Cited by 883Language, Metaphor, and CognitionCategorization, perception, and languageAI-based Problem Solving and Planning

Structure-mapping: A theoretical framework for analogy

Open access
GENTNER, D·1983·Cognitive Science

Abstract

A theory of analogy must describe how the meaning of an analogy is derived from the meanings of its parts. In the structure-mapping theory, the interpretation rules are characterized as implicit rules for mapping knowledge about a base domain into a target domain. Two important features of the theory are (a) the rules depend only on syntactic properties of the knowledge representation, and not on the specific content of the domains; and (b) the theoretical framework allows analogies to be distinguished cleanly from literal similarity statements, applications of abstractions, and other kinds of comparisons. Two mapping principles are described: (a) Relations between objects, rather than attributes of objects, are mapped from base to target; and (b) The particular relations mapped are determined by systematicity, as defined by the existence of higher-order relations.

AI snapshot

Written from the paper’s full text.

TL;DR

This paper presents structure-mapping as a framework for how analogies are interpreted, arguing that analogical mapping transfers relations between objects rather than object attributes, and that relations belonging to interconnected higher-order systems are preferred. It distinguishes analogy from literal similarity, abstraction, and mere appearance matches, and cites prior empirical work as support.

Key findings

  • Analogy maps relational predicates from base to target while discarding object attributes; for example, the atom/solar-system analogy maps relations such as revolving around, attraction, and relative mass, but not attributes such as color or temperature.
  • The systematicity principle holds that a predicate is more likely to be mapped if it belongs to a mutually interconnected system of relations, especially higher-order causal or constraint structures; this explains why a relation like “more massive than” is preserved while a formally similar relation like “hotter than” is not.
  • The framework separates comparison types by predicate overlap: literal similarity maps many attributes and many relations; analogy maps few attributes and many relations; abstraction maps few attributes and many relations with abstract base objects; mere appearance maps attributes but not relations; anomaly maps neither.
  • Empirical support is cited from prior studies: subjects’ interpretations of analogical comparisons were predominantly relational, and aptness ratings correlated with the amount of relational information produced; no numeric values are reported in this paper.

Methods

  • Theoretical paper: knowledge is represented as propositional networks with syntactic predicate types—attributes take one argument, relations take two or more, and predicates can be first-order or higher-order.
  • Worked examples (Rutherford atom/solar system, heat/water, arithmetic analogy) are used to derive mapping rules and a taxonomy of domain comparisons.
  • Empirical support is summarized from prior studies in which subjects wrote interpretations of analogical comparisons and naive judges rated assertions as attribute-based or relation-based; developmental and science-learning studies are also cited.

Limitations

  • The paper states a simplifying assumption that the hearer begins with object correspondences and then maps relations; it notes that actual processing order is variable and that object correspondences may be settled only before relational mapping is finished.
  • The paper assumes predicates are carried across as identical matches and notes that treating predicates as merely similar risks an infinite regress.
  • Evident from the text: empirical support is brief and relies on cited studies rather than new experiments or statistics; the paper is primarily theoretical and does not report detailed sample sizes or quantitative results.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/s0364-0213(83)80009-3

Cites

11 of 11 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 20 cited sources checked · 0 supported · 0 flagged

  • Findingcitation
    § EMPIRICAL SUPPORT

    Subjects judged as most apt those analogies for which they produced the most relational information.

    “Further, a correlation of aptness ratings and relationality ratings revealed that subjects rated as most apt those analogies for which they wrote the greatest amount of relational information.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gentner, 1980b

      Not checked in the free run.

  • Premisecitation
    § Abstract

    A model based on the degree of featural overlap gives a good account of literal similarity comparisons.

    “Such a “degree of overlap” approach seems reasonably correct for literal similarity comparisons.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Tversky (1977)

      Not checked in the free run.

  • Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    Changing one relation in a mappable system affects the others.

    “One symptom of this systematicity is that changing one of these relations affects the others.”
    Quote in the paperSubstantial claim
  • Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    Postulating that analogies select for dynamic causal knowledge would be a mistake because it unreasonably limits the scope of analogy.

    “We could accomplish this by postulating that analogies select for dynamic causal knowledge, or more generally, for appropriate abstractions.Either of these would be a mistake: The former course limits the scope of analogy unreasonably,”
    Quote in the paperSubstantial claim
  • Premiseown data
    § EMPIRICAL SUPPORT

    Research exists that supports the structure-mapping approach.

