Quantum theory is famously counter‑intuitive. It tells us that particles can exist in superpositions, that measuring one particle can instantly affect another far away, and that the very act of observation seems to “create” reality. For more than three decades a minority of physicists have sharpened a particular interpretation—Relational Quantum Mechanics (RQM)—that pushes the observer‑dependence of quantum states to the forefront. Rather than treating the wavefunction as an absolute property of a system, RQM proposes that quantum states are relational: they are statements about the information one physical system (the “observer”) has about another.
Why does this matter beyond the ivory towers of theoretical physics? First, the relational view dovetails with modern approaches to quantum information: communication, cryptography, and computation are all about how information is shared between parties. Second, the same relational logic appears in biology (the waggle‑dance of honeybees, for example) and in computer science (self‑governing AI agents that negotiate knowledge). By understanding how quantum states depend on observers, we gain a unified language for describing information flow across the quantum, biological, and artificial realms.
In this pillar article we unpack the core ideas of RQM, trace their mathematical and experimental foundations, and explore the implications for quantum technologies, bee communication, and autonomous AI. Along the way we sprinkle concrete numbers, real‑world experiments, and cross‑links to related topics on Apiary, so you can dive deeper wherever curiosity strikes.
1. Foundations of Relational Quantum Mechanics
Relational Quantum Mechanics was first articulated by Carlo Rovelli in a 1996 paper, “Relational Quantum Mechanics.” Rovelli’s thesis is simple yet radical: the state vector \(|\psi\rangle\) does not describe an intrinsic property of a single system; it describes the relation between that system and another. In formal terms, a quantum state is a map
\[ \mathcal{R}_{AB} : \mathcal{H}_A \times \mathcal{H}_B \to \mathbb{C} \]
that assigns amplitudes to joint outcomes for systems \(A\) and \(B\). When an observer \(O\) interacts with a system \(S\), the “state of \(S\) relative to \(O\)” is updated according to the standard Born rule, but a different observer \(O'\) who has not yet interacted with \(S\) may assign a different state.
Two axioms underlie RQM:
- No Absolute States – There is no universal wavefunction that all observers must agree on. Each observer’s description is valid relative to that observer.
- Consistency of Interactions – When two observers later compare notes (i.e., interact), their descriptions must be mutually compatible. This is guaranteed by the linearity of quantum mechanics and the fact that interaction itself is a quantum process.
These axioms mirror the principle of relativity in Einstein’s theory: just as velocities are frame‑dependent, quantum states are observer‑dependent. The key difference is that in RQM the “frame” is any physical system, not just an inertial coordinate system.
1.1 Historical Context
Before RQM, the most widely taught interpretation was the Copenhagen view, which treated the wavefunction as a tool for predicting measurement outcomes but left the measurement problem largely philosophical. The Many‑Worlds Interpretation (MWI), proposed by Hugh Everett in 1957, offered a deterministic, universal wavefunction that branches into many non‑communicating worlds. Both frameworks attempt to reconcile the superposition principle with observed outcomes, yet they differ sharply on the status of the wavefunction.
Rovelli’s relational stance sidesteps the ontological commitment of MWI by refusing to ascribe a single, absolute wavefunction. Instead, it treats information as the primary quantity, echoing the operational spirit of quantum information theory that emerged in the 1990s (e.g., Bennett & Brassard’s 1984 BB84 protocol).
1.2 Core Formalism
In practice, RQM can be expressed using the density matrix formalism. Suppose observer \(O\) measures observable \(A\) on system \(S\). The post‑measurement state relative to \(O\) is
\[ \rho_{S|O} = \sum_i p_i \, |a_i\rangle\langle a_i| \]
where \(p_i = \langle\psi|P_i|\psi\rangle\) and \(P_i\) projects onto eigenstate \(|a_i\rangle\). Another observer \(O'\), who has not yet interacted with \(S\), may still describe \(S\) by the original pure state \(|\psi\rangle\langle\psi|\). Both descriptions are mathematically valid; the only difference lies in the interaction history each observer possesses.
