Neuroscience

Dating Apps, Dopamine, and the ADHD Brain

July 18, 2026 · by Brendon Kennedy

In short: Dating apps hold your attention with variable, slot-machine-style rewards — and divided attention is exactly what keeps your brain from encoding what a match told you, which lands hardest on the ADHD brain. Here's the neuroscience, the design, and where it went in court.

A few weeks ago I was on a first date — I’ll call her Jane — and somewhere between the second drink and a walk to nowhere in particular, she asked me a small follow-up about a story she’d told me twenty minutes earlier. And I had nothing. Not a fuzzy version, not a wrong guess — a clean blank where the memory of a person I actually liked was supposed to be. I covered for it, the way you do. But the drive home turned into one long, quiet interrogation of myself: what is wrong with me, that I can be that interested and still lose the thread?

I’ve built a career on paying very close attention to systems — how they’re wired, where the incentives hide, what they’re really optimized for. So the question that stuck with me afterward wasn’t really about that evening. It was two questions, actually, and they turned out to be the same shape: Is something wrong with my memory? Or is something wired into the app I met her on?

This is my attempt to answer both honestly, with the papers open. It’s a personal essay with citations, not medical advice — and everything I say about anyone I’ve dated stays at the level of “a person, on a date,” on purpose.

Is ADHD a memory problem or an attention problem?

Short answer: it’s an attention problem that looks like a memory problem. ADHD — attention-deficit/ hyperactivity disorder, a condition roughly 6% of U.S. adults now carry a diagnosis for¹ — is usually described as a difference in attention regulation and reward timing. What’s less discussed is how directly that spills into memory.

The distinction that reorganized my thinking is encoding versus storage. Encoding is how an experience gets attended to, organized, and written down in the first place. Storage is holding it. Retrieval is getting it back. When researchers pool the studies, ADHD’s long-term-memory difficulty sits largely at the encoding stage — reduced spontaneous, organized, deep encoding — with storage and retrieval comparatively intact for everyday episodic memory.² Nothing is falling out of the vault. Less is being written down to begin with.

That single reframe changes the emotional weight of forgetting what a date said. It isn’t “I don’t care enough to hold onto this.” It’s “the note got taken faintly, in a distracting room, by a brain whose attention gate is quieter than average.”

Why divided attention wrecks a memory before it even forms

Here’s the part that isn’t about ADHD at all — it’s about everyone. In a classic set of experiments, Craik and colleagues had people learn material while doing a second task. Splitting attention during encoding sharply reduced what they could later recall or recognize. Splitting the very same attention during retrieval barely touched accuracy.³ Attention is most protective at the input stage. Distraction while you’re taking something in is far more expensive than distraction while you’re trying to remember it.

Underneath that is an interplay: prefrontal control biases what gets selectively encoded, and the hippocampus rapidly binds the moment’s context — the face, the room, the feeling — into something you can later retrieve.⁴ Starve that process of attention and you get a thin, under-bound trace with few handles on it.

Now put a phone on the table. Every buzz, every half-thought about the three other conversations waiting in an app, is a second task running during encoding. The memory of what she actually said is being written in the worst possible conditions. (I break down the ADHD-specific version of this in why you forget what your matches told you.)

A parallel from an unlikely place: why none of us remember being babies

I want to bring in one study carefully, as a parallel — not the same mechanism. In 2025, a team scanned awake infants in an fMRI and found that greater hippocampal activity while a baby looked at a new image predicted whether the baby later showed memory for it, with the effect emerging around age one.⁵ The provocative read: infantile amnesia — the fact that none of us remember our first years — may be less about failing to form those memories and more about later being unable to retrieve them.

The reason I find it clarifying isn’t that a first date is like infancy. It’s that the study pries apart two things we usually collapse: a memory can be genuinely encoded and still be unreachable, and a memory can feel “forgotten” for reasons that have nothing to do with whether it was ever properly written down. When I can’t recall a detail from a conversation, the useful question isn’t “why is my memory broken” — it’s “was this ever encoded well enough to have a way back to it?”

What sitting across a table does that a chat thread can’t

A retrieval cue only works insofar as it recreates something that was present when the memory formed — the encoding specificity principle.⁶ That has a blunt consequence for how you meet someone.

In person, an exchange is encoded through many channels at once: a face, a voice, the space, the timing, the feeling in the room. Semantically congruent multisensory encoding produces stronger later memory than a single channel — and the advantage survives even when you’re later tested on just one of those channels.⁷ More channels at encoding means more distinct cues, each a potential route back.

I want to be honest about the limits of this, because the tidy version is too tidy. Multimodal richness is not an automatic memory booster: when researchers added faces and voices to the learning of biographical facts, recall didn’t reliably improve.⁸ The benefit is real for congruent perceptual material, not a blanket “in-person always wins.” But a text thread is close to the floor of this effect — one channel, stripped of voice, face, timing, and shared space. It is a genuinely impoverished place to encode a person.

The app isn’t a neutral pipe: variable rewards and the slot-machine loop

Everything above is about my brain. This section is about the interface’s design, which is engineered in the opposite direction.

