massive disvalue, it may be beneficial to screen for the AI’s desire and capacity to attain power and their propensity towards hallucination/psychotic behavior in addition to screening for malevolence.

Which Scales to Use

Not all scales are created equal, so for those who wish to implement screening processes, I have some recommendations.

General Malevolence

The D Scales

I recommend the D scales developed by Moshagen et al. (2020); in particular, the D-16 appears to be a good choice. There are a few advantages that the D scales possess which make them a standout candidate.

First, the D scales have demonstrated generally acceptable to excellent psychometric properties. The scales are highly internally consistent, reliable across time, and appear to be interpreted with sufficient similarity across gender and culture. The D scales have been shown to have robust and largely expected relations with specific malevolent trait measures, self-reports of aggression, crime, and empathy, and behavior in economic games (e.g., Moshagen et al., 2020, Table 4).

Second, they are relatively holistic measures. The D scales are composed of items drawn from a pool of 12 representative specific malevolent traits (Moshagen et al., 2020, Table 1). They have been examined against 58 specific malevolent traits, with only 1 trait significantly predicting behavior in an economic game beyond D (Hilbig et al., 2023, Figure 2). D thus appears to represent the dark personality space sufficiently well.

Finally, they are not incredibly long. Some scales can reach upwards of 200 items in length, and one may be concerned with the time and/or effort required for individuals to complete the scale. By contrast, measures of D are relatively short, with the full scale being 70 items long and reduced versions being 35 and 16 items long. The shorter versions correlate very highly (r > .90) with the full version. Furthermore, because the shorter versions—particularly the D-16—tend to keep the items that yield the most information and are the least problematic, the psychometric properties of the shorter versions are generally similar to (and in some ways better than) the psychometric properties of the full scale.

They’re also very accessible! For non-commercial purposes, you are free to use the D scales as long as you give credit to the authors. Simply go to www.darkfactor.org and scroll down to “Download the item sets and translations”. Bibliographic citations for attribution of credit can be found under the heading “Measuring and modeling D”.

Unfortunately, I am aware of no empirically derived cutoff points for the D scales. In lieu of this, I propose that roughly the top 30% of the scoring range should be considered “high malevolence” (e.g., on a scale from 0 to 100, anyone scoring above 70). This would appear to capture approximately the 95th %ile and above of scorers. So for the D-16, I propose that a score of 11 or greater be considered “high D”. For the D-70 and D-35, I propose the cutoff scores (inclusive) to be 49 and 25, respectively. I stress that these cutoffs are not empirically based, and I hope that future analyses will be able to propose more robust cutoffs.

The Dark Side of Humanity Scale

The Dark Side of Humanity Scale (DSHS) is a very new measurement instrument of dark personality (Katz et al., 2022). Nevertheless, my opinion is that it is quite promising, as it possesses acceptable to excellent psychometric properties (see Section 3). The Dark Side of Humanity is composed of 4 factors: Successful Psychopathy, Grandiose Entitlement, Sadistic Cruelty, and Entitlement Rage. Of these 4 factors, the most relevant are Successful Psychopathy and Sadistic Cruelty. Successful Psychopathy is largely a blend of Psychopathy’s callousness and the manipulativeness/planfulness of Machiavellianism (pp. 9–10), while Sadistic Cruelty draws from conceptions of Sadism and is most related to Sadism, spitefulness, and contempt (p. 10).

There are a few key advantages to this scale. It is perhaps the only one which I am aware of which has an explicit focus on identifying “successful” Psychopaths: those who can mask their antisocial tendencies and attain social status/power. Successful Psychopathy and Sadistic Cruelty thus appear to capture exactly the sort of disposition we want to be screening for; most other measures appear to be more geared towards capturing overt antisociality/criminality. It is also relatively short, with the full scale only being 42 items and the Successful Psychopathy + Sadistic Cruelty subscales together being only 26 items.

Unfortunately, I am aware of no empirically derived cutoff points for the DSHS. I’m also unaware of any data regarding the distribution of scores for this scale. Data and statistics regarding the distribution of scores of representative samples might therefore be helpful in establishing appropriate cutoff points for this instrument.

You can read the article and contact the authors here.

The Self-Report Psychopathy Scales

If one prefers a more established measure of malevolence, then I recommend using the Self-Report Psychopathy Scale (Paulhus, Neumann, & Hare, 2009), 3rd edition (SRP-III) or 4th edition (SRP-IV). These scales are the closest analogue to self-report versions of the widely-used PCL-R, which has been used to diagnose Psychopaths in clinical/offender settings. These scales have short forms as well, which may be comparable or superior to the longer versions. I recommend focusing exclusively on the Interpersonal Manipulation facet and Callousness facet (these may be grouped under the higher-order factor of Primary Psychopathy); the other two facets are focused on impulsivity and erratic lifestyle, which are of lesser concern for our purposes.

Though there are concerns regarding the underlying structure of the SRP scales, it appears that their psychometric properties are largely acceptable (Boduszek & Debowska, 2016; Gordts et al., 2017). Many of the concerns with the SRP scales regard the impulsive and antisocial facets and their integration into the larger Psychopathy construct, neither of which are of substantial interest to us. Additionally, the SRP scales benefit from widespread study and usage, such that their relations with other measures are more well-understood and their results potentially subject to greater and more impartial scrutiny.

