Ned Block (2002), The Journal of Philosophy 99(8), 391–425. A longer version appeared as “Petrus Hispanus Lectures 2003: The Harder Problem of Consciousness”, Disputatio 1(15), 4–49 (2003), and the paper is reprinted in Block’s Consciousness, Function, and Representation (MIT Press, 2007).
Key takeaways
- Two perspectives (§I). Deflationism gives armchair analyses of consciousness in terms of function, representation or thought. Phenomenal realism denies that and holds that consciousness is real, so there may be facts about its distribution that are not accessible to us. Naturalism, as defined here, is the view that it is the default that consciousness has a scientific (physical) nature.
- The hard problem partly dissolved (§§II–III). Subjective and objective are better seen as properties of concepts than of properties. Jackson’s Mary learns a new concept of an old fact. What remains is that we have no idea what objective concept could pick out the same thing as a subjective one, like a pre-Einsteinian thinking about mass and energy.
- Commander Data (§§V–VI). A merely superficial functional isomorph of us, whose physical realizers share nothing with our brains beyond what toasters share, and which is not homunculi-headed. Functional likeness is a reason to ascribe consciousness, but a defeasible one (C2). We have “no conception of a ground of rational belief” either way (C4): his consciousness is meta-inaccessible.
- Disjunctivism and superficialism (§§VI–IX). Naturalism could allow Data to be conscious if consciousness were a disjunctive physical property or the superficial functional role itself. Theoretical identity is accepted by inference to the best explanation (§VII), and neither identity would explain anything, so neither could be known.
- The harder problem (§§IX, XIII). Physicalism is the default and also inaccessible and meta-inaccessible. It is the “subjective default” that Data is not conscious, yet the phenomenal realist must hold the question open. Block lays out the tension and does not resolve it.
Standing in the record
Filed on 2026-10-03 at the owner’s request, with six other papers by Block on consciousness. No anthology topic holds the epistemology of other minds or of mind–body identity. The paper names no machine-learning system and carries no instruction for practice.
It extends “Troubles with Functionalism” (LIT-407, read in NOTE-366). Commander Data is a superficial functional isomorph in the sense of Troubles’ Functionalism, defined over folk psychology, and the paper places Psychofunctionalism, which Troubles distinguished, among the phenomenal- realist options (§IV, n. 13). Its fourth condition on Data rules out a “homunculi-headed” realization, the case Troubles made famous. It carries Troubles’ problem of liberalism and chauvinism over from metaphysics to evidence. Block also corrects his earlier framing: Troubles’ discussion of “universal psychology”, he says, assimilated the question of exotic creatures to the explanatory gap, and “I am guilty, too” (§IX).
It rivals “Facing Up to the Problem of Consciousness” (LIT-406, read in NOTE-336). Chalmers argues that because reductive explanation leaves experience unexplained, experience should be taken as fundamental and tied to the physical by psychophysical laws. Block’s §III takes the gap to be one between concepts, keeps physicalism as the default, and says plainly that “the standard arguments against physicalism do not work” (§X(B)). Both accept the explanatory gap; they disagree on whether it shows a gap in nature. Block cites Chalmers’s The Conscious Mind, not this paper.
Bearing on THEORY-023. That theory holds that current evidence cannot settle whether an AI system is conscious. This paper argues for a stronger claim about a narrower case. For a perfect functional duplicate on a different substrate, we lack not only the evidence but any conception of what evidence would be. It also explains why both sides of THEORY-023’s deadlock are stuck. Shared laws would not settle it either: Data’s states will obey some of our laws and not others, and we cannot tell laws of consciousness from laws of the human realizer (§X(C)). It is the philosophical source the theory’s “dispute over what counts as evidence” clause lacks.