I have been happy to see Relational Quantum Mechanics (RQM) become the focus of a great deal of research in recent years (for a brief introduction to RQM see here). This post is concerned with the metaphysics of RQM, prompted by a recent paper from Andrea Oldofredi and Antonio Vassallo called “Too Many Perspectives, Not World Enough: A Metaphysical Appraisal of Relational Quantum Mechanics.” But their metaphysical critique is closely connected to a question about the epistemic implications of RQM, so I’ll begin with comments on that topic.
Does RQM undermine scientific knowledge?
The key worry here is that a merely relational characterization of measurement outcomes undermines the notion that QM properly delivers scientific knowledge. We typically view science as offering objective facts. And while attempting to define an appropriate notion of scientific “objectivity” is fraught, we assume, at a minimum, that intersubjective agreement about observations and experimental results is needed to underpin scientific facts. If RQM implies this is problematic, because every measurement outcome is relative to the observer performing it, then we seem to have a problem.
Whether RQM can deliver intersubjective agreement in principle is thus controversial.[1] One positive response begins by acknowledging that an interaction has a physical impact on the systems involved, and in the case of large persisting systems we call “observers”, this impact is discoverable by other observers via subsequent interactions. One paper discussing this idea is Adlam and Rovelli’s “Information is Physical: Cross-Perspective Links in Relational Quantum Mechanics” (2023). First, the authors offer an interpretation of the physical nature of interactions in the way I just described: these are events that alter the systems involved, and in principle this should be accessible to third parties via appropriate further interactions.[2] They describe this addition to the assumptions of RQM as follows:
Cross-perspective links (CPL): In a scenario where some observer Alice measures a variable V of a system S, then provided that Alice does not undergo any interactions that destroy the information about V stored in Alice’s physical variables, if Bob subsequently measures the physical variable representing Alice’s information about the variable V, then Bob’s measurement result will match Alice’s measurement result. (Adlam and Rovelli, 2003, p.7)
With this addition, we can be satisfied that we will not have disagreements among observers, and intersubjective agreement is safeguarded.
But some may think scientific knowledge demands a stronger foundation than this. If an additional observer accesses the information in the way envisioned, one might think the problem is just pushed back a step—the information just gleaned is still relative to the new observer. And since this acquired information is only available to a further observer if they engage in yet another appropriate interaction, we have a chain of relational facts, never reaching a point where we can just declare the information is not relative to a single perspective. Given this concern, Adlam and Rovelli go on to offer a stronger interpretation of what it means to assert cross-perspective links:
But with the addition of CPL, it no longer seems possible to insist that everything is relational—or at least, it is no longer necessary to do so—because this postulate implies that the information stored in Alice’s physical variables about the variable V of the system S is accessible in principle to any observer who measures her in the right basis; so at least at an emergent level, this information about V is an observer-independent fact. This suggests that the set of “quantum events” should be regarded as absolute, observer-independent features of reality in RQM. (Adlam and Rovelli, 2023, p. 11)
This assertion extends beyond pragmatic intersubjective agreement, reaching a stronger, and more metaphysical conclusion. If interaction events are not merely relational but rather absolute (whether or not they are accessed by any particular observer), then RQM describes an observer-independent world. If this idea is accepted, one can affirm that the world described by RQM is one that concerns objective scientific facts. But I note that this move to posit absolute facts appears to add something that doesn’t necessarily to follow from RQM as originally presented. In a subsequent paper, Rovelli endorses the weaker relational reading of CPL, where it does not require that facts are treated as objective. Rather, it offers a way to explain how observers’ relative facts achieve consistency. He goes on to argue this is all we need:
To understand the world, ourselves, our science, and the others, we do not need such ultimate perspective. We can simply be happy of always being, talking and describing the state of the world from within a perspective. (Rovelli, 2024, p. 5)
The metaphysical dimension of the problem
Rovelli’s view is critiqued from a metaphysical perspective in the paper by Oldofredi and Vassallo. The authors begin with a conception of the kind of metaphysics they argue is needed to support scientific realism:
Traditional metaphysical realism demands the existence of a single, unified world— an objective reality that science investigates. This demand is a structural requirement that underwrites the coherence of the realist package as a whole. Scientific realism is not merely the claim that there are facts, nor even that such facts are mind-independent, but that they collectively belong to one and the same world…In the absence of such a world, truth threatens to become perspective-relative, and scientific objectivity risks being replaced by mere intersubjective coordination. While intersubjective agreement is undoubtedly an important feature of scientific practice, it is ordinarily understood as epistemic access to a shared reality, not as a substitute for it. (Oldofredi and Vassallo, 2026, p.11)
They conclude that the ontology implied by RQM fails to meet the standard. In particular, the authors’ concern is as follows: “In our view it [RQM] does not provide a sufficiently developed account of how this ontology supports a coherent, mind-independent physical world (p. 3).”
