There has been a trend in recent years to solidify around the claim that quantum theory forbids universal facts and instead implies “observer relative” facts. I recently was asked to review a paper that obediently and uncritically accepted this claim and discussed various implications of it, including such constructs as “zombies” and “mindless hulks”. Below is my review. It points out the serious problems with this recent fashion of so-called “Quantum Relativism” and why the “theorems” on which it is based are not at all the solid ground that they are widely assumed to be.
Open Review of a manuscript about “Quantum Relativism”
Since the author’s entire project seems to be be prompted by ‘no-go theorems for absolute facts,’ it’s important to include at least some mention of the distinct weaknesses of such theorems. In short, all these theorems are based on the unquestioned assumption that QM ‘really’ allows only unitary physical evolution–i.e., that quantum mechanics must be Unitary-Only Quantum Mechanics (UOQM). As I’m using the term, UOQM encompasses both ‘textbook QM’ and Everettian approaches. UOQM does not necessarily deny a “projection postulate’ (PP), but if PP is invoked (as in the ‘textbook theory’), it’s assumed that there is no quantified physics that describes it. In other words, under UOQM, the ‘shifty split’ is in play and one simply invokes the PP as a mathematical recipe for consistency with empirical results (i.e. the fact that one gets measurement outcomes with probabilities predicted by the Born rule in a well-defined measurement basis, even if the latter is not accounted for under UOQM).
Under the prevailing UOQM convention, an “objective collapse theory,” in which unitarity is physically broken in a quantified manner, is considered a category exemplified only by ad hoc-engineered models, such as GRW, that deviate from ‘real quantum theory’. But importantly, even though perhaps surprisingly, this is not actually the case: arguably, ‘real QM’ already contains implicit physical non-unitarity and is thus already a collapse theory! (Although of course that sounds crazy to most of us raised on ‘textbook QM’.) Indeed, the non-unitarity inherent in the standard theory is generally not recognized, and that lack of recognition is the key blind spot in the UOQM orthodoxy. So, what exactly am I talking about? Here is some specific non-unitary physics that UOQM neglects:
- The Feynman propagator of standard quantum field theory is complex, which makes the associated action complex. This is discussed in connection with decoherence based on a non-unitary component of the action in Breuer and Petruccione 2000, pp. 40-41: https://www.researchgate.net/publication/235426843_The_Theory_of_Open_Quantum_Systems
- Decay rates from Fermi’s Golden Rule, based on a decaying exponential, imply that a non-unitary process is in play. Indeed, the concept of a decay rate, a number of absorptions/detections per unit time, offers a prediction of an outcome that is missing in the conventional UOQM interpretation. This prediction is routinely confirmed in the lab, even as UOQM fails to identify the outcome as the empirically corroborated detection warranting the appropriate collapsed state description–such as |’detector fired’>—and instead (mis)represents the state evolution as unitary (thus leading to the arguably erroneous prediction of a “Schrodinger’s experimenter” such as Wigner’s Friend ).
Further details on this point—that non-unitarity is ‘already there’ in quantum theory (even if not reflected in the conventional interpretation of UOQM) can be found in this reference, Section 3.3: https://arxiv.org/pdf/2605.17196
So, to be clear, what I’m calling UOQM is the conventional interpretation of quantum theory that assumes—despite the above quantified theoretical features such as non-unitarity in the action already in standard QM—that the only physically quantifiable form of state evolution or transformation is unitary. It’s important to be aware that the claim QM = UOQM is simply a dogma, and worse—-based on the above points, a dogma not consistent with the actual theory itself. Thus, UOQM not only lacks any independent empirical or logical support but is de facto inconsistent with empirically corroborated data predicted by the theory (e.g. detections in accordance with Fermi’s Golden Rule).
The submitted manuscript buys into this UOQM convention by uncritically defining quantum mechanics as lacking any real, physical non-unitarity. Given the above points, hopefully it is understood now why the present reviewer is pushing back on the project right out of the starting gate, since the ‘race’ is being run on a ‘track’ that never should have been laid down in the first place. For starters, it is this (arguably incorrect) unitary-only assumption about quantum theory that leads to the measurement problem: i.e., “measurement,’ in the sense of a process precipitating an outcome, cannot be consistently defined under that assumption. The notion of “quantum relativism” then only arises due to such ill-definedness of the concept of an outcome-yielding measurement, and this raises the question of whether “quantum relativism” merits treatment as a well-formulated notion in the first place. Again, from standard QM and the relevant perturbation we get Fermi’s Golden Rule and a non-unitary component, a decaying exponential, that yields a decay rate. That rate predicts time-dependent outcomes (detections) subject to a relativistically covariant description. Such detections can be considered determinate/invariant events–Einstein’s ‘point-coincidences’, if you will, and as such, may be considered physically well-defined universal facts, even if their relativistic descriptions may vary in accordance with standard relativity theory. Since standard QM in the form of Fermi’s Golden Rule evidently provides such facts, one may reasonably question the conventional dogma that quantum theory doesn’t provide universal facts. Hopefully the above helps to make clear what may be wrong with the assumed premise of QM=UOQM and its seeming implication of “quantum relativism.”
