Integrated Guide · Chapter 5

Philosophy and ethics — what remains open even if the engineering works.

Suppose every one of the six axes in Chapter 3 were completed. Three questions would still not resolve themselves: is this the survival of the original person, is anything experienced inside it, and what may be done to it. This chapter does not adjudicate. It lays out the major positions and what each entails — a map against which readers can inspect their own intuitions.

Stance: neutral, both sides presented18 references

In this chapter

5.1 Personal identity: is a copy you?
5.2 Consciousness: can a machine experience?
5.3 Ethics: what may be done to an emulation?
5.4 Law and social institutions
5.5 A map of positions
5.6 Chapter summary
References

5.1 Personal Identity

Personal identity: is a copy you?

Stating the problem

Suppose a brain is scanned and a complete functional duplicate is started on a computer. The duplicate has the memories, the personality and the self-awareness, and insists it is the original. Has (a) the person survived, or (b) has a very similar different person been created? This is the copy problem. If the scan is destructive, the stakes are life and death; if it is non-destructive, it becomes a fission problem in which there are now two of you.

The main positions

Psychological continuity (patternism). The self is essentially a pattern of memory, personality and psychology; if that is sufficiently continuous, the substrate — neurons or silicon — does not matter. Parfit went further, arguing that there is no deeper fact of identity beyond psychological continuity, and that identity itself is not what matters in survival¹. On this view uploading can be survival in principle.
Biological (animalist) views. I am a particular organism; when the organism dies, I die. The copy is an elaborate record of me. Here destructive uploading is always death plus the creation of someone new.
Physical continuity. Identity resides in spatiotemporal continuity of matter and process. An instantaneous copy breaks that continuity, but gradual replacement (below) might preserve it.

The thought-experiment toolkit

Thought experiment	Setup	What it forces
The teleporter (Parfit)¹	Your body is scanned and destroyed, then rebuilt on Mars. One day the destruction fails and the original remains on Earth	The intuition that "the copy is survival" collapses the moment the original persists. Destructive uploading is logically the same device
Gradual replacement (Moravec)²	Neurons are replaced one at a time with functional equivalents. Consciousness never lapses and the subject notices nothing	At which single replacement did you become someone else? The Ship of Theseus. Most people find this more survival-like than instantaneous copying — but is there a theoretical basis for that intuition?
Fission (Egan)³	A non-destructive scan starts a duplicate, so original and copy coexist	Identity is supposed to be a one-to-one relation. If both are "you", the concept breaks down and pushes you toward Parfit's "identity is not what matters"
Hemisphere connection (Watanabe)⁴	While alive, one brain hemisphere is coupled to a machine hemisphere via an interface, experience is shared as one integrated consciousness, and the biological side is later lost	The claim that connect-integrate-migrate preserves subjective continuity. An engineering instantiation of gradual replacement; its testability is the live question

Where this actually stands

The problem is two millennia old, but the practical consensus in the uploading context is an asymmetry: intuitions split evenly on whether an instantaneous destructive copy counts as survival, while gradual migration is accepted as survival by more positions. That asymmetry is why engineering keeps reinventing stepwise replacement and connection — Moravec's procedure, the hemisphere proposal — as the identity-friendly route.

5.2 Consciousness

Consciousness: can a machine experience?

An emulation could be behaviourally perfect and still raise a separate question about whether there is subjective experience inside. The leading theories make opposite predictions, so this is not merely "unsolved" — the choice of theory is itself unsettled.

Theory / position	Would an emulation be conscious?	Grounds and difficulties
Functionalism / computationalism (Chalmers and others)	Yes	Consciousness supervenes on functional organisation. The fading/dancing qualia arguments hold it is incoherent for experience to gradually vanish or oscillate as neurons are replaced one by one⁵. The theoretical pillar of substrate independence.
Global neuronal workspace (GNWT) (Dehaene and others)	Possibly	Consciousness is a property of an architecture that broadcasts information globally. Implement the architecture and the substrate is irrelevant⁶.
Integrated information theory (IIT) (Tononi and others)	Probably not	Consciousness resides in the integration of causal structure (Φ), and its proponents state explicitly that a simulation on a von Neumann machine could reproduce function while having near-zero Φ⁷. Note the corollary: a neuromorphic implementation might be a different case.
Biological naturalism (Searle) / life-based views (Seth and others)	No	Consciousness depends on causal properties of living systems, or on life processes themselves; implementing the computation is not enough⁸. Hard to prove or falsify.

