The word "attention" makes it sound like a quantity — some people have more, some less. That framing does not survive contact with the experience most people actually describe, which is not a shortage of attention but a problem with moving it.
Two versions of the same difficulty:
Starting. The gap between deciding to do something and beginning it, which can be minutes or weeks, and which does not shrink in proportion to how much the thing matters.
Stopping. Being unable to leave a task that has hold of you, even when it stopped being the right thing to do two hours ago. The name usually given to this is hyperfocus, which makes it sound like a superpower and understates the cost of coming out of it.
Both are transitions. It is the switch that is expensive, not the attending.
Why transitions cost so much
The honest position is that the mechanisms are debated and no single account is settled. But some observations hold up reliably enough to be useful:
The current state has inertia. Once engaged, staying engaged is easy; once disengaged, engaging is very hard. This explains why the same day can contain both "could not start for four hours" and "worked for six hours without eating".
Switching requires holding two things at once — where you are and where you are going — and that is precisely the kind of load that is hardest to sustain.
Interest, urgency, novelty and challenge make transitions cheap. A genuinely interesting task can be entered instantly. This is why "you can concentrate when you want to" is observably true and explains nothing: the variable is not want, it is what the task supplies.
The cost is not fully paid at the switch. After an interruption, attention does not snap back. There is a tail, and during it you are working at a fraction of capacity while feeling like you are working.
What this means for a day
A few consequences that follow directly:
Batching is worth more than it looks. Not because doing similar things together is tidier, but because you pay one transition instead of six.
An interrupted hour is not an hour. A meeting at 11am does not remove one hour from the morning; it removes the morning. Planning as though it costs an hour is how days disappear with nothing done.
Switching to something small is not free. Answering one email in the middle of deep work is a full transition, both ways.
Ending a task is a task. Landing a piece of work — saving it, noting where you were, leaving a way back in — takes effort at exactly the moment you have least. Doing it anyway is the single highest-return habit, because the next start is priced by how well you stopped.
What actually reduces the price
Nothing on this list is a cure and all of it is mundane. Nothing here replaces treatment where treatment is indicated.
Make the first move absurdly small. The barrier is at the boundary, not inside. "Write the report" is not enterable. "Open the document and type the heading" is. This is a well-worn piece of advice and it works for a specific reason: it moves the transition to the cheapest possible place.
Leave a deliberate handhold. Stop mid-sentence rather than at the end of a section. An unfinished thought is an entry point; a clean ending is a locked door.
Use an external boundary. A timer, a scheduled call, someone else in the room. The point is not discipline — it is putting the transition outside your head, where it costs less.
Protect one uninterrupted block rather than optimising the whole day. One block with no transitions in it will do more than six hours of fragments.
Decide the next task before you need it. The gap between finishing and starting is where hours vanish, and it is much worse when the question "what now" has to be answered from scratch.
Plan the exit from hyperfocus, not just the entry. An alarm that is across the room. A hard stop imposed by something external. The state does not self-terminate at a sensible point.
The cost that gets ignored
Transitions are tiring in a way that is disproportionate to the work done, and the resulting fatigue is routinely misread as laziness — by other people and, more damagingly, by the person experiencing it. A day spent switching constantly can produce almost nothing and leave you flattened, and the conclusion drawn is usually about character.
It is more useful, and more accurate, to treat transitions as a resource that runs down. If a day is going to be full of them, it is not also going to contain your hardest thinking, and planning as though it will is setting up an evening of self-recrimination that changes nothing.
What is worth reporting to a clinician
If treatment is being adjusted, the useful observation is not "focus was better". It is which part changed. Starting, sustaining, or stopping are three different things, and a change that makes starting easy while leaving you unable to stop is a specific finding with a specific response.
Writing down which one moved — and on which days — is more useful than any score.