Hand-cranked focus

The observation

Sharp, sustained focus does not arrive for me on its own. I have to manufacture it, by triggering in myself something that feels like an adrenergic response.

The triggers are small and mostly physical. I open my eyes wider than necessary. I feel each word I think, write or read, one at a time, as if beaming it serially into my context so it does not dissolve into the surrounding stream. My legs move constantly. I pursue rich information streams and fast-paced short feedback loops. I attach a private sense of urgency to each task, a microurgency, so that every step feels like it has a deadline seconds away.

Sometimes I do the opposite. I close my eyes to cut the visual noise, until what remains is pure abstraction: concepts without the page.

When the triggers work, focus is real and often deep. When they stop, focus collapses. And sometimes, after a long session, I cannot come down at all.

This essay takes that self-observation seriously as data. The claim is simple: what I experience as “trying to focus” is better described as hand-cranking an arousal system that does not idle at the right level by itself. Focus is not free for this kind of brain. It is produced, and production has a cost.

The levers

The levers fall into two classes. Gain levers raise arousal through sensory gain, motor activation, event rate, reward density and temporal compression. Filter levers lower noise, so the signal that remains is easier to hold.

Lever	Class	Channel	What it does	Evidence
Widening the eyes	Gain	Sensorimotor, autonomic	Pupil size tracks locus coeruleus activity; widening likely feeds back into arousal and raises visual input gain	Pupil and LC coupling: strong, though not LC-specific.¹ Voluntary feedback: untested
Feeling each word, one at a time	Gain	Cognitive, embodied	Gates each word as a discrete serial event and refreshes it in working memory before the surrounding stream overwrites it; recruits the articulatory loop	Event-rate effect and articulatory rehearsal: strong. Felt word-by-word gating as a strategy: untested
Fidgeting, restless legs	Gain	Motor	Gross movement raises catecholamine tone; in ADHD it tracks better working-memory performance	Preliminary: one small pediatric study²
Rich information streams, fast feedback loops	Gain	Reward, novelty	Dense input and short action-outcome cycles keep reward signaling active and minimize delay to reward	Novelty seeking and delay aversion in ADHD: strong³
Microurgencies	Gain	Temporal, motivational	Compresses a distant goal into a proximal one, shortening delay to reward and triggering mild stress	Delay aversion: strong. Anger, a related approach state, helps on hard tasks only: one multi-study paper
Closing the eyes	Filter	Sensory	Removes competing visual input, freeing resources for internal, abstract representation	Eye closure improves recall by reducing load: moderate, general population⁴

Widening the eyes and closing them look contradictory, but they serve one target: a high signal-to-noise state. Gain levers raise the signal; filter levers lower the noise. Word-by-word feeling does both at once, amplifying the current word while suppressing its neighbors.

None of these were learned from a book. They were found by trial, kept because they worked, and became partly automatic. That is what makes them interesting: they are a self-discovered control system.

The mechanism

Four established models converge on the same picture: in ADHD the problem is less the capacity to attend and more the energy needed to hold the right state.

The inverted U. Prefrontal function depends on norepinephrine and dopamine along an inverted-U curve. Too little catecholamine and attention leaks to every stimulus. Too much, as under strong stress, and the prefrontal cortex goes offline again.⁵ Focus lives in a narrow middle band.

Only the part of the curve above the threshold supports cognitively complex work. Resting ADHD arousal sits far to the left of that window, so gain levers or stimulants must carry it right, and a beta-blocker pulls it back when it overshoots. Filter levers act on the other side of the problem: by removing noise, they let a given arousal level clear the threshold more easily.

Adaptive gain. The locus coeruleus switches between a phasic mode, tight bursts locked to the task, and a tonic mode, diffuse and exploratory.⁶ Distractibility looks like a system stuck in tonic mode or unable to hold phasic mode for long.

State regulation. The cognitive-energetic model locates the ADHD deficit in effort allocation.⁷ Performance degrades most in slow, low-event tasks and recovers when pace or incentives rise. The bottleneck is energetic, not structural.

Optimal stimulation. Zentall and Zentall proposed that ADHD brains idle below their optimal arousal point and seek stimulation to reach it.⁸ Restlessness, novelty-seeking and risk are then not noise but regulation.

Challenge-gated activation. Across six studies, Lench and colleagues induced anger and found it improved goal attainment only when the task was hard. Angry participants solved 39% more difficult anagrams than neutral ones (d = 0.81) and skied a harder slalom better (d = 0.72), with no gain on the easy versions of either task.⁹ Anger also outperformed plain physiological arousal from ten minutes of cardio at moving people to act: 89% versus 69% signed a petition. Microurgencies look like a mild, private version of that state: an approach emotion aimed at an obstacle, not arousal for its own sake. It fits the threshold in the chart above, where activation pays off only on work demanding enough to need it. The samples were students and voters, not people with ADHD, and the findings have not, to my knowledge, been independently replicated.

Read through these models, my levers stop looking like quirks. Gain levers push catecholamine levels up the curve. Filter levers attack the same target from the other side: phasic mode is itself a high signal-to-noise state, and removing competing input makes it easier to hold. This is also, mechanistically, what stimulant medication does chemically.

