Chapter 44: Scaffolding Intuitive Rationality

The chapter "Scaffolding Intuitive Rationality" explores the study of intuitive judgment, which occupies a "murky zone" of psychological processing—implicit, tacit, and nonlinguistic behaviors that fall between explicit, language-dependent deliberation and basic, stimulus-driven reflexes. This intuitive cognition is critical for tasks like reading emotions or driving on autopilot, but its structure is typically opaque to introspection. The author proposes the "build-up" strategy, starting from basic associative learning, to understand this complex behavior, contrasting it with the "revise-down" strategy, which attempts to adapt formal logical models to nonlinguistic agents. The chapter presents a novel hybrid model for intuitive practical inference that combines internalism and externalism. Judgments are deemed rational in the explanatory sense (internalism) when they are intensionally sensitive to the modes of presentation in the agent’s category representations, often relying on formal models of similarity judgments to govern actions. Conversely, these judgments are rational in the justificatory sense (externalism) only when they achieve ecological validity—a successful fit with the informational structure of the environment. This externalist view is necessary because classically rational, optimizing strategies are often instrumentally ineffective and vulnerable to overfitting when dealing with noisy, unpredictable environments. The process of "building up" complexity involves making similarity-based inferences more sophisticated by matching increasingly abstract features, which technically means developing representations resistant to nuisance variables like rotation or scale. When abstraction is combined with predictive learning based on future contingencies, agents can reshape their representational similarity spaces, allowing for insightful generalizations, such as ravens inferring hidden competitors through a peephole based on their own prior pilfering experience. Abstraction may eventually yield unsaturated representational functions, or "tintamarresque slots," which are placeholders for individuals in an equivalence class. Although full planning ability benefits from linguistic scaffolding, a limited form of mental time travel can be achieved by chaining abstract associations. Ultimately, the chapter contends that researchers should reject applying abstract logical norms to nonlinguistic thought—a bias termed anthropofabulation—because classical logic and decision theory are highly scaffolded scientific idealizations, not descriptive models of everyday psychological reasoning.