Unfortunately, the SRP scales are less accessible, and the SRP-IV in particular is proprietary . Nevertheless, if you wish to use the SRP scales, you can go here to purchase the SRP-IV. I am unsure of the status of the SRP-III, but you may contact Drs. Delroy Paulhus (dpaulhus@psych.ubc.ca), and/or Craig Neumann (craig.neumann@unt.edu) for more information/access regarding the SRP scales.

Power-Seeking/Attaining

Admiration

The Admiration subscale of the Narcissistic Admiration and Rivalry Questionnaire (Back et al., 2013) may be used for this purpose. The facet of Agentic Extraversion from the Five-Factor Narcissism Inventory—Short Form (FFNI-SF; Sherman et al., 2015) also appears to capture admiration-seeking

Boldness

The Boldness subscale of the Triarchic Psychopathy Measure (Tri-PM; Patrick et al., 2009) may be used for this purpose.

Conscientiousness

The Conscientiousness subscale of the NEO-FFI-3 (a shorter version of the NEO-PI-3; McCrae & Costa, 2007) may be used for this purpose

Intellect

A gold-standard measure of intellect is the Wechsler Adult Intelligence Scale (Wechsler, 1997) and various short-form versions (Jiménez Bascuñán et al., 2021).

Psychoticism

A scale which appears to be appropriate to use in nonclinical settings, which has the most widespread usage, and captures symptoms of psychosis is the Prodromal Questionnaire (PQ; Loewy et al., 2005). Since the original scale is 92 items long, it may be more practical to use one of the shorter versions: the PQ-B (21 items; Loewy et al., 2011) or the PQ-16 (16 items; Ising et al., 2012).

These scales balance sensitivity (avoiding false negatives) against specificity (avoiding false positives; see section 2.3). For optimal balance of these, the full PQ scale has a recommended cutoff of 14, the PQ-B has a recommended cutoff of 3, and the PQ-16 has a recommended cutoff of 6. All cutoffs are inclusive (>=).

The PQ-16 in particular seems to have evidence of not only good sensitivity/specificity in a help-seeking population (Ising et al., 2012, Table 2), but also excellent psychometric properties when administered to a non-help-seeking population (Howie et al., 2022, Table 4). This, along with it being the shortest of the scales, makes it a good candidate for measuring Psychoticism in the general population. For this scale, I recommend deferring to the proposed cutoff of 6 and above.

How to Intervene

On the Societal Level

Screening

There are already precedents for evaluating the competency of individuals for high positions. The U.S. President, for example, must be at least 35 years old and a resident of the U.S. for at least 14 years. Many job applications evaluate individuals based on tests of aptitude, degrees and certifications earned, and prior (criminal) background. For any position with significant duties to the public interest, a person’s disposition towards malevolence is just as relevant, if not more. 

I recommend using non-clinical psychometric measures where possible for two reasons. First, the general population has a greater proportion of subclinical individuals than the forensic population, so it is likely more useful to have instruments which are sensitive to subclinical levels of relevant traits. Second, screening for clinical disorders in a workplace setting carries a higher risk of litigation under the Americans with Disabilities Act (ADA) of 1990, which views personality disorders as constituting disabilities/impairments (LeBreton et al., 2018, see “Delimiting the Scope of the Dark Triad”). Using non-clinical measures is therefore both more appropriate and less risky.

A major disadvantage of self-report methods is that they are vulnerable to gaming. While such things as social desirability bias appear not to be a large issue, I worry about the possibility of learning to deliberately game the measures. Evidence suggests that individuals can intentionally “fake good” on dark trait measures (Walker et al., 2022, Section 4). Additionally, many psychometric measures (along with scoring criteria) are publicly available, which means it is possible to “cheat the exam”. It might therefore be beneficial for researchers to more tightly control the flow of information to maintain an epistemic advantage. Closing this loophole would likely make it much more difficult to deliberately game psychometric measures. I suspect that self-report measures will always be gameable for agents who have sufficient intelligence and access to information (such as a highly intelligent AI with access to the entire internet). Even the best self-report measures should not be seen as walls which keep out all evil, but rather as fences which make it more difficult for malevolent individuals to enter. Note that this might just be paranoia on my part.

Can We Trust Ourselves to Identify Malevolent Actors?

Short answer: don’t bet on it

Theoretically speaking, I think there are a few things that make proficient identification of malevolent agents “with the naked eye” difficult.

First, people are beset by all kinds of biases and preconceptions. For example, tribalism probably substantially colors the glasses through which one appraises others. Their terrorist becomes our freedom fighter, their fraudster our Robin Hood, and their evil dictator our messianic strong-man. Even good-faith individuals may dupe themselves or be duped into believing that a highly malevolent individual is good or can be used for the greater good.

Second, a certain subset of malevolent individuals may be skilled at impression management. The Successful Psychopath who is bold, charismatic, conscientious, and self-controlled may tend to make a favorable impression on others. Their impression management may be particularly successful in environments with relatively few repeat/long-term interactions and/or in less personal settings. If they do exhibit any symptoms of malevolent behavior, these may go unnoticed, unreported, or unbelieved. It may be only after the damage has been done that anecdotes emerge and their true character is recognised.

