Contribution
This is the first sustained philosophical reply to LIT-097 and to the wider “transcranialist” camp (Dennett, Donald, Hutchins). It moves the debate from where cognition is to what cognition is. The bounds of cognition “must be determined by finding the mark of the cognitive, then seeing what sorts of processes in the world have the mark” (p. 46). It proposes two necessary marks, and it gives the “contingent intracranialism” position a name.
Key insight
Grant parity, then ask whether anything outside the head actually does what the head does. On their view nothing does: notebook marks mean what they mean by convention, and the physics of a screen or a pencil is nothing like the processes in a brain. Like muscle contraction and unlike some digestion, cognition happens to stay inside the body.
Assumptions
- The mental is not the cognitive. “We assume without argument that the mental is not the same as the cognitive” (p. 48). Paradigm cognitive processes are learning, remembering, sensing, perceiving and thinking.
- Consciousness is inessential to cognition. “We assume without argument that there can be cognitive processes that lack associated quale and are not phenomenally conscious” (p. 48).
- Mark 1: non-derived content (§2.1). Cognitive states “must involve intrinsic, non-derived content”. The support is a claimed “fairly broad consensus” (Dretske, Fodor, Searle, Millikan) and arguments against Dennett’s stance view and against purely syntactic parsing.
- Mark 2: causal individuation (§2.2). Sciences partition phenomena into causally homogeneous kinds, so cognition “must be discriminated on the basis of underlying causal processes”. The analogies are Bacon on heat and neo-Lamarckians confusing inheritance with infection. A look-up table passing the Turing test, and chess programs’ search, are behaviour without cognition.
- Cognitive science is immature. “Relatively little can be said about the difference between the cognitive and the non-cognitive” (p. 53). The confidence that a processing condition exists comes from what mature sciences have found.
Key results
- Parity granted (p. 46). They “don’t propose to challenge” the parity principle. It shows only that “being inside the brain cannot be the mark of the cognitive”, which “seems to us to be true and obvious”.
- Contingent intracranialism (pp. 46–47, 53). Transcranial cognition is “logically and perhaps nomologically possible”, but as “boring contingent empirical fact” there is none. Cognition is like muscular contraction, which stays in the arm when shears are used, and unlike some digestion, which is transcorporeal in spiders and in coprophagous rabbits and chickens.
- Tetris (§3.1). Case (1) uses mental representations of the blocks. In case (2) the blocks on screen “are not representations at all … they are the blocks to be fit together”, and button-driven rotation is a circuit and a CRT, “surely not the same as any cognitive process”. Case (2) adds decision, attention, memory and motor processing that case (1) lacks. Case (3) is underdescribed: it can be aligned with (1) or with (2), not with both.
- Otto and Inga (§3.1). Otto’s notes have “merely derived content”, so they “do not … constitute beliefs or memories”. Otto’s retrieval uses motor and visual processing that Inga’s lacks, and Inga uses a capacity Otto has lost. The differences cannot be unimportant “save by adopting a behavioral conception of cognition”.
- The coupling argument (§3.2). Quoting LIT-097’s coupled-system passage, they reply that causal coupling does not extend a process: the expansion of a thermostat’s bimetallic strip stays in the strip, and filtration stays in the kidney, though both are coupled to more. They grant that Clark and Chalmers “are under no illusions” about this and actually claim the external processes are like the internal ones. That claim, they say, begs the question, and is false. Without a mark of the cognitive, coupling threatens “pancognitivism”, called “cognitive bloat” or “cognitive ooze”.
- Explanatory simplicity (§3.3). Easier explanations are not better science (“one’s plants are thirsty”). “Isomorphism is a relatively weak relation”: melting and evaporating are isomorphic and distinct.
- Donald and Hutchins (§§3.4–3.5). Donald’s own table of engrams versus exograms (Table 1) lists the processing differences that defeat his claim. Hutchins studies “naturally occurring computation, rather than cognition”. Even a perfect computational duplicate of the brain’s program would lack cognition unless its states had non-derived content.
- No science of tool use (§4). Functional kinds do exist in science (heritability, oscillators), but “there are no laws covering humans and their tool use over and above the laws of intracranial human cognition and the laws of the physical tools”. Memory aids, perceptual instruments and computing devices are each “an unscientific motley”. This is offered as speculation and as “our bet”.