In 2025, a large adversarial collaboration (Cogitate) pitted IIT against GNWT in the same experiment and reported that core predictions of both theories were partly wrong⁹. Consciousness science is advancing, but no theory capable of adjudicating whether an emulation is conscious has yet been selected. Beneath that sits the more fundamental problem of other minds: there is no way to establish the presence of subjective experience from the outside, from behaviour and structure alone. Schneider and others have proposed practical tests based on subjective report during chip replacement, but these are not regarded as decisive¹⁰.

5.3 Ethics

Ethics: what may be done to an emulation?

Suffering in failed attempts. An incomplete emulation under development might have a broken subjectivity and persistent suffering. Sandberg argued that animal and human emulation research should pre-emptively adopt a framework analogous to laboratory-animal welfare — estimating suffering, defining stopping criteria, controlling states in a way equivalent to anaesthesia¹¹. Metzinger goes further, proposing a moratorium on creating synthetic phenomenology at all until 2050¹².
The first subjects are animals. The fly emulation of 2026 already stands at the entrance to this problem. Whether a fly has morally relevant experience is unknown — but the question arrives in the form of whether it is acceptable to run millions of copies while it remains unknown. Arguments that AI welfare must be taken seriously have become a mainstream research topic since 2024¹³.
Consent. What form of advance consent is necessary and sufficient for post-mortem scanning and for starting an emulation? Preservation coupled to assisted death (the 2026 Nectome protocol¹⁴) introduces an option of highly uncertain prospect into end-of-life decision-making. Designing consent that does not smuggle in inflated expectations is one of the few genuinely present-tense ethical tasks.
Rights over shutdown, deletion and copying. If there is consciousness, is shutdown anaesthesia or homicide? Is restarting from backup resuscitation or a new birth? How are property, responsibility and relationships allocated among copies? May an emulation be copied, edited or experimented on without its own consent?
Inequality and coercion. Expensive preservation and uploading becoming a class good; employers or states effectively compelling emulation — a pressure Hanson's em economy models explicitly¹⁵.

Present-tense ethical work — not futurism

1. Monitoring the gap between preservation marketing and scientific evidence. 2. Consent design and review for assisted-death-compatible protocols. 3. Drafting welfare guidelines for animal emulations. 4. Privacy of connectome data — brain data is the ultimate personal information. All four are live practical problems as of 2026.

5.4 Law

Law and institutions: no framework exists yet.

As of 2026, no jurisdiction has law that addresses emulations directly. The foreseeable disputes are: conferring legal personhood (by analogy to natural persons or to corporations); consistency with the definition of death (is a preserved brain a corpse or a patient); the criminal status of shutdown and deletion; voting rights and how "one person" is counted; labour law (variable-speed work and unlimited copying against minimum wage); inheritance (does it occur at all if the person "continues"); and jurisdiction (server location or nationality). The nearest existing precedent at the intersection of neuroscience and law is "neurorights" legislation protecting brain data, such as Chile's 2021 constitutional amendment. For institutional design as a thought experiment, Hanson's analysis of em society¹⁵ and recent policy research on the rights of digital minds¹³¹⁶ are the starting points.

5.5 Position Map

A map of positions: where does your intuition sit?

Cluster	Identity	Consciousness	Consequence	Representative figures
Optimistic patternism	Psychological continuity suffices	Functionalist	Uploading is survival; destructive scanning is acceptable in principle	Kurzweil, Koene, Zeleznikow-Johnston¹⁷
Gradualism	Continuity is what is essential	Functionalist-leaning, attentive to implementation	Instantaneous copying is dangerous; gradual replacement or connection can be survival	Moravec, Chalmers (conditionally)⁵, Watanabe⁴
Identity deflationism	"Identity" is not what matters	—	Reframe the question: what should persist is psychological continuity and relationships	Parfit¹
Substrate scepticism	—	IIT / biological naturalism	A digital emulation may be an unconscious duplicate — a zombie	Tononi⁷, Searle⁸
Biological conservatism	Animalism	Life-based	Uploading is, by definition, death plus the creation of a record	Olson-style metaphysics; Seth on consciousness⁸

5.6 Summary

Chapter summary.

Key points

The copy problem is unresolved but its structure is clear: intuitions about instantaneous copying and gradual migration are asymmetric, and engineering is following that asymmetry toward identity-friendly routes.
On consciousness, theories predict opposite answers and the 2025 adversarial experiment did not settle the matter. Whether an emulation is conscious is currently undecidable — and that uncertainty is precisely the grounds for ethical caution.
Ethics is not a future topic: consent design for preservation, welfare of animal emulations, and brain-data privacy are real problems as of 2026 (see Chapter 6).

References