Science of Attraction, Compatibility & Romance | Dr. Paul Eastwick

2026-06-22 - 170 min - source - Read full transcript

Andrew Huberman (host)Paul Eastwick

Key insights

Attraction starts as a marketplace but stops being one once people actually get to know each other.

Eastwick's forehead-number classroom demo shows people behaving like a strict hierarchy market when judging strangers at a glance. But real acquaintances only agree about two-thirds of the time on who is attractive, and the longer two people spend together, the more their impressions diverge from consensus, opening room for someone rated a 'six' on average to be seen as a 'nine' by one particular person.

attraction-formation-over-time

Attraction typically builds slowly over weeks, not in an instant spark.

Contrary to the idea that chemistry must be maximal on first contact, first impressions of eventual partners are usually 'middling.' Eastwick describes a window of uncertainty that slowly collapses toward a stable, positive impression over roughly ten minutes to thirty days, driven by an accumulation of small positive moments across different trait categories (funny, smart, a good listener) rather than one big signal.

attraction-formation-over-time

Men and women want the same things once tested against real people, not self-reported preferences.

Twenty years of speed-dating studies across more than 40 countries find no gender difference in how much attractiveness or earning potential/ambition sways actual liking of a real person met face to face, even though men and women report different priorities when just rating traits in the abstract. Eastwick argues self-report scales capture stereotypes, not revealed behavior.

compatibility-and-similarity-myths

The 'older man, younger woman' pairing pattern doesn't reflect an asymmetric preference at the point of attraction.

In a roughly 4,400-person professional matchmaking dataset where the searching men were about four years older than the available women on average, both men and women who ended up wanting a second date showed the same-sized preference for a younger partner. The age gap in successful couples is therefore explained by pool composition, not a stronger male preference for youth.

compatibility-and-similarity-myths

Actual trait similarity barely predicts compatibility; perceived similarity, which is self-serving, does.

Matching two strangers on real, measured similarity in interests and values performs close to a coin flip at predicting whether they will click. What tracks with relationship satisfaction is perceived similarity, which is elastic: people in happy relationships selectively emphasize whatever they do share and downplay mismatches, a form of motivated reasoning Eastwick says is functional rather than delusional.

compatibility-and-similarity-myths

Objective income has a small effect on how a partner feels about someone; lack of free time is the bigger risk.

A high earner's actual income level shows very small effects on partner satisfaction. The more consequential variable is whether that income came with enough free time to be present in the relationship; a demanding, high-earning partner who is rarely around creates more strain than the income itself resolves. Separately, the historical finding that marriages were more fragile when a wife out-earned her husband stopped holding after the 1990s.

compatibility-and-similarity-myths

Dating apps mechanically amplify the marketplace dynamic that fades in ordinary in-person life.

Because swiping is a snap judgment on strangers, apps reproduce a 'kleptocracy' where the most swiped-right users capture a hugely disproportionate share of attention, and women swipe yes on roughly 5% of men versus men's roughly 50%. That gap narrows to about 2x, not 20x, in studies of real-world propositions among people who already know each other, showing the extreme skew is an artifact of stranger-judgment contexts rather than a fixed sex difference in selectivity.

dating-app-market-dynamics

Feeling like a good lover is one of the strongest predictors of relationship satisfaction, not a secondary factor.

Across both established couples and people rating someone they're initially attracted to, the subjective sense that a partner is (or would be) a good lover is one of the strongest correlates of how positively someone feels about the relationship and whether they want it to continue. Eastwick is explicit that this doesn't fade in importance once other compatibility is established.

relationship-protective-mechanisms

Derogation of alternatives protects monogamous relationships, and desire can rebound from outside attraction.

