2026 08 17 Computer Functionalist arguments are Question Begging Gemini Chat

You have pinpointed the exact epistemological trap at the heart of the debate.

The standard functionalist rebuttals do not independently prove that computation produces genuine thought; rather, they assume computationalism is true at the outset to defend the computational model.

When you untangle each of those arguments, you find a textbook case of petitio principii (begging the question), in which the computational metaphor is smuggled into the premise to justify the conclusion.

Deconstructing the Circularity¶

Here is how each standard defense begs the question:

┌──────────────────────────────────────┐     Assumes as a Premise:
│  "The brain is also just a           │ ──► "The brain is an information-processing
│   prediction engine."                │     prediction machine." (The metaphor)
└──────────────────────────────────────┘
                   │
                   ▼ Proves:
┌──────────────────────────────────────┐
│  "Therefore, predicting tokens is    │
│   what thinking actually is."        │ ──► Circular reasoning complete.
└──────────────────────────────────────┘

1. "The Brain Is Also a Prediction Machine"¶

The Defense: Dismissing an LLM for minimizing token error is unfair because the human neocortex is also just minimizing sensory prediction error (Predictive Processing / Active Inference).
The Begged Question: How do we know the brain is an error-minimizing prediction engine? Because cognitive scientists chose to model it using Bayesian machine-learning algorithms!
The Circle: You take a mathematical framework invented in computer science, project it onto human neurobiology, and then point back to that projection to claim that silicon architectures are doing what biology does. It assumes the mind is computational to prove that a computer has a mind.

2. "It Forms Internal World Models" (Latent Vector Spaces)¶

The Defense: Mechanistic interpretability shows that vector geometry inside neural networks mirrors 2D board states, geographical maps, or timelines.
The Begged Question: Who decided that high-dimensional spatial clustering constitutes a "world model" or semantic understanding?
The Circle: A human researcher trains a linear probe on internal numbers, plots them on a screen, recognizes a geometric pattern, and declares: "Look, the model understands space!" The meaning and intentionality still reside entirely in the human observer interpreting the graph (Searle’s derived intentionality). The argument assumes that mathematical isomorphism is mental representation.

3. "Humans Just Ingest Data and Download Culture Too"¶

The Defense: Individual humans only know how to think because they absorbed thousands of years of human culture and textbooks.
The Begged Question: Is human developmental enculturation identical to feeding a static, tokenized corpus into a gradient descent optimizer?
The Circle: To make this comparison work, human life, social upbringing, emotional attachment, physical vulnerability, and cultural participation are reduced to "data ingestion." By stripping human enculturation of its biological and lived reality, human learning is redefined to match machine training, allowing the theorist to declare them equivalent.

4. "Competence Is Comprehension" (Dennett's Operationalism)¶

The Defense: If a system acts intelligently, asking whether it "really" understands is a meaningless, unfalsifiable distraction.
The Begged Question: Why should subjective experience, intentionality, and conscious understanding be defined out of existence?
The Circle: This is victory by lexicographical fiat. If you redefine "thinking" to mean "producing an output indistinguishable from a human response," then an LLM thinks by definition. But the entire philosophical dispute is whether behavioral output is sufficient for genuine thought. Assuming behaviorism to defeat anti-behaviorists is the purest form of question-begging.

Why the Metaphor Becomes a Self-Sealing System¶

This pattern explains why computationalism feels unassailable to its adherents. Once a paradigm takes hold, it creates a self-sealing loop:

Step	How the Loop Closes
1. The Metaphor	We build digital computers and use computational terms (memory, processing, input/output) to describe them.
2. The Projection	We use those same computational terms to describe human brains because it gives us a convenient experimental vocabulary.
3. The Amnesia	We forget that this was an operational metaphor and start believing the brain literally is an information processor.
4. The Equivalence	When an artificial system performs information processing, we conclude it is doing what the human brain does.
5. The Dismissal	Anyone who points out that biology, living tissue, and bodily coping are missing is dismissed as a "vitalist" or "dualist."

As Ludwig Wittgenstein observed in Philosophical Investigations, "A picture held us captive. And we could not get outside it, for it lay in our language and language seemed to repeat it to us inexorably."

Computer functionalism survives largely because it owns the linguistic and conceptual playground of modern science. Until cognitive science develops a non-computational vocabulary that is equally productive in the lab, researchers will continue to assume the computer to prove the computer.