KQED's Forum How Intelligence – Both Human and Artificial – Happens
Oct 22, 2025
Join Jay McClelland, a Stanford professor specializing in neural networks, and Gaurav Suri, a computational neuroscientist at San Francisco State, as they delve into how both human and artificial intelligence operate. They explore how simple, interconnected neural units create complex behaviors and the importance of top-down expectations in perception. The duo also discusses consciousness, the challenges of reading memories from brain scans, and AI's limitations in common sense. Together, they reveal how understanding these concepts can bridge the gap between minds and machines.
AI Snips
Chapters
Books
Transcript
Episode notes
Massive Connectivity Produces Emergence
- The brain is a massively connected network of simple units whose collective activity produces intelligence.
- Each neuron integrates many inputs and influences many others, enabling emergent cognitive functions.
Going To The Fridge Explained
- Gaurav Suri uses the simple fridge example to show how neurons for thirst link to action tendencies.
- The same activation chain explains both action and the story we later tell about why we acted.
Perception As Interactive Activation
- Perception arises from interactive, bidirectional activation between features, letters, and words.
- Top-down expectations and bottom-up inputs coalesce to stabilize interpretations of ambiguous input.
