The Research

What the evidence supports, what it does not, and the study that argues against us.

You came here to find out whether this is real.

Good. Most of this category does not survive that question. Search "affirmations" and you get a wall of confident claims with no papers behind them, or papers cited so loosely that the citation is doing decoration work rather than evidence work.

So here is the deal for this page. Neuralingual is not going to tell you that affirmations are proven. They are not. What there is instead is a real, uneven, sometimes contradictory body of research about how self-talk operates, and a set of design decisions we made from it. Some of that research is strong. Some of it fell apart when other people tried to reproduce it. One of the findings you are about to read argues directly against the thing we sell.

You get all of it. Then you decide.

Start with the study that argues against us

In 2009, Joanne Wood, Elaine Perunovic and John Lee ran the experiment the whole genre would rather you not read. Participants repeated the statement "I'm a lovable person." Among people who scored low on self-esteem, the ones who repeated it felt worse than the ones who did not. Among people who scored high, the repeaters felt better, but only slightly.

Their own summary: repeating positive self-statements "may benefit certain people but backfire for the very people who need them the most."

That is the paper. It is real, it is famous, and if you are selling repeated positive statements about yourself, it is the most inconvenient thing in the literature.

Now the part almost nobody who quotes it mentions. In 2020, Sean Flynn and Michael Bordieri ran two studies to reproduce that result. Both failed. No difference between conditions on any outcome, and no interaction with trait self-esteem.

So where does that leave you? Somewhere honest and a little unsatisfying. The backfire effect is plausible enough to design around and not established enough to treat as settled. Which cuts both ways: it means the scariest finding about affirmations is shakier than the headlines, and it also means the evidence base here is thinner than anybody in this category admits.

The rule we took from it: an affirmation has to be believable to the person hearing it. Not modest. Believable. A line you cannot get anywhere near just measures the distance to a place you are not standing. Aim big. Aim at something you could actually become.

The one choice everything else rests on: "you," not "I am"

Neuralingual does not write "I am confident." It writes "You trust your preparation."

That is the central design decision, and it is the one place where the evidence is genuinely good.

Ethan Kross and colleagues, 2014. Seven studies, 585 people total, published in the Journal of Personality and Social Psychology. The manipulation was almost embarrassingly small: while reflecting on a stressor, use your own name and non-first-person pronouns (including "you") instead of "I." That was it.

The results held up across a range of measures. In a public-speaking study and a first-impressions study, condition-blind raters watching video scored the non-first-person group as performing better. That group also reported less distress, and in the speech study did less of the grinding replay afterward. Two further studies found the same language shift changed how people sized up a stressor before it happened, toward "this is a challenge" and away from "this is a threat."

Dolcos and Albarracín, 2014. Three experiments (95, 143 and 135 introductory psychology students) in the European Journal of Social Psychology, and this one is specifically about the second person rather than the third. People who prepared for a task using "you" solved more anagrams and reported stronger intentions than people who used "I." The effects were small to moderate (d = 0.43 in the first experiment).

Moser and colleagues, 2017. This is the one that made us stop hedging. Using EEG and fMRI, third-person self-talk reduced a neural marker of self-referential emotional reactivity, and it did so without the increase in cognitive-control activity you would expect if the person were effortfully white-knuckling their way to calm. In other words: it looks cheap to run. That matters enormously for something you are supposed to do on a walk, half-distracted, on a Tuesday.

Orvell and colleagues, 2021. Two high-powered experiments found distanced self-talk reduced emotional reactivity across experiences that varied a lot in intensity.

Four independent research groups, converging behavioral, neural and self-report evidence, replicated more than once. In this field, that is a strong hand.

Where that evidence runs out

Here is what those studies do not show, stated plainly, because you would find it yourself and then discount everything else on the page.

They are not studies of affirmations. Every one of them manipulated how a person referred to themselves while thinking about their own situation. Nobody in any of these studies listened to a set of pre-written statements in someone else's voice. That is the gap between the evidence and our product, and it is not small.

Kross found the effect in a specific context. Participants self-distanced and then analyzed their feelings. The authors say so directly, and they note that self-distancing used for other purposes (to avoid your feelings, say) may not help and could hurt. The mechanism is not "say 'you' and good things happen."

