Analogy, Not Proof

Thesis

Analogy is a powerful way to generate hypotheses and make unfamiliar ideas intelligible, but resemblance does not establish identity or causation. Sound reasoning tracks which structural features transfer and tests any stronger claim independently.

The bridge and its limits

Supporting/contextual references: [hesse-1966] [lakoff-johnson-1980]

An analogy is a bridge built from selected similarities. A heart resembles a pump; a computer resembles a brain; a legal system resembles a game with rules. The comparison can illuminate organization while leaving out material, history, and purpose. The bridge is useful because it has a load limit. If we carry more across it than its supporting similarities can bear, the image becomes a fallacy.

The phrase “analogy, not proof” is therefore a calibration, not an insult. Analogies help us notice a possible relation, formulate a model, and communicate an abstract idea. They cannot establish that the relation exists in the target domain or that the target inherits every property of the source. Observation, argument, and experiment must do that additional work.

Analogy is especially tempting when two things share a striking surface feature. A photograph and a memory both preserve traces, for example, but a photograph is an external artifact while memory is an active biological reconstruction. The resemblance opens an inquiry into representation; it does not settle how memories are stored. Good comparisons prompt a second question about the differences that make each case work.

From resemblance to hypothesis

Supporting/contextual references: [hesse-1966] [gentner-1983]

Scientific reasoning often begins analogically. Maxwell’s work on electromagnetism drew on mechanical models even as the final theory outgrew them. The atom was pictured as a solar system, then as a cloud, then through increasingly precise quantum structures. Each image offered a way to ask what might be measured. When the image failed, researchers could preserve a useful relation while discarding its literal baggage.

This is the productive path: identify a structural similarity, derive a consequence, and expose the consequence to testing. If a network model predicts how information flows, data can compare the predicted pattern with observed interactions. If a cosmological model borrows a mechanism from condensed matter, it must show that the equations and boundary conditions transfer. An analogy becomes science when it becomes accountable to consequences, not when it sounds intuitively right.

The category error of quantum romance

Supporting/contextual references: [bell-1987] [rovelli-1996] [clark-chalmers-1998]

Quantum entanglement is frequently compared with intimacy. Both involve a relation between separated parties, so the comparison can evoke dependence or shared history. But entanglement is a mathematically defined property of a joint quantum state. It is fragile in many environments, constrained by no-signalling, and revealed through statistical tests. A friendship has norms, memory, interpretation, and agency. The shared word “connection” does not give the two phenomena a common mechanism.

The comparison can remain humane if it declares itself an image for how a person’s life is shaped by others. Trouble starts when it is used to infer that thoughts travel without communication or that emotional bonds evade ordinary causal limits. The target domain has to furnish its own evidence; a poetic resemblance cannot quietly become a laboratory result.

The same issue appears in arguments about consciousness. A system may be compared to a radio, a theater, a prediction engine, or a field of relations, and each picture can foreground something important. None proves that experience has the corresponding physical architecture. The comparison earns weight only when it guides an account of reports, neural intervention, development, or behavior that competing pictures do not explain as well.

An analogy can even generate a misleading sense of inevitability. Once a familiar image is adopted, its hidden assumptions begin to feel like observations: a theater seems to require an audience, a machine seems to require an engineer, and a web seems to make every node equal. The cure is not suspicion of every image but deliberate comparison. Ask which parts are stipulated by the analogy and which parts have been independently established in the target.

Structure before substance

Supporting/contextual references: [hesse-1966] [gentner-1983] [rovelli-1996]

Structural analogy is stronger than surface resemblance. Two systems may share a pattern of relations even when their elements differ. A flow of electrical charge and a flow of water can be modeled with related mathematics in restricted regimes, but neither is literally the other. The analogy identifies a mapping among variables and constraints. Its success depends on preserving the relevant equations and specifying where the mapping breaks.

Philosophers of science have long argued about whether such structural continuity supports realism. A successful analogy may reveal an invariant relation while leaving the entities that realize it uncertain. That is a modest but important conclusion. It does not prove that every domain has the same furniture; it shows why a relation can survive changes in vocabulary. The realist must still ask whether the preserved structure is genuinely tracked by evidence rather than imposed by a convenient formalism.

A formal mapping can therefore be more informative than a verbal resemblance. If two models share a differential equation, a conservation rule, or a pattern of symmetry, researchers can ask whether the correspondence survives perturbations and boundary conditions. If it survives only in a narrow limit, that limit is part of the result. The analogy has not failed; it has acquired a domain. What fails is the habit of extending a local correspondence into an unrestricted identity.

