[ CogSci Summaries home | UP | email ]
http://www.jimdavies.org/summaries/

Gentner, D. (1983). Structure-mapping: A theoretical framework for analogy. Cognitive Science. 7 (2), pp155-170.

```@Article{Gentner83,
author =       "Dedre Gentner",
title =        "Structure-Mapping: {A} Theoretical Framework for
Analogy",
journal =      "Cognitive Science",
year =         "1983",
volume =       "7",
number =       "2",
pages =        "155--170",
month =        apr # "--" # jun,
}
```

Author of the summary: Jim Davies, 1999, jim@jimdavies.org

Cite this paper for:

• an analogy is a relational structure that normally applies in one domain can be applied in another domain.
• attributes: predicates with one argument e.g. red(apple)
• relations: predicates with more than one argument. e.g collide(x,y)
• Abstraction: is like analogy but there are few object attributes in both base and target domains. e.g. The atom is a central force system.
• Order of a relation: "The order of a relation is determined by the order of its arguments. A first-order relation takes objects as its arguments. A second-order relation has at least one first-order relation among its arguments; and in general an nth order relatino has at least one (n-1)th order argument."
• Systematicity: The principle is this-- "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."
Some (Tversky 1977) say that a good analogical mapping is a good feature overlap. This is true for "literal similarity" but is not good for analogy. In Gentner's view, an analogy is a relational structure that normally applies in one domain can be applied in another domain.

attributes: predicates with one argument e.g. red(apple)

relations: predicates with more than one argument. e.g collide(x,y)

In structure-mapping theory (SMT), a good analogy is one where the relations match well, without regard to what they are relating. The deeper the structure of the relation (high order relations) the more promising the analogy is. For example, an electron revolving around a nucleus is a good analogy with a planet revolving around a sun even though nucleus does not equal sun-- they both share the revolves-around(x,y) relation.

Literal similarity is where many relations map and many attributes map as well. e.g. our solar system is like the K5 solar system.

Analogy is when few attributes match but many relations match. e.g. The atom is like our solar system.

Abstraction is like analogy but there are few object attributes in both base and target domains. e.g. The atom is a central force system.

Anomaly is when there are few attribute and few relations mapped. e.g. coffee is like a solar system.

Metaphor is like analogy in most cases, but is sometimes used jusr for one specific attribute. e.g. "She's a giraffe" means only that she's tall, and probably not that she is furry.

Order of a relation: "The order of a relation is determined by the order of its arguments. A first-order relation takes objects as its arguments. A second-order relation has at least one first-order relation among its arguments; and in general an nth order relatino has at least one (n-1)th order argument."

Systematicity: The principle is this-- "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." So the sun being hotter than the planets is not really relevent to the solar system's behavior like the relative masses are. Because the relative masses affect the other relations in the system, it is a more important relation. Matching that is more valuable and has more systematicity than mapping the hotter-than relation.

There is empirical support for SMT. Subjects asked to write about analogies such as "a cigarette is like a time bomb" were found to focus heavily on relational information rather than objects or attributes.

Summary author's notes:

• none

Back to the Cognitive Science Summaries homepage
Cognitive Science Summaries Webmaster:
JimDavies (jim@jimdavies.org)