The cost of context switching: what the research measured

"Context switching" is one of the most quoted ideas in productivity writing and one of the most misquoted. The laboratory evidence that switching has a cost is strong. The numbers attached to it in blog posts, 23 minutes to refocus, 40 % of productive time lost, a ten-point IQ drop, are not from the studies they are pinned to. This page separates what was measured from what was embroidered, and gives you a way to estimate your own switching cost.

Switch costs in the lab

The core finding is more than thirty years old and has been replicated many times. When people alternate between two simple tasks, say classifying a digit as odd or even and then as high or low, responses on the trial right after a switch are slower and less accurate than on a repeat trial. In Rogers and Monsell's alternating-runs experiments the switch cost was a few hundred milliseconds, and although giving people time to prepare reduced it, a residual cost survived even with ample warning A^[1]. Monsell's review of the literature attributes the cost to two mechanisms: actively reconfiguring the mental "task set", and interference from the task set you just left, which does not switch off cleanly A^[2].

Rubinstein, Meyer and Evans added the moderators that matter for real work: switch costs grow when tasks are more complex and less familiar, and shrink with practice and with clear cues about what comes next A^[3]. That is the one lab result that transfers most directly to the office. Switching between two routine admin tasks is cheap; switching between a budget model and a legal draft is not.

These are experiments measured in seconds, on tasks learned in minutes. They establish that switching is never free. They do not say how much of your day it costs; that extrapolation is the one every popular figure makes without saying so.

Attention residue: why the last task follows you

Leroy's 2009 experiments added a second mechanism with a workplace flavour. When people had to leave a task unfinished and move to another, part of their attention stayed with the first one, and performance on the second task suffered. Finishing the first task under time pressure, so that it felt closed, reduced the residue B^[4]. The study is small and from a single lab, so we grade it B, but it formalises something most people recognise: an open loop costs attention even after you have stopped working on it.

The practical implication is unusual. The cheapest switch is not the fastest one; it is the one that closes a loop first, even with a one-line note of where you stopped and what comes next.

The 23-minute figure: what Mark and colleagues actually found

The claim that "it takes 23 minutes and 15 seconds to get back on task" is usually attributed to Gloria Mark's research at UC Irvine. Two papers are involved, and neither contains the number.

In a 2005 field study, Mark, Gonzalez and Harris shadowed information workers and found work fragmented into short episodes across many "working spheres", with over half of those spheres interrupted before completion C^[5]. In the 2008 lab study, Mark, Gudith and Klocke interrupted participants during office-style tasks: the interrupted groups finished their work faster, but reported significantly higher stress, frustration, time pressure and effort B^[6]. The 23-minute figure comes from interviews in which Mark summarised earlier field observations about how long people took to return to an interrupted project; a 2024 interview gives it as "almost half an hour" B^[7].

So what is the honest interruption cost? It is a range, and the context decides where you sit in it:

Context	What was measured	Recovery or cost	Grade
Lab, simple task switch	Reaction time on the trial after a switch	A few hundred milliseconds, partly reducible by preparation [1]	A
Lab, procedural task	Errors after a 2.8 s or 4.4 s interruption	Errors roughly doubled and tripled respectively B[8]	B
Lab, office-style tasks	Completion time, stress, effort	Faster completion, higher stress and workload [6]	B
Field, information workers	Time to return to an interrupted project	Tens of minutes, as reported by the researcher in interviews; not a published measurement [7]	B/C

If you quote a single number, you are choosing one row of this table and discarding the rest. The interactive switching simulation lets you set switch frequency and task complexity and see how the ranges add up over a day under different assumptions.

Media multitasking and the brain

The other popular claim is that heavy multitaskers damage their ability to filter distraction. Ophir, Nass and Wagner's 2009 study compared 19 heavy and 22 light media multitaskers and found the heavy group worse at filtering irrelevant stimuli and slower to switch C^[9]. Four years later Alzahabi and Becker found the opposite on switching: heavy media multitaskers had smaller switch costs C^[10]. Wiradhany and Nieuwenstein ran two direct replications of Ophir's paradigm, both of which failed, and their meta-analysis of 39 effect sizes found only a small pooled association between media multitasking and distractibility B^[11]. Uncapher and Wagner's review concludes that heavy media multitasking is associated with poorer sustained attention and working memory in the lab, but the evidence is correlational and the effects small to medium B^[12].