The samples are narrow. Undergraduates and online panels, single lab sessions, no long-term follow-up. Kross's team explicitly called for work in non-undergraduate samples and in daily life.

They measured minutes, not months. Not one of them tells you what happens if you do this every day for a year.

Self-efficacy, and the part Bandura ranked last

Albert Bandura's self-efficacy work (1977, and the 1997 book) is the most-cited framework in this whole territory. He identified four sources of the belief that you can do a thing:

Mastery experience. You did it before.
Vicarious experience. You watched someone like you do it.
Verbal persuasion. Someone told you that you could.
Physiological and affective states. How your body reads to you in the moment.

Bandura ranked these. Mastery is the strongest by a distance. Verbal persuasion is the weakest of the four.

We are, mostly, verbal persuasion.

Worse, in Bandura's model verbal persuasion means persuasion from someone else, from a credible source. A coach, a teacher, a parent. Whether a recording of language you commissioned about yourself functions the same way is a question the model does not answer, and we are not going to pretend it does.

So why build on it at all? Because Bandura also found that verbal persuasion gets considerably stronger when it is paired with the other three rather than substituted for them. That is why the practice is designed to sit alongside the work, not replace it. The playlist you listen to before a hard conversation is not doing the job. The preparation is doing the job. The playlist is one of four inputs, and by the founder of the field's own ranking, the weakest one.

If a self-talk product tells you Bandura proves it works, it either has not read Bandura or is counting on you not having.

"Self-affirmation" is not what you think it is

This is the most common sleight of hand in the category, and it is worth twenty seconds of your time.

There is a large, respected literature called self-affirmation theory (Claude Steele, 1988; reviewed by Geoffrey Cohen and David Sherman in Annual Review of Psychology, 2014). Products cite it constantly as evidence that affirmations work.

It is about something else. In that research, you affirm a value you already hold: being a good parent, being honest, your friendships. You typically write about it for a few minutes when something has threatened your sense of yourself. That is the intervention. Reconnecting with a value, not asserting a trait.

It has nothing to do with repeating "I am successful" in a mirror. Wikipedia keeps them on separate pages for exactly this reason.

And that literature has its own hard problem. In 2017, Hanselman, Rozek, Grigg and Borman ran a well-powered replication with 2,109 seventh-graders across two cohorts, using the same procedures as an earlier study that had found large effects in the same district. They found nothing, with estimates precise enough to rule out benefits larger than an effect size of 0.10. Extensive tests for moderators failed to explain the gap.

We use the Steele and Cohen research as a design principle (ground a playlist in something the person genuinely values, not in a trait they are trying to talk themselves into). We do not use it as proof that affirmations work, because it is not about affirmations, and because the field is still arguing about when it holds.

The strongest effect on this page has nothing to do with what you say

Peter Gollwitzer's work on implementation intentions is the best-established thing in this entire document, and it is about when and where, not what.

The format is "when situation X happens, I will do Y." That is the whole trick. Gollwitzer and Sheeran's 2006 meta-analysis pooled 94 independent tests and more than 8,000 participants and found d = 0.65, a medium-to-large effect on goal attainment, with a particularly strong effect on the specific failure of never getting started.

Read that number against everything above. The best-supported finding in this area is not about the content of your inner voice at all. It is about deciding in advance exactly when you are going to do the thing.

Which is why the practice we describe is stubbornly specific about logistics. On the walk. In the car on the way over. The ten minutes before you fall asleep. That sounds like a lifestyle detail and it is actually the part with the best evidence behind it. A playlist you generated and never played is not a practice. Listening is.

Noticing, in Langer's sense

Ellen Langer has spent decades on a definition of mindfulness that is not the one you know. Hers (see Langer and Moldoveanu, Journal of Social Issues, 2000) is actively drawing novel distinctions: noticing new things about a situation, which by definition puts you in the present and makes you sensitive to context. No meditation, no breathing practice, no sitting still.

