Believing In Yourself: Does It Actually Help Kids Learn?
A famous study said yes. A fake brain scan on young kids says otherwise.
Here’s a pep talk I’ve probably given a hundred times, to my kids, myself, and my students: “You’ve got this. I know you can do it.” It’s the kind of thing we say instinctively, right alongside teachers, coaches, and anyone who has ever handed a kid (of any age) a worksheet. The logic feels good: believe in yourself a little more than the evidence strictly supports, try harder, get better. Confidence as fuel. Add in a power pose and we’re even better.
For almost two decades, one especially famous study seemed to back that logic up with real data. It’s been cited hundreds of times. It shows up in textbooks. I’d bet a fair number of you have heard some version of its conclusion repeated as settled fact: overconfident kids learn better.
A big new investigation just took that finding apart and what’s left standing is a lot more useful than “believing in yourself works” (and it involved a fake brain scan).
Research Spotlight
The original story was appealing. Back in 2007, Shin et al. had preschoolers and young grade-schoolers study several lists of words, predicting each time how many they’d remember before actually being tested. Kids, as usual, guessed higher than they scored (kids are usually a little overconfident about what they can do). The researchers then split the group into “more overconfident” and “less overconfident” kids and tracked how their memory changed over the course of the task. The more overconfident kids seemed to improve more. The conclusion: overconfidence must give kids some kind of edge, maybe by keeping them motivated to keep trying.
A team of researchers decided to run it again, more carefully. When Hu et al. (2026) repeated the exact same setup with a new group of kids, they got the same headline result: the more overconfident kids did look like they improved more. But when they dug into why, the story fell apart.
It turned out the “less overconfident” kids just happened to score unusually well on the very first test, and the “more overconfident” kids happened to score unusually low. That’s not a coincidence - it’s baked into how “overconfidence” gets calculated in the first place.
And here’s the thing about kids who score unusually low or high on any one test: the next time you test them, their scores tend to drift back toward the middle, just by chance, the same way a lucky bowling frame tends to be followed by a more average one.
Once the researchers accounted for that, the apparent “improvement” from overconfidence disappeared. The high scorers came back down. The low scorers came back up. Nobody actually got better because they were overconfident.
So the researchers double-checked it with a computer. They built a simulation of thousands of pretend children, and deliberately programmed overconfidence to have zero real effect on learning. Then they ran the exact same kind of group comparison the original study used and it still produced the same fake “overconfidence helps” pattern, purely from the quirks of how the groups got sorted.
In other words, this way of analyzing the data can’t actually tell a real effect from a statistical illusion. That’s a pretty significant problem for a finding that’s shaped how a lot of us think about kids and confidence.
Then came the fake brain scans. To get a truly clean answer, the researchers ran one more experiment - this time assigning confidence levels at random, rather than sorting kids after the fact. Each child wore an EEG headband and watched their own brain waves appear on a screen. Then they were told (this was a cover story, explained to the kids afterward) that the scan showed their memory was either excellent or poor, with a specific score they’d supposedly hit on the next round. Half the kids got the confident version, half got the modest version, assigned by coin flip rather than by how they’d actually performed. It worked — the “excellent memory” kids became more overconfident. And because the groups were randomly assigned instead of sorted by their own scores, both groups started out identical.
The result: no difference at all in how much either group’s memory improved. None. Being told you’re great at something, with nothing about your actual ability changed, didn’t move the needle. When the researchers ran the same data back through the original, flawed method of analysis, the fake “overconfidence helps” pattern reappeared right on schedule - a tidy demonstration of exactly how that illusion gets manufactured.
And this isn’t just a memory-game story. A separate study of fourth graders found something that fits neatly alongside this: what predicted how well kids actually did in school wasn’t how confident they were, it was how accurate their self-assessment was. Kids who had a realistic sense of their own abilities (e.g., whether that sense was modest or bold) did better than kids whose self-image was off in either direction (Freeman et al., 2017).
