What intelligence needs

Intelligence did not emerge as a luxury, it emerged as a survival mechanism. For any creature in a complex and uncertain universe, survival depends on understanding the environment and shaping it to its advantage:

• To understand the environment means to detect regularities—patterns that reveal how the world works.
• To shape the environment means to use those regularities to invent strategies, tools, and behaviors that improve the odds of survival.

If you were to design an intelligent creature from scratch, with no prior exposure to earth’s species, what cognitive mechanisms would you build into it so it could adapt, survive, and dominate its environment?

This post argues that intelligence, at its core, requires only three fundamental capacities: association, abstraction, and analogy. These three A’s constitute the minimal architecture of general intelligence.

1. Association: Detecting patterns in experience

The universe is full of patterns and structures: things happen not randomly but according to causal or statistical regularities. An intelligent agent must first be able to associate experiences, linking sensory inputs, actions, and outcomes into coherent patterns, so that next time when they face similar situations, they know how to respond properly.

There would be no learning without association. The creature would constantly experience the world as a chaotic stream of disconnected events. Association allows the creature to compress its experiences - linking stimuli and fragments of experiences with known consequences and patterns - to predict future events.

2. Abstraction: Generalizing beyond the particular

However, association alone is insufficient. No two situations are ever exactly the same. If a creature can only react to previously seen patterns, it will fail when faced with novelty. Abstraction solves this.

Abstraction filters out superficial details and preserves invariant structures and properties of events across a variety of situations, extracting the underlying principles shared across different experiences. By clustering or representing experiences at a higher level, the creature builds mental models of the world that can handle variation and uncertainty.

Abstraction is the leap from memory to model, the key to flexible, adaptive behavior.

3. Analogy: Applying knowledge to the unknown

However, even abstraction isn't enough on its own. To thrive, a creature must apply what it has learned to novel situations. This is the role of analogy-mapping patterns from past experiences to new contexts.

Survival depends on acting effectively in unfamiliar scenarios. Early humans used the behavior of falling stones to design tools or weapons. An engineer applies fluid dynamics to design airplane wings by analogy to bird flight. Analogy activates relevant abstractions in response to novel inputs, allowing rapid transfer of knowledge.

Analogy is the engine of innovation and reasoning. It lets intelligence extend its reach beyond direct experience.

Why are these three sufficient (and necessary)?

These three A's are not arbitrary cognitive skills. They constitute a minimal and recursive architecture that enables survival in a universe woven from patterns yet filled with uncertainties:

- Association organizes raw sensory input into structured experiences.

- Abstraction distills general principles from those experiences.

- Analogy reuses and recombines these principles in novel contexts.

Together, they give rise to what we call "intelligence", the ability to adapt and act in ways that maximize survival, whether in an animal brain, an alien organism, or an artificial system. Even advanced cognitive behaviors like language, planning, and creativity can be understood as complex compositions of these three foundational abilities.

Importantly, the boundaries among the three A's should not be treated as rigid nor hierarchical. Rather, they should be seen as interwoven processes within a single dynamic system. For example, analogy depends on associating one situation with another, while abstraction itself involves analogical comparison between events to extract shared patterns. The distinction among the three A's is thus a conceptual tool, not a literal separation. Intelligence in practice is always a continuum of inseparable processes.

In cognitive science, there are plenty of discussions on the architecture of intelligence and its necessary components. Some may argue that intelligence also requires elements like memory, planning, communication etc. While these are indeed crucial aspects of human cognition, they are higher-level behaviors that either derive from or depend on the three A's:

- Memory is a substrate that enables association and abstraction.

- Planning is analogy applied over temporal sequences.

- Language is a cultural system built upon abstraction and analogy.

This makes the three A's both necessary (every intelligent system must have them) and sufficient (they're minimal to generating higher functions).

The "simplicity" of intelligence

Intelligent behavior is often seen as something deeply complex. Yet, the fundamental requirements of any intelligent mechanism are surprisingly simple. A being that can associate experiences, abstract patterns, and draw analogies across contexts possesses the essential ingredients for adaptation, innovation, and survival. Whether we seek to understand the origins of human intelligence, design artificial intelligence, or imagine alien minds, these three A’s offer a unifying framework for what it truly means to be intelligent.

This leads to a deeper question: If we want to create an intelligent system, what would it take to model these core features properly, and why has it been so difficult to achieve?