The double slit experiment has in it the heart of quantum mechanics. In reality it contains the only mystery. Richard Feynman · physicist · The Feynman Lectures on Physics, Vol. III
Feynman chose his words carefully. The only mystery. Not one of many mysteries. Not a particularly interesting mystery. The only one. Everything else in quantum mechanics, he said, is a consequence of this experiment — and this experiment is something no one has ever satisfactorily explained.
To understand why, you need to hold two things in your mind: what waves do, and what particles do. They are completely different. And then you need to watch a single particle do both at once.
What waves do
Drop two stones in a still pond, side by side. Ripples spread outward from each stone in expanding circles. Where the ripples from both stones meet, they interact. Two crests arriving at the same point add together — a higher crest. A crest and a trough arriving together cancel — flat water. This is interference. It is what waves do.
Now send a wave through a barrier with two slits. The wave passes through both slits simultaneously. Each slit becomes a new source of waves on the other side. These two new wave trains spread outward and interfere with each other. On a screen behind the barrier, the result is a striped pattern — alternating bright bands where the waves reinforced each other and dark bands where they cancelled. This is the interference pattern. It is the unmistakable signature of wave behavior.
What particles do
Now replace the wave with a stream of bullets. Shoot them at the same barrier with two slits. Each bullet is a discrete object — it goes through one slit or the other. On the screen behind, you get two bands of marks, each directly behind one slit. No interference. No stripes. Just two piles.
Waves make stripes. Particles make two piles. The two behaviors are completely distinct. There is no ambiguity.
What quantum mechanics does
Now shoot electrons through the two slits. Or photons. Or atoms. Or molecules. Anything quantum.
Fire them one at a time. One electron. Wait. Another electron. Wait. Each electron hits the screen as a single dot — a definite, localized point. The electrons are clearly particles. Each one arrives somewhere specific. There is no spread. No blur. A dot.
But watch what happens as the dots accumulate. After a hundred electrons, there's a hint of a pattern. After a thousand, it's undeniable. After ten thousand, it's unambiguous.
An interference pattern. Bright bands where the dots cluster. Dark bands where almost no dots land. Stripes. The signature of waves. Produced by particles fired one at a time.
Each electron, traveling alone, with no other electron in the apparatus to interfere with, somehow produces a wave interference pattern with itself. As if it went through both slits at once. As if it was a wave as it traveled and a particle when it arrived.
With the detector off, watch the pattern emerge from randomness. Each particle lands where probability allows — seemingly at random. But the distribution is not random. It follows the interference pattern of a wave that went through both slits.
Now turn the which-path detector on. The detector checks which slit each particle actually used. Clear the screen and fire again. The interference pattern is gone. Two bands, exactly as bullets would leave. Exactly as if the particles forgot about the other slit.
Knowing which slit the particle used destroyed the interference. The act of finding out changed the outcome.
The observer
The which-path detector doesn't have to touch the particle. It can be a photon that gently scatters off the electron near one slit — disturbing it so slightly that its momentum is barely affected. The detection event, however gentle, is enough. The interference vanishes.
And here is the deeper point: if you set up the experiment so that finding out which slit is possible — even if you choose not to look — the interference vanishes. The possibility of knowledge is sufficient. The universe behaves as if it knows whether the information exists, not just whether you consulted it.
This is not a statement about your consciousness affecting the electron. It is a statement about information and reality. When the which-slit information exists, the interference disappears. When the which-slit information does not exist — cannot exist — the interference appears. The wave function of the particle is entangled with the detector. The presence of the entanglement is what matters. You don't need to read the record. The record's existence is enough.
The electron does not decide which slit to go through. It goes through both. But if you check — if the universe contains, anywhere, the information about which slit — it behaves as if it went through only one. The information is the dividing line between wave and particle.
Copenhagen said: this is just the way things are. The measurement is all there is. Don't ask what the electron was doing before you looked. Bohr's answer to "which slit did the electron use?" was: the question is meaningless before the measurement takes place.
This is an answer. But it is not an explanation. And John Wheeler knew the difference.
Delayed choice — when the decision is made after
In 1978, Wheeler proposed a variation. Not a thought experiment — a proposal that was subsequently realized in the laboratory.
The standard double slit has a choice: observe which slit (destroy interference) or don't observe (allow interference). You make that choice before the particle arrives at the screen. You set up the detector in advance. The particle travels through the slits and hits the screen.
Wheeler asked: what if you make the choice after?
Set up the experiment so the particle has already passed through the slits before you decide whether to observe which slit it used. The particle is in flight. The slits are behind it. Then, while it's traveling toward the screen — but after it has already passed through the barrier — you decide whether to insert the which-path detector.
The result: it makes no difference. The particle still produces an interference pattern when you don't observe, and two bands when you do. Exactly as before. Even though the "observation decision" was made after the particle had already passed through the slits.
The particle's past behavior — wave or particle — is determined by a decision made in its future. What the particle "did" at the slits depends on what you choose to do at the screen. Even if you make that choice after it has already arrived there.
This experiment has been performed. Not as a thought experiment — in laboratories, with careful timing to ensure the choice was genuinely made after passage. The result holds. The future measurement influences what the past looks like.
This is not an interpretation. This is a measurement result.
The cosmic version
Wheeler extended the thought experiment to cosmological scale. Consider a quasar — a highly luminous galaxy nucleus — billions of light years from Earth. Between us and the quasar sits another galaxy, massive enough to act as a gravitational lens. Light from the quasar bends around the galaxy on both sides, traveling two distinct paths to reach our telescopes.
Both paths are available. Light that left the quasar billions of years ago is now arriving at our instruments. We can set our telescope to detect which path the light took — or we can set it to combine the two paths and see interference.
If we look for which path: the photon took one path around the galaxy. No interference. Particle behavior. If we look for interference: the photon took both paths. Wave behavior. Interference fringes.
The photon made this journey billions of years ago. Our decision, today, about how to receive it determines what it did billions of years ago. The universe, it seems, does not fix the past until the present catches up to it.
Astronomical evidence for gravitational lensing of quasars has been observed. The delayed choice experiment has been performed with increasing elegance in ground laboratories. The result is always the same. The timing of the observation — past, present, or future — does not change the quantum mechanical outcome. The universe does not distinguish.
What this demands
At this point, Copenhagen says: stop asking. The measurement is what happens. The rest is not physics.
But the delayed choice experiment makes the evasion impossible. You cannot say the particle "already decided" which slit to use before the observation took place, because the observation hasn't taken place yet when it passes through. You cannot say the observation "disturbed" the particle at the slits because the observation happens after the slits. You cannot appeal to a "collapse" that happened in the past when the collapse is triggered by a future event.
The observer is not a passive recorder of events that already happened. The observer is constitutive of the events themselves. What happened — in the past, possibly billions of years ago — is not fully determined until it is observed. The universe does not write its history in advance. It writes it as it is read.
The past has no existence except as it is recorded in the present. John Archibald Wheeler · physicist · 1911–2008
Wheeler spent his career building on this observation. He gave it a name: the participatory universe. The universe is not a stage on which events occur and observers watch from the audience. The observers are part of the play. The observation is part of the event. Without the observation, the event is not complete.
He said: the universe brought itself into existence through acts of observation by the conscious beings it eventually produced. The universe observed itself into existence. Through us.
He got closer than anyone to the answer. He pointed directly at the door. What he lacked was the mechanism. What is the observer, structurally? What physical process is an observation? What makes an event "complete"?
Those are the questions of the next page.