This experiment pushes the delayed-choice quantum eraser further by testing common "what if" scenarios. While exploring forums and FAQs, we encountered many questions people ask about the experiment:
- What if we block one of the paths?
- What if a fly looks at the beam instead of me?
- What if the detector is unplugged?
The underlying question is: what really counts as a detection? If someone else observes $D_2$ but I don't, do we see the same results? If a fly or a wall "detects" the photon, what happens?
The answer has two parts: first, to see interference, we must record coincidences between $D_1$ and $D_2$. Second, the idler measurement must be performed in a basis that allows the wavefunction terms to recombine (e.g., diagonal polarization). Blocking the idler or leaving $D_2$ unplugged satisfies neither condition. Furthermore, which-path information cannot be destroyed—it may change form (e.g., if a photon is absorbed by a wall), but the correlation with the signal path remains encoded in the environment.
In this section, we test two scenarios:
- Wall on path 2: A physical barrier blocks the idler photon before $D_2$.
- $D_2$ unplugged: The idler photon reaches the detector, but no signal is recorded.
We conclude with a FAQ addressing additional "what if" questions.
Data files:
Each file contains 7 columns:
- Polarizer angle $P_1$
- Polarizer angle $P_2$
- Actuator position ($\mu$m)
- Sweep number
- Counts at $D_1$ (signal path)
- Counts at $D_2$ (idler path)
- Coincidences between $D_1$ and $D_2$
At each actuator position, we recorded multiple samples to compute an average and standard deviation. We performed several sweeps to further reduce noise.
Singles at $D_1$: In both scenarios, the $D_1$ singles show only diffraction—no interference fringes. This matches all previous experiments: the signal path alone never reveals interference because which-path information is encoded in the idler.
Coincidences:
- Wall: Near-zero coincidences. The small residual count is experimental noise (stray light, dark counts).
- Unplugged: Exactly zero coincidences. No electrical signal means no coincidence logic trigger.
The key result: without coincidence counting, interference cannot be recovered.
The mathematics follows directly from Experiment 5.
The BBO produces the Bell state:
The signal photon enters a spatial superposition:
Vertical polarizer at slit $A$ transmits $\ket{\updownarrow}_1$; horizontal polarizer at slit $B$ transmits $\ket{\leftrightarrow}_1$. After filtering:
The $\ket{\text{lost}}$ term represents blocked photons; we drop it in subsequent steps.
Photons from slits $A$ and $B$ travel distances $a$ and $b$:
In both scenarios (wall or unplugged detector), the idler photon provides no usable information. However, the which-path information still exists:
- $\ket{\updownarrow}_2$ $\Rightarrow$ signal came through slit $A$
- $\ket{\leftrightarrow}_2$ $\Rightarrow$ signal came through slit $B$
The idler's polarization state is correlated with the signal's path, whether or not we measure it.
To find the intensity at $D_1$, we must account for both idler states. The idler polarizations $\ket{\updownarrow}_2$ and $\ket{\leftrightarrow}_2$ are orthogonal—they cannot combine. Each term contributes its own intensity independently:
The intensity is independent of position $y$:
The pattern at $D_1$ is the sum of two single-slit diffraction envelopes—no fringes.
The which-path information is encoded in the idler photon's polarization. Blocking or ignoring the idler does not erase this information—it merely prevents us from accessing it. Since the information exists (in the absorbed photon, in the environment, etc.), interference cannot appear. To recover interference, we would need to:
- Detect the idler photon
- Measure it in a basis that erases which-path information (e.g., diagonal)
- Post-select coincidences based on that measurement
None of these are possible when the idler is blocked or the detector is unplugged.
A: The results at $D_1$ are unchanged—you see only diffraction. To observe interference, you must record coincidences. If another person has the $D_2$ data and you have the $D_1$ data, you could later combine them to find interference in the coincidence subset. But looking at $D_1$ alone will never show fringes.
A: Any interaction that could, in principle, reveal which-path information counts as a "measurement." A fly's eye, a camera sensor, or even air molecules scattering the photon—all constitute measurements. The key is whether the which-path information becomes encoded in the environment. If it does, interference is lost at $D_1$ (unless you can somehow access and erase that information).
A: No. The wall can be 1 cm or 1 km from the BBO. As long as it blocks the idler before reaching $D_2$, the result is the same: no coincidences, no interference. This reinforces that the effect is not about timing—it's about whether coincidence data can be collected.
A: Even with an extremely long delay (using fiber optic storage, for example), the physics is unchanged. When you eventually detect the idler and compute coincidences, you will find interference if you measured in the erasing basis ($\pm 45°$), or no interference if you measured in the which-path basis ($0°/90°$). The "delayed choice" aspect works regardless of the delay length.
A: No. The $D_1$ singles never show interference—they always show diffraction. Only the coincidence subset reveals fringes, and computing coincidences requires classical communication between $D_1$ and $D_2$ (which is limited to light speed). There is no way to use this experiment for faster-than-light signaling.
A: No. The $D_1$ detection is unaffected by what happens to the idler. The signal photon's behavior at $D_1$ is fixed the moment it's detected. What changes is our ability to sort the $D_1$ data into interference and non-interference subsets—and that sorting requires the idler data, which we don't have if it's blocked.
These "what if" experiments confirm the core principle: interference appears only in coincidences, and only when which-path information is erased.
Blocking the idler or unplugging the detector doesn't change the physics—it simply removes our ability to perform the coincidence measurement that reveals interference. The which-path information still exists; we just can't access or erase it.