The currently popular fashion of assuming that QM precludes universal or invariant facts based on the habitual yet uncritical identification of quantum theory with UOQM (despite the above non-unitarity implicit in standard applications of quantum theory) is a key target of the criticism in Kastner (2024). Yet the submission seems to not only miss this point, but to misrepresent it. (I’ll elaborate further on this below.)
In any case, at the outset, it needs to be noted that in view of the inability of UOQM to define “measurement” yielding any sort of “fact” in the first place, such “no go theorems” are hardly to be regarded as firm results binding quantum theory. A specific publication pointing out the flaws in these “theorems” should be included up front: “Unitary Interactions Do Not Yield Outcomes: Attempting to Model “Wigner’s Friend.” https://arxiv.org/abs/2105.01773, Kastner, R. (Found Phys 51, 89 (2021)). This publication addresses not only the larger issue above, but a specific experiment about which the ‘no absolute facts’ claim is made. It points out not only the lack of a criterion for measurement leading to an outcome under the unitary-only assumption, but also misleading statements about the alleged measurement process itself in the authors’ discussion of their experiment. Specifically, the authors imply that photons were in a well-defined pure state prior to the assumed measurement transition when instead they were in an improper mixed state, according to their collective prepared state. This crucially relevant material is discussed on p. 7 of the above reference (https://arxiv.org/abs/2105.01773).
The point here is that such ‘theorems’ are often based on untenable handwaving about ‘measurement’ that doesn’t actually define a measurement transition at all, and in worst cases (like the above) misrepresent the states of the relevant systems in order to portray unitary processes as yielding outcomes when there is no physical basis for that claim. One doesn’t get a legitimate ‘theorem’ out of these sorts of gesticulations, and it’s important for any study of allegations of ‘no absolute facts’ to make explicit the demonstrably tenuous foundation on which such assertions are based.
Of additional concern is the apparent misrepresentation (however unintentional) of the content of one of the cited references, Kastner 2024 (https://philosophyofphysics.lse.ac.uk/articles/10.31389/pop.158. The authors say: “In some frameworks, self-relative properties are explicitly ruled out (Rovelli 1996, 1666). This is partly related to consistency issues that ensue when agents endeavour to model themselves quantum-mechanically (Kastner 2024)…”
However, this is not what is said in Kastner (2024). Indeed, this reference is crucially relevant to the topic of the manuscript, but as a dissent from the authors’ embrace of UOQM and its ostensible implication of ‘quantum relativism’. Kastner (2024) certainly does not rule out self-relative properties, so the author’s mention of the reference starts out misleading on that proferred context. Unfortunately, the quoted comment further misrepresents the content of Kastner’s critique, which raises a counterexample to Rovelli’s “Relational Quantum Mechanics” (RQM). In the counterexample, observers are modeled as quantum systems. That is already done in the basic Wigner’s Friend setup, so this is not a new or illicit move. In the counterexample, two different observers (the “Friend” and “Wigner”) must consequentially disagree about the probabilities of outcomes on a particular system, where there is an empirical-level comparison (not ‘hidden’ and ‘private’ to each observer). To instead portray this inconsistency as arising from ‘observers modeling themselves quantum-mechanically’, as the submission does, is not only to offer to RQM a gratuitous evasion from the counterexample, but to deny the basic premise of the Wigner’s Friend experiment in the first place. Moreover, the counterexample is not limited to ‘observers’ at all, since it can be put in form of an ‘automatic machine’ subject to the same inconsistency, as explicitly noted in Kastner, 2024:
“The above points should also be kept in mind when evaluating the preferred options of Baumann and Brukner (2019), who treat quantum theory in an instrumentalist fashion, as a tool for making predictions. They refer to the inner observer as being subject to “updating degrees of belief,” but as noted above, under that approach there is still, in principle, an infinite regress of observers whose “observed outcomes” are neither predicted by Unitary Quantum Theory nor can be regarded as facts about reality. Moreover, the latter treatment implicitly smuggles in the observer-independent occurrence of outcomes in assuming that an “automatic machine,” could play the part of the outer observer. Yet in that case, the infinite regress still applies, and the “automatic machine” could somehow be “mistaken” about its outcome. Thus, once again, Unitary Quantum Theory is faced with a fatal inconsistency among outcomes that no resort to “updating” can cure, since a machine is not subject to “degrees of belief.” (Kastner, 2024)