We lean on this for one specific reason. A worn-in reaction is a reaction you stopped noticing. Langer's framing says the useful move is not to relax about it but to notice it again, freshly, and see that the situation has features your automatic response is not tracking.

Honest caveat: this is the loosest connection on the page. Langer's construct informs how the guide is written. There is no study showing that listening to second-person audio produces Langerian mindfulness, and we are not aware of one.

Attention, and the claim you should stop repeating

This one arrives with baggage, so the debunk comes first.

You will meet a version of this everywhere in the affirmation world: your reticular activating system filters eleven million bits per second down to about fifty, so tell it what matters and it will show you what matters.

The anatomy is wrong. The reticular activating system is a brainstem network that regulates arousal and wakefulness. It has no access to meaning. It cannot recognise that you have decided your writing matters this year. A system that governs whether you are awake cannot sort your day for relevance.

We are blunt about this because we had that line in our own internal writing for months before we checked it. The longer version is in You Are What You Weight.

What survives. The bit-rate figures trace to Manfred Zimmermann's work on sensory systems: roughly 11 million bits per second arriving through the senses against something on the order of 40 to 50 bits per second consciously. Treat that as a gesture at a real asymmetry rather than a specification. Measuring a mind in bits per second borrows a frame built for communication channels, and a brain is not one. The honest version of the claim is qualitative: almost everything you take in, you never consciously touch.

What actually does the work is selective attention, which needs no exotic mechanism and has been well established for decades. Capacity is finite and it is allocated rather than spread evenly. What you have primed yourself to treat as important is weighted more heavily and surfaces faster.

The frequency illusion is the everyday form of it. Arnold Zwicky named it in 2005: you notice something once, and then it appears everywhere. Two ordinary processes stacked, selective attention plus confirmation bias. Nothing in the world changed. Your sort order did.

Honest caveat, and it is the same one that governs this whole page: none of this is a study of affirmations. Selective attention research does not tell you that listening to composed second-person audio reweights what you notice. It tells you that attention is finite and primeable. The step from there to this product is an inference we are making, not a finding we are reporting.

Self-perception: you infer yourself from your own behaviour

Daryl Bem's self-perception theory (1972) holds that people work out their own attitudes partly by observing their own behaviour, in much the way they would infer a stranger's. You are not consulting a fixed internal record. You are reading the evidence of what you did.

This matters here for a reason that is not obvious. If it holds, then the act of doing the practice is itself evidence to you about who you are, somewhat independently of the content of the lines. Showing up is data.

Honest caveat: Bem's theory is well cited and long standing, and it is a theory about attitude inference in general. It has not been tested on this. We use it to explain why consistency seems to matter more than intensity, which is a plausible reading and not a demonstrated one.

What the sport psychology literature says, including the part that cuts against us

Self-talk has been tested seriously in athletics. Hatzigeorgiadis and colleagues (2011, Perspectives on Psychological Science) meta-analyzed 32 studies yielding 62 effect sizes and found a moderate positive effect on performance (ES = .48).

Two moderators matter more than the headline number:

Self-talk worked better for fine motor tasks than gross ones.
Self-talk worked better for novel tasks than well-learned ones.

Sit with the second one. If you are already excellent at the thing, this literature predicts you get less out of it, not more. That is inconvenient for anyone selling self-talk to elite performers, and it is in the meta-analysis in plain sight.

The other thing to know: much of the sport self-talk literature studies short instructional cues ("elbow up," "smooth"), not identity statements. Instructional and motivational cues did not differ significantly overall in this meta-analysis, but instructional cues outperformed motivational ones on fine tasks. What we build is closer to the motivational end.

What we built from this, and what it is not

Here is what actually happens, described the way an engineer would describe it rather than the way a brochure would.

You write a few sentences about what you are working on. Before a single affirmation is written, Neuralingual writes you a guide: a written methodology, three to five principles, the traditions and teachers it is drawing on, the themes your playlist will be organized into, a practical note on how to use it, and a one-line takeaway. Then it writes the affirmations from that guide, grouped into those themes. You can read the guide any time. (Composition is model-driven, which is what makes a guide this specific affordable to produce for one person's exact situation. That is the whole of the AI story here and it is not the interesting part.)