Before evaluating this argument, I want to recall the basic ontological framework offered in Rovelli’s presentations of RQM (and assumed in the Adlam and Rovelli paper just discussed): it is one that limits reality to the interaction events between physical systems.[3] While the existence of systems is implied (they are, after all, the entities interacting), Rovelli takes a strongly anti-realist stance on the wave functions/quantum states used to model these systems in QM. These (relational) states are instrumentally useful from the point of view of an observer, encoding the results of past interactions with a system and allowing one to calculate probabilistic predictions about it. But the states are not to be taken as representations of systems out there in the world.
Given this ontological framework, along with the original RQM interpretation of interaction events as relational and not absolute, Oldofredi and Vassallo say that “reality is fragmented into a multitude of relative facts, each corresponding to an interaction between specific systems (p. 12).” They further note that “What is missing here is some “ontological glue” that renders the many relative facts postulated by RQM a unique, all-encompassing entity (p.13).” They go on to try to make this point more precise by referencing David Lewis’ ideas about the “world-building relations” between fundamental ontological elements that are needed to form a coherent reality (see p. 13). RQM falls short by not providing these.
Oldofredi and Vassallo then spend a part of the paper examining the CPL proposal of Adlam and Rovelli discussed above. This seems to add the needed objectivity to the interaction events. At the same time, given RQM’s view of quantum states, there is a puzzle regarding how to assess the impact of these absolute events between interactions. Oldofredi and Vassallo ask: does the addition of CPL’s change RQM’s anti-realist assumptions about quantum states or not (p. 19?) They argue that systems would need to be described by a temporally extended “ontic” state in order to denote the ability of third-parties to access the physical characteristics associated with the past “absolute” interaction event. But such a notion conflicts with the basic RQM framework, where the state takes on merely relational bookkeeping role, and we do not have any further resources to describe systems between interactions.[4] So, there is a possible inconsistency here, and it seems to derive from Adlam and Rovelli’s move to affirm the presence of objective facts based on absolute events without altering RQM’s other assumptions.[5] It doesn’t seem we can have such objective facts without a different interpretation of systems and states. And if we fall back on the relational reading of CPL, we are back to a reality that is arguably too metaphysically fragmented.
Providing the glue
With this background, I’ll now describe a different path toward a satisfying metaphysics for RQM. It will still differ from the traditional realism described by Oldofredi and Vassallo. But, to invoke Wilfrid Sellars,[6] it will show how things hang together. The picture begins by reviving a role for quantum states: they do represent real systems between interactions, albeit incompletely. Then, it posits the mutual dependence of these states on each other as the basis for a pluralistic, but metaphysically connected, world-network.
To begin: famously, quantum systems do not have definite values for state-dependent observables prior to measurement. A number of thinkers pondering QM have conceived of these observables as representing dispositions or propensities of systems to manifest definite values in interactions. Mauro Dorato (2016) suggests this picture in the context of RQM:
In other words, such systems S have intrinsic dispositions to correlate with other systems/observers O, which manifest themselves as the possession of definite properties q relative to those Os. (Dorato, 2016, 239; emphasis original)
As Dorato explains, such manifestations only occur mutually, among two or more interacting systems. With this idea in place, the quantum state can be taken to represent the dispositions of a system to probabilistically manifest in interactions with a specified reference “observer” system. Likewise, the Schrödinger equation may be taken to represent the propagation of the system and its dispositions between such interactions. We thus take a realist stance toward non-interacting systems, characterized by their dispositional profile, in addition to the interaction events. A difference, of course, between the two is non-interacting systems cannot be localized in three-dimensional space: only the measurement-like interaction events are localized.[7] But the systems and their dispositional profiles are real.
Since we are still working in the context of RQM, it remains the case that the interaction events only feature manifestations of participating systems. Likewise, the quantum states modeled by QM only represent dispositions of the system relative to particular interaction partners. However, we can be less austere and view this latter relational limitation as a feature of quantum mechanics, not of reality. Systems can interact with any system they encounter, it is just that QM only models these for one potential interaction context at a time.
With this picture in place, let’s return to the question of the supposed fragmentation of reality. What is the glue? The glue is the mutual dependence of the systems, whose potential to interact embodies reciprocity.[8] Between interactions, any system, via its dispositional profile, has the potential to interact with whatever partners it might encounter. This web of potential weaves all systems into one reality. The impression of fragmentation is the result of limiting our view to the interaction events themselves. It is understandable that our attention is drawn to interaction events as the source of facts (definite values for observables at spacetime locations). But it is nonetheless a mistake to restrict our ontology to the events.
