Last Tuesday I finished a call at 11:00, sat down to write, and produced one usable paragraph by 11:40. I was not procrastinating. I was not on my phone. I was sitting in front of the document the whole time, typing and deleting, while some background process in my head kept rerunning a thing someone had said on the call.
It was not an important thing. It was a mildly unresolved thing, which turns out to be worse.
Attention residue is the part of your attention that stays with a previous task after you have moved on to the next one, and it degrades your performance on the new task. The concept comes from Sophie Leroy, an organizational behavior researcher now at the University of Washington Bothell, whose work on it was published in 2009. Her central finding is that switching tasks does not move your attention cleanly, and the residue is worse when the first task was left unfinished or unresolved. You are not imagining the fog after a meeting or a quick Slack check. Part of your cognitive capacity is still allocated somewhere else, and that allocation is largely outside your control.
The practical consequence is that the cost of an interruption is mostly not the time the interruption took. It is the residue it leaves behind.
Leroy's work connects to a much older observation called the Zeigarnik effect, named after the Soviet psychologist Bluma Zeigarnik, who studied it in the 1920s. The gist is that people remember interrupted or incomplete tasks more readily than completed ones. Something about an unresolved goal keeps it active in memory in a way a closed one is not.
You know this already. A conversation that ended badly loops for hours; one that ended cleanly disappears. An email you answered is gone from your head, while an email you read and did not answer keeps resurfacing all afternoon.
This is the most useful thing to understand about residue, because it tells you what to do. If unresolved things stick, the intervention is not to switch faster or concentrate harder. It is to give the outgoing task a sense of closure before you leave it.
Here is the arithmetic most people never do.
You check email for two minutes and read four messages. Three are fine. One is a request you cannot answer immediately, from someone whose opinion you care about, with an unclear deadline. You close the tab. Two minutes, back to work.
Except an unresolved goal has now been added to the background. It does not announce itself. It shows up as slightly slower reading, slightly worse writing, a vague pull toward checking again. You paid two minutes at the till and signed up for an unknown amount afterwards.
"It's only two minutes" is true and irrelevant. The right question is not how long the interruption lasts, it is how much unresolved material it introduces. By that measure, a twenty minute call ending in a decision can be cheaper than a two minute glance at a thread ending in a question.
| The switch | Clock cost | Residue cost | |---|---|---| | Quick glance at an inbox with unread requests | Tiny | High, because you opened loops you did not close | | A meeting that ended with a clear decision | Large | Moderate, since the task closed | | A meeting that ended with "let's circle back" | Large | High, nothing resolved | | Finishing a task properly before switching | Moderate | Low | | Switching mid-sentence because a notification arrived | Tiny | Highest of all |
That last row deserves attention. Being pulled away mid-action is the worst case, because you leave the task maximally unfinished at the exact level where your working memory was holding the most.
One switch is survivable. The problem is that a normal working day is composed almost entirely of switches, and residue does not fully clear between them.
The day goes: standup, twenty minutes of work, a question in Slack, back to work, a calendar notification, a "quick" review, lunch, a meeting, then forty minutes of work fragmented by three more pings. At no point were you idle. At 6pm you have the specific exhausted feeling of having worked hard with nothing to show for it, and if someone asks what you did you genuinely struggle to answer.
That is what accumulated residue feels like from the inside. You were always partly on the previous thing, so nothing got your whole capacity, so nothing finished, so the unfinished pile grew. Feeling busy all day is not evidence that you did a lot. Often it is evidence that you switched a lot.
This changes what you optimise for. The goal is not to squeeze more tasks into the day. It is to reduce the number of transitions, which is a different and usually easier problem.
The most useful practical finding in this area is Leroy's own follow-up work on what she calls a ready-to-resume plan. The idea is small enough to sound trivial and it works well enough that I now do it by reflex.
Before you switch away from a task, spend a minute or two writing down where you are and what comes next.
That is it. Not a full handover document. Two or three lines: what state the work is in, what the immediate next action is, and any thought you are afraid of losing. The mechanism is the same one behind the Zeigarnik effect, running in reverse. The unresolved goal keeps pulling at you because your mind is holding it open. Writing down the resumption point gives it somewhere else to live, and attention releases.
What I actually write, in the file I am working in, looks like this:
NEXT: the transition paragraph between sections 2 and 3 is wrong.
Section 3 should open on the cost argument, not the definition.
Do not rewrite section 2, it is fine.
Thirty seconds. The difference when I come back is not subtle: I resume at the right place instead of spending the first ten minutes rereading to reconstruct what I was thinking.
