An intuitive solution to three puzzling quantum experiments
How the absorber perspective fixes quantum weirdness
In this excellent article, Chris Ferrie proposes a series of mind-bending experiments using a Mach-Zehnder interferometer (an experimental setup involving lasers, mirrors, beamsplitters and detectors), daring the readers to find an intuitive explanation.
I took up the challenge and posted a comment explaining the results in a simple way, so I decided to turn that comment into an article, modifying the images that Chris used (I hope he doesn’t mind).
I was only able to explain these experiments in such a way because my quest to understand reality has led me to conclude that the universe is not first elements and laws, then conservation and consistency checks, but first conservation and consistency checks, then elements and laws.
From this point of view, we can describe all the proposed experiments with a single mechanism, without having to define photons as waves or particles, or having to choose between different paths for no reason.
The basic idea is that, in the absence of constraints, nature always considers all possible ways to construct reality. This is related to the principle of least action. Nature is always as efficient as possible, but also consistent with all global, present and past conditions. Events in spacetime always connect through all possible light-like paths that keep the self-consistency of the universe, and it is the presence and evolution of matter in spacetime that limits which light-like paths are possible, constraining reality in such a way that whatever unfolds must be consistent with the conditions imposed up to that point.
So instead of waves or particles, we’ll describe photons as possible light-like relations between emission and absorption. Photons are possibilities “filtering” through spacetime at the speed of light, with the final outcome that emerges fully consistent with one of the possibilities allowed by the constraints imposed by the whole experiment.
We usually describe these experiments as evolving forward in time from the source to the detectors, but that often forces us to go back and change something retroactively. But in reality, we can never say that something left the source until it’s detected elsewhere, so it’s best to analyze these experiments by assuming an end result, while looking back in time until we either arrive at the source unscathed or find an inconsistency that makes that end result impossible. That way, we don’t have to go back and change anything after the fact, because the sequence of events is either consistent and the result can unfold, or inconsistent and the result cannot unfold.
We’ll also define light or photons by the way they arise to reality. Whatever they are during their journey at the speed of light, they always begin and end in matter vibrating back and forth, so the emission and absorption events are actual matter oscillations that imply real motion in one direction or the other, and wherever their propagation requires them to “choose a path” in free space, we’ll force them to take all possible paths, as explained.
So let’s begin:
1. Light as a wave
This is the first configuration of the experiment, showing the final outcomes when light is considered a wave:
Before explaining why the top observer is always the one detecting light, we need to make clear how beamsplitters work: They only invert oscillations when they’re reflected from the dielectric side (the top branch in the experiment), leaving them as they were in any other case, as shown below:
Then, the usual explanation begins by analyzing how light propagates from the laser to the different observers. In this view, the laser oscillations are split at the first beam splitter and travel along both branches until they reach the second splitter.
For the bottom observer, the first splitter flips the oscillations in the top branch while the second one does not, so the oscillations in both branches are incompatible, and this observer receives opposite oscillations that cancel each other out by destructive interference. Thus, although the final path to the bottom observer has no obstacles, it can never detect light.
Meanwhile, for the top observer, both splitters flip the oscillations in the top branch leaving them as they were, so the oscillations from the top and bottom branches are compatible and can add up constructively.
The end result is that light always reaches the top observer, and never the bottom one. This description is all well and good as long as we don’t consider what happens to the opposing oscillations in the final bottom path when actual light is detected at the top. Somehow, the opposing waves must dissapear for the top ones to add up!
We will now look at this case from the new perspective, starting from one of the possible absorbers, and looking back in time while considering all possible paths to the laser, showing how this new view makes much more sense.