A coherent world
An important role for mutual dependence is defended in a paper from Mauro Dorato and Matteo Morganti, titled “What Ontology for Relational Quantum Mechanics?” In this paper, the authors examine some proposals about the metaphysical character of the cosmos and evaluate their compatibility with RQM. They look at the views called “priority monism” and “ontic structural realism” and conclude they do not provide a good fit. Very briefly (please check out the paper for details), priority monism is the view that the parts of the cosmos depend metaphysically on the whole. RQM, however, describes symmetric relations among systems, with no dependence on any notion of the whole. In fact, I would point out that RQM is inconsistent with the notion that the cosmos is an entity at all: systems can only be described from the perspective of a further system, and there is no perspective outside the cosmos. Ontic structural realism, which posits a metaphysics of relations without relata, is upon reflection also a poor fit for RQM, which emphasizes relationality, but does not suggest an ontology built solely on relations.
The right fit, according to the authors, is an idea called metaphysical coherentism:[9]
Metaphysical coherentism is essentially a rejection of (i) The traditional foundationalist conceptions of reality in terms of levels, ‘hierarchical’ dependence-chains and fundamentality; (ii) The Hume–Lewis idea that fundamental entities must be absolutely independent. (Dorato and Morganti, 2022, p.13)
The first point means all entities that exist are on a par with one another. The second, rejecting what is sometimes called “Hume’s Dictum,” allows for the fact that, given RQM, entities in fact symmetrically depend for their existence on one another. The result is a pluralistic picture, but not a fragmented one. The mutual dependence of entities implies that the shape of reality is that of a connected network, or woven web.[10]
Returning to Oldofredi and Vassallo’s claim that scientific realism requires a unified world: the metaphysically coherent world is unified in that its pluralism of perspectives is interwoven by ties of mutual dependence. On the other hand, we must give up our intuition that the world abides completely untouched by our participation in it, but that seems to be an unavoidable lesson of taking quantum mechanics seriously.
An Even Tighter Weave
In my work, I flesh out this ontology of systems and their interactions by marrying it to the account of causal structure introduced by Salmon (1984). Then we have a combined ontology and theory of causation that can be used to underpin scientific explanations across the natural sciences (see this past post or my recent paper; my prior paper on molecular structure that invokes RQM is here). I’ll make just a couple of points pertinent to the present topic.
As in the discussion above, the ontology has two features: propagating entities (causal processes) characterized by dispositions to interact, and the interaction events themselves. I then add an account of composition whereby a pattern of repeated interaction gives rise to a larger scale process (but with no implication of asymmetric “levels of reality”). One interesting feature: all processes are defined in part by reference to something external to them. All processes propagate dispositions toward interactions with other processes. In addition, composite processes form when a group of processes interact with each other more frequently than with other processes outside the group. The existence of every entity depends on something else: nothing is complete on its own. Also, a tight link between the entities and the interactions is emphasized: the events are mutual manifestations, and these take the form of changes in the dispositional profiles of participating processes.
What RQM brings to the picture, of course, is the co-existence of multiple perspectives on the causal web. All perspectives are views from the inside of reality, of course—we don’t even need RQM to see this. But RQM means that direct participation in interactions delivers a view that is distinct from that of non-participating systems. This is not intuitive, but it is consistent with a coherent and I think attractive metaphysics.
References
Adlam, E., Rovelli, C. (2023). Information is physical: cross-perspective links in relational quantum mechanics. Philosophy of Physics. 1, 4. https://doi.org/10.31389/pop.8
Dorato, M. (2016). Rovelli’s relational quantum mechanics, anti-monism, and quantum becoming. In: A. Marmodoro, A. and Yates, D. (eds.) The Metaphysics of Relations, pp. 235-262. Oxford, Oxford University Press. https://doi.org/10.1093/acprof:oso/9780198735878.003.0014
Dorato, M., Morganti, M. (2022). What ontology for relational quantum mechanics? Foundations of Physics (2022) 52:66. https://doi.org/10.1007/s10701-022-00581-x
Esser, S. (2024). Relational quantum mechanics, causal composition, and molecular structure. Foundations of Chemistry. 26, 429–446. https://doi.org/10.1007/s10698-024-09513-1
Esser, S. (2026). Turning Salmon’s causal process theory into process ontology. Synthese 208, 30. https://doi.org/10.1007/s11229-026-05691-x
Frauchiger, D., Renner, R. (2018) Quantum theory cannot consistently describe the use of itself. Nature. Communications. 9, 3711. https://doi.org/10.1038/s41467-018-05739-8
Martin, C. (2008). The Mind in Nature. Oxford, Oxford University Press.
