I believe you're correct. I have an email from a GM engine control engineer that states MAF is used exclusively above what we know as the B0120 data point and that the system uses a combination of MAF and VE below that point, depending on whether the engine is in steady state or transient mode. The only time it uses VE exclusively is in the event of a MAF failure. The actual e-mail is on another computer-- I'll post it tomorrow.
For what it's worth, I've felt for a long time that below the B0120 point, the switch between VE table and MAF input operates along the lines of the high and low octane scalar-- it's never exclusively one or the other, but biased one way or the other depending on operating conditions. In this case, engine speed rate of change alters the bias. I don't know this for a fact, but all my experimenting points in that direction. After being the recipient of flack and other foreign matter about my cat-skinning procedure, and the rough edges it might leave, I put a vehicle that I had tuned using only changes to the VE table, into speed/density. The changes I see in long term fuel trims range between 0 and 12%, but there doesn't seem to be any consistency in the degree of change during steady state operation, as opposed to during throttle transitions. I need to do some more logging to see if any type of pattern exists.