We have contributed to a new preprint: "Learning from imagined experiences via an endogenous prediction error". This study was led by Aroma Dabas in collaboration with Heidrun Schulz, and Roland Benoit.