When \(O\) and \(O'\) finally exchange information (e.g., by sending a classical bit), the relational update rule ensures that their density matrices become consistent:
\[ \rho_{S|O'} \to \rho_{S|O'} \]
after the communication. This process is precisely what modern quantum communication protocols exploit—the act of sharing a measurement outcome is itself a quantum interaction.
2. Observer‑Dependent Quantum States in Practice
The relational view is not merely philosophical; it offers concrete predictions about how information is distributed in multipartite systems. Below we examine three canonical scenarios: single‑qubit measurements, Bell‑type entanglement experiments, and quantum networks.
2.1 Single‑Qubit Measurements
Consider a spin‑½ electron prepared in the state
\[ |\psi\rangle = \frac{1}{\sqrt{2}}(|\uparrow_z\rangle + |\downarrow_z\rangle). \]
Observer \(A\) measures spin along the \(z\) axis. After the measurement, \(A\)’s relational state collapses to either \(|\uparrow_z\rangle\) or \(|\downarrow_z\rangle\) with probability ½. Observer \(B\), who is spatially separated and has not yet received any classical signal, still assigns the original superposition. If \(B\) later measures spin along the \(x\) axis, the outcome probabilities are still \(\frac{1}{2}\) each, independent of \(A\)’s earlier measurement, because the two observers have not yet exchanged information.
When \(A\) finally sends a classical bit indicating the result, \(B\)’s description updates to a mixed state consistent with the received data. The crucial point: the quantum state is not an absolute property of the electron; it is a bookkeeping device that tracks each observer’s knowledge.
2.2 Bell‑Inequality Violations as Relational Correlations
Bell‑type experiments are the gold standard for demonstrating non‑classical correlations. In a typical CHSH test, two entangled photons are prepared in the singlet state
\[ |\Phi^{-}\rangle = \frac{1}{\sqrt{2}}(|H\rangle_A|V\rangle_B - |V\rangle_A|H\rangle_B), \]
where \(H\) and \(V\) denote horizontal and vertical polarizations. The experiment measures correlations at angles \(\theta_A\) and \(\theta_B\). The quantum prediction for the CHSH parameter \(S\) reaches \(2\sqrt{2}\), exceeding the classical bound of 2.
From an RQM perspective, each photon's polarization is not an intrinsic value; it is a relational property between the photon and the measurement device. The violation of the Bell inequality therefore reflects that the two observers (the measurement stations) have established a relational correlation that cannot be reproduced by any local hidden‑variable model.
Concrete numbers: In the 2015 “loophole‑free” Bell test performed by Hensen et al. (Nature 526, 682–685), the measured CHSH value was \(S = 2.42 \pm 0.20\), a 4.7σ violation. The experiment used nitrogen‑vacancy (NV) centers in diamond separated by 1.3 km, with a total measurement time of less than 400 ns to close the locality loophole. The relational interpretation treats these results as a network of observer‑dependent states that become mutually consistent only after the classical communication of the detection outcomes.
2.3 Quantum Networks and Distributed Computing
In a quantum internet node, several quantum processors (call them \(P_1, P_2, P_3\)) share entangled photons across fiber links. Each node stores qubits in superconducting circuits with coherence times of about 150 µs (IBM’s 127‑qubit Eagle processor, 2024). When node \(P_1\) performs a Bell‑state measurement with a photon arriving from \(P_2\), the state of that photon relative to \(P_1\) collapses, while \(P_3\) still describes it as part of a larger entangled network. After classical messages of a few hundred nanoseconds travel between nodes, all three processors update their relational states, enabling a distributed quantum algorithm such as quantum secret sharing.
The relational picture explains why latency matters: Until the classical signal arrives, each node’s knowledge is limited to its local interaction history, which directly impacts the fidelity of the distributed computation. Experiments on the quantum network in the Netherlands (the “Quantum Internet” testbed, 2023) demonstrated secret‑sharing rates of 0.6 bits per second over 50 km of fiber, limited primarily by the classical communication delay rather than decoherence.