Third, it seems nontrivially difficult for everyone to sufficiently agree on definitions of “good” and “bad”, and this may complicate evaluations of someone’s character. For example, some individuals may admire/value “dominant” traits such as conflict-proneness, toughmindedness, and boldness. Some individuals may be indifferent to or even contemptuous of “weak” traits such as an empathetic disposition or forgiving nature. While there may still be substantial broad agreement between individuals on what “good” and “bad” traits are, I emphasise these potential differences to make the point that the differences may be substantial, which likely makes it more difficult for at least some individuals to identify malevolent actors.

There doesn’t appear to be much empirical evidence regarding individuals’ abilities to detect Psychopathy (which I use as a stand-in for malevolence). The best and most recent I found was the analysis conducted by Olderbak et al. (2021). Their findings indicate that individuals appear, on the whole, to be only slightly better than chance in detecting Psychopathy in others in a first impression (Section 4.2.1). The range of individual correlations was substantial, with some individuals being worse than chance in detecting Psychopaths. In line with this, most of the variance in judgements tends to be due to the responders (i.e., noisy responding) and not the target. Finally, individuals appear to be less able to detect Primary Psychopathy than Secondary Psychopathy (Table 1). This suggests that, while individuals on the whole may be able to detect malevolence slightly better than chance, our detection abilities are a far cry from being a sufficient safeguard against malevolent individuals.

General Awareness

While I think lay perception of malevolence is “on the right track”, I think there are a few aspects where it may fall short.

It seems that the lay public perceives Psychopathy (and thus most likely malevolence) to be inevitably connected with some form of overtly violent antisocial behavior (e.g., serial killing; Furnham et al., 2009, see “Discussion”). Even the construct of Psychopathy itself is intimately tied to impulsivity and antisocial behaviors. I think the term “Psychopathy” should be reserved for those individuals who are callous, manipulative, and instrumentally antagonistic. For individuals who have impulsive, reactively antagonistic, and violent/criminal tendencies, I might use the term “Sociopathy”. Having two actually different names (rather than “Factor I” and “Factor II”) will help with distinguishing the two as having separate etiologies. Additionally, the term “Sociopathy” gives a hint to its own etiology, as it is theorised that high levels of reactive aggression and impulsivity can be due to environmental factors (i.e., early stressors increasing reactivity to threats; Blair, 2005, pp. 880–882). Essentially, it is possible and perhaps even likely for a person to be malevolent without being overtly and clearly antisocial/violent.

There may also be a perception that malevolence is inherently linked to some kind of mental illness, trauma, or psychosis. While high levels of malevolence may sometimes coincide with a disordered mind, I do not think it is generally useful to think of high malevolence as a mental illness. Mental illness carries all sorts of connotations, including irrationality, personal distress, and inability to function in society. I worry that this may result in us believing that rational, undistressed, and sufficiently functional individuals will not act malevolently. A powerful narrative in our society seems to be that there must be some external cause for someone’s malevolent behavior—a bad parent, a war-torn environment, unresolved internal trauma or pain. We must emphasise that this is not always the case: the evidence shows that some people are just born to be malevolent by default. It may be most useful to think of this form of malevolence as a maniaque sans délire—a mania without madness (Kiehl & Hoffman, 2011, Section I part B).

On the Individual Level

Based on this research, my intuition is that—at least for the purposes of making individuals substantially less malevolent—interventions will need to work at the neurological and biological levels. This does not mean that there is no place for cognitive/behavioral aspects in an intervention, but only that a purely cognitive/behavioral approach will be insufficient. Since there are no significant data (as far as I am aware) regarding the neurology or biology of the Dark Factor, I will rely on the literature on Psychopathy instead.

In the perception of the lay public, and in the minds of many clinicians (Pickersgill, 2011), Psychopathy is a notoriously difficult and tricky pathology to treat successfully. This reputation is largely deserved, as clinicians and researchers have noted Psychopaths’ high recidivism rates, tendencies to (successfully) deceptively game treatment attempts, and that therapy risks making Psychopaths more rather than less dangerous (Kiehl & Hoffman, 2011, Sections III & V). This does not mean treatment of Psychopathy is impossible; only that interventions must be “thoughtful, targeted, well-designed, and controlled” (and preferably targeting juveniles; see Section V) to have substantial positive impact.

However, even if careful cognitively-/behaviorally-based interventions are successful in reducing antisocial behavior, I don’t think they will be able to effectively address the issues which lie “beneath the surface”. Psychopathy is known to have a large heritable component, while there appears to be no known significant environmental factor. While traumatic early life circumstances may contribute to greater impulsivity and (reactive) aggression, there is no sign that they cause the callousness and instrumental aggression that is unique to Psychopathy (Blair, 2005, pp. 867–868). Differences in empathy and morality between children who develop into Psychopaths vs. typical children have been observed as early as 3 years of age (pp. 873–874), and a leading neuroscientific theory of Psychopathy proposes that specific and subtle deficiencies in learning further stunt the Psychopath’s development of empathy and morality (pp. 874–878). All this strongly suggests that the most effective times of intervention are before birth and in the first few years of life, which is often beyond the pale of traditional psychotherapies.