The accurate summary: heavy multitasking is weakly associated with poorer attention in some lab tasks; whether it causes the difference or attracts people who already have it is unknown; nothing in this literature measures IQ or brain volume.

Calculator: your weekly switching cost

A number for your own week is more useful than any published average. The method below uses the measured ranges and makes its assumptions visible. It takes about ten minutes with a two-day interruption log.

Step 1. Count switches. For two typical days, tally every time you leave one piece of work for another, whether a notification, a colleague, a meeting, or your own impulse to check something. Multiply the daily average by five.

Step 2. Classify them. Mark each switch as simple (routine tasks, you return within seconds and lose no thread) or complex (you have to rebuild where you were, re-read, or re-plan).

Step 3. Apply a range, not a point. The lab literature supports a per-switch cost of well under a second for simple, well-practised switches ^[1][3]. For complex switches the only evidence is Mark's field work, which supports a range of a few minutes to around 25 minutes to fully resume a demanding task ^[5][7]. Use 3 minutes as a low bound and 20 minutes as a high bound.

Step 4. Add the error term. For procedural work, interruptions of a few seconds doubled or tripled sequence errors ^[8]. Count the reworks you can trace to an interruption during the two days; that time goes on top.

Example week	Switches	Low estimate	High estimate
60 simple switches	60 × ~0.5 s	about 30 s	about 1 min
40 complex switches	40 × 3–20 min	2 h	13 h
Interruption-traced rework	logged	0.5 h	2 h
Total		about 2.5 h	about 16 h

The width of that range is the finding. It tells you two things: your simple switches are nearly free and not worth fighting, and your complex switches are where the week goes, by a factor that depends mostly on how hard you find it to get back in. Reducing complex switches from 40 to 20 saves between one and six hours even on these rough bounds, which is why batching is a robust recommendation even though the exact figure is not knowable.

Reducing switches without quitting Slack

The evidence supports interventions that attack the switch count or the resumption lag, and it is quiet on everything else.

• Batch communication into fixed windows. In a 124-person crossover trial, checking email three times a day lowered daily stress relative to unlimited checking; output was not measured B^[13]. In a 13-person field experiment, five days without email halved window switching, from roughly 37 to 18 switches per hour, and lowered heart-rate-based stress C^[14].
• Remove notification sources during protected blocks. Receiving a notification without checking it impaired sustained attention about as much as active phone use B^[15]. Silence applies to the device, not only the app.
• Close loops before you switch. Attention residue falls when the previous task feels finished ^[4]. A one-line "parking note" of where you were and your next action is the cheapest version.
• Treat interruptions by type. The management literature distinguishes intrusions, breaks, distractions and discrepancies, with different effects; a self-chosen break is not the same event as a colleague's question B^[16].
• Keep the small stuff small. Simple switches between routine tasks carry lab-scale costs. Spend your effort on protecting complex work.

What the evidence does not say

• It does not say switching costs you "40 % of your productive time". That figure is not in Rubinstein, Meyer and Evans or any APA publication; lab switch costs are fractions of a second on simple tasks ^[3].
• It does not say recovery takes 23 minutes. That is an interview figure; the measured range runs from seconds to tens of minutes depending on context ^[6][7].
• It does not say multitasking lowers IQ or shrinks the brain. The media-multitasking literature is small, correlational and partly failed replication ^[11].
• It does not say every interruption is harmful. Interruptions that match a natural break point, or that bring information you need, can have neutral or positive effects ^[16].

What to do next

Run the two-day log, compute your low and high bounds, then decide which complex switches you can convert into batches. The focus planner takes your meeting hours, energy pattern and main interruption sources and places protected blocks where your own answers say they will survive, with a batched admin window for the communication you cannot drop.

To go deeper: the broader science of attention, why cognitive load at work is the term the research actually uses, and a data-first guide to improving deep work.

Frequently asked questions

What is context switching in productivity terms?

Moving your attention from one task or project to another before the first is finished. The research literature calls it task switching and measures two costs: the time and errors on the first trial after a switch ^[2], and the attention residue that stays with an unfinished task ^[4].

Does it really take 23 minutes to refocus after an interruption?

The figure is not in any peer-reviewed paper. It comes from interviews in which Gloria Mark summarised field observations of how long workers took to return to an interrupted project ^[7]. Measured recovery depends on context: fractions of a second for simple lab switches ^[1], doubled or tripled errors after interruptions of a few seconds ^[8], and tens of minutes in field self-reports ^[5].

