Possible Selves: How Imagining Who You Could Become Shapes Behavior

Imagining who you could become changes behavior when the image is specific, when it is paired with a matching version of who you are afraid of becoming, and when it comes attached to something you can do this week. Strip out any of those three and the effect thins out fast. That conditional structure is the actual finding, and it is considerably more demanding than the version that reaches most people, which is roughly "picture your best self."

The construct has a precise origin. Hazel Markus and Paula Nurius introduced possible selves in 1986 in American Psychologist, defining them as a person's ideas of what they might become, what they would like to become, and what they are afraid of becoming. The move was to treat the self-concept not as a summary of who you are now but as a repertoire that extends into the future, with the future entries doing real work: supplying incentives for behavior and supplying the frame you use to interpret how the present is going.

Why the feared self matters as much as the hoped-for one

This is the part that almost never survives into popular treatments, and it is the part with the most interesting evidence behind it.

Daphna Oyserman and Hazel Markus reported the key result in 1990 in the Journal of Personality and Social Psychology, studying adolescents across different levels of involvement with the juvenile justice system. What distinguished the groups was not how positive their hoped-for selves were. It was balance: whether the expected self and the feared self sat in the same domain. A youth who expected to graduate and feared dropping out has a balanced pair. A youth who hoped to be wealthy and feared being lonely does not, because the hope and the fear point at different parts of life and neither one constrains the other.

Balance was associated with less delinquency. The proposed reason is that a matched pair does something a lone hope cannot: it makes the same behavior relevant in two directions at once. Going to class both approaches the wanted self and avoids the feared one. A hoped-for self by itself has no failure state attached, so a bad week does not register as anything in particular.

That study is cross-sectional and correlational, and it is a specific population. It does not establish that installing a feared self causes better behavior. It established the pattern that later intervention work went looking for.

What happened when they tried to install one

Oyserman's School-to-Jobs program is the most-cited attempt. The 2006 report with Deborah Bybee and Kathy Terry in the Journal of Personality and Social Psychology describes a brief school-based program for low-income eighth graders, built to make school-focused possible selves feel plausible, to attach concrete strategies to them, and to change how students interpret difficulty. Reported outcomes at follow-up included improvements in grades, attendance and classroom behavior, with the possible-selves measures mediating the effect.

Here is where honesty is required. Most of the intervention evidence in this literature comes from a small number of research groups, much of it from Oyserman's own. Independent, large-scale, preregistered replications are scarce. The measurement is open-ended, which is a strength conceptually and a weakness practically, because it means outcomes depend on how researchers code free-text responses. And the field this belongs to has a specific track record worth knowing about.

When brief identity and mindset interventions have been tested at scale by independent teams, the effects have consistently come in smaller than the founding studies. The best example is the National Study of Learning Mindsets, published by David Yeager and colleagues in 2019 in Nature: a nationally representative sample of more than 12,000 ninth graders across 65 schools. The intervention worked, but modestly and unevenly. It raised grade point average by about a tenth of a grade point among lower-achieving students, did essentially nothing for higher-achieving ones, and produced larger effects in schools where peer norms already supported taking on challenges.

That is not a debunking. A tenth of a grade point from a short exercise is a real result, and the study is a model of how this work should be done. But it is the realistic ceiling for this category of intervention, and it is a long way from the transformation implied by most writing about imagining your future self.

The interpretation-of-difficulty problem

Oyserman's later framework, identity-based motivation, adds a wrinkle that explains a lot of the mixed results. A possible self does not simply pull you forward. It also supplies an interpretation of how hard things feel.

If difficulty reads as evidence that you are doing something important, the identity holds and you keep going. If difficulty reads as evidence that this is not who you are, the same identity now argues for quitting. Which reading you get depends on context, and that is a genuinely uncomfortable finding for anyone selling future-self imagery as a straightforward intervention. The same vivid image can cut either way depending on what the first hard week feels like.

What the evidence supports, in order

1. Specificity over grandeur. "A person who reads for thirty minutes before bed" outperforms "a well-read person" for the same reason specific goals beat vague ones. There is nothing to act on in an abstraction.
2. A matched feared self in the same domain. The balance finding is the most distinctive contribution of this literature. If your hoped-for self is about finishing a degree, the feared self should be about not finishing it, not about something unrelated.
3. Attached strategies. Every intervention in this area that reported effects included a step where students converted the image into specific actions. The imagery on its own is not the active ingredient, and if-then planning is the best-documented way to make that conversion, as covered in the piece on implementation intentions.
4. Plausibility and proximity. A possible self that sits decades away has the same problem the future-self literature keeps finding: the brain does not reliably treat distant-you as you, which is the subject of the piece on future-self continuity. Nearer, more credible versions have more traction.
5. A prepared reading of difficulty. Deciding in advance what a hard day means is the least intuitive item here and possibly the most important one.

How to hold this

Possible selves is a good construct with a real empirical base and an intervention literature that has not yet been tested the way it needs to be. The honest summary is that the descriptive work, showing that people carry future self-images and that the structure of those images correlates with behavior, is fairly solid. The prescriptive work, showing you can install a better one and get durable behavior change, is promising and unresolved.

Which puts it in the same category as most of what gets bundled into future-focused self-improvement: a real mechanism, oversold. The wider version of that argument runs through the piece on whether manifestation actually works, and the specific question of whether picturing a future self helps or costs you is handled in the piece on what visualization actually does.