Rohin Shah comments on Imitative Generalisation (AKA ‘Learning the Prior’)