Take 400 Brazilian adults, ages 18 to 60, hand them a Dark Triad scale twice — once rating their own Machiavellianism, narcissism, and psychopathy, once rating how desirable each trait would be in an ideal partner — and see how closely the two sets of numbers track. That’s the design behind Dark side of mate choice: homogamy drives attraction to dark triad, but desirability of antisocial traits and criminal behavior is not enhanced by fame, wealth, or looks, published Oct. 19, 2026 in Evolution and Human Behavior by Katharina P. da Silva, Bruno H. Amaral, Marco Antonio C. Varella, and Jaroslava V. Valentova, and covered the same day by PsyPost. The headline number: self-ratings predicted desired-partner ratings at 0.59 for Machiavellianism, 0.66 for psychopathy, and 0.69 for narcissism. People higher in dark traits want a partner higher in the same ones. But the same dataset found a plain prosocial profile beating every antisocial profile it was pitted against — fame, wealth, and good looks included. Aristotle named the first half of that finding in the Rhetoric. He named the second half in the Nicomachean Ethics. About twenty-three centuries before anyone ran a correlation on it.

The Quick Verdict

Question	What we found
What was studied?	Whether people’s own Dark Triad traits predict what they find desirable in a partner, and whether fame, wealth, or looks can rescue the desirability of antisocial or criminal profiles
Who ran it?	Katharina P. da Silva, Bruno H. Amaral, Marco Antonio C. Varella, and Jaroslava V. Valentova
Where and when?	Evolution and Human Behavior, DOI 10.1016/j.evolhumbehav.2026.106889; covered Oct 19, 2026 by PsyPost
Sample	400 Brazilian adults (193 women, 176 men, 31 non-binary/travesti), ages 18-60, average age roughly 28
Core finding	Self-ratings correlated with desired-partner ratings at 0.59 (Machiavellianism), 0.66 (psychopathy), 0.69 (narcissism) — strong homogamy for dark traits
The ceiling on that finding	A prosocial profile scored 5.92 of 7 in desirability; the same profile loaded with Dark Triad traits but no crime dropped to 2.40; adding a criminal assault history dropped it to 1.52; pairing that criminal profile with fame, wealth, or attractiveness topped out at 1.60
What Aristotle already argued	Rhetoric II.4: people feel friendly toward those who resemble them; Nicomachean Ethics IX.4: the wicked can’t sustain real friendship, because there’s nothing in them genuinely lovable, even to themselves

The Quick Version

A 2026 study of 400 Brazilian adults found strong evidence that people with Dark Triad traits want partners who share those traits — correlations as high as 0.69 for narcissism. But likeness didn’t rescue desirability once crime entered the picture. A prosocial profile beat every Dark Triad profile tested, and no amount of fame, wealth, or attractiveness pulled a criminal profile back into competitive territory. Aristotle predicted both halves: similarity breeds liking, but vice still can’t produce the thing a real bond is built from.

What the Study Actually Found

Most research on “bad boy” or “femme fatale” appeal picks a side before it starts. Either people are drawn to dark personalities (hence the fan mail Ted Bundy received in prison, the premise the researchers say partly inspired this project) or they aren’t, and the apparent attraction is really something else — status, confidence, a thrill that has nothing to do with the cruelty itself. Da Silva, Amaral, Varella, and Valentova tested both the “like attracts like” question and the “can other goods offset vice” question in the same sample, rather than treating them as the same claim.

The homogamy result is the cleaner of the two. Participants rated their own standing on Machiavellianism, narcissism, and psychopathy, then rated how desirable each trait would be in a partner. Self-ratings in this Brazilian sample ran low on average — 1.62 for Machiavellianism, 1.58 for psychopathy, 2.45 for narcissism, all on a 1-to-7 scale — meaning most people don’t describe themselves as particularly dark. But wherever a person landed on that scale, their partner preferences tracked it, tightly: 0.59, 0.66, and 0.69. Narcissism pulled hardest toward a matching partner. The researchers call this the first clear evidence of active assortment for antisocial traits specifically in a Brazilian sample, rather than something inferred from already-paired couples after the fact.

The second half of the design is where the “bad boy” myth runs into trouble. Participants rated the desirability of several hypothetical partner profiles: a plain prosocial description, the same description rewritten with Dark Triad traits but no criminal history, and that antisocial profile again with a history of physical assault added. Then — aimed squarely at the Bundy-letters phenomenon — the researchers paired the criminal profile with fame, wealth, and physical attractiveness separately, to see whether any of those three classic desirability boosters could pull a criminal record back up to competitive territory.

