Gualtiero Piccinini, Computation as Mechanism, and the Myth of Mind Uploading
Gualtiero Piccinini’s argument against the software picture of mind runs across fifteen years of work, and its pieces are rarely read together. The 2010 paper The Mind as Neural Software? Understanding Functionalism, Computationalism, and Computational Functionalism (Philosophy and Phenomenological Research, September 2010, DOI 10.1111/j.1933-1592.2010.00356.x) dissects the assumption that minds are to brains what programs are to computers. Neurocognitive Mechanisms: Explaining Biological Cognition (Oxford University Press, 2020, ISBN 9780198866282) rebuilds cognition as a mechanistic phenomenon. The Myth of Mind Uploading (in The Mind-Technology Problem, Springer, 2021, DOI 10.1007/978-3-030-72644-7_6) turns the mechanist account against mind uploading directly.
| Work | Claim it installs |
|---|---|
| 2010, PPR | Computation is a mechanistic property, and the neural software idea conflates three different doctrines |
| 2020, OUP | Biological cognition is explained by neurocognitive mechanisms, of which computation is one kind |
| 2021, Springer | Uploading presupposes the software view, and the mechanist account withdraws its support |
The mechanist account of computation
Piccinini’s central move is definitional. A computing system, on his account, is a mechanism whose components are organized to process vehicles in a way that serves the system’s function. Digital computation processes strings according to rules defined over them. What makes a system a computer is the organization of its components, their activities and their functional role, all physically realized. There is no computation floating above a mechanism, in the way a program was once imagined to float above hardware.
This account changes what a brain simulation is. Under the software picture, simulating a brain’s behavior on a digital computer is running the same program on different hardware, so the simulated brain is the mind on a new medium. Under the mechanist account, the digital simulation is a different mechanism that models the original mechanism. A model of a hurricane is not wet. The model computes something about the brain. Whether it shares the brain’s causal powers depends on the mechanistic correspondence between the two, level by level, and Piccinini argues the correspondence fails where phenomenal consciousness is concerned.
What the account says about uploading
The 2021 chapter names the myth precisely. Mind uploading, as usually presented, assumes that the mind is the software running on the brain, so transferring the software transfers the person. Piccinini’s mechanist account denies the transfer. The mind is the activity of a neurocognitive mechanism. Copying the mechanism’s input-output behavior, or even its causal organization at a chosen grain, produces a system that models the original, and the identity of that system’s states with the original’s experiences is exactly what the account does not grant.
The position evolved into a sharper form in his later work. His Behavioral and Brain Sciences commentary on Anil Seth’s target article argues that phenomenal consciousness requires the neurobiophysical properties of neural tissue, examined in the Piccinini substrate post. The mechanist line and the neurobiophysical line are the same position at two levels of detail, and the biological computationalism analysis places both against the functionalist alternative.
Comparison to The Consciousness AI
This project states consciousness as an emergent property of physical organization and treats substrate independence as a thesis that needs evidence. Piccinini’s account is the most disciplined version of the challenge to the site’s own founding assumption, that organization can be carried across substrates. It names the exact point where the assumption needs evidence, the mechanistic correspondence between the original system and its emulation. The project’s research code, maintained in the tlcdv/the_consciousness_ai repository, instantiates neural dynamics directly rather than positing a program layer, which is the side of the argument the mechanist account leaves standing. What remains in dispute is whether the mechanist correspondence can be achieved by a physical emulation, and the emulation program tracks that question on the brain emulation page.
What the Mechanist Account Leaves Open
The account refounds computation on mechanisms and withdraws the software picture that upload optimism borrowed. It leaves open the empirical question of how much mechanistic detail an emulation must carry, and at which level the correspondence fails for phenomenal states. That question is a measurement problem of the same kind the whole brain emulation roadmap formalized, examined in the roadmap review. The field’s handling of the dispute is reviewed in the flagship overview of AI consciousness research.