Contribution
The second round of Adams and Aizawa’s exchange with Clark, after LIT-438 and Clark’s replies. It adds one new thing: a name and a pattern for the objection that coupling is not constitution, which LIT-438 argued without naming. The rest restates and clarifies the 2001 marks of the cognitive against Clark’s criticisms. The authors say as much: they are “disappointed that we were unable to convey our objections clearly enough to forestall Clark’s criticisms”.
Key insight
The opening joke carries it. “Question: Why did the pencil think that 2 + 2 = 4? Clark’s Answer: Because it was coupled to the mathematician.” If being coupled to a cognitive agent in the right way is what makes a thing cognitive, then the thing’s own nature does no work, and that is the mistake. What makes X part of a cognitive process has to be something about X: a mark of the cognitive.
Assumptions
- The extended-mind case runs on coupling. The draft reads Clark as holding that “the nature of the processes internal to a pencil … has nothing to do with whether that thing carries out cognitive processing”, and grants that Clark “does shy away from this from time to time” (n. 1).
- The 2001 marks. Cognition involves non-derived content, and is a kind of causal processing that, as a contingent empirical fact, happens only in the brain. Neither mark is argued afresh here; both are taken from LIT-438.
- Naturalistic content for mental representations. Mental representations of trees and rocks mean what they do “in virtue of satisfying some naturalistic conditions on meaning”, for which Fodor’s asymmetric causal dependence is offered as one candidate. No theory is defended.
- Scientific orthodoxy as evidence. The authors “side with what is by all accounts scientific orthodoxy”, and take that to give prima facie support.
Key results
- The coupling-constitution fallacy (opening section).
- Pattern: draw attention to cases where an object or process is coupled to a cognitive agent; “slide to the conclusion that the object or process constitutes part of the agent’s cognitive apparatus or cognitive processing”.
- Counter-case: the neurons into a neuromuscular junction are coupled to the muscle, but not part of it; neurotransmitter release is coupled to contraction, but not part of that process.
- Called “the most common mistake that extended mind theorists make”, charged to Van Gelder and Port (1995), Clark and Chalmers (1998), Clark (2001), Gibbs (2001) and Haugeland (1998) (n. 3).
- A fuller treatment is deferred to “Adams & Aizawa (in progress). Problems for process externalism”.
- Intrinsic content (§ “The Intrinsic Content Condition”).
- The condition has “calculated openness”: cognitive states must involve non-derived content, not consist of it. A Turing machine’s non-representational head states are offered as a reason (n. 6).
- Clark is said to attack both a too-strong reading (all content intrinsic), which they “explicitly rejected”, and a too-weak one (vacuous). Their reading: a process with no intrinsic content is non-cognitive, as Tetris blocks on a screen “are not representations of blocks to be rotated, they are the blocks to be rotated”.
- Euler circles: the convention that gives meaning to circles on paper is “logically separate” from what makes a mental image of them mean what it does. The dependence on a prior convention is historical, like a mental image of a car depending on the car’s invention.
- Against convention fixing mental content: a convention needs access to “the would-be syntactic item” it gives meaning to, and “we have no way to identify particular tokens of brain states qua syntactic items”.
- Causal processing (§ “The Causal Processing Condition”).
- First flaw (armchair): the authors deny doing armchair philosophy. That a brain-tool science is possible does not touch a claim about what the evidence shows. Clark offers no regularity they missed.
- Even granting information-processing regularities across the skull: DVD players, radios and phones process information differently from brains, which is why brains do language and face recognition and they do not. What Clark needs is “a cognitive level of description”, not just a unified level such as physics.
- Tools “do not constitute a natural kind”, so a brain-tool science would have to cover “an open disjunction”, “a motley”, not “a multiplicity”.
