Dualities in physics have challenged traditional forms of scientific realism by undermining the idea that theories describe a unique underlying ontology. In this paper, we develop a new perspective on scientific realism that responds to this challenge. We argue that while realist commitment remains appropriate at the level of a theory's full formal structure, ontological commitment should be treated as tied to specific empirical contexts rather than to a fixed, real ontology. Our proposal draws inspiration from Dennett's notion of a "compression algorithm" as a defining criterion of a scientific theory. On this basis, we separate realism from ontological commitment. To clarify the stakes of this distinction, we contrast our approach with common core realism, which locates ontology in the invariant structure shared by dual models. Focusing on dualities in quantum mechanics and string theory, we show how our view accommodates ontological pluralism while preserving a robust form of structural realism.
Deep Dive into Realism and Ontology in Quantum Mechanics and String Theory.
Dualities in physics have challenged traditional forms of scientific realism by undermining the idea that theories describe a unique underlying ontology. In this paper, we develop a new perspective on scientific realism that responds to this challenge. We argue that while realist commitment remains appropriate at the level of a theory’s full formal structure, ontological commitment should be treated as tied to specific empirical contexts rather than to a fixed, real ontology. Our proposal draws inspiration from Dennett’s notion of a “compression algorithm” as a defining criterion of a scientific theory. On this basis, we separate realism from ontological commitment. To clarify the stakes of this distinction, we contrast our approach with common core realism, which locates ontology in the invariant structure shared by dual models. Focusing on dualities in quantum mechanics and string theory, we show how our view accommodates ontological pluralism while preserving a robust form of struct
Dualities are one of the most conspicuous characteristics of fundamental physics. They play a pivotal role in string theory and show up in various contexts of quantum field theory. In a slightly broader sense, quantum mechanics itself is characterized by dual representations.
One of the central challenges dualities pose for scientific realism is the apparent decoupling between a theory’s fundamental structure and its ontological content. Dual models often differ dramatically in their ontology-featuring different fundamental entities, interactions, and even different spacetime structures-while remaining empirically equivalent. This has led to proposals that may be subsumed under the name common core realism (Matsubara and Johansson, 2018, Rickles, 2017, Huggett, 2017), and more recently (De Haro and Butterfield, 2025), which seek to identify invariant structure across dual models as the proper target of both realist and ontological commitment.
In this paper, we propose a different approach. We argue that realist commitment regarding a theory that has different dual representations should be decoupled from the endorsement of a specific ontology. Our proposal is related to and bears some resemblance to Dennett, 1991’s defense of “real patterns”. There, Dennett argues that realist commitment should not depend on whether a structure is fundamental, but on whether it supports a successful compression algorithm, i.e., whether it yields explanatory and predictive power. Dennett’s proposal amounts to separating realism from fundamentalism: one can be a realist about a pattern without insisting that it reflects the ultimate building blocks of reality. Dennett’s point is that explanatory success, not metaphysical depth, should guide our realism. We adopt Dennett’s concept of a compression algorithm but deploy it in a slightly different way: whereas Dennett’s account uses it to separate realism from fundamentalism, our use of it serves to separate realism from ontology.
A scientific theory, on our account, amounts to a compression algorithm in Dennett’s sense. Realist commitment, on this account, expresses trust in the theory’s viability (empirical adequacy) in a given empirical domain or, if one endorses a universal final theory claim about a given theory, with regard to the entire world. But realist commitment does not imply the endorsement of an ontology of the world.
Although ontology is thus separated from scientific realism, it still plays an important role in the understanding of the world. It is tied to the specific contingent characteristics of the theory’s dynamics, degrees of freedom, solutions, and ground state that provide the basis for the human observer’s empirical access to the world. Thereby, it points at a specific observational perspective on the world, representing the theory in terms of human intuitions about elementary objects that are rooted in that perspective. In other words, we endorse a view of ontology that is, in a deep sense, perspectival (Adlam, 2024), sensitive to the empirical interface between theory and world. We will call the described take on ontology observer-based ontology.
To avoid ambiguities in what follows, it is helpful to introduce a few clarifying distinctions between our terminology and nearby positions in the literature. In our account, realism refers to commitment to the approximate truth of a theory’s full formal structure, including the invariant relations preserved across dual models. By contrast, ontology designates the entities or structures that arise only in regimes where the theory affords empirical access, such as subsystem decompositions in quantum mechanics or near-classical limits in string theory. Because these regimes exist only relative to the physical presence of observers, ontology in our sense is observer-based. Crucially, this does not mean that ontology is observation-dependent (varying with specific measurement outcomes), nor perspectivalist in the sense of Massimi, 2022, because the human perspective, and therefore the adequate choice of ontology, is enforced by the universe’s dynamics and the place in the universe we happen to be in. Extracting the ontology, therefore, is not a matter of pragmatic choice. Realism thus targets the truth of the theory’s observer-independent invariant structure, while ontology reflects the structures that become salient only when empirical access is physically realized.
The described separation between realist commitment and ontology allows us to retain the epistemic virtues of scientific realism even in the presence of dualities, without committing to a unique or real ontology. It also provides a basis for distinguishing between two equally important ways of speaking about the world: an observer-independent commitment to the (approximate) truth of a theory can be expressed in the form of realist commitment, and a conception of the world in terms of the empirical perspective we happen to have on the world that c
…(Full text truncated)…
This content is AI-processed based on ArXiv data.
