The science

Changing who you are, not just what you do

"Become the kind of man who doesn't" is identity language. What the research on identity and behaviour change supports, and where it backfires.

By the WINNR team · 17 October 2026 · 11 min read

You have tried rules. No porn on weeknights, no phone in the bedroom, a blocker on the laptop. They hold for a while and then they do not, and somewhere around the third collapse you met a different idea: stop changing the behaviour and change the person. Do not aim to watch less porn. Become the kind of man who does not watch porn.

There is a real research literature underneath that idea, and a specific way it goes wrong for this behaviour, which almost nobody repeating the advice mentions.

Two things first. There is essentially no research on identity change and pornography: one qualitative study, below. Everything else here is transferred from smoking cessation, from alcohol and drug recovery, and from educational psychology, and transfers can be reasonable and still be wrong. And the strongest single finding in the whole literature runs against the way the advice is usually given.

What "identity" means in the research

Identity splits into two measurable things. Self-identity is how strongly you endorse a label about yourself: "I see myself as a smoker", "I think of myself as someone who is quitting". Social identity is how strongly you see yourself as part of a group whose norms feel like your own.

The Social Identity Model of Recovery, set out by Best and colleagues, argues that getting out of addiction means moving from a social identity built around using to one built around recovery, and that this transition is socially negotiated, not privately decided [1]. Frings and Albery's Social Identity Model of Cessation Maintenance makes a related case for staying stopped: identifying with a group that treats abstinence as normal helps hold the change in place [2]. Best and colleagues' is a model paper. Frings and Albery set out a model and offer initial evidence for it. Neither is a trial.

The strongest evidence comes from smoking and from drug recovery

Smoking

This is the best-developed literature, and much of it is prospective.

A Dutch study followed 3,411 smokers, measured their identity as a smoker and as someone who was quitting, and looked at who had made a quit attempt four months later. Both mattered and did different work: "quitting" identity was especially important for intention, and both predicted actually trying [3].

A survey of recent ex-smokers in England is more sober. Of 574 people who had quit within the past year, 80.3 per cent thought of themselves as a non-smoker rather than a smoker. That self-label predicted abstinence three months later (adjusted odds ratio 2.64, 95% CI 1.09-6.42). At six months it no longer reached significance (adjusted odds ratio 2.05, 95% CI 0.83-5.04), and the authors concluded the protective effect may be limited to the early period after quitting. Follow-up was poor, 179 and 163 people, and abstinence was self-reported [4].

A cohort of 943 smokers and ex-smokers, measured yearly from 2009 to 2014, found smoking identification rose among continuing smokers and fell among ex-smokers, with lower socioeconomic groups resisting change more [5].

Drug and alcohol recovery

The clearest work tracked 132 adults entering a residential therapeutic community, measuring their identity as a user and as someone in recovery at admission, every two weeks, and at exit, with 60 followed up later. User identity fell for about 76 per cent during the first month, recovery identity rose, and the size of that shift predicted later outcomes after controlling for severity at entry. It accounted for roughly 34 per cent of the variance in how much people drank, 41 per cent in how often, and 49 per cent in life satisfaction. For other drug use it was about 5 per cent [6].

That is one study, in residential treatment, for substances, with no control group. It shows identity shift travelling with recovery, not producing it.

The one study on identity and pornography

A 2026 qualitative study interviewed 27 people in structured recovery groups, 26 men and one woman, about recovery from problematic pornography use. The authors describe a movement from what they call a "spoiled identity" towards something more authentic, and report accountability within the group as the mechanism participants named most often [7].

It is not evidence that identity work causes change. The sample was self-selected from men already committed to group recovery, there was no comparison group, and the method describes lived experience rather than testing a mechanism. It tells you men in recovery talk about themselves in identity terms. That is not a result.

Identity-based motivation, and its one randomised test

The theory with the most developed experimental base is Oyserman's identity-based motivation, and its claim is narrower than "be the person you want to be". Which identity is active depends on the situation, and that changes what difficulty means. When an action feels identity-congruent, difficulty reads as a signal that the thing matters. When it feels incongruent, the same difficulty reads as evidence that this is not for people like you [8]. The thought "this is harder than I expected" is not self-interpreting.

It was tested in a randomised trial of a school programme with 264 low-income Detroit eighth-graders, eleven sessions over seven weeks, followed for two years. The intervention group did more homework, missed fewer days, had higher grades and reported lower depression, and modelling indicated that changes in their future-self images mediated the behavioural effects [9]. Reported effect sizes included d = 0.74 for homework time and d = 0.30 for grades, both at the second-year follow-up; the first-year homework effect was d = 0.16 and not significant [9].

That is a genuine randomised test of an identity mechanism changing behaviour, on schoolwork in teenagers, about as far from compulsive pornography use as adjacent evidence gets.

The popular version, and where it outruns the evidence

Most people meet this idea as "identity-based habits", a framing that reached a very large audience through the 2018 bestseller Atomic Habits rather than a journal. The book is not a research paper and does not claim to be, and the advice in it is not foolish. But there is no trial of it, and where the underlying idea has been tested directly, the results are mixed.

The cleanest experimental test of identity wording is three small studies in which survey questions described voting with a noun ("being a voter") or a verb ("voting"). Turnout was higher in the noun condition: 95.5 against 81.8 per cent in one experiment of 88 people, and 89.9 against 79.0 per cent in another of 214 [10].

A later large field experiment replicated the design and found no significant difference between the wordings, with an estimated effect near zero. Both versions did worse than a standard get-out-the-vote message [11]. The original authors argued the replication was not psychologically comparable [12]. The disagreement is unsettled.

