Manning 2015 — Delayed Choice with a Single Atom

In plain language

John Wheeler asked a strange question in 1978. Take a quantum experiment where a particle can take one of two paths. Whether the particle behaves like a wave (showing interference between the paths) or like a particle (taking one definite path) depends on what measuring equipment you have in place. Wheeler asked: what if you decide which measurement to make after the particle has already entered the experiment? Has the particle already "chosen" to behave one way or the other? If so, can your later decision change what it already did?

For decades this was a thought experiment because the timing was technically impossible. By 2015 it wasn't. Manning, Truscott, and colleagues at the Australian National University ran Wheeler's experiment with a single helium atom, making the wave-or-particle decision after the atom had passed the beamsplitter. The result was unambiguous and matches what quantum mechanics predicts: the atom's behaviour is fixed by the entire experimental configuration, including the part that hadn't happened yet when the atom entered.

This is not magic and it is not faster-than-light signaling. It is the experimental confirmation that the picture of "a definite history that happens, then gets observed" does not work at the quantum level. The atom did not "decide" earlier and the experimenter did not "change the past." Instead, there is no fact of the matter about the atom's path until the entire measurement chain — including the late choice — is complete. The quantum state is, in a precise sense, constrained by both its preparation and its eventual measurement.

The trilogy's symmetric architecture — where past and future jointly determine present quantum states, where Lucía Reyes's pre-event cymatic window and the Libet readiness potential are two halves of the same time-symmetric structure — rests on results exactly like Manning's. The atom-level experiment is the cleanest demonstration available that time-symmetry is real physics rather than philosophical speculation.

The rest of this page walks through the experimental setup, what was actually measured, and the philosophical implications.

The core idea of Wheeler's delayed choice

Wheeler's proposal considers a single quantum (originally a photon) sent through a Mach–Zehnder interferometer, where the experimenter decides after it has entered whether the interferometer is "open" (which-path information, particle-like) or "closed" (recombined paths, interference, wave-like). The key conceptual point is that the late choice of measurement setting appears to retroactively decide whether the quantum "was a wave" or "was a particle" in the past, challenging naive realist pictures of its trajectory.

For the full setup and the photon implementation by Jacques and colleagues (2007), see the Wheeler delayed-choice explainer. What Manning and colleagues add is the most direct possible answer to a natural follow-up question: is this strange behavior peculiar to light, or does it generalize to particles with rest mass?

What Manning and colleagues actually did

Manning and collaborators implemented Wheeler's scheme using a single metastable helium atom passing through a matter-wave Mach–Zehnder interferometer formed by coherent beam-splitting and mirror pulses. The setup:

• An ultracold (sub-nanokelvin) source of single atoms, released from an optical dipole trap and falling under gravity onto a delay-line detector that records individual He* atoms with full 3D spatial and timing information.
• The two interferometer "arms" are realized by coherent momentum-space splitting and recombination of the atomic wavefunction using Bragg or Raman-type light pulses, which act as matter-wave beam splitters and mirrors.
• By appropriately timing these pulses, they either close the interferometer and recombine the paths (interference possible) or leave it open so that the detection outcome reveals which path the atom took (no interference).

The principle is identical to the photon experiment. The implementation, with massive particles instead of light, is the new and conceptually striking part.

The delayed choice and the randomness

The "delayed choice" in this experiment is implemented by triggering a random number generator only after the atom has passed the first beam splitter (the initial splitting pulse). That random bit then determines whether the final pulse configuration will close the interferometer (enabling interference) or leave it open (providing which-path information), ensuring the atom has no causal access to the future choice under ordinary relativistic assumptions.

Because the atoms move far more slowly than light, the experimenters have an extended time window — compared to photonic experiments — to make and implement the random choice between the two measurement settings. This slow-massive-particle regime makes the atomic realization conceptually clean and technically flexible in comparison with earlier large-scale optical tests. The delayed-choice loophole has more breathing room when the carrier of the quantum is not racing at the speed of light.

Results: interference vs which-path

When the interferometer is closed, the detected atoms show high-visibility interference fringes as a function of the relative phase between the two arms, displaying unambiguously wave-like behavior of the single atom. When the interferometer is left open, the phase-dependent interference pattern disappears, and detections correlate with one arm or the other, corresponding to particle-like which-path information.

Crucially, which of these two behaviors appears is decided by the delayed, random choice that occurs after the atom has already entered the interferometer — in accord with Wheeler's original thought experiment. The observed statistics match the standard quantum prediction that the measurement context completely determines whether interference emerges, with no need to posit any retrocausal influence or pre-existing "wave" or "particle" character.

Conceptual upshot and Bohr's view

The authors emphasize that their data support Bohr's complementarity view: it is simply not meaningful to attribute "wave" or "particle" behavior to the atom independently of the measurement arrangement. In other words, asking "what was the atom really doing between source and detector?" in classical terms is a category error; only the full experimental context and resulting outcomes have well-defined physical meaning.

Their realization also demonstrates Wheeler's delayed-choice scheme for massive particles, extending such tests beyond photons and highlighting that wave–particle duality and context-dependent behavior are not restricted to light. This is encouraging for further experiments involving entanglement, Bell tests, and macroscopic quantum phenomena with massive systems, where similar delayed-choice and contextuality arguments can be probed.

Why this matters for the trilogy

The Manning result is the cleanest experimental form of the question Limen asks about time. If the Jacques photon experiment leaves any room for the objection "but photons are not massive — perhaps the strangeness is specific to light," Manning closes that escape hatch. Massive atoms behave the same way. The delayed measurement choice determines whether the atom was a wave or a particle. The pattern of behavior is a feature of quantum systems generally, not a quirk of electromagnetic radiation.

Read together with the photon-version explainer and the two-state vector formalism, Manning's result anchors the trilogy's central time-symmetric claim: that the present is constructed at the interface between forward-propagating and backward-propagating quantum constraints. The trilogy's symmetric 300-millisecond gaps — Libet's readiness potential on the past side, Lucía Reyes's cymatic pre-event window on the future side — are not fictional flourishes. They are exactly the kind of phenomenon that contemporary quantum mechanics, taken at face value across both photons and massive particles, has been showing us how to think about for forty years.

For the original 2015 paper, see Nature Physics 11, 539–542. For the photon implementation that started the empirical program, see the Jacques 2007 explainer. For the formal language that organizes these results, see the two-state vector formalism explainer. For the synthesis, see What the Evidence Shows So Far.