Retrocausality is the idea that a future event can influence something in the past, flipping the usual direction of cause and effect. It sounds like science fiction, but a growing number of physicists argue it deserves serious consideration as a feature of how the quantum world works. The concept shows up in several well-known quantum experiments, in at least one formal mathematical framework within standard quantum theory, and in proposals meant to resolve some of the deepest puzzles about entanglement and nonlocality. Retrocausality does not mean sending messages backward in time or predicting next week’s lottery numbers. The version physicists discuss is far more constrained and far stranger.
Why the Laws of Physics Allow It in Principle
The fundamental equations governing physics are, with few exceptions, symmetric in time. Run the equations of motion backward and they still work. This is true in classical mechanics, in electromagnetism, and in the core equations of quantum theory. A system described by these laws does not “know” which direction the clock is running. The technical way to say it is that these laws are invariant under time reversal: swap the direction of time and apply an appropriate transformation, and the equations hold just as well.
This symmetry is not some obscure mathematical curiosity. It sits at the foundation of modern physics and raises a genuine question: if the laws themselves treat past and future equally, why should causation be a one-way street? The everyday experience of time flowing forward, of causes always preceding their effects, is something we impose on top of the equations. It comes from thermodynamics and the special low-entropy state of the early universe, not from the microscopic laws that govern particles and fields. The standard explanation for why we experience a direction to time at all relies on a cosmological assumption, often called the Past Hypothesis, which holds that the universe began in an extraordinarily low-entropy state at the Big Bang. Given that starting point, entropy almost always increases, giving us a thermodynamic arrow of time.
But at the level of individual quantum events, before thermodynamic averaging kicks in, the time symmetry of the underlying laws leaves the door open for influences that run backward. Retrocausality takes that open door and asks what happens if you walk through it.
Delayed-Choice Experiments and Quantum Erasers
The experiments that most vividly illustrate why physicists entertain retrocausality are the delayed-choice experiments first proposed by John Archibald Wheeler. The setup is a variation on the classic double-slit experiment. A photon can behave as a wave, producing an interference pattern, or as a particle, showing up at one slit or the other. Wheeler asked: what if you delay the choice of which measurement to perform until after the photon has already passed through the slits? Does the photon “decide” in advance whether to be a wave or a particle?
The answer, confirmed in laboratories and even scaled up to satellite-to-ground distances spanning thousands of kilometers, is that it does not matter when you choose. The photon’s behavior matches whatever measurement you eventually make, even if the measurement setup was configured while the photon was still in flight. A satellite-ground experiment demonstrated exactly this, observing complementary wave-like or particle-like behavior at a ground station depending on the measurement apparatus selected while photons were still propagating from the satellite.
1PubMed Central. Extending Wheeler’s delayed-choice experiment to spaceA related setup is the quantum eraser. Here, “which-slit” information is recorded for each photon, destroying the interference pattern as expected. But then a second measurement “erases” that information, and the interference pattern reappears in the correlated data. In one thermal-light version of the experiment, the interference reappeared even after the first photon had already been detected and annihilated, with the explanation framed as a nonlocal interference phenomenon in which a random photon pair interferes with itself at a distance.
2PubMed Central. Delayed-choice quantum eraser with thermal lightThese experiments do not prove retrocausality is the correct interpretation. Orthodox quantum mechanics can explain each result without invoking backward causation, typically by emphasizing that no usable signal travels backward and that the patterns only emerge when you compare records from both detectors. But the results sit uncomfortably with any picture where events in the past are fixed and independent of future measurements. They are one of the reasons some physicists find retrocausal models appealing rather than merely provocative.
The Two-State Vector Formalism
Retrocausality is not only an interpretive gloss applied to surprising experiments. It also appears in a rigorous mathematical framework within quantum theory itself. The two-state vector formalism, developed originally by Yakir Aharonov and collaborators, describes a quantum system at any given moment using two quantum states: one evolving forward in time from a past measurement and one evolving backward in time from a future measurement.
