Source
Rupert (2004), “Challenges to the Hypothesis of Extended Cognition”, LIT-422, read in NOTE-343 from the author’s version.
What was actually shown
Two hypotheses (§II). HEMC, embedded cognition: cognitive processes “depend very heavily, in hitherto unexpected ways, on organismically external props”. HEC, extended cognition: the unit of analysis is organism plus environment, and cognitive states “literally comprise—as wholes do their proper parts—elements in that environment”. Inferring HEC from HEMC by dependence is invalid. Understanding Nazi Germany’s invasion of Poland needs 1920s German economics, which is not part of the invasion (NOTE-343, C2, strong).
The past-endorsement dilemma (§IV). Directory assistance satisfies the first three of LIT-097’s criteria, so without the fourth Otto believes every listed number: “It is absurd”. With it, it is arbitrary to require endorsement for extended but not internal belief, and it makes the organism’s consciousness “the ultimate source of cognitive authority”, so the notebook is “best accommodated by HEMC” (C4 strong, C5 moderate).
Memory differs (§§V–VI). Holding a model of a long conversation needs internal working memory, and keeping it on paper would “destroy the dynamics of normal conversation”. In A-B/A-C paired-associate learning, internal memory shows interference; a written list shows none, and no learning curve. The generation effect has no external analogue. These are differences in learning time and access time, “at the very heart” of memory research.
No useful generic kind (§VII). A kind like “has some sort of access to the information that P” would cover notebooks. But internal systems share power laws of learning and forgetting and Rescorla-Wagner learning that external stores lack, and psychologists who split memory into systems tend to drop the superordinate kind, not widen it.
Functionalism does not help (§VIII). Commonsense functionalism forbids seeing one’s own memory. Psychofunctionalism takes its roles from memory science, whose roles external stores do not fill.
What could have come out otherwise. An external store could show interference or a generation effect; the paper argues that one does not, by thought experiment.
What this does not say
- That no coarse kind could be useful. The case against a generic kind is a challenge plus conservatism, not a demonstration (NOTE-343, Limitations; C7 weak to moderate).
- That external stores were tested. No experiment on external memory is cited or run. The external side of each comparison is argued (“there is no reason to think…”).
- That HEMC wins without conservatism. The second horn of the dilemma needs the conservatism principle to get from “the organism is privileged” to “prefer HEMC”.
- Anything about consciousness or groups. Socially distributed cognition is set aside (§I). Note 38’s remark that corporate bodies have no plausible theory of consciousness is a point about endorsement, not about phenomenal group consciousness.
- That it is a claim about AI practice. §IX’s appeal to AI’s aims is a 2004 claim about what the field wants, offered as weak support (C9).
Connections
- Rival. Chalmers’s sensorimotor statement (THEORY-512) answers the “different psychological profile” objection, citing Rupert’s 2009 book, in one line: “implausible human chauvinism”. Clark’s book (LIT-412, unread) takes Rupert’s labels for a chapter title.
- The coupling point. Rupert’s dependence-is-not-constitution argument reaches a conclusion close to Adams and Aizawa’s (THEORY-582) from a different premise.
- Adams and Aizawa. Rupert’s n. 87 says this critique was developed independently, rejects their derived-content argument as question-begging, and agrees that internal and external processes have different cognitively relevant properties (THEORY-591).
- The extended self. Note 6 says criticising HEC “strike[s] a blow” to the extended self of LIT-097 §5; Floridi (LIT-136) and Register (LIT-138) are the record’s other readings on that claim.