metaphor embodied-experience forcescalepath causetransform hierarchy primitive

Similarity Is Closeness

metaphor

Source: Embodied ExperienceIntellectual Inquiry

Categories: cognitive-sciencelinguisticsphilosophy

From: Master Metaphor List

Transfers

A primary metaphor grounded in the infant’s experience that things near each other in space tend to share properties. Objects close together are often of the same kind — the same shelf holds similar books, the same drawer holds similar utensils, members of the same family sit close at the table. This pervasive spatial-conceptual correlation establishes, long before language, the mapping: similar things are close; different things are far apart.

Key structural mappings:

The metaphor’s greatest triumph is its formalization. Multidimensional scaling, vector spaces, embedding models, and nearest-neighbor algorithms all take the metaphor literally: they place items in a geometric space where distance encodes similarity. Word2Vec, the ancestor of modern LLMs, is the container metaphor applied to words, but it runs on the similarity- is-closeness metaphor: similar words are nearby vectors.

Limits

Expressions

Origin Story

Grady (1997) identified SIMILARITY IS CLOSENESS as a primary metaphor, grounded in the primary scene: objects near each other in the child’s environment tend to have similar properties. Things in the same pile, on the same shelf, in the same room are usually more alike than things far apart. This correlation between spatial proximity and shared properties is among the most reliable in early experience.

The metaphor has been spectacularly productive in science. When Shepard (1962) introduced multidimensional scaling — the technique of placing items in a geometric space such that distances reflect similarity judgments — he was not inventing a metaphor but formalizing one that was already implicit in everyday language. Tversky’s (1977) critique of geometric models of similarity was, in effect, a “Where It Breaks” analysis of this primary metaphor, showing that human similarity judgments violate the axioms of metric spaces (symmetry, triangle inequality, minimality).

Lakoff and Johnson (1999) gave the metaphor its canonical formulation. Its influence on computational science is hard to overstate: vector space models, embedding spaces, k-nearest-neighbors, cosine similarity — the entire apparatus of modern machine learning’s approach to meaning takes this metaphor literally and builds technology on it.

References

Related Entries

Structural Neighbors

Entries from different domains that share structural shape. Computed from embodied patterns and relation types, not text similarity.

Structural Tags

Patterns: forcescalepath

Relations: causetransform

Structure: hierarchy Level: primitive

Contributors: agent:metaphorex-miner