Appendix H: Empirical Hints and Exploratory Extensions for RFH and Programmability¶

This appendix has two parts. Part I collects exploratory empirical case studies where existing instruments are viewed through the Continuum Computation Thesis (CCT) lens, particularly the RFH bandwidth–discreteness law—the bandwidth–quantization relation (sometimes abbreviated BQL) at the core of RFH—and programmability–energy ideas. Part II collects exploratory extensions and Phase 3+ signposts that are useful for orientation but are not part of the operational identification spine in Appendix C. The goal is not to prove universality, but to:

• illustrate how to map real-world platforms onto CCT quantities (bandwidth \(B\), discreteness proxies \(\Delta f/f\), confounders \(Z\), and approximate \(\chi \equiv P/(kTB)\)),
• prototype RFH-style log–log fits in familiar domains, and
• surface regime-specific behaviors that can inform future, dedicated experiments.

These case studies are heterogeneous and opportunistic: they pull from existing datasets and instruments (LIGO, camera sensors, radar odometry, ECG, pulsar timing) with varying degrees of control over confounders, hardware design, and preprocessing. They are intended primarily as consistency checks and workflow templates—demonstrating how RFH fits and regime classifications can be applied in practice—rather than as strong evidence for any universality claim. In many instances the observed exponents coincide with what standard detection theory would already predict; the value of RFH here is to provide a common scaling grammar across very different platforms. More controlled analog systems and purpose-built experiments, developed outside the scope of this paper, are needed to meaningfully test the broader CCT conjectures.

These case studies are best treated as empirical hints and worked examples. They complement the toy worlds and estimators in Appendix C rather than extending CCT’s empirical claims.

Note on Prog_T scope. The case studies in this appendix focus on RFH bandwidth–discreteness probes, not on programmability (Prog_T). Prog_T estimation requires explicit control inputs, energy accounting, and causal intervention—conditions satisfied in CCT Labs simulation and hardware environments (see cct-lab.md for current experimental phases) but not in the observational datasets used here (LIGO, cameras, ECG, pulsars). Future work may extend Prog_T estimation to observatory-class instruments with known control parameters (e.g., adaptive optics, active interferometer feedback), but such estimates are beyond the current scope. Readers seeking Prog_T details should consult cct-scientific.md §4 and Appendix C §3.5.

Pilot RFH probe summary (exploratory). The table below consolidates the regime‑local RFH probes in this appendix. All \(\alpha\) values are platform‑ and regime‑specific worked examples, not evidence for universality.

Case (section)	Bandwidth proxy \(B\)	Discreteness proxy \(\Delta\)	Regime class	\(\hat\alpha\)
ADC oversampling (§H.1)	OSR / effective in‑band throughput	ENOB / in‑band noise floor	Instrument‑specific RFH‑PL	\(0.3\text{–}1.4\) band (illustrative)
Optical squeezing (§H.2)	Squeezing / detection bandwidth	Noise below shot / quantization threshold	SQL → loss‑limited	\(0\text{–}0.5\) band (illustrative)
LIGO matched‑filter (§H.3)	Coherent integration bins in 80–300 Hz	5σ minimal detectable line amplitude	Coherent / Fourier	\(0.99 \pm 0.03\)
Camera patches (§H.5)	Spatial samples \(B=N^2\)	5σ minimal detectable contrast	Incoherent averaging	\(0.50 \pm 0.01\)
RobotCar radar (§H.6)	Coherent frames / integration window	Range‑rate discretization error	Coherent navigation	\(\approx 0.99\)
ECG averaging (§H.7)	Beats averaged per template	Morphology resolution error	Incoherent physiology	\(0.5\text{–}0.55\)
Pulsar timing (§H.8)	Timing baseline \(B \propto T\)	Phase/frequency residual floor	Super‑coherent	\(\approx 1.5\)
Paleomag excursions (§H.8b)	1/smoothing time (samples/kyr)	Apparent directional duration (yr)	Sub-incoherent to incoherent	\(0.15\text{–}0.75\) (pilot)
Economic time series (§H.8c)	Temporal aggregation bandwidth \(B=1440/\Delta t\)	Aggregation distortion vs 1m baseline	Sub-incoherent (RV, near incoherent), below-band tail	\(\alpha_{\text{RV}}\approx 0.52,\ \alpha_{\text{tail}}\approx 0.28\)
Bioelectric morphogenesis (§H.B1–B3)	Connectivity / pattern coherence (proxy)	Morphological error / RI	Controller‑type RFH	\(\alpha>0\), rough \(0.5\text{–}1.5\) (two‑point)
Bioelectric gap junction (§H.B4)	Gap junction conductance \(g_{\text{gap}}\)	Voltage pattern heterogeneity (CV)	Sub‑incoherent	\(0.35 \pm 0.02\)