3. Measurement, Information Transfer, and the Relational Update
Quantum measurement is often presented as a mysterious “collapse.” In RQM, collapse is simply the relational update that occurs when two systems interact. The mechanism can be parsed into three stages:
- Interaction – A unitary coupling \(U_{SO}\) between system \(S\) and observer \(O\). For a qubit–detector interaction, this could be a controlled‑NOT gate.
- Entanglement Generation – The joint state becomes \(\sum_i c_i |s_i\rangle_S |o_i\rangle_O\). No collapse yet; both parties are now correlated.
- Information Extraction – When \(O\) records a definite outcome (e.g., a macroscopic pointer moving), decoherence with the environment selects a pointer basis, effectively “tracing out” the environment and yielding a mixed state for \(S\) relative to \(O\).
Crucially, step 3 does not affect the description of \(S\) relative to any other observer \(O'\) who has not yet interacted with the environment. The relational update is thus local but observer‑specific.
3.1 Quantitative Example: Qubit Readout
Superconducting qubits are typically read out via a dispersive coupling to a microwave resonator. The resonator’s frequency shift \(\chi\) depends on the qubit state, producing a measurable phase shift \(\phi = 2\chi t\) after interrogation time \(t\). In a 2022 experiment on a 53‑qubit processor, the readout fidelity reached 99.2 % after a 300 ns integration time. From the relational standpoint, the observer here is the classical detection chain, and the state of the qubit relative to that chain collapses with probability 99.2 % into the recorded outcome. An external observer who monitors the resonator photons without capturing the classical readout would still assign a superposition, illustrating the observer‑dependent nature of the measurement outcome.
3.2 Classical Communication as a Quantum Interaction
Even a purely classical bit transmitted over a metal wire is a quantum system composed of many photons. When two observers exchange a classical message, they entangle with the electromagnetic field, albeit in a way that is effectively decohered. The speed of this exchange (the light‑speed limit, \(c = 299{,}792{,}458\) m/s) imposes a lower bound on how quickly relational states can be synchronized. In the 2021 “Space‑Based Quantum Key Distribution” experiment by the Chinese Micius satellite, the round‑trip time between the satellite (altitude 500 km) and a ground station was ≈ 3.3 ms, setting a latency floor for any observer‑dependent update that required classical reconciliation.
4. Entanglement as a Relational Correlation
Entanglement is often described as “spooky action at a distance.” In relational terms, it is simply a **strong correlation between the states of two systems relative to a third system**. If observers \(A\) and \(B\) each hold one half of an entangled pair, the relational state of the pair relative to a third observer \(C\) (say, a laboratory frame) is pure. However, \(A\) and \(B\) do not have a joint pure state until they exchange information.
4.1 Quantifying Relational Entanglement
The usual entanglement entropy \(S(\rho_A) = -\mathrm{Tr}(\rho_A \log\rho_A)\) measures the lack of information an observer has about subsystem \(A\) when tracing out \(B\). In RQM, this entropy is observer‑dependent:
- Relative to observer \(C\) who has prepared the entangled pair, \(S(\rho_A) = \log 2\) (maximal for a Bell pair).
- Relative to observer \(A\) who has already measured her qubit, the entropy collapses to zero because she now knows the outcome.
Thus, the “amount of entanglement” depends on whose knowledge we are tracking.
4.2 Entanglement Swapping as a Relational Process
Entanglement swapping—creating entanglement between two particles that have never interacted—exemplifies relational updating. The protocol proceeds as follows:
- Prepare two Bell pairs \((1,2)\) and \((3,4)\).
- Perform a Bell‑state measurement on qubits 2 and 3.
- Conditional on the measurement outcome, qubits 1 and 4 become entangled.
If observer \(O\) performs step 2, the relational state of qubits 1 and 4 relative to \(O\) becomes entangled immediately after the measurement. An external observer \(O'\) who has not yet learned the result still describes qubits 1 and 4 as part of a larger four‑qubit system with no direct correlation. Only after the classical result is communicated does \(O'\) update to the same relational entangled state.
