The concept behind a GFCI is simple: It takes two conductors to make a circuit, hot and neutral.
The GFCI looks at the current traveling along each conductor to be sure they are the same. If they are not the same, it trips and shuts off power to the circuit.
Regardless of where in a system the GFCI is placed, it will look at the downstream conductors for balance.
If power travels along the hot conductor, through the GFCI, through the load, and then has a pair of neutrals to return through, only half of the current that the GFCI expects to see will be on the neutral the GFCI is looking at. Therefore, it will trip; as it is assumed that such a condition entails a ground fault. Normally, when a GFCI sees 1A leaving on the hot, and 1/2A returning on the neutral, then the missing half amp would be returning to the source along a ground fault current path.
So, if you're protecting a whole feeder (not sure why you would) then you would need to ensure that all neutral current is returning to the neutral point of the GFCI supplying that feeder.