3arXiv. The Two-State Vector FormalismIn standard quantum mechanics, you assign a state based on how the system was prepared and then predict what future measurements will find. The two-state vector formalism says that a complete description requires also knowing the outcome of a later measurement, with a second state propagating backward from that outcome. This is not a different theory from quantum mechanics. It is a time-symmetrized reformulation of the same theory, producing the same predictions.
4Journal of Physics Communications. Stories in the two-state vector formalismThe framework gained additional attention through the concept of weak measurements, where a quantum variable is measured so gently that the system is barely disturbed. Weak measurements can yield “weak values” that sometimes lie outside the range of eigenvalues you would normally expect. Aharonov and others have argued that these results are naturally explained by the retrocausal structure of the two-state formalism, where the future measurement outcome is part of what determines the system’s properties at intermediate times.
5AIP Conference Proceedings. The Retrocausal Nature of Quantum Measurement Revealed by Partial and Weak MeasurementsFor many physicists, the two-state vector formalism is the strongest argument that retrocausality is at least mathematically natural within quantum theory, even if you are free to interpret the formalism without taking the backward-evolving state literally.
Resolving Quantum Nonlocality Without Spooky Action
One of the most compelling motivations for retrocausality comes from the problem of quantum nonlocality. When two particles are entangled, measuring one instantly fixes the outcome for the other, no matter how far apart they are. Bell’s theorem tells us that no theory based on local hidden variables can reproduce all the predictions of quantum mechanics, which seems to force a choice: either accept instantaneous influences across space (what Einstein called “spooky action at a distance”) or give up the idea that particles carry pre-existing hidden properties.
Retrocausal models offer a third option. If hidden variables are allowed to depend on future measurement settings, not just past preparation, then the correlations between entangled particles can be explained locally. The measurement choice you make here, now, is what “set” the hidden variable back when the particles were created. For the small cost of accepting backward causation, this approach provides a hidden-variables interpretation of entanglement that is genuinely local, with no need for faster-than-light influences.
6Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics. Retrocausal models for EPRThis is not merely a philosophical reframing. Locality is a cornerstone of both special relativity and quantum field theory. Any interpretation that preserves it while still accounting for Bell-inequality violations has a real theoretical advantage. The price, of course, is accepting that future choices can constrain past states, which many physicists find just as uncomfortable as nonlocality.
How Retrocausality Differs from Superdeterminism
Retrocausality is sometimes confused with superdeterminism, and the two ideas do share a surface similarity: both reject the assumption, built into standard Bell-inequality experiments, that the experimenter’s measurement choices are independent of the system being measured. In a superdeterministic universe, everything, including your choice of which button to press, was fixed at the Big Bang by initial conditions. The measurement settings and the particle properties are correlated not because the future reaches back, but because a common cause in the distant past predetermined both.
Retrocausality is different. It preserves the experimenter’s freedom to choose measurement settings. The correlation between the setting and the earlier hidden variable arises because the choice genuinely influences the past, not because both were locked in from the start. Proponents have argued that this loophole is missed precisely because it gets confused with superdeterminism on one side and with spooky action at a distance on the other, even though it is distinct from both and has clear advantages over each.
7arXiv. A Live Alternative to Quantum SpooksThe distinction matters because superdeterminism, taken to its logical end, undermines the whole enterprise of experimental science. If your experimental choices are not free, then the results of any experiment could be a coincidence baked in by initial conditions, and no empirical test can be trusted to mean what it seems to mean. Retrocausality sidesteps that problem by keeping the experimenter’s choices genuinely free, at the cost of letting those choices have backward-reaching consequences.