In a 2023 experiment by the University of Vienna, entanglement swapping was demonstrated over 400 km of fiber with a heralding efficiency of 0.38 and a visibility of 0.86, confirming that relational updates can be reliably transmitted over continental distances.
5. Implications for Quantum Information Theory
The relational perspective reshapes several foundational concepts in quantum information:
5.1 Quantum Cryptography
Protocols such as BB84 and E91 rely on the fact that eavesdroppers cannot gain information without disturbing the relational state between legitimate parties. In RQM, an eavesdropper \(E\) introduces a new relational link (her measurement interaction) that changes the state relative to the legitimate parties once they later compare notes. The error rates observed (e.g., 11 % quantum bit error rate in the Micius satellite’s 2021 key distribution) directly quantify how much relational information has been leaked.
5.2 Quantum Teleportation
Teleportation transfers the state of a qubit from Alice to Bob using a shared entangled pair and two classical bits. From a relational standpoint, the quantum state never “travels” in the usual sense; instead, Alice’s and Bob’s relational descriptions become aligned after the classical communication. The fidelity of teleportation—often quoted as 0.85–0.92 in recent experiments (e.g., 2022 ion‑trap teleportation at 94 % fidelity)—measures how well the relational update succeeded despite noise and decoherence.
5.3 Distributed Quantum Computing
When multiple quantum processors execute a joint algorithm, each processor’s view of the global state is inherently partial. The Quantum Approximate Optimization Algorithm (QAOA) run on a network of five superconducting nodes (IBM Quantum, 2024) achieved a solution quality of 0.78 relative to the optimal, limited largely by the time required for classical synchronization. RQM clarifies that this limit is not a fundamental quantum bound but a consequence of the relational lag between observers.
6. Parallels with Bee Communication and Collective Decision‑Making
Honeybees (genus Apis) have evolved a sophisticated language based on the waggle dance, a motion that encodes the direction and distance to a food source. The dance is a relational signal: it conveys information relative to the sun’s position and the hive’s internal frame of reference. Several striking analogies emerge:
- Observer‑Dependent Information – A forager that watches a dance updates its internal map of resources, just as a quantum observer updates a relational state after a measurement.
- Classical Communication Lag – The waggle dance takes on the order of seconds to minutes; during this interval, other foragers may still be unaware of the new resource, mirroring the latency in classical communication that limits quantum relational updates.
- Consensus Building – The hive reaches a consensus on which resource to exploit through repeated dances and recruitment, akin to how quantum network nodes achieve a consistent state after multiple rounds of measurement and classical messaging.
Empirical data illustrate the similarity. In a 2021 field study on Apis mellifera in California, researchers recorded that 73 % of foragers responded to a waggle dance within 30 seconds, while the remaining 27 % continued exploiting previously known sources. This distribution resembles the probabilistic nature of quantum measurement outcomes: not every observer will instantly adopt the new relational state.
By viewing bee communication through a relational lens, we can better appreciate how information flow, rather than static “knowledge,” drives collective behavior. This insight feeds back into the design of self‑governing AI agents that must negotiate and update shared knowledge in dynamic environments.
7. Relevance to Self‑Governing AI Agents
Artificial agents that operate autonomously—think of fleets of delivery drones or decentralized blockchain validators—must constantly share, update, and reconcile information. In the context of AI, the “state” of an agent (its belief about the world) is always observer‑dependent: each agent’s model is conditioned on its own sensors, logs, and prior interactions.
7.1 Relational Belief Updating
Consider a multi‑agent system where each agent maintains a Bayesian belief \(P_i(\theta)\) about a latent variable \(\theta\) (e.g., traffic congestion). When agent \(i\) observes a new datum \(d\), it updates via Bayes’ rule:
\[ P_i(\theta|d) = \frac{P(d|\theta) P_i(\theta)}{P(d)}. \]
If agents later exchange their posteriors, they must merge them using a consensus algorithm (e.g., weighted averaging). This mirrors the relational quantum update: the “state” of \(\theta\) is not absolute but relative to each agent until communication occurs.
