Controlled and automatic uses of memory in depressed patients: effect of retention interval lengths.