Why you need to know this: if you have ever issued the same warning to two difficult people and watched it land on one and slide straight off the other, this explains why, and what to do instead of issuing it again, louder.
A popular picture of the dangerous colleague is almost always the calm one: composed, charming when it suits, never visibly troubled by anything, and the word most people reach for is psychopath. The research on psychopathic traits has spent decades arguing that the word covers two quite different patterns, and that the calm one is only half of it.
The other half is anxious, impulsive, quick to anger and easily overwhelmed, and it rarely gets pictured at all. The largest synthesis of the workplace evidence, pooling 166 samples and 49,350 psychopathy scores, found that the harm psychopathic traits do to performance, to helpfulness, and to conduct at work was mostly more pronounced for this anxious, impulsive form than for the cold one (Roth & Klehe, 2025).
The two forms do not appear to respond to the same things. In questionnaire studies, people high in the cold form report lower sensitivity to the threat of punishment, while both forms report sensitivity to reward (Newman, MacCoon, Vaughn & Sadeh, 2005; Ross et al., 2007). Nor is fear simply absent in psychopathy. In 125 prisoners high in the trait, the fear response was normal when they were attending to the threat and reduced when their attention was on something else (Newman, Curtin, Bertsch & Baskin-Sommers, 2010).
That has consequences for how an organisation responds. Across three studies of leaders, it was clear that rules, in particular, of the three organisational features tested, weakened the link between the cold form and self-serving or abusive behaviour (Laurijssen, Wisse, Sanders & Sleebos, 2024). In a scenario study and a field study, employees high in the cold form reported more positive mood and engagement under an abusive boss, and in the field study, less anger than their peers, which is a poor advertisement for getting tough (Hurst, Simon, Jung & Pirouz, 2019).
And change is not ruled out. Among 152 high-risk offenders followed for about five years, those who changed more in treatment went on to reoffend violently less, although change was smaller among those highest in the cold, emotionally detached core of the trait (Olver, Lewis & Wong, 2013).
None of this licenses deciding which kind of person a colleague is, and the article argues at length that you do not need to. What it offers instead is a way of reading how somebody responds to what you have already done, and a set of responses chosen because they reach both patterns at once: write the rule down before raising the stakes, put the consequence in front of the decision at the moment it is taken, take load off before adding pressure, and stop assuming that the one who does not flinch is the one who matters most.