They couldn’t. The prosocial profile scored 5.92 out of 7. Antisocial traits alone, no crime, dropped that to 2.40. A criminal assault history dropped it further, to 1.52. Fame nudged it to 1.46 — statistically indistinguishable, arguably worse. Wealth brought it to 1.49. Attractiveness, the single most commonly cited explanation for criminal fan mail, got it to 1.60 — still nowhere near a quarter of what plain prosociality scored alone. The researchers call this the desirability by-product hypothesis: the rare cases of attraction to real criminals likely ride on traits that travel alongside darkness — boldness, confidence, social dominance — rather than on the cruelty itself. As the lead author told PsyPost, “despite evidence showing some attraction to antisocial traits and criminal individuals, prosociality still remains a far more desirable trait in the context of mate choice.”

What Did Aristotle Mean By This?

In the Rhetoric, Book II, Chapter 4, Aristotle lists what makes people feel philia — friendly goodwill — toward each other, and resemblance tops the list: people feel friendly “towards those who are like ourselves in character and occupation, provided they do not get in our way.” In the Nicomachean Ethics, Book IX, Chapter 4, he turns to why that liking fails between wicked people specifically: “the bad man does not seem to be amicably disposed even to himself, because there is nothing in him to love.”

These are two different books, written to do two different jobs, and the study happens to land on both of them at once. The Rhetoric is a manual for persuasion — Aristotle is explaining to a speaker how an audience’s existing attachments work, not endorsing those attachments as good ones. His claim there is purely descriptive: similarity is a mechanism of liking, full stop, no judgment attached. It explains why a Machiavellian would feel a pull toward another Machiavellian without needing any claim about whether that pull is healthy. That’s the 0.59-to-0.69 half of the study, almost exactly.

The Nicomachean Ethics half is a different kind of argument, and a sharper one. Aristotle isn’t denying that two cruel people can resemble each other, team up, or even feel something like fondness for each other — he’s denying that what they’d have qualifies as real friendship, philia in its complete sense, the kind built in his three-tier account covered here before. His reasoning is almost clinical: a friend is, in a sense, another self, and you can only extend that outward if there’s something in your own character worth extending. A person at war with their own impulses — pulled one way by appetite, another by whatever conscience is left — has nothing stable to offer anyone else, including themselves. That’s not a moral slap on the wrist. It’s closer to a structural diagnosis: vice doesn’t just make you worse to be around, it makes you a worse foundation for the thing a bond needs to be built on. Which is close to the shape of a profile that scores 1.52 even before anyone adds a dollar of wealth or a magazine cover to it.

Why Didn’t Fame, Wealth, or Looks Rescue a Criminal Profile?

1. The attraction people report to real criminals likely isn’t attraction to the crime. The desirability by-product hypothesis holds that admirers are responding to confidence, boldness, or social dominance that happens to correlate with Dark Triad traits — not to the cruelty itself. Strip away the plausible cover story and the antisocial profile has nothing left to trade on.
2. Resemblance predicts attraction to traits, not tolerance for consequences. The 0.59-to-0.69 correlations describe people drawn to a personality style — manipulative, bold, self-focused. A criminal assault history is a documented outcome of that style, not a restatement of it, and the data treats them as separable.
3. A desirability floor this low leaves little room for a booster to work on. Once a profile sits at 1.52 out of 7, fame, wealth, or attractiveness have almost nowhere to move it — the ceiling for any one variable to drag a profile back toward competitive territory is mathematically narrow once the baseline has collapsed this far.
4. Aristotle’s framework predicts exactly this ordering, because fame and wealth were never his criteria for friendship to begin with. He ranks friendships of utility and pleasure below friendships of virtue, and money, status, and looks are utility-and-pleasure goods in his system — add more to a vice-based profile and you’ve made a better utility friendship, not a virtue one.

Why the Study and the Philosophy Rhyme

The honest version of this isn’t “a Greek philosopher predicted a 2026 Brazilian mate-preference study.” It’s narrower, and it holds up better for being narrower: Aristotle’s specific claims were that resemblance generates ordinary liking, full stop, and that liking still can’t manufacture the thing virtue friendship requires if vice is what’s being resembled. Da Silva, Amaral, Varella, and Valentova weren’t testing Aristotle. They were testing whether Dark Triad self-ratings predict partner preferences, and whether external goods offset crime. But the shape of what they found — strong resemblance-driven attraction on one axis, a hard ceiling that money and fame can’t lift on the other — is close to exactly the split Aristotle drew between what makes people like each other and what makes a bond worth having.

