The Delayed-Choice Quantum Eraser
The delayed-choice quantum eraser is one of the most mind-bending experiments in quantum mechanics. It demonstrates that whether we observe wave-like interference or particle-like which-path information depends not on when we make a measurement, but on what information is available to us. This experiment challenges our classical notions of causality and demonstrates the profound role of quantum entanglement.
The Setup
Entangled Photon Pairs
A nonlinear crystal converts pump photons into entangled pairs: a signal photon and an idler photon. The signal photon enters a double-slit interferometer and is detected at D0. The idler photon carries which-path information - it's entangled with the path the signal photon took. But crucially, we have a choice about what to measure on the idler photon.
Two Measurement Choices
Which-Path Mode (D3/D4): We can measure which slit the signal photon went through by detecting the idler at D3 or D4. This gives us definite path information, and when we look at signal detections correlated with D3 or D4 clicks, we see no interference pattern - just two blobs.
Eraser Mode (D1/D2): Alternatively, we can send the idler through a beam splitter that "erases" the which-path information by making the paths indistinguishable. When we look at signal detections correlated with D1 or D2 clicks, we see complementary interference fringes - wave behavior is restored!
The Delayed Choice
Retrocausal Appearance
Here's what makes this experiment so striking: the idler photon can be measured long after the signal photon has already been detected at D0. The "choice" of whether to measure which-path or erase that information happens after the signal has hit the screen. Yet this choice determines whether we see interference or not in the post-selected data.
Crucially, this is not retrocausality. Looking at all D0 detections (without post-selection), you always see a featureless blob - no interference. The pattern only emerges when you sort the D0 detections based on which idler detector clicked. The information revealed by the idler measurement determines what pattern we find in the already-recorded signal data.
What This Reveals
Complementarity is About Information
The key insight is that wave-particle duality isn't about physical disturbance or temporal ordering of measurements. It's about information. If which-path information exists anywhere in the universe (even in the idler photon that hasn't been detected yet), interference is destroyed in the joint probabilities. If that information is erased (made inaccessible even in principle), interference returns.
Entanglement and Nonlocality
This experiment showcases quantum entanglement's nonlocal correlations. The signal and idler are in a shared quantum state. Measuring the idler doesn't "send a signal" to the signal photon (that would violate causality). Instead, the measurement reveals which subset of the entangled state you're looking at, and different subsets exhibit different statistical properties.
Things to Try
No Selection: Run the experiment without enabling either measurement mode. All signal detections at D0 pile up in a featureless distribution - no interference visible. This is what you'd see if you didn't use the idler information.
Which-Path Measurement: Enable which-path mode and send photon pairs. Watch the total D0 pattern (still no fringes). But look at the post-selected patterns: events coincident with D3 form one blob, coincident with D4 form another. The which-path information destroys interference.
Quantum Eraser: Switch to eraser mode. Now D0 detections coincident with D1 show interference fringes, and those coincident with D2 show complementary anti-fringes. The patterns are phase-shifted by π. Together they wash out to no net interference, but separately each subset exhibits perfect wave behavior.
Phase Dependence: In eraser mode, adjust the phase slider. Watch how the D1-correlated and D2-correlated patterns shift. The interference visibility depends on the optical phase difference, just like a standard interferometer, even though we're post-selecting on entangled pairs.
Delayed Choice Thought Experiment: Imagine the signal photon travels 1 light-year to D0, while the idler is delayed locally. You could choose whether to measure which-path (D3/D4) or eraser (D1/D2) long after the signal was detected. The pattern in the D0 data would retroactively depend on your choice - but only when you later correlate the data. No faster-than-light signaling, but a profound demonstration that "what happened" depends on "what information is available."
Historical Context
This experiment was first proposed by Scully and Drühl in 1982 as a thought experiment to clarify complementarity, and later realized experimentally by Kim et al. in 2000 using spontaneous parametric down-conversion. It's often called the "delayed-choice quantum eraser" following Wheeler's earlier delayed-choice experiments with single photons.
Why This Matters
The delayed-choice eraser forces us to abandon certain classical intuitions:
- The past is not fixed in the sense of classical definiteness - quantum superposition persists until measured.
- Complementarity is enforced by information-theoretic constraints, not mechanical disturbances.
- Entanglement allows correlations that seem acausal but never violate relativity.
- Post-selection can reveal patterns that are invisible in the raw data.
- Quantum mechanics is fundamentally about correlations in joint measurements, not individual events.
This experiment remains one of the clearest demonstrations that quantum mechanics isn't just a theory about our ignorance of classical reality - it describes a genuinely different kind of reality where information, entanglement, and measurement play foundational roles.