Contribution
The paper is the first reply by either author of LIT-097 to the published criticism. It sets the terms of the later debate in three ways. It names “the Parity principle”. It says the thesis concerns material vehicles, not contents (note 1). And it recasts the argument as a challenge: show why Otto’s notebook is not just an unusual realiser. It also introduces the “ecological control” picture of the brain that Clark’s book develops.
Key insight
The critics granted parity and then looked for something the notebook lacks. Each candidate (intrinsic content, active updating) is either missing from some states we already count as cognitive, or would not matter if it were missing inside the head. So the question returns to parity, and parity is meant to “command rational assent” as a release from “bio-chauvinistic prejudices”.
Assumptions
- Vehicles, not contents. Having a contentful state is a property of a whole active system. The thesis concerns the material aspects that enable it (note 1).
- Parity is not argued. “We did not argue for this principle, except by providing illustrations and defences against counter-arguments” (p. 2).
- Functionalism with hybrid realisers. Following Weiskopf, functionalists “have long been open to unusual realizers”, including silicon-biological hybrids (p. 3).
- Content is “as content does” (Prinz and Clark 2004). Intrinsic content is “essentially fuzzy and indistinct” (p. 4).
Key results
- Parity, restated (p. 2). “If, as we confront some task, a part of the world functions as a process which, were it to go on in the head, we would have no hesitation in accepting as part of the cognitive process, then that part of the world is (for that time) part of the cognitive process. (Clark and Chalmers 1998: 8)”. Note 3 calls it a “veil of metabolic ignorance”.
- Intrinsic content, first reply (p. 4). It is unclear the notion is intelligible. Where it is, nothing makes it neural: artificially evolved creatures, and a silicon neuron that worked within 14 biological neurons of a Californian spiny lobster (Szucs et al. 2000), suggest silicon can carry it.
- Second reply (p. 5). In-the-head problem solving with an imagined Venn diagram uses a conventional representation, and is cognitive. For the harder case of Otto’s dormant entry: Martians who store bit-mapped images of text in biological memory would count as remembering, so Otto should. “All that matters, for some conventional encoding to count as the vehicle of a dispositional belief, is that it be appropriately linked, at run-time, to representations whose content is … intrinsic.”
- Third reply (pp. 5–6). Adams and Aizawa concede (2001: 50) that it is unclear how much of each cognitive state must involve non-derived content. “This concession, I submit, removes the entire force of the appeal to intrinsic content”, since EM never claimed a cognizer made wholly of notebooks.
- Active memory (pp. 6–7). The worry (O’Brien, Dartnall, Weiskopf): biological memory revises itself and propagates updates, and a notebook does not. Reply: a system that updated only at recall time (“compilation” rather than “interpretation”) would be “every bit as rational as we are”. Otto updates at run-time too. Concession: the flow of revision is “restricted and local”, and “the notebook’s lack of inbuilt links and pathways makes it unlike biological memory”.
- The weakened claim (p. 7). Otto-and-the-notebook “exhibit enough of the central features and dynamics of a normal agent having … the dispositional belief”. A world in which face memory were never reconstructive would still count those memories as cognitive, so the static notebook can count too.
