Why Notifications Break Your Focus Worse Than Interruptions Do

A colleague walking into your office and asking a question is an interruption. It costs you time and breaks your train of thought. Research on interruptions consistently finds that it takes an average of 23 minutes to return to a task after a significant interruption. That number gets cited constantly in productivity discussions, often as a case against open offices and meetings.

But that 23-minute figure is specifically about interruptions that require a real response — someone asking you something that demands attention. Notifications are different. A notification that you notice but don’t act on turns out to do something more insidious than a direct interruption: it keeps your attention partially occupied even after you’ve dismissed it. The cognitive costs of the notification you glance at and put down are higher than the cognitive costs of the interruption you actually handle and finish.

Understanding why this is — and what the actual mechanisms are — is more useful than the usual advice to “turn off notifications.” Not all notifications are equal, the research is more nuanced than most productivity articles suggest, and the practical fixes depend on understanding the specific problem rather than just eliminating sources of interruption.

The Research on Partial Attention Costs

Gloria Mark’s lab at the University of California, Irvine has produced much of the foundational research on interruption and attention in workplace contexts. Her work distinguishes between external interruptions (something happens that demands your attention) and internal self-interruptions (you choose to check something). Both types carry recovery costs, but they’re different in character.

External interruptions that are handled to completion — you stop, deal with the interruption, and return — have a known cost but a recoverable one. The brain processes the interruption as a completed task and can re-engage with the prior task after a lag. External interruptions that are only partially handled — you glance at a notification, register that it exists, decide not to respond yet — create an open task in working memory. The Zeigarnik effect: unfinished tasks continue to occupy cognitive resources in a way that finished tasks don’t.

A 2015 study by Stothart, Mitchum, and Noushine found that receiving phone notifications — even notifications that were not opened or responded to — produced attention lapses comparable to actively using the phone. The study measured performance on a sustained attention task and found that participants who received notifications while working made significantly more errors than control participants, even though the notification group never looked at or responded to their phones. The mere awareness that a notification was waiting was sufficient to degrade performance.

This is the mechanism that makes notifications worse than interruptions in certain conditions. An interruption that gets handled is a completed task. A notification that you notice and defer is an open loop. If you’re interrupted three times in an hour and handle each interruption, you have three completed tasks and three recovery periods. If you receive fifteen notifications that you see and defer over the same hour, you have fifteen open loops running simultaneously in your working memory — even if you didn’t respond to any of them.

Why Notification Design Exploits This

App developers and platform designers know the research on notifications, even if the people implementing them don’t always think about it in these terms. Notification design is deliberately uncertainty-inducing: the banner that appears briefly and then disappears creates a “what was that?” response even when you’ve nominally dismissed it. The red badge on an app icon is a persistent open loop — it keeps generating a small pull toward resolution every time you see the app. The notification preview that shows enough text to create curiosity but not enough to resolve it is designed to maximize the probability of opening the app.

Social media notifications are the most deliberately engineered version of this. A notification that says “someone liked your photo” creates uncertainty about who, what, and how many. The only way to close the loop is to open the app. A notification that says “You have 3 new followers” creates uncertainty about who those followers are. Slack’s red badge on a channel doesn’t tell you whether it’s 1 message or 20. The design is not accidentally incomplete — incomplete information is more motivating than complete information for eliciting engagement behavior.

The notification permission system on both iOS and Android was partly intended to give users control over this, and in principle it does. In practice, most users grant notification permissions to most apps during onboarding (when the app is asking right after delivering some value) and never audit them again. The result is that the typical smartphone has dozens of apps with notification permissions, most of which deliver low-value interruptions regularly.

The Distinction That Actually Matters: Actionable vs Informational

A more useful frame than “notifications good or bad” is distinguishing between notifications that have a clear, appropriate action you should take now versus notifications that are informational updates you don’t need to act on immediately.

Actionable, time-sensitive notifications are genuinely valuable: a message from someone you need to respond to, a calendar reminder for something starting in 15 minutes, a security alert requiring immediate attention. These are legitimate interruptions with clear completions. Handling them closes the loop. Receiving them at the right time reduces the cognitive load of tracking those things manually.

Informational notifications — someone liked your post, an app has updated, your package is on the way, a news headline, a weekly digest — are not time-sensitive and carry no meaningful required action in the moment. These are the notifications most worth turning off, because they create open loops without offering the completion that handles them. They’re also the majority of what most people receive.

The practical framework this suggests: audit your notification settings not by app but by notification type and ask whether each type is actionable and time-sensitive. Most apps offer granular control — Instagram lets you distinguish between direct messages (potentially actionable) and likes and follows (purely informational). Email clients distinguish between a specific sender’s messages and bulk/promotional mail. Using that granularity rather than all-or-nothing notification toggles is both more practical and more effective.

Platform-Level Tools That Help and Don’t Help

iOS Focus modes and Android’s Do Not Disturb have matured significantly. iOS’s Focus feature allows creating custom profiles — Work, Personal, Sleep — with specific apps and contacts allowed to break through, and these profiles can be scheduled and location-triggered. The mechanism is sound: you get urgent, important alerts from people who matter and nothing else during focus periods. The failure mode is setup friction; most users never configure Focus beyond the basic settings, and the default “all or nothing” is less useful than the customized version.

The notification summary features on both iOS (Scheduled Summary) and recent Android versions batch non-time-sensitive notifications for delivery at set times — once in the morning, once in the evening — rather than immediately. For truly informational notifications, this is the right model: receive the information, but at a time you’ve chosen rather than whenever the app decides. In practice, the categorization of “what belongs in summary” isn’t perfect and users sometimes miss things they should have seen sooner, which erodes trust in the feature.

The Behavioral Change That Actually Works

The research on notification management and focus consistently points to one behavioral change as more effective than any app setting: physically separating the phone from the work environment during focused work sessions.

A 2017 study (Ward, Duke, Gneezy, and Bos) found that participants’ cognitive capacity was measurably higher when their phone was in a different room compared to when it was face-down on the desk. The mere presence of the phone — even when it was off, even when participants reported not thinking about it — reduced available cognitive capacity. The mechanism is the same as notification effects: the potential for notification creates persistent low-level attention allocation to monitoring for it.

This is a more radical suggestion than adjusting notification settings, and for many people it’s practically difficult — phones serve legitimate work functions. But for work sessions that genuinely require sustained focus, “phone in the other room” or “phone in a bag out of sight” is the highest-impact single change the research supports. Notification settings are the second-best option: better than nothing, effective for the specific notifications you configure, but unable to address the ambient awareness effect of having the device present at all.

The notification flood problem is ultimately a design problem dressed up as a user behavior problem. Platform designs that make notification permission easy to grant and hard to revoke, app designs that use incomplete information to maximize engagement, and notification UIs designed to create urgency rather than inform — these are the systems producing the focus degradation that individuals are then told to manage through personal discipline. Both the system-level and the individual-level responses matter. But being clear about which problem each solution addresses is at least a more honest starting point than treating notification overload as purely a user habits issue.