Eliminative materialism in its Churchland-style form relies on an invalid inference from representational inadequacy to ontological elimination: the fact that a vocabulary is ill-suited to one explanatory task does not show that its categories are unreal. We argue that dissatisfaction with folk psychology licenses, at most, revision, supplementation, or relocation to a different explanatory level. It does not license the conclusion that beliefs and desires should be eliminated. We develop the argument in three steps. First, we argue that eliminativism trades on a double category mistake: it treats folk psychology as though it were simply a failed proto-science and treats its possible displacement as same-level ontological liquidation rather than cross-level re-description. Second, drawing on companion work on task-sufficient representation, we distinguish epistemic verdicts about representational limits from metaphysical verdicts about what exists, and from that distinction formulate a double burden on eliminativism: a successful eliminative argument must both provide a successor that captures or outperforms the legitimate explanatory and practical work of folk psychology and explain why that framework appeared to succeed. Third, we argue from scientific practice, especially the retention of higher-level kinds under deeper theories, that layered ontology is standard and anticipatory elimination exceptional. The result is a general criterion for resisting eliminativist overreach and a modest realism about higher-level kinds compatible with scientific revision.
Lorand Bruhacs (Mon,) studied this question.