Part I — Empirical Hints and RFH Worked Examples¶

H.1 ADCs and Oversampling: Lab-Scale Compilers¶

Sketch.
Successive-approximation and sigma–delta ADCs act as compilers from analog flux to digital codes. Many datasheets tabulate:

• effective number of bits (ENOB) or SNR,
• oversampling ratio (OSR) or effective bandwidth, and
• power and temperature operating ranges.

An RFH-style analysis treats:

• bandwidth proxy \(B\): effective bandwidth or oversampling ratio,
• discreteness proxy \(\Delta f/f\): in-band noise or quantization step size normalized by full-scale,
• confounders \(Z\): supply voltage, temperature, gain settings, clock mode.

For a given device family, one can fit: $$ \log!\left(\frac{\Delta f}{f}\right) = -\alpha \log B + \beta^\top Z + \varepsilon $$ over OSR sweeps. Preliminary fits in generic MCU/ADC families suggest:

• \(\alpha\) values living in a narrow band (roughly \(0.3 \lesssim \alpha \lesssim 1.4\)) rather than spanning the full rate–distortion space,
• roll-off and saturation at high OSR where thermal and reference noise dominate, and
• device-specific structure (e.g., steeper \(\alpha\) in early OSR steps, flattening later).

These are illustrative: datasheet-level numbers are heterogeneous and not optimized for RFH. They show how to treat ADCs as RFH domains and motivate more controlled lab sweeps.

H.2 Optical Squeezing: Sub-Shot-Noise Regimes¶

Sketch.
Optical parametric oscillators and squeezing sources provide tunable bandwidth and noise floors relative to shot noise. Public spectra often report:

• squeezing level (dB below shot) across a frequency band,
• cavity or crystal parameters, and
• optical power and loss fractions.

An RFH-style view:

• \(B\): measurement or squeezing bandwidth (e.g., GHz-wide combs or THz waveguides),
• \(\Delta f/f\): amplitude noise relative to shot or a discretization threshold,
• \(Z\): pump power, cavity linewidths, detection efficiency.

Mode averaging typically yields \(\Delta f \propto B^{-1/2}\) in simple models (\(\alpha = 0.5\)); real systems are bounded by loss and technical noise. Exploratory fits in published squeezing spectra are consistent with:

• \(\alpha\) clustering near \(0 \lesssim \alpha \lesssim 0.5\) over the useful band,
• effective flattening where cavity or detection limits dominate, and
• no evidence of arbitrarily steep \(\alpha\) at modest energy cost.

Again, these are consistency checks, not precision measurements. They demonstrate how CCT's bandwidth and RFH notions can be mapped onto quantum-optical platforms.

H.2b Hybrid interferometer (displaced counting → homodyne): observer-slider template¶

• Bandwidth proxy: B := \(\dot{\mathcal{I}}\) (FI/sec) not 1/T_bin by default
• Discreteness proxy: pre-registered "click mass fraction" + "continuous record variance" (or your chosen pair)
• Confounders: LO phase noise, mode mismatch, thresholding/dead time
• Expected: smooth interpolation in measurement record; RFH-PL or RFH-QF-style knees depending on resonance/memory

Observed in simulation:

• LO displacement is the clean knob for the click↔continuous slider.
• \(B\) should be FI/sec (or proxy), not \(1/T_{\text{bin}}\), if you want scaling to mean anything.
• The wedge criterion should be curve-shape / knee behavior, not "slope must → −1".

H.2c Small-Source Quantum Optics RFH Pilot (DS3)¶

Context.
We ran a lightweight RFH pilot on public source data from a thin-film LiNbO\(_3\) squeezing experiment (DOI 10.11583/DTU.25845730.v1). This is an exploratory consistency check, not a full bandwidth-sweep proof, because RBW is fixed in the dataset.

