Focus vs Multitasking: What the Research Supports

Multitasking feels like doing two things at once. In practice it is doing one thing, stopping, doing the other, and paying to get back into the first. That matters because the popular claim that multitasking costs 40 percent of productivity has no solid source, and using a made-up number weakens a case that the real research already supports strongly enough.

Attention switches rather than splits

For attention-demanding tasks, the mind processes one stream at a time and switches between streams. The exception is pairings where one task is automatic and does not compete for the same resources: walking while talking, folding laundry during a call. Two tasks that both need language, planning or working memory do not pair, however much the calendar suggests otherwise.

That is why talking on the phone while writing a document goes badly, and why listening to a familiar instrumental album usually does not. The test is resource competition, not how the tasks look from the outside.

The switch-cost evidence

Rubinstein, Meyer and Evans reported in 2001 that switching between tasks carries a measurable time cost, and that the cost is largest when the rules of the tasks are unfamiliar or unmapped, because you must reconfigure what you are doing before you can continue. Later work by Ophir, Nass and Wagner in 2009 tested heavy media multitaskers and found they performed worse on tasks requiring them to filter out irrelevant information and to switch cleanly. Practice at heavy switching did not produce better switching.

Add Leroy's attention residue and the picture is consistent across three labs with different methods: switching between demanding tasks is slower than the work itself, and the time you think you saved is spent remembering where you were.

Why the 40 percent figure should be dropped

The line that context switching costs 40 percent of productive time is repeated everywhere and cannot be traced to a peer-reviewed measurement. It appears to come from a 1998 magazine summary rather than an experimental result, and citing it invites the argument to collapse on a technicality.

The defensible version is narrower: switching costs time and accuracy, and the size depends on the tasks, the person and how buried they were in the interrupted work. Gloria Mark's observational studies found rapid task switching across the working day and a return to the original task averaging around 23 minutes after a real interruption. Use that instead, and describe it as an average from workplace observation rather than a universal law.

How to single-task on purpose

Focus is a plan before it is a feeling. Before starting a block, write the one task at the top of the page, in a sentence with a verb: 'draft the section on failure costs'. A named task is harder to abandon than a feeling of getting on with things, and it is what you check against when you notice your attention drifting.

Then remove the switch triggers rather than resisting them one at a time. Phone in another room, notifications off, mail client closed, one browser window with the tabs the task needs. If a stray thought arrives, write it in one line and return. A 25-minute work and 5-minute break cycle is a reasonable way to train attention, and longer 90-minute blocks suit work that needs a long warm-up. Choose the length that matches the task, not the app you like.

• One named task - A written verb phrase beats a vague intention. It also gives you something to check when you drift.
• No second screen - A second screen is a second task waiting for a reason. Use it only if both screens serve the same task.
• Capture, do not act - One line in a notes file per stray thought. Acting on it mid-block is a switch you chose and still paid for.
• Finish the block cleanly - End on a finished sentence and write the next step. Coming back is then one minute of reading, not ten.

When multitasking is the right choice

Some roles require overlapping demands, and pretending otherwise causes more harm than good. A parent working from home, a support agent on a queue and a nurse on a ward all switch by design. In those settings the useful discipline is to pair only tasks that can share attention safely, and to plan recovery rather than pretending the interruptions do not cost anything.

Even in knowledge work, some multitasking is efficient. Two low-stakes tasks with different resource demands, such as a routine import running while you write meeting notes, cost little. The line is whether either task needs your full attention. If it does, it gets the block to itself.