Analogical reasoning
Solving a new problem by mapping it onto a familiar one with the same structure.
What it means
Analogical reasoning is the process of transferring relational structure from a known source domain to a novel target domain, allowing inference and problem solving by alignment of corresponding roles rather than surface similarity. Structure-mapping theory holds that good analogies match systems of relations, not isolated attributes, and that the systematicity of the mapping governs which inferences are projected. The central difficulty is access: people readily use an analogy once it is pointed out but often fail to retrieve a structurally similar source spontaneously when surface features differ, the classic transfer failure. Analogy underlies scientific discovery, metaphor, legal precedent, and much instruction, and abstraction across multiple examples improves later transfer. It is a hallmark of flexible, relational human thought.
How the mapping works
Structure-mapping treats an analogy as an alignment problem. The reasoner sets two situations side by side and looks for a one-to-one correspondence between their elements, constrained by parallel connectivity: if two relations match, the things they relate must match too. General maps to doctor, army to rays, fortress to tumor. Attributes drop out: the army's size and the fortress's stone walls have no counterpart in the tumor, and are not asked to have one. What survives is the relational skeleton. Systematicity then does the selecting: mappings that hang together in a deep, interconnected causal system beat piles of unconnected matches. What crosses over are candidate inferences, the payoff of the exercise. They are also guesses. The mapping proposes them; it cannot certify them.
What the evidence shows
The radiation problem supplies the canonical numbers. Roughly 10 percent of people solve it cold. Reading the fortress story minutes earlier lifts that to about 30 percent, and a nonspecific hint to use it pushes it to about 75 percent. The gap between 30 and 75 is the whole finding: the analogy sits in memory, encoded minutes earlier, and most people never reach for it. Gick and Holyoak then tested what helps. Stating the principle or diagramming a single story barely moved transfer, and summarizing it did no better on a second target problem. Two sources did: 45 percent solved the problem unprompted after comparing two analogs, against 21 percent after one, and pairing the two with a diagram took total solutions to 92 percent. The quality of the schema people wrote out predicted whether they transferred.
The retrieval bottleneck
Holyoak and Koh varied a source's surface and structural resemblance to a target independently. Both mattered for whether a source came to mind at all: surface resemblance plays no causal role in the solution, yet it shaped spontaneous retrieval alongside structure. Structure did the heavy lifting later, once relevance was pointed out, where it had the greater effect on whether the source could actually be used. So retrieval answers partly to a cue that carries no explanatory weight. That is why an analyst recalls another case from her own field rather than the paper with the identical feedback structure. The lab may overstate the failure. Trench observes that working experts produce distant analogies constantly, and argues the paradigm only tests retrieval of a source someone else planted, while real reasoners can also build a base analog to fit.
Using it in practice
The lever is comparison, not exposure. Gentner, Loewenstein and Thompson taught negotiation to novices and found that comparing two cases against each other beat studying the same two in sequence, with the advantage surviving into a face-to-face negotiation exercise. So do not hand over one exemplar and hope. Show two, unalike on the surface, and ask explicitly what they share; the abstraction that question forces is the part that travels, and naming it gives memory a cue that is not skin-deep. Then interrogate the analogy rather than trust it. Ask which differences touch a causal element of the source's outcome. A difference that leaves the causal structure intact is harmless. One that breaks it makes the analogy misleading rather than merely incomplete, and the mapping will not tell you which you have.
Examples
Most people fail to solve a tumor-radiation problem until reminded of a story about an army splitting to attack a fortress from many directions, whose 'converge weak forces' structure maps directly onto the cure.
Students grasp a circuit faster when told it is water in pipes: pressure maps to voltage, flow to current, a narrow pipe to resistance. The relations transfer, the water does not.
A lawyer wins by mapping a dispute over a rented car park onto an old ruling about a rented stable: nothing on the surface matches, but every role does.
An engineer examining burrs clinging to fur maps hook-catches-loop onto woven fabric and gets a fastener. Seeds and fur carry nothing over; only the geometry of the catching relation does.
Interface designers map the office onto the screen: documents nest in folders, deleted work lands in a bin. Users reach into the bin for something they discarded without ever having tried it — a candidate inference the mapping handed them, correct but untested. The same mapping hands them a wrong one: that a file, like a sheet of paper, can sit in only one folder at a time.
First described in Gentner (structure-mapping, 1983); Gick & Holyoak (1980).
Key references
- Trench, M. (2014). Explaining the abundance of distant analogies in naturalistic observations of experts. Frontiers in Psychology, 5, 1487. doi.org/10.3389/fpsyg.2014.01487
- Gentner, D., Loewenstein, J., & Thompson, L. (2003). Learning and transfer: A general role for analogical encoding. Journal of Educational Psychology, 95(2), 393-408. doi.org/10.1037/0022-0663.95.2.393
- Holyoak, K. J., & Koh, K. (1987). Surface and structural similarity in analogical transfer. Memory & Cognition, 15(4), 332-340. doi.org/10.3758/BF03197035
- Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15(1), 1-38. doi.org/10.1016/0010-0285(83)90002-6
- Gentner, D. (1983). Structure-mapping: A theoretical framework for analogy. Cognitive Science, 7(2), 155-170. doi.org/10.1207/s15516709cog0702_3
- Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12(3), 306-355. doi.org/10.1016/0010-0285(80)90013-4