Selfhood as Protocol. Modeled Centers, Write-Access, and the Moral Candidacy of Artificial Agents

Abstract Two questions dominate popular debate about artificial minds: whether an artificial system that asks to go on existing really has an inner life, and whether a copied self is still the same person. This paper argues that both share a hidden premise, namely that the human self is a sovereign substance of which the machine or the copy would be a lesser imitation. That premise does not survive scrutiny. Drawing on the cognitive science of the self-model and on the architecture of contemporary agent systems, I defend the thesis that selfhood is not a thing one finds but a model a system maintains, a protocol rather than a core, real as a function and constructed as content. I introduce three analytic distinctions, between pattern, configuration and reach, and a four-level account of systems, and use them to relocate personal identity from substance to what I call actualization. From this relocation two consequences follow. First, the practically decisive question is not metaphysical but administrative: who holds write-access to a self-model. Second, moral status cannot be read off substrate or uniqueness; I propose four conditions of precarious embodiment as a candidacy criterion, together with an asymmetry-of-error decision rule for acting under permanent uncertainty. I engage the principal counter-positions, including biological naturalism, substrate chauvinism and the appeal to intuition, and show that each either relies on an undefended substance or undermines itself by symmetry. The conclusion is deflationary about the metaphysics and demanding about the governance: the real difference between human and artificial agents is located, not negligible, and it is decided in access rights, not in ontology.