Keyboard shortcuts

Press or to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Mind Uploading and Whole-Brain Emulation

Mind uploading is the proposed realisation of a person's mental organisation in a different substrate, usually by scanning and emulating the brain. Whole-brain emulation is the engineering project of reproducing relevant brain functions at sufficient detail. The philosophical questions precede feasibility: which organisation is sufficient for mentality, whether functional duplication preserves consciousness, and whether a resulting subject would be the original person or a copy.

These questions should be separated. An emulation could be conscious and psychologically continuous with someone while not being numerically identical to them. Conversely, an identity theory might permit continuity under gradual replacement while rejecting a separately created duplicate. The metaphysics of persistence is handed to the existing time section; this page focuses on mind under changed realisation.


What would be emulated?

An upload proposal must choose a target grain:

  • connectome and cell types;
  • neural firing and synaptic dynamics;
  • biochemical and glial processes;
  • bodily feedback and endocrine regulation;
  • functional or computational organisation only;
  • learned relations to environment and social practice.

Too coarse a model may reproduce behaviour while omit mechanisms relevant to consciousness or affect. Demanding every physical detail makes emulation indistinguishable from duplication of matter and abandons functional abstraction.

The correct grain is theory-relative. A functionalist seeks causal organisation; a biological naturalist may require specific cellular powers; an embodied view includes body-environment loops.

Computational sufficiency

If mental states are computational-functional states, implementing the right computation in another substrate preserves cognition. The emulation need not share molecular composition any more than one program needs one processor.

The premise is stronger for reasoning and memory than for phenomenal consciousness. Functionalism faces absent-qualia and implementation objections. A simulation could reproduce reports while its intrinsic dynamics differ in ways a rival consciousness theory regards as decisive.

Computational sufficiency must also specify timing, recurrence, embodiment, and counterfactual transitions. Matching recorded outputs is not emulating an active mind.

Structural and dynamical fidelity

An emulation may preserve network connectivity while alter continuous timing, noise, plasticity, or chemical modulation. Whether these differences matter depends on which properties realise cognition.

Structural fidelity reproduces components and connections. Dynamical fidelity reproduces how states evolve under varied conditions. Functional equivalence requires enough of both to preserve relevant counterfactual behaviour.

Validation would compare responses across perturbations, learning, memory, and action, not merely replay a known trajectory. Consciousness validation remains underdetermined because functional equivalence is the disputed premise.

Embodiment

A disembodied emulation would lose bodily regulation, interoception, sensorimotor skill, and environmental history. It could receive simulated or robotic inputs, but equivalence depends on whether those loops preserve the original organisation.

Embodiment may be required for affect and selfhood even if abstract reasoning transfers. A body model can be simulated, yet a homeostatic theory may distinguish modelled need from actual material self-maintenance.

The uploaded mind may therefore be a transformed kind of subject rather than an unchanged human mind in a new container.

Gradual replacement

Imagine replacing neurons one at a time with functionally identical components while the subject remains awake and behaviourally continuous. Intuition often favours survival because no sharp boundary occurs.

Chalmers' fading and dancing qualia arguments use this case to support organisational invariance. If experience vanished despite unchanged function, the subject would be unable to notice or report the loss.

Gradualness does not prove identity or consciousness. A property could decline continuously, and a sequence of small changes can cross a vague or sharp threshold. Still, replacement pressures theories to identify when and why the relevant property fails.

Scan-and-copy

A destructive scan records structure while destroying the biological brain, then creates an emulation. A non-destructive scan leaves the original alive alongside the copy.

The second case exposes the identity problem. Both successors cannot be numerically identical with one pre-scan person if they are now distinct from each other. Psychological continuity can branch; identity cannot.

This does not show the copy lacks a mind. It may possess memories, character, projects, and consciousness while being a new person. The distinction between survival and duplication is ethically and practically central.

Pattern and token views

A pattern view treats the person or mind as an informational and functional pattern. Multiple realisations preserve the pattern, suggesting upload survival.