And it echoes something we already know about praise. A study following 120 parent-child pairs over a year and a half found that when parents’ praise ran a little too rich (e.g.,“That was incredibly good!”) it actually predicted lower self-esteem in kids over time, not higher (Brummelman et al., 2017). Inflating a child’s sense of their own ability, whether it comes from the outside or from within, doesn’t seem to build the thing we’re hoping it’ll build.
Why This Matters
None of this means confidence is bad, or that kids shouldn’t be encouraged. What it challenges is a much narrower claim: that puffing up a child’s belief in their own ability, separate from anything true about their actual skill, makes them learn better. Across a replication, a simulation, and a randomized experiment, that claim didn’t hold up.
What does seem to matter is accuracy. Kids who have a realistic read on their own performance tend to do better in school. Parents whose praise tracks what actually happened tend to raise kids with steadier self-esteem. This is what is actually useful from the literature because it points to something concrete you can do, instead of a vague instruction to “be more positive.”
Helping a child notice the gap between what they expected and what actually happened (gently, without shame) is teaching them a real skill. That’s different from talking them into feeling capable regardless of the evidence.
It’s also worth saying plainly: this was one kind of task (a memory game) with one age group, so it’s not the final word on confidence in every part of a kid’s life. But it lines up with other research on adults and older kids showing that inflated self-belief tends not to translate into real gains - it just feels good in the moment.
Long Story Short
A famous study claiming overconfident kids learn better turned out to be a statistical illusion, created by comparing groups that were never equal to begin with. When researchers removed that flaw by randomly assigning kids’ confidence levels instead, the “benefit” disappeared completely. What actually seems to help kids is accurate feedback that helps them see clearly where they stand, not inflated belief in their own ability.
Quick Takeaways
A famous, widely taught study claiming overconfidence helps kids learn didn’t hold up when researchers tested it properly.
The original result was a statistical illusion, not a real effect. It came from comparing groups that were unevenly matched from the start.
A new study found that kids with an accurate sense of their own abilities did better in school than kids whose self-image ran too high or too low.
Inflated praise from parents has separately been linked to lower self-esteem in kids over time (the same pattern, from a different angle).
The takeaway isn’t “stop encouraging your kid.” It’s that specific, honest feedback tied to what actually happened does more good than generic confidence-boosting ever could.
Development Decoded is built on a simple belief: science should be clear, practical, and accessible to anyone who cares about kids. I write, research, and create resources outside of my day job and family life to make that happen. If you’d like to help fuel the work, you can now do so with one coffee at a time.
References
Brummelman, E., Nelemans, S. A., Thomaes, S., & Orobio de Castro, B. (2017). When parents’ praise inflates, children’s self-esteem deflates. Child Development, 88(6), 1799–1809. https://doi.org/10.1111/cdev.12936
Freeman, E. E., Karayanidis, F., & Chalmers, K. A. (2017). Metacognitive monitoring of working memory performance and its relationship to academic achievement in Grade 4 children. Learning and Individual Differences, 57, 58–64. https://doi.org/10.1016/j.lindif.2017.06.003
Hu, M., Zhao, W., Cao, M., Shanks, D. R., Hu, X., Luo, L., & Yang, C. (2026). Does overconfidence really confer adaptive benefits to children’s learning? Psychological Science, 37(7), 477–492. https://doi.org/10.1177/09567976261455283
Shin, H., Bjorklund, D. F., & Beck, E. F. (2007). The adaptive nature of children’s overestimation in a strategic memory task. Cognitive Development, 22(2), 197–212. https://doi.org/10.1016/j.cogdev.2006.10.001
Xia, M. T., Poorthuis, A. M. G., & Thomaes, S. (2024). Children’s overestimation of performance across age, task, and historical time: A meta-analysis. Child Development, 95(3), 1001–1022. https://doi.org/10.1111/cdev.14042
Isn’t there a distinction between confidence in a task and belief in yourself? I’d guess that kids who feel/know their worth, that they matter, that they are loveable and loved tend to do better than kids with low self esteem. And that this becomes a self fulfilling prophecy unless there’s an intervention.