The hidden cost: a second loop

The expensive part is not the levers themselves but the process that decides when to pull them.

Self-generated focus runs two loops at once:

The task loop. Reading, coding, reasoning: the work.
The supervisory loop. Continuous monitoring of my own arousal, detecting the drop, and firing a lever before focus fails.

The second loop is executive monitoring, which is precisely the function ADHD handles worst. So the system that compensates for the deficit is built out of the deficit. It works, but it taxes the same scarce resource it protects.

This explains a pattern that otherwise looks contradictory. Focus can be deep and still feel exhausting. And when fatigue arrives, it is the supervisor that fails first: the levers stop being pulled, arousal drifts down, and only then does the task collapse.

The literature supports pieces of this. In healthy adults, methylphenidate increases willingness to choose demanding cognitive work by shifting its perceived benefits against its costs, and it does so more in people with lower striatal dopamine synthesis capacity.¹⁰ What remains largely unmeasured is the cumulative physiological cost of running the supervisory loop for hours, day after day.

Overshoot

The same machinery that produces focus can push past the optimum, and then it does not switch off when the work does.

Three properties make this likely.

Positive feedback. The levers activate the sympathetic nervous system. Rising heart rate and muscle tension are read by the brain as evidence of demand, which raises central arousal, which raises sympathetic output again. A loop meant to lift me to the middle of the curve has no built-in stop at the middle.

Summation. Each lever produces a small burst. Over hours they stack, and slower stress hormones such as cortisol join in.

Hysteresis. Arousal rises in seconds and clears over tens of minutes to hours. When the task ends, the system is still running at task intensity with nothing to aim at. In ADHD the off switch is as poorly regulated as the on switch.

The result is the state I know well: done with the work, unable to come down. What helps is a beta-blocker. Propranolol blocks adrenaline and norepinephrine at beta receptors, silencing the bodily signature of arousal. With the body no longer reporting alarm, the loop breaks and the mind can settle. It addresses the adrenergic arm, not cortisol or rumination, which is why it lets me zone out rather than fully reset.

Implications

The goal is not more focus. I can already reach the optimum. The goal is a lower cost per minute of staying there, and a gentler exit.

That suggests three moves.

Externalize the levers. Every lever that lives in the environment instead of in the supervisory loop is effort returned to the task.

Motor. A standing desk, a pedal, a fidget object. Movement becomes ambient.
Event rate. Very small visible units of work, rhythmic sound.
Temporal. Visible countdown timers per chunk, replacing manufactured urgency.
Social. Body doubling, which supplies arousal and accountability at almost no monitoring cost.

Raise the baseline. Stimulants such as methylphenidate and lisdexamfetamine raise catecholamine tone continuously, which should mean fewer levers and a quieter supervisor. A long-acting, smooth formulation may also reduce overshoot, since pulsed manual triggering is what produces the summation. The risk is the same inverted U: too high a dose reproduces the wired state. This is a clinical decision, not a self-experiment.

Design the descent. Taper lever intensity near the end of a session instead of running at full until the last minute. Add a short buffer of low-stakes activity before rest. Use extended-exhale breathing or brief exercise to clear the sympathetic state rather than only suppressing it.

Thesis and open questions

For an ADHD brain, focus is not a default state that gets disrupted. It is a manufactured state, held in place by a self-built control system whose operating cost is paid in executive resources and whose failure mode is overshoot.

Seen this way, the useful question changes. Not “why can’t I focus” but “what is my focus costing, and where is that cost being paid.”

Open questions worth testing:

Can the cost of the supervisory loop be measured directly, for example with pupillometry or heart rate variability across a long session?
Does voluntary eye widening raise locus coeruleus activity, or only follow it?
Does externalizing levers measurably reduce post-task residual arousal?
How much of what people call ADHD burnout is the cumulative cost of hand-cranked focus?

References

Joshi, S., Li, Y., Kalwani, R. M., & Gold, J. I. (2016). Relationships between pupil diameter and neuronal activity in the locus coeruleus, colliculi, and cingulate cortex. Neuron, 89(1), 221–234. doi:10.1016/j.neuron.2015.11.028. Recordings in monkeys; pupil also tracks other brain areas. ↩
Sarver, D. E., Rapport, M. D., Kofler, M. J., Raiker, J. S., & Friedman, L. M. (2015). Hyperactivity in attention-deficit/hyperactivity disorder (ADHD): impairing deficit or compensatory behavior? Journal of Abnormal Child Psychology, 43(7), 1219–1232. doi:10.1007/s10802-015-0011-1. Boys aged 8–12; 29 with ADHD, 23 typically developing. ↩
Sonuga-Barke, E. J. S. (2003). The dual pathway model of AD/HD: an elaboration of neuro-developmental characteristics. Neuroscience & Biobehavioral Reviews, 27(7), 593–604. doi:10.1016/j.neubiorev.2003.08.005. ↩
Vredeveldt, A., Hitch, G. J., & Baddeley, A. D. (2011). Eyeclosure helps memory by reducing cognitive load and enhancing visualisation. , 39(7), 1253–1263. .