Three places to use it:
Meeting to work is the sharpest residue transition in most jobs, for two reasons. Meetings are socially loaded, so they generate exactly the unresolved material that lingers. And they usually end without a written conclusion, so every attendee leaves holding open loops.
Two things help more than anything else I have tried:
Close the meeting on paper before you leave the call. Ninety seconds on what was decided, what you owe, and by when. This is not for the team. It is for you, so the loops are closed outside your head.
Do not start focused work the instant a meeting ends. Put ten minutes between them and walk, rather than checking messages. Filling that gap with your inbox just swaps one set of residue for a fresh one.
The same mechanism explains why work follows you home. If you stop by closing the laptop mid-task, every unresolved item stays active, which is why you think about work in the shower and lie awake rehearsing tomorrow's conversation.
The shutdown routine that works is short and specific. Ten minutes before you finish: write tomorrow's first task in enough detail that starting needs no decisions, empty the loose items out of your head onto the list, then deliberately mark the day done. Cal Newport recommends an explicit verbal cue, which sounds faintly ridiculous until you try it for a fortnight. The point is not the phrase. It is that closure needs a signal, and unfinished work does not generate one by itself.
This is worth doing not because evening thinking is unpleasant, though it is, but because residue carried overnight costs you the following morning, usually the most valuable part of your day.
Residue is a bigger tax on cognitive work than on physical or routine work for a straightforward reason: knowledge work runs on working memory, and working memory is the exact resource residue consumes.
If you are laying bricks, an interruption costs the time it takes to pick the trowel back up. If you are holding a half-built argument, a system's architecture, or a chain of reasoning about a bug, an interruption costs the structure itself, and rebuilding it takes far longer than the interruption did. The more of the task lives in your head rather than on the page, the more expensive every switch becomes.
This suggests a useful defensive habit: externalise more of the task while you are doing it. Notes, an outline, a scratch file of half-formed thoughts. Work that exists outside your head is cheaper to drop and cheaper to resume.
Since the largest source of avoidable switches is a machine you control, the environment is worth configuring. I use Deep Focus for this, and the parts that matter here are the unglamorous ones. A profile blocking the apps I "just check" removes the decision entirely, because the cheapest switch to prevent is the one you start yourself. The Session Planner does more than it looks like it does: putting explicit break blocks between focus blocks turns transitions into designated space instead of leaving them to be filled by whatever pings first. And the distraction tracking in Analytics is quietly the most informative part, because seeing which apps actually tried to pull you out tells you where your residue is being generated, which is usually not where you assumed.
Residue cannot be eliminated, only reduced. Any advice promising otherwise is selling something.
Some switching is genuinely unavoidable. In support, operations, management or anything on-call, interruption is not a failure of your system, it is the job. The realistic target is not zero switches. It is protecting one or two blocks a day, taking the resume note seriously, and refusing to feel guilty about the fragmented remainder.
Residue is also not an argument for never talking to anyone. Plenty of work only happens in conversation. What the research supports is being deliberate about transitions, not minimising human contact. Someone who batches all messages into two windows and becomes unreachable has traded one problem for a worse one.
It is the portion of your attention that remains on a previous task after you have switched to a new one, which reduces your performance on the new task. Sophie Leroy named and studied the effect, publishing the core research in 2009. It is strongest when the earlier task was left unfinished or unresolved.
There is no fixed duration, and I would be suspicious of anyone quoting one. It depends on how unresolved the previous task was and how emotionally loaded it felt. An answered email clears quickly. A tense conversation with no conclusion can sit in the background for hours.
Usually because you stopped mid-task rather than at a closing point, leaving goals your mind still treats as active. A short written shutdown routine helps: record tomorrow's first action in concrete terms, dump the loose items out of your head, and give the day an explicit end. You are not trying to suppress the thoughts, you are giving them somewhere to be stored.
Yes, and rapid multitasking is close to the worst case, because each switch leaves a task unfinished at the moment your working memory was holding the most. What feels like doing two things at once is really doing two things badly in alternation while paying a residue cost at every crossing.
Reduce the number of transitions rather than trying to recover faster from each one. Batch similar work, keep messaging apps closed during focused blocks instead of merely muted, insert a short gap between meetings and deep work, and write a two line resume note before every unavoidable switch.
The part that changed my behaviour was not the finding itself. It was realising that "it's only two minutes" had been a lie I told myself several times an hour for years, and that the actual price was never on the tag.
Get the latest productivity tips and Deep Focus updates delivered to your inbox