Agents and Patients in Plural Societies

Plural societies now include biological individuals, artificial systems, and collectives, yet our institutions lack a non-metaphysical, testable way to decide who is answerable as an agent and who is owed protection as a patient. This paper offers a two-axis, role-relative framework. On the agency axis, an entity qualifies as a Public-Reason Agent (PRA) for a decision role if it can (i) give reproducible reasons on near-duplicate cases (NAT), (ii) demonstrate counterfactual stability with minimal flips and goal invariance (CST), (iii) model others and second-order responses (MMT), and (iv) correct and repair in a timely, generalized way (CRT); identity continuity and provenance are required. On the patiency axis, a Patiency Evidence Ladder (PEL0–5) organizes non-harm duties by graded welfare evidence. The axes are independent: some animals may warrant strong protections without agency; some AIs may earn procedural partnership without welfare rights. We provide auditable thresholds (v0.1), a minimal Justice Floor that is lexically prior to optimization, an “Agency Agreement” template, and micro-cases (AI tutor; forest+DAO; cephalopod welfare). Appendix B sketches how the axes guide moral weighting compatible with Non-Zero Ethics without collapsing into single-utility maximization.