People in happy relationships subjectively downgrade the desirability of potential alternative partners relative to outside observers' ratings, a protective bias against straying. In parallel, recent studies show that fantasizing about someone other than a partner also raises sexual desire for the current partner (but not the reverse), supporting Esther Perel's argument that outside attraction is not itself the danger; unresolved risk comes specifically from repeated, secretive escalation with that other person.

relationship-protective-mechanisms

Relationship duration alone is a mild negative predictor of satisfaction; shared narrative is what actually helps.

On average, controlling for other factors, people report being happier earlier in a relationship than later, making raw duration a weak or even negative marker of quality. What buffers against that decline is a felt shared story built through obstacles overcome together, which is why breakups are hard partly because they sever continuity with one's own narrative, not just because they end a source of support.

relationship-protective-mechanisms

Repeated small-group activities build relationships more reliably than value-matching or filtering.

Historical mechanisms for meeting partners (church groups, recurring beach volleyball games, potlucks, improv classes) work primarily because they force repeated low-stakes exposure to the same people over time, not because the specific activity filters for shared values. Eastwick is candid that his field cannot say which single activity maximizes odds of finding a partner within a set window; the general prescription is to be around the same people on repeated occasions.

social-infrastructure-for-meeting-partners

Couple friends provide relationship validation without direct advice-seeking, which carries its own risk.

Directly soliciting friends' opinions about a struggling relationship risks either suffering in silence or getting biased outside input that doesn't reflect what two partners privately experience. Research on couple friends and double dates suggests a better mechanism: being observed as a couple by trusted friends provides a background sense of social support (an unused 'bank account') without anyone formally weighing in.

social-infrastructure-for-meeting-partners

Books referenced

Bonded by Evolution: The New Science of Love and Connection - Paul Eastwick - Eastwick's book, referenced at the close of the episode as the fuller treatment of the research discussed throughout.
Protocols: An Operating Manual for the Human Body - Andrew Huberman - Huberman's own forthcoming book, mentioned in the end-of-episode promotional segment, unrelated to the interview content.

Media referenced

True Romance - movie - Used as an example of the 'we're both screwed up but good people' attraction model, where two people from hardscrabble backgrounds fall for each other.
Say Anything - movie - Discussed as a depiction of old-school dating: the couple attends a party together but also watches each other interact with other people, rather than interviewing each other one-on-one.
The Town - movie - Cited for the Ben Affleck 'who's driving' scene as an idealized model of loyal, no-questions-asked male friendship.
Call Me by Your Name - movie - Referenced as an example of how social danger historically delayed disclosure of romantic feelings in same-sex relationships.
Mad Men - show - Cited as an illustration of the historical risk (job loss, social ruin) that came with disclosing same-sex attraction.

Companies

Instagram - Described by Huberman, based on an informed source, as the largest dating app in the world by usage, despite not being marketed as one; direct-message engagement functions like a dating app layer.
David Protein - Sponsor; protein bar.
Lingo - Sponsor; continuous glucose monitor.
AG1 - Sponsor; vitamin/mineral/probiotic drink.
Element - Sponsor; electrolyte drink.
Function - Sponsor; lab testing membership.

Techniques and frameworks

Forehead-number classroom demo - Eastwick's demonstration of pure mate-value marketplace behavior: students wear a hidden number and try to pair with the highest number possible, illustrating how initial-attraction settings behave like a market until people get to know each other.
36 Questions / Fast Friends procedure - A structured question set (e.g., 'what's one thing you've never told anyone and what's stopping you') used on longer dates to elicit reciprocal self-disclosure and accelerate genuine closeness.
Derogation of alternatives - The tendency of people in happy relationships to subjectively rate potential alternative partners as less desirable than they objectively are, functioning as a psychological defense of monogamous commitment.
Couple friends / double-date validation - Spending time with another couple as an indirect way to feel relationship support without directly soliciting friends' opinions of a partner, which research links to better relationship outcomes.
Attachment framework (secure, anxious, avoidant) - Evolutionary theory of bonding; discussed as capable of shifting over time, with avoidant partners becoming more secure through a sustained relationship with the right partner.

