borys
← Writing

ADHD · ∩

2026-07-11~15 minSolertio · esej

Engineer the identity, not the willpower

Willpower is the least reliable input a person has, and for an ADHD-shaped mind it is close to useless as a strategy. The more durable lever is the one underneath behaviour: identity. Change who you believe you are and the actions follow with far less friction. This is real, it is engineerable, and — held wrong — it becomes a cage you build yourself.

Behaviour flows downhill from identity

James Clear's central move in Atomic Habits is to separate three layers of change: outcomes (what you get), processes (what you do), and identity (what you believe you are). Most people try to change from the outside in, starting with the outcome. The durable direction is the reverse: identity first. His compression of the mechanism is the useful part — every action you take is a vote for the type of person you wish to become. You are not trying to write a book; you are casting votes for being a writer, and the book is a byproduct of enough votes.

This is not a slogan dressed as psychology. It rests on a real finding. Daryl Bem's self-perception theory holds that we infer our own attitudes and identity partly by observing our own behaviour — we watch what we do and conclude who we are. Festinger's dissonance work adds the pressure: once you hold an identity, acting against it is uncomfortable, so consistency becomes a force that carries the behaviour for you. Put those together and you get a loop you can deliberately drive: act, accumulate evidence, update the self-image, and let the updated identity make the next action cheaper.

Why this matters more if your executive function is unreliable

For the ADHD-shaped mind, the standard advice — try harder, be disciplined — asks the one system that is least dependable to carry the load. If attention and follow-through swing with a miscalibrated reward loop, then a plan that requires steady willpower is a plan built on sand. Identity engineering routes around that. It does not ask you to summon effort in the moment; it changes which actions register as "like me," so the right move becomes the low-resistance move rather than the one you have to force.

Every action you take is a vote for the type of person you wish to become.James Clear

The practical form is small and evidence-based. You do not install "I am disciplined" by declaring it; you install it by manufacturing a steady trickle of small, undeniable evidence — a two-minute version of the behaviour, done often enough that the self-perception loop has something to observe. For a profile that is strong at starting and weak at finishing — the initiator who leaves a trail of half-built things — the target identity is specifically the finisher, and the votes are deliberately cast on completion, not initiation: the smallest possible closed loop, repeated, until "I finish things" has enough evidence to become load-bearing.

Design the identity and the room together

Identity work fails when it floats free of the environment. The self-perception loop needs actions to observe, and actions come cheapest when the room is arranged to produce them — which is why identity engineering and environment design are the same project from two ends. You set the identity you are voting for, then you lower the friction on the specific action that counts as a vote, so the evidence accumulates whether or not you feel motivated. Motivation, if it shows up, is a bonus, not a dependency.

A worked identity: manufacturing a finisher

Take the specific profile this blog keeps circling: strong at starting, weak at finishing — the initiator who leaves a comet-trail of half-built things and a private conviction that they cannot complete anything. The willpower prescription is to force completion by discipline, which asks the least reliable system to carry the load. The identity move is different: you decide the target identity is "a person who finishes," and then you manufacture the evidence for it, because self-perception runs from behaviour to belief, not the other way.

The mechanics have to stay small, or the loop has nothing to observe. Define "done" so tiny it is almost embarrassing — not "ship the product" but "close one issue and mark it done," not "write the essay" but "publish the two-paragraph version." The point is not the size of the output; it is producing a steady stream of undeniable instances of finishing for the self-perception system to watch. Each closed loop is a vote, and the votes are cast on completion specifically, because that is the identity under construction. An initiator has years of evidence for starting and almost none for finishing; you are correcting the ledger on purpose.

Then write it down, because the ADHD narrator does not keep reliable books. An external record — a visible log of finished things — does the remembering that an unstable self-image cannot, so that on the week the internal narrator insists you never complete anything, the ledger says otherwise and the identity holds. Over enough weeks, "I finish things" accumulates enough evidence to become load-bearing, and completion starts to feel like something you do rather than something you fail at. None of this is motivation. It is environment and identity engineered together, so the evidence accrues whether or not the loop is cooperating that day.

Where a manufactured identity turns on you

This is a powerful tool, and powerful tools cut both ways. Three failure modes deserve to be named plainly.

One: identity is a prior that resists disconfirmation. The same stability that makes an identity carry behaviour makes it hard to update, and a strong identity in the wrong place is a trap. "I am a founder" makes it feel like dying to shut down a venture that should be shut down; "I am a runner" keeps you running on the injury. The startup domain rewards the ability to pivot, and a rigidly engineered identity is precisely what makes pivoting feel like self-betrayal. The tool that lowers the cost of the right action also raises the cost of abandoning it.

Two: the causal arrow requires the action, or it is theatre. Bem's mechanism runs from behaviour to belief, not the other way. Affirmations, vision boards, and declaring your new identity without casting the votes are the failure — self-narrative with no evidence underneath, which the mind quietly discounts. Identity-first does not mean identity-only. If the behaviour is not there, you are not engineering an identity; you are role-playing one.

Three: the ADHD substrate may not hold still. Identity engineering assumes a self-narrative stable enough to accumulate votes into. The same unstable reward loop that makes willpower unreliable can make the self-image swing — energised founder one week, impostor the next — and votes cast into a shifting ledger do not compound the way the theory assumes. The honest version is that identity is a more reliable lever than willpower for this kind of mind, not a fully reliable one, and it works best braced by an external record that remembers who you are when your internal narrator forgets.

So stop trying to out-discipline a system that was never going to cooperate. Choose the identity, cast the smallest honest vote for it today, and write the evidence down — then stay willing to lose the identity the day it stops serving you.

Sources

  1. primaryJames Clear, Atomic Habits (2018) — identity-based habits.
  2. primaryDaryl J. Bem, "Self-Perception Theory," Advances in Experimental Social Psychology (1972).
  3. primaryLeon Festinger, A Theory of Cognitive Dissonance (1957).
  4. secondaryRussell A. Barkley, on ADHD and self-regulation — why willpower-based plans underperform.
  5. secondaryCarol Dweck, Mindset (2006) — self-beliefs and behaviour change.