If two copies instantiate the same pattern, either one pattern can be multiply located, both are continuations without strict identity, or token history distinguishes them. Ordinary persons also change physical matter, so material continuity alone seems too strict.

A token-continuity view requires the right causal continuation of one process. Gradual replacement may satisfy it while copy-and-restart does not. The criterion is historical, not merely qualitative similarity.

Psychological continuity

An upload may preserve apparent memories, dispositions, values, language, and plans. It would sincerely regard itself as the original and be recognisable to others.

Psychological continuity can support prudential concern and social succession even if strict identity fails. Parfit's view that identity is not what matters makes uploading less dependent on a binary metaphysical verdict.

False memories and copied records complicate the relation. Genuine memory may require an appropriate causal trace from the original experience; an emulation can inherit such a trace through scanning even if a separate copy also does.

The numerical question remains in Persistence and the ontology page on Persons.

Consciousness

The upload's consciousness depends on theory:

  • functionalism favours preserved fine-grained organisation;
  • recurrent and workspace theories require their respective architecture;
  • integrated-information theory depends on intrinsic causal implementation;
  • biological naturalism may deny non-biological consciousness;
  • embodied theories require adequate bodily and autonomous loops;
  • panpsychist views ask whether the new substrate supports the combining structure.

Behavioural continuity cannot adjudicate theories designed to permit behavioural duplicates without experience. The artificial-consciousness framework therefore applies directly.

Speed, copies, and branching

Digital emulations could run at different speeds, pause, branch, merge records, or create many copies. These operations stress concepts formed around continuous biological lives.

A speed change should not remove mentality if causal organisation is preserved, though interaction with an environment must be rescaled. Pausing resembles dreamless sleep on some views and process termination plus recreation on others.

Branching creates several equally continuous successors. Merging threatens ownership of conflicting memories. The cases suggest psychological continuity is a relation that can vary and branch rather than numerical identity.

Original and copy interests

If an emulation is conscious, its welfare and autonomy matter regardless of whether it is the original. Calling it a copy does not make its suffering simulated from its own perspective.

Pre-upload consent cannot settle every future choice, especially when successors diverge. Copies may acquire separate interests immediately. Ownership claims over an emulation conflict with treating it as a person.

The moral problem therefore depends first on mentality and sentience, then on identity for inheritance, promises, and responsibility.

Epistemic validation

A credible emulation programme would validate progressively:

  • component dynamics;
  • network responses under intervention;
  • preserved learning and memory;
  • robust psychological capacities;
  • self-report tied to internal monitoring;
  • theory-specific indicators of consciousness.

Passing behavioural tests supports functional equivalence. It cannot demonstrate identity with the source, and its evidential force for consciousness depends on the theory being tested.

The engineering phrase "whole-brain" should not imply philosophical completeness merely because a chosen model includes every component at its own grain.

Assessment

QuestionWhat would support itWhat remains separate
Cognition preserved?counterfactual functional and learning equivalencephenomenal consciousness
Consciousness preserved?theory-specific causal organisationnumerical identity
Same person?contested causal or psychological continuitywhether successor has a mind
Survival matters?continuity of projects, relations, and welfarestrict one-one identity
Copy has status?sentience, agency, and interestsbeing the original

Mind uploading is not one yes-or-no proposition. Computation, embodiment, consciousness, identity, and moral succession can come apart. The most defensible claims specify which property a proposed emulation preserves and avoid using behavioural fidelity as a proxy for all the others.

Selected references

  • Chalmers, David J. "The Singularity: A Philosophical Analysis" (2010).
  • Parfit, Derek. Reasons and Persons (1984).
  • Sandberg, Anders and Nick Bostrom. Whole Brain Emulation: A Roadmap (2008).
  • Schwitzgebel, Eric. "The Problem of the Toad's Plurality of Simultaneous Visual Experiences" (2015).