The arrow of time and retrocausality

1. The puzzle in plain language

The basic situation can be stated in one paragraph. The laws of physics governing how matter and energy behave are time-symmetric. If you record a video of a single molecule moving through empty space, and you play it backwards, the backwards video still obeys the same laws of motion. There is nothing in the fundamental equations that picks out a preferred direction of time. Yet the world we experience is wildly time-asymmetric. Cups fall and break; they never spontaneously unbreak. Heat flows from hot things to cold things; it never reverses. We remember the past but not the future. The arrow of time is one of the most obvious features of experience and one of the strangest things physics has to explain — because the underlying laws don't put it there.

Most attempts to resolve this go through three steps:

• The fundamental laws are time-symmetric. (Established.)
• The macroscopic arrow of time comes from statistics, not from the laws. The second law of thermodynamics — that entropy tends to increase — emerges from counting: there are vastly more disordered arrangements than ordered ones, so a system left alone is overwhelmingly likely to drift toward disorder. (Established, mostly.)
• The reason the universe is currently moving toward greater disorder is that it started in an extraordinarily ordered state — the low-entropy state of the early Big Bang. This is sometimes called the Past Hypothesis. We don't know why the universe started this way. We just know that if it did, everything we observe about the arrow of time follows. (Less established. The Past Hypothesis is a brute fact that physics needs but cannot derive.)

This is the standard story. It is mostly satisfying. It is also incomplete — because at the quantum level, the apparent forward-only causation we take for granted in daily life appears to break down. The present, at small enough scales, is constrained by what happens later. That is the strangeness this page is really about.

2. Thermodynamic, electromagnetic, psychological — three arrows, one direction

Physicists distinguish several "arrows of time" that all happen to point the same way:

• The thermodynamic arrow: entropy increases. Heat flows from hot to cold; structures decay; mixed things tend to stay mixed.
• The radiative arrow: electromagnetic waves are emitted outward from sources after they are generated, not inward toward sources before they are absorbed. A pebble dropped in a pond creates expanding ripples, not contracting ones.
• The cosmological arrow: the universe is expanding, and presumably it began with a big bang that established a clear "before" and "after."
• The psychological arrow: we remember the past and not the future. We make decisions about the future based on the past, not vice versa.
• The causal arrow: causes precede effects, never the other way around.

All five point the same direction. Whether they are five independent arrows that happen to align, or one arrow with five different aspects, has been debated for a century. The mainstream view is that the thermodynamic arrow is fundamental and the others are downstream consequences. Memories form by entropy-increasing processes (you imprint a pattern on a recording medium by raising its entropy at the cost of the surroundings); waves expand outward because the alternative requires precisely tuned initial conditions that almost never occur; the cosmological arrow set the thermodynamic arrow by providing a low-entropy initial state.

The reduction is mostly successful. What it does not explain is the causal arrow at the quantum level — the assumption that causes always precede effects. That assumption has been getting weaker, not stronger, over the last sixty years of foundations work.

3. Wheeler's delayed-choice — the past, after the fact

The cleanest experimental window onto retrocausality is the delayed-choice experiment, discussed at greater length in the wave-particle duality page. The setup: a photon traverses a Mach-Zehnder interferometer (an experimental arrangement with two possible paths and a recombining stage at the end). After the photon has passed the first beamsplitter and is already committed to one path or the other — or, in quantum terms, is already in a superposition of both paths — the experimenter decides whether to install the second beamsplitter that will let the paths interfere. The decision can be made arbitrarily late, even, in principle, after the photon should already have been detected.

What the experiments show, repeatedly and unambiguously: if the second beamsplitter is installed, interference appears and the photon's path is undefined. If it is not installed, no interference appears and the photon has a definite path. The character of the photon's earlier journey appears to be fixed by the experimenter's later choice.

This is not magic. It is not a violation of relativity. It is, instead, a sign that the classical picture of the photon "having a definite history that was already determined before the experimenter chose" is wrong. The photon's quantum state is not a record of a definite past. It is a probability amplitude for a range of possible histories, and the history that gets realised depends on the entire experimental context — including the parts of the context that have not yet occurred when the photon first enters the interferometer.

The cleanest reading of this result, and the one favoured by Aharonov and others, is that quantum mechanics is genuinely time-symmetric at the level of fundamental states. The state of a system at any intermediate time is constrained both by its preparation in the past and by its final measurement in the future. The past does not unilaterally cause the present; past and future jointly determine present quantum states.

4. The Two-State Vector Formalism

The technical framework that makes this picture rigorous is the Two-State Vector Formalism (TSVF), developed by Yakir Aharonov, Lev Vaidman, and collaborators since the late 1960s. (See the dedicated TSVF page for a deeper walk-through.) The core idea:

In standard quantum mechanics, a system is described at each moment by a single state vector evolving forward from its preparation. In TSVF, the same system is described at each intermediate moment by two state vectors — one evolving forward from preparation, one evolving backward from a final measurement. The system's behaviour at intermediate moments is determined by both. This is not a different theory; it is the same quantum mechanics, rewritten so that time-symmetry is manifest.

The empirical payoff is that TSVF predicts phenomena that look anomalous in the forward-only picture but appear natural in the two-state picture:

• Weak values. Aharonov, Albert, and Vaidman (1988) showed that "weak" measurements made between a system's preparation and its final post-selection can yield expectation values outside the spectrum of the observable being measured. A spin-1/2 particle weakly measured at an intermediate time can yield a measured spin of 100, or -50, or any value — values that are impossible under standard "strong" measurement. Weak values are anomalous from the forward-only perspective and natural from the two-state perspective.
• Particles in three places. The Aharonov-Vaidman three-box paradox: a single particle, suitably prepared and post-selected, can be shown by weak measurement to have been with certainty in two different boxes at the same intermediate time. Anomalous forward-only; natural two-state.
• The quantum Cheshire cat. A system's properties (spin, position, charge) can be separated from the system itself during the intermediate evolution — the cat is in one place, its grin in another. Demonstrated experimentally with neutron interferometry (Denkmayr et al., 2014).

