Inductive reasoning involves drawing general conclusions from specific observations. Unlike deduction, inductive conclusions are probable rather than certain — observing that every swan you have seen is white supports but does not prove that all swans are white. Induction underlies scientific hypothesis formation, everyday generalization, and category-based inference, making it arguably more important for real-world cognition than formal deduction.
Types of Induction
Categorical induction involves generalizing properties from known to unknown category members ("Robins have property X; therefore sparrows likely have X"). Causal induction involves inferring causal relationships from observed correlations and temporal patterns. Analogical induction involves transferring knowledge from a familiar domain to a novel one based on structural similarity.
Factors Affecting Inductive Strength
Osherson and colleagues (1990) identified several phenomena in category-based induction. The typicality effect: arguments with typical premises are stronger (robins having a property is better evidence for birds in general than penguins having it). The diversity effect: diverse premises are stronger (knowing that both robins and penguins have a property is stronger evidence than knowing that robins and sparrows both have it). The monotonicity effect: more premises generally make for stronger arguments.
Bayesian models formalize inductive reasoning as probabilistic inference: the reasoner updates prior beliefs about hypotheses (general conclusions) based on observed evidence (specific cases). Tenenbaum and Griffiths have shown that a Bayesian framework accounts for human inductive generalization remarkably well, predicting both the typicality and diversity effects from rational principles of probabilistic reasoning.