Understanding Learning Through Uncertainty and Bias

Resumen

Learning allows humans and animals to make predictions about the environment that facilitate adaptive behavior. Casting learning as a predictive inference problem can shed light on cognitive mechanisms of learning, including normative aspects that improve predictions in the face of uncertainty. Drawing on normative learning models, we illustrate how learning should be calibrated to different sources of uncertainty, including perceptual uncertainty, risk, and uncertainty arising from environmental changes. We show that such models explain many hallmarks of human learning in terms of specific statistical considerations that come into play when updating predictions in the face of uncertainty. However, humans also display systematic learning biases, leading to deviations from normative learning studied in computational psychiatry. In some cases, such biases can be explained as normative inference conditioned on inaccurate prior assumptions about the environment. However, in other cases, biases may reflect specific approximations to optimal Bayesian inference that people employ in order to reduce cognitive demands in the face of limited cognitive resources. We argue that learning biases provide a useful lens through which to understand the exact cognitive mechanisms underlying learning and how these mechanisms might go awry in psychiatric illness.

Áreas y palabras clave

Cómo citar (APA 7)

Bruckner, R., Heekeren, H. R., & Nassar, M. (2022). Understanding Learning Through Uncertainty and Bias. https://doi.org/10.31234/osf.io/xjkbg