- Not a verbal dispute (§4). They define “cognition_cs” (non-derived content plus processing) and “cognition_r”, the radical notion, and predict that only the first will yield a natural science.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Being in the brain is not the mark of the cognitive; the parity principle is true | assertion | ”true and obvious”, p. 46 |
| C2 | Cognitive states must involve non-derived content | weak to moderate | claimed consensus; replies to Dennett and to syntactic parsing; “we assume” |
| C3 | Cognitive kinds are individuated by underlying causal processes | moderate as a philosophy of science | analogies (heat, inheritance, look-up tables) |
| C4 | Actual external aids lack non-derived content | assertion | conventional meaning of marks; note 5 on natural meaning |
| C5 | Otto’s and Inga’s retrieval differ cognitively in ways that matter | moderate | lists motor, visual and lost-capacity differences; “not unreasonable to say” |
| C6 | Causal coupling does not extend a process | strong as a general point | thermostat and kidney counterexamples |
| C7 | Clark and Chalmers have no mark of the cognitive, and without one coupling leads to bloat | moderate | §3.2; LIT-097 indeed offers none |
| C8 | Human-plus-tool systems will support no science | speculation | ”Slim to none, we speculate”; lists of tools |
| C9 | Cognition is brain-bound as a contingent fact | conclusion | rests on C2–C5 and C8 |
Concepts
- mark of the cognitive — the conditions that separate cognitive from non-cognitive processes. The paper offers two necessary conditions, not a theory.
- non-derived (intrinsic) content — meaning a state has not by convention or social practice.
- transcranial / extracranial / transcorporeal cognition — the labels for the view opposed. “Extended mind” is not their term.
- contingent intracranialism — cognition could extend and does not.
- cognitive bloat (“cognitive ooze”) — the slide to pancognitivism if coupling suffices.
- cognition_cs vs cognition_r — the commonsense notion and the radical one.
Connections
- LIT-097 (NOTE-180). The paper quotes LIT-097’s Tetris cases, the parity sentence, the coupled-system passage, the burden-shifting passage and the Scrabble passage, with page numbers 1–6 that match no published pagination (they appear to be a preprint’s). It supplies what NOTE-180 found missing: an explicit challenge to the move from coupling to constitution. NOTE-180 names it as the “coupling–constitution fallacy” objection; that phrase is not in this paper.
- Clark 2005 (LIT-403). The reply argues against Mark 1 with the silicon-neuron case, conventional mental imagery and Martians, and uses the concession on p. 50 against the authors.
- Rupert 2004 (LIT-422). Rupert’s note 87 says his critique was developed independently, rejects the derived-content argument as question-begging, and agrees with the worry that internal and external processes have different cognitively relevant properties (pp. 58, 61–62 here).
- Chalmers 2008 and 2019 (LIT-421, LIT-415). The foreword dismisses “the fact that external cognitive processes work differently” without naming anyone. The chapter answers the derived-content and coupling-constitution objections in one line each, citing the authors’ 2008 book.
Bearing on the record
- THEORY-043. The paper takes consciousness off the table by assumption, so it says nothing about phenomenal group minds. Its natural-kind argument (C8) would, if sound, count against treating organised groups as cognitive systems, but it is not applied to groups, and List (LIT-401) grants groups agency on different grounds.
- Mereology. C6 is the paper’s contribution to the parts question: coupling is not parthood. It offers no account of parthood for processes either.
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
- Both marks are assumed. The authors say so, and they say little about what non-derived content is.
- The processing differences are asserted, not measured. The paper cites no experiment on external memory. Rupert (LIT-422) later supplies the empirical version.
- The “no science” argument is a bet. The authors call it one.
- The concession on p. 50 (it is unclear how much of each cognitive state needs non-derived content) weakens the derived-content objection against a hybrid like Otto-plus-notebook. Clark exploits it.
Open questions
- Is there a non-question-begging test for non-derived content that a silicon replacement for a neuron passes and a notebook fails?
- Would a cognitive science of human-plus-tool systems find law-like regularities? The paper bets not; nobody here has measured it.
Corrections
- none to a seeded skim (there was no seed)
- NOTE-180’s attribution. NOTE-180 says the reply that coupling is only causal “was pressed later, notably as the ‘coupling–constitution fallacy’ objection of Adams and Aizawa”. The substance is in this 2001 paper (§3.2). The name is not; it is in the authors’ later work (cited by LIT-415 to Adams and Aizawa 2008). Reported here, not edited there.
- The brief’s gloss. The brief describes this paper as “the coupling–constitution and ‘mark of the cognitive’ critique”. The mark of the cognitive is right. The coupling–constitution label is the later name for its §3.2 argument.