One version of why humans are as intelligent as we are, proposed by the psychologist Geoffrey Miller, holds that a meaningful part of it came from sexual selection rather than survival pressure alone. The idea is that intelligence works like a costly, hard-to-fake signal, similar in principle to a peacock’s tail, and that people have long preferred more intelligent partners because intelligence is difficult to produce without an underlying healthy, well-functioning brain. It’s a genuinely interesting theory. Two research teams have now tested a fairly direct prediction that falls out of it, and in both cases, the prediction didn’t hold up.
What two studies actually tested
Julie Driebe, Morgan Sidari, and six co-authors published the more direct test in 2021, in the journal Evolution and Human Behavior. Their first study had 88 men complete a real intelligence test, then had 179 women rate each man’s intelligence, funniness, physical attractiveness, and overall mate appeal from short video clips. Their second study put 763 participants through actual speed dating, two to five sessions each, with intelligence measured beforehand and mate appeal rated after each three minute interaction.
A separate team, Gabriela Hofer, Roman Burkart, Laura Langmann, and Aljoscha Neubauer, ran something similar and published it the same year in the Journal of Research in Personality. Their version measured verbal, numerical, and spatial intelligence, creativity, and emotional competence in 87 women and 88 men, who then went on real speed dates with up to 14 people each, 1,094 speed-dating encounters in total.
The distinction that mattered
Both teams found the same basic pattern. People could accurately tell how intelligent someone else was, even from a short video or a brief conversation, so intelligence itself wasn’t hiding. But that accurate perception didn’t translate into being seen as a more appealing date. What did predict mate appeal was perceived intelligence, meaning how smart a person came across as, which is a related but genuinely separate thing from how they actually scored on a test. Driebe and her colleagues also checked whether funniness explained the connection, since Miller’s theory treats humor as a kind of intelligence display. It didn’t. Measured intelligence had no relationship with how funny someone was rated as being, which undercuts that particular piece of the theory too. In the Hofer team’s data, the pattern held for men across the board, and for women, only creativity showed any direct link to mate appeal among the abilities they measured, not intelligence and not emotional competence.
Hofer’s team added one more layer worth naming plainly. Once they statistically controlled for physical attractiveness, the already modest effects of both measured and perceived abilities shrank further still. That’s a useful check against reading too much into any of this. It isn’t that intelligence secretly beats looks once you measure things properly. If anything, the opposite showed up in the data: physical attractiveness kept coming out as the strongest single predictor of mate appeal across both papers, and whatever intelligence was contributing had to compete with that for what little variance remained.
Why “too smart” might cut both ways
A separate line of research suggests the picture gets stranger, not simpler, at the very top end. Gilles Gignac and Zoe Callis, publishing in the journal Intelligence in 2020, asked people to rate how desirable a hypothetical partner would be at different IQ percentiles. On average, a partner at the 90th percentile, an IQ of around 120, was rated the most desirable, and the 99th percentile came out lower than that, a small but real dip rather than the number simply climbing forever. When they asked people why they’d rated the very top tier lower, about 60 percent pointed to general compatibility concerns and about 40 percent specifically worried that an exceptionally intelligent partner would come with interpersonal skill difficulties.
What this doesn’t settle
I don’t think this closes the book on why humans are as intelligent as we are, and neither do the researchers behind these studies. Both speed-dating papers were explicit that they measured initial attraction over a handful of short interactions, not what matters once two people actually know each other. Driebe and her colleagues note that intelligence still predicts real, practical relationship benefits over time, things like earning capacity, problem solving, and being a reliable long-term partner, even if it doesn’t move the needle on a first three minute impression. Whether those benefits eventually show up as attraction once someone has had more than a speed-dating round to notice them is a different question, and one none of these three studies were actually designed to answer.
What I’d take from this
I wrote recently about how strangers tend to agree on someone’s looks, but that agreement dissolves the longer people actually know each other, and this feels like a companion finding rather than a contradiction. Intelligence, it turns out, might be a similarly slow variable, something that plays almost no role in the first three minutes and possibly a real one much later, which would make it a strange thing to expect a first date, or a short video clip, to reveal.
