Robert D. Rupert (2004), Journal of Philosophy 101(8), 389–428 — DOI-10.5840/jphil2004101826
Standing in the record
Filed on 2026-10-02 at the owner’s request, as a critic of Chalmers’s extended-mind programme. NOTE-343 is the close reading of the same day, and it placed the work: Active.
How it bears on LIT-097. It accepts the case and attacks the arguments for it. Its alternative is not that the mind stops at the skin by definition, but that a conservative hypothesis (embedded cognition, HEMC) explains the same cases. On LIT-097’s §5 criteria it builds the dilemma NOTE-180’s open question about “cognitive bloat” anticipated: without past endorsement, anyone who can dial directory assistance believes every listed number; with it, conscious endorsement inside the organism is “the ultimate source of cognitive authority”, and the notebook looks like a tool. On LIT-097’s appeal to explanatory simplicity and natural kinds, it answers from memory research. NOTE-180 recorded that simplicity “is power” was assumed, not argued, and this paper argues the opposite case: the complex, internal-plus-input explanation is the illuminating one.
What the critics show and do not. The paper shows that the 1998 argument from kinds needs a kind that does causal-explanatory work, and that fine-grained memory kinds do not cover notebooks. It does not show that no coarse kind is useful. Its case there is a challenge plus conservatism. Chalmers’s later answer to its first objection is one line, calling it “an implausible human chauvinism about cognition” (LIT-415 §2, citing Rupert’s 2009 book); Clark’s book answers it under the chapter title “HEMC, HEC, HEMC…” (LIT-412, unread).
On consciousness and on groups. The paper keeps consciousness out except at one point. In footnote 38 it argues that making conscious endorsement itself extended would need “an extended theory of conscious acts”, and would face the same bloat. It sets socially distributed cognition aside, saying that with group mental states “many of us, quite rightly, lose our bearings”, and it notes that corporate bodies have neither “a plausible theory of consciousness” nor “paradigm cases of conscious acts”. That is a point about extended belief in groups, and it bears on THEORY-043 only as a reason why group agency is easier to grant than group endorsement. It does not discuss phenomenal group consciousness.
It carries no instruction for machine-learning practice. Its §IX argues that AI research aims at self-contained systems that work in many environments, as a reason to doubt that extended systems are cognitive science’s proper objects. That is a claim about the aims of AI as of 2004, not a recommendation for practice.