Conference on Cognitive Computational Neuroscience

The ability to generalize knowledge to unobserved situations is a fundamental cognitive facette in humans. Scientific research on human generalization suffers from a severe segmentation into subfields that propose different mechanisms. Although some correlates in attention networks show a promising role in one of the proposed mechanisms, overall the neural underpinnings remain vague. We combined computational modeling with model-based fMRI to identify behavioral and neural commonalities between generalization in Pavlovian and instrumental learning. Here we show that the behavior can be well explained as a combination of previously suggested mechanisms and extend the role of the frontoparietal attention network from representation learning to stimulus generalization.