Let’s start with the top observer. There’s a free path to the second beamsplitter, so light from there can reach it without problems:
Then light could have merged at the second beamsplitter by passing through both branches. For this to happen, the oscillations must conform to the rules of the splitter and be compatible with each other so that the top observer can actually detect light. Therefore, the splitter had to flip the oscillations in the top branch, but other than that, there are no obstacles or inconsistencies here:
Finally, there’s only one free path from the laser to the first splitter. This splitter also has to comply with the rules, so it also inverts the laser oscillations reflected to the top branch. The double inversion in the top branch is compatible with the oscillations in the bottom branch, so light could indeed come out of the laser, propagate to the first splitter, split into two branches, hit the mirrors, merge in the second splitter and reach the top observer without any problems or inconsistencies, so this outcome is possible and does happen:
Now we need to explain why light doesn’t take the path to the bottom observer. Let’s assume that it actually receives some light:
Since there is no constraint preventing light from going from the second splitter to the bottom observer, the reason light doesn’t take this path must be found elsewhere, so we keep looking back. We make sure that the oscillations from both branches satisfy the rules at the second beamsplitter and are also compatible between them so that the bottom observer can actually detect light. The second splitter transmitted the oscillations, so they are not inverted there this time. Then we keep looking back until we reach the first splitter, as before:
Again, there’s only one free path from the laser to the first splitter, and again, this splitter must flip the oscillations as before. But since the second splitter didn’t flip the oscillations, the top branch is now incompatible with the bottom branch, and there lies the problem: For the bottom observer to actually receive some light, the laser particles would have to physically oscillate back and forth at the same time! This, of course, never happens, and that’s why no light ever reaches the bottom observer:
Both explanations (old and new) have a free path that light does not take, but where the old explanation requires a variety of light contributions propagating near observers, somehow always collapsing at the top, the new explanation has no oscillations at all that can reach the bottom observer, since there is no physical means by which actual light can emerge from the laser to reach the bottom observer in this setup.
So although it seems there are two possible outcomes, there was only one physical outcome all along (light reaching the top observer), and we don’t need light to propagate through all possible (and impossible) paths through spacetime to always collapse at the top, since from the new perspective, the physical constraints imposed by the elements of the experiment (including the final observer) shape the path that light has no choice but to take.
2. Light as a particle
In the second experiment, we use the same setup, but now the laser fires one photon at a time, so we consider light to be a particle. If we now assume that the splitters either transmit or reflect the photons, but they cannot be split, then we’ll get detections at both observers, since both splitters will randomly transmit or reflect photons no matter where they come from.
This image shows a photon taking the top branch after being reflected by the first splitter, but it would be the same if it were transmitted to the bottom branch:
But actually, when we do the experiment, all the photons end up back on top, even when they’re fired one at a time. To explain this, we have to remember that photons are not really indivisible particles, but rather wave packets or trains of oscillations, and by treating them as “unbreakable”, we changed how beamsplitters react to them. So this is a more accurate picture of what’s going on now:
If a photon is simply a set of oscillations, both explanations naturally lead to the same result as before: In the old explanation, two sets of reinforcing oscillations add up on their way to the top observer, while two opposing oscillations cancel each other out on their way to the bottom observer…
… while in the new explanation, the laser particles can physically oscillate to release a photon that can reach the top observer, but cannot oscillate in two opposite ways to release a photon that can reach the bottom observer:
Still, the old explanation has to face the same problem of opposing oscillations that must retroactively collapse or disappear when the photon hits the top observer, while in the new interpretation, no ad hoc mechanism is needed, since the only way to prevent a photon from taking a free path is some previous inconsistency elsewhere in the experiment, such as requiring matter to do incompatible things, asking photons to retroactively change their nature, or forcing them to “choose” one path or the other for no reason.
Thus, photons don’t change from wave to particle to adapt to our actions retroactively, they always take all possible paths through spacetime while limited by both material obstacles and consistency rules that prevent certain outcomes from becoming real.
3. Another observer
The most interesting case, which perfectly illustrates the pitfalls of the standard explanation, happens when we introduce another observer between the beamsplitters. By doing this alone, when the new observer detects no light… both final observers may end up detecting light!
Chris suggests that this is counterintuitive and that the results should be the same as before because the new observer did not disturb the light, but this is not entirely true. The new observer, whether detecting light or not, always blocks the bottom branch, so light cannot propagate the same way as before.
