Neuroscientists at Princeton University report that the brain achieves flexible learning by reusing modular cognitive components across tasks. In experiments with rhesus macaques, researchers found that the prefrontal cortex assembles these reusable “cognitive Legos” to adapt behaviors quickly. The findings, published November 26 in Nature, underscore differences from current AI systems and could eventually inform treatments for disorders that impair flexible thinking.
Researchers at Princeton University investigated why biological brains adapt to new tasks more effectively than many artificial intelligence systems. In a new study, they report that the brain repeatedly reuses shared neural patterns, or cognitive "blocks," to build complex behaviors rather than learning each task from scratch.
According to Princeton’s account of the work, published November 26, 2025, in the journal Nature, the team trained two male rhesus macaques to perform three related visual categorization tasks while recording brain activity.
In the tasks, the monkeys viewed colorful, balloon-like blobs on a screen and had to judge either whether each shape looked more like a bunny or the letter "T" (shape categorization) or whether it appeared more red or more green (color categorization). To indicate their choices, the animals reported their decisions by looking in one of four directions on the screen. In one task, for example, looking left signaled that the blob resembled a bunny, while looking right indicated that it looked more like a "T." Some images were clearly one category or another, while others were ambiguous and required finer judgment.
A key feature of the design was that each task had distinct rules but still shared elements with the others. One of the color tasks and the shape task required the same mapping between eye movements and choices, while both color tasks used the same rule for judging color (more red vs. more green) but required different eye-movement responses. This structure allowed the researchers to test whether the brain reused the same neural patterns – its cognitive building blocks – whenever tasks shared specific components.
Analysis of brain activity showed recurring patterns in the prefrontal cortex, a region at the front of the brain involved in higher cognition and decision-making. These patterns appeared when groups of neurons worked together toward shared goals, such as discriminating colors, and could be flexibly combined with other patterns to support different tasks.
“State-of-the-art AI models can reach human, or even super-human, performance on individual tasks. But they struggle to learn and perform many different tasks,” said Tim Buschman, Ph.D., senior author of the study and associate director of the Princeton Neuroscience Institute. “We found that the brain is flexible because it can reuse components of cognition in many different tasks. By snapping together these ‘cognitive Legos,’ the brain is able to build new tasks.”
Buschman compared a cognitive block to a function in a computer program: one set of neurons might determine the color of an image, and its output can then feed into another block that guides an action such as a particular eye movement. For one of the color tasks, for instance, the brain assembled a block that evaluated color with another that controlled gaze direction. When the animal switched to judging shapes while using similar eye movements, the brain instead combined a shape-processing block with the same movement block.
Lead author Sina Tafazoli, Ph.D., a postdoctoral researcher in the Buschman lab, said the prefrontal cortex also appeared to suppress irrelevant blocks, helping the animals focus on the current goal. “The brain has a limited capacity for cognitive control,” Tafazoli said. “You have to compress some of your abilities so that you can focus on those that are currently important. Focusing on shape categorization, for example, momentarily diminishes the ability to encode color because the goal is shape discrimination, not color.”
The researchers interpret this compositional organization – assembling new behaviors from reusable neural components – as a key reason why humans and other animals can learn new tasks so rapidly. By contrast, many machine-learning systems suffer from “catastrophic interference,” in which acquiring a new skill overwrites older ones. “When a machine or a neural network learns something new, they forget and overwrite previous memories,” Tafazoli said.
According to Princeton’s report and related coverage of the study, understanding how the brain reuses and recombines these cognitive blocks could help engineers design AI systems that learn new tasks without erasing prior knowledge. The same principles may eventually guide clinical approaches for conditions such as schizophrenia, obsessive-compulsive disorder and some forms of brain injury, in which people often struggle to shift strategies or apply existing skills in new contexts.
Funding for the research was provided by the U.S. National Institutes of Health, including grants R01MH129492 and 5T32MH065214.
