Article · 11 min read

Your brain on infinite scroll: normative dissociation, explained

Forty minutes gone and nothing remembered. That state has a name, a literature, and an intervention that works with the mechanism instead of against your willpower.

The search that brings people here is a small, specific kind of alarm: you look up and forty minutes are gone, you can’t recall a single thing you scrolled past, and the honest question isn’t “how do I use my phone less” but “what just happened to me?” The productivity internet answers with willpower and dopamine, neither of which explains the memory hole. The research literature has a better answer, and it even has a name for the thing that happened. It’s called normative dissociation, and it is not a character flaw. It’s a described psychological state, it has been measured, and understanding it changes which interventions are worth your time.

The thing that happened has a name

Dissociation sounds clinical, and the clinical version, the kind associated with trauma, is a different and more severe animal. What happens on a feed is the ordinary, non-pathological cousin: normative dissociation, total cognitive absorption with self-awareness dimmed and your sense of agency reduced⁹. You get pulled into a task so completely that the part of you that monitors what you’re doing goes quiet. It’s the same family of experience as getting lost in a book, missing your highway exit, or driving home and not remembering the drive. Nothing is wrong with you; a very ordinary attentional mechanism has been engaged, and the feed is engineered to engage it.

The construct entered this field through Baughan et al. (2022), whose title, “I Don’t Even Remember What I Read”, is the symptom stated as data. They built a custom Twitter client, Chirp, gave it to 43 U.S. participants, and found that design interventions (custom lists, labels showing what you had already read, time-limit dialogs, usage statistics) measurably reduced the dissociation⁹. Three components do the work: absorption (you’re deep in it), diminished self-awareness (the monitor is off), and a reduced sense of agency (the part of you that chooses has stopped choosing).

The later work that operationalizes it most usefully for our purposes is Ruiz, Molina León and Heuer (2024, Mensch und Computer), which builds directly on Baughan’s account and then tries to interrupt it: thirty participants, made to react to each post before the next one arrived rather than scrolling freely. Recall of what they had seen improved significantly under the friction⁸. The memory hole, measured from the other side. It was not free: 53% of them were frustrated by it, a third felt demotivated from carrying on using the app, and one in five wanted more apps to work that way⁸. The memory is the tell. If you remembered every post, you weren’t dissociated. You were just choosing to scroll, which is a different problem with a different fix.

Naming it matters because it relocates the failure. You are not weak-willed for scrolling past your intended stop. You are absorbed, and absorption is precisely the state in which intention has no one home to talk to. Willpower is a function of the self-monitoring system, and normative dissociation is that system going quiet. Asking it to try harder is asking a sleeping person to be more alert.

The loop inside the loop

If self-awareness is switched off during a session, the obvious question is what ever switches it back on. A 2023 study in Proceedings of the ACM on Human-Computer Interaction (the MHCI issue), a one-week field study with 46 participants, looked directly at what ends an infinite-scrolling session, and offered a model with two nested loops¹⁰. The inner loop is the tight one: see content, react, scroll, repeat. It’s a self-sustaining cycle that supplies its own next reason to continue and, crucially, contains no natural exit. Infinite scroll is designed to keep you there; the whole point of removing the page break is removing the moment where you might have stopped. The outer loop is the decision to open or close the app at all, and that’s where sessions actually terminate.

The finding worth carrying out of that paper: most of the reasons people end a session are contextual. Something outside the app intervenes, rather than anything the app did¹⁰. You arrive somewhere. Someone talks to you. The bus stops, the kettle boils, your arm falls asleep. Sessions rarely end because the feed ran out or because you decided, mid-absorption, that you’d had enough. They end because the world got more interesting than the screen and reached in to collect you. This fits the dissociation account exactly: if the inner loop suppresses self-monitoring, then the signal to stop can’t reliably come from inside. It has to come from outside, from a context strong enough to break the absorption. The trance ends when reality interrupts it, not when you decide it should.

Which is a quietly damning fact about any intervention that lives inside the loop.

Why the pop-up stops working

The intuitive fix is to manufacture an interruption: a reminder, a “you’ve been scrolling for 20 minutes” card, a mindfulness prompt that appears mid-feed. Stage an outer-loop moment from inside. It works. For about a week.

