Attention residue
The evidence is real and the argument about it is still running: how strong it is, how far it travels, or whether it repeats. Trust the direction, and hold the numbers loosely.
Sophie Leroy, an organisational researcher now at the University of Washington, started from an observation about how knowledge work is actually arranged. Almost nobody does one thing. People go from meeting to meeting, start work on one project and soon after transition to another, and the working assumption underneath all of it is that attention moves when the body does. Her 2009 paper asked whether that assumption is true, and the title asked the practical version of the question: why is it so hard to do my work?
The experiments had a simple shape. Participants worked on one task, word puzzles, and then switched to a second, unrelated one, reviewing CVs and making hypothetical hiring decisions. Between the two, she measured what was still active in their minds with a lexical decision test, strings of letters flashed on a screen, judged quickly as words or not, because the words a mind responds to fastest reveal what it is still processing. People who had been interrupted mid-task were still processing the first task while working on the second. That persistence is attention residue, her term, defined as cognitive activity about a prior task continuing while you perform the current one. And the residue was not free: performance on the second task suffered while part of the mind stayed on the first.
The second experiment produced the detail that separates this research from the folk advice it resembles. The obvious remedy is to finish one thing before starting the next, and it turned out to be insufficient. Participants who finished the first task calmly, with ample time, still carried residue into the next one. The condition that actually released attention was finishing under time pressure. A deadline while closing something forces a kind of cognitive settling of accounts, the open questions get resolved or written off, and the mind lets go. Completion alone does not close a task. Closure does, and time pressure is what manufactures it.
Leroy then tested a practical version of that closure. In work with Theresa Glomb published in 2018, people who were interrupted mid-task made a ready-to-resume plan before switching: a minute or two noting where they had stopped and how they would pick the work up again. That small act reduced the residue and restored performance on the interrupting task. The mind does not need the task finished. It needs to know the task is safely parked.
The honest limits are the usual ones for this kind of work. These are laboratory experiments with short tasks and student participants, the follow-up comes from the same research programme, and the concept reached general audiences through Cal Newport's book Deep Work, which built more on top of it than the studies alone carry. What the research itself shows is narrower and solid: attention does not switch cleanly, unfinished work holds a claim on it, and the claim has a measurable cost on whatever you do next.
The version of this that costs the most money is not the interrupted hour. It is the standing arrangement: the old business kept running while the new one gets built, the side project beside the main role, the person staffed across three initiatives at a third of their time each. The arithmetic of those arrangements assumes attention divides like hours divide. This research says it does not. Each switch carries residue, so the two halves of a split day do not add up to a whole one, and the tax is invisible because it shows up in quality rather than in time. Both things get done. Neither gets a full mind.
There is a sharper version for anyone running two ventures at once. Every unresolved question in the first venture is a claim on attention during the second, and the claims do not announce themselves. They surface as the meeting where you were present and absorbed nothing, the decision made with half the care it needed, the draft that reads fine and is not. The cost lands on whichever thing you switched to, which is usually the newer, harder, more important one.
What the research licenses is not the advice to do one thing, which for most operators is fantasy. It is a discipline about transitions. Closing a work session under a small self-imposed deadline releases more attention than drifting out of it. And when something must be left unfinished, the ready-to-resume note, two minutes on where you stopped and what happens next, is the cheapest cognitive tool this library contains: it converts an open loop into a parked one, and parked loops release their claim. The organisational reading is the same one: a person on three projects is not three thirds of a person, and the missing fraction went to the switching. For what running two things at once did to Sony, see the silo effect. For why the older commitment keeps its grip, see the sunk cost fallacy.
Brooks says adding people to a late project makes it later, because the new person's first weeks are paid for by the people with no time and the coordination never leaves. Attention residue says splitting one person across several things does not give you fractions of a person, because each switch leaves part of their mind behind. Same arithmetic error from two directions: hours divide, and neither coordination nor attention does. Brooks is the cost of adding heads. Residue is the cost of splitting one. A plan that staffs a rescue with three people at a third each has made both mistakes in one line, and the tax shows up in quality rather than time, which is why nobody's budget will ever show it.
Source: Leroy, Why is it so hard to do my work? The challenge of attention residue when switching between work tasks, Organizational Behavior and Human Decision Processes, volume 109, 2009, pages 168 to 181. The ready-to-resume plan: Leroy and Glomb, Tasks Interrupted, Organization Science, volume 29, 2018, pages 380 to 397.
Cal Newport, 2016
The book that carried this research to a general audience, and the fullest argument for designing work around undivided attention. Newport builds more on the finding than the finding alone proves, which he is open about, and the building is worth reading.
Draw your own card. It does not take long, and it rewards taking your time.