Does multitasking lower your IQ?

No study in this literature measured IQ change from multitasking. A 2009 study found heavy media multitaskers filtered distraction worse ^[9], a 2013 study found they switched tasks better ^[10], and two direct replications of the original failed, with a meta-analysis finding only a small association ^[11].

How much time does context switching cost per week?

There is no valid single number. Using the measured ranges, a week with 40 complex switches costs somewhere between about 2 and 13 hours in resumption time, plus any interruption-caused rework. The calculator section on this page shows how to compute your own low and high bound from a two-day log.

What is the fastest way to reduce switching costs?

Cut the number of complex switches, not the simple ones. Batch email and chat into fixed windows, which lowered stress in a two-week trial ^[13], silence devices rather than single apps during protected blocks ^[15], and leave a one-line note before you switch so the next resumption is shorter ^[4].

Sources and evidence

1. Costs of a predictable switch between simple cognitive tasks — Rogers, R. D.; Monsell, S. (1995), RCT A doi.org/10.1037/0096-3445.124.2.207 (retrieved 2026-11-09)
2. Task switching — Monsell, S. (2003), review A doi.org/10.1016/S1364-6613(03)00028-7 (retrieved 2026-11-09)
3. Executive control of cognitive processes in task switching — Rubinstein, J. S.; Meyer, D. E.; Evans, J. E. (2001), RCT A doi.org/10.1037/0096-1523.27.4.763 (retrieved 2026-11-09)
4. Why is it so hard to do my work? The challenge of attention residue when switching between work tasks — Leroy, S. (2009), RCT B doi.org/10.1016/j.obhdp.2009.04.002 (retrieved 2026-11-09)
5. No task left behind? Examining the nature of fragmented work — Mark, G.; Gonzalez, V. M.; Harris, J. (2005), conference paper C doi.org/10.1145/1054972.1055017 (retrieved 2026-11-09)
6. The cost of interrupted work: more speed and stress — Mark, G.; Gudith, D.; Klocke, U. (2008), conference paper B doi.org/10.1145/1357054.1357072 (retrieved 2026-11-09)
7. How to sharpen your attention and meet your goals (interview with Gloria Mark, UC Irvine) — University of California Newsroom (2024), report B www.universityofcalifornia.edu/news/how-sharpen-your-attention-and-mee (retrieved 2026-11-09)
8. Momentary interruptions can derail the train of thought — Altmann, E. M.; Trafton, J. G.; Hambrick, D. Z. (2014), RCT B doi.org/10.1037/a0030986 (retrieved 2026-11-09)
9. Cognitive control in media multitaskers — Ophir, E.; Nass, C.; Wagner, A. D. (2009), observational C doi.org/10.1073/pnas.0903620106 (retrieved 2026-11-09)
10. The association between media multitasking, task-switching, and dual-task performance — Alzahabi, R.; Becker, M. W. (2013), observational C doi.org/10.1037/a0031208 (retrieved 2026-11-09)
11. Cognitive control in media multitaskers: Two replication studies and a meta-analysis — Wiradhany, W.; Nieuwenstein, M. R. (2017), meta-analysis B doi.org/10.3758/s13414-017-1408-4 (retrieved 2026-11-09)
12. Minds and brains of media multitaskers: Current findings and future directions — Uncapher, M. R.; Wagner, A. D. (2018), review B doi.org/10.1073/pnas.1611612115 (retrieved 2026-11-09)
13. Checking email less frequently reduces stress — Kushlev, K.; Dunn, E. W. (2015), RCT B doi.org/10.1016/j.chb.2014.11.005 (retrieved 2026-11-09)
14. A pace not dictated by electrons: an empirical study of work without email — Mark, G.; Voida, S.; Cardello, A. (2012), conference paper C doi.org/10.1145/2207676.2207754 (retrieved 2026-11-09)
15. The attentional cost of receiving a cell phone notification — Stothart, C.; Mitchum, A.; Yehnert, C. (2015), RCT B doi.org/10.1037/xhp0000100 (retrieved 2026-11-09)
16. Pardon the Interruption: An Integrative Review and Future Research Agenda for Research on Work Interruptions — Puranik, H.; Koopman, J.; Vough, H. C. (2020), systematic review B doi.org/10.1177/0149206319887428 (retrieved 2026-11-09)