Structure-Mapping Theory
Dedre Gentner's structure-mapping theory (1983) provides the most influential cognitive account of how analogies work. The central insight is that analogical thinking involves mapping relational structures from a known source domain onto a target domain, preserving the relationships between objects even when the objects themselves change.
When you understand that "the atom is like a solar system," you're not claiming electrons are literally planets or that nuclei are literally suns. You're mapping relational structure: smaller objects orbit larger objects in stable patterns. The theory explains why some analogies are good and others poor: good analogies preserve relational structure while poor analogies break or distort the mapped relationships.
Gentner distinguished between analogies, metaphors, and similes based on the type of mapping. Pure analogies map structural relations without object-level claims. Metaphors often map attributes in addition to relations—the " Achilles heel" of a system maps the attribute of vulnerability, not just a relational structure.
Holyoak and Thagard's Research
Keith Holyoak and Paul Thagard extended analogical reasoning research in several important directions. Their "multi-constraint theory" proposed that analogical mapping is guided by multiple factors:
Semantic constraint: The content of the domains matters. Analogies are more accessible when the domains share underlying semantic features (both involve physical objects, causation, agents with goals).
Structural constraint: The relational structure of the source and target must be compatible. More overlapping relations enable smoother mapping.
Pragmatic constraint: The purpose of the analogy matters. We're more likely to retrieve and apply analogies that address current goals and concerns.
Holyoak's research on analogical problem-solving demonstrated that providing target problems with analogous source problems substantially improved solution rates. Duncker's radiation problem exemplifies: subjects shown an analogous military siege problem solved the medical problem at dramatically higher rates than control groups.
Insight Problem Solving
Insight problems—those that require a sudden "aha" moment rather than gradual progress—have been linked to analogical processing. The constraint relaxation view (Dietrich, 2004) proposes that insight occurs when problem solvers suddenly recognize that an assumed constraint doesn't actually exist.
Analogical thinking enables insight by providing alternative frames. When problem solvers encounter an analogy that maps structure from a different domain, they may recognize that constraints they assumed were inherent to the problem are actually artifact of their current framing.
Research by Bowden and colleagues (2005) using the "aha" experience as a marker found that insight solutions showed different neural signatures than non-insight solutions. Insight was associated with sudden bursts of gamma activity in temporal regions, suggesting a distinct cognitive process rather than merely faster version of analytical processing.
Constraints on Successful Mapping
Not all analogies are equally valuable. Several factors constrain analogical success:
Surface similarity: People tend to retrieve analogies based on surface features rather than structural features. This leads to missed opportunities when structurally similar domains have dissimilar surfaces, and to misleading mappings when superficially similar domains have different deep structures.
Spontaneous access: Analogies that are obvious to experts are often invisible to novices because they require background knowledge to recognize the relevant structure. Novices map based on surface features; experts can access deeper structural representations.
Mapping fidelity: Even when an analogy is retrieved, mapping errors can occur—importing the wrong relations, over-mapping (applying source relations that don't apply to target), or under-mapping (failing to apply relevant source relations).
Practical Applications
Deliberately seek cross-domain analogies: When stuck on a problem, actively search for structurally similar problems in domains you know well. The structure you're trying to create or discover in the target may already exist in another domain.
Surface features vs structural features: When considering an analogy, ask: are we matching the surface features or the relational structure? Surface matches may mislead; structural matches provide insight.
Make analogies explicit: Articulate the mapping in detail—which elements correspond, which relations map, which don't. This prevents over-mapping and enables critical evaluation of the analogy's validity.