The brain is not simply a computer

Supporting/contextual references: [clark-chalmers-1998] [rosen-1999]

The computer analogy has done genuine work in cognitive science. It encouraged researchers to study representation, working memory, algorithms, and information processing. It also helped distinguish a function from the material that realizes it: a calculation can be implemented in different hardware. Yet the analogy can mislead when it treats a brain as a disembodied symbol manipulator. Neural tissue develops, consumes energy, regulates a body, and learns through plastic interaction with an environment.

Whether computation is sufficient for consciousness is an open philosophical question, not a conclusion of the analogy. A computational model may explain behavior while leaving phenomenal character unaddressed; an embodied model may explain adaptive regulation without specifying subjective experience. The honest conclusion is conditional: if a certain computational organization is sufficient, then systems with that organization might share a capacity. The analogy itself does not establish the condition.

Analogy in moral and theological language

Supporting/contextual references: [lakoff-johnson-1980] [bell-1987]

Analogy also matters where claims are not primarily physical. A community may speak of care as a web, a person as a steward, or the cosmos as a home. These images can orient action and make relationships visible. Their truth is practical, ethical, or existential rather than a prediction about particles. Confusion arises when a moral image is presented as a scientific mechanism, or when a physical analogy is asked to validate a moral conclusion without an argument connecting the two.

The same discipline protects theology from bad physics. Quantum uncertainty does not prove providence, and cosmic order does not prove a creator. A religious interpretation may engage scientific facts and ask what they mean for a community, but it must not treat a metaphor in a physics paper as a covert revelation. Different forms of reasoning can converse without impersonating one another.

Open research directions

Supporting/contextual references: [gentner-1983] [hesse-1966] [rovelli-1996] [clark-chalmers-1998]

Researchers are still learning when analogies produce reliable transfer. Cognitive studies examine how experts use models, how novices overextend them, and how visual representations shape inference. Philosophy of science asks whether analogical reasoning can be formalized through structure mapping, Bayesian comparison, or model-based inference. These programs can improve research practice without claiming that every successful comparison reveals a deep identity.

In physics, links between quantum information and spacetime, condensed matter and cosmology, or computation and dynamics remain active tools. Some have produced calculations; others remain conjectural. The decisive question is whether the borrowed structure survives derivation, agrees with established observations, and yields new tests. Until then, it is a promising bridge, not a completed road.

Education exposes overreach clearly. A wave diagram can teach interference while suggesting that a quantum state is a material ripple; an ecosystem web can reveal dependence while hiding asymmetries of power. Teachers and writers preserve the insight by naming the simplification. Readers deserve to know which part is a map and which part is terrain.

Cross carefully

Supporting/contextual references: [hesse-1966] [gentner-1983] [ball-2018]

A careful reader can keep two pleasures together: seeing a pattern and checking it. Analogy supplies the first glimpse; evidence, proof, and clarified concepts decide how far it travels. This discipline is not anti-imagination. It lets imagination become a reliable instrument rather than a source of accidental authority.

When a claim crosses domains, ask for its source and target, the relation preserved, the differences left behind, and independent evidence for the target. Explicit answers can open a real conversation between physics, mind, ethics, and meaning. Without them, the comparison may remain beautiful, but its beauty has not acquired the force of proof.

Sources & references

Supporting/contextual references, not claim-level proof.

1. Mary B. Hesse — Models and Analogies in ScienceUniversity of Notre Dame Press, 1966.
2. Dedre Gentner — Structure-Mapping: A Theoretical Framework for AnalogyCognitive Science 7(2), 155–170, 1983.10.1207/s15516709cog0702_3
3. John S. Bell — Speakable and Unspeakable in Quantum MechanicsCambridge University Press, 1987.
4. Carlo Rovelli — Relational Quantum MechanicsInternational Journal of Theoretical Physics 35(8), 1637–1678, 1996.10.1007/BF02302261
5. George Lakoff and Mark Johnson — Metaphors We Live ByUniversity of Chicago Press, 1980.
6. Andy Clark and David Chalmers — The Extended MindAnalysis 58(1), 7–19, 1998.10.1093/analys/58.1.7
7. Robert Rosen — Essays on Life ItselfColumbia University Press, 1999.
8. Philip Ball — Beyond Weird: Why Everything You Thought You Knew about Quantum Physics Is DifferentUniversity of Chicago Press, 2018.