While AI systems excel at performing individual jobs with precision, the human mind remains unmatched in its ability to transfer knowledge and adjust to new challenges, thanks to a system of mental flexibility researchers are now calling “cognitive legos.”
Led by neuroscientist Tim Buschman at Princeton University, the study didn’t test humans directly but instead relied on a close biological proxy—rhesus macaques (Macaca mulatta). These monkeys were asked to complete tasks involving shape and color identification on a screen, using their gaze to respond. While they performed the tasks, scientists monitored their brain activity to investigate how different areas of the brain worked together during learning and adaptation.
Brain Scans Reveal Modular Thinking In Action
The Princeton team used brain imaging to observe what happened inside the monkeys’ heads as they moved through a sequence of three related cognitive tasks. According to ScienceAlert, the brain scans showed that the monkeys’ brains reused the same blocks of neurons—described by the researchers as “cognitive legos”—in different combinations depending on the task at hand.
These building blocks were found to be concentrated in the prefrontal cortex, the brain region associated with high-level cognition, such as planning, decision-making, and problem-solving. The researchers discovered that the brain would selectively activate these blocks as needed, while reducing activity in blocks that weren’t currently useful. This suggested that the brain doesn’t just build new behaviors from scratch—it reconfigures existing mental components efficiently.
“We found that the brain is flexible because it can reuse components of cognition in many different tasks,” said Buschman. “By snapping together these ‘cognitive legos’, the brain is able to build new tasks.”
Monkeys Show How Flexible Thinking Works
In the experiment, the macaques had to discriminate between shapes and colors over three separate but related tasks, all of which required them to apply knowledge from one task to the next. This setup tested whether they could adapt previously learned information to new situations—something modern AI still struggles with significantly.
As the monkeys progressed, the scans revealed overlapping patterns of neural activity, indicating that existing cognitive blocks were being reused and repurposed in real time. According to the study published in Nature, this reusability supports a kind of mental flexibility that artificial systems haven’t replicated yet.
Buschman compared the process to how functions work in computer programming: one set of neurons might handle color discrimination, while another might use that output to drive a physical response. “That organization allows the brain to perform a task by sequentially performing each component of that task,” he explained.
AI Still Suffers From “Catastrophic Forgetting”
Despite major improvements in AI, including the ability to achieve superhuman performance on specific tasks, machine learning models still suffer from catastrophic forgetting—a condition where learning a new task can erase the memory of a previous one. This makes it difficult for AI to handle multiple tasks consecutively without retraining.
The study’s findings underscore the current limitations of AI systems when compared to biological intelligence. The human brain, and by extension the monkey brain, has evolved a system that enables it to adapt rapidly to new environments by retrieving previously learned components from long-term memory or through reward feedback.
The researchers also believe that their work could eventually inform training strategies for AI to become more adaptable, and even contribute to the development of treatments for neurological or psychiatric disorders that involve difficulties in transferring knowledge to new settings.
For now, though, this research confirms one thing: when it comes to learning across tasks, the human brain—flexible, modular, and efficient—is still ahead of the smartest machines.