Sample frequency and sample duration as sources of stimulus control in delayed matching to sample