Roots in Classical Physics and Absorber Theory
The idea that future states might influence present ones did not originate in quantum mechanics. In the 1940s, John Wheeler and Richard Feynman developed their absorber theory of electrodynamics, which invoked both retarded waves (traveling forward in time, as you would expect) and advanced waves (traveling backward in time) to describe how charged particles emit and absorb radiation. The advanced waves were needed to solve a longstanding problem: an accelerating electron radiates energy and experiences a damping force, but the standard classical equations did not cleanly account for where that damping came from without advanced solutions.
8arXiv. Wheeler-Feynman Absorbers on the Light HorizonWheeler and Feynman showed that if you include both forward-traveling and backward-traveling electromagnetic waves, and if the universe contains enough absorbing matter (a “complete absorber”), the advanced waves cancel out in a way that reproduces the standard results. The backward-in-time waves are not observable in the final answer, but they are essential to making the math work. This elegant trick made retrocausality respectable in at least a limited sense decades before the quantum debates brought it back into the spotlight.
Indefinite Causal Order
A more recent line of research pushes the question of causality even further. In ordinary physics, events have a definite causal order: A happens before B, or B before A. Quantum mechanics allows a stranger possibility called indefinite causal order, in which two operations can be applied in a quantum superposition of both possible orderings. The device that realizes this is called a quantum switch.
In a quantum switch, a control qubit determines whether operation A is performed before B or B before A, but the control qubit can itself be in superposition, meaning neither ordering is definite until a measurement is made. Experimental teams have verified that such processes genuinely lack a definite causal order. One experiment demonstrated this by measuring a so-called causal witness, confirming by almost seven standard deviations that the implemented process did not have a definite causal order.
9PubMed Central. Experimental verification of an indefinite causal orderMore recent theoretical work has shown that the indefinite causal order in a quantum switch can even be certified in a device-independent way, meaning you do not need to trust the internal workings of the apparatus, provided you assume that faster-than-light influences are impossible.
10Quantum. Possibilistic and maximal indefinite causal order in the quantum switchIndefinite causal order is not identical to retrocausality. It is a broader and arguably even more radical idea: that the question “which came first?” may simply not have an answer in certain quantum situations. But it shares conceptual DNA with retrocausality in challenging the assumption that causal order is always fixed and observer-independent.
Self-Consistency and the Paradox Problem
The most immediate objection to retrocausality is the grandfather paradox: if the future can influence the past, can you create contradictions? Could you set up an experiment where a future measurement prevents the conditions needed for that very measurement to occur?
The answer from the physics literature is a firm no, at least within the frameworks where retrocausality is discussed. A key principle is that only self-consistent causal loops can occur. A specific laboratory proposal examined closed causal cycles within quantum mechanics and found agreement with the principle that inconsistent loops have zero probability: the only loops that survive are those where everything fits together without contradiction.
11Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics. Retrocausality and quantum mechanicsThis self-consistency constraint is crucial because it means retrocausality, as physicists discuss it, cannot be used to create paradoxes, send messages to the past, or violate causality in any observable way. The backward-in-time influences are real in the sense that they appear in the mathematical description and help explain otherwise puzzling correlations, but they are tightly constrained so that they never produce a logical contradiction or allow superluminal communication.
One way to think about this: imagine the universe as a completed crossword puzzle. Each square must be consistent with its row and its column. Retrocausality says that the “column” constraints (coming from the future) are just as real as the “row” constraints (coming from the past). But the puzzle, by the time it is filled in, always makes sense. You never get a contradiction because only self-consistent solutions exist.
The Block Universe and Post-Determination
Retrocausality fits naturally into what physicists call the block universe, the idea that past, present, and future all exist equally and that our sense of time “flowing” is a feature of our perspective, not of the universe itself. In a block universe, asking whether the cause comes before or after the effect is like asking whether the left side of a painting comes before or after the right side. Both are just parts of the whole.
