From mind to machine: How human biology is informing new breakthroughs in AI

Neural networks have evolved from early cognitive science research into the foundation of modern artificial intelligence, demonstrating the unexpected ways that basic research into human cognition can lead to transformative technological advances.

Origins and early breakthroughs: The groundwork for today’s AI systems was laid in the late 1970s and early 1980s through NSF and Office of Naval Research funded projects exploring human cognitive abilities.

• James McClelland, David Rumelhart, and Geoffrey Hinton developed pioneering neural network models to understand human letter and word perception
• Their 1986 publications introduced parallel distributed processing theory and the revolutionary backpropagation algorithm
• This foundational work earned the team a 2024 Golden Goose Award for its profound impact on modern technology

Challenging conventional wisdom: The team’s approach to understanding cognition represented a significant departure from prevailing theories of the time.

• While researchers like Chomsky viewed language processing as rule-based symbol manipulation, McClelland’s background in neurophysiology suggested a more fluid, context-dependent process
• Their neural network model demonstrated how context influences letter recognition and word perception
• Hinton’s contribution moved the field away from discrete symbolic representations toward distributed neural activity patterns

Technical evolution: The development of the backpropagation algorithm marked a crucial turning point in neural network capabilities.

• Earlier models could only adjust errors in their final output layer, limiting their learning ability
• Backpropagation enabled error signals to flow backward through the network, allowing deeper learning
• Modern AI systems have expanded from simple networks to thousands of intermediate layers using this same fundamental principle

Brain-AI connections: Current research reveals fascinating parallels and differences between human cognition and artificial neural networks.

• Both human brains and AI systems demonstrate similar language processing capabilities and context-dependent reasoning
• Language models can now maintain contextual information similar to human working memory
• However, AI systems require significantly more training data than humans – approximately 100,000 times more to learn language

Future directions: Understanding the differences between biological and artificial neural networks is driving new research initiatives.

• The human brain’s ability to learn efficiently with minimal data remains a key area of investigation
• Researchers are exploring alternatives to backpropagation that better reflect biological learning processes
• The bidirectional, multi-sensory nature of brain processing offers potential insights for improving AI systems

Looking ahead: While artificial neural networks have achieved remarkable success, their biological counterparts still hold valuable secrets that could lead to more efficient and capable AI systems, highlighting the ongoing symbiotic relationship between cognitive science and artificial intelligence research.