Clearly, the machine envisioned by Baumann and Brukner is not an ‘observer,’ yet it remains subject to the inconsistency highlighted by the counterexample. Thus, the submission’s portrayal of Kastner, 2024 as allegedly being about an inconsistency in ‘observers modeling themselves quantum-mechanically,’ implying that the inconsistency could be sidestepped by avoiding that, is not only contrary to the argumentation in the reference (which is a critique of RQM, not a salvaging of it), but is explicitly contradicted by its observation that the inconsistency still obtains for outcomes of a non-observer. Thus, the issue of ‘modeling an observer quantum mechanically’ is a ‘red herring,’ since the same inconsistency arises under UOQM for a non-observer machine, as quoted above. The submission’s misrepresentation of Kastner, 2024 is tendentious in that it shields RQM from the counterexample by offering an ‘out’ of denying that systems subject to quantum states are ‘observers’, while also neglecting the point that this wouldn’t rescue RQM from the counterexample anyway, in view of the non-observer ‘machine’ case. This misrepresentation is serious, unacceptable, and needs to be corrected, at minimum.
In any case, there is no a priori reason that an observer could not be described by a quantum state. The only reason this seems to create a problem is because of Wigner’s Friend-type inconsistencies, arising under UOQM, in which ‘measurement’ is not rigorously defined. In contrast, as noted in the cited Kastner (2024), under a well-defined reduction formulation (one which has been in the published literature for years), we don’t run into Wigner’s Friend-type problems and there is nothing stopping us from assigning an appropriate quantum state to an observer (or to a machine acting as a substitute for an observer).
Of course, it’s understandable that adherents of UOQM might wish to dispute Kastner’s inconsistency finding against RQM by arguing that we shouldn’t represent observers by quantum states. But that’s not what Kastner said, which was the opposite (and in any case, that move wouldn’t help for the non-observer ‘machine’ discussed in the cited reference). Further, denying that observers should be described by quantum states also of course contradicts the whole premise of the Wigner’s Friend experiment, which describes the Friend by a quantum state. It thus seems a rather desperate way to try to get rid of the problem raised by WF scenarios for UOQM: i.e., just deny that we can set up the WF scenario in the first place. But then, the burden is still on those attempting to evade UOQM inconsistencies through denial that an observer should be described by a quantum state to specify what sort of system counts as an ‘observer,’ and why. And again, we have the non-observer machine, whose state is taken as reflecting an outcome, also subject to the inconsistency. These points, I think, show that the problem is much deeper than considerations about what observers can know about themselves, or relative to whom or what, and associated psycho-physical digressions. The problem is, arguably, with UOQM.
Finally, I have put a few additional comments into the manuscript concerning a neglected formulation of quantum mechanics that inherits none of these sorts of problems; in particular, no ‘zombies’ or ‘mindless hulks’ arising from (at best) tenuously grounded forays into psychophysical parallelism–which only arose in the first place because ‘measurement’ is not defined in UOQM. In the humble opinion of this reviewer, it is long past time that the philosophy of physics community began to take into account alternative coherent formulations that define ‘measurement’ in clear physical terms, instead of restricting itself to the arguably incoherent UOQM that founders endlessly on the ‘shifty split’. In fact, demonstrable and published progress has been made in the direction of a consistent physical account of measurement in quantum mechanics–one which defines the conditions for ‘measurement’ unambiguously, and which does not make ad hoc changes to the basic theory. (Details are given in the commented ms.; a starting point is here: www.cambridge.org/9781108830447; and among the fruits is a unification of quantum theory and relativity, here: https://iopscience.iop.org/article/10.1088/2399-6528/acd6d7.) It is a formulation that dissents from UOQM; such dissents should not continue to be ignored, misrepresented, and/or disallowed in discussion of condundra involving ‘observers,’ where such conundra arise only from the UOQM inability to define ‘measurement’.