A recent proposal adds a twist to this picture. In a block universe subject to global consistency constraints, each new quantum observation eliminates some of the possible solutions that were consistent with all previous observations. For an observer experiencing the flow of time, this looks like either superdeterminism or retrocausality, but it does not manifest in a way that would allow anyone to violate causality. The block universe appears “post-determined,” meaning it is not fully fixed from the beginning but is progressively narrowed down by each quantum measurement.
12arXiv. The post-determined block universeThis framing softens the strangeness of retrocausality. The future is not “reaching back” to change the past in a way you could detect. Rather, the global pattern of events, spanning past and future, must be self-consistent, and what looks like backward influence is really the universe enforcing that consistency.
Why You Cannot Use Retrocausality to See the Future
Given how often retrocausality is confused with precognition, it is worth being explicit about what it cannot do. No retrocausal model in physics allows usable information to travel backward in time. The backward-evolving states, advanced waves, and future-dependent hidden variables are all hidden from direct observation. Their effects show up only in statistical correlations that can be seen after the fact, once you compare records from different detectors or different branches of an experiment. You cannot use them to place a bet, dodge a disaster, or send a message to your younger self.
The psychological claim of precognition, the idea that people can sense future events before they happen, has been tested experimentally and has not held up. A well-known set of experiments by Daryl Bem in 2011 reported small effects suggesting precognition, but large-scale replication attempts have found no evidence supporting those claims. Whatever retrocausality turns out to mean for the foundations of physics, it does not validate the folk notion that some people can see the future.
Why the Arrow of Time Still Points Forward
If the microscopic laws are time-symmetric and retrocausality is at least mathematically legitimate, why does the world stubbornly look like causes precede effects? The standard answer involves the thermodynamic arrow of time. The early universe had extremely low entropy, a special initial condition often called the Past Hypothesis. Given that starting point, the overwhelming majority of possible histories show entropy increasing over time, giving us the familiar direction of cause and effect, memory, aging, and all the rest.
13PubMed Central. The Decoherent Arrow of Time and the Entanglement Past HypothesisThe arrow of time is not written into the fundamental laws. It is an emergent feature of our universe’s boundary conditions. Retrocausal effects, if they exist, operate beneath the thermodynamic surface, at the quantum level where the time symmetry of the laws is still manifest. They are too constrained by self-consistency to bubble up into anything you could observe as “the future causing the past” in everyday life. The arrow of time remains intact as a statistical phenomenon even if the microscopic reality is time-symmetric.
Information-theoretic approaches to the arrow of time reinforce this picture. When you treat information-processing agents, like human experimenters, as physical systems subject to the same thermodynamic laws, the psychological experience of time flowing forward and memories pointing backward emerges naturally from the entropy gradient, not from any fundamental asymmetry in the laws themselves.
14arXiv. Information dynamics and the arrow of timeWhere the Debate Stands
Retrocausality is not mainstream consensus in physics. It is a minority position, but a respectable one, championed by serious researchers and published in leading journals. Its chief appeal is parsimony in certain contexts: it can preserve locality in quantum mechanics, provide a natural explanation for delayed-choice experiments, and fit cleanly with the time symmetry that already exists in the fundamental laws. Its chief cost is conceptual: most people, including most physicists, find it deeply counterintuitive that the future could constrain the past.
No experiment has yet been devised that would definitively prove retrocausality over other interpretations of quantum mechanics. The different interpretations, Copenhagen, many-worlds, Bohmian mechanics, and retrocausal models, all make the same predictions for every experiment performed so far. The disagreement is about what those predictions mean, not about what will happen in the lab. That may change if someone designs a test that distinguishes retrocausal hidden-variable models from their competitors, and proposals along those lines exist, but none has yet produced a decisive result.
What makes retrocausality worth taking seriously is not that it has been proven but that it solves real problems in the foundations of physics without introducing anything exotic like extra dimensions, many worlds, or instantaneous long-range influences. It simply takes the time symmetry that is already there in the equations and follows it to its logical conclusion. Whether the universe actually works that way remains genuinely open.