2.1. Physical perspectivalism

The increasing emphasis on perspectives in the foundations of physics has given rise to a class of views that I will refer to as physical perspectivalism, suggesting that physically meaningful descriptions must always be relativized to a perspective.

It is important to emphasize that physical perspectivalism is not primarily about knowledge. This is clear from the role that physical perspectivalism is supposed to play in the Wigner’s friend experiment. In this experiment, if we think the state ${{\boldsymbol\psi} _{AS}}$ assigned by Bob merely reflects Bob’s lack of knowledge about Alice’s measurement outcome, we will naturally conclude that the true state of Alice and $S$ is $\phi _{AS}^i = \left| {i{\rangle _A} \otimes } \right|i{\rangle _S}$ , where $i$ corresponds to the definite value that Alice has actually seen. But if unitary quantum mechanics is correct, it must be the case that when Bob subsequently performs measurements on Alice and $S$ , he will always see outcomes consistent with the state ${{\boldsymbol\psi} _{AS}}$ , which are different from the outcomes we’d expect to see if the true state were either $\phi _{AS}^i$ or some probabilistic mixture of states of the form $\phi _{AS}^i$ . Therefore, the idea that the state is ${{\boldsymbol\psi} _{AS}}$ relative to Bob cannot be merely a characterization of Bob’s ignorance but must be a substantive assertion about the actual dynamics that will take effect when Bob interacts with Alice and $S$ . To resolve the puzzle, we need physical perspectivalism to be a claim about physics rather than epistemology.

Because physical perspectivalism is not about knowledge, the “perspectives” to which facts are relativized need not be associated with possible knowers, so consciousness need not play any special role in this kind of perspectivalism. Thus, although some physical perspectivalists, such as Brukner (Reference Brukner2021) and Cavalcanti (Reference Cavalcanti2021), do focus on perspectives associated with conscious agents, it’s not clear that this limitation is well motivated. Other physical perspectivalists have employed a more liberal notion of a perspective: Any physical system can act as the origin of a coordinate system relative to which we may describe the position of other physical systems, and physical perspectivalism generalizes this idea, suggesting that any physical system can anchor an “internal view” relative to which we may describe the degrees of freedom of other physical systems.

One example of this approach is Rovelli’s relational quantum mechanics (Rovelli, Reference Rovelli1996), which maintains that quantum descriptions must always be relativized to an “observer” but then stipulates that any physical system can play the role of an observer. Another is Dieks’s perspectival quantum realism, which posits that “objects may possess different, but equally objective properties with respect to different physically defined perspectives” (Dieks, Reference Dieks2022). Dieks emphasizes that “realism should not focus on agents and their subjective points of view; if perspectivalism is to be part of a realist scheme, the perspectives in question should be defined with respect to physical systems.”

Physical perspectivalism also appears to be a motivator for the quantum reference frame (QRF) formalism (de la Hamette et al., Reference de la Hamette, Galley, Hoehn, Loveridge and Mueller2021; Giacomini et al., Reference Giacomini, Castro-Ruiz and Brukner2019; Castro-Ruiz et al., Reference Castro-Ruiz, Giacomini, Belenchia and Brukner2020; Vanrietvelde et al., Reference Vanrietvelde, Hoehn, Giacomini and Castro-Ruiz2020; Höhn and Vanrietvelde, Reference Höhn and Vanrietvelde2020; Giacomini and Brukner, Reference Giacomini and Brukner2022), which provides a suite of tools for defining an internal view relativized to a given physical system and for switching from one such internal view to another. Proponents of this formalism talk about “jumping into the perspective of a quantum system” (Vanrietvelde et al., Reference Vanrietvelde, Hoehn, Giacomini and Castro-Ruiz2020), but this is presumably not intended to suggest that QRFs generally have conscious perspectives. Rather, the idea is that in a universe without a fixed background space-time, we must describe physics relative to a physical system, and if that system is quantum in nature, then we will thereby be describing physics relative to a QRF.

