The brain doesn't learn every task from scratch: it transfers experience from one activity to another. But how do you know which tasks are truly related? Neuroscientists came up with a clever trick: let an artificial network "peek" at brain activity during one task and try to predict activity during another. If it works well, the tasks are close. If not, the brain engages completely different mechanisms.
It turned out that motor tasks (those involving moving hands or feet) form a tight mutual-aid circle. They transfer skills to each other excellently but barely help, say, solving math problems or recognizing faces. This is like athletes: a basketball player can easily pick up volleyball, but that won't speed up learning chess. Movement and thought live in different "neighborhoods" of the brain.
Then the researchers moved from pairwise comparisons to a more realistic scenario: when resources are limited and you need to choose just a few tasks for direct training to then handle the rest. Using Boolean integer programming (a method of enumerating options with clear yes/no conditions), they sifted through possibilities and found an unexpected leader — working memory tasks, where you need to hold several items in mind and respond to specific cues. These most often made it into the "golden set."
This approach resembles scheduling classes at school: instead of teaching everything, you can single out a few key subjects that will pull the rest along. The study shows for the first time how experience transfer works not just in pairs but across an entire network of tasks, and which nodes in that network are most valuable.
Tasks involving finger, tongue, or foot movements barely help with emotion or arithmetic tasks, even though within their own "motor family" they support each other perfectly, like siblings who speak the same language but don't understand their neighbors.