It connects to the La Rochefoucauld piece on strategic kindness and Machiavellianism — that study found competitive niceness tracking manipulation rather than real warmth, the same trait this study found people actively seeking out in a partner once they score high in it themselves. It sits close to the Spinoza piece on long-term singlehood: both push back on the assumption that any pairing beats none — here a criminal profile dressed in fame and wealth still loses badly. And it rhymes with the Rousseau piece on amour propre: a self that needs external validation to feel real is close to the engine behind narcissism scoring the strongest homogamy correlation of the three traits measured here.

The Gap Aristotle Didn’t See Coming

Here’s where the comparison gets more interesting than a clean match.

Aristotle was describing friendship — a bond he treated as separate from erotic partnership, with its own chapter and its own logic in the Ethics. The study is about mate choice specifically, layered with things his framework doesn’t address: sexual attraction, reproductive strategy, the appeal some people report toward risk and intensity in a partner they wouldn’t necessarily want in a friend. He had no model for why narcissism, of the three dark traits, would show the single strongest homogamy correlation (0.69) — a question the researchers don’t fully resolve either, beyond noting that narcissistic self-regard may make a similarly self-regarding partner feel less threatening rather than more.

The scale of the measurement is also a real gap. Aristotle wrote in absolutes — virtue friendship, complete and whole, versus its lesser imitations. The study works in correlations and averages across 400 people, built to catch population-level tendencies, not to predict any one couple. Some people here surely chase prosocial partners regardless of their own Dark Triad standing; some presumably chase antisocial partners despite scoring low themselves. A correlation of 0.69 is strong. It is not 1.0, and it was never going to be.

What This Study Doesn’t Show

Be precise about the size of this. The framing invites more confidence than the data earns.

Hypothetical Profiles, Not Real Relationships

Participants rated written descriptions of partners, not people they’d actually dated or married. A stated preference for a hypothetical profile and the partner someone actually chooses, under real constraints and real chemistry, aren’t guaranteed to match.

One Country, One Cultural Context

400 adults from Brazil is a genuine sample, but it’s one country with its own norms around dating, status, and crime. Whether the same ceiling on fame, wealth, and attractiveness holds in cultures with a different relationship to celebrity or wealth is untested here.

Self-Reported Traits on Both Sides

Every score — a person’s own Dark Triad standing and their partner preferences — comes from self-report. People high in Machiavellianism are, almost by definition of the trait, not guaranteed to self-report accurately.

Correlational, Not Causal

The design measures whether self-ratings and preference ratings move together, not whether one causes the other. A third factor — general comfort with one’s own personality, say — could plausibly drive both scores at once.

What Actually Follows From This

1. “Opposites attract” and “like attracts like” are both true, just for different things. This study is squarely in the like-attracts-like camp for personality style — the stronger someone’s own Dark Triad traits, the more they want a partner who shares them. That’s homogamy, documented here with real numbers, not folk wisdom either way.
2. Status and looks are real currency, but they aren’t infinite currency. Fame, wealth, and attractiveness each nudged a criminal profile’s desirability by a few tenths of a point, not by the several points it would have taken to make it competitive. Aristotle’s ranking of utility and pleasure beneath virtue friendship said as much, without needing a 1-to-7 scale to say it.
3. Ask Aristotle’s actual question about anyone drawing you in by resemblance: resemblance in what, specifically? Sharing a sense of humor or a view of the world is one kind of likeness. Sharing a willingness to manipulate people is another. The data says both produce liking. Only one of them produces something durable enough to be worth building on.

Aristotle never filled out a personality inventory, and da Silva, Amaral, Varella, and Valentova weren’t out to vindicate a Greek philosopher’s account of friendship. But a study built to separate “do dark personalities pair up” from “can money and fame paper over the cost of vice” landed close to the exact split Aristotle drew between what produces liking and what produces something worth keeping: similarity pulls people together reliably, in a lab and in a two-thousand-year-old treatise alike. It just never promised, in either place, that what gets pulled together would be good for anyone involved.

---

This post discusses a 2026 study published in Evolution and Human Behavior (da Silva, Amaral, Varella, and Valentova, DOI: 10.1016/j.evolhumbehav.2026.106889), reported by PsyPost, and Aristotle’s Rhetoric and Nicomachean Ethics. The study relies on self-reported traits and hypothetical partner profiles from a single Brazilian sample, and cannot establish that stated preferences predict who people actually choose as partners.