A flexible and generalizable model of online latent-state learning
Fig 2
Simulated associative strengths of cues during blocking, overexpectation, and conditioned inhibition experiments, which the Rescorla-Wagner (RW) model famously could explain.
Built upon the RW model, our latent-state model and the Gershman (2017) model can also explain these experiments. Gray dashed lines demarcate experimental stages.