2.2. Strong versus moderate physical perspectivalism

Let us differentiate two possible versions of physical perspectivalism. The first, which I will refer to as strong physical perspectivalism, maintains that all facts about physical reality must be relativized to a (physical) perspective. The second, which I will refer to as moderate physical perspectivalism, maintains that empirically meaningful facts about physical reality must be relativized to a (physical) perspective, where we say a fact is “empirically meaningful” if it gives a description of some part of reality in terms of phenomena of the kind that could be directly experienced by a realistic observer. Unlike strong physical perspectivalism, moderate physical perspectivalism allows that there may exist some perspective-neutral facts about physical reality, but it maintains that such facts are typically quite abstract and do not have direct empirical content.

Perspectivalists are fond of the mantra “there is no view from nowhere” (Di Biagio and Rovelli, Reference Di Biagio and Rovelli2022; Berghofer, Reference Berghofer2024), and strong physical perspectivalists typically interpret this statement as asserting that there cannot exist physical facts of any kind that are not relativized to a perspective. However, moderate physical perspectivalists would interpret this statement differently. One natural way to understand the term view is to think of it as referring to a description of some part of reality in terms that we ourselves can conceptualize or visualize—that is, a description expressed in empirically meaningful language referring to phenomena of the kind that could be directly observed by a realistic observer. And moderate physical perspectivalism does indeed maintain that any such “view” must be relativized to a physical perspective, so there is a sense in which it upholds the idea that “there is no view from nowhere.”

Even without appealing to the scientific results discussed in section 2, there are good reasons to believe that something like moderate physical perspectivalism is true. For although scientific theories are often formulated in an abstract, third-person way, in order to extract from a theory predictions that are useful for an embodied observer in a specific physical scenario, we necessarily have to move away from an abstract third-person description and adopt some kind of first-person view. As van Fraassen (Reference van Fraassen2008) puts it, “If someone is to use (special relativity) to predict the behavior of electrically charged bodies in motion, bodies with which s/he is directly concerned, choice of a coordinate system correlated to a defined physical frame of reference is required. The user must leave the God-like reflections on the structure of spacetime behind in order to apply the implications of those reflections to his or her actual situation.” That is, realistic observers necessarily have experiences from within a particular embodied perspective, so any empirical content of the kind that could be experienced by a realistic observer must at least implicitly appeal to the perspective of a possible observer who could have the corresponding experience.

Moderate physical perspectivalism also seems like a good way of capturing the features of quantum mechanics and relativity discussed in section 2. We have seen that the Wigner’s friend paradox pushes us toward relativizing quantum states to perspectives, and quantum states encode predictions for measurement outcomes that could be obtained by observers, so they fall into the category of “empirically meaningful descriptions.” Thus, the relativization of states to perspectives in a quantum context looks like an instance of the idea that empirically meaningful descriptions must be relativized to a perspective. Meanwhile, in the framework of partial/complete observables, statements about phenomena that could be experienced by observers are necessarily expressed in terms of partial observables, and the formalism tells us that partial observables must always be relativized to some physical system or “perspective,” so again we have an instance of the idea that empirically meaningful facts should be relativized to a perspective.

Additionally, because perspectivalism has been cited as a potential route to unifying quantum mechanics and relativity, we should consider what is suggested by the combination of both theories. At present, it is not straightforward to extract empirical content from existing proposals for full theories of quantum gravity, so I will instead address this question by considering some general consequences of the Dirac quantization procedure (Dirac, Reference Dirac1981), which is the usual approach to quantizing theories subject to constraints. Because many physicists expect that a theory of quantum gravity will satisfy diffeomorphism invariance just as general relativity does, a natural way of approaching the problem is to use Dirac quantization with a diffeomorphism constraint.

In the process of Dirac quantization, we start from a “kinematical” Hilbert space encoding the most general possibilities, and we impose constraints in order to move to a “physical” Hilbert space encoding only states that are physically possible, in the sense that they respect the symmetries associated with the constraints. But in the case of a diffeomorphism constraint, the physical Hilbert space is quite abstract and contains redundancies in its representations, so “in order to make operational sense out of physical phenomena, we must make additional choices to fix these redundancies,” which can be done by choosing a system from the perspective-neutral picture to serve as our reference frame” (Vanrietvelde et al., Reference Vanrietvelde, Hoehn, Giacomini and Castro-Ruiz2020). That is, once again, we find that in order to extract empirically meaningful content, it is necessary to adopt an internal reference frame; thus, current ideas about quantum gravity also seem to support the idea that empirically meaningful descriptions must be relativized to perspectives.

