Mathematical Theory Of Intelligence
Mathematical theory of intelligence is Nina Miolane’s term in Claire Isabel Webb & Nina Miolane: The Geometry of Consciousness for a research program that seeks equations explaining how brains and artificial neural networks represent and act in the world. The source frames the project as analogous to moving from observed planetary orbits to laws that explain why those orbits appear.
Key Claims
- A theory should explain data and make predictions for future experiments, not only visualize existing recordings.
- The comparison between brains and machines should happen at the level of computation and algorithm, not biological substrate.
- Neural Geometry offers one mathematical language for describing collective activity in high-dimensional neural systems.
- Spatial Navigation Torus and Fourier Spatial Encoding are the talk’s concrete example of a geometric pattern and candidate explanation.
- The theory remains incomplete in more complex domains such as social behavior, affect, and Consciousness Measurement.
Connections
- Nina Miolane - researcher presenting the framework.
- Neural Geometry and Population Coding - data representation layer that makes the theory possible.
- Spatial Navigation Torus and Fourier Spatial Encoding - worked example and explanatory hypothesis.
- AI Interpretability By AI and Representation Learning - adjacent AI branches concerned with explaining learned representations.
- World Models - related AI direction where representations must support prediction and action in an environment.