But what’s puzzling about the usual explanation is that the mere presence of the new observer seems to change the behavior of the splitters, forcing them to treat photons as particles (although we usually say that the photons themselves change from waves to particles). The first splitter now either sends photons to the bottom branch, where the new observer detects them preventing any further propagation to the top branch, or it sends them to the top whenever the new observer didn’t detect anything. As for the second splitter, it now has no choice but to randomly transmit or reflect all incoming photons.
This change in behavior stems from the misleading view of interactions as oscillations that propagate forward in time from the emitter until they collapse at a given detector, either by the rules of wave interference (as in the previous experiments) or by random chance (as seems to happen here).
But with the new explanation, the rules are always the same and don’t change depending on the setup: interactions always connect an emission event to an absorption event by all possible light-like paths through spacetime, and the end result is always one of the consistent possibilities allowed by the constraints imposed by the physical elements and processes that define the experiment.
So we’ll finally explain why placing a new observer after the first splitter allows photons to reach both final observers. We start by analyzing a detection by the new observer, since this is a new result to consider. There’s a free path connecting this observer to the first beamsplitter…
… and another free path connecting this splitter to the laser, so this is a perfectly valid outcome. When this happens, the other two observers had no chance to detect anything, and in fact, the presence or existence of those observers is completely irrelevant here:
Note that the new observer could also be placed in the top branch, since it doesn’t matter whether the laser oscillations are flipped or not, as there are no other oscillations that should match them.
Now, let’s see what’s needed for the top observer to detect a photon. There’s a free path from the second splitter to the top observer, so no problem here:
But now that there’s an obstacle in the bottom branch (the new observer), it is no longer a free path. If laser light were to go to the bottom branch, it would be detected by the new observer, and we’d be considering the result we just explained. But now we’re considering that light has already reached the top observer, which automatically rules out a detection by the new observer and is only possible if light came through the top branch:
So the new observer, by its very presence, splits what was one outcome with two paths into two outcomes with one path each. Since there’s only one possible chain of free paths connecting the laser to the top observer, and there are no other oscillations that must match, it doesn’t matter that the photons are flipped twice, and this is a possible outcome that can happen and does happen:
Now, we analyze if a detection at the bottom observer is possible. There’s a free path between this observer and the second splitter, so this is totally OK:
But now we see that in this case, too, the presence of the new observer blocks what was previously a free path, so now light can’t reach the bottom observer from below, and the second splitter can only send light to it from above:
So the free path that imposed conflicting constraints on the laser is “disabled”, and light can now reach the bottom observer only through the top branch, which doesn’t require the laser to oscillate in two incompatible ways. Then, since there’s only one chain of free paths connecting the laser to the bottom observer, it doesn’t matter that the laser oscillations are only inverted once, and photons can reach the bottom observer without problems, so this outcome becomes possible and then happens:
As you can see, we were able to explain all these experiments without having to rely on wave-particle duality or wavefunction collapse. We only required that information must always travel through all possible light-like paths available, and that the final outcome that emerges must be one among the many consistent possibilities allowed by the global, past and present constraints of the experiment.
The old explanation, by analyzing the situation forward in time, considers all possible outcomes in superposition, as if all observers were competing for the same photon (new observer AND top observer AND bottom observer), but this is where puzzling behavior and retroactive changes arise, while the new perspective, by analyzing the situation backwards in time, considers the results one at a time (new observer OR top observer OR bottom observer), and each separate outcome perfectly represents a different realization according to the constraints that also determine the different rates at which these realizations unfold.
Thus, analyzing things from one end or the other fundamentally changes the logic to be applied, along with the concepts and mechanisms needed to explain each situation. We, as material observers, are naturally immersed in the flow of time, so we are used to understand possibilities as evolving in time as well. But the fact is that nature always realizes only one outcome, perfectly aligned with the moment it unfolds, and always showing the result after the fact. And that’s why I argue that the observer’s perspective is closer to how nature shapes reality.