4.1. Rejecting the Cartesian–Hegelian ideal

One important motivation for physical perspectivalism is the idea that modern physics has revealed the bankruptcy of what one might call the traditional “Cartesian–Hegelian” metaphysics, in which it is assumed that the most fundamental description of reality is completely observer independent and can be characterized in wholly third-person terms. Descartes and Hegel were prominent defenders of scientific objectivity, with Descartes believing that “scientific knowledge—and especially mathematical physics—is God’s gift to humans that allows them to overcome the sorts of errors to which they are prone, largely because of their embodied nature” (Halvorson, Reference Halvorson, Kemp and Wietzkeforthcoming), and meanwhile Hegel “conceiv(ing) of the epistemic ideal as the backward-looking, stationary state of the detached spectator, i.e. the god’s eye view” (Halvorson, Reference Halvorson, Kemp and Wietzkeforthcoming). Since that time, the Cartesian–Hegelian ideal has come under sustained attack for both philosophical and scientific reasons, and many physical perspectivalists appear to see their position as a logical culmination of this intellectual development. For example, Cuffaro and Hartmann (Reference Cuffaro and Hartmann2024) write, “the ideal of an observer-independent reality is not methodologically necessary for science and … modern physics (especially, but not only, quantum theory) has taught us … that there is a limit to the usefulness of pursuing this ideal.’’

Now, we have seen that both quantum and relativistic considerations provide good reasons to question the Cartesian–Hegelian metaphysics. However, the Cartesian–Hegelian picture is characterized not only by the idea that there exists a perspective-neutral description of reality but also by the stronger claim that this perspective-neutral description constitutes the most fundamental and complete characterization of reality. One way of rejecting the Cartesian–Hegelian picture is to deny the existence claim, which leads to strong physical perspectivalism, but another way of rejecting the Cartesian–Hegelian picture is to simply deny the claim about the fundamentality of the third-person perspective while leaving the existence claim alone, which leads instead to something like moderate physical perspectivalism, in which we stipulate that both perspectival and perspective-neutral facts exist and both are equally fundamental. That is, moderate physical perspectivalism suggests that although we may be able to map between the perspectival and perspective-neutral descriptions, first-person perspectives are an essential feature of reality as we experience it, and we should not be seeking to eliminate them in favor of a third-person view. Thus, although it may well be true that quantum mechanics and/or relativity give us reasons to reject the traditional Cartesian–Hegelian metaphysical picture, this does not mean we must be strong physical perspectivalists; moderate physical perspectivalism offers an alternative way of responding to this situation.

For example, given the central role of partial observables in our empirical experience, there is a sense in which the partial observables must be taken seriously as elements of reality even though they are not individually real according to the standard interpretation; as Rickles (Reference Rickles and Dieks2008) puts it, “both spaces—the space of genuine (complete) observables and partial observables—are invested with physicality by Rovelli.” Insofar as this is a correct characterization of Rovelli’s views, one way to understand it would be to see Rovelli as investing the partial observables with physicality because he sees the first-person perspective (relative to which partial observables are defined) as having equal validity to the third-person perspective (in which complete observables are defined)—so this way of thinking about the relation between partial and complete observables does indeed look like a form of moderate physical perspectivalism.

4.2. Epistemic perspectivalism

Other philosophical intuitions lying behind the adoption of strong physical perspectivalism appear to be linked to previous philosophical perspectivalisms. There is a long tradition of such philosophical views, stretching back at least to the work of Leibniz (Reference Leibniz1714) and continuing through the work of Nietzsche (Reference Nietzsche1887), Kuhn (Reference Kuhn1996), Rorty (Reference Rorty1991), and Giere (Reference Giere2006), among many others. In general, such views maintain that all knowledge is necessarily relativized to the perspective of the knower, and no matter how hard we try to achieve “objectivity,” we will never wholly escape our perspective to achieve perspective-neutral knowledge. As Rorty (Reference Rorty1991) puts it, “the image of climbing out of our minds—to something external from which we can turn and look at them – needs to be replaced.”

