Breakthrough in Understanding How Brains Learn: A Different Approach Than AI

Researchers have discovered a fundamental difference in how the human brain learns compared to artificial intelligence systems. This breakthrough could lead to faster, more efficient learning algorithms in AI and a deeper understanding of how the brains learn.

The study, published in Nature Neuroscience, focused on how the brain adjusts connections between neurons during learning. Unlike AI, which relies on trial-and-error adjustments, the brain seems to take a more strategic approach.

The Key Difference between how AI and human brains learn: Prospective Configuration

Instead of directly modifying connections based on errors, the brain first establishes a balanced activity pattern among neurons. This “prospective configuration” acts as a foundation for learning, preventing interference between existing knowledge and new information.

Benefits of the Brain’s Approach:

• Faster Learning: This method allows the brain to learn new things more quickly and efficiently compared to AI’s error-correction approach.
• Preserved Knowledge: New information doesn’t disrupt existing knowledge, unlike in some AI systems.
• Robustness: The brain’s learning process is less prone to errors and distractions.

Implications for AI and Brain Research:

• Developing Faster AI Algorithms: Understanding the brain’s learning principles could inspire new AI algorithms that learn faster and more effectively.
• Bridging the Gap Between AI and Brains: Researchers aim to translate the abstract principles of prospective configuration into real-world applications for AI.
• Understanding the Brain: This discovery could shed light on how the brain processes information and learns, potentially leading to new insights into neurological disorders.

Challenges Ahead:

• Simulating Prospective Configuration in AI: Current computers struggle to implement this method efficiently, requiring new hardware or specialised algorithms.
• Bridging the Gap to Real Brains: Understanding how the brain implements prospective configuration in its complex neural networks is a major challenge.

This research marks a significant step forward in understanding how the brain learns and has the potential to revolutionise both artificial intelligence and neuroscience.

Written by
LarsGoran Bostrom©