Partitioning may actually help your brain learn faster

Scientists who recorded 90,000 neurons also found that minds wander aimlessly through the environment, creating internal maps that accelerate future learning when tasks emerge.

When you “just walk around” or explore aimlessly, your brain may be harder than you think. Researchers at Janelia Research Campus of HHMI found that seemingly aimless mental wandering actually constructs detailed internal maps of the world that can significantly speed up learning when concrete tasks appear later.

This discovery challenges traditional wisdom about how the brain learns. By simultaneously recording the activity of thousands of neurons, scientists found that even without training, reward or specific targets, visual cortical neurons even encode environmental features, the process prepares animals to learn new tasks faster.

Unconscious learning

The study began with neuroscientist Lin Zhong noting something unexpected while studying mice in the virtual reality corridor. Some visual textures in the corridor are associated with the rewards, others do not. But even if the reward is completely eliminated, the animal’s brains show significant plasticity—changing the usual signal learning in neural connections.

“As we think more and more about this, we ended up exploring the question of whether the task itself is necessary,” explained Marius Pachitariu, head of the group. “It’s entirely possible that a lot of plasticity happens essentially through animal exploration of the environment.”

When the team explicitly tested this hypothesis, they found different brain regions that deal with different types of learning. During passive exploration, some areas of the visual cortex encode visual features, while others are activated only when a specific task is introduced.

Two parallel learning systems

Research shows that the brain runs two learning algorithms at the same time:

• Unsupervised learning extracts without external descriptions
• Supervised learning assigns meaning and goals to those preexisting patterns
• The medial visual area is specialized in unsupervised function detection
• Pre-brain area processing reward prediction and target-oriented response

It took weeks to explore the mice in the virtual corridor without any training to connect texture to rewards much faster than animals trained only for a specific task. The difference is dramatic – Exploring mice mastered the discrimination task, making it take other people longer to learn.

“I was surprised,” said Zhang, the lead author of the study. “I have been doing behavioral experiments since my PhD, and I never expected that mice were not trained to do a task and you would find the same neuroplasticity.”

Psychological diagram without a teacher

The team used advanced visualization tools they developed, including raster diagrams, to analyze patterns across large numbers of neural datasets. They found that the visual cortical region is actively building internal environmental models during passive exploration, laying the foundation for future rapid task learning.

“Even if you want to partition or just walk around, or you think you are doing anything special or hard, your brain is still trying to help you remember where you are and organize the world around you,” Pachitariu explained.

This unconscious preparation proves invaluable. When the concrete target was later introduced, animals with exploration experience showed accelerated learning compared to animals without such background exposure.

What is outside the laboratory

The results of the study show that seemingly ineffective periods of exploration (carrying through cities, browsing new environments or random observations) do actually have important cognitive functions. These activities may be building neural scaffolding that can support faster learning when concentration is required.

The study also reveals why different brain regions specialize in different learning processes. Although the medial visual area excels in extracting patterns from sensory experiences, the front area focuses on connecting these patterns to rewards and goals.

Zhang concluded: “This means you don’t always need a teacher to teach you: You can still learn your environment without knowing it, and this kind of learning can prepare you for your future.”

This study opens new research directions for how unsupervised and supervised learning systems interact and how environmental exploration can integrate with the spatial navigation networks of the entire brain. Understanding these processes can inform educational methods and cognitive training methods, thereby leveraging the brain’s natural abilities to learn unconsciously.

Related