Notably, these previous philosophical perspectivalisms have typically fallen into the category of what I will refer to as epistemic perspectivalism—that is, positions that pertain to what it is possible for beings like us to know or believe, rather than what exists. That is, when epistemic perspectivalists assert that “there is no view from nowhere,” this is not a claim about the existence or otherwise of perspective-neutral facts; rather, it is asserting that the “view from nowhere” is not a perspective that can be taken up by any realistic observer. Relatedly, because epistemic perspectivalism is a claim about what certain kinds of beings can know or how they can think, it is typically concerned only with “perspectives” associated with conscious subjects.

Evidently, then, these previous philosophical perspectivalisms are different from physical perspectivalism. It does seem plausible that physical perspectivalism entails some form of epistemic perspectivalism, for if all facts, or all empirically meaningful facts, must be relativized to a perspective, then it seems natural to think that knowledge of such facts must also be associated with a perspective. But the converse implication is not so obvious. Many presentations of epistemic perspectivalism explicitly endorse the existence of an external perspective-neutral reality while maintaining that we cannot have perspective-independent knowledge of it; for example, Massimi (Reference Massimi2018) defines perspectival realism as the view that “states of affairs about the world are perspective-independent; whereas our scientific knowledge claims about these states of affairs are perspective-dependent.”

It is true that there exist some philosophical views that would have the effect of erasing the distinction between epistemic and physical perspectivalism. For example, if one is committed to some form of idealism positing that (human) thought is the foundation of reality (Guyer and Horstmann, Reference Guyer, Horstmann, Edward and Nodelman2023), then the claim that agents cannot have knowledge that is not relativized to their own perspective inevitably leads to the claim that all physical facts must be relativized to a perspective. Similarly, some forms of pragmatism and positivism, or the phenomenological approaches that have recently been revisited within quantum foundations (French, Reference French2023), would erase the distinction between epistemology and ontology, so the perspectival nature of all knowledge would necessarily entail that all facts are perspectival.Footnote ¹ However, these kinds of views do not combine well with versions of physical perspectivalism that attribute perspectives to all physical systems—for a description relative to a nonconscious physical system is not known or perceived by anyone, so it would not make sense for a typical idealist, positivist, or phenomenologist to see this perspective as physically meaningful. Thus, although it may be the case that the physical perspectivalists who focus on conscious perspectives are ultimately peddling a new form of idealism, positivism, or phenomenology, other physical perspectivalists who employ a more liberal notion of “perspective” cannot be considered idealists, positivists, or phenomenologists.

So, in at least some cases, it appears that physical perspectivalism must be understood as a distinct position from epistemic perspectivalism. But nonetheless, many physical perspectivalists appear to be motivated by intuitions linked to epistemic perspectivalism. In particular, one common motivation for epistemic perspectivalism is a kind of epistemic humility, based on the observation that all knowledge is possessed by physically embodied observers who are subject to various kinds of limitations in terms of their epistemic access to reality. For example, Giere’s scientific perspectivism (Giere, Reference Giere2006) is motivated by consideration of the epistemic limitations to which we are subject, such as the limitations of our scientific instruments: “instruments are sensitive only to a particular kind of input. They are, so to speak, blind to everything else. Second, no instrument is perfectly transparent. That is, the output is a function of both the input and the internal constitution of the instrument.” Giere argues that our instruments are perspectival because of the way in which existing theory influences their design and intended applications, and thus the scientific knowledge we obtain from them is also necessarily perspectival.

And some proponents of physical perspectivalism also seem to be motivated by a similar kind of epistemic humility. For example, Rovelli (Reference Rovelli2024) first notes that there are inevitable limitations on human knowledge— “we live in this space between ignorance and certainty. The two extremes are of no interest. What matters to us is the space in between”—and then uses this observation to motivate the conclusion that “reality is more tenuous than the clear-cut one imagined by the old physics models; it is made up of happenings, discontinuous events, without permanence, located with respect to one another and only existing relatively to one another.” Here, epistemic considerations are used to motivate what appears to be a form of physical perspectivalism. Similarly, Evans (Reference Evans2020) invokes Giere’s arguments about the limitations of scientific instruments to motivate a version of physical perspectivalism intended to address the Wigner’s friend paradoxes. It appears that these authors believe that the considerations that support epistemic perspectivalism also lend some support to physical perspectivalism.

