New Books in Science Tom Griffiths, "The Laws of Thought: The Quest for a Mathematical Theory of the Mind" (Henry Holt and Co., 2026)
Feb 4, 2026
Tom Griffiths, cognitive scientist and head of Princeton’s AI Lab, blends history and math to track attempts to formalize thought. He covers three frameworks—symbols, neural networks, and probability. Stories range from Boole and Turing to Chomsky, neural learning rules, backpropagation, and the strengths and limits of large language models.
AI Snips
Chapters
Books
Transcript
Episode notes
Recovering Mind Geometry From Judgments
- Multidimensional scaling recovers people's psychological spaces from similarity judgments.
- Shepard used it to show internal color representation is a wheel, unlike the physical wavelength line.
Networks As Space Transformers
- Neural networks compute by transforming points between high-dimensional spaces.
- Learning adjusts weights to discover functions that map inputs to desired representations and actions.
Hebb Vs. Rosenblatt On Learning
- Hebb proposed 'neurons that fire together, wire together' as associative learning.
- Rosenblatt designed the perceptron and updated weights based on mistakes, not mere co-activation.
