Do notifications and interruptions actually destroy reading comprehension?
They damage it in a specific way, measured on two clocks. A notification you never touch still raises errors on an attention task, and a phone that buzzes lowers how well you understand a text. The famous 23-minute figure times how long a task waited, not how long your brain needed.
The damage is narrow, and it lands exactly where reading happens. An interruption usually leaves recognition intact: you still know the words you passed. What collapses is the build across sentences, where you assemble an argument and hold its examples and exceptions at once.
The alert alone is enough. In 2015, 166 students at Florida State did a deliberately dull attention task in two blocks (Stothart, Mitchum and Yehnert). Midway, some were sent a call notification and some a text, and none touched a phone. The notified group's error rate climbed more than three times as fast as the control group's, and calls looked worse than texts, though not significantly. An alert pulls up task-irrelevant thoughts that outlast the sound.
The reading version is newer and smaller. In a November 2026 study, 36 people read with no phone visible, with a silent phone on the desk, and with a phone buzzing with WhatsApp alerts (Guimarães and colleagues). Only the buzzing phone lowered comprehension, and it raised self-reported anxiety and frontal beta activity, read as effort spent resisting the urge to check. Caution: 36 people, one messenger, one lab.
How long does it take your brain to recover focus after a notification?
Seconds, not minutes. Across three experiments, resumption lag grew steadily as interruptions ran from 3 to 13 seconds, then levelled off, and a demanding filler task cost more than a dull one of the same length. The measured lag was around 1.5 seconds against a 0.9-second baseline: roughly 600 milliseconds of reorientation.
That demand term is the useful part. An interruption that asks you to think about something else costs more to leave than one that just takes time. A letter-matching filler left resumption at 1,789 milliseconds, a tracking task at 1,605, and a blank screen at 1,322 (Monk, Trafton and Boehm-Davis). That is why a quick question is rarely quick. These are computerised tasks, not reading, so read them for the shape of the cost.
One small mercy: the cost is front-loaded. An eight-second warning before a planned interruption let people prepare and resume faster (Trafton and colleagues, 2003), and even a quarter-second break delays the next action (Monk and colleagues, 2004), so an interruption you see coming is cheaper than one you do not.
Now the famous number, misdescribed for years. "It takes 23 minutes to refocus" traces to a 2006 Gallup interview with Gloria Mark about fieldwork shadowing 36 office workers, where interrupted work picked up the same day came back after an average of 23 minutes 15 seconds. The published study reports a smaller, longer-observed sample: 24 information workers, over 700 hours, 77.2% resuming the same day after 25 minutes 26 seconds on average.
That spread is the tell. The measure is an interval, abandoning a piece of work partway and getting back to it, and the distribution is skewed enough that no median is published. It is not a mental recovery time; the 2008 paper most often credited with it never mentions the number. The same fieldwork coded 41% of interruptions as internal, which changes what you can control.
Does the phone just sitting there matter?
Less than the internet says, and the alert matters more. In the one reading study that tested both, a silent phone in view performed like no phone at all, and only the buzzing phone lowered comprehension. Distance is a cheap habit, but muting is the intervention.
The belief is everywhere: even face down or on silent, your brain still dedicates attention to it. Mere-presence effects do appear in other tasks, so this is a contradiction, not a settled correction, but the reading evidence points at the buzz.
Should you turn off every notification?
No, and the trial that tested this is unusually clean. Over two weeks, 237 people used four schedules, and batching three times a day beat normal, hourly and never. That group unlocked less, felt more in control, reported less inattention and better concentration, and felt less stressed. Hourly batching did nothing at all.
The never group gained few of the benefits and reported more anxiety and more fear of missing out (Fitz and colleagues, Computers in Human Behavior, 2019). Removing more is harder than it sounds. A month-long trial asked people to block mobile internet for two weeks: of 467 who agreed, 266 installed the blocking app and 119 hit ten compliant days out of fourteen (Castelo and colleagues).
What do you actually do when the interruption is your job?
Stop defending the block and rehearse the return instead. Put your stopping point at a boundary you choose, write one line about where you were and what comes next, and let alerts arrive on a schedule. Four in ten interruptions are self-inflicted anyway, and a note beats a willpower rule.
The one-line note earns its place from the goal literature. Masicampo and Baumeister found that unfinished goals pushed intrusive thoughts into an unrelated reading task, and that a specific plan for finishing removed the interference.
A queue, a message and a colleague are not notifiable, and some jobs are interrupt-shaped by design, so a silent phone will not save a support shift. Absorb's daily comprehension workout is on the App Store, ten to twelve minutes with separate reading, listening and analysis scores.