Creating Intelligence: A Computational Foundation for AGI

2026-06-30 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Expert, quick

Summary

A new computational theory of mind, "Creating Intelligence: A Computational Foundation for AGI," grounds artificial general intelligence in set theory and hyperdimensional computing, departing from traditional neural networks' continuous weights. This framework uses sparse binary data to represent information as discrete sets, directly modeling biological neural population codes. Associative memory naturally emerges from network topologies featuring a combinatorially expanded hidden layer, with learning driven by topological plasticity rather than scalar weight adjustments. The core algorithm unifies auto-associative and hetero-associative learning through information retrieval via subset pattern matching and exact nearest-neighbor search, operating with constant-time complexity. This mechanism bridges perceptual data and symbols and is proposed to be implemented by the cerebellum and neocortex, translating directly into in-memory hardware for human-level energy efficiency.

Key takeaway

For AI Scientists exploring novel AGI architectures, this work suggests a fundamental shift from continuous weights to discrete logic. You should investigate set-theoretic models and topological plasticity as a path to human-level energy efficiency. This approach, leveraging in-memory hardware, offers a promising alternative to traditional neural networks for building synthetic intelligence.

Key insights

A computational theory of mind grounds AGI in discrete set theory and topological plasticity for efficient, biologically-inspired cognition.

Principles

• Information represented as discrete sets models biological neural population codes.
• Associative memory emerges from combinatorially expanded hidden layers.
• Cognition fundamentally relies on subset pattern matching.

Method

Information retrieval via subset pattern matching and exact nearest-neighbor search, driven by topological plasticity, unifies associative learning.

In practice

• Design networks with combinatorially expanded hidden layers for associative memory.
• Implement cognitive algorithms using discrete logic for in-memory hardware.
• Explore sparse binary data representations for neural models.

Topics

• Computational Theory of Mind
• AGI
• Set Theory
• Hyperdimensional Computing
• Topological Plasticity
• In-Memory Hardware

Best for: AI Scientist, Research Scientist, AI Hardware Engineer

Related on AIssential

• A Dynamical Theory of Sequential Retrieval in Input-Driven Hopfield Networks — A two-timescale IDP Hopfield network enables analytically tractable sequential memory retrieval with explicit transition conditions.
• The Thinking Machine That Google and Anthropic will Ship Together — True AI cognition necessitates a new mathematical blueprint, moving beyond current LLM scaling and flat embeddings.
• Asynchronous AI cuts computing energy by orders of magnitude while learning continuously — Asynchronous Neural Turing networks (ANT) significantly cut AI energy by mimicking brain's selective neuron updates.
• The Mathematical Foundations of Intelligence [Professor Yi Ma] — Intelligence can be formalized through parsimony and self-consistency, enabling principled AI architecture design.
• Scaling Neuro-symbolic Problem Solving: Solver-Free Learning of Constraints and Objectives — A neuro-symbolic architecture with a probabilistic loss efficiently learns to solve NP-hard problems from natural inputs.
• Working Paper: Towards a Category-theoretic Comparative Framework for Artificial General Intelligence — Category theory offers a formal, algebraic framework to compare and analyze diverse AGI architectures structurally.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.