In a neuroethics seminar, the class agrees quickly on one thing: brain devices today are mostly for recovery, not enhancement. Cochlear implants restore hearing. Deep brain stimulation treats Parkinson’s. The FDA path is built around disease.
Then someone mentions the experience machine — Nozick’s tank that gives you any life you want — and the room splits. A few utilitarians say: if wellbeing is what matters, plug in. Others refuse on instinct: something about real friendship, real achievement, real contact with the world matters, even if the qualia are identical.
That split never went away. It got louder. Neuralink is implanting write-capable BCIs in humans. Frontier labs talk about merging with AI. Longtermist tracts describe a “posthuman” future as the point of the AGI project. And a friend recently said: if the BCI is strong enough, you might not be able to press “remove” — the coupled system becomes the agent, and the pre-implant self gets replaced, not upgraded.
This essay is a map of that territory: what “posthuman” even means, why therapy/enhancement is philosophically contested (not just regulatory), how the experience machine connects to BCI coupling, and why the Ship of Theseus is suddenly an engineering question.
”Posthuman” is at least three different projects
Before ethics, disambiguate:
| Thread | Core claim | Typical names |
|---|---|---|
| Transhumanist posthuman | We should enhance; posthuman modes of being can be vastly better | Bostrom, More, Kurzweil |
| Critical posthumanism | Deconstruct human/nature/machine dualisms; anti-anthropocentrism | Haraway, Braidotti |
| Cultural/sci-fi posthuman | Uploads, digital minds, post-biological civilization | Egan, Vinge |
Cyborg is not synonymous. Haraway’s cyborg is a political metaphor for blurred boundaries. Clark’s “natural-born cyborg” is the extended-mind thesis: we already think through tools. In AI alignment circles, “cyborgism” often means human–AI collaborative cognition — yet another usage.
Posthuman ethics is unsolvable if you argue across these definitions without noticing.
Who is for it, who is against it?
Not a monoculture. Three live camps:
- Transhumanists — enhancement is morally permissible and often desirable; posthuman life could be wonderful (Bostrom, 2008).
- Bioconservatives — therapy restores; enhancement masters nature and corrupts dignity or authenticity (Sandel, 2004; Fukuyama; Habermas).
- Critical posthumanists — neither “preserve human nature” nor “Silicon Valley upgrade”; ask who holds power in human–tech entanglements (Haraway, 1985).
Mainstream neuroethics and clinical practice default bioconservative-adjacent: therapy-first, enhancement needs extra justification. That is not because philosophers proved enhancement wrong. It is a stack of normality, justice, history (eugenics), and regulatory path dependence.
Meanwhile, a separate critique — Gebru & Torres’s TESCREAL bundle — links posthuman utopianism to AGI investment narratives and 20th-century eugenic genealogies. Whether you buy that genealogy or not, it is now part of the public sentiment landscape.
Therapy vs enhancement: the philosophy, not just the FDA
Why did your class land on recovery?
Regulatory story: indications, reimbursement, equipoise.
Philosophical stories (independent):
| Argument | Idea |
|---|---|
| Normal functioning | Medicine restores species-typical capacity; enhancement pushes beyond |
| Harm vs optional good | Treating disease satisfies negative duties; enhancement is supererogatory |
| Giftedness | Therapy fixes bad luck; enhancement treats life as raw material to engineer |
| Coercion / fairness | Enhancement arms races; inequality by wallet |
| Authenticity | Enhancement changes who you are; therapy returns you to yourself |
The transhumanist rebuttal (Bostrom & Sandberg, 2009) is sharp: the line is leaky. Glasses, vaccines, IVF, and growth hormone for short healthy children blur therapy and enhancement. The same intervention can be both depending on baseline. Morally, they argue, we should not rest policy on a distinction that cannot be drawn cleanly.
Both sides can be right about different things: the distinction is fuzzy and politically load-bearing.
The experience machine vs the brain in a vat
Undergrad confusion often mashes these together. They target different questions.
Nozick’s experience machine (1974 excerpt): if only felt experience matters, you should plug in forever. Nozick’s punchline: we care about (1) actually doing things, (2) being a certain kind of person, (3) contact with reality — not just feeling as if we do.
Putnam’s brain in a vat (1981 excerpt): a skeptical scenario about knowledge and reference, not wellbeing. Different homework.
The utilitarian in your class is not stupid. Hedonistic utilitarianism really does push toward the machine. The dissenters are often running Nozick, or preference utilitarianism (respect reflective refusal to plug in), or relationship goods that do not factor into a pleasure ledger.
