I think mathematics has taught us that 'foundations' are an inappropriate way to think of intellectual structures.
The 'foundation' of mathematics from a functional point of view is really more around the 'middle'. It is really based in geometry and number theory, things humans easily grasp. But those seem complicated in other ways, so there are deep attempts to construct everything from very basic concepts. A short way in, these attempts just flushed out Russell's Paradox, which pointed out that what we look at as very basic concepts are, in fact, quite complicated in other ways.
Physics, and all the other sciences have the same problem, but it is in some ways both more obvious and easier to hide. We agree that physics is constructed by agreement of those practicing it, in a way that mathematics does not seem, at least on the surface, to be. We can look at mass or gravitation as a more basic concept, we can prefer to consider either charge or nuclear composition a simpler way of seeing things. 'Down' in complexity is clearly more relative.
But at the same time, within the field if physics, the idea that 'down' really exists somehow persists better. As we see particles and integrating theories become more and more embedded in assumptions about the more complex processes that are used to detect them and prove their existence, it does not occur to us that we are getting deeply embedded in layers of assumptions. Each layer seems so obvious and clear. They are the same sets of assumptions we rely on to keep our day-to-day technology working, so why would we question them?
Only when something really deep goes 'wrong' do we question this kind of thing: when we cannot rush into a beam of light and see our relative velocities add up; when we need something continuous to act less like a continuum in order for there not to be infinitely much heat created by a finite quantity of entropy. Then we need extensions of theories that really seem to almost destroy the theories themselves.
Kuhn's model of paradigms captures this well. Faced with a theory that eats itself in this manner, we renegotiate what is more basic, and what is more derived, so that we can rearrange our view of the subject and get a sense of complete coverage back. In the process, we discover some of what we thought was reliable data is just wrong, or that we no longer have ways of explaining it.
In that sense, all of these theoretical constructs could be totally reframed in completely different terms if the discoveries that created them happened in different orders, and were rearranged on different terms. They very well may be rearranged extensively in the future.
This puts science in the same position where Set Theory put math: what is common experience is really what needs to be maintained as the 'foundation' of the science, the part that cannot be lost in a shift of paradigms. The things that scientists within the field see as 'more basic' are in fact almost infinitely renegotiable and likely to be seen differently in some future generation.
There is in that sense no 'downward' direction in complexity in any science.