Source
Chalmers (2019), “Extended Cognition and Extended Consciousness”, LIT-415, read in full in NOTE-356 from the author’s copy. Chalmers (2008), the foreword to Clark’s Supersizing the Mind, LIT-421, read in NOTE-369. Both revise Clark and Chalmers (1998), LIT-097, read in NOTE-180.
What was actually shown
The 1998 thesis is too weak (LIT-415 §1). LIT-097 “never quite state[s] a single official thesis”. Reconstructed, it is constitution by external entities “when those entities play the right role in driving cognitive processes”. Following Farkas (2012), that is “near-obviously true in extended circuit cases”, such as Clark’s Diva, whose damaged division circuit is replaced by an external silicon one. The chapter says the main opponents accept such cases. It does not quote them saying so (NOTE-356, C2).
The proposed thesis (LIT-415 §1). Constitution “in virtue of the subject’s sensorimotor interaction with these entities”. It counts Otto and Tetris in, and Diva, brain-computer interfaces and intravenous stimulation out, with telepathy borderline. Chalmers calls it “extensionally correct” for the paradigm cases and offers it as “the retrospectively official statement”. It has no necessary-and-sufficient test (NOTE-356, C3).
The strongest objection, kept (LIT-421; LIT-415 §2). If the mental lies between perception and action, Otto’s notebook is outside it, since Otto must read and write. Both texts withdraw the 1998 reply that Otto’s access “is not perceptual at all”: “there is no denying that Otto sees the notebook and reads from it” (LIT-421). The reply Chalmers favours is revisionism. If an extended state functions in explanation as belief does, it is belief. Whether it counts as extended can flip with the question asked, like a Necker cube (LIT-421, C4). The objection is “not a knockdown”, and neither is the reply.
Consciousness does not extend this way (LIT-421; LIT-415 §3). The foreword’s reason is that no Otto-style twin case, internal duplicates who differ in experience, can be built (LIT-421, C6). It floated a bandwidth explanation (C7). The chapter withdraws that, granting Vold (2015) that vision is a very high-bandwidth channel. In its place it puts directness: consciousness needs information directly available for global control, and direct access is “a relatively short and unmediated causal pathway” that “requires access that is not mediated by other mental states”. Standing belief needs only two-step availability, which Otto’s perception supplies. Hence “even if there is extended cognition, there is no extended consciousness.”
What could have come out otherwise. The reconstruction could have found the 1998 thesis already controversial in the circuit cases, or a twin case for consciousness could have worked. The first is settled only by an unquoted claim about the opponents; the second by an unshown search.
What this does not say
- That consciousness cannot extend. An extended circuit could carry consciousness, since directness is no obstacle where access is not perceptual (LIT-415, C10), and its physical basis could extend “in some possible worlds” (LIT-421). The denial concerns notebooks, phones and screens.
- That directness is defined. The author has “no necessary and sufficient conditions for directness”. Unconscious perception might escape the mediation argument, and a verdict there “would require more detailed criteria for directness than I have given here”.
- That the consciousness criterion is argued here. It is imported from The Conscious Mind and the neural-correlate paper (NOTE-356, Assumptions). It is a correlate, offered to explain an independently held verdict (“there seem to be no plausible cases”).
- That the critics are answered. LIT-415 §2 gives five objections one sentence each. NOTE-356 rates those replies weak (C5). The coupling-constitution reply assumes that parity carries the argument, and NOTE-180 found parity stated in LIT-097, not argued.
- Anything about the hard problem. “Partly constituted” is read as partial grounding, or as nomological dependence for a dualist. The foreword says the thesis needs only the parity principle’s “very weak functionalism” and is compatible with dualism.
Connections
- Rivals. Adams and Aizawa’s contingent intracranialism (THEORY-591) and Rupert’s embedded cognition (THEORY-528) both deny that Otto’s notebook is part of his cognition. Their shared logical point, that coupling does not by itself make parthood, is stated separately (THEORY-582); this account accepts it and rests on parity instead.
- Group consciousness. THEORY-043 names the directness criterion as the most promising candidate the record holds for an architectural line between a society of mind and a nation. That application is the record’s, not Chalmers’s (NOTE-356, Bearing on the record).
- AI and tools. Direct availability for global control is kin to the global-workspace indicators in LIT-056 and so to THEORY-023’s dispute. On this account a user’s phone or assistant, reached by perception, can at most extend belief.
- Implants. A brain-computer interface falls with the extended circuits, which bears on Register’s account of BCIs as body parts (THEORY-540, LIT-138).
- Clark’s book (LIT-412) is filed unread.