Interrupting recognition memory: Tests of familiarity-based accounts of the revelation effect