Morganti, M. (2019). From ontic structural realism to metaphysical coherentism. European Journal for Philosophy of Science 9, 7. https://doi.org/10.1007/s13194-018-0233-5
Oldofredi, A. (2021). The bundle theory approach to relational quantum mechanics. Foundations of Physics (2021) 51:18. https://doi.org/10.1007/s10701-021-00407-2
Oldofredi, A., Vassallo, A. (2026). Too many perspectives, not world enough: A metaphysical appraisal of relational quantum qechanics. Philsci-archive preprint. https://philsci-archive.pitt.edu/id/eprint/30014/1/RQM_Oldofredi_Vassallo_090626.pdf
Rovelli, C. (2024). Can Alice do science and have friends, in a relational quantum world? Solipsism and Relational Quantum Mechanics. ArXiv preprint. https://arxiv.org/pdf/2410.20012
Rovelli, C. (2026). Relational quantum mechanics. The Stanford Encyclopedia of Philosophy (Fall 2026 Edition), Edward N. Zalta & Uri Nodelman (eds.). https://plato.stanford.edu/archives/fall2026/entries/qm-relational/
Salmon, W. C. (1984). Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press.
Sellars, W. (1963). Science, Perception, and Reality. New York: The Humanities Press.
Terris, B. (2026). The informational observer in relational quantum mechanics. Foundations of Physics 56:39. https://doi.org/10.1007/s10701-026-00939-5
Williams, N.E. (2010). Puzzling Powers: the problem of fit. In: The Metaphysics of Powers: Their Grounding and Their Manifestations Anna Marmodoro (Ed.), pp. 84-105. Routledge Press, New York.
[1] Given the fact that we all effectively share an environment, the phenomenon of decoherence means that intersubjective agreement is not a practical concern. It is rather a concern in principle, typically analyzed using idealized thought experiments, such as extended versions of the so-called Wigner’s friend scenario. A paper that helped spur this debate is Frauchiger and Renner (2018).
[2] Here I want to flag a complication, which is that this information recovery process assumes the physical systems involved (observers) have some degree of size and complexity. This was recently explored by Terris 2026 (“The Informational Observer in Relational Quantum Mechanics”). Part of this complexity must involve persistence across time, which touches on another tricky issue for interpreting RQM’s ontology, which in its traditional form has little to say about the nature of systems between momentary interactions. This issue comes up again below.
[3] See the SEP article (Rovelli, 2026), which describes the ontology as “sparse”.
[4] The Adlam and Rovelli paper has further discussion of ways to interpret states and systems, but the discussion keeps the assumption that the events are the only real entities (Adlam and Rovelli, 2023, pp. 11-14).
[5] The authors note that if Adlam and Rovelli do allow for a particular observer’s post-measurement situation to be described using an ontic state, this still doesn’t seem to extend to states describing other observers who have yet to interact with her (p. 21).
[6] “The aim of philosophy, abstractly formulated, is to understand how things in the broadest possible sense of the term hang together in the broadest possible sense of the term (Sellars, 1963, 1).”
[7] Interestingly, Oldofredi has also proposed a realist ontology for RQM (Oldofredi, 2021). Here, systems are objects defined as “mereological bundles” that include observer-dependent properties. He doesn’t use the framework of dispositions to explicate observer dependence, but rather the idea of “metaphysically indeterminate” properties. While this is intriguing, I prefer the dispositional framework, which allows for the fusing of ontology with an attractive theory of causation. Also, while I cannot explore the proposal in depth here, I am concerned that it describes these objects as existing in spacetime (the bundle includes spatio-temporal properties): “A remarkable consequence of this proposal is that, under this reading, RQM is committed to a realism towards properties; thus, from the perspective of mereological bundle theory, one can properly speak about material objects in motion in spacetime, whose attributes vary in relation to different observers (Oldofredi, 2021, p. 18).” This seems difficult to square with quantum theory, where systems arguably cannot be viewed as residing in spacetime (even in some indeterminate way). Generally, the state of a multi-particle system can only be modeled in a high-dimensional configuration space.
[8] Philosopher C.B. Martin used the term “reciprocal disposition partners (Martin 2005)”.
[9] Discussed in a previous paper by Morganti (2018).
[10] The authors mention that the argument for the symmetric dependence could be furthered by attributing reciprocal dispositions to the systems, but they don’t emphasize the dispositional ontology in this paper (see footnote 17, p. 10). I’ll also note that this non-Humean principle complements a non-Humean approach to causation centered on dispositions, as I mention below. Finally, I want to shout out a paper from Neil. E. Williams from 2010 called “Puzzling powers: the problem of fit.” Williams explores the metaphysical implications of the fact that reciprocal dispositions (a.k.a. powers) need to harmoniously work together. I think approaching the issue via quantum mechanics helps clarify the issues (My old blog has a post discussing this paper).