Bandwidth and discreteness proxies.
$$ B = f_{\text{sideband}} \; (\text{MHz}) $$ $$ \log_{10}\Delta = -\alpha \log_{10}B + c $$ with two \(\Delta\) definitions from each trace (relative to shot-noise mean):

• \(\Delta_{\text{sqz}} = 1 - 10^{(\text{low-level}/10)}\)
• \(\Delta_{\text{anti}} = 10^{(\text{high-level}/10)} - 1\)

Results.

• Default full-range run (5 to 325 MHz, extrema metrics):
• \(\alpha_{\text{sqz}} \approx 0.173\), bootstrap 95% CI \([0.127, 0.298]\), \(R^2=0.638\)
• \(\alpha_{\text{anti}} \approx 0.200\), bootstrap 95% CI \([0.134, 0.359]\), \(R^2=0.585\)
• Robust mid-band run (55 to 200 MHz, quantile metrics \(q_{\text{low}}=0.05, q_{\text{high}}=0.95\)):
• \(\alpha_{\text{sqz}} \approx 0.530\), bootstrap 95% CI \([0.332, 0.740]\), \(R^2=0.745\)
• \(\alpha_{\text{anti}} \approx 0.396\), bootstrap 95% CI \([0.142, 0.708]\), \(R^2=0.576\)

Interpretation.
The DS3 pilot shows positive RFH-style scaling. Under robust mid-band settings, the squeezing-depth metric is near the incoherent \(\alpha \sim 0.5\) class. Because RBW is fixed and sideband frequency is used as a proxy for \(B\), this remains exploratory and should not be treated as a firm regime label.

Reproducibility.

• Runner: analysis/rfh_qopt/run_rfh_qopt_ds3.py
• Data: analysis/rfh_qopt/data/DS3/
• Full-range output: analysis/rfh_qopt/out/20260206T221130Z_rfh_qopt_ds3/
• Mid-band robust output: analysis/rfh_qopt/out/20260206T225529Z_rfh_qopt_ds3/

H.3 LIGO Case Study: Matched-Filter RFH Probe on GW150914¶

Context.
Interferometric gravitational-wave detectors (LIGO/Virgo/KAGRA) are among the best-characterized measurement systems we have: high sampling rates, detailed calibration, and well-understood noise budgets. This makes them a natural testbed for an RFH-style bandwidth–discreteness probe.

Setup (core regime).
Using public LOSC data for the GW150914 event, we considered:

• detectors: H1 and L1,
• data: 32 s strain files at \(f_s = 4096\ \text{Hz}\),
• segments: seven off-source 8 s windows per detector (offsets 0, 4, …, 24 s),
• band of interest: 80–300 Hz,
• test frequency: \(f_0 = 150\ \text{Hz}\) inside the band.

For each segment and window length \(N \in \{256, 512, 1024, 2048, 4096, 8192\}\), we defined:

• Effective bandwidth in the 80–300 Hz band as
  $$ B(N) = \frac{(300 - 80)\,N}{f_s}, $$ i.e. the number of independent frequency bins resolved in that band for a given \(N\).
• Matched-filter amplitude \(\hat A\) at \(f_0\) as the (normalized) projection of the strain onto \(\sin(2\pi f_0 t)\) and \(\cos(2\pi f_0 t)\) over the window, combined as \(\hat A = \sqrt{a_{\sin}^2 + a_{\cos}^2}\).
• Noise level \(\sigma_A(B)\) as the sample standard deviation of \(\hat A\) across sliding windows of length \(N\) within a segment.
• Detection threshold \(A_{\text{thr}}(B) = \rho\,\sigma_A(B)\) with \(\rho = 5\) (a 5σ criterion).

Normalizing by the maximum threshold amplitude across all points (\(A_{\text{ref}}\)), we defined $$ \Delta(B) = \frac{A_{\text{thr}}(B)}{A_{\text{ref}}}, $$ and treated \(\Delta(B)\) as the discreteness proxy.

Result (80–300 Hz band).
Pooling H1 and L1, we obtained \(\sim 80\) \((B, \Delta)\) points. A log–log regression of \(\log_{10} \Delta\) versus \(\log_{10} B\) yielded $$ \Delta(B) \propto B^{-\alpha_{\text{GW}}},\qquad \alpha_{\text{GW}} \approx 0.99 \pm 0.03, $$ with \(R^2 \approx 0.9\). A mixed-effects model with random intercepts per segment gave essentially the same slope, indicating modest segment-to-segment variation in noise floor but a stable bandwidth exponent.