Reducing expressions of antisocial behavior and/or improving impulse control in individuals high in aversive traits may ultimately be relatively insignificant or even negative in value. This is because it ignores and potentially exacerbates the less salient but more consequential harm done by “successful” Psychopaths. I hypothesise that individuals who have malevolent tendencies would tend to enter into arenas which allowed them to use such tendencies to their advantage—such as business, politics, finance, and the military. These individuals would form the basis of a supply which met a “Molochian” demand for callousness (e.g., conducting preemptive strikes to incapacitate enemies, driving down animal welfare conditions to maximise profits, state-sanctioned torture and killing, deception and betrayal of all kinds). A person has the ability to be substantially more harmful in this capacity—contributing to intense suffering for thousands, millions, or billions of beings where even the most prolific serial killers “only” reached the hundreds in terms of victim count. Treating Psychopathy with only the cessation of symptoms in mind appears not unlike treating many other chronic disorders with such a mindset: it merely removes the symptoms which made identification of the disorder easy and leaves the underlying disorder present to manifest later in an even more devastating fashion.

Thoughts on Malevolence in EA

Epistemic status: mostly my own inferences and personal experience, hopefully somehow based on insights from empirical data

How well should we expect EA to compare to other broad umbrella institutions (such as academia and government) in its ability to resist malevolent actors?

There are some things that support EA doing much better. My experience is that the vast majority of people who are seriously involved in EA (e.g., have gone to an EAG conference) whom I met were very warm, compassionate, and well-intentioned individuals. I think this may be due to something of a double selection effect. Since EA so explicitly values doing good and selflessness, it seems substantially less likely that malevolent individuals would be attracted to EA as opposed to, say, Wall Street or the military. On the flip side, the way into the EA community is gated by individuals who tend to care disproportionately about positive impact on the world. For example, the university group I was a part of plays a big role in deciding who can break into the EA community at large. I’m sure informal selection criteria among group members include aspects such as high compassion, high honesty/humility, and low bullshit/deceptiveness. I’m ~70% confident that we do so to a substantially greater degree than the average organisation (student or otherwise). So in a lot of ways, I think EA does a pretty good job of selecting people who have good character.

However, not all is well. There are some worries that are (relatively) specific to the EA community. EA being a high-trust community filled with above-average agreeable and non-retributive actors may make it more vulnerable to exploitation by bad agents. This is especially true if there are not strong institutions to reliably deter bad actors (recall Nowak et al., 2016). Whether or not EA has (on balance) better or worse institutions than the average broad organisation is a question which I don’t feel capable of answering.

Additionally, EA’s focus on rationality may also incidentally align with the tendencies of bad actors, and thus be advantageous for them. For example, EA tends to be highly tolerant of instrumental deceptiveness and instrumental harm, so long as it can be justified by a greater end. This is unfortunately baked-in to utilitarianism, which appears to be the predominant moral philosophy of EA. EA also famously favors risk-neutrality, which I think on balance favors malevolent actors. Some evidence does partially suggest that more malevolent actors will tend to be somewhat more willing to take risks (Tiwari et al., 2021). At least in certain parts of the community, these rationalist tendencies may favor individuals who are relatively callous and ideologically motivated over individuals who are affectively empathetic. I note that these tendencies within EA may not be bad on the whole; however, they may by their nature have “side-effects” which select for malevolent tendencies.

Even if we assume that EA does a substantially better job at resisting malevolence than average, there is still the question of whether EA does a sufficient job in absolute terms. Here I am much more pessimistic. With great power comes great responsibility, and EA as a movement is necessarily highly power-hungry (agentic). So we cannot be content with just doing better than the average organisation. Having excellent culture, institutions, and individuals is not an option, but a requirement for securing the long-term flourishing and alignment of EA.

Coda: Compassion Towards Malevolent Individuals

It is true that the line between good and evil is drawn through all hearts, and especially that no one wants to destroy a piece of their own. So who are we to cast stones when we ourselves are not sinless? And how do you or I know that it is not we who will end up doing great evil in the world?

And yet we must still draw lines in the sand. For it is too simplistic to say that all people have evil in their hearts; that all people would react in the same way under the same circumstances. For example, a sample of 1500 Nazi biographies (mainly focused on officers and top leaders) found that approximately 65% of them were “Old Nazis”: those who had joined the Party before the massive expansion in Nazi power in 1933 (Mann, 2017, Table 6). Furthermore, eventual Nazis were overwhelmingly overrepresented in the post-WWI Freikorps, which was a loose association of paramilitary organisations known for its murderous, nationalist, and racist acts. Finally, there was little evidence to suggest that familial or environmental trauma was a significant factor for the vast majority of Nazis (pp. 341–342). Of course, many individuals were coerced or forced by pressures beyond their control to commit atrocities they could hardly bear to watch. At the very least, however, this is evidence that those within the bureaucracies which perpetrate massive harm-doing tend not to be “just like the rest of us” or “a victim of circumstances”. It is not only incorrect but an insult to the innocent and suffering to say otherwise.