- Second flaw (cognition may itself be a motley): intracranial processes may be many, but genome and development bound them; the brain is “at least in the running to be a natural kind”. And even a motley intracranial cognition, if distinct from extracranial processes, would still make cognition intracranial.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Inferring constitution from coupling is a fallacy, and the commonest error of extended-mind theorists | moderate for the logical point; weak for “commonest” | one counter-example by analogy (neuromuscular junction); the attributions in n. 3 are not argued case by case here |
| C2 | Cognitive states must involve, not consist entirely of, non-derived content | assertion, clarified | restates LIT-438 pp. 48–50, which assumed it |
| C3 | A mental image of Euler circles need not get its content by convention | moderate | a distinction between the diagram’s and the image’s content; a positive account is only gestured at (Fodor 1994) |
| C4 | Conventions cannot fix the content of brain states, because brain-state tokens are not accessible as syntactic items | moderate, and dated | stated relative to “the state of current science”; the authors concede science-fiction cases |
| C5 | Brains process information by different principles from brain-tool combinations | assertion | an inventory of consumer electronics and what they cannot do; no study cited |
| C6 | Tools are an open-ended motley, so no brain-tool science is to be expected | weak to moderate | the artifact/natural-kind contrast; LIT-438 pp. 60–61 on fragmented sciences |
| C7 | Even a motley intracranial cognition would leave cognition intracranial, if it is distinct from extracranial processes | follows from the definition | a conditional; the antecedent is what is in dispute |
Concepts
- Coupling-constitution fallacy. Inferring that X is part of a cognitive process or system from X’s being causally coupled to it.
- Mark of the cognitive. A criterion that says what makes a process cognitive rather than non-cognitive; the authors’ is non-derived content plus a distinctive causal process.
- Non-derived (intrinsic) content. Content a state has not by convention or by an interpreter’s say-so.
- Multiplicity vs. motley. Many distinct principles in a bounded domain, against an open-ended collection with nothing in common. Only the second defeats a science.
- Intra-, extra- and transcranialism. Cognition inside the brain, outside it, or spanning the boundary.
Connections
- LIT-438 (Adams and Aizawa 2001). The paper defended. Its §3.2 argument that “the mere causal coupling of some process with a broader environment does not … extend that process” is what this text names a fallacy.
- LIT-097 (Clark and Chalmers 1998). Charged with the fallacy (n. 3), and its pencil-and-paper arithmetic (p. 8) is the target of the opening joke (n. 2).
- LIT-403 (Clark 2005). Not the text answered, but it shares the Venn-diagram and Martian cases with “Memento’s Revenge”. The Euler-circle reply here is the record’s only direct answer to LIT-403’s use of the Venn case.
- LIT-422 (Rupert 2004). Not cited. Rupert’s embedded-cognition alternative makes a similar point about coupling from a different premise, conservatism.
- LIT-415 (Chalmers 2019). Answers “the coupling-constitution objection” in one line, citing the 2008 book.
Bearing on the record
- NOTE-180 found LIT-097’s step from coupling to parthood unargued. This text gives that finding its standard name, and the objection a counter-example. It does not add an argument that coupling never suffices, only that it does not entail parthood.
- Groups. The motley argument (C6) is about artifacts. It is not applied to groups of people, and organisms are not artifacts, so it does not by itself count against a nation as a cognitive system (LIT-159).
Limitations
- A draft. The published chapter may differ, and was not seen.
- The fallacy is shown by a single analogy, and the attributions to five authors are not supported in this text.
- The processing contrast rests on everyday examples of devices, not on cognitive science.
- The marks of the cognitive are restated, not argued; the text’s force is as a clarification of LIT-438.
Open questions
- Does the published chapter keep the name and the n. 3 list of offenders?
- Where was the name first printed: the 2008 book, or an earlier paper such as the “Problems for process externalism” announced here?
Corrections
- NOTE-180 says the name “coupling–constitution fallacy” is from the authors’ “later work (2008)”. The name is older than 2008 in circulation: this draft, which predates December 2005, uses it. 2008 may still be its first appearance in print; the draft does not settle that.