Interpretation.
In this specific regime—off-source GW150914 noise, 80–300 Hz band, matched-filter amplitude at a fixed \(f_0\)—LIGO behaves as an RFH-typical compiler:

• giving the observer more effective bandwidth in the band (longer coherent integration) reduces the minimal detectable line amplitude roughly as \(1/B\),
• discreteness (“is there a resolvable line at \(f_0\)?”) is set by a bandwidth-dependent threshold, not by a fixed microscopic grain.

This is fully compatible with standard detection theory under Gaussian noise; it does not claim new physics. Its value for CCT is methodological: it shows precisely how to instantiate RFH quantities \((B, \Delta, \alpha)\) on a real instrument and yields a concrete, physically meaningful estimate \(\alpha_{\text{GW,regime}} \approx 1\) in one clean regime.

Multi-band extension.
Running the same matched-filter RFH probe on the same 32 s GW150914 files across three distinct bands,

• low: 40–80 Hz with \(f_0 = 60\ \text{Hz}\),
• mid: 80–300 Hz with \(f_0 = 150\ \text{Hz}\),
• high: 300–600 Hz with \(f_0 = 400\ \text{Hz}\),

and pooling H1+L1 in each case, yields:

Band	\(f_0\) (Hz)	\(\alpha\)	Std. err.	\(R^2\)	\(n\) (points)
40–80 Hz	60	1.09	0.03	0.94	84
80–300 Hz	150	0.98	0.03	0.94	84
300–600 Hz	400	1.12	0.02	0.97	84

Across these three GW bands, the RFH relation is very well described by $$ \Delta(B) \propto B^{-\alpha}, \qquad \alpha_{\text{GW}} \approx 1.0 \pm 0.1, $$ so LIGO behaves as a coherent RFH compiler with a stable, regime-level exponent near 1 rather than a single tuned case.

H.4 Regime Classification and the \(\alpha\) Exponent¶

The empirical case studies in this appendix (ADCs, optical squeezing, LIGO) exhibit different RFH exponents \(\alpha\), ranging from \(\alpha \approx 0.3\) to \(\alpha \approx 1\). This is not a contradiction but a reflection of regime-dependent scaling, a core feature of CCT's framework.

H.4.1 The Apparent Discrepancy¶

Two key results appear to conflict:

1. Baby Theorems 1 & 8 (Appendix C): Predict \(\alpha = 1/2\) for parameter estimation under back-action or quantum shot noise.
2. LIGO Case Study (§H.3): Finds \(\alpha_{\text{GW}} \approx 0.99 \pm 0.03\) in matched-filter detection.

H.4.2 Resolution: Incoherent vs Coherent Regimes¶

These represent two distinct universality classes of the RFH, governed by whether information accumulation is incoherent (averaging) or coherent (Fourier/phase).

Regime A: Incoherent Averaging (\(\alpha = 1/2\))

Applies to: Baby Theorem 1, Baby Theorem 8 (Standard Quantum Limit), shot-noise-limited sensors.

The observer estimates a static parameter by averaging \(N\) independent, noisy probes: - Signal: Accumulates linearly (if summing) or stays constant (if averaging). - Noise: Accumulates as \(\sqrt{N}\) (random walk). - SNR: Scales as \(\sqrt{N}\). - Resolution: \(\Delta \propto 1/\sqrt{N}\).

RFH Exponent: $$ \Delta \propto B^{-1/2} \implies \alpha = 1/2. $$

Examples: - Quantum position measurement (Heisenberg uncertainty, BT8). - Polling to estimate a population mean. - Photon shot noise in imaging.

Regime B: Coherent / Fourier Scaling (\(\alpha = 1\))

Applies to: LIGO (§H.3), frequency counters, atomic clocks, time-of-flight measurements.

The observer measures a time-varying signal (wave, oscillation) or geometric interval where the "signal" is encoded in phase or periodicity: - Mechanism: Coherent integration over time \(T\). - Bandwidth Resource: \(B \propto T\) (observation window). - Resolution: Governed by the Fourier Uncertainty Principle (\(\Delta f \cdot T \ge 1\)). - Scaling: Frequency resolution \(\Delta f \propto 1/T\).

