Following is the information supplied to me by a GM engine control specialist:
The engine control module uses a combination of the MAF
sensor and the MAP sensor to compute trapped cylinder air mass, which is
used as a critical operating parameter for much of the control system.
When the engine is in a "steady state" situation, the MAF sensor
measurement is an accurate representation of trapped cylinder air mass. A
"steady state" is defined as when actuators (such as the throttle or cam
phasing) that control air flow into or out of the manifold are not moving.
In this circumstance, the mass of air in the intake manifold is stable.
If one of those actuators is moving, it will generally cause the mass of
air in the intake manifold to change. Because of this change of air mass
in the manifold, the MAF sensor (which measures air entering the manifold)
is not correctly measuring the air in the cylinder (which is the mass of
air exiting the manifold). This transient phenomenon is called "manifold
filling (or emptying)".
During these transient maneuvers, the MAP sensor is a more accurate
representation of the state of the air entering the cylinder. Therefore, a
speed density calculation (using the MAP as its basis) is used during these
transient situations.
The only scenario where the MAP sensor is used exclusively is in case of a
MAF sensor failure.