Barbara Tversky (Synthesis)

Among the findings that Barbara Tversky's research program has established over four decades, one is especially pertinent to the AI transition: that the mind does not build tidy, metrically accurate models of the domains it reasons about. It builds cognitive collages — spatial representations that are partial, perspectival, assembled from multiple viewpoints and encounters with the domain, distorted in ways that are systematic and functional rather than random and erroneous. The person who navigates a city is not consulting a survey map. She is consulting a collage of routes walked, landmarks noted, and partial views glimpsed — a representation that sacrifices accuracy for usability, that emphasizes the relationships that matter for movement while simplifying the ones that do not. This is not a limitation of human cognition. It is one of its most successful features: the cognitive collage preserves the information needed for the task while shedding the overhead of information that is