← Studies Suggest 🧠 Psychology

You're Convinced You've Stopped Changing. So Were 19,000 People, at Every Age From 18 to 68

Harvard and Virginia psychologists asked more than 19,000 adults how much they'd changed in the past ten years and how much they'd change in the next ten. Everyone gave the same answer: a lot behind me, not much ahead. A follow-up proved the error has a price: $129 for a future concert ticket versus $80 for the same nostalgia today.

By Daniel Voss, Psychology & Neuroscience · September 26, 2026

Ask a thirty-five-year-old about the person they were at twenty-five and you will hear a conversion story full of abandoned tastes, discarded values, and embarrassing haircuts. Ask the same person who they will be at forty-five and the story changes genre, because the future self sounds suspiciously like the present self with better lighting. This asymmetry feels like ordinary optimism, but a team of psychologists suspected something more systematic: a miscalibration so reliable you could set your watch by it, and so expensive it shows up in wallets.

In 2013, Jordi Quoidbach, Daniel Gilbert, and Timothy Wilson published the test in Science. More than 19,000 people between 18 and 68 took part across six studies, each sorted into one of two conditions. Reporters described how much they had changed over the past ten years, while predictors estimated how much they would change over the next ten. The researchers compared 28-year-old reporters with 18-year-old predictors, 29 with 19, and so on across 41 decade cohorts, so every prediction faced people who had actually lived through that decade.

The pattern held everywhere they looked, starting with a Big Five inventory completed by 7,519 adults, where predictors expected less change than reporters a decade older remembered (B = 0.14, P < 0.001). The Schwartz values inventory, taken by 2,717 adults, showed a wider gap (B = 0.46, P < 0.001), and preferences showed it too, in 7,130 adults asked about music, food, vacations, hobbies, and best friends. The gap did not fade with age, since the decade-by-condition interaction was essentially zero (B = 0.01, P = 0.68). Teenagers believe their personal history is over. So do grandparents.

The obvious objection is memory, because people might exaggerate how much they used to differ from their present selves, inflating the past and making the future look stable by comparison. The authors checked against reality, turning to MIDUS, a panel that had measured 3,808 adults twice, ten years apart, expressing change as an intraclass correlation where higher means less change. Actual change came out at 0.52, remembered change at 0.51, and predicted change at 0.65: memories nearly perfect, forecasts far too stable. The error sits in the forecast.

Then the researchers put a price on it, asking 170 adults to name their current favorite band and state the most they would pay today to see that band in ten years, while a separate group named the band that was their favorite ten years ago and priced a ticket to see that band this week. The future ticket fetched $129 against $80 for the nostalgia ticket, a 61 percent premium for the privilege of indulging, a decade hence, tastes they will probably have shed (B = 0.16, P < 0.05). Do the subtraction the paper leaves to the reader: $49 per ticket is payment for the belief that your current preferences are permanent, so a couple buying two tickets hands over roughly $98 for the illusion.

The regression table hides a second result the authors never spell out: the forecasting gap stays constant across the lifespan while total change shrinks with age (B = 0.13 for decade, P < 0.001). A fixed gap against a shrinking base means the miss becomes a larger share of the change that remains, so the twenty-year-old underestimates a large change by a fixed amount while the sixty-year-old underestimates a smaller change by the same amount, making the underestimate proportionally bigger. The illusion does not mellow with experience. If anything, experience makes it a larger fraction of what is left to get wrong.

The Strongest Counterargument

The sharpest attack on this paper does not dispute a single number but the interpretation, and it comes in two parts. First, the design is cross-sectional: nobody was followed for ten years and asked to compare forecast with reality, because different people supplied the past and the future, and autobiographical memory is a known fabulist. The MIDUS comparison answers this for personality but only for personality and only approximately, so a skeptic is entitled to want the full longitudinal version, predictions recorded and checked a decade later, before calling anyone deluded.

Second, and more lethal, is the mathematics of uncertainty: suppose you are certain your tastes will change but have no idea whether you will drift toward jazz or away from it. Your best point forecast of signed change is zero, not because you believe nothing will move, but because the movements cancel in expectation. On that reading the predictors were not deluded but doing statistics correctly, and the researchers mistook a rational forecast under uncertainty for an illusion. This objection has real force against every trait measured with a direction attached.

The authors anticipated the uncertainty objection before the critics formalized it, and in a follow-up with 1,163 adults they dropped trait ratings and asked one direction-free question, how much have you changed as a person and how much will you, and the gap survived (B = -0.74, P = 0.007). With direction removed, the math critique loses its footing, yet the effect stands, and Harris and Busseri's reanalysis of MIDUS life-satisfaction data, written as a direct reply to the original team, found the same for absolute change: experienced shifts averaged 0.97 points against predicted 0.82, an underestimate both sides could agree on while arguing about what averages obscure. The honest scorecard is that the phenomenon is real, the word illusion does some work the data cannot fully carry, and the truth sits between delusion and rationality.

What We Didn't Prove

The honest boundaries, stated plainly: the core design compares different people rather than tracking the same people over time, so it infers a forecasting error without ever watching anyone make one. The concert study behind the dollar figures ran on 170 Mechanical Turk workers, a thin base for a 61 percent claim, and willingness-to-pay answers are noisy. The samples skew American, female, and internet-recruited, which limits generalization, and the paper cannot say why the gap exists: the authors offer two guesses, self-flattery and the difficulty of imagining new things, and test neither. The effect describes averages, while plenty of individuals predict their own changes just fine.

The Bottom Line

Two facts fit in one head. You have changed more than you remember planning to. You are planning to change less than you will. The first is a memory. The second is a forecast, and forecasts are where the money leaks: $49 a ticket, tattoo-removal clinics, marriages that seemed permanent at the altar. The present feels like a destination because every past self felt the same way on arrival. History, as the authors put it, is always ending today.

What You Can Do

Tax your future indulgences: before buying the ten-year version of anything, a timeshare, a sports car, a tattoo, price the equivalent indulgence of your ten-years-ago self and ask whether you would pay full fare for it now. Run the best-friend test, because the researchers found people reliably swapped best friends across decades while predicting the current one would last, so audit your own decade-scale commitments for the same asymmetry. Schedule a disagreement with your future self by writing down three tastes you consider permanent, sealing the list, and opening it in five years, since the data suggest at least one will embarrass you. Finally, treat certainty about your future values as a budget item rather than a fact, because the people in these studies were wrong at every age, and you are probably some age.