The study built on years of research into how learning disabilities affect children's brains. By using new AI tools, the team was able to explore these problems in greater detail. The study created "digital twins" — AI models that copy the brain activity of real students as they solve math problems.
The researchers tested 45 students, aged 7 to 9, including 21 with math learning disabilities. While the students solved basic addition and subtraction (减法) problems, their brain activity was tracked by a brain scanning machine. The AI models, or "digital twins", were then trained to copy the students' answers and brain activity.
The results were surprising. It turned out that the problem wasn't too little brain activity, as many had thought, but too much. The students with learning disabilities showed signs of too much excitement in brain areas connected to math. The digital twins showed the same patterns, which means that too much neural activity could cause confusion in solving math problems.
The study suggests that students with learning disabilities may need more training to better their performance. The researchers hope that their "digital twins" can be used to create personalized learning plans for students, helping to design more effective educational programs.
However, a researcher said that we cannot overread the results. The model needs refinement. There is more work to do, but it does point in some promising new directions for further research.
"The study makes us have a system to test targeted strategies before trying them in real classrooms," Vinod Menon, a professor at Stanford, said. "That can speed up the ability to design effective educational programs and make real progress for kids with math learning disabilities."
