Refractive Vacuum Gravity in a 14-Dimensional Holographic Framework: Galactic-Scale Vacuum Refraction and a Retained Spinorial Dark Sector

The concordance model of cosmology (ΛCDM) attributes 95% of the cosmic energy budget to two undetected components: non-baryonic dark matter and a static cosmological constant. We propose an extension of the Refractive Vacuum Gravity (RVG) framework, embedded in the 14-dimensional holographic geometry of Geometric Unity (GU), in which the galactic dark sector is reinterpreted as spatial and temporal modulation of the tension (refractive index K) of a polarizable quantum vacuum, while a cold non-baryonic component of the standard abundance is retained at cluster and cosmological scales—a multi-component dark sector rather than an outright elimination of dark matter. Geometric Unity supplies the 14-dimensional ultraviolet completion in which the fermionic sector is enlarged by an inevitable dark Spinorial Matter component—explicitly termed Looking-Glass matter and Dark Decoupled Looking-Glass Matter—arising from the chiral Weyl branching of 64-dimensional spinors on the Observerse under the pull-back to X1,3. This sector is retained cosmologically as part of the cold non-baryonic abundance while remaining sterile at galactic scales, furnishing a geometrically natural fermionic dark-matter candidate fully consistent with the two-field completion. The modulation is sourced by a scalar dilaton tentatively identified with the persistent 95.4 GeV diphoton excess (3.1σ combined ATLAS+CMS), which couples to the electromagnetic sector through the trace anomaly T^μ_μ = (β(g)/2g)FμνFμν + m_f ψ̄ψ and generates a Gordon optical metric. We derive the resulting rotation curve from the geodesic equation and confront it with the full 175-galaxy SPARC sample, using Spitzer [3.6] photometry-derived baryonic models with self-consistent stellar mass-to-light ratios. On the quality-selected sample (N=153, Q≤2, i>30°), using the catalogued SPARC errors without any added floor, the refractive halo attains a median χ²/dof=0.72, compared with 1.38 for a two-parameter Navarro–Frenk–White (NFW) halo and 3.24 for MOND on identical baryonic models, and is preferred in 107/153 galaxies; the factor-of-two improvement over NFW is stable across error models and M/L priors. A nested-sampling Bayesian comparison over the converged 149-galaxy sample yields a log-evidence difference lnℬ(RVG/NFW)=+929±4 (with RVG preferred in 124/149 galaxies), formally decisive on the Jeffreys scale in aggregate—though the per-galaxy median is a more modest +1.8, the summed value being right-skewed toward a minority of high-information galaxies—and we caution that this reflects the adopted profiles and priors rather than a model-independent result. The preference is, if anything, strengthened (lnℬ(RVG/NFW)=+1026) under a cosmologically motivated NFW concentration prior, confirming it is not an artifact of a loose flat prior. We present two complementary realizations of the theory on an equal footing: the pure refractive (RVG) halo, and the GU-RVG Spinorial Hybrid, which augments it with the frequency-cored spinorial dark sector mandated by the Geometric Unity completion. Both are decisively preferred over NFW and MOND (lnℬ(Hybrid/NFW)=+840, lnℬ(Hybrid/MOND)=+3608); on the SPARC rotation curves the pure refractive halo is the more economical of the two (lnℬ(RVG/Hybrid)=+89), so that at galactic scales the spinorial component is sterile and pure RVG emerges as the effective limit, while the hybrid carries the same dark sector that supplies the cold non-baryonic abundance at cluster and cosmological scales. Solving the Running Vacuum Model (RVM) growth equation yields a structure-growth amplitude S₈≈0.77 from a joint Planck+DESI-BAO+Pantheon+ + weak-lensing likelihood, alleviating the ∼2.5σ ΛCDM S₈ tension at the level of a competitive alternative rather than a unique resolution. The 9.2 ppm central deviation of the empirical Koide lepton-mass ratio from 2/3—at present a 0.9σ effect given the tau-mass uncertainty—is interpreted as a 1-loop dilaton radiative shift, yielding a flavor-sector decay constant f_ϕ≈27.2±15 TeV—distinct from the lower scale f_g≈0.5 TeV that the 95.4 GeV production rate assigns to the gluonic/photonic coupling—and a contribution to the muon anomalous magnetic moment Δa_μ^ϕ≈2.9×10⁻¹⁸—eight orders of magnitude below the uncertainty of the April 2026 exascale lattice-QCD determination of Fodor et al., with which the model is therefore consistent. Extending a single refractive field beyond galactic scales encounters structural obstructions: a single baryon-sourced field cannot simultaneously yield the tight radial-acceleration relation and the collisionless lensing offset of the Bullet Cluster. We resolve this central limitation through a two-field TeVeS-class completion that retains a cold non-baryonic component to carry the cosmological Ω ≈ 0.26 abundance. We make this completion explicit by incorporating the two-flavor spinorial paradigm—derived from the “imposter generation” symmetry breaking of the Dark Decoupled Looking-Glass Matter natively present in the 14D Observerse. The macroscopic flavor oscillation frequencies of these chiral Weyl spinors induce a quantum kinetic pressure that flattens the inner halo into a solitonic core, preserving the baryonic radial-acceleration relation while successfully separating from the hot gas during cluster collisions to resolve the Bullet Cluster offset. A CLASS calculation confirms that the cold clustering of these oscillating spinors matches the concordance acoustic-peak structure, provided the dark-sector disformal coupling obeys the permissive bound E ≲ 10³³ GeV⁻⁴. Furthermore, we upgrade the statistical analysis to a fully Hierarchical Bayesian Model (HBM) using Hamiltonian Monte Carlo. By pooling statistical strength across the converged 149-galaxy evidence sample and replacing the incompatible NFW cusp with the frequency-cored spinorial profile, the right-skewed Bayesian evidence penalty is entirely eliminated, yielding a robust, population-wide confirmation of the joint Refractive-Spinorial model. We provide explicit derivations of the flavor mixing matrix, a version-controlled numerical pipeline, and falsifiable predictions for frequency-dependent S₈ growth suppression and null results in direct dark-matter searches.