Analogical Reasoning
Evaluating Analogy Strength
Master advanced techniques for stress-testing analogies, including identifying hidden structural assumptions, measuring analogical distance, and understanding how metaphorical framing shapes thought and policy.
Context
Why this exercise
At the advanced level, evaluating analogies stops being a matter of vibe and becomes a procedure. You learn to enumerate the mappings explicitly, measure the analogical distance, locate the structural breakdowns, and ask whether a competing analogy would have steered the conclusion in a different direction. The stakes for this skill are highest when analogies govern public policy and scientific research programs, because the chosen metaphor quietly determines which solutions seem natural and which become invisible. This exercise drills the systematic evaluation techniques that separate sophisticated analogical thinking from impressionistic comparison.
Before you start
Lakoff and Johnson's conceptual metaphor theory, refined by experimental work from Paul Thibodeau and Lera Boroditsky, has shown that metaphors are not decorative — they are constitutive of how we reason about abstract domains. When crime is framed as a 'beast,' participants in their experiments favor enforcement; when it is framed as a 'virus,' the same participants favor social reform, even when the underlying statistics are identical. The War on Drugs frame channeled trillions of dollars into law enforcement and incarceration; the later opioid 'epidemic' frame redirected resources toward treatment and harm reduction. In science, the governing analogy for the central object of study (brain as telegraph, then as computer, now as predictive engine) determines which phenomena researchers think to study and which they fail to notice. Thomas Kuhn's 'The Structure of Scientific Revolutions' documented this dynamic across the history of physics, chemistry, and biology.
The systematic evaluation framework has four steps. First, enumerate the structural mappings — exactly which features of the source correspond to which features of the target — and refuse to evaluate an unanalyzed analogy. Second, assess the causal relevance of each mapping: a similarity that has no causal relationship to the conclusion does no work, no matter how vivid it feels. Third, identify the disanalogies and ask whether any of them undermine the specific conclusion. Fourth — and this is the move beginners almost always skip — consider what alternative analogy would have been available and what conclusion it would have supported. The 'Uber of X' pitch is persuasive partly because no competing analogy is offered; once you ask 'what if this is actually the regulated-utility of X?', the conclusion shifts.
Two concepts are worth carrying with you. Analogical distance, studied by Gavetti and Rivkin and by Dunbar, measures how conceptually remote the source is from the target — near analogies are reliable but incremental, far analogies are risky but generative. And the 'similarity counting' error: ten irrelevant similarities never outweigh one relevant dissimilarity. A car and a horse share many features (both transport, both have moving parts, both need maintenance) but differ on every dimension relevant to emission regulation. The advanced thinker evaluates analogies surgically, targeting the specific structural mapping the argument depends on, and treats every analogy as a hypothesis about structural similarity rather than as an established truth. For the broader skill of testing arguments and assumptions, see Dialectical Thinking.