Why is it that descriptions arising from different contexts do not fragment, but instead remain coherent as a single world at all?
I’d say it’s ultimately because not all things are relative. Some things do acquire consistent values in different perspectives. Velocity of a train might be relative but acceleration is not, it’s an absolute property which everyone can agree on who is accelerating and who is not. These absolute properties exist sort of as anchor points that make perspective transitions consistent between one another. Quantum mechanics is relative/contextual for all the variable properties of particles, but the intrinsic properties are absolute, like charge and spin, which again serves as an anchor point.
This paper also goes into detail on how the logic of quantum mechanics also pushes relative facts to become “stable” facts on macroscopic scales through the process of decoherence. The logic of the theory guarantees that even if you are making a purely relative prediction you will always predict that if you observe something and immediately ask someone else to observe it then they would perceive the same thing, and so the more things that “observe” it (not necessarily people but even the environment interacting with it) causes the relative property to become more stable among all observers involved.
But there is method in this madness. If I know that you have looked at the butterfly’s wings, and you tell me that they were blue, I know that if I look at them I will see them as blue: this is what the theory predicts, despite the fact that properties are relative. The fragmentation of points of view, the multiplicity of perspectives opened up by the fact that properties are only relative, is repaired, made coherent, by this consistency, which is an intrinsic part of the grammar of the theory. This consistency is the basis of the intersubjectivity that grounds the objectivity of our communal vision of the world.
— Carlo Rovelli, Helgoland
I tend to side with Francois-Igor Pris who argues that you cannot meaningfully have a philosophy that both includes an objective world and also describes it as dependent upon observation/engagement without it running into an infinite regress.
We can only ever know that our act of observation disturbs a system by comparing it to a more subtle form of observation which presumably does not or perturbs it very little. We know the Hawthorne effect is real because you can compare an experiment where a person is observed in a very obvious way where they are clearly aware of it to one where the observation is secretive, and interviewing them later you can confirm they were not aware of it.
But if we are talking about reality being fundamentally dependent upon observation, then this dependence applies to all observations by definition, and thus there is no such thing as a non-subtle observation that you could compare it to. You could never actually empirically confirm that your act of observation is something that is active that physically alters the system. You would have no control to compare
Indeed, there is kind of an infinite regress here that Pris explains in his book Контекстуальный квантовый реализм и другие интерпретации квантовой механики.
Observation at a fundamental level has to be treated as passive, and thus must always be treated as observer-independent or else it leads to a vicious logical circle. An objective world (one that contains “facts”) that is observer-dependent is just not logically consistent.
This does seem a bit strange, because clearly what I observer is different from what you observe, so it seems like there is a kind of “observer-dependence.” But what the philosopher Jocelyn Benoist has argued as well as Pris is that we should distinguish between observer-dependence and contextuality.
If I am sitting on a bench and see a moving train go by, and you are inside the train, we will both perceive the train to be traveling at different velocities. Velocity is sometimes described as “observer-dependence,” but this implies that velocity somehow depends upon the existence of conscious observers and is thus subjective, when this is wrong. Velocity is clearly an objective feature of the world and has no fundamental dependent upon conscious observers, as you can define velocity even relative to inanimate things like a rock that is sitting beside the track.
The physical reality of velocity would be obvious if I stepped onto the tracks in the path of the train. You would not be harmed, as you are riding the train thus its velocity from your perspective is zero, but I would be harmed because the velocity of the train in my perspective is non-zero. No one would be surprised if I got harmed because “velocity is subjective” or “observer-dependent,” we all intuitively understand that velocity is a real, objective feature of the world.
The main premise of “contextual realist” philosophy as Benoist and Pris call it is to remove the anthropomorphic character from this kind of difference in perspective, to stop calling it “observer-dependent” as if conscious observers play some sort of fundamental role here. It is more accurate to call it contextual. You and I perceive the train to travel at different velocities because we are perceiving it under different contexts.
Contextual realism takes all physical properties of the world to be contextual in the sense that they only can be meaningfully assigned an ontology once you specify the context under which it is realized. The train’s velocity, between you and I, is realized in a different context, and so it really is ontologically different. Contextual realism extends this to all things. All of reality is context-dependent, but not observer-dependent. The conscious observer plays no fundamental role, and what they observe is always, on a fundamental level, passive and just identifies what is really there, but what is there depends upon the context under which the observation is made.