Research suggests the brain doesn't process scrolling through a feed like settling into a book — the constant change repeatedly asks attention to reset and decide what matters, which helps explain why an hour online can leave you feeling more mentally exhausted than informed

An hour disappears into the phone and the honest inventory afterwards runs to three items: a man cutting marble, somebody’s strong opinion about interest rates, a dog on a skateboard.

None of it is retained in any way worth having. The hour still lands like work.

That mismatch, between the feeling of effort and the absence of anything learned, has a reasonable research explanation behind it, and willpower barely features in it.

A feed is a run of cold starts

Reading a book is one long continuous job. The brain loads a setting, a cast, an argument, and then keeps using that same setup for as long as the reading lasts.

A feed offers nothing of the kind.

Each post arrives with its own subject, tone, format and rules, which means the machinery of attention has to reconfigure itself before any of the content can be processed.

A feed adds another recurring job: a verdict. Each item has to be sized up for whether it is true, whether it is funny, whether it is aimed at anyone in particular. That call gets made before the thing can be dismissed, and then made again at the next one, and the one after that.

Psychologists have taken that reconfiguration apart. In a paper published in the Journal of Experimental Psychology: Human Perception and Performance, Joshua Rubinstein, David Meyer and Jeffrey Evans described two stages: goal shifting, meaning the decision to do this now instead of that, and rule activation, meaning the swap of mental instructions from the old job to the new one. Both stages take time. A single switch might cost only a few tenths of a second. That sounds trivial until you count the switches inside an hour of scrolling.

Their experiments used arithmetic and shape-sorting rather than phones, so applying the finding to a feed is an inference, not a measurement.

The part that stays behind

Switching also leaves something on the floor. Sophie Leroy, now a professor at the University of Washington Bothell, coined the term attention residue in a 2009 paper in Organizational Behavior and Human Decision Processes, after experiments in which people were pulled off one task and pushed onto another. Performance on the second task dropped, because a slice of their attention was still parked on the first, and the effect was sharpest when the first task had been left unfinished.

Scrolling has the shape of that setup. Post after post is dropped mid-thought and replaced by the next one, which looks a great deal like the unfinished condition Leroy identified as the costly one. Her participants were working through lab tasks, though. The parallel to a phone is drawn by analogy, and nobody has run the experiment there.

Forty-seven seconds

Gloria Mark, a professor of informatics at the University of California, Irvine, has been logging how long people stay on a single screen since 2004. Back then the average was about two and a half minutes. In an interview with the University of California, she put the recent figure at roughly 47 seconds, with a median of 40, meaning half of all observed stretches were shorter than that.

That number needs handling with care. It is assembled from separate studies across two decades, not one group of people followed the whole way, and what it measures is screen behaviour. People still read long books. They just do it somewhere other than a browser tab.

Why a book behaves differently

Does the medium itself change how much gets understood? Pablo Delgado of the University of Valencia and three co-authors pooled 54 studies covering more than 171,000 participants in a meta-analysis for Educational Research Review, and found a consistent, modest advantage for paper. The gap widened when reading was timed instead of self-paced, held for informational texts but not narrative ones, and grew across the years studied, an unhelpful trend for anyone expecting digital natives to close it.

Pooling dozens of experiments makes that result sturdier than any single study, though the effect itself stays small.

Maryanne Wolf, who studies the reading brain at UCLA, argues in an op-ed for the Guardian that skimming squeezes out the slower processes that make reading useful: drawing inferences, making analogies, testing whether a claim is true, stepping into another person’s point of view. Move the eye fast enough and there is no room left for any of it.

Where the tiredness comes from

Fatigue needs its own explanation, and it comes from a separate strand of Mark’s research.

In a study presented at the 2008 CHI conference, Mark, working with Daniela Gudith and Ulrich Klocke, interrupted people mid-task and measured what happened. The interrupted group finished faster, with no drop in quality, which was the surprise. They also reported more stress, more frustration, more time pressure and more effort.

Compensation, in other words.

The brain appears to cover the cost of fragmentation by spending harder. Mark’s participants were working rather than scrolling, and the equivalent test has not been run on a feed, so the link is suggestive rather than established. Even so, the flatness and irritability at the end of a long scroll, the vague sense of having been mugged, match what those participants described.

Which puts the whole exercise in a different light. An hour on a feed involves several hundred small decisions made at speed, each one demanding a quick ruling on what the thing is and whether it matters. The content was the least of it. The deciding was the job.

Nobody sends an invoice for that. The tiredness is the receipt.