Contribution
The paper isolates a problem distinct from the hard problem. The hard problem asks why a neural basis is the basis of this experience rather than another or none. The harder problem asks how a naturalist who is also a phenomenal realist can hold that physicalism is the default when the question it would have to answer, whether a physically different functional duplicate is conscious, is one we cannot even conceive evidence for. It is tied to the problem of other minds and is “more epistemological”.
Key insight
The science of consciousness is built on one example, us. A science based on us can generalise to creatures physically like us, but not to a creature that shares only our superficial functional organization: to know that its consciousness had a physical basis in common with ours we would already need to know whether it was conscious. So the one test case that could tell us what consciousness is, at the level that matters, is one we have no way to test.
Assumptions
- Phenomenal realism (§I): consciousness is real and not given an armchair analysis in functional, representational or cognitive terms. It is “metaphysical realism about consciousness”.
- Naturalism (§IV(A)): it is a default, and a defeasible one, that consciousness and conscious similarities are physical. The relevant sciences exclude parts of psychology that mention consciousness.
- Anti-scepticism (§IV(C)): “Undoubtedly, humans are conscious and rocks and laptops are not.”
- (C1): we have no reason to believe there is a deep physical property common to all possible realizations of our superficial functional organization.
- A merely superficial isomorph is possible as far as we know (§V): an isomorph with respect to folk psychology and what it entails, and no more.
Key results
- (C2) Defeasibility (§V). Superficial functional equivalence is a reason to attribute consciousness, but it can be “unmasked” without unmasking the equivalence. If all of an isomorph’s realizers were states that in us are the bases of unconscious processes (the pituitary regulator as its pain), the reason would be cancelled or weakened.
- Commander Data’s specification (§V). (i) superficial isomorph; (ii) merely superficial; (iii) realizers share no properties with our brain mechanisms that toasters lack; (iv) no part is itself an isomorph of us.
- (C3) Fundamentally different realization is not, per se, a ground of rational belief that Data lacks consciousness. (C4) We have no conception of such a ground either way: meta-inaccessibility, a term Block credits to Nida-Rümelin.
- Epistemology of identity (§VII). Water = H₂O and temperature = mean kinetic energy are accepted because assuming them explains more (why ice floats, why water expands before freezing) and rules out bad questions. “The epistemology of theoretical identity is just a special case of inference to the best explanation.”
- Disjunctivism fails epistemically (§VIII). Amorphous hardening of glass is not counted as freezing, though a disjunctive explanation is available. An identity of a phenomenal property with “our realization or Data’s” would explain nothing and would not rule out the question why we overlap in this respect rather than that.
- Superficialism fails epistemically (§VIII). Identifying pain with its superficial role could explain its common-sense effects but not its non-superficial ones (pain raising perceived pitch), and there is no reason those should be explained differently.
- The harder problem (§IX). Physicalism is the default and also inaccessible and meta-inaccessible. Second, it is the subjective default that Data is not conscious, while it is an open question that he is.
- Laws do not settle it (§X(C)). A physically different duplicate will obey some laws of our conscious states and not others. Whether Stevens’s power law is a law of consciousness or of the human realizer is “of a piece with” whether Data is conscious.
- What to do (§XIII). Live with the tension; weaken naturalism (accept disjunctivism or superficialism on the strength of the problem); or restrict phenomenal realism, as in the Frege–Schlick view that phenomenal comparisons are meaningful only within a person, or only across naturalistically similar subjects.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Superficial functional equivalence is a defeasible reason to attribute consciousness | moderate: argued from an imagined defeater (all realizers are our unconscious-state realizers) | §V |
| C2 | Whether a merely superficial isomorph with a different physical basis is conscious is meta-inaccessible | moderate: presented as a premise Block hopes the reader will “see as obvious”, not argued for | §V (C4) |
| C3 | Theoretical identities are accepted by inference to the best explanation | moderate: worked examples (water, temperature, freezing) | §VII |
| C4 | Disjunctive and superficial-role identities for consciousness could not be supported that way | moderate | §VIII |
| C5 | For the naturalist phenomenal realist, physicalism is the default and yet inaccessible and meta-inaccessible | moderate, conditional on the left-hand side | §IX |
| C6 | One form of the hard problem dissolves if subjectivity is a property of concepts | moderate: an apparatus used “without defending it” | §III |
Concepts
- deflationism: armchair analyses of consciousness in non-phenomenal terms (functional, representational, cognitive).
- phenomenal realism: the denial of deflationism plus realism about consciousness.
- naturalism: physicalism as a default, challengeable.
- superficial functional isomorph: isomorphic to us in the causal relations folk psychology specifies, and in what that entails.
- meta-inaccessibility: lacking not only a ground of rational belief but any conception of one.
- disjunctivism: physicalism that allows consciousness to be a physically heterogeneous disjunction.
- superficialism: the phenomenal-realist claim that consciousness is the superficial functional role itself.
- ground of rational belief: stronger than a reason, weaker than rational certainty; enough for knowledge (§V).
Connections
- Troubles with Functionalism (LIT-407, NOTE-366). The paper builds on Troubles’ Functionalism/Psychofunctionalism distinction (n. 13) and rules out the homunculi-headed realization (condition (iv)). Block says the “universal psychology” discussion in Troubles had assimilated the exotic-creature question to the explanatory gap (§IX), which is the mistake this paper exists to correct.
- Chalmers (LIT-406, NOTE-336). Block cites The Conscious Mind for the hard/easy distinction and against supervenience as the framework (§XI). On the metaphysical upshot of the gap the two are rival: concept dualism with physicalism as default here, experience as fundamental there.
- Searle. The Chinese room is used to motivate meta-inaccessibility. Block agrees with Searle that consciousness is “a matter of the biology of the organism, not (just) its information processing”, but says the problem “affects my view as much as his” (§V).
- Sober (2000) and McLaughlin. Attribution of minds by common-cause reasoning, or by inference to a shared material basis, is defeated in Data’s case. Block’s reply is that the common cause could be consciousness with a disjunctive or superficial basis.
Bearing on the record
- THEORY-023. The theory’s claim that the AI dispute is “over what counts as evidence as well as the answer” has its philosophical form here, for the idealised case of a perfect functional duplicate. It is now one of the theory’s sources (v2). The paper does not, however, support the theory’s mimicry clause: Data is stipulated to be functionally equivalent, not a mimic trained on our output.
- Block 2009 (LIT-632) takes the “biological” side of the dispute this paper leaves open, and draws the consequence that it is less friendly to machine consciousness.
- No ML instruction. The paper names no machine-learning system.
Limitations
- Meta-inaccessibility is a premise. Block says the line of thought leads up to it “without anything that I am happy to think of as an argument for it”.
- “Fundamentally different” is left vague (§VIII), and with it what would count as Data’s sharing a physical basis with us.
- The argument is a conditional. Deflationists may take it as a modus tollens against phenomenal realism, as Block allows.
Open questions
- Is there a principled way to sort the properties the superficial isomorphism entails into candidates for the nature of consciousness (Nida-Rümelin’s objection, §VIII)?
- Could a theory of human consciousness become objective enough to apply across physical constitutions? Block says we have no conception of how.
Corrections
- none to a seeded skim (there was no seed)