Using ecphoric confidence ratings to discriminate seen from unseen faces: The effects of retention interval and distinctiveness