Summary

Andrew Huberman hosts psychologist Paul Eastwick, a professor at UC Davis who studies attraction and relationship formation through speed-dating experiments, longitudinal couple studies, and matchmaking-company datasets, to challenge the evolutionary "marketplace" narrative that dominates popular dating discourse. Eastwick agrees the marketplace model captures something real about how strangers size each other up at first glance, illustrated by a classroom demo where students wear hidden numbers and pursue only the highest-value partner they can find. But he argues this dynamic largely dissolves once people spend real time together: agreement about who is attractive drops from a strong consensus among strangers to something closer to idiosyncratic, and a person rated average by the group can be seen as uniquely compelling by someone who has actually gotten to know them.

Much of the middle of the conversation dismantles specific gender-difference claims using data instead of self-report. Speed-dating studies spanning 20 years and more than 40 countries find no gender difference in how much attractiveness or ambition/earning potential drives real liking once people meet face to face, even though men and women report different priorities when simply rating traits on a scale. A large matchmaking dataset similarly shows no asymmetric preference for youth between men and women once real dating outcomes are examined, undercutting the standard older-man/younger-woman narrative. Eastwick also walks through why dating apps distort behavior relative to ordinary acquaintance: because swiping forces snap judgments on strangers, it reproduces an extreme, "kleptocratic" concentration of attention on the most conventionally desirable users, a skew that shrinks dramatically once people already know each other in real contexts.

A recurring thread is that attraction is a slow accumulation process rather than an instant read. Early impressions of eventual long-term partners are typically "middling," and what actually builds attraction is a string of small, idiosyncratic moments (unscripted reciprocal disclosure, banter, watching how someone treats other people) collected over roughly ten minutes to thirty days. Eastwick connects this to why matching people on real, measured similarity performs close to a coin flip at predicting compatibility, while perceived similarity, which people construct selectively around whoever they already like, tracks much better with satisfaction. He frames this motivated reasoning as functional rather than a flaw: the happiest couples are actively doing it.

The back half turns to what sustains relationships once formed. Feeling like a good lover, or being perceived as one, is among the strongest predictors of overall relationship satisfaction and is not a secondary consideration that fades once other compatibility is settled. Eastwick describes "derogation of alternatives," the tendency of happy partners to subjectively downgrade other potential partners' desirability, as a real protective mechanism, and cites recent research showing that fantasizing about someone else can even boost desire for a current partner, aligning with Esther Perel's argument that outside attraction itself is not the danger; the risk is specifically repeated, secretive escalation. He also flags that raw relationship duration is, on average, a weak or mildly negative predictor of satisfaction; what protects against decline is a felt shared narrative built through difficulty together.

Throughout, both host and guest return to the value of repeated, low-stakes group activities (church groups, recurring pickup sports, potlucks, improv classes) as the most reliable way to meet people, arguing the mechanism is repeated exposure to the same people over time rather than any particular activity filtering for compatible values. Eastwick is candid that relationship science cannot currently specify which single strategy maximizes someone's odds of forming a relationship within a set window. The episode closes on a broader argument against "heteropessimism": Eastwick contends men and women want fundamentally the same things from relationships, and that online, algorithm-amplified narratives of extraction and failure on both sides obscure how much cooperative, benevolent behavior actually characterizes real romantic relationships.

Notable Quotes

"The more that people spend time together getting to know each other, it reduces some of those market forces that give the desirable people all the advantages." - Paul Eastwick

"Humans are stories, right? We're narratives. And we want other people to be privy to that narrative and then maybe eventually be a part of it." - Paul Eastwick

"The problem is not that your partner was sexually attracted to somebody else. The problem is usually in the escalation." - Paul Eastwick

"The way you destroy a society is to get the men and the women to hate each other." - Andrew Huberman

"I think the key lesson here is like, believe your subjective experience when you're interacting with somebody and you're getting to know them." - Paul Eastwick