The clearest evidence on the shape of that decay is Meinhardt et al. (2026, N=104, seven-day within-subject), still in press. It compared gradual friction interventions on short-form video and found that the pop-up was initially effective but quickly lost impact⁷, something your thumb learns to dismiss before your eyes have read it. The condition that sustained subjective ratings the longest was not the loudest one. It was a gradually intensifying visual friction: something that ramps rather than announces, that changes the environment slowly enough to stay below the threshold where you’d learn to swat it away⁷. The same study found that self-regulation traits moderated the response: for the most impulsive participants, the plain pop-up was the most effective condition of all⁷. No single intervention lands identically for everyone, and that includes anything we’d recommend.

The reason a pop-up habituates while a slow ramp doesn’t maps back onto dissociation cleanly. A pop-up is an event (discrete, repeated, predictable), and the absorbed brain is extraordinarily good at filing predictable events as noise without ever surfacing them to awareness. You can dismiss a modal without leaving the trance. A gradual environmental shift offers nothing discrete to habituate to; there’s no single moment to learn to ignore. It doesn’t try to wake the monitor with a bang. It just makes the room less worth staying in, continuously, until you drift out on your own: the outer loop reached from the environment rather than staged as an interruption inside the inner one.

The self-awareness cue, and what it tells us

There’s a striking recent demonstration of the diminished-self-awareness component being load-bearing. Kim et al. (2026, CHI EA, N=84) ran a laboratory experiment in which an app periodically showed people one of four self-related cues while they scrolled short-form video (a live camera image, a selfie, the person’s own name, or a black screen), and the cues disrupted the mindless viewing¹¹. Being reminded that there is a you here doing this, apparently, reintroduces exactly the self-monitoring that dissociation switches off.

To be very clear, because the mechanism is so vivid it invites the wrong inference: this is not what Dwindle does. Dwindle does not show you your face, does not use your camera, and has no interest in staring at you. The Kim result is cited here for one reason only. It’s clean experimental confirmation that the second component of normative dissociation is real and load-bearing. Restore self-awareness and the absorption breaks. That’s a finding about the mechanism, and it points at a broad family of interventions that work by making you briefly aware of yourself again. A live camera image is one, fairly confrontational, member of that family. There are gentler ones, and the study hints at it from its own control condition: the black screen, included as the control, was the one people most wanted to keep using¹¹.

Grayscale, and the stimulation the trance runs on

Absorption needs fuel. The inner loop sustains itself because each post pays out a small hit of visual salience (the saturated thumbnail, the red badge, the vivid cut every half-second), and salience is what captures attention pre-consciously, before the deliberate self has a vote. That’s the reward density the trance runs on. Strip the color and you thin the fuel.

This is where the evidence gets unusually tidy for a wellbeing intervention. Across published field studies, running a phone in grayscale reduces objectively measured screen time by roughly 20 to 40 minutes per day^{1 2 3}, logged from the device, not self-reported, and consistent across different countries and study designs. The detail that matters most for the dissociation story comes from Dekker and Baumgartner (2024, Mobile Media & Communication): about 20 minutes per day less screen time, and no change in the number of daily unlocks². The reach for the phone survives entirely. Grayscale doesn’t touch the reflex that picks it up. What shrinks is the session. You still open Instagram; you put it down sooner, and participants reported more control over their phone use and less sense of overusing it².

That split is the mechanism in one finding. Grayscale doesn’t fight the outer loop (you open the app as often as ever) and it doesn’t block anything (there’s nothing to override). It drains the inner loop: it lowers the payout that keeps absorption self-sustaining, so the trance is shallower and the contextual off-ramps that end sessions arrive sooner and land harder. It works at the layer where the trance actually lives, which is neither access nor willpower but the moment-to-moment reward density of the feed itself.

The trance, not the task

Every intervention above is aimed at the same target from a different angle: the absorbed, low-self-awareness, memory-poor state, not your legitimate reasons for picking up the phone. That distinction is the whole game. You have real tasks on that device: the message, the map, the recipe, the actual reason you opened it. A blocker can’t tell the trance from the task, so it gates both. Grayscale can’t either, exactly. But it doesn’t need to, because it never blocks the task; it only thins the stimulation, and the task survives thinning where the trance doesn’t.

If forty missing minutes is the search that brought you here, the state that ate them has a name, a literature, and, for once, an intervention that works with the mechanism instead of against your willpower.

Related guides: the science of grayscale, why de-stimulation beats blocking, and the stimulation stack.