Digital Distraction and Focus: What the Research Actually Shows

Almost every claim you have read about phones and attention falls into one of three groups: measurements that hold up, measurements that were never taken, and measurements that were taken but do not mean what the headline says. This guide sorts the evidence into those three groups, and points you to the article that goes deeper on whichever part is your problem.

It carries no new research. Every claim below is a summary of something already sourced in one of the four articles it links to, and each links back so you can check the working.

The number everyone quotes, and where it does not come from

The eight-second attention span, usually paired with a goldfish, is the most repeated statistic in this field. It is also the one we could not source. A documented retrieval did not locate an originating study. The figure traces to a 2015 Microsoft Canada marketing report, which credited a commercial statistics aggregator that has never produced the underlying research. The goldfish comparison has no research behind it at all.

What has actually been measured is narrower, and it needs more care than it usually gets, because the numbers do not all come from the same kind of source. Gloria Mark's group at UC Irvine has spent two decades logging how long people stay on one screen before switching. The early figure is peer-reviewed: González and Mark's 2004 observational study of information workers reported roughly two to three minutes on a single tool or document before a switch. The most recent peer-reviewed figure is a median of about 40 seconds, from Mark and colleagues' 2016 in-situ study of 40 workers. The widely repeated 75-second and 47-second figures do not appear in any paper a documented retrieval could locate; Mark reports them in her 2023 book and in interviews.

So the honest summary is this. Several observational studies report shorter switching intervals in later datasets, but the estimates are not directly interchangeable because the studies used different definitions, samples and logging methods. And none of it measures attention capacity. It measures what people do, which is not the same thing.

How to fix your attention span works through that distinction in detail.

What an interruption actually costs

Switching between tasks has a measurable price. Rubinstein, Meyer and Evans (2001) measured the time cost of switching rules mid-task in a series of laboratory experiments; Monsell's 2003 review set out how consistently switch costs appear, while noting that their interpretation is still contested. Leroy (2009) described attention residue: in two lab experiments, part of a person's attention stayed with the previous task after they had moved on, with the effect reduced when the earlier task was finished.

Mark, Gudith and Klocke (2008) found something more uncomfortable. In a lab study of 48 participants, interrupted work was completed at least as fast as uninterrupted work, but people reported significantly more stress, frustration, time pressure and effort. Speed is not the cost. The cost is what the recovery takes out of you. Mark's field data also puts a little over half of all interruptions down to self-interruption, which means the notification is only half the problem.

The work time you lose to task switching covers the mechanism, including Altmann and Trafton's (2002) computational model of why resuming a task is harder than starting one.

Why "distracted" and "bored" turn out to be the same problem

Boredom is usually treated as the opposite of distraction. The research treats it as a close relative. Eastwood and colleagues (2012) defined boredom as the unpleasant state of wanting, but being unable, to engage attention. Westgate and Wilson's MAC model (2018) split that into two failure modes: attention that will not settle, and an activity that carries no meaning.

Tam and Inzlicht (2024) then found the result that matters most here. Across seven experiments with more than 1,200 participants, switching and skipping between digital videos made people more bored, not less, and watching one longer video uninterrupted was rated less boring and more satisfying. Participants expected the opposite. That was tested on digital video specifically, not on boring tasks in general, so read it as a finding about feeds rather than about work.

Why am I so bored? sets this out, including Wilson and colleagues' 2014 finding on how uncomfortable many people find unstructured time with their own thoughts.

What actually helps, and how much

The honest summary is that the mechanism is better understood than the remedy. We know from Mark (2008) and Leroy (2009) why removing an interruption before it arrives should beat resisting it once it has. What we do not have is a strong independent evidence base for the specific commercial tools sold on that logic: peer-reviewed efficacy trials of named focus apps are limited, and much of what circulates is vendor-reported.

The platform-level controls are at least documented by their makers. Apple publishes what Screen Time does and Google publishes what Digital Wellbeing does, which is a lower bar than efficacy evidence but a higher one than marketing copy.

Focus apps that actually block distraction goes through the category and is explicit about where the evidence stops.

What the evidence does not settle

• The eight-second figure. A documented retrieval did not locate an originating study. Treat it as unsourced rather than as disputed.
• How far switching intervals have actually fallen. Later datasets report shorter intervals, but the size and direct comparability of the change remain uncertain. The most recent peer-reviewed number is a 40-second median, and the more widely quoted 47 seconds comes from a trade book rather than a paper.
• Whether heavy media multitasking damages attention. Ophir, Nass and Wagner's 2009 finding that heavy media multitaskers filtered distractions less well became the field's anchor result. Wiradhany and Nieuwenstein's 2017 work ran two replications and a meta-analysis: most of the replicated effects did not hold, and the meta-analytic association, while weakly significant at first, became non-significant after correction for small-study bias. The authors say this leads them to question whether the association exists in laboratory tasks at all. So the evidence here is smaller and more statistically fragile than the original result suggested, which is not the same as saying the effect is absent.
• Whether switching behaviour reflects attention capacity. Mark's screen-switching data measures what people do. Nothing in it establishes what people are still able to do.
• Whether any specific focus app works. The mechanism is supported. Independent efficacy evidence for individual products is limited.
• Where interruptions come from. A little over half are self-generated, which means an intervention aimed only at external notifications is aimed at half the problem.

Where to start, depending on what is actually wrong

Most people arrive here with one of four problems, and they need different things.

If you have absorbed the idea that your attention span has collapsed and you want to know what is really established, read How to fix your attention span. It is the piece that separates the measured findings from the myth.

If the problem is work that keeps getting broken up, and the cost shows in your stress rather than your output, read The work time you lose to task switching. Start with the self-interruption half, because that is the half you control.

If you are not being interrupted so much as fleeing something dull, read Why am I so bored? The counter-intuitive finding there is that switching away makes it worse.

And if you have decided the answer is software, read Focus apps that actually block distraction before you pay for anything. The mechanism is sound; the evidence for individual products is thinner than the marketing suggests.

If none of those is quite it, the single highest-value change most people can make needs no article and no app: decide in advance what the next task is, and remove the trigger rather than resisting it. That is the one move every study on this page points at.