It takes the force of the upper portion to break the slab to column connections. The floor connections resist that force, applying force to the upper portion. The columns also resist the downward force of the upper portion, applying more force to the upper portion.
You write of "the force of the upper portion", as if that was an intrinsic property of the upper portion. It isn't. A force arises (and changes) in response to things hitting things and deforming them elastically or inelastically.
But yes - the upper perimeter exerts a force on a floor assembly as the two collide, and the floor assembly exerts the same force, only in the opposite direction, on the upper perimeter.
That force increases and increases as both parts deform elastically (and yes, this in turn exerts a force on the lower columns via truss seats) ... until ... Well, until when? -> Until one of the elements involved reaches its load capacity, its maximum elastic deformation, and starts deforming inelastically.
Now which element would that be - the upper column, the lower column, or the floor truss, or the truss seat?
Now plastic deformation continues - until either the fall of the top part is halted, or a first element no longer can deform plastically and instead breaks, ruptures, tears out - in other words: Fails and splits in (at least) two parts.
Now which element would that be - the upper column, the lower column, or the floor truss, or the truss seat?
Theory, observation and even intuition tell us that the truss seat is the weakest element that will fail - long before the columns get even near their elastic limit.
When the truss seats fails, the lower column unloads, the upper column unloads mostly (still some lesser force to accelerate, perhaps deform the floor slab), and the floor goes down.
(Ok, floor goes down after ALL truss seats have failed - which they will do in very rapid succession, as the same drama plays out at all truss seats within a short period of time).