Should You Say "I Can Do This" or "You Can Do This"?

On the available evidence, "you" beats "I" - but the evidence is a good deal thinner than the confidence with which that answer circulates. The claim rests principally on one 2014 paper reporting three experiments with undergraduates, in which second-person self-talk produced better anagram performance and stronger stated intentions than first-person self-talk. The effects were around half a standard deviation. Two of the three critical tests came in at p values of .044 and .049. A separate and larger body of work on psychological distance points in a compatible direction, but it mostly tests something adjacent rather than the same thing. If you want a practical answer, "you" is the better default. If you want an honest one, this is a small literature carrying a large amount of advice.

The study everyone is citing

Sanda Dolcos and Dolores Albarracin published "The inner speech of behavioral regulation: intentions and task performance strengthen when you talk to yourself as a You" in the European Journal of Social Psychology in 2014. Three experiments, all with introductory psychology students.

In the first, participants read a vignette about someone facing a choice, imagined themselves as that person, and wrote self-advice with every sentence beginning with either "You" or "I." Then they were given an unrelated anagram task. The "You" group solved more anagrams: an average of 17.53 against 15.96, a difference of about four tenths of a standard deviation, reported at p = .044.

The second experiment moved the self-talk directly in front of the task. Participants wrote ten self-directed thoughts about how to approach the anagrams, again constrained to start with "You" or "I," and a third group did the anagrams with no writing at all. The "You" group averaged 17.58 solved, against 15.61 for "I" and 15.51 for the control group, with effects of about half a standard deviation over each.

That last comparison is the most informative number in the paper and it is almost never mentioned. The "I" group and the do-nothing control group were indistinguishable. First-person self-talk did not hurt. It simply did nothing, while second-person self-talk did something.

The third experiment swapped domains entirely. Participants wrote ten pieces of self-advice about exercising over the next two weeks, then reported how many hours they intended to exercise. The "You" group reported 12.59 hours against 10.34 for the "I" group, at p = .049.

Reading those results honestly

Three findings, in the same direction, from three different setups, is a real pattern. It is also worth being precise about what kind of pattern it is.

The samples were introductory psychology students. Two of the three outcomes were anagrams solved in a laboratory session. The third was self-reported intended hours of exercise, collected immediately after the manipulation, with no follow-up on whether anyone actually exercised. That distinction is not a technicality: the gap between what people intend and what they do is one of the best documented findings in behavioral science, and it is the subject of the piece on the intention-behavior gap. A pronoun that moves stated intentions has not yet been shown to move behavior.

The p values in experiments one and three, .044 and .049, sit just inside the conventional threshold. That does not make them wrong. It does mean that a study of that size run again could easily land on the other side of the line, and it is a reason to treat the pooled direction as more trustworthy than any single result.

The authors also ran mediation analyses suggesting that attitudes toward the task carried the effect: second-person self-talk made people feel more positively about the anagrams and about exercising, and the improved performance followed from that. The bootstrapped confidence intervals for those indirect effects had lower bounds of 0.01, 0.05 and 0.05, which is to say they excluded zero by a margin you could describe as narrow.

Where "you" comes from in the first place

The pronoun choice is not an invention of experimenters. Ethan Zell, Amy Beth Warriner and Dolores Albarracin reported in Social Psychological and Personality Science in 2012 that people spontaneously address themselves as "you," often in the imperative, and that this pattern shows up most in situations demanding direct self-regulation: negative events, experiences of autonomy, and moments of acting rather than planning or reviewing.

The proposed explanation is developmental. Directive speech arrives from outside first. Parents and teachers say "you need to stay focused" and "you can do this," and that grammar gets internalized along with the instruction. When self-guidance is required, the internalized voice comes back in the form it was learned in.

The distancing literature next door

The better-powered work in this neighborhood is about psychological distance rather than the second person specifically.

Ethan Kross and colleagues published "Self-talk as a regulatory mechanism: how you do it matters" in the Journal of Personality and Social Psychology in 2014: seven studies, 585 participants total. Using non-first-person pronouns and one's own name during introspection increased self-distancing. Under social stress, including making a first impression and public speaking, the non-first-person group performed better according to objective raters, showed less distress, and ruminated less afterward. A meta-analysis within the paper found none of this was moderated by trait social anxiety.

Jason Moser and colleagues added physiological evidence in Scientific Reports in 2017. Third-person self-talk reduced an event-related potential marker of self-referential emotional reactivity within the first second of viewing aversive images, and did so without raising markers of cognitive control. The suggestion is that distancing is cheap rather than effortful, which matters because effortful strategies tend to fail exactly when you need them.

Note the gap. Kross and Moser mostly manipulate name and third person. Dolcos and Albarracin manipulate second person. The two are usually cited together as though they were one finding, and they are not. The mechanism proposed for third-person distancing is that you process the situation more like an observer would. The mechanism proposed for second-person self-talk is that you reproduce the encouraging voice of someone else. Those are related intuitions with different logics, and no study has cleanly pitted "you" against your own name against "I" with adequate power. The distancing side of the story is covered in the piece on third-person self-talk.

What to actually do, ranked by how well it is supported

1. Use your own name or the third person when you are trying to calm down. Best supported item here. Seven studies, 585 participants, converging ERP and fMRI evidence, effects that do not require effort.
2. Use "you" when you are about to do something and need direction. Supported by three experiments with modest effects and borderline p values, and consistent with how people spontaneously talk to themselves under self-regulatory demand.
3. Do not assume "I" is harmful. In the one experiment that included a no-self-talk control, first-person self-talk was indistinguishable from saying nothing at all.
4. Do not expect a pronoun to close the gap between intending and doing. The only intention outcome measured was stated intention. Nobody followed up.
5. Do not treat this as settled. The whole second-person case is one paper. It should be replicated at scale before it is repeated with the confidence it usually receives.

The honest summary

Something about grammatical person appears to matter, and it plausibly matters through two different routes: distance from your own reaction, and the borrowed authority of a voice that sounds like it came from someone who was on your side. Both are modest effects. Neither has been shown to change anything that happens outside a laboratory session. Against the wider question of what self-talk can and cannot do, addressed in whether positive self-talk actually works, the pronoun is a small dial on a machine whose overall output is still being argued about.