Is respiratory control more sensitive to small changes in arterial PO2 or in arterial PCO2?