Now add BCI: a write-capable implant that shapes mood, memory, and desire is an experience machine you cannot leave without brain surgery. The thought experiment became industrial.
Strong coupling, IIT, and the off button
Clark & Chalmers (1998) argued cognition extends into notebooks and environments. Clark’s later work (2025 Nature Communications) applies the same lens to generative AI: we are already hybrid thinkers.
The worry about strong BCI coupling is extended mind at extreme gain:
- Read-out BCI (decode intent): mostly tool-like.
- Write-in BCI (stimulate, edit, smooth neural activity): plasticity may weave the device into the circuit. Removing it is cognitive amputation, not closing a laptop.
Friends invoke IIT here speculatively: if consciousness attaches to integrated causal structure, the moral subject might be the combined brain–device system. “Possession” is melodramatic language for a simpler claim: agency may migrate, and the pre-implant person may not have standing to consent for the post-coupling self.
Recent ethics work on enhancement BCIs (PLOS Biology, 2024) separates:
- Bioconservative authenticity — changing yourself via device is betrayal of a fixed true self.
- Existential authenticity — fine if reflectively endorsed; terrifying if the desire to change was induced by the device (e.g. you want to remove the implant, but the implant makes you want to keep it).
That is the Ulysses problem: you signed up, but the signer is gone.
Ship of Theseus: when do we stop being human? Does it matter?
Replace neurons with silicon one at a time. Swap ship planks. At which plank is it no longer the same ship — or the same person?
Parfit-style answer: psychological continuity matters more than identity facts; “same person” may be the wrong question.
Bioconservative answer: continuity is not enough; crossing human dignity thresholds matters even if memories persist.
Policy answer: we may not need to resolve metaphysics to regulate. Protect off switches, withdrawal rights, non-write defaults, open repair — regardless of whether the coupled entity is “still you.”
Illich’s Tools for Conviviality adds a design criterion: convivial tools are reversible; industrial tools lock you in. A BCI you cannot safely remove fails conviviality before it fails any single moral theory.
Speciesism and the AI safety overlap
Peter Singer’s speciesism: discounting interests because of species membership alone. The principle is equal consideration of interests, not identical treatment.
Connections you will hear in alignment circles:
- Substrateism — denying moral status to AI systems that might have interests, because they are silicon not carbon.
- Posthuman hierarchy — enhanced posthumans vs unenhanced humans as a new species caste.
- TESCREAL — posthuman astronomical value may marginalize present humans and animals.
These debates are not finished. Consciousness indicators for LLMs (Butlin et al., 2025), neurorights movements, and animal welfare law are all moving while philosophy departments still argue about Nozick.
Where I land (provisionally)
I do not think “posthuman” is a single research program, and I do not think enhancement ethics was settled in a seminar.
The load-bearing questions for the next decade look practical:
- Coupling strength — read vs write; removable vs woven-in.
- Consent over time — Ulysses contracts when the self is editable.
- Therapy/enhancement — treat the distinction as politically necessary and philosophically leaky.
- Off buttons — conviviality as cross-ideological minimum (Illich meets neurorights).
- Species / substrate boundaries — AI safety cannot ignore Singer-style arguments if we build systems that plausibly have interests.
Before debating whether we should become posthuman, we should ask who gets to design the coupling, whether exit is real, and whose interests count when the coupled system votes.
Sources
- Bostrom, N. (2008). Why I Want to Be a Posthuman When I Grow Up.
- Bostrom, N. & Sandberg, A. (2009). Ethical Issues in Human Enhancement.
- Sandel, M. (2004). The Case Against Perfection. The Atlantic.
- Haraway, D. (1985). A Cyborg Manifesto.
- Clark, A. & Chalmers, D. (1998). The Extended Mind. Analysis.
- Clark, A. (2025). Extending Minds with Generative AI. Nature Communications.
- Nozick, R. (1974). Experience machine excerpt from Anarchy, State, and Utopia.
- Putnam, H. (1981). Brain in a vat excerpt from Reason, Truth and History.
- Singer, P. (1989). All Animals Are Equal.
- Gebru, T. & Torres, É. (2024). The TESCREAL bundle. First Monday.
- PLOS Biology (2024). Ethical considerations for BCIs for cognitive enhancement.
Research pack: notes/posthuman_ethics_map.md, PDFs in readings/posthumanism/.