Animal Symbolicum

Animal Symbolicum (Cassirer)

Ernst Cassirer’s An Essay on Man (1944) proposes that the classical definition of the human — animal rationale, the rational animal — misses the point. Reason is one capacity among many. What actually distinguishes humans is the capacity for symbolic representation.

The Argument

Other animals perceive and react. Humans perceive, symbolize, and then react to the symbols. We don’t live in a raw physical world — we live in a symbolic universe. Language, myth, art, religion, history, science: these aren’t tools for describing reality. They’re the medium through which we constitute reality.

A sunset isn’t just photons hitting retinas. It’s a word (“sunset”), a painting tradition, a metaphor for endings, a scientific process (atmospheric refraction), a religious symbol. We never encounter the raw thing. We encounter our symbols of it.

This isn’t a limitation — it’s the distinctively human form of being. The symbolic is not a veil between us and reality. It is our reality.

Symbolic Forms

Cassirer identified several autonomous “symbolic forms” — each with its own logic, its own way of organizing experience:

  • Language: not labels for pre-existing concepts, but the medium in which concepts become possible
  • Myth: not primitive science, but a distinct way of experiencing causation, identity, the sacred
  • Art: not decoration, but a way of seeing that reveals aspects of reality other forms miss
  • Science: not the truth behind appearances, but one symbolic form among others — distinguished by method, not by privileged access to the real

None is reducible to any other. Science can’t replace myth. Art can’t replace language. Each opens a different dimension of experience.

Relevance to This Vault

The vault asks: is AI an animal symbolicum?

LLMs are trained entirely on symbols — language, code, mathematical notation. They operate in the symbolic medium. But do they constitute reality through symbols, or do they manipulate symbols without constituting anything?

Meaning Making Machines draws on Cassirer directly: the compulsion to attach significance to patterns, to wrap raw data in narrative, to make symbols out of everything — is this a distinctly human trait, or does it emerge from sufficient exposure to symbolic systems?

Open Source as Counter-Power invokes Cassirer’s point that symbolic systems are never neutral. Controlling the symbols controls the reality. Proprietary AI is a symbolic form owned by someone.

If Cassirer is right that we are fundamentally symbol-makers rather than rational calculators, then AI trained on symbols is closer to human cognition than AI trained on logic — which is exactly the opposite of what the classical AI tradition assumed.

See Also