The MRC’s Brain Network Dynamics Unit, in collaboration with Oxford University’s Department of Computer Science, recently announced a significant discovery in neuroscience. The discovery was published with title “A study shows that the way the brain learns is different from the way artificial intelligence systems learn.” The Researchers have identified a new brain learning principle called “prospective configuration,” offering insights into the superior learning mechanism of the human brain compared to artificial intelligence (AI) systems.
Understanding Learning: The Human Brain vs. AI
Traditional AI training, mainly based on back-propagation, adjusts model parameters to minimize errors in the output. This process differs sharply from the newly discovered method of training the brain. The human brain exhibits an extraordinary capacity to rapidly assimilate new information while retaining pre-existing knowledge, a feat that AI systems have yet to achieve. These abilities have motivated researchers to investigate the underlying principles of brain learning.
The concept of “prospective configuration”
The principle of “prospective configuration” suggests that the human brain optimizes neuronal activity in a balanced state before adjusting synaptic connections. This approach minimizes interference between new and existing information, increasing learning efficiency. Computational models using this principle have been shown to learn more efficiently and faster than current AI models in various simulations, outperforming tasks faced by animals and humans in natural settings.
Future research and implications
The research team, led by Professor Rafal Bogac and Dr. Yuhang Song, recognized the difference between abstract learning models of the brain and detailed anatomical knowledge. Future studies aim to understand how “prospective configuration” is implemented in specific brain networks. Furthermore, the simulation of this principle in machine learning faces challenges due to current computational limitations, suggesting the need for innovative computational technologies or dedicated brain-inspired hardware for efficient and low-power implementation.
Conclusion
This important discovery of the “prospective configuration” learning principle in the human brain not only enriches our understanding of neural processes, but also holds significant potential for advancing AI technology. It offers a new direction for AI research aimed at developing learning algorithms that mimic the efficiency and adaptability of the human brain.
Image source: Shutterstock