Source
Gentner (1983), LIT-1810, pp. 156–168; read in NOTE-1349 from a scan of the published article.
What was actually shown
The rules (p. 158): “Discard attributes of objects”; “Try to preserve relations between objects”; and “(The Systematicity Principle) To decide which relations are preserved, choose systems of relations”. Attributes are one-place predicates and relations two-or-more-place ones; a higher-order relation takes propositions as arguments (p. 157). The principle: “A predicate that belongs to a mappable system of mutually interconnecting relationships is more likely to be imported into the target than is an isolated predicate” (p. 163).
The argument. In the Rutherford analogy, MORE MASSIVE THAN (sun, planet) is mapped and HOTTER THAN (sun, planet) is not, because only the former belongs to the system of relations tied together by the equations of central-force motion (pp. 162–164). In “heat is like water”, a causal scenario is preferred to a static one because causal knowledge is represented as a more deeply embedded structure (pp. 164–165). The analysis of the 3:6::2:4 analogy shows feature overlap is irrelevant (p. 156).
The evidence. Interpretations written for comparisons such as “A cigarette is like a time bomb” were judged relational or attributional: “Relational information predominates over attributional information in analogy interpretations, but not in object descriptions generated by the same subjects”, and aptness ratings went with the amount of relational information written (pp. 165–166). No numbers are reported; details are in a conference paper (Gentner 1980b).
What could have failed. Interpretations of analogies could have been as attribute-laden as descriptions of the objects; they were not, by the one study’s report. Nothing in the paper could have shown systematicity false: no experiment varies a predicate’s place in a higher-order system.
What this does not say
- That the attribute/relation split is fixed by the world or by logic. “Logically, a relation R(a,b,c) can perfectly well be represented as Q(x) … Psychologically, the representation must be chosen to model the way people think about the domain” (n. 4, p. 157). The claim is about people’s representations, and it inherits whatever choices built them.
- That systematicity determines the mapping. The text’s own statement is a preference (p. 163); “determined by” in the abstract overstates it.
- That a good analogy is a true one. Acceptance depends on the plausibility of the candidate inferences, “a separate matter” (n. 11, p. 165).
- That matching one relation’s values is structure-mapping. An alignment of two spaces by a single distance relation, as in Gromov–Wasserstein matching (LIT-1818), meets the first two rules for that one relation and has no higher-order relations, so it has no analogue of the systematicity principle (NOTE-1349).
- A processing model. The order of processing “is clearly variable” (n. 10, p. 164), and no algorithm is given here.