I’ve read and reread the passages on quantum erasers and delayed choice in Brian Greene’s wonderful book The Fabric of the Cosmos. I think I’m finally getting a feel for what it means. And I also think the existence of quantum erasers says something amazing about, believe it or not, free will.

Stay with me here, because the going will be a bit treacherous. I’ve written before about the double slit experiment. I’ve also written a little bit about entanglement. When you combine those two crazy ideas together, you get, well, craziness. But a very particular kind of craziness, the kind that makes the whole mind-blowing subject, I think, a little more human.

OK start with the basic idea. You can’t know both the particle-nature of a photon (which slit it went through) and the wave-nature of that same photon (interference pattern). Anything you do that reveals “which path” information automatically makes the interference pattern disappear.

What’s interesting about the quantum eraser is that you can set up things so that which path information exists, but is still hidden. One way Greene describes this is by polarizing the photon either clockwise or counter-clockwise. If you do this, even if you don’t collect the information, the interference pattern is destroyed. This is key: the mere potential of collecting which path information destroys the interference pattern. This is important for later.

The quantum eraser takes that information back out. The polarization is erased by removing the tag. Now the photons can interfere once more. Remember, as long as the potential for which path information exists, the interference pattern disappears. But (and again, this is key) once that potential is erased, the interference pattern comes back. This is important, too.

OK, now let’s think about another way to make which path information potentially available. This is a method that is even less invasive than the polarized tag. This time, we use something called a down-converter. Where at first one photon existed, we now create two, each with half the frequency of the original photon, and each entangled with the other. Call one photon the signal photon, the other the idler photon.

Suppose we send the signal photons toward a double slit. Will they form an interference pattern? The answer is, it depends on the idler photons. As long as these idler photons exist and give us the potential to discover which path information, the answer is no, no interference pattern will form. If, on the other hand, these idler photons are recombined in such a way that which path information becomes impossible to retrieve, then the interference pattern, in a certain specific sense, reappears.

Here’s the incredible part, though. The choice about what to do with the idler photons, whether to measure them directly or to recombine them to erase which path information, can happen after (even well after) the signal photons have formed their pattern. Note that, due to some clever bookkeeping by nature, you can never actually “see” the interference pattern until you obtain some extra information about the recombined idler photons. It’s all part of that amazing way that nature has of covering her tracks. It’s pretty clever, and if you want to know more about it I urge you to read Greene’s book, this Wikipedia article, or even the original paper (look at how graphs 3 and 4 each show an interference pattern, but when you combine them into graph 5, as always happens in the real experiment, the pattern disappears. Nature is shrewd!)

But astounding as all that is, it’s still not the most amazing piece of this. Here’s the most amazing piece. We already saw that we can reveal honest-to-goodness, no-mistaking-them interference patterns if we erase the which path information before the photons hit the screen. Remember the important points bolded above. What we can’t do is reveal an honest-to-goodness, no-mistaking-it interference pattern if the which path information is erased after the signal photons hit the screen. There’s a before-after dichotomy. The universe behaves differently regarding a choice that’s been made and a choice that might be made. And while it would be mind-blowing either way, I find this result (the future can’t affect events that have already happened) just as mind-blowing, maybe even more mind-blowing, than the alternative. Why? Because as long as the idler photons exist, as long as the potential for which-path information is there, the signal photons will not form an interference pattern. It’s as if they’re saying, “You might find out, you might not, so we’re not taking any chances.” But if we might find out and we might not, then we have the choice. We aren’t constrained by the past, and all future paths are open to us.

The refusal of those stubborn signal photons to form an interference pattern, no matter how we promise that no, really, we’d never go and detect your idler friends, demonstrates to me an amazing fact. We have free will. The future is open. We can make our own choices.

This is a big deal, because there are philosophies that say the future is just as determined as the past, that time is an illusion, that all future events are controlled by conditions right now. But the delayed choice quantum eraser denies that idea. It says that I control my own destiny, and the universe, recognizing my autonomy, must behave accordingly. And that’s pretty cool. Damned be determinacy! Long live the idler photons, for they have set us free!