Take a moment to think about this. There are two kids inside an MRI machine. Two numbers are shown on a screen to both of them. Both of them press the right button. They both get it right. But something very different is going on inside their heads.
That is the main point of a study from Stanford Medicine that was published in the Journal of Neuroscience. It sounds like a simple finding, but it makes us think about how we think about kids who have trouble with math in a really uncomfortable way.
The study looked at 87 kids in second and third grade. Based on standard fluency scores, 34 of them were found to have a math learning disability. Kids with math learning disabilities did about the same as their peers on a simple task that asked them to compare two numbers and pick the bigger one. The brain activity, on the other hand, told a completely different story.
Kids who were learning normal math skills took longer to solve harder problems. They changed after making mistakes. Kids who had trouble learning math were much less likely to do any of those things, especially when working with Arabic numbers instead of groups of dots. It wasn’t firing the same way as the brakes, which are like an internal pause that says, “Wait, something went wrong.” The middle frontal gyrus and the anterior cingulate cortex were less active on MRI scans. These are parts of the brain that are involved with executive function, focused attention, and finding mistakes.
The difference between a dot and a number adds depth to this finding. When math problems were shown as groups of dots, kids who had trouble learning math were more careful after making mistakes, not less. So it’s not that these kids can’t fix their mistakes on their own. It’s a more specific problem—one that seems to have to do with processing symbolic numbers and the abstract language of math. The lead author of the study, Vinod Menon, put it this way: “They can tell the difference between five and ten dots with no trouble, but when you ask them to reason with the number itself, something changes.”
Menon clearly talks about how this has a cascading quality to it. When a child is having trouble and doesn’t know why, they lose motivation. They get nervous when they have to solve problems. The gap between them and their classmates grows slowly. What began as a difference in their brains turns into an emotional and academic burden that they carry into every classroom.
Right here is where things get tricky for AI tutoring tools. At the moment, most edtech platforms work with feedback loops that are based on right and wrong answers. If a child gives the wrong answer, the system lets them know and gives them the chance to try again or get the answer fixed. A few tools keep track of speed and repetition. But Stanford’s research shows something that simple answer checking can’t find: a child whose process isn’t aligned even when their output is right. What needs attention is the mechanism for making changes, the metacognitive layer that watches performance in real time. And an algorithm that is based on being correct can’t easily see that.
There’s a chance that AI tutoring tools in the future will get better at this. It’s getting better for adaptive learning systems to keep track of error patterns and response times across sessions. But making something that can watch not only what a child says but also how they think is a whole other level of difficult. It takes the kind of careful observation that a trained teacher develops over months with a student—noticing, for example, that a certain child always works faster than slower on hard problems and never seems to stop and think about their answer, even when it seems uncertain.
The study suggests that interventions that directly work on metacognitive skills can help kids develop habits of self-monitoring, being aware of mistakes, and changing their strategies. That’s not an app for flashcards. There is a relationship there, and it takes time.
The quiet weight of the word “hidden” is what stays with me after reading this research. For years, these kids have been getting the answers right. They don’t always show how hard things are for them. And if you only look at results to see how much someone has learned, you might not notice what is wrong until the gaps get too big to ignore.