None of these are mere mathematical curiosities. They are confirmed in actual experiments. They make sense only in a picture in which past and future jointly constrain present quantum states — which is to say, in a picture in which the world is time-symmetric at the fundamental level and the asymmetry we experience is emergent.

5. Retrocausality — the live position

"Retrocausality" is the name for the explicit claim that the future can influence the past at the quantum level. It is a minority position among working physicists but it has been gaining ground for thirty years. Three arguments support it:

The Bell-theorem argument. Bell's theorem (see the Bell page) rules out local hidden-variable theories. The standard response has been to abandon locality. But there is another response: abandon the assumption that hidden variables are uncorrelated with future measurement settings. If hidden variables are allowed to depend on what will be measured later, Bell's inequality can be evaded while keeping locality. This is the retrocausal "way out" pursued by Costa de Beauregard, Cramer, Kastner, and others, often in combination with the transactional interpretation of quantum mechanics.

The delayed-choice argument. As above. The experiments demonstrate that the photon's character is fixed by the experimental context including the future. The retrocausal reading is the most natural and least mystical reading of this fact.

The time-symmetry argument. All fundamental laws are time-symmetric. The asymmetry we experience is statistical and emergent. There is no in-principle barrier to causation running backward in time at the level where the asymmetry has not yet emerged — the level of individual quantum events.

The mainstream view remains that "retrocausality" is a confusing way of describing what time-symmetric quantum mechanics already implies, and that no actual signal can be sent backward in time (which is true — macroscopic information cannot be transmitted into the past without violating thermodynamics). But the question of whether quantum correlations can be retrocausal is increasingly open. Cramer's transactional interpretation, Kastner's Possibilist Transactional Interpretation, Aharonov's TSVF, Sutherland's retrocausal hidden-variable models — all are live, all are taken seriously, all imply that the future has the same status in physics as the past, at least at the quantum level.

6. The Past Hypothesis — why the universe started ordered

If the fundamental laws are time-symmetric, the world we observe asks for an explanation: why did the universe start in such a wildly improbable low-entropy state? Boltzmann calculated that the initial conditions of the universe required a precision of about 1 part in 10¹⁰¹²³ to give us the entropy growth we observe. (This is Roger Penrose's famous number, derived from the Bekenstein-Hawking entropy of the observable universe versus the maximum possible entropy.) That is not a number any random fluctuation could plausibly produce.

This is the Past Hypothesis: the assumption, sometimes derided as cheating but invoked by every working cosmologist, that the universe began in a state of vanishingly low entropy and the second law of thermodynamics is the consequence of that initial condition rather than of fundamental physics.

The Past Hypothesis is a strange object. It is needed by physics, but physics cannot derive it. It is one of the cleanest pointers toward something fine-tuned about the initial conditions of the universe — something that the standard cosmology has no internal mechanism to explain. The competing accounts (Penrose's conformal cyclic cosmology, Carroll's reverse-direction-of-time argument, the Boltzmann brain problem and its resolutions) are all live, all unsatisfying, and all gesture at the same gap: why does time have a direction at all, in a world whose fundamental laws don't pick one out?

7. What this means for the trilogy

Three touchpoints, each of them visible in the books.

First, the readiness potential mirror in Anima. The Libet readiness potential is the EEG signal that precedes voluntary action by 300 ms; it is the most famous evidence against classical free will, used by determinists to argue that the brain has already decided before the conscious "you" is aware of deciding. But the readiness potential's symmetric partner — the EEG signal that follows a decision by 300 ms, the moment when consciousness has caught up and now consciously experiences having decided — is also real, also measurable, also part of the picture. The trilogy treats the 300-ms window as a symmetric opening, not a one-way determinism. In a time-symmetric framework, the decision is constrained from both directions; what looks like "the brain deciding before consciousness" is one half of a temporally symmetric process.

Second, Lucía Reyes's cymatic pre-event window. Lucía perceives events in a roughly 300-ms window before they happen — the cymatic patterns of the events resolve in her perception before their causal antecedents have completed. The trilogy frames this as her being able to read the back-evolving state vector that TSVF posits everyone has but that most receivers cannot decode. The "premonition" is not magic; it is access to the future-boundary-condition information that the time-symmetric formalism says is always present. Lucía's gift is reading what the formalism allows everyone, in principle, to read.

Third, the field cosmology of Limen takes the time-symmetric framework as foundational. Causation in Limen is not unidirectional; the field is read in both directions and what we call "the present" is the constructed interface between past and future. The retrocausal interpretation of quantum mechanics is not used as a mystical device; it is used as a literal description of how the field renders local experience. Past and future are equally fundamental; the felt directionality of time is the way the field appears from inside a single localised receiver.

The honest summary: the arrow of time is not a fundamental feature of physical law. It is a feature of statistics, of initial conditions, and of how receivers like us happen to access the substrate. The fundamental world is two-way. Causes can be future as well as past. Quantum experiments have been hinting at this for sixty years. The trilogy takes the hint seriously and builds the books around it.

This page is part of the Reading companion essays. For the formal framework, see the Two-State Vector Formalism; for the delayed-choice experiments, see Wheeler delayed-choice; for the measurement problem that sits underneath all this, the measurement-problem page; for the synthesis, The Evidence.