    “There is research supporting the structure-mapping approach.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim. 21
  • May combine two claims. 8
  • The quote only partly states this claim. 1
  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given
    § THE ANALOGICAL SHIFT CONJECTURE

    An informal observation is that the earliest comparisons in learning a domain are mostly literal-similarity matches, then analogies, then general laws.

    “An informal observation is that the earliest comparisons are chiefly literalsimilarity matches, followed by analogies, followed by general laws.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim.

    Findingown data
    § THE ANALOGICAL SHIFT CONJECTURE

    In an observed case of a subject trying to understand water flow through a constricted pipe, the subject first made similarity matches, later analogies, and finally a general statement of the Bernoulli principle.

    “His first comparisons were similarity matches, e.g., water coming through a constricted hose.Later, he produced analogies such as a train speeding up or slowing down, and balls banging into the walls and transferring momentum.Finally, he arrived at a general statement of the Bernoulli principle, that velocity increases and pressure decreases in a constriction.”
    Quote in the paperSubstantial claim
  • May combine two claims.

    Findingcitation
    § EMPIRICAL SUPPORT

    In analogy interpretation, relational information is more prevalent than attributional information, but this is not true for object descriptions produced by the same subjects.

    “The results indicated a strong focus on relational information in interpreting analogies.Relational information predominates over attributional information in analogy interpretations, but not in object descriptions generated by the same sub-”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gentner, 1980b

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    The systematicity principle can satisfy both the requirement to prefer chain (2) over chain (1) and the requirement to operate initially without appealing to specific content or appropriateness.

    “The systematicity principle offers a way to satisfy both requirements.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    Analogy conveys a system of connected knowledge rather than an assortment of independent facts.

    “Part of our understanding about analogy is that it conveys a system of connected knowledge, not a mere assortment of independent facts.”
    Quote in the paperSubstantial claim
  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    In the "Heat is like water" analogy, the dynamic causal scenario is more interesting than the static spatial one.

    “Intuitively, the second scenario is more interesting than the first: (1) conveys merely a static spatial description, while (2) conveys a dynamic causal description.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § STRUCTURE-MAPPING: INTERPRETATION RULES FOR ANALOGY

    In the mapping, relations between objects are preserved.

    “2. Try to preserve relations between objects: Nbi, WI --> ML, t,),”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § STRUCTURE-MAPPING: INTERPRETATION RULES FOR ANALOGY

    Which relations are preserved is decided by choosing systems of relations, according to the systematicity principle.

    “3. (The Systematicity Principle) To decide which relations are preserved, choose systems of relations:”
    Quote in the paperSubstantial claim
  • May combine two claims.

    Premisecitation
    § EMPIRICAL SUPPORT

    Feature-list and multidimensional-space representations handle object attributes well but are very limited in expressing relations between objects, especially higher-order relations.

    “These kinds of representations can deal well with object attributes, but are extremely limited in their ability to express relations between objects, and especially higher-order relations.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Ortony, 1979

      Not checked in the free run.

    • cites Rumelhart & Abrahamson, 1973

      Not checked in the free run.

    • cites Tourangeau & Sternberg, 1981

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationnone given
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    The equations embody higher-order relations that connect lower-order relations into a mutually constraining structure.

    “These equations embody higher-order relations that connect the lowerorder relations (1) through (4) into a mutually constraining structure.”
    Quote in the paperSubstantial claim
  • May combine two claims.

    Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    MORE MASSIVE THAN is preserved while HOTTER THAN is not because only the former participates in the central-force system.

    “This is why MORE MASSIVE THAN is preserved while HOTTER THAN is not: Only MORE MASSIVE THAN participates in the central-force system of predicates.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim.

    Interpretationreasoning
    § STRUCTURE-MAPPING: INTERPRETATION RULES FOR ANALOGY

    Predicates from the base are carried to the target using the node substitutions dictated by the object correspondences.

    “Predicates from B are carried across6 to T, using the node substitutions dictated by the”
    Quote in the paperSubstantial claim
  • May combine two claims.

    Interpretationreasoning
    § THE ANALOGICAL SHIFT CONJECTURE

    Potential analogies are less likely to be noticed because they require relational matches to access the database, and object matches do not help.

    “Potential analogies are less likely to be noticed, since they require accessing the data base via relational matches; object matches are of no use.”
    Quote in the paperSubstantial claim
  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationnone given
    § Abstract

    An analogy asserts that a relational structure normally applying in one domain can be applied in another domain.

