Your Attention Span Dropped From 2.5 Minutes to 47 Seconds

In 2004, UC Irvine professor Gloria Mark started timing how long people could look at one screen before switching to something else. She used a stopwatch and shadowed office workers through their day, logging the exact second their attention moved. The average was two and a half minutes. By 2012, using the same method, it had fallen to 75 seconds. Measured again over the last five years - by Mark's team and, separately, by other researchers who replicated the work - it has settled at around 47 seconds.

That's not one study with a catchy number. It's the same measurement, repeated for two decades, converging on the same answer each time.

A 20-Year Decline, Measured the Same Way Each Time

What makes Mark's numbers hard to dismiss is the method. This isn't a survey asking people to guess how distracted they feel - it's observed behavior, timed. Early studies used stopwatches and in-person shadowing; later ones used computer-activity logging to track window switches automatically. Mark ran the measurement across different workplaces, different job types, and college students, and independent researchers replicating her method have landed in the same 44-to-50-second range. Twenty years, several different research groups, one consistent trend line.

What's Actually Driving It

Mark's own explanation points to task-switching load in general - open-plan offices, notifications, juggling more digital tools at once. But a 2026 meta-analysis published in Psychological Bulletin, one of psychology's most rigorously reviewed journals, narrows the picture considerably. Researchers pooled data from 98,299 people across 71 independent studies (full study on PubMed) and found that attention and impulse-control problems weren't just correlated with "screen time" broadly - they were specifically and consistently correlated with short-form video use. Attention showed a correlation of r = -.38; inhibitory control (the ability to stop an impulse, like scrolling to the next clip) showed r = -.41 - the two strongest associations found anywhere in the analysis. In psychology research, where most effects are small and inconsistent across studies, a moderate correlation holding steady across 71 separate samples is unusually strong evidence.

Why Short-Form Video Specifically

The likely mechanism isn't just "too much screen time" - it's the absence of a natural stopping point. A short-form feed autoplays the next clip before you've decided whether you want it, and every clip is a small, unpredictable reward: sometimes forgettable, occasionally exactly the right joke or the right piece of information. That unpredictability is the same variable-reward pattern that makes slot machines, rather than fixed payouts, the most compelling form of gambling - the brain keeps going not because the last video was good, but because it can't predict whether the next one will be. A book, a long-form video, even a normal social media post has a natural pause built in, a moment that requires an active decision to continue. An infinite short-form feed removes that decision point entirely.

What the Same Research Says It Isn't

The Psychological Bulletin meta-analysis is also useful for what it didn't find. Despite a common assumption in pop-science coverage, the pooled data showed no significant association between short-form video use and body image or self-esteem. The measurable effects were specific - attention, impulse control, stress and anxiety - not a blanket "social media ruins everything" finding. That precision is worth taking seriously in both directions: the attention effect is real and well-replicated, but it's not evidence for every claim made about short-form video.

What Actually Helps

Neither body of research points to willpower as the fix. Mark's own work on task-switching suggests that structural changes - fewer competing demands on attention, fewer forced interruptions - matter more than trying to simply concentrate harder. Applied to short-form feeds specifically, the same logic holds: since the problem is the missing stopping point, not the content itself, the more durable interventions are the ones that restore a stopping point rather than the ones that rely on remembering to quit. A daily time limit doesn't do that - it only intervenes after the scrolling has already happened. Something that interrupts the autoplay loop itself, at the moment it would otherwise continue automatically, addresses the actual mechanism instead of the symptom.

Twenty years ago, switching your attention away from a screen took two and a half minutes to happen naturally. Today it takes 47 seconds - and the research increasingly points to a specific, identifiable reason why.