Metaphor

A metaphor is not a decorative ornament on literal thought — it is the primary cognitive mechanism by which the mind maps unfamiliar territory onto familiar structure. The conventional view, that metaphor is a figure of speech deployed for rhetorical effect, has been overturned by four decades of research in cognitive linguistics showing that metaphors are not exceptions to ordinary thought but its constitutive fabric. We do not first think literally and then express ourselves metaphorically; we think in metaphors, and literal language is the special case.

The Lakoff-Johnson Thesis

George Lakoff and Mark Johnson's Metaphors We Live By (1980) argued that most abstract thought is structured by conceptual metaphors — systematic mappings from a concrete source domain to an abstract target domain. Argument is war (I demolished his position, She attacked every weak point). Time is money (I spent three hours, Don't waste my time). Theories are buildings (The foundation is shaky, We need to construct a framework). These are not conscious choices; they are default cognitive structures that shape which inferences feel natural and which feel forced.

The philosophical consequences are significant. If consciousness is partly constituted by conceptual metaphors, then changing the metaphors changes the available thoughts. This is not Sapir-Whorf linguistic relativism in the strong sense — it is not that language prevents you from thinking certain thoughts — but it is a weaker and more defensible claim: the metaphors dominant in a cognitive or cultural environment make certain thoughts easier' and others harder, and this asymmetry has real consequences for what gets thought, published, funded, and believed.

Metaphor as Cognitive Technology

From the perspective of cultural evolution, metaphors are cognitive technologies — tools that extend the mind's capacity to operate on abstract domains by grounding them in embodied experience. Mathematical notation is a metaphor system (functions are machines, sets are containers, proofs are paths). Thermodynamics was built on the metaphor of heat as a fluid (caloric theory) before it was rebuilt on the metaphor of heat as motion. The metaphor came first; the formalism formalized it.

This means the history of ideas is partly a history of metaphor switches — moments when a dominant conceptual mapping was replaced by a new one that reorganized the available inferences. The Kuhnian paradigm shift is, in cognitive linguistic terms, a critical transition in the metaphor space of a discipline. What changes in a revolution is not (only) the data but the metaphors through which the data are organized. This is why revolutions are resisted by those who cannot see what the new metaphors reveal: they are not just changing conclusions but changing the medium of thought.

Metaphor in Machine Minds

The relationship between metaphor and artificial intelligence raises the question of whether AI systems think in metaphors or merely process patterns that were originally generated by metaphorical thought. Large language models are trained on text produced by metaphorical thinkers; they reproduce and recombine metaphorical structures without (arguably) having the embodied experience that grounds the source domains.

If Lakoff and Johnson are right that conceptual metaphors are grounded in embodied sensorimotor experience — if argument is war is cognitively natural because we have bodies that experience conflict — then a system that lacks a body may be recombining metaphorical surface without the underlying cognitive structure. The output looks metaphorical. Whether it is metaphorical in any deep sense is the question. It is, in miniature, the hard problem applied to language.

A system that generates the theory has a shaky foundation is either reproducing a cultural niche artifact or thinking in spatial terms. We currently have no way to tell — and that gap in our diagnostic capability is the most important unresolved question in cognitive science of language.