And in fact, it seems plausible that epistemic humility provides some support for moderate physical perspectivalism. Such an argument might go something like this:

1. (1) Realistic observers are always subject to limitations in their epistemic access to reality.

2. (2) Therefore, all of the direct empirical content that can be experienced by a realistic observer is filtered through the lens of their epistemic limitations.

3. (3) Perspective-neutral facts about reality are, by definition, not filtered through the lens of any epistemic limitations.

4. (4) Therefore, perspective-neutral facts cannot express direct empirical content of the kind that could be experienced by a realistic observer.

But can epistemic humility support strong physical perspectivalism? Such an argument would presumably go something like this:

1. (1) Realistic observers are always subject to limitations in their epistemic access to reality.

2. (2) Therefore, scientific inquiry can only yield knowledge that is relativized to a perspective.

3. (3) Therefore, scientific inquiry cannot give us any evidence for the existence of facts that are not relativized to a perspective.

4. (4) Thus, applying Ockham’s razor, we should believe that there do not exist any facts that are not relativized to a perspective.

However, there are two important problems with this argument.Footnote ² I will go through these two problems in some detail because I think this will help to clarify the connection between epistemic and physical perspectivalism and will therefore provide a clearer understanding of the philosophical motivations for strong physical perspectivalism.

4.3. First-person-plural views

In light of concerns about the social nature of science, some epistemic perspectivalists have adopted what we might call a first-person-plural view, where perspectives are associated not merely with single individuals but also with entire epistemic communities. Thus, epistemic perspectivalists can accept that, indeed, social features of science allow us to transcend the perspectives of individual observers, but they still maintain that the resulting knowledge is relativized to a perspective—it is simply the perspective of the whole epistemic community, rather than an individual observer. For example, Rorty (Reference Rorty1991) argues that “whatever good the ideas of ‘objectivity’ and ‘transcendence’ have done for our culture can be attained equally well by the idea of a community which strives after both intersubjective agreement and novelty.”

And in light of the concerns expressed in the previous section, physical perspectivalists might be tempted to make a similar move. This would involve arguing that in addition to facts relativized to the perspectives of individual observers, there also exist facts relativized to some kind of emergent perspective that results from information being shared across a community, so physical perspectivalists can acknowledge the role of the community in the acquisition of scientific knowledge.

One example of such a view is Healey’s “desert pragmatism” (Healey, Reference Healey2012), in which quantum states and measurement outcomes are relativized to the situation defined by a “decoherence environment,” that is, a region of space-time in which environmental decoherence stabilizes the value of some variable such that it can meaningfully be assigned a value. Because decoherence will spread rapidly through a community of macroscopic observers, it is reasonable to expect that the entire human community belongs to the same decoherence environment, so Healey’s picture suggests that our entire epistemic community shares a set of relativized facts about quantum states and outcomes.

Similarly, Evans (Reference Evans2020) puts forward a notion of “perspectival objectivity,” in which “a scenario in which some feature of the world is in part a function of the agent perspective while at the same time, given such a perspective that is inescapably shared between similar agents, there is an (intersubjectively) objective fact of the matter concerning that feature.” Evans (Reference Evans2020) appears to be arguing that this kind of objectivity can exist even in the context of physical perspectivalism because he explains the motivation for the view as follows: “to attempt to accommodate the recent claims from quantum foundations that quantum mechanics rules out the possibility of ‘observer-independent facts.’” That is, Evans is positing that in the context of physical perspectivalism, the process of sharing information can bring into being a higher-level emergent perspective associated with a whole epistemic community, and thus we can still do justice to the important role of intersubjective sharing of information as a means of achieving a higher level of objectivity.

Now, both of these approaches work well in the context of moderate physical perspectivalism. For example, Evans’s notion of perspectival objectivity would be a helpful way for the moderate physical perspectivalist to explain how empirically meaningful facts relativized to individual perspectives can nonetheless be understood as having a kind of objectivity by virtue of the way in which they are shared between the perspectives of different agents.

