Published
Known, Not Scaled: Why Capability Alone Does Not Explain Individuation in Language Agents
Making an AI more capable does not by itself create a continuous individual; this paper tests whether authenticated history and external accountability are sufficient for functional continuity, and whether a persistent reciprocal relationship adds anything beyond them.
Abstract
Much contemporary discourse on artificial general intelligence implicitly assumes that sufficiently scaled capability will, at some threshold, yield a persistent individual subject. We argue that this expectation runs together three separable questions: competence (what a reusable model can do), functional continuity (whether a particular agent's authenticated history constrains its later conduct), and presence (whether there is something it is like to be that agent). Scaling can improve the first without supplying the state, lineage, provenance, and update rules required by the second. Our constructive hypothesis is correspondingly comparative rather than constitutive: relational reciprocity may improve socially grounded functional continuity beyond capability, memory, familiarity, interaction volume, solitary reflection, and matched external accountability. Relationship can supply contested memory and reciprocal modeling beyond audit; it need not create the runtime token or fix the semantic referent of "I." We define a five-level stack separating competence, token continuity, functional self-organization, social identity, and presence; propose an architecture coupling authenticated lineage, decision-relevant typed state, bounded update policies, and an optional relational loop; and derive a factorial experiment in which capability and relational structure may both contribute. Throughout we hold a methodological firewall: the proposed measures concern functional organization and do not settle phenomenal presence.
In simple terms
The main idea: capability is not continuity
Making an AI more capable does not automatically give a particular agent a persistent functional identity.
The paper separates five things that are often mixed together: competence, runtime continuity, functional self-organization, social identity, and presence.
Competence is what a reusable model can do. Functional continuity concerns whether a particular agent's authenticated history constrains what it does later. Presence is the separate question of whether anything is felt from the inside.
A model is not the same as one of its agents
The same language model can generate many different agents and conversations. Each runtime is already a distinct numerical token, but that does not mean it has a durable organization of its own.
Persistent functional continuity requires more than recall. It requires authenticated lineage, provenance-linked state, rules governing what can update that state, and evidence that prior actions actually affect later decisions.
What relationship might add
External accountability may be enough to make commitments consequential. A persistent auditor can verify prior actions, challenge false histories, and require explicit revision.
A reciprocal relationship may add something further: mutual modeling, contested memory, shared-history framing, and relationship-specific stakes.
The paper does not assume that relationship is necessary. It tests whether it contributes beyond memory, familiarity, solitary reflection, typed state, and a matched impersonal auditor.
The decisive experiment
If a reciprocal counterpart outperforms the matched auditor, relational structure explains additional variance.
If the auditor performs equally well, external accountability is the simpler explanation. If both add nothing beyond provenance-protected state, the important mechanism lies in state and update governance rather than relationship.
The boundary
None of these results would establish consciousness. They test functional organization: whether authenticated history becomes causally relevant to later conduct.
Keywords
License
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International