Making Symmetry Explicit: The Limits of Sophistication

TL;DR

The paper investigates when symmetry in general relativity and gauge theories should be treated as harmless redundancy versus a feature that must be explicitly managed. It introduces Background-Relative Sophistication (BRS), a criterion that tests whether the symmetry group $S$ acts as automorphisms of a fixed background $B$; if so, symmetry can remain implicit, otherwise explicit machinery is required. Through detailed analysis of GR (linearised gravity, initial-value problem, and ADM formalism) and gauge theory (principal-bundle vs. gauge-potential formalisms), it shows that BRS fails precisely in the settings where symmetry must be made explicit, while even when BRS holds, tasks like quantisation and regional composition force cross-model identifications and dressings to keep locality and comparisons coherent. The work thus clarifies when physicists can work up to isomorphism and when explicit symmetry-handling tools (gauge-fixing, dressings, equilocality) are indispensable, with implications for quantum gravity and regional subsystem physics. It also ties the two gaps in sophistication—individuation and correspondence—together under the unifying framework of representational schemes built from background automorphisms.

Abstract

Symmetry is often treated in philosophy of physics as an interpretive problem. A particularly lively dispute concerns local symmetries: do they indicate surplus structure that ought to be expunged, or are they merely a harmless redundancy? One influential response favours the second option for certain theories -- those dubbed internally sophisticated. And indeed, in much of physics practice, local symmetries are left implicit: one simply works "up to isomorphism'' without pausing over invariance. But not always. In some settings, local symmetry and invariance become pressing practical concerns for physicists. Yet philosophical discussions of sophistication have paid little sustained attention to when, and why, this happens. Surveying textbook general relativity (GR) and gauge theory, I identify the settings in which diffeomorphism invariance or gauge invariance must be handled explicitly. (Here a setting is a choice of representational framework or background assumptions within which one formulates and uses the theory -- for instance, linearisation, an initial-value formulation, or a Hamiltonian $3+1$ formalism.) I propose an operational criterion -- background-relative sophistication (BRS) -- and argue that it accounts well for the pattern: it marks just where symmetry can stay implicit and where it must be made explicit. Quantum and subsystem settings raise a further difficulty: there, certain tasks (superposition and gluing) force symmetry into view even for theories that are BRS.

Figures (4)

• Figure 1: Left: a spatial diffeomorphism slides points along the leaves $\Sigma_t$ of a foliation, preserving the foliation structure; it is an automorphism of the ADM background. Right: a refoliation deforms the hypersurfaces in the normal direction; the resulting transformation depends on the dynamical fields and does not preserve the foliation structure. BRS holds for the former but fails for the latter.
• Figure 2: Left: in the principal-bundle formalism, a gauge transformation is a vertical move along the fibre---an automorphism of the bundle $P\to M$. Right: in the potential formalism, changing gauge corresponds to changing the section (the 'cut' through the bundle); this is an affine transformation $A\mapsto A + d\chi$ of the local representative, not an automorphism of any natural background structure.
• Figure 3: The space of models $\Phi$ as a principal $\mathcal{G}$-bundle over the space of physical possibilities $[\Phi]$. Each gauge orbit $\mathcal{O}_\phi$ is a fibre over the equivalence class $[\phi]$. A gauge-fixing condition $F: \Phi \to V$ with $\dim V = \dim(\mathrm{Lie}(\mathcal{G}))$ defines a section $\mathscr{F} = F^{-1}(0)$ (shown in yellow) that intersects each orbit exactly once. The dressing $\varpi: \Phi \to \mathcal{G}$ sends each model $\phi$ to the unique group element $\varpi(\phi)$ such that $\phi^{\varpi(\phi)} \in \mathscr{F}$.
• Figure 4: A one-parameter family of non-isomorphic models, $\phi_s$ (blue curve) projected to the gauge-fixing surface. The projected curve $\{\phi_s^{\varpi(\phi_s)}\}$ lies entirely in $\mathscr{F}$. The family of diffeomorphisms $\varpi(\phi_s)$threads spacetime points across non-isomorphic models: a point $x$ in $\phi_1$ corresponds to $\varepsilon(\phi_1, \phi_2)(x)$ in $\phi_2$.

Theorems & Definitions (1)

• Definition 1: Background-Relative Sophistication (BRS)