Understanding error-driven learning in infancy and early childhood

Countless times each day people make predictions about what will happen next. Although these predictions are often right, sometimes they end up being wrong. Instances where predictions do not match actual outcomes are ripe opportunities for learning, because learners can focus on building a more accurate mental model of the world.

This project asks how such “prediction errors” affect human learning early in life, focusing on infancy, when learning is especially important. The project tests the hypothesis that noticing an inconsistency between what was predicted and what was observed enhances young children’s learning across a range of contexts, and that individual differences in responses to such expectancy violations contribute to later cognitive outcomes.

Although prediction errors have been much studied in animal models and in human adults, nearly nothing is known about how they affect thinking in very young children, even though understanding this important form of learning may have broad relevance for understanding learning deficits, and for developing educational strategies. Therefore, this project will characterize how individual infants differ from one another in their ability to learn from prediction errors, and will chart the longitudinal consequences of these early differences, using a combination of measures including infants’ looking time and exploratory behaviors, as well as standardized batteries to index memory, vocabulary, and IQ.

In addition, the project will determine what kinds of predictions, and what kinds of errors, enhance learning in infants and young children. Overall, this research will help reveal the mechanisms behind a fundamental form of learning, with implications for understanding knowledge acquisition in children, adults, and artificial systems.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.