Chanissara Viboonlarp

Moral judgments characteristically present themselves as truth-apt and stance-independent, and this phenomenology is often taken to support moral realism. This paper develops a structural explanation of these features within a predictive processing framework. I assume that predictive operations shape phenomenology in accordance with their structural commitments, and extend this to the case in which the system predicts its own global experiential state, including action-guiding dispositions. On this basis, I argue that predictive processing yields (i) normativity from the fact that predictive operations determine action, (ii) truth-aptness from the structure of prediction as answerable to an assumed actual outcome, and (iii) an appearance of stance-independence when, in predicting its own action-guiding states, the system both presupposes a determinate outcome and brings it about, without representing this relation. I then argue, by inference to the best explanation, that this account explains common features of moral phenomenology, including resistance to revision, agent-generality, and social conformity. If correct, the account shows that the appearance of stance-independence is systematically misleading and therefore cannot serve as evidence for a stance-independent normative domain, thereby undercutting a prominent phenomenological route to moral realism.

Contemporary social epistemology struggles to explain cases of deep disagreement that persist despite evidence-sharing and rational deliberation. Existing approaches, including hinge epistemology and social-ontological accounts, illuminate the role of fundamental commitments but do not fully explain why certain disagreements generate structural instability rather than stable pluralism. We develop a structural account of legitimacy boundaries grounded in the functional role of non-negotiable normative commitments, using a framework that distinguishes three axes: negotiability, target scope, and actionability. Our central result, the Legitimacy Trilemma, is a structural limit: once a violation of a universal, non-negotiable legitimacy norm is publicly acknowledged within a community, the community cannot simultaneously maintain universal standards, insider inclusion, and the absence of acknowledged internal contradiction. It must instead make a repair move - excluding the violator, enforcing compliance, revising the norm, or blocking acknowledgment of the breach. This constraint is independent of the merits: it holds whether the norm is just or unjust and whether the challenge is warranted or baseless, identifying a structural form without delivering a normative verdict. The instability of deep disagreement therefore arises not primarily from irrationality or bias, but from this constraint on admissibility within shared epistemic spaces. We further argue that a specific subclass of phenomena commonly described as epistemic injustice can be modeled as such reactive boundary defense - complementary to, not a replacement for, accounts of oppression as a constitutive feature of social structures. Where deep disagreement appears intractable, the difficulty is typically not that no resolution exists but that every resolution exacts a cost - to identity, inclusion, or coherence - even among competent, good-faith agents.