Every affirmation is written in second person. All twelve coach voices default to it, and the instruction runs through every layer of composition rather than being applied as a find-and-replace at the end. First person stays available as a setting for people who already have a practice built on it. It is not the default, because of the evidence above.

Now the parts you should hold us to.

The second person is a composition rule, not a hard filter. There is no final check that rejects a stray "I am" line. It is rare. It is not impossible. If you catch one, that is a bug, and we would like to hear about it.

The names in your guide are not citations. They are the traditions and practitioners the composition drew on, and they render as a name, sometimes a work, and a sentence about the contribution. No year, no journal, no DOI, no link. The system is instructed to name real people only where it is confident and to fall back to a general attribution ("classical sports psychology," "the cognitive-behavioral tradition") when it is not. A separate check flags sources that look obviously invented, and today that check logs rather than blocks.

So nothing in the product verifies that a named source is real. Treat those names as pointers worth following, not as a reference list you can lean on. If one of them is wrong, we want to know.

And not every playlist carries a guide. Older playlists and some first-person playlists do not have one.

That is the honest version. It is less impressive than "backed by peer-reviewed research," which is what we would say if we were willing to say it.

What we do not know

Whether listening to second-person audio produces the same effect as generating second-person self-talk yourself. We have not found a study that tests it. It is the biggest open question about this product.
What the right dose is. We suggest eight to ten listens across a week in different contexts, because that is what practice looks like, not because a study established it.
What happens over months. The evidence base is lab sessions.
Whether the backfire risk from Wood et al. is real. Two attempts to reproduce it failed. We design as though it might be.
Whether any of this holds outside undergraduate and online-panel samples. Kross's own team asked for that work.
Whether reweighting attention through composed audio works the way selective-attention research suggests it should. The attention findings are solid; applying them to this product is our inference.
The largest gap on this page, stated plainly: no study has tested this product, or anything structurally like it, over time. The research manipulates which pronoun you use while thinking about your own stressor, once, in a lab. This is composed audio you listen to repeatedly over weeks. That distance is bigger than any of the others listed here, and nothing below it in this section should distract from it.
How well the central paper holds up. Kross 2014 is seven small-sample studies published in 2014, which places it inside the window social psychology has since spent a decade re-examining. We have not found an independent replication of the second-person result at scale. That does not make it wrong. It does mean it is one paper, and we are telling you that before someone else does.

If you find research that sharpens or contradicts anything here, send it. This page gets updated, and the updates get dated.

What this adds up to

There is a real mechanism here, and it is narrower than the category claims. Talking to yourself in the second person appears to change how you appraise and regulate a hard moment, at low cognitive cost, with converging evidence from four research groups. Deciding in advance when you will do a thing has the strongest evidence of anything on this page. Belief has to be reachable or the statement just measures the gap. Language is the weakest of Bandura's four sources of self-efficacy, so it belongs next to the work rather than instead of it.

We built around that, including around the parts that say "this does not work the way you think."

What the evidence supports is a practice worth running. Not a result anyone can promise you.

Sources

Each entry has enough detail to look up and check. Where a finding failed to replicate, the replication is listed with it.

Second-person and distanced self-talk

Kross, E., Bruehlman-Senecal, E., Park, J., Burson, A., Dougherty, A., Shablack, H., Bremner, R., Moser, J., & Ayduk, O. (2014). "Self-Talk as a Regulatory Mechanism: How You Do It Matters." Journal of Personality and Social Psychology, 106(2), 304–324. DOI: 10.1037/a0035173. Seven studies, total N = 585.
Dolcos, S., & Albarracín, D. (2014). "The inner speech of behavioral regulation: Intentions and task performance strengthen when you talk to yourself as a You." European Journal of Social Psychology, 44(6), 636–642. DOI: 10.1002/ejsp.2048. Three experiments, N = 95, 143, 135.
Moser, J. S., Dougherty, A., Mattson, W. I., Katz, B., Moran, T. P., Guevarra, D., Shablack, H., Ayduk, O., Jonides, J., Berman, M. G., & Kross, E. (2017). "Third-person self-talk facilitates emotion regulation without engaging cognitive control: Converging evidence from ERP and fMRI." Scientific Reports, 7, 4519.
Orvell, A., Vickers, B. D., Drake, B., Verduyn, P., Ayduk, O., Moser, J., Jonides, J., & Kross, E. (2021). "Does Distanced Self-Talk Facilitate Emotion Regulation Across a Range of Emotionally Intense Experiences?" Clinical Psychological Science.

