How focus works in the brain
Play with a conceptual model of attention: three networks, a working-memory meter that overflows at about four items, and an hour that fragments as you add interruptions. Every number is labelled as estimate or illustration.
What the diagram is, and is not
The animation above is a conceptual diagram. The three coloured nodes are not brain regions drawn to scale and the pulses are not neural signals; they are a way to make three well-supported ideas visible at once: attention has separable components, working memory is small, and interruptions cost more than the seconds they occupy. Every number in the panel is either a textbook estimate or an illustrative range, and the text under each panel says which.
What this shows / what it does not show
Shows: how three functions of attention interact conceptually, how a small working-memory budget fills up, and why recovery after an interruption is better described as a range than as a single number.
Does not show: your own attention, any measured brain activity, or a prediction of how long you will need to recover. The 15-second and 10-minute recovery bounds are deliberately wide bookends from different kinds of study, not findings from one experiment.
Three networks, not one spotlight
Posner and Petersen proposed in 1990 that attention is carried out by at least three anatomically separable networks: an alerting network that keeps you ready to respond, an orienting network that selects where attention goes, and an executive network that resolves conflict and keeps a goal active against competing input A[1]. Twenty years of imaging later, the same authors reviewed the evidence and kept the three-way division, adding detail on the neuromodulators involved and on a second executive network A[2].
For a working day the useful translation is this: feeling alert is not the same thing as being able to hold a goal. You can be wide awake (alerting) and still have your orienting network pulled towards every new notification. The cognitive control toggle in the diagram contrasts the two outcomes that executive control makes possible: the goal stays active and the pull fades, or the goal weakens and attention is captured. Both outcomes are normal; which one wins depends on how salient the distractor is, how practised the task is, and how much of your capacity is already committed.
About four things at once
Cowan's review of the working-memory literature puts the typical adult capacity at about four chunks, with a range of roughly three to five, when rehearsal and grouping strategies are prevented A[3]. The slots meter uses that figure literally so that you can watch it overflow. Two caveats matter. First, a "chunk" is whatever your experience lets you treat as one unit, so experts hold more content per slot, not more slots. Second, the slot-versus-resource debate is still open; the meter is a teaching device, not an endorsement of one model over the other [3].
What the meter is good for is making an everyday failure visible: when a fifth concern arrives (a message you must reply to later, a number you are holding for the next step), something either has to be written down or is likely to be dropped.
Interruptions: the honest range
The slider fragments one hour. Each interruption is drawn as a one-minute gold block, followed by a recovery band drawn twice: a short band at lab scale and a long band at the scale people report in the field. The two differ by two orders of magnitude, and that is the point.
In laboratory tasks, resumption lags after an interruption are measured in seconds and grow with the length and mental demand of the interruption B[6]. Even interruptions of under three seconds roughly doubled sequence errors in a procedural task, and interruptions of about four and a half seconds roughly tripled them B[5]. The well-known study by Mark and colleagues found that interrupted participants actually finished their task faster, but reported more stress, frustration, time pressure and effort B[4]. The much-quoted figure of "23 minutes to refocus" is not in that paper; it comes from an interview in which Mark described earlier field observations. We therefore show the long end of the band as a field-report scale and label it illustrative.
The two readings in the summary text ("intact focus between X and Y minutes") are the same hour computed under each bound. If the long-recovery estimate makes the recovery windows overlap, the panel says so, because at that point the model has no clean stretch left to show.
How to use this
- Protect the executive network's job: decide before a block what you will do with incoming requests (defer, note, ignore), so the decision is not made under capture.
- Externalise anything beyond three or four live items; a visible list is working-memory insurance, not a sign of weakness.
- Treat recovery time as personal and variable; log a few interruptions and your own return time instead of adopting any single published number. The focus planner can turn such logs into a schedule.
- For what happens when the interruptions are self-inflicted task changes, see the cost of context switching.
Frequently asked questions
Is the brain diagram anatomically accurate?
No. It is a schematic. The three networks described by Posner and Petersen involve distributed regions (for example parietal and frontal areas for orienting, anterior cingulate and lateral prefrontal cortex for executive control). The node positions in the diagram are chosen for legibility only.
Where does the '23 minutes to refocus' number come from?
From a media interview with researcher Gloria Mark describing earlier field observations, not from the 2008 CHI paper that is usually cited for it. That paper found interrupted participants worked faster but reported more stress. We show recovery as a range for that reason.
Is working memory really four slots?
About four chunks is Cowan's estimate for adults when rehearsal and grouping are prevented, with a typical range of three to five. Whether capacity is best described as discrete slots or a shared resource is still debated; the meter uses slots because they are easy to see.
Can I train my attention networks?
The sources on this page are anatomical and cognitive descriptions, not training studies. They do not show that attention can be strengthened like a muscle, and this page makes no such claim.
Sources and evidence
- The Attention System of the Human Brain — Posner MI, Petersen SE (1990), narrative review A doi.org/10.1146/annurev.ne.13.030190.000325 (retrieved 2026-11-09)
- The Attention System of the Human Brain: 20 Years After — Petersen SE, Posner MI (2012), narrative review A doi.org/10.1146/annurev-neuro-062111-150525 (retrieved 2026-11-09)
- The Magical Mystery Four: How Is Working Memory Capacity Limited, and Why? — Cowan N (2010), review A doi.org/10.1177/0963721409359277 (retrieved 2026-11-09)
- The cost of interrupted work: more speed and stress — Mark G, Gudith D, Klocke U (2008), lab experiment (CHI conference paper) B doi.org/10.1145/1357054.1357072 (retrieved 2026-11-09)
- Momentary interruptions can derail the train of thought — Altmann EM, Trafton JG, Hambrick DZ (2014), lab experiment B doi.org/10.1037/a0030986 (retrieved 2026-11-09)
- The effect of interruption duration and demand on resuming suspended goals — Monk CA, Trafton JG, Boehm-Davis DA (2008), lab experiment B doi.org/10.1037/a0014402 (retrieved 2026-11-09)