Source
Dennett (1971), “Intentional Systems”, LIT-440, read in NOTE-350. Dennett (1980), “Reply to Professor Stich”, LIT-456, read in NOTE-371 from the author’s final draft.
What was actually shown
The stance (LIT-440 pp. 87–90). An Intentional system is one whose behaviour can be predicted by ascribing beliefs and desires and assuming rationality. Rationality “so far means nothing more than optimal design relative to a goal or optimally weighted hierarchy of goals” and constraints (p. 89). Intentional predictions fail on “design weaknesses and lapses”, and consistently failed predictions push the predictor to the design stance, as with the insane (pp. 93–96).
Belief’s two norms (LIT-440 pp. 101–103). Normally, if x believes p, p is true; normally, if x avows p, x believes p. Hence “a false belief system is a conceptual impossibility” (p. 101). The norms break the belief–desire circle “in the long run and in general” (p. 103). NOTE-350 rates this moderate (C7): an evolutionary argument plus a conceptual claim.
Irrational avowals (LIT-440 pp. 103–106). When avowals are irrational, leaning on rationality overrides the speaker and leaning on avowal grants him irrational beliefs: “there is no coherent Intentional description of such an impasse”. Neither phenomenology nor neurophysiology breaks the tie, since both presuppose a rationality standard. An “Official Tolerance Rule” for error would confuse “the norm they follow with what gets by in their world” (C8, weak to moderate).
Contradiction (LIT-456 p. 74). Following Quine, evidence for ascribing the belief that p is the best evidence against ascribing a contradictory belief that q. Researchers on inference who ascribe contradictory beliefs uncautiously “will indeed be confused — not ‘simply’, but subtly”. Ordinary practice tolerates the slack (“he sort of believes that p”) at a cost, notably in judgments of responsibility (NOTE-371, C3, moderate).
The concession (LIT-456 p. 75). Against Stich’s charge that a norm of full rationality would disqualify every natural believer: “The conception of ourselves as intentional systems is granted to be an unrealizable idealization”, and so is the concept of a person built on it. It is the idealisation that figures in ethics (C4, stated, not argued).
What could have come out otherwise. A stable intentional description of a creature with contradictory beliefs could have been given. The argument says it cannot be, from the interdependence of the evidence; no case is examined.
What this does not say
- That people are rational. The account says ascription assumes it. Dennett grants that no one is fully rational.
- That experiments on reasoning error find nothing. The reply says they find something other than “unproblematic cases of people believing contradictions”, without saying what (NOTE-371, Open questions).
- That a norm-free theory of belief is impossible. Such a theory is “not ruled out”, but it would revise our concepts, and Dennett would accept it only if it kept most stance-based ascriptions (LIT-456).
- That the evolutionary grounding is worked out. The 1980 reply admits it is left “open-ended”, “one of those disconcerting bits of soft-pedalling of which Stich rightly accuses me”. Its reliance is minimal: belief structures are extensions of control systems “selected by nature because they are benign in the long run”.
- That the stance is observer-relative. The 1971 paper says so (C1). “True Believers” (LIT-398) withdraws the relativity; this account does not depend on either position.
- Anything about pain or consciousness. The stance is “unmetaphysical” about consciousness (LIT-440 p. 100), and the 1980 reply gives pain no such protection, offering policy instead of theory.
Connections
- The intentional stance itself. “True Believers” (LIT-398, NOTE-353) is the record’s full reading of the stance as a criterion for having beliefs. This account takes only the rationality presupposition from it and its 1971 predecessor.
- Why agency is cheap to agree on. THEORY-043 uses the stance’s blindness to composition and consciousness (LIT-440, C2) to explain why every reading it cites grants agency to wholes. The idealisation conceded here is part of why that grant is cheap (NOTE-371, Bearing on the record).
- Behaviour is not enough. Block’s argument (THEORY-092) that behavioural capacity does not fix intelligence draws on the same intentional-system papers for the contrast.
- Loans of intelligence. LIT-440’s other thesis, that any content- ascribing theory takes out a loan repaid only by analysing away its readers (p. 96), is a methodological rule and is not stated here. Where it continues, “Artificial Intelligence as Philosophy and as Psychology” (LIT-393), is filed unread.