The Princeton study shows that brains reuse modular neural components—"cognitive Legos"—to build and adapt to new tasks. Recording from rhesus macaques, researchers found these modules concentrated in the prefrontal cortex and observable as reduced activity when unused. This modular organization enables transfer learning and helps avoid catastrophic forgetting, offering insights for more flexible AI and therapies for cognitive disorders.
How 'Cognitive Legos' Let Brains Outperform AI — New Princeton Study
human brain pink background
Despite rapid advances in artificial intelligence, new research shows the biological brain still has a clear advantage when it comes to transferring skills across tasks. A Princeton-led study found that the brain builds new behaviors by recombining reusable neural components the authors call "cognitive Legos."
Study Design
Researchers recorded neural activity from rhesus macaques (Macaca mulatta), whose brain organization closely resembles humans'. The animals completed three related tasks that required discriminating shapes and colors and indicating their answers by looking in particular directions. While the monkeys performed these tasks, scientists scanned their brains to identify overlapping activity patterns and reusable neural modules.
YouTube Thumbnail
Key Findings
The scans revealed distinct clusters of neurons that the team describes as modular building blocks — "cognitive Legos" — which the brain can snap together and reconfigure to perform new tasks. These modules were concentrated in the prefrontal cortex, a region associated with planning, decision-making, and higher cognition.
"State-of-the-art AI models can reach human, or even super-human, performance on individual tasks," said Princeton neuroscientist Tim Buschman. "But they struggle to learn and perform many different tasks. We found that the brain is flexible because it can reuse components of cognition in many different tasks."
The researchers also observed that when a module was not needed for a given task, its activity dropped — suggesting the brain can downregulate or "shelve" components to sharpen focus on the functions required at the moment.
Subscribe to ScienceAlert's free fact-checked newsletter
Why This Matters
This modular reuse helps explain two major features of biological learning: rapid transfer of skills to new situations and resistance to catastrophic forgetting (the tendency of many artificial neural networks to lose previously learned behaviors when trained on new tasks). By recombining preexisting modules rather than overwriting them, the brain can adapt quickly without sacrificing earlier learning.
Implications For AI And Medicine
Understanding how the brain organizes and reuses computation could guide the design of more adaptable AI that better supports continual learning. The findings may also inform treatments for neurological and psychiatric disorders in which patients struggle to apply learned skills across different contexts.
The work was published in Nature.
