Infants' Domain-General Responses to Expectancy Violations.

The proposal that the human mind contains modular systems dedicated to processing distinct types of input has been influential for many decades. These systems are thought to function early in development; for example, infants recruit different brain areas and have different expectations about the behavior of physical objects versus social agents. Yet whether these specialized systems drive behavioral responses independently of one another early in life remains unknown. To find out, we examined individual differences in infants' responses to events involving physical objects versus social agents. Unlike previous studies, in which infants are tested with an event from a single domain, we showed 14- to 27-month-old children 16 different events that either accorded with or violated expectations about object physics or social behavior. We found that infants' increased visual interest in the surprising over expected events was better fit by a single latent factor than by separate factors for physical and social expectations. This similarity in surprise responses to physical and social expectancy violations increased with age. Finally, individual differences in infants' surprise across domains predicted later curiosity, highlighting the role these differences play outside the laboratory and across time. These findings suggest that although separate bodies of knowledge support infants' expectations about objects versus social entities, violations of those expectations also recruit domain general processing. SUMMARY: Infants' responses to expectancy violations across physical and social domains reflect a single, domain-general prediction error mechanism. Domain-generality of prediction error responses strengthens with age, suggesting growing generalization of error detection across knowledge systems. Early individual differences in prediction error responses predict later curiosity-driven exploration, but not other behavioral traits. Findings bridge core-knowledge and predictive-processing frameworks, identifying surprise as a developmental mechanism linking error-detection to curiosity.