The engine is the variable-ratio reward — a reward delivered after an unpredictable number of attempts. It’s the schedule behind a slot machine, and its signature is a high, steady rate of trying. Swiping mirrors it: a match arrives after some unknowable number of swipes.¹⁰ Dopamine is central here not as a pleasure chemical but as an anticipation signal — it tracks the prediction of reward and peaks under maximum uncertainty, which is exactly the state an unpredictable match schedule manufactures.⁹

I’ll hedge where the evidence hedges: a dating match isn’t a pure variable-ratio schedule, because it also depends on another person’s independent choice, and there’s no neuroimaging of someone mid-swipe. It’s an analogy — but a load-bearing one, and not an accidental one. A former Google design ethicist has argued for years that these interactions borrow the slot machine’s core trick, “intermittent variable rewards.”¹¹ (I pull that mechanic apart on its own in why dating apps feel like slot machines.) The point isn’t that anyone is cackling in a boardroom. It’s that the mechanic is well understood, and the interface leans on it.

Hinge’s “Your Turn Limits,” explained

Which makes one specific feature genuinely interesting. In September 2024 Hinge rolled out Your Turn Limits globally.¹² The mechanic: once you have eight or more conversations waiting on your reply, Hinge stops you sending or receiving new Likes until you answer — or end — some of them.¹³ (Full walkthrough: Hinge’s Your Turn Limits and the 8-chat cap.) Try it:

Read one way, it’s the rare humane move: a cap that pushes you to actually engage the people in front of you instead of collecting an unmanageable pile of open threads. Hinge frames it as quality over quantity, and — connecting it back to the memory science — eight parallel conversations is precisely the attention fragmentation that guarantees you’ll encode almost none of them well. A cap on open threads is, incidentally, a cap on how badly you can fragment your own attention.

”Designed to be deleted”? The incentive underneath

Hinge’s tagline, adopted in 2018, is “Designed to be Deleted.”¹⁴ Match Group says its brands are each built to increase users’ likelihood of connecting and forming meaningful connections.¹⁶ Taken at face value, Your Turn Limits fits that story.

But there’s a structural tension that academics have named directly as a conflict of interest: a company whose stated goal is to help you leave still answers to a for-profit incentive to maximize engagement and lifetime value.¹⁵ (I sit with that tension in “designed to be deleted”? incentives vs. your goals.) That’s not an accusation of bad faith — a genuinely satisfied user is also great word-of-mouth marketing. It’s just the honest shape of the incentives, and it’s why a single pro-wellbeing feature is worth reading closely rather than cheering reflexively.

The lawsuit that put this in a courtroom

In February 2024 this stopped being a philosophical debate. A putative class action — Oksayan v. Match Group, the company behind Hinge, Tinder, and The League — alleged that those apps were designed to be addictive, pleading consumer-protection, product-liability, and negligence theories, and asking for warnings and changes to how the apps are marketed.¹⁷ I’ll be precise, because it matters: these are allegations, never tested on the merits. (The full case breakdown is its own piece: was Hinge designed to be addictive?)

Here’s what actually happened to it, which most write-ups never circle back to. The court granted Match Group’s motion to compel arbitration on December 10, 2024, and stayed the case — a ruling about where the dispute is heard, not whether the claims are true.¹⁸ By late 2025 the matter had dropped out of Match’s securities filings, and the public docket shows it terminated on December 3, 2025 via voluntary dismissal without prejudice.¹⁹ For balance, Match Group’s position throughout was blunt: a spokesperson called the suit meritless and said the company’s model “is not based on advertising or engagement metrics.”²⁰

So the courtroom didn’t resolve the question. It redirected it into private arbitration and then closed. Which means the interesting judgment is still yours to make — with the mechanics in front of you rather than a verdict handed down.

So — am I broken?

No. And neither, probably, are you.

What I have is a regulation-and-timing difference in attention that makes encoding quieter than average. What the app has is an interface tuned to fragment attention with unpredictable rewards. Put those two together on a date where a phone is buzzing, and “I forgot what she told me” needs no story about a broken memory. It’s two well-documented forces meeting at the exact worst moment — the moment of encoding.

The practical move that falls out of this is almost embarrassingly simple, and it’s the one thing I’d stand behind: when it’s a person I actually want to remember, the phone goes away, and the app’s pile of open threads doesn’t get to run as a second task while I’m trying to encode a human being.

Here’s the small thing I changed after Jane, and I’m holding it as an experiment, not a cure: when I’m actually across from someone I want to remember, the phone goes face-down and out of reach, and I don’t let the app’s pile of unanswered threads sit in the back of my head like a second conversation running in parallel. Whether that’s genuinely improved my recall or just made the evenings feel less fractured, I honestly can’t tell you yet — it’s too soon, and I distrust clean before-and-after stories about my own attention. But I’ve stopped reading the blank as proof that something’s broken in me, and started reading it as two systems — mine and the app’s — both pulling attention away from the one moment that was actually worth encoding.

← All writing