The 'Participatory Horizon': Causal Limits and the CMB Low Power Anomaly

Several persistent anomalies in the Cosmic Microwave Background (CMB) appear concentrated at the largest angular scales, most notably an anomalous suppression of temperature correlations for separations $\theta \gtrsim 60^\circ$. Curiously, the scale at which these correlations vanish corresponds to a comoving distance comparable to the radius of the observable universe.  Plausible explanations invoke statistics, while tuned primordial initial conditions have also been considered. Here, we explore an alternative hypothesis motivated by Wheeler's participatory view of quantum measurement. We propose that classical cosmological records are instantiated only to the extent that correlations can be resolved and irreversibly registered within an observer's causal domain. In this stance, the CMB is treated as a relational reference frame rather than a fixed fossil record, with correlations on scales comparable to the 'participatory horizon' remaining under-instantiated as classical records, effectively filtering their contribution to the realised sky.

This concept is explored phenomenologically by applying a smooth horizon-scale infrared suppression to a fiducial $\Lambda$CDM spectrum while leaving the background cosmology unchanged. Fitting to Planck TT bandpowers over $2 \leq \ell \leq 30$ yields a preferred cutoff scale $k_{\rm cut} = \alpha (\pi / \eta_0)$ with $\alpha \simeq 0.69$ (for a representative sharpness $p=6$), corresponding to $k_{\rm cut} \simeq 1.53 \times 10^{-4} \, \text{Mpc}^{-1}$. This result indicates a preference for a suppression scale anchored to the observer's present horizon. Using a diagonal bandpower $\chi^2$ diagnostic, the fit is improved by $\Delta \chi^2 \simeq -5.7$. A complementary per-multipole inverse-gamma likelihood, which accounts for the non-Gaussian distribution of low-$\ell$ power spectrum estimators and for cosmic variance via effective degrees of freedom, yields a consistent best-fit $\alpha$ with $2\Delta\ln\mathcal{L} \simeq +3.1$ The large-angle statistic $S_{1/2}$ is reduced by ${\sim}\,85\%$ relative to fiducial $\Lambda$CDM, while the acoustic peak structure at $\ell \gtrsim 50$ remains unchanged.

The model offers potenially falsifiable predictions for the low-$\ell$ $EE$-mode spectrum and temperature--polarisation cross-correlations, with the precise strength of the polarisation suppression depending on whether the filter operates in angular or comoving $k$-space. These exploratory results cautiously suggest that the low power anomaly might be feasibly reframed as a dynamic participatory effect, providing a viable quantitative overlay to the $\Lambda$CDM model.