You can read or listen Don Hoffman and Rob Reid conversation explaining the experiment and its conclusions. Below is a shortened and captioned version of the transcript. To listen to it start at at 38:40.
A consistent Experiment
"This is an experiment it (...) has been carried out probably thousands of times now. (...) Highly, highly, consistent from experiment to experiment. This is an utterly non-controversial statement about the way reality works. Specifically, light, electrons and certain other things sometimes behave like a bunch of particles and at other times behave like waves. (...)
Sometimes light behaves as a particle
So first to visualize the particular nature of light. Imagine you have one of those machines that shoots tennis balls, and they're all soaked in blue paint. And you're standing in front of a wall that has two large vertical slits in it, and behind the wall is a canvas, like an artist's canvas, a couple feet behind the wall. And you start firing these tennis balls. What'll happen is that most of the tennis balls are just gonna bounce off the wall, but a certain set of them are gonna go sailing on some angle through the left slit, and some of them are gonna go sailing on another set of angles through the right slit. After a certain amount of time, if you go in and you see your artwork that's resulted, you're going to see a cluster of blue spots where the tennis balls came through the left slit, and another cluster where they came through the right slit. And that is basically what would happen if light were behaving like a particle.
So in a sense, these tennis balls are behaving very particle-like, very much like lone photons should behave, photons being the tiniest indivisible units of light. Particles flying through two slits should behave like our tennis balls and create two clusters of dots. With photons we use photographic film or a digital sensor to record where they land rather than a canvas.
Sometimes light behaves like a wave
Now to visualize what happens when light acts like a wave. Imagine we have a huge deep pan of blue paint. It's so deep we can lower the wall with the two openings halfway into it. So now it's like we have two arched doorways on this little sea of paint instead of two slits. And the pan is filled right to the rim. One more drop of paint and it's overflowing. And the far edge of the pan is touching a fresh canvas. So we're gonna make so more art.
Now on our side of the pan across from the canvas, we start dropping rocks into the paint at regular intervals to make a series of waves. Those waves radiate out in the form of expanding semicircles toward the wall with the two arched doorways. Now when a wave hits, most of it's stopped by the wall. But the parts that go through the doorways become two mini waves, which themselves start radiating out as semicircles. Now when these two new sets of waves meet, they're going to interfere with each other. Sometimes the crest of a left side wave will meet with the crest of a right side wave, and they'll join up and become a higher crest. Other times a crest will meet a trough, and they'll cancel out. At that point of the wave front, there will be no wave. Eventually these much more complicated wave come to the far end of the pan, and they splash over the side. Where two crests have teamed up, they'll make a higher and deeper mark. Where a crest and a trough meet, there's nothing, so they'll make no mark at all. And the result will be a banded pattern of paint on our canvas, light dark, light dark, light dark, or a banded pattern of light on our sensor if we're using light instead of paint. And here our light is acting like waves.
Why one behavior or another? Let's try to shoot one photon at a time
Okay, now imagine you're a grad student replicating the famous double slit experiment for the thousandth time. If you just shine a steady light at the two slits, on the far side you're going to get the classic banding pattern of waves. So how do you get the two shotgun patterns? Well a logical answer might be to start firing the light just one photon at a time, because then you're only sending over solitary packets. And you'd think a solitary packet has to either go through the left slit or the right one. Once you've fired enough single photons to create a discernible pattern on the sensor, it's gonna be two shotgun clusters, right? Well, wrong. You actually get the banded pattern again. It's like the photons, despite going through one slit or the other one at a time, somehow choreographed themselves. They coordinated their landings on the sensor so that instead of producing the dual shotgun pattern of random particles, they made this very specific pattern which should only be made by waves of light. (...) It's like a conspiracy of photons that are ganging up to play a trick on you.
So let's record them to see which slit photons went through
So to figure out what's going on, you put a detector on each slit, which will detect whether or not a photon goes through it. This way, you'll know which side each photon goes through before making its contribution to the banding pattern. But now, the photons suddenly start acting like particles, like tennis balls, and they obediently create the two shotgun patterns rather than the banded interference pattern. It's like they know they're being watched, or measured to be more precise. And sure enough, if you turn off the detectors, the photons go back to making the banded patterns of interfering waves. Now we could go on for days about what might be causing this, but the cliff notes are that many decades after discovering this, thousands of the brightest minds on earth have no idea. (...)
Rob Reid: So what that says, and I'm going to quote something that you've written, we don't passively observe a preexisting objective reality, but actively participate in constructing reality by our actions. And that's I guess how you would tie the double slit experiment to this whole argument that we create reality by looking at it. But you're talking about chairs not photons.
What is realism and why it is false? Why observing things changes them?
Don Hoffman: That's right. So the double slit experiment is one of the weirdest things in quantum mechanics, and it indicates that when you don't observe, you cannot say which slit the particle went through. You can't say what its position is, and for most particles you can't say what it's momentum is, what it's spin, all these properties. Unless you observe, you cannot say what those properties are.
Rob Reid: And those properties don't actually take a property until they are observed.
