Negative Analogy

A negative analogy is a known difference between two things being compared. Such a difference can limit what you infer, without making the entire comparison useless. A cardboard model and a steel bridge may share a layout, but their different materials mean the model alone does not establish the real bridge’s strength. Ask this: which known difference matters to the conclusion I am trying to draw?

Distinguish known similarities, differences and unknowns

In the terminology discussed by Bartha, a positive analogy consists of accepted similarities and a negative analogy of accepted differences. A neutral analogy concerns a feature whose counterpart is not yet known to match. Not knowing whether a feature transfers is different from knowing that it does not.

Tie the difference to an inference

Imagine a cardboard model of a proposed steel bridge. Its layout can help communicate where the supports and span will be. The material difference matters if someone tries to infer the real bridge’s load-bearing strength from the model alone, but need not defeat the model’s use for illustrating the layout.

Use social comparisons cautiously

A manager may compare a company to a family to emphasize care and belonging. That comparison does not establish that employment and family relationships have identical obligations or expectations. Identify the particular conclusion being proposed, rather than assuming every feature of one relationship belongs to the other.

State the model’s assumptions

If a city model assumes fixed, centrally specified interactions, residents’ independent choices may limit what can be inferred from it. This is an illustration of checking assumptions, not a claim that every machine is static or that one planning metaphor inevitably causes a particular social outcome.

Check before rejecting

List known differences, then ask which are relevant to the proposed inference. An irrelevant difference does not by itself refute the analogy. Where a feature is unknown, seek evidence instead of treating the absence of information as a known mismatch.

Known differences help delimit inferences, but the relevance of the difference matters.

• Bartha’s2019 archived entry distinguishes positive, negative and neutral analogies: accepted similarities, accepted differences, and features not yet known to have counterparts. It uses these distinctions to describe what an analogical argument asks us to infer. [1]
• Gentner and Markman (1997) describe structural alignment in analogy and similarity, emphasizing the organization of relational correspondences rather than the mere presence of matching object attributes. This provides a way to examine which correspondences an inference relies on. [2]

A known difference need not be relevant to the conclusion, and an unknown feature is not automatically a known difference. The classification helps organize a comparison; it does not by itself establish whether an inference is justified. Check the purpose, relationships and assumptions being transferred, and seek additional evidence where needed. A useful model can retain important limitations.