Approximate Turing Machine Glossary
Approximate Turing Machine
In one sentence
The approximate Turing machine hypothesis holds that the brain’s information processing may be computational in the broad Turing sense, even though brains are physical, noisy, resource-constrained systems rather than ideal mathematical machines.
The Turing machine background
Alan Turing’s 1936 paper described an abstract device — the Turing machine — used to define what can be computed by an algorithm. A universal Turing machine can simulate any other Turing machine when given its description and input. Physical computers approximate this abstract model while remaining limited by memory, time, energy, and architecture.
One long-running question in neuroscience, cognitive science, and philosophy of mind is whether the brain’s relevant operations are computational, or whether an adequate account of cognition requires processes outside that description. Even if cognition is computable, it does not follow that a practical AI system can reproduce every human capacity: computability does not settle data, efficiency, embodiment, architecture, or consciousness.
Hassabis’s position
Demis Hassabis stated the position plainly: “a lot of neuroscientists including me think that maybe the brain… is an approximate Turing machine.” The qualifier approximate matters because a Turing machine is a mathematical model, whereas a brain is a physical system constrained by metabolism, wiring, noise, and finite resources. The comparison proposes a computational description of cognition; it does not establish that brains and present AI systems use the same mechanisms.
If the hypothesis is correct, human cognition does not depend on a form of information processing that is uncomputable in principle. That still leaves substantial empirical questions about which architectures can reproduce particular capacities, at what cost, and with what limitations.
What the hypothesis leaves open
The approximate Turing machine frame does not resolve the consciousness question. It brackets it. Whether a system that computes approximately as a Turing machine computes thereby has subjective experience — whether there is something it is like to be that system — is a question the frame deliberately does not answer. Hassabis was explicit about this: “I’m quite open-minded about what the answers might be.” That open-mindedness is itself a philosophical position: the avyākata move, the deliberate refusal to resolve a question before the evidence warrants it.
This Dictionary uses the phrase to name the computational hypothesis without deciding whether computation is sufficient for subjective experience or a self. See Descartes Was Wrong for the related philosophical discussion.
The practical consequence
For practitioners, the hypothesis suggests that claims about tasks AI can never perform should be stated as empirical claims with specified evidence and limits. A task may remain beyond current systems because of architecture, data, cost, reliability, or embodiment even if it is computable in principle. Those claims should be tested again as the systems change.
See also
Consciousness Calculator · Move 37 · The CERN Alternative · Sovereign Compute
Proposed May 9, 2026. Source: Demis Hassabis interview, Huge Conversations / Cleo Abram, May 2026; Alan Turing, “On Computable Numbers” (1936).