Don Hoffman: And that's the wild thing. And Einstein was worried about this. This is why Einstein didn't like quantum mechanics. He wanted a world in which particles really had positions and momenta even when they weren't observed, and quantum mechanics doesn't allow you to specify what those positions and momenta are unless you observe. So Einstein said, "Well, whatever quantum mechanics is, it's incomplete. Maybe it's not false, but it's not complete. There's really a position, really a momentum." That's called realism. That even if you don't measure particle as a real position, a real momentum, real definite values, that's called realism. And there's never-
Rob Reid: That reality exists whether you measure it or not.
Don Hoffman: Exactly right. The other concept is locality, that these properties like position and momentum have influences that don't propagate faster than the speed of light. And so local realism is the claim.
Rob Reid: And that is an intuition that would feel to a reasonably physics savvy person about as strong as the proposition that the earth is flat. It just seems overwhelmingly logical.
Don Hoffman: That's right. It seems like of course it's gonna be true, and it turns out it's false. It's been tested many, many times. There was a theorem by John Bell in 1963 that showed us how we could in principle test whether local realism is true, and local realism is false. It's been tested many, many times, all sorts of loopholes have been proposed and closed, and everything we close a loophole we still get the same effect. Local realism is false.
Delayed double slit experiment. Can we make decisions on past events?
Rob Reid: And then John Wheeler, another great physicist who I think did a lot of collaboration with Niels Bohr and other people in the early 20th century on nuclear work, he came up with a thought experiment which make this weird thing slightly weirder.
Don Hoffman: That's right. So Wheeler has what's called the delayed choice experiment. It's the double slit experiment, but after you've shot that photon to the double slits, and you wait until it seems like the photons should already be past the double slits, and then you decide am I going to measure which slit it came through, or am I not going to do that and I'm just going to let an interference pattern occur. So I wait until after the photon should've already been passed to make that decision. And quantum mechanics says it shouldn't affect things. It shouldn't matter when you make that decision. So Wheeler's delayed choice experiment has been done, and quantum mechanics is right. You can make the choice after quote-unquote the particle should already gone through, past the slits, and you still get the same effect.
Rob Reid: So you're firing photons in this experiment from a significant distance. Light travels, what is it, about a foot per nanosecond. (...) And so in this experiment, you fire your photon a long enough distance that the scientific apparatus has enough nanoseconds to actually act. And it has gone through a slit, one of the two slits, or it has not, or it's done whatever weird quantum things it's gonna do. But the point is, after it's past the hurdle of the slits, then and only then is the decision effectively made to determine if the photon has just gone through a specific slit in the very recent past. That decision's made just as the photon's about to hit the sensor, and if we decide retroactively that we're going to pin it down to having gone through one slit or the other, the photon will then hit the sensor in a way that contributes to that shotgun pattern that suggests it was acting like a particle all along. Whereas if we don't measure it, it's going to hit the sensor in a way that contributes to a banding pattern.
But because it's just about to hit the sensor when the decision's made, it either has to know what's going to be chosen in advance so that it can position itself to land at the correct spot and contribute to the correct pattern, or it has to rewrite its own history since passing through the slit in order to position itself correctly.
And this incredibly creepy thing has in fact been done in the lab, and at least in theory if we were looking at some very, very distant object, like a galaxy that's ten billion light years away, had that experiment been done with those photons, they will presumably behave in the same manner, which means that on a photon that left its home star nine billion years ago, we can in effect impose nine billion years of history on that thing by observing or not observing something that it does in its last instant of travel from here to our detector.
Don Hoffman: That's right. John Wheeler actually proposed this kind of experiment, a cosmological delayed choice double slit experiment. So if you have a quasar that's ten billion light years away and you have a black hole between us and that quasar, or a big galaxy, according to Einstein's theory of general relativity, it bends space and you can get, if circumstances are right, a gravitational lens. From Earth, it could look like there are actually two quasars when in fact there's only one quasar, but it's an optical illusion created by bending of space.
The ability to act nine billion years back
So you can now ask, for each photon that comes to me, do I want to decide whether it came on the left side of the gravitational lens billions of years ago, or did it come from the right side, or do I want to just measure the interference pattern. So suppose I make the choice now, and I decide to measure which side it came on, and I find out that it went on the left side. That means I can say for the last ten billion years, that photon has been on a path that started from the quasar and went around the left side of the gravitational lens. But if instead today I had chosen to not measure that and just measure the interference pattern, then it would not be true that for the last nine billion years or ten billion years that photon had gone around the left side. So the choice I make today determines the ten billion year history of that photon.
Rob Reid: I should point out that gravitational lensing has been overwhelmingly and demonstrably proven. There's lots of images that astrophysicists have captures, and you can just google them. We see double images of galaxies. There's even some triple and quintuple images that are out there.
Is consciousness a foundational part of reality?
So now, I think you've argued pretty vociferously that we are seeing a user interface that looks like space time. Space time is not objective reality. What is objective reality?
Don Hoffman: The right answer is I don't know, but as a scientist, I'm going to try to propose a theory and try to make it precise and make it testable and see where we go. So the theory that I'm proposing is that consciousness is fundamental. And by that I mean, conscious experience is like experiencing the taste of chocolate, the pain of a headache, the feel of velvet, the smell of a rose, the sound of a trumpet. All these things as conscious experiences, that these are not late comers in the universe. They are the foundational entities.