Neural Mechanisms Underlying Overgeneralization of Valence-Based Learning

I will describe our studies in aversive/fear-conditioning that demonstrate that learning with punishments can lead to increase in perceptual thresholds and hence contribute to over-generalization. This is robust to different modalities, both of the unconditioned and of the conditioned stimulus, strongly suggesting a central neural mechanism. We then show that the Amygdala serves as such a central hub that modulates the width of generalization, as well as describe a functional neural network that supports this. We use both imaging (fMRI) in humans and anxiety patients, as well as animal models using electrophysiology and optogenetics. Finally, I will describe new data about information-gain during aversive learning, and the role of uncertainty and expected value underlying individual differences in over-generalization following aversive learning.