Attention Switching Cost

A simple explanation

Every time you move your attention from one task to another, a piece of the first task does not move with you. It stays — humming under the surface of working memory, still half-loaded, still pulling resources. The next task begins not on a clean desk but on a desk with the previous task's papers still scattered across it. The cost is small for any single switch. It is rarely small by the end of the day.

Sophie Leroy called this attention residue, and the laboratory finding is dull and devastating: people who switch tasks under time pressure carry measurable performance decrements into the next task for minutes afterward. The Reward System, asked to feel productive, supplies the satisfying sensation of motion. The motion is real. The deposit is not.

An everyday example

You sit down to draft a proposal. Two minutes in, a Slack notification surfaces — you glance, decide it can wait, return to the document. Four minutes in, an email preview slides across the corner of the screen — you glance, decide it can wait, return. Six minutes in, a calendar reminder pulses — you dismiss it, return. By minute fifteen, the proposal is three sentences long and you are reading the email anyway because the residue from the half-look has been pulling at you the whole time.

You finish the email. You go back to the proposal. You re-read the three sentences. You try to remember the shape of the argument you had been building before the first glance. The shape is gone. You start again.

Why does it take me so long to get back into something after a small interruption?

Because the interruption did not interrupt one thing — it interrupted a stack. You were holding the document, the argument inside it, the next sentence, the next paragraph, and the felt sense of the reader you were writing for. The glance unloaded part of that stack. When you came back, you had to rebuild it — not from zero, but from a partial state where some pieces are still there and you cannot tell which.

The Reward System reads the glance as a tiny success — a problem checked, an inbox tended, a notification cleared. It does not see the cost because the cost lives in the next task. The trade looks like plus one small win, minus nothing, when the actual ledger reads plus one small win, minus eight minutes of context rebuild.

The behavioral loop

A loop that is easy to enter and hard to see:

1. Engagement — you settle into a task that requires loading working memory: writing, designing, debugging, deciding.
2. Cue — a notification, a tab, a thought about another task surfaces.
3. Reward verdict — the System classifies the cue as a tiny reward opportunity: quick win, then back.
4. Switch — you move attention. The previous task's stack partially unloads.
5. Tiny win — the cue is handled. A small satisfaction is logged.
6. Return — you come back to the original task with residue still attached to the previous one.
7. Rebuild — working memory reassembles the stack. This takes longer than the switch did.
8. Re-entry — another cue arrives. The loop runs faster because the path is now grooved.

Emotional drivers

Four feelings, often stacked:

• The faint anxiety of an un-checked notification, which the body reads as a small open loop.
• The dopaminergic pleasantness of a small task closed, which the System over-weights.
• A diffuse sense of being on top of things, which is the felt experience of motion mistaken for the felt experience of progress.
• An evening dissatisfaction — I was busy all day and nothing got finished — which arrives too late to inform the next morning's choices.

What your nervous system does

The Posner executive-attention network — anchored in the anterior cingulate and lateral prefrontal cortex — is what actually performs the switch. Each switch is a measurable metabolic event: glucose is spent, the network briefly reconfigures, the new task's representations get loaded while the old ones get partially flushed. The body does not log this as effort because no single switch is dramatic. The cost is in the integral.

Over a day of heavy switching, the executive network shows degraded performance — slower reaction times, more errors, a felt sense of cognitive heaviness that arrives by mid-afternoon and gets misattributed to caffeine, lunch, or sleep. The body knows it worked. It is just unsure on what.

The DojoWell interpretation

Attention switching cost is one of the clearest cases of the Reward System substituting a feeling for a system. The original system was focus — sustained attention on a task long enough for the work to consolidate. The substitute was the feeling of being productive while switching. They share a surface property: both involve effort, both involve completion of small units, both leave the body tired. They are opposite in deposit.

The density signature is effort_without_deposit because the equation is uniquely lopsided. Effort is high and visible. Residue is high but distributed across micro-tasks, so no single residue is large enough to be named. Deposit is low because consolidation requires sustained attention, and the day never gave any one task enough of it.

This is also why the closure pattern is deferred. Each individual task gets closed — the email sent, the Slack answered, the notification cleared. But the meta-task — the proposal, the design, the decision — gets deferred again and again. By Friday, the week's most important work is the work that never happened, and the work that did happen does not feel load-bearing.

The Reward System is not wrong that small wins matter. It is wrong about which wins were available. A morning of three deep hours would have produced more deposit than a day of forty small switches. The trade is rational only if you measure in dopamine.

How do I stop the small switches from eating my morning?

You do not eliminate switching. You raise the cost of unintentional switching while keeping intentional switching cheap. The System will still issue the route; what is workable is whether the route is one click away.

Three moves, in order of difficulty:

1. Close the cues. Notifications off, tabs closed, phone face-down or in another room. Not forever. For the block. The cost of re-opening a tab is the friction that breaks the half-second reflex.
2. Name the block. Decide in advance what the next ninety minutes is for. The decision pre-loads the executive network and makes the residue from non-block tasks easier to flush.
3. Track residue, not time. At the end of a block, write one sentence about what was unfinished and where you stopped. The sentence becomes the loading dock for the next session — the residue gets externalised so working memory does not have to carry it.

Practical steps

1. Design your first ninety minutes. The morning has the cleanest executive network. Spend it on the task that benefits most from no residue. Email and Slack will still be there at 10:30.
2. Batch the small. Group switching-cheap tasks into a single block. The cost of one switch into "admin mode" is much smaller than the cost of forty switches into and out of it.
3. Install one friction per recurring switch trap. The browser homepage, the Slack autoload, the notification badge. Each one removed reclaims a half-second decision per occurrence.
4. End the block by externalising state. A two-line note — where I am, what's next — closes the working-memory loop and lowers the residue you carry into lunch.
5. Track the felt cost weekly. Not minutes. Energy. A day of fragmentation produces a recognisable evening signature. Naming it converts the cost from invisible to legible.

Reflection questions

• Which task did you most consistently fail to give a clean block to this week?
• Which notification or tab is the single biggest source of residue in your day?
• Where did you mistake the feeling of motion for the feeling of progress?
• What would it cost — and what would it free — to make the first ninety minutes of your day uninterruptible?

Frequently Asked Questions