As a holder of an Industrial Engineering degree, I am very sympathetic to systems thinking, and respect the commandment that thou shalt not attribute to malice that which can be attributed to moderately rational individuals following incentives in a complex system. Nevertheless, I think seminal works like Arendt’s Eichmann in Jerusalem, Milgram’s Obedience Experiments, and Zimbardo’s Stanford Prison Experiment have driven us too far in this regard. They foster a sense that everyone is equally capable of evil, when the reality is, by all evidence, substantially different.

Therefore, while I still think that all beings are deserving of compassionate treatment to the extent that we can give it to them, we must always be aware of the practical limitations which face us. Highly malevolent individuals present too great a risk to general welfare to be left as they are, and current treatments appear to be largely ineffective in reducing this risk. Until there is a “cure” for pathological malevolence, we must be unwavering in refusing to allow highly malevolent individuals to ever attain positions of significant power.

Acknowledgements

A giant thank-you to David Althaus, who was my supervisor during the writing of this post. I am also grateful to Polaris Ventures more broadly for funding this research.

Author Contributions

This post was written by me (Kenneth Diao) under the supervision of David Althaus.

Bibliography

Ackerman, R. A., Witt, E. A., Donnellan, M. B., Trzesniewski, K. H., Robins, R. W., & Kashy, D. A. (2011). What does the narcissistic personality inventory really measure?. Assessment, 18(1), 67–87. https://doi.org/10.1177/1073191110382845

Althaus, D., & Baumann, T. (2020). Reducing long-term risks from malevolent actors. Publications of Center of Long Term Risk. https://longtermrisk.org/files/Reducing_long_term_risks_from_malevolent_actors.pdf

Back, M. D., Küfner, A. C., Dufner, M., Gerlach, T. M., Rauthmann, J. F., & Denissen, J. J. (2013). Narcissistic admiration and rivalry: disentangling the bright and dark sides of narcissism. Journal of personality and social psychology, 105(6), 1013–1037. https://www.researchgate.net/profile/Mitja-Back/publication/257837687_Narcissistic_Admiration_and_Rivalry_Disentangling_the_Bright_and_Dark_Sides_of_Narcissism/links/00b7d52f8ba1f96482000000/Narcissistic-Admiration-and-Rivalry-Disentangling-the-Bright-and-Dark-Sides-of-Narcissism.pdf

Bader, M., Hartung, J., Hilbig, B. E., Zettler, I., Moshagen, M., & Wilhelm, O. (2021). Themes of the dark core of personality. Psychological Assessment, 33(6), 511–525. https://www.researchgate.net/profile/Martina-Bader/publication/348306977_Themes_of_the_Dark_Core_of_Personality/links/5ff728b0299bf140887d59ce/Themes-of-the-Dark-Core-of-Personality.pdf

Bader, M., Hilbig, B. E., Zettler, I., & Moshagen, M. (2023). Rethinking aversive personality: Decomposing the Dark Triad traits into their common core and unique flavors. Journal of Personality, 91(5), 1084–1109. https://doi.org/10.1111/jopy.12785

Blair, R. J. R. (2005). Applying a cognitive neuroscience perspective to the disorder of psychopathy. Development and psychopathology, 17(3), 865–891. https://www.cambridge.org/core/services/aop-cambridge-core/content/view/3283E1226BF1BE3020392F3B2660E675/S0954579405050418a.pdf/div-class-title-applying-a-cognitive-neuroscience-perspective-to-the-disorder-of-psychopathy-div.pdf

Blötner, C., & Beisemann, M. (2022). The Dark Triad is dead, long live the Dark Triad: An item-response theoretical examination of the Short Dark Tetrad. Personality and Individual Differences, 199, 111858. https://doi.org/10.1016/j.paid.2022.111858

Boduszek, D., & Debowska, A. (2016). Critical evaluation of psychopathy measurement (PCL-R and SRP-III/SF) and recommendations for future research. Journal of Criminal Justice, 44, 1–12. https://doi.org/10.1016/j.jcrimjus.2015.11.004

Browning, C. (1992). Ordinary men: Reserve police battalion 101 and the final solution in Poland. https://www.ias.edu/sites/default/files/sss/pdfs/Ordinary%20Men.pdf

Chen, P., Pruysers, S., & Blais, J. (2021). The dark side of politics: Participation and the dark triad. Political Studies, 69(3), 577–601. https://doi.org/10.1177/0032321720911566

Collison, K. L., Vize, C. E., Miller, J. D., & Lynam, D. R. (2018). Development and preliminary validation of a five factor model measure of Machiavellianism. Psychological assessment, 30(10), 1401. https://psyarxiv.com/teg8j/download?format=pdf

Coolidge, F. L., Davis, F. L., & Segal, D. L. (2007). Understanding madmen: A DSM-IV assessment of Adolf Hitler. Individual Differences Research, 5(1). https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=791353e929de11bc3fea2f4b67f8f8cac68c01df

Davis, A. C., & Vaillancourt, T. (2023). Predicting dominance and prestige status-striving from the dark tetrad: The mediating role of indirect aggression. Current Psychology, 42(16), 13680–13692. https://doi.org/10.1007/s12144-021-02492-y

Dinić, B. M., Wertag, A., Sokolovska, V., & Tomašević, A. (2023). The good, the bad, and the ugly: Revisiting the Dark Core. Current Psychology, 42(6), 4956–4968. https://doi.org/10.1007/s12144-021-01829-x

