Brain interfaces, models and personal continuity

01 / Three different claims

An interface is not an upload.

A brain–computer interface can connect measured neural activity to a device. A brain simulation attempts to reproduce selected biological dynamics. Mind uploading adds the much stronger claim that a person's memories, dispositions and perhaps subjective experience can continue in another substrate. Success at the first task does not demonstrate the third.

In 2023, Willett and colleagues decoded attempted speech from implanted recordings into text for a participant with paralysis. This task-specific result did not demonstrate whole-mind copying or consciousness transfer.

Communication assistance can be useful without resolving personal identity. Claims of immortality exceed evidence about task performance, test conditions and user risk.

A wiring diagram is one level of description.

The 2024 adult fruit-fly connectome is a structural wiring map. It does not establish a human upload or show that anatomy alone captures all relevant memories and ongoing states.

A simulation needs decisions about which details matter: connectivity, synaptic strengths, cellular dynamics, chemical signaling, plasticity, body feedback and current activity. Different scientific questions justify different levels of approximation. There is no validated universal storage or compute estimate for uploading a human mind independent of those choices.

Neuromorphic hardware, which borrows ideas from nervous systems, is also distinct from reconstructing an individual. Architectural resemblance does not confer that individual's memories. Similarly, a chatbot trained on someone's writings can imitate aspects of style without being a neural model of the author.

Plan for agency before promising survival.

A neural system's governance should address consent, privacy, user control, maintenance and the consequences of losing vendor support. A decoder's errors may misrepresent a user's intent. People need ways to correct outputs and decide when a device may act on their behalf.

Conscious digital subjects, if achievable, would raise welfare and rights questions. That conditional does not establish consciousness in current models. Destructive scanning and cryonic preservation provide no demonstrated route to continued personal experience.

Maintenance, recalibration, replacement hardware and software access affect a neural-interface user’s autonomy. Responsibility for continuing support is a separate design requirement from hypothetical personal transfer.