    “The central idea is that an analogy is an assertion that a relational structure that normally applies in one domain can be applied in another domain.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § Additional information

    The structure-mapping theory describes the implicit interpretation rules of analogy.

    “The strucfure-mapping theory describes the implicit interpretation rules of analogy.”
    Quote in the paperSubstantial claim
  • May combine two claims.

    Interpretationreasoning
    § Abstract

    The strength of an analogical match does not depend on overall featural overlap, and only certain kinds of mismatches count for or against an analogy.

    “The strength of an analogical match does not seem to depend on the overall degree of featural overlap; not all features are equally relevant to the interpretation.Only certain kinds of mismatches count for or against analogies.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § INTERPRETATION:

    An analogy maps relational predicates but few or no object attributes from the base to the target.

    “(2) An analogy is a comparison in which relational predicates, but few or no object attributes, can be mapped from base to target.”
    Quote in the paperSubstantial claim
  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationcitation
    § KINDS OF DOMAIN COMPARISONS

    A literal similarity comparison maps many predicates, including both object attributes and relations, relative to the number of nonmapped predicates.

    “(1) A literul similurity statement is a comparison in which a large number of predicates is mapped from base to target, relative to the number of nonmapped predicates (e.g., Tversky, 1977).The mapped predicates include both object-attributes and relational predicates.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Tversky, 1977

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationnone given
    § Additional information

    The structure-mapping rules depend only on syntactic properties of the knowledge representation, not on the specific content of the domain.

    “These rules have the desirable property that they depend only on syntactic properties of the knowledge representation, and not on the specific content of the domain.”
    Quote in the paperSubstantial claim
  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    Dynamic causal information is usually represented at a deeper level of embedding than simple stative information.

    “Dynamic causal information [e.g., (2)] will usually be represented in a more deeply embedded structure than simple stative information [e.g., (l)].”
    Quote in the paperNot a substantial claim
  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    Such a system can be represented by an interconnected predicate structure in which higher-order predicates enforce connections among lower-order predicates.

    “Such a system can be represented by an interconnected predicate structure in which higher-order predicates enforce connections among lower-order predicates.”
    Quote in the paperNot a substantial claim
  • The quote only partly states this claim.

    Interpretationreasoning
    § Abstract

    In structure-mapping theory, predicate types are distinguished syntactically so that it can be stated which predicates will be mapped.

    “In the structuremapping theory, a simple but powerful distinction is made among predicate types that allows us to state which ones will be mapped.”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § STRUCTURE-MAPPING: INTERPRETATION RULES FOR ANALOGY

    In the mapping, attributes of objects are discarded.

    “1.Discard attributes of objects: A(b,)I -•t -> LWJ”
    Quote in the paperSubstantial claim
  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationnone given
    § EXAMPLE

    Abstraction differs from analogy and other comparisons by having few object attributes in both the base and the target domains.

    “Wbstraction differs from analogy and the other comparisons i n having few abject-attributes i n the bose domain as well as few object-attributes i n the target domain.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim.

    Interpretationreasoning
    § Additional information

    The structure-mapping framework allows one to state the differences between analogies and literal similarity statements, abstractions, and other comparisons.

    “Further, this theoretical framework allows us to state the differences between analogies and literal similarity statements, abstractions and other kinds of comparisons.”
    Quote in the paperSubstantial claim
  • May combine two claims.

    Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    Postulating that analogies select for appropriate abstractions would be a mistake because appropriateness is vague and analogies can convey inappropriate abstractions.

    “and the latter course is both vague, in that “appropriateness” is difficult to define explicitly, and incorrect, in that analogies can also convey inappro-e”
    Quote in the paperNot a substantial claim
  • May combine two claims.

    Premisecitation
    § STRUCTURE-MAPPING: INTERPRETATION RULES FOR ANALOGY

    Most explanatory analogies are one-to-one mappings, with m equal to n, though exceptions exist.

    “Most explanatory analogies are l-1 mappings, in which m = n.However, there are exceptions (Gentner, 1982).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gentner, 1982

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationreasoning
    § HIGHER-ORDER PREDICATES AND SYSTEMATICITY

    The number of relations that can be mapped in an analogy may have no definite upper limit.

    “indeed the number of mappable relations may have no clear upper bound.”
    Quote in the paperNot a substantial claim
  • Classified as probably not a substantial claim.

    Interpretationnone given
    § STRUCTURE-MAPPING: INTERPRETATION RULES FOR ANALOGY

    An analogy maps the object nodes of the base domain onto the object nodes of the target domain.