Furnham, A., Daoud, Y., & Swami, V. (2009). “How to spot a psychopath” Lay theories of psychopathy. Social Psychiatry and Psychiatric Epidemiology, 44, 464–472. https://doi.org/10.1007/s00127-008-0459-1

Gatner, D. T., Douglas, K. S., Almond, M. F. E., Hart, S. D., & Kropp, P. R. (2023). How much does that cost? Examining the economic costs of crime in North America attributable to people with psychopathic personality disorder. Personality Disorders: Theory, Research, and Treatment, 14(4), 391–400. https://gwern.net/doc/psychology/personality/psychopathy/2022-gatner.pdf

Glenn, A. L., Kurzban, R., & Raine, A. (2011). Evolutionary theory and psychopathy. Aggression and violent behavior, 16(5), 371–380. https://lust-for-life.org/Lust-For-Life/_Textual/AndreaLGlenn-RobertKurzban-AdrianRaine_EvolutionaryTheoryAndPsychopathy_2011_15pp/AndreaLGlenn-RobertKurzban-AdrianRaine_EvolutionaryTheoryAndPsychopathy_2011_15pp.pdf

Glenn, A. L., & Sellbom, M. (2015). Theoretical and empirical concerns regarding the dark triad as a construct. Journal of personality disorders, 29(3), 360–377. https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=f368702648967cd06c1495785297cc1b6cfc7493

Gordts, S., Uzieblo, K., Neumann, C., Van den Bussche, E., & Rossi, G. (2017). Validity of the Self-Report Psychopathy Scales (SRP-III full and short versions) in a community sample. Assessment, 24(3), 308–325. https://doi.org/10.1177/1073191115606205

Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life. Intelligence, 24(1), 79–132. https://www1.udel.edu/educ/gottfredson/reprints/1997whygmatters.pdf

Hilbig, B. E., Thielmann, I., Klein, S. A., Moshagen, M., & Zettler, I. (2021). The dark core of personality and socially aversive psychopathology. Journal of Personality, 89(2), 216–227. https://doi.org/10.1111/jopy.12577

Hilbig, B. E., Thielmann, I., Zettler, I., & Moshagen, M. (2023). The Dispositional Essence of Proactive Social Preferences: The Dark Core of Personality vis-à-vis 58 Traits. Psychological Science, 34(2), 201–220. https://doi.org/10.1177/09567976221116893

Horsten, L. K., Moshagen, M., Zettler, I., & Hilbig, B. E. (2021). Theoretical and empirical dissociations between the Dark Factor of Personality and low Honesty-Humility. Journal of Research in Personality, 95, 104154. https://doi.org/10.1016/j.jrp.2021.104154

Howie, C., Hanna, D., Shannon, C., Davidson, G., & Mulholland, C. (2022). The Structure of the Prodromal Questionnaire—16 (PQ-16): Exploratory and confirmatory factor analyses in a general non-help-seeking population sample. Early Intervention in Psychiatry, 16(3), 239–246. https://pureadmin.qub.ac.uk/ws/portalfiles/portal/245027777/The_Structure_of_the_Prodromal_Questionnaire_16_EIP_2020_035_R3_March_9th_2021_Clean_Version_2_.pdf

Ising, H. K., Veling, W., Loewy, R. L., Rietveld, M. W., Rietdijk, J., Dragt, S., ... & van der Gaag, M. (2012). The validity of the 16-item version of the Prodromal Questionnaire (PQ-16) to screen for ultra high risk of developing psychosis in the general help-seeking population. Schizophrenia bulletin, 38(6), 1288–1296. https://doi.org/10.1093/schbul/sbs068

Jasch, H. C., & Kreutzmüller, C. (Eds.). (2017). The Participants: The Men of the Wannsee Conference. Berghahn Books. link here (click PDF)

Jiménez Bascuñán, A., Fuentes Durá, I., & Dasí, C. (2020). Short forms of the Wechsler adult scales: a systematic review. https://repositorio.ual.es/bitstream/handle/10835/8990/2765-12557-1-PB.pdf?sequence=1

Jonason, P. K., Koenig, B. L., & Tost, J. (2010). Living a fast life: The Dark Triad and life history theory. Human Nature, 21, 428–442. https://doi.org/10.1007/s12110-010-9102-4

Kajonius, P. J., Persson, B. N., & Jonason, P. K. (2015). Hedonism, achievement, and power: Universal values that characterize the Dark Triad. Personality and Individual Differences, 77, 173–178. https://www.researchgate.net/profile/Peter-Karl-Jonason/publication/271274905_Hedonism_Achievement_and_Power_Universal_values_that_characterize_the_Dark_Triad/links/5fa2891a299bf1b53e613031/Hedonism-Achievement-and-Power-Universal-values-that-characterize-the-Dark-Triad.pdf

Katz, L., Harvey, C., Baker, I. S., & Howard, C. (2022). The dark side of humanity scale: A reconstruction of the Dark Tetrad constructs. Acta Psychologica, 222, 103461. https://doi.org/10.1016/j.actpsy.2021.103461

