The problem of detecting long-term forgetting: Evidence from the Crimes Test and the Four Doors Test