Agent-Level Mapping of Intelligence Types in the Algorithmic Information Theory of Consciousness

A short reference note that slots classical intelligence taxonomy into a formal theory of mind, showing how four cognitive faculties interlock to make a complete agent.

The Kolmogorov Theory (KT) of consciousness treats the mind as a compression engine: a system that builds compact, generative models of the world and uses them to predict, plan, and act. This note asks a simple but useful question — if you accept that framework, where do the familiar categories of human intelligence fit inside it?

The answer is a clean four-way decomposition. Crystallised intelligence (Gc) is the stored model itself — all the learned structure about the world and the body, accumulated slowly over time. Fluid intelligence (Gf) is the mechanism that revises that model when predictions keep failing — it's the induction engine, running structure-learning and hypothesis generation on the fly. Executive function (EF) is the planner that reads the model and decides what to do next, operating on the fast timescale of seconds. Emotional intelligence (EI) is the piece that grounds everything in bodily states and social value — it supplies the cost functions that tell the system what actually matters, shaping both what gets learned and what gets done.

What makes this more than a labelling exercise is the dependency structure the note spells out. Planning (EF) is useless without a model to plan over, so EF depends on Gc. Model updates (Gf) need to know which errors are worth fixing, so Gf depends on EI for prioritisation signals. And EI without a world model is just raw sensation with no context. The four faculties aren't parallel modules — they form a loop, with each one needing the others to function.

The note also highlights a hierarchy of timescales: Gc changes over years, Gf over minutes to hours, EF over milliseconds to seconds, and EI fluctuates with interoceptive cycles but also accumulates into long-term utility functions. This layering is not incidental — it's what lets a slow, stable knowledge base coexist with fast, flexible action.

This is a short working note, essentially a structured reference table with commentary, rather than a full empirical or theoretical paper. Its value is organisational: it gives the KT framework a clean vocabulary for discussing agent cognition and points toward testable operationalisations (Raven's matrices for Gf, Wisconsin Card Sorting for EF, and so on). The source does not develop new formal results, but it does provide a useful conceptual scaffold for future work in the KT programme.