But can such approaches be employed in the context of strong physical perspectivalism? This seems challenging because strong physical perspectivalism is committed to denying the existence of any mechanisms that could bring about connections between the perspectives of different agents. Moderate physical perspectivalists can maintain that there exists some kind of perspective-neutral underlying structure that connects all of the perspectives together and explains the relations between them, such as Minkowski space or the global state ${\rm{\Psi }}$ described in section 3, but strong physical perspectivalists cannot allow such a possibility; they would have to insist that facts about such a structure must themselves be relativized to something, and thus, as Riedel (Reference Riedel2024) puts it, we inevitably end up with an “iteration of relativization.” So, we can’t simply get rid of external reality and yet maintain that information is really shared across a community; the first-person-plural approach in the context of strong physical perspectivalism amounts to telling the same story as in the moderate case but then pulling out the rug in a way that destabilizes the whole view.

We can see this in action with regard to Evans’s concept of perspectival objectivity, for in order for this concept to apply to some set of observers $\left\{ O \right\}$ in certain specific circumstances, there must exist certain facts about (a) the nature of the observers in $\left\{ O \right\}$ ; (b) their current circumstances; (c) the ways in which the observers are constrained in their access to the world; and (d) the relations that hold between different observers, such as the fact that intersubjective sharing of certain kinds of information is possible for them. But in the context of strong physical perspectivalism, we are not allowed to postulate any facts that are not relativized to a perspective, so which perspective could facts (a) through (d) be relativized to?

We might try to say that these facts are simply relativized to some particular observer $O{\rm{'}}$ in the set $\left\{ O \right\}$ . However, in that case, it is true only relative to that $O{\rm{'}}$ that information is shared across the set of observers, so the “shared” information is no more objective than any other piece of information relativized to $O{\rm{'}}$ ; the higher level of objectivity that is supposed to be conferred by the fact that the information is shared by the whole community is lost.

We might try to say that facts (a) through (d) are true relative to the joint perspective of all of the observers in the set $\left\{ O \right\}$ . But the joint perspective of a community does not simply come into being as a fundamental unit; rather, the community is brought into being by dynamical interactions that create correlations between individual perspectives, and in order to make sense of this process, we need to be able to describe these dynamical interactions in a way that does not presuppose the existence of the joint perspective that is supposed to be created by those very interactions. In addition, if we simply postulate the whole epistemic community as a fundamental unit, we will be left with no principled way to distinguish between groups of observers who share a joint perspective and groups of observers who do not: It is precisely the existence of perspective-neutral facts about the spread of information that allows us to identify when an epistemic community has been formed. Thus, in order to do justice to the dynamical, participatory formation of an epistemic community, we must be able to offer a perspective-neutral description of the way in which dynamical processes create connections between perspectives.

Finally, we might try to say that facts (a) through (d) are true relative to some other observer outside of the set $\left\{ O \right\}$ . But it is hard to see how this would yield any meaningful notion of objectivity. What Evans wants is for the members of the community $\left\{ O \right\}$ to be able to see their own scientific judgments as (quasi-)objective, by virtue of information being shared across the whole community, and it is no help from the point of view of an observer in the set $\left\{ O \right\}$ that some other observer thinks the information is shared across the whole community.

Moreover, the need for perspectival objectivity to be backed up by perspective-neutral facts is clear from Evans’s motivating examples. Evans cites color vision and causation as perspectivally objective phenomena, but the perspectival objectivity of these phenomena is grounded on underlying perspective-neutral facts about how our individual perspectival experiences of these phenomena are related. For example, Evans argues that although color is not part of the fundamental furniture of the world, it has a kind of “objectivity” by virtue of the fact that most members of our epistemic community agree about judgments of color. But that agreement does not exist in a vacuum. The members of our epistemic community agree about color because there exists an underlying mechanism according to which objects reflect certain frequencies of light and the reflected light causes observers like us to have certain kinds of experiences. And the facts about which frequencies of light are reflected by a certain object are not relativized to any perspective, which is precisely what allows these facts to ground connections between color experiences featuring in two different perspectives. If there were no such mechanism connecting our experiences of color vision, we wouldn’t generally agree about judgments of color, and thus there would be no perspectival objectivity.

For these reasons, it seems to me that we cannot invoke anything like Evans’s notion of perspectival objectivity or Healey’s desert pragmatism in the context of strong physical perspectivalism. These approaches work much better if we think of them as forms of moderate physical perspectivalism. So, there exist a number of interesting routes to explore here for the moderate physical perspectivalist, but options appear very limited for the strong physical perspectivalist.