Attention, priming, and self-perception

Zimmermann, M. (1986/1989). Estimates of sensory information throughput (~11 million bits/second) versus conscious processing (~40 to 50 bits/second), in Human Physiology (Schmidt & Thews, eds.). Widely requoted; treat as an order-of-magnitude gesture, not a measurement. The mind is not a communication channel and the units are borrowed.
On the reticular activating system: the RAS regulates arousal and wakefulness. It is not a semantic filter and does not select for personal relevance. The popular affirmation-world claim attributing attentional selection to it is an anatomical error, not a contested finding.
Zwicky, A. (2005). "Just Between Dr. Language and I." Coined the frequency illusion: selective attention plus confirmation bias, not a change in the world.
Bem, D. J. (1972). "Self-Perception Theory." Advances in Experimental Social Psychology, 6, 1–62.

Self-efficacy

Bandura, A. (1977). "Self-efficacy: Toward a unifying theory of behavioral change." Psychological Review, 84(2), 191–215.
Bandura, A. (1997). Self-Efficacy: The Exercise of Control. W. H. Freeman. (Source of the four-source model and the ranking in which verbal persuasion is weakest.)

Self-affirmation theory (the values construct, not positive self-statements)

Steele, C. M. (1988). "The Psychology of Self-Affirmation: Sustaining the Integrity of the Self." In L. Berkowitz (Ed.), Advances in Experimental Social Psychology, Vol. 21, 261–302.
Cohen, G. L., & Sherman, D. K. (2014). "The Psychology of Change: Self-Affirmation and Social Psychological Intervention." Annual Review of Psychology, 65, 333–371.
Sherman, D. K. (2013). "Self-Affirmation: Understanding the Effects." Social and Personality Psychology Compass, 7(11), 834–845.
Hanselman, P., Rozek, C. S., Grigg, J., & Borman, G. D. (2017). "New evidence on self-affirmation effects and theorized sources of heterogeneity from large-scale replications." Journal of Educational Psychology. N = 2,109 across two cohorts; null result precise enough to reject effects larger than 0.10.

The counter-evidence, and its replication

Wood, J. V., Perunovic, W. Q. E., & Lee, J. W. (2009). "Positive Self-Statements: Power for Some, Peril for Others." Psychological Science, 20(7), 860–866.
Flynn, S. M., & Bordieri, M. J. (2020). "On the failure to replicate past findings regarding positive affirmations and self-esteem." Journal of Contextual Behavioral Science. Two studies, both failed to replicate Wood et al. (2009).

Implementation intentions

Gollwitzer, P. M. (1999). "Implementation intentions: Strong effects of simple plans." American Psychologist, 54(7), 493–503.
Gollwitzer, P. M., & Sheeran, P. (2006). "Implementation Intentions and Goal Achievement: A Meta-Analysis of Effects and Processes." Advances in Experimental Social Psychology, 38, 69–119. 94 independent tests, N > 8,000, d = 0.65.

Mindfulness as active noticing

Langer, E. J., & Moldoveanu, M. (2000). "The Construct of Mindfulness." Journal of Social Issues, 56(1), 1–9.
Langer, E. J. (1989). Mindfulness. Addison-Wesley.

Self-talk and performance

Hatzigeorgiadis, A., Zourbanos, N., Galanis, E., & Theodorakis, Y. (2011). "Self-Talk and Sports Performance: A Meta-Analysis." Perspectives on Psychological Science, 6(4), 348–356. 32 studies, 62 effect sizes, ES = .48.

Last reviewed: 2026-08-02.

Get posts like this in your inbox

Occasional writing about affirmations, intentional self-development, and building AI tools. No spam, unsubscribe anytime.