François Kammerer (2015), Philosophia 43(4), 1047–1057 — DOI-10.1007/s11406-015-9653-z
Key takeaways
- Nesting as such is allowed. A conscious whole may have conscious parts, so the antheads and Block’s tiny people pass. What is excluded is a conscious state that the whole has only because some part consciously represents the whole, in a role no non-representing part could fill.
- Applied to the US: citizens represent the US (what it is, does, should do), and a “country” of neurons or simple computers, or of inhabitants unaware of the group, would lack its complexity. So the US fails both conditions and has no conscious states. That counterfactual is asserted, not argued.
- The principle works state by state, not whole by whole. A system can keep its “old” conscious states while being denied new ones that depend on self-representing parts.
Standing in the record
Filed on 2026-10-02 at the owner’s request, as one of the works bridging Minsky’s Society of Mind (LIT-429, with the memos LIT-399 and LIT-402 and Singh’s exposition LIT-444) to Schwitzgebel’s “If materialism is true, the United States is probably conscious” (LIT-159, read in NOTE-131). NOTE-362 is the close reading of the same day, and it placed the work: Active.
Where it sits on the seam. It is the seam: an anti-nesting principle, rebuilt so that Schwitzgebel’s §2 objections do not touch it. Against Schwitzgebel’s claims as NOTE-131 numbers them:
- C3 (a whole of conscious parts can be conscious): concedes it in general, contests it for the US. The SAP explicitly allows group consciousness when the whole’s organisation does not depend on members representing the whole.
- C4 (Putnam’s principle is unjustified): concedes it. Kammerer accepts that the antheads and Block’s neuron-replacement case refute Putnam’s principle. He keeps only its “intuitive pull”, and diagnoses Putnam’s error as reading “part” spatially rather than functionally.
- C5 (IIT’s exclusion postulate has absurd consequences): neutral. IIT appears only in a footnote (n. 2), as another principle Schwitzgebel discusses. The SAP is offered as an add-on to any supervenience theory, not as a replacement for exclusion.
What it gives the bridge. The SAP draws its line exactly where Minsky’s and Schwitzgebel’s pictures part. Minsky’s agents are mindless and represent nothing about the mind they compose, so condition (A) never holds and the SAP never fires on a society of mind. Kammerer says as much of global-workspace modules, which “are not supposed to have any conscious mental states at all” (p. 7). Schwitzgebel’s US is the opposite case: its parts are minded and their minding is about the whole. Hofstadter’s “Ant Fugue” colony (LIT-437) falls on Minsky’s side, since the ants do not represent the colony. So the SAP turns the decomposition/composition contrast into a criterion: not “are the parts minded?” but “does the whole’s organisation run through parts’ representations of the whole?”
Schwitzgebel’s replies.
- He answered in “Is the United States Phenomenally Conscious? Reply to Kammerer”, Philosophia 44(3), 877–883 (online 20 June 2016), DOI-10.1007/s11406-016-9725-8. Per its abstract, he raises three concerns: that the SAP may not exclude actual groups, that it makes group consciousness depend on internal details of individuals, and that it is ad hoc. That reply is not filed here.
- In The Weirdness of the World (LIT-216; NOTE-188) he answers Kammerer again, alongside Dretske, as “motivated post-hoc”, with the Planck-sized person counter-case.
- NOTE-188 glosses the principle as “a whole is not conscious if its conscious parts represent the whole”. That leaves out condition (B), indispensability, and the state-by-state scope. Both matter, because they are what let the SAP keep Block’s tiny people (who represent the whole but are replaceable) conscious.
Other neighbours. The SAP’s justification, that one should not ascribe consciousness where “other, distinct subjects” explain the behaviour (p. 8), is an explanatory-parsimony criterion. Set beside Dennett’s “Real Patterns” (LIT-220), it asks a different question: not whether a group-level description earns its keep, but whether the subjects inside already explain what it describes. Kammerer does not cite Dennett on this. Kammerer explicitly allows spatially scattered conscious beings (Schwitzgebel’s supersquids), so the SAP is no contiguism and bears nothing on LIT-124’s composition principles. It is compatible with List’s IIT reply (LIT-401), which reaches the same verdict on the US by a different and independent route.
It carries no instruction for machine-learning practice, so nothing belongs in the Anthology of the SOTA.