Alternate link

Kaufman, S. B., Yaden, D. B., Hyde, E., & Tsukayama, E. (2019). The light vs. dark triad of personality: Contrasting two very different profiles of human nature. Frontiers in psychology, 10, 467. https://doi.org/10.3389/fpsyg.2019.00467

Kiehl, K. A., & Hoffman, M. B. (2011). The criminal psychopath: History, neuroscience, treatment, and economics. Jurimetrics, 51, 355–397. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4059069/

Kowalski, C. M., Rogoza, R., Saklofske, D. H., & Schermer, J. A. (2021). Dark triads, tetrads, tents, and cores: Why navigate (research) the jungle of dark personality models without a compass (criterion)?. Acta Psychologica, 221, 103455. https://doi.org/10.1016/j.actpsy.2021.103455

LeBreton, J. M., Shiverdecker, L. K., & Grimaldi, E. M. (2018). The dark triad and workplace behavior. Annual Review of Organizational Psychology and Organizational Behavior, 5, 387–414. https://doi.org/10.1146/annurev-orgpsych-032117-104451

Lo, C. F. (2017). Is there a relationship between high IQ scores and positive life outcomes? a critical review. Psychology, 8(4), 627–635. https://doi.org/10.4236/psych.2017.84040

Loewy, R. L., Bearden, C. E., Johnson, J. K., Raine, A., & Cannon, T. D. (2005). The prodromal questionnaire (PQ): preliminary validation of a self-report screening measure for prodromal and psychotic syndromes. Schizophrenia research, 79(1), 117–125. https://doi.org/10.1016/j.schres.2005.03.007

Loewy, R. L., Pearson, R., Vinogradov, S., Bearden, C. E., & Cannon, T. D. (2011). Psychosis risk screening with the Prodromal Questionnaire—brief version (PQ-B). Schizophrenia research, 129(1), 42–46. https://doi.org/10.1016/j.schres.2011.03.029

Alternate link

Mann, M. (2017). Were the perpetrators of genocide “ordinary men” or “real Nazis”? Results from fifteen hundred biographies. Nazi Germany, 381–416. https://www.sscnet.ucla.edu/soc/faculty/mann/Doc3.pdf

Marcus, D. K., Preszler, J., & Zeigler-Hill, V. (2018). A network of dark personality traits: What lies at the heart of darkness?. Journal of Research in Personality, 73, 56–62. https://doi.org/10.1016/j.jrp.2017.11.003

McCrae, R. R., & Costa, Jr, P. T. (2007). Brief versions of the NEO-PI-3. Journal of individual differences, 28(3), 116-128. https://www.researchgate.net/profile/Paul-Costa/publication/232591551_Brief_versions_of_the_NEO_PI-3/links/54ebeabb0cf2a03051951d5e/Brief-versions-of-the-NEO-PI-3.pdf?_sg%5B0%5D=started_experiment_milestone&origin=journalDetail&_rtd=e30%3D

McDonald, M. M., Donnellan, M. B., & Navarrete, C. D. (2012). A life history approach to understanding the Dark Triad. Personality and individual differences, 52(5), 601–605. https://www.researchgate.net/profile/Melissa-Mcdonald-16/publication/224827812_A_life_history_approach_to_understanding_the_Dark_Triad/links/5a1c2b4baca272df0810bf10/A-life-history-approach-to-understanding-the-Dark-Triad.pdf

McHoskey, J. W., Worzel, W., & Szyarto, C. (1998). Machiavellianism and psychopathy. Journal of personality and social psychology, 74(1), 192–210. link to pdf

Michels, M. (2021). General Intelligence and the Dark Triad. Journal of Individual Differences. https://doi.org/10.1027/1614-0001/a000352

Moshagen, M., Hilbig, B. E., & Zettler, I. (2018). The dark core of personality. Psychological review, 125(5), 656–688. https://wildbeimwild.com/wp-content/uploads/2018/11/Psychologie-Der-dunkle-Kern-der-Pers%C3%B6nlichkeit.pdf

Moshagen, M., Zettler, I., & Hilbig, B. E. (2020). Measuring the dark core of personality. Psychological Assessment, 32(2), 182–196. https://www.darkfactor.org/Moshagen,%20Zettler,%20Hilbig%20-%20Measuring%20the%20dark%20core%20of%20personality.pdf

Muris, P., Merckelbach, H., Otgaar, H., & Meijer, E. (2017). The malevolent side of human nature: A meta-analysis and critical review of the literature on the dark triad (narcissism, Machiavellianism, and psychopathy). Perspectives on psychological science, 12(2), 183–204. https://doi.org/10.1177/1745691616666070
Alternate link

Neumann, C. S., Jones, D. N., & Paulhus, D. L. (2022). Examining the Short Dark Tetrad (SD4) across models, correlates, and gender. Assessment, 29(4), 651–667. https://doi.org/10.1177/1073191120986624

Nowak, A., Gelfand, M. J., Borkowski, W., Cohen, D., & Hernandez, I. (2016). The evolutionary basis of honor cultures. Psychological science, 27(1), 12–24. https://doi.org/10.1177/0956797615602860

O’Boyle, E. H., Forsyth, D., Banks, G. C., & Story, P. A. (2013). A meta-analytic review of the Dark Triad–intelligence connection. Journal of Research in Personality, 47(6), 789–794. https://doi.org/10.1016/j.jrp.2013.08.001