    “The analogy maps the object nodes of B onto the object nodes of T: M: bi --> t,”
    Quote in the paperSubstantial claim
  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § EXAMPLE

    Many attribute-based metaphors use conventional vehicles or conventional dimensional matches.

    “Many such metaphors involve conventional vehicles, such as “giraffe” above, or conventional dimensional matches, such as “a deep/shallow idea”.[Glucksberg, Gildea, & Bookin, 1982;Lakoff & Johnson, 19801).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Glucksberg, Gildea, & Bookin, 1982

      Not checked in the free run.

    • cites Lakoff & Johnson, 19801

      Not checked in the free run.

30 of 30
  • The authors state this as an assumption.

    Predicates are carried across as identical matches, and similarity can be restated as identity among some number of component predicates.

    “The assumption that predicates are brought across as idenficul matches is crucial to the clarity of this discussion.”
    “I will assume instead that similarity can be restated as identity among some number of component predicates.”
  • The authors state this as an assumption.

    Partial knowledge of a system is often sufficient for a person to gauge its interconnectedness.

    “I make the assumption here that partial knowledge of the system is often sufficient to allow a person to gauge its interconnectedness.”
  • The authors state this as an assumption.

    In comprehending an analogy, the hearer begins with object correspondences and then maps relations across.

    “Yn this discussion I have made the simplifying assumption that, in comprehension of analogy, the hearer starts with the object correspondences and then maps across the relations.”
3 of 3

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Higher-level relational reasoning, such as analogy and inference, involves mapping correspondences and creating higher-order relations between schemas.

    “Higher-level forms of relational reasoning, such as analogy and inference, involve mapping correspondences and creating higher-order relations between schemas (i.e., relations between relations; Gentner, 1983;Holyoak, 2012).”

    From Mental Time Travel? A Neurocognitive Model of Event Simulation. (2020) · cites this paper as Gentner, 1983

Cites

Flagged first, then by citations

1 of the 12 references couldn’t be matched to a known paper.

Research works printed but not linked: 7

  1. Rumelhart, D. E., &Norman. D. A. Analogical processes in learning. In J. R. Anderson (Ed.), Cognirive skills and fheir acquisifion. Hillsdale, NJ: Lawrence Erlbaum Associates, 1981.
  2. Schustack, M. W.. & Anderson. J. R. Effects of analogy to prior knowledge on memory for new information. Journal of Verbal Learning and Verbal Behavior, 1979, 18, 565-583.
  3. Steels. L. An applicaiive view of object oriented programming (MIT A.I. Memo No. 15). Cambridge, MA: Massachusetts Institute of Technology, March 1982.
  4. Stevens, A., Collins, A., & Goldin, S. E. Misconceptions in student's understanding. Journal o/Man-Machine Sfudies. 1979, II. 145-156.
  5. Tourangeau, R., & Sternberg, R. J. Aptness in metaphor. Cogniriue Psychology, 1981, 13, 27-55.
  6. Tversky, A. Features of similarity. Psychological Review, 1977, 84, 327-352.
  7. Winston, P. H. Learning and reasoning by analogy. CACM, 1980. 23, No. 12. Winston, P. H. Learning new principles from precedenrs and exercises. (MIT Artificial Intelli- gence Memo No. 632). Cambridge, MA: Massachusetts Institute of Technology, May 1981.

Cited by

Sorted by citations

Authors
GENTNER, D.
Corresponding author
GENTNER, D
Venue
Cognitive Science
Publisher
Wiley
Published
1983-06
Volume
7
Issue
2
Pages
155-170
ISSN
0364-0213 (ISSN-L)
OpenAlex
W4229954498
Cited by
816
References
12
Citations by year
2026: 60 · 2025: 55 · 2024: 69 · 2023: 63 · 2022: 50 · 2021: 53 · 2020: 45 · 2019: 38 · 2018: 27 · 2017: 22 · 2016: 28 · 2015: 23 · 2014: 30 · 2013: 17 · 2012: 14
FWCI
5.03
Type
journal article
Language
en
Open access
Locations
Cognitive Science (publishedVersion)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Language, Metaphor, and Cognition
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Categorization, perception, and language
  • Physical Sciences › Computer Science › Artificial Intelligence › AI-based Problem Solving and Planning
Keywords
structure-mapping theory, analogy, analogical reasoning, mapping rules, knowledge representation, systematicity, higher-order relations, abstraction

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-07.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-07.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review