David J. Chalmers (2016), in Godehard Brüntrup and Ludwig Jaskolla (eds), Panpsychism: Contemporary Perspectives, Oxford University Press, pp. 179–214 (volume dated 2017 on the book). DOI 10.1093/acprof:oso/9780199359943.003.0008. Author’s draft at https://consc.net/papers/combination.pdf.
The brief’s citation is right in title, editors, publisher and venue. Its year is the book’s; Crossref dates the volume 29 December 2016. Its “earlier version circulated from 2013” is plausible but unverified: the paper was written alongside the 2013 Amherst Lecture, and the posted draft is the April 2015 one.
Key takeaways
- Three combination problems, not one (§3). The subject problem: how do micro-subjects make a macro-subject? Its sharpest form is subject-summing: given any group of subjects, it seems possible for them to exist without any further subject. The quality problem, whose sharp form is the palette problem: a handful of micro-qualities and a vast array of macro-qualities. The structure problem, whose sharp form is structural mismatch: the structure of experience is not the structure of the brain. There are also the unity, boundary, awareness and grain problems. A solution has to address all of them and should say which it addresses.
- The arguments (§4): anti-aggregation (James’s nihilism about composites, easy to reject), subject-summing, a conceivability argument for a “panpsychist zombie” world, a knowledge argument, the palette argument, the revelation argument and the structural mismatch argument. Chalmers rates subject-summing and structural mismatch as the serious ones.
- Non-combinatorial responses (§5). Emergent panpsychism (IIT read as a fundamental law is one) and autonomous panpsychism avoid combination but inherit dualism’s problem of mental causation. Identity panpsychism makes the macro-subject a fundamental entity; its best form is quantum holism, where an entangled system is the subject, and it faces doubt about brain-scale entanglement and a structural mismatch.
- Combinatorial responses (§§6–8). Deflating the subject only relocates the problem. Combinatorial infusion collapses into identity or emergent panpsychism. Phenomenal bonding is “one of the more promising approaches” but must steer between one giant subject (if bonding is transitive) and fragmentary subjects (if not). Functional accounts of awareness meet the conceivability objection. Panqualityism avoids micro-subjects but leaves all three problems open. For structure, the only viable line is that phenomenal structure follows informational structure, and it is unclear how that could be constitutive rather than a brute law.
- Conclusion (§9): the problem may be unanswerable, or answering it may point to the right theory.
Standing in the record
Filed on 2026-10-02 at the owner’s request, as part of the Chalmers works he asked for, and for the composition side of THEORY-043. NOTE-358 is the close reading. Active: it is the reference statement of the problem, by the author whose taxonomy the literature uses.
On THEORY-043. The paper is about micro-subjects composing a macro-subject, not about groups. But three things in it bear directly on that account.
- James’s argument is a group-mind argument. The epigraph that names the problem is James’s twelve men, each thinking one word of a sentence, with “nowhere … a consciousness of the whole sentence”, and “the private minds do not agglomerate into a higher compound mind”. Subject-summing formalises it (§4.2). Its premise is general: no group of subjects necessitates a further subject. If it held, it would refuse Schwitzgebel’s upward move (LIT-159) as firmly as any anti-nesting principle. Chalmers aims it only at constitutive panpsychism, though. It bites where the whole’s experience is grounded in the parts’ experiences. Schwitzgebel grounds a nation’s experience in its organisation, which is a materialist’s claim and is not Chalmers’s target here.
- An objection to exclusion that does not mention groups. Footnote 7
likens a view Nagel suggested to IIT’s exclusion postulate, which it
glosses as “a system is conscious iff it is not part of (and does not
contain) a system with a higher degree of integrated information”.
Chalmers’s objection is that both make consciousness extrinsic:
intrinsically identical systems might or might not be conscious
depending on their surroundings. This is an argument against the
account’s first refusal route, anti-nesting, and it is independent of
what the principle says about nations. It cuts the other way from what
the account’s
promote_whenasks for: it is an independent objection to exclusion, not an independent justification of it. Schwitzgebel’s ballot reductio (NOTE-131, C5) is a concrete case of the same extrinsicness, though the footnote does not cite it. My own remark, not the paper’s: Kammerer’s functional principle (LIT-409) does not obviously meet the objection, since it rules on a whole from what its parts do, not from what surrounds it, and it leaves the parts’ consciousness alone. - The boundary problem is the same seam. Phenomenal bonding must bond the micro-subjects of one organism and not those elsewhere. Chalmers’s candidate for a relation that does this is “informational integration along the lines of Tononi’s hypothesis”. He notes that such a relation is derivative, not fundamental, so it opens a new combination problem. That is THEORY-043’s second route, an architectural criterion, reached from panpsychism, and it is left unprincipled here too.
On Levin’s “stand on the combination problem”. Levin (LIT-439, read in NOTE-341, C8) says his model “takes a stand on the perennial ‘combination problem’ and can perhaps be seen as a form of panpsychism”, citing Chalmers 2013. Two findings.
- Levin is not citing this paper. His reference list (PMC6923654) gives “Chalmers D. (2013). Panpsychism and panprotopsychism. Amherst Lect. Philos. 8”, that is, Chalmers’s companion paper. NOTE-341 records only “citing Chalmers 2013”, so the record has not said otherwise, but the brief assumed this paper.
- By this paper’s standard it is not yet a stand. Chalmers asks that a proposed solution say which of the subject, quality, structure, unity, boundary, awareness and grain problems it addresses (§3), and say how the macro-subject is constituted. Levin names no sub-problem and offers no constitution relation. His nested goal-pursuing Selves are a candidate answer to the boundary problem for agency. For experience they would need a functional account of awareness, and §6.4 argues that such accounts face the same conceivability objection as functionalism. The nearest reading of his remark in Chalmers’s terms is a panprotopsychist view with functional bonding. Chalmers rates that family as open, not as solved.
Neighbours. IIT 3.0 (LIT-445) states the exclusion postulate that footnote 7 objects to, and Barrett et al. (LIT-185) defend IIT’s panpsychist reading. Chalmers here classifies IIT, read as a fundamental law, as emergent panpsychism (§5.1). Among the core consciousness works being filed alongside this one, Panpsychism and Panprotopsychism is this paper’s companion and supplies its terminology (§2). The hard-problem conceivability argument is the template for the panpsychist zombie argument of §4.3. It carries no instruction for machine-learning practice.