- The larger picture (pp. 7–10). Against Dartnall’s spade (“my-spade-and-I do not get the prize”), the line between a constantly available tool and a limb is “fuzzy at best”. Macaque rake use (Iriki et al. 1996) and stick-induced near-space remapping (Berti and Frassinetti 2000) show tools entering body maps. Brains are “open-ended controllers” practising “ecological control”. Note 7 reports MIT “memory glasses” that cue names below conscious awareness and improve memory in mildly Alzheimic patients “by about 50%”, credited to Chalmers. Persons are better identified with Rovane’s “rational point of view”, and their identification with the biological is a Lockean “forensic matter”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | The extended-mind thesis concerns vehicles, not contents | assertion (clarification) | note 1 |
| C2 | No non-question-begging notion of intrinsic content picks out all and only the neural | moderate | silicon-replacement argument; lobster neuron (one artificial neuron, cited) |
| C3 | A conventional encoding linked at run-time to intrinsically contentful states can be a belief’s vehicle, inside or outside the head | informal argument | Venn diagram; Martians (parity) |
| C4 | Adams and Aizawa’s p. 50 concession removes the force of their objection | moderate | the concession is quoted accurately in substance |
| C5 | When memory integrates new information is below the level that matters for belief | informal argument | ”compilation” thought experiment; Dennett’s left-handers |
| C6 | Otto-plus-notebook has enough of a believer’s features, though unlike biological memory | moderate as a reformulation | conceded differences; face-memory counterfactual |
| C7 | Brains are open-ended controllers that assimilate tools | suggestive evidence | Iriki et al. 1996; Berti and Frassinetti 2000; no extension from body maps to belief is argued |
| C8 | Persons are forensic, not biological, items | assertion | Locke and Rovane, one paragraph and notes 9–10 |
Concepts
- material vehicle — the enduring material aspect of a system that enables it to have a contentful state.
- Parity principle — the name given here to LIT-097’s §2 sentence.
- active dispositional memory — the critics’ point that biological memory is integrative and reconstructive.
- ecological control — conscious control that devolves detail to non-conscious mechanisms, body mechanics and external devices.
- open-ended controller — a brain that computes, moment to moment, which resources are available and under its control.
Connections
- LIT-097 (NOTE-180). The paper is Clark’s restatement of the 1998 argument, with new parity cases. It confirms NOTE-180’s finding that parity was stated, not argued, in its own words. On the quotation, see the corrections below.
- Adams and Aizawa (LIT-438). Answered directly. The paper accepts their framing (a mark of the cognitive) only so far as to deny that non-derived content draws the line.
- Rupert (LIT-422). Not mentioned. The active-memory section answers the same kind of argument, a functional difference between notebook and memory, and its reply (differences “below the level”) is the position Rupert’s §VII argues does no explanatory work without a generic kind that earns its keep.
- Chalmers. Thanked for exchanges on drafts and credited with the memory-glasses example. The 2019 chapter (LIT-415) compresses this paper’s reply into one line (“the content of items in biological memory is derived from relations to conscious states”), which is a different reply from Clark’s.
- Persons. Floridi’s informational self (LIT-136) and Register’s body parts (LIT-138) are this record’s other accounts of persons beyond the biological boundary.
Bearing on the record
- THEORY-043. Indirect only. Rovane’s “rational point of view”, which Clark favours for persons, comes from a book (not held here) that is usually cited for allowing group persons. That bears on the agency half of the account, which the readings there already grant, and not on phenomenal consciousness, which the paper does not discuss.
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
- Imagined cases throughout. Martians, compiled updating and non-reconstructive face memory are parity arguments, and their force is parity’s.
- The empirical material is suggestive. One silicon neuron in a lobster ganglion, and body-map remapping from tool use, are about sensorimotor integration, not belief.
- It answers Adams and Aizawa’s first mark, not their second. Their claim that the processes differ in kind is met only by the active-memory concession and the “below the level” reply.
Open questions
- What is the level “to which the philosophical functionalist should attend”, and does memory science’s level (interference, learning curves) fall below it? Rupert says no.
- Does the “rational point of view” criterion for persons extend to groups, and if so, what does it say about group consciousness?
Corrections
- none to a seeded skim (there was no seed)
- The parity quotation. Clark quotes LIT-097 (p. 8) as ending “that part of the world is (for that time) part of the cognitive process”, with “were it to go on in the head” and “accepting as”. NOTE-180, reading the authors’ web version, gives “were it done in the head”, “recognizing as” and “(so we claim)”. Adams and Aizawa (LIT-438, p. 46) quote the same wording as NOTE-180, with an ellipsis where the hedge stands. So this paper’s version is a paraphrase presented as a quotation, or it follows a typeset text that differs from the web version; I could not check the typeset 1998 text. It should not be cited as the 1998 wording.
- The concession, quoted. Clark quotes Adams and Aizawa as “it is unclear to what extent every cognitive state of each cognitive process”; their text (p. 50) has “each cognitive state”. The sense is the same.