Something similar happened when researchers tried to strengthen identity deliberately. In a randomised experiment with 339 daily smokers, a writing exercise designed to build a "quitter" self-identity produced at best a small, marginally significant effect, and nothing carried through to quit intention or behaviour [13]. That is the direct test of talking yourself into an identity, and it did not work. Identity measurably tracks behaviour change; attempts to manufacture it with a brief exercise have mostly failed. Both are true.

The direction of causation is the inconvenient part

A study of 1,036 smokers and ex-smokers used cross-lagged modelling across repeated waves, which lets you look at both directions at once, and the pattern held when the authors re-ran it on earlier waves of the same cohort. Identity and behaviour were reciprocally related: each predicted change in the other. But the stronger path ran from behaviour to identity. The authors put it plainly: behaviour appears more important for identity change than identity for behaviour change [14].

That reverses the advice. You do not decide you are the kind of man who does not watch porn and then find the behaviour follows. You accumulate weeks of evidence and the self-concept catches up, after which it starts contributing something of its own. Identity is better treated as a lagging indicator that becomes a protective one.

The engine is elsewhere: in the plan you make for the specific moment (if-then plans), in working out what the behaviour does for you (CBT methods), and in what you do while an urge is present (ACT).

When identity language turns into shame

Identity framing works by creating a gap between who you take yourself to be and what you are doing. For pornography, that gap has its own literature, and it does not look benign. It is called moral incongruence: the distress produced when a man's use conflicts with his own moral view of it.

A systematic review and meta-analysis by Grubbs and colleagues pooled the available samples. The association between moral incongruence and self-perceived problems from pornography was large across seven samples, aggregate Fisher's rz = .66. The association between how much pornography people actually used and self-perceived problems was considerably smaller across ten samples, rz = .28. The authors' own comparison made moral incongruence the much better predictor (Fisher's z = 17.55, p < .001) [15]. In a nationally representative US sample of 2,075 adults, about 11 per cent of men agreed that "I am addicted to pornography", and agreement was predicted by religiousness and moral incongruence as well as use [16]. In a second representative sample of 2,519, the link between use and self-reported addiction was stronger among more religious participants [17].

Read that with the identity advice in mind. "Become the kind of man who doesn't" manufactures exactly the configuration this literature associates with distress and with feeling addicted. If you hold a strong self-concept as a man who does not do this, and you do it on Thursday night, the identity has given you no mechanism. It has given you a contradiction, and contradictions about the self are the part that hurts.

Two qualifications. The meta-analytic samples are cross-sectional, so they cannot establish direction. And a nationally representative Polish study of 1,036 adults, of whom the 880 reporting any pornography exposure entered the models, found frequency of use the strongest predictor of both self-perceived addiction (β = 0.52) and problematic use (β = 0.43), with moral incongruence distress weaker though still significant [18]. How much you watch is not irrelevant. Saying so would be an overcorrection.

Our pages on whether porn addiction is real and telling whether you have a problem go further into self-labelling.

What separates a useful identity shift from a punishing one

None of this has been tested on pornography. Treat it as reasoning from adjacent evidence.

It points towards something, not away from something. The smoking work repeatedly finds that identifying with quitting does more than un-identifying with smoking, and recommends strengthening the first rather than attacking the second [14]. In the therapeutic community study, the rise in recovery identity tracked outcomes alongside the fall in user identity [6]. "Someone who is working on this" has evidence behind it. "Someone who would never" does not.

It is earned, not declared. Behaviour drives identity more strongly than the reverse [14], and the exercise designed to install an identity failed [13], so let the self-concept follow a record of behaviour. A single lapse is then not disconfirming evidence about who you are. Treating it as proof of character is what converts a slip into a return to the old pattern, and it is why a counter that resets to zero is a bad instrument.

It is provisional, not a rule. An identity that works as a law about yourself breaks loudly. ACT offers a different relationship to self-description, self-as-context, in which you hold self-concepts lightly rather than defending them [19]. That is the real alternative to identity-as-rule, and our post on ACT covers it.

It has other people in it. Every model here treats identity change as social rather than internal [1][2], and the qualitative pornography study put group accountability at the centre [7].

Common questions

Does changing your identity actually help you quit porn?

Nobody knows. There is one qualitative study of 27 people and no trial. In smoking, identity measures predict quit attempts and short-term abstinence; in residential drug treatment, identity shift travels with better outcomes. Whether engineering the shift helps is a separate question, and the one smoking experiment that tried largely failed.

Is "become the kind of person who" advice backed by research?

Partly. The mechanism has a real literature and one randomised trial behind it, in schools. The slogan comes from a popular book, and the cleanest experimental test of identity wording failed to replicate in a large field experiment.

Should I call myself a porn addict?

The research on moral incongruence suggests labels loaded with moral failure track distress more closely than how much you use. You do not need a diagnostic label to decide you want to use less. If use is seriously affecting your life, speak to a GP or psychologist rather than settling the label question yourself.

What if I slip after telling myself I am "someone who doesn't"?

Expect it to feel worse than a plain slip, because a rule about yourself has broken, not a plan. Treat it as information about a situation, not your character.

WINNR is a science-based app, expected mid-November 2026, that draws on ACT and CBT. It is built around the parts of this with a mechanism: planning for specific moments, tracking what you do rather than days passed, and your relationship to urges. It does not ask you to declare a new identity, because the evidence runs the other way round.

A note on what this is

WINNR is not a treatment or therapy. If porn use is seriously affecting your life, speak to a licensed professional; if you are in crisis, contact your local emergency services.

Want a plan for your own pattern? Take the private 4-minute check

Fighting an urge right now? Open the urge page

#

Want a plan for your own pattern? Take the private 4-minute check. Fighting an urge right now? Open the urge page.