Olderbak, S., Bader, C., Hauser, N., & Kleitman, S. (2021). Detection of psychopathic traits in emotional faces. Journal of Intelligence, 9(2), 29. https://doi.org/10.3390/jintelligence9020029

Patrick, C. J., Fowles, D. C., & Krueger, R. F. (2009). Triarchic conceptualization of psychopathy: Developmental origins of disinhibition, boldness, and meanness. Development and psychopathology, 21(3), 913–938. https://www.cambridge.org/core/services/aop-cambridge-core/content/view/172BC63ED5C4C4C295C47DDCB01E838D/S0954579409000492a.pdf/triarchic_conceptualization_of_psychopathy_developmental_origins_of_disinhibition_boldness_and_meanness.pdf

Paulhus, D. L., Neumann, C. S., & Hare, R. D. (2009). Manual for the self-report psychopathy scale. https://storefront.mhs.com/collections/srp-4

Paulhus, D. L., & Williams, K. M. (2002). The dark triad of personality: Narcissism, Machiavellianism, and psychopathy. Journal of research in personality, 36(6), 556–563. https://doi.org/10.1016/S0092-6566(02)00505-6

Pickersgill, M. (2011). ‘Promising’therapies: Neuroscience, clinical practice, and the treatment of psychopathy. Sociology of Health & Illness, 33(3), 448–464. https://doi.org/10.1111/j.1467-9566.2010.01286.x

Reicher, S., & Haslam, S. A. (2006). Rethinking the psychology of tyranny: The BBC prison study. British Journal of Social Psychology, 45(1), 1–40. https://doi.org/10.1348/014466605X48998

Roberts, B. W., Lejuez, C., Krueger, R. F., Richards, J. M., & Hill, P. L. (2014). What is conscientiousness and how can it be assessed?. Developmental psychology, 50(5), 1315. http://timbates.wdfiles.com/local--files/impulse/Roberts%20et%20al%202012%20Conscientiousness%20review.pdf

Sellbom, M., & Drislane, L. E. (2021). The classification of psychopathy. Aggression and Violent Behavior, 59, 101473. https://doi.org/10.1016/j.avb.2020.101473

Sherman, E. D., Miller, J. D., Few, L. R., Campbell, W. K., Widiger, T. A., Crego, C., & Lynam, D. R. (2015). Development of a short form of the five-factor narcissism inventory: the FFNI-SF. Psychological Assessment, 27(3), 1110. https://www.researchgate.net/publication/273638115_Development_of_a_short_form_of_the_Five-Factor_Narcissism_Inventory_The_FFNI-SF

Solzhenitsyn, A. (1974). The Gulag Archipelago. https://ia801301.us.archive.org/11/items/TheGulagArchipelago-Threevolumes/The-Gulag-Archipelago__vol2__III-IV__Solzhenitsyn.pdf

Spurk, D., Hirschi, A., Wang, M., Valero, D., & Kauffeld, S. (2020). Latent profile analysis: A review and “how to” guide of its application within vocational behavior research. Journal of vocational behavior, 120, 103445. https://doi.org/10.1016/j.jvb.2020.103445

Stanley, J. H., Wygant, D. B., & Sellbom, M. (2013). Elaborating on the construct validity of the Triarchic Psychopathy Measure in a criminal offender sample. Journal of personality assessment, 95(4), 343–350. https://www.tandfonline.com/doi/pdf/10.1080/00223891.2012.735302?casa_token=yfHA8ktffjEAAAAA:Ne1YxY-YAfqhzRNfnfD1uGaPtjWLmLXT6BuykTTUz9qy1pX8MxAtz7GANtaV-KNnelRRQzHN7tjR6w

Tiwari, S., Moshagen, M., Hilbig, B. E., & Zettler, I. (2021). The dark factor of personality and risk-taking. International journal of environmental research and public health, 18(16), 8400. https://doi.org/10.3390/ijerph18168400

Trahair, C., Baran, L., Flakus, M., Kowalski, C. M., & Rogoza, R. (2020). The structure of the Dark Triad traits: A network analysis. Personality and Individual Differences, 167, 110265. https://www.radoslawrogoza.com/uploads/1/2/3/1/123141801/trahair_et_al._2020.pdf

Van Dongen, J. D. (2020). The empathic brain of psychopaths: From social science to neuroscience in empathy. Frontiers in psychology, 11, 493832. https://doi.org/10.3389/fpsyg.2020.00695

Walker, S. A., Double, K. S., Birney, D. P., & MacCann, C. (2022). How much can people fake on the dark triad? A meta-analysis and systematic review of instructed faking. Personality and individual differences, 193, 111622. https://doi.org/10.1016/j.paid.2022.111622

Wechsler, D. (1997). Wechsler Adult Intelligence Scale--Third Edition (WAIS-III) [Database record]. APA PsycTests. https://doi.org/10.1037/t49755-000

Wilmot, M. P., & Ones, D. S. (2019). A century of research on conscientiousness at work. Proceedings of the National Academy of Sciences, 116(46), 23004–23010. https://doi.org/10.1073/pnas.1908430116

*preview image was generated using ImagineArt AI