Pointing Is Not Defining: The Definition of Consciousness Is Not Settled

It is widely held that the explanandum of consciousness science is adequately fixed by phenomenological characterisation (Nagel's "something it is like" and its kin), so that no further definition is needed before theories are compared, and functional or architectural formulations are to be judged as hypotheses about that fixed target. This holds for the uncontested cases. For the cases that motivate much of the science (unresponsive patients, infants, animals, machines) the characterisation does not decide membership, and whoever carries it to those cases supplies the decision from premises about which features of reporting humans matter: a stated theory in a few recent programmes, a weighted cluster of properties in the natural-kind strategy, and an unstated analogy in most clinical, developmental and comparative practice. A definition is called settled when researchers share a procedure for deciding the cases the science investigates, whether or not the verdicts of that procedure are yet agreed; a shared procedure is what a programme of tests for consciousness needs if its tests are to be validated beyond the consensus cases (cf. Bayne et al., 2024). Whether the definition of consciousness is settled in this sense is an empirical question about the field, and the evidence reviewed here answers it in the negative. Each of the twelve formulations examined rests on terms it does not define, none can be applied to a new case without first-person report, inference from reporting cases or a prior theory, and the one systematic survey of the field finds that the definition of what is to be explained is not shared among its respondents. Conditions on a formulation used to extend the term to systems that do not report express the obligation the survey's authors already recommend, to state the premises that do the extending; the natural-kind strategy meets it by stating a feature set and a rule for weighting it. Applied to the perturbational complexity index, conditions C3 to C7 show which steps of an existing extension are stated and which are inherited from report.