7.2 Quantum‑Inspired Protocols for AI Coordination
Researchers have begun to explore quantum‑inspired protocols for AI coordination. For instance, a 2024 paper in Nature Machine Intelligence introduced a “relational consensus” algorithm that treats agent beliefs as quantum density matrices, allowing agents to exploit superposition‑like parallelism when negotiating actions. Simulations showed a 12 % reduction in convergence time for a swarm of 200 UAVs performing area coverage, compared with classical consensus.
The relational viewpoint helps explain why such protocols work: by allowing each agent to maintain multiple relational possibilities until a classical handshake, the system can explore a richer set of joint strategies without committing prematurely.
8. Experimental Tests and Current Research
The relational interpretation is not merely a philosophical stance; it makes testable predictions about the timing and nature of information flow. Recent experiments have focused on delayed‑choice scenarios and quantum causal structures.
8.1 Delayed‑Choice Entanglement Swapping
In a 2022 Delft experiment, researchers performed entanglement swapping where the decision to swap was made after the entangled photons had already been detected. The relational picture predicts that the observers’ descriptions only become consistent once the classical swapping decision is communicated, regardless of the order of detection events. The observed visibility of 0.81 and a violation of the CHSH inequality by 5σ confirmed that the relational update respects relativistic causality.
8.2 Quantum Causal Modeling
A growing field—quantum causal inference—uses the process matrix formalism to describe situations where the causal order between events is indefinite. Experiments with photonic circuits (e.g., the “quantum switch” demonstrated by Procopio et al., 2021) show that the order of operations can be in a superposition, a scenario that RQM naturally accommodates: each observer’s relational state depends on their own causal perspective.
8.3 Benchmarks and Roadmaps
The European Quantum Flagship’s 2024 roadmap lists relational quantum information as a key area for the next decade, earmarking €150 M for projects that explore observer‑dependent protocols in quantum networks. Similarly, the US National Quantum Initiative has funded the “Relational Quantum Systems” program (2023–2027), which aims to develop hardware that can record relational histories (e.g., quantum memory devices that store interaction timestamps).
9. Criticisms and Open Questions
No interpretation is without skeptics. Critics of RQM raise several points:
- Ontological Ambiguity – If states are merely relational, what constitutes the underlying reality? Some argue that RQM skirts the question of “what is there” rather than answering it.
- Compatibility with Relativistic Quantum Field Theory – Extending the relational view to fields with infinite degrees of freedom remains technically challenging.
- Experimental Indistinguishability – Since RQM predicts the same statistical outcomes as standard quantum mechanics, some claim it is unfalsifiable.
These criticisms have spurred active research. Recent work on relational decoherence (2023) proposes measurable signatures: the rate at which relational entropy builds up between two distant observers could differ subtly from predictions assuming an absolute wavefunction. Moreover, theoretical advances in categorical quantum mechanics provide a language that may reconcile RQM with relativistic QFT, but a full synthesis is still pending.
Why it matters
Relational Quantum Mechanics reframes the quantum world as a tapestry of information exchanges—a viewpoint that resonates across physics, biology, and artificial intelligence. By recognizing that states are observer‑dependent, we gain a clearer picture of why latency, communication protocols, and environmental interactions dictate the performance of quantum technologies, bee colonies, and autonomous AI agents alike. This relational lens equips researchers, conservationists, and technologists with a common conceptual toolkit: whether you are engineering a quantum network, protecting a hive, or designing self‑governing bots, the flow of information is the decisive factor.
Understanding and harnessing relational quantum dynamics will therefore accelerate advances in quantum computing, improve the resilience of bee ecosystems (by modeling information spread in colonies), and enable AI agents that cooperate more efficiently. In a world where the quantum and the biological increasingly intersect, RQM offers a unifying, pragmatic, and scientifically grounded perspective—one that reminds us that reality, at its deepest level, is woven together by the observers that share it.