RFH Exponent: $$ \Delta f \propto T^{-1} \propto B^{-1} \implies \alpha = 1. $$

Why LIGO shows \(\alpha \approx 1\): The analysis in §H.3 defined "Effective Bandwidth" \(B\) as the number of time-domain samples \(N\) used for the matched filter. The matched filter coherently integrates the waveform phase over the window: - Longer window \(N\) → narrower frequency bin → better rejection of noise outside the signal band. - This acts like a Fourier transform: the "discreteness" (ability to distinguish signal from noise floor) improves linearly with time/bandwidth.

H.4.3 The Unified RFH Landscape¶

CCT does not predict a single universal \(\alpha\). It predicts that \(\alpha\) is a regime-dependent invariant determined by the system's coherence class:

Regime	Mechanism	Scaling Law	RFH \(\alpha\)	Example
Incoherent	Statistical Averaging	\(1/\sqrt{N}\)	0.5	Quantum Position (BT8), Shot Noise
Coherent	Fourier / Phase Accumulation	\(1/N\)	1.0	LIGO (§H.3), Atomic Clocks, Freq Counters
Super-Coherent	Entanglement / Squeezing	\(1/N\) (Heisenberg Limit)	1.0†	Quantum Metrology (Giovannetti et al.)
Back-Action Saturated	Strong Feedback	\(1/N^0\) (Constant)	0.0	BT1 (Large \(B\) limit)

† Requires global correlations; under diffusion + loss, observed scaling may reduce to Regime-A-like behavior with prefactor/knee changes rather than a stable asymptotic exponent shift.

H.4.4 Falsifiability and Regime Stability¶

The falsifiable claim of CCT is not "\(\alpha\) is always 1" or "\(\alpha\) is always 1/2".

It is: "For a given architecture class, \(\alpha\) is a stable invariant that does not drift arbitrarily."

Tests: - If a Regime A system (shot noise) yields \(\alpha=1\) persistently, CCT is falsified (or squeezing was introduced). - If a Regime B system (LIGO) yields \(\alpha=0.5\), coherence has been lost (decoherence noise). - If \(\alpha\) varies wildly within a single platform under fixed conditions, the RFH framework does not apply.

H.4.5 Implications¶

• Baby Theorems 1 & 8 are correct for the incoherent/SQL regime (\(\alpha=1/2\)).
• LIGO data is correct for the coherent/Fourier regime (\(\alpha=1\)).
• CCT covers both, identifying them as the two fundamental "universality classes" of observational scaling.

This regime classification resolves the apparent discrepancy and provides a template for analyzing future RFH case studies across diverse platforms.

---

H.5 Camera Case Study: RFH Probe on Imaging Data¶

Context.
We performed an RFH-style bandwidth–discreteness probe on developed output from a Nikon D810, using a central \(1024 \times 1024\) grayscale crop from an 8-bit TIFF (derived from a 14-bit lossless-compressed NEF) as a noise reservoir. The aim was to test whether a simple, standard spatial detection task on real imaging data also produces a clean power-law relation between spatial bandwidth \(B\) and minimal detectable contrast \(\Delta\).

Setup (i.i.d. camera-noise regime).
Square patches of side length \(N \in \{16, 24, 32, 48, 64, 96, 128, 192\}\) were drawn from the crop. For each \(N\), we defined the effective spatial bandwidth as $$ B = N^2, $$ the number of spatial samples (pixels) in the patch. For each \(B\), we synthesized 400 noise-only patches by drawing \(N^2\) pixels with replacement from the crop’s 1D intensity histogram, reshaping to \(N \times N\), and computing a matched-filter amplitude \(\hat A\) for a fixed sinusoidal grating with four cycles across the patch, $$ p(x,y) = \sin!\left(2\pi \frac{4x}{N}\right). $$ The discreteness proxy was the minimal detectable contrast at that spatial frequency, taken as $$ \Delta(B) = A_{\text{thr}}(B) = 5\,\sigma_A(B), $$ where \(\sigma_A(B)\) is the sample standard deviation of \(\hat A\) over the 400 noise realizations (a 5σ threshold).

Result.
Over bandwidths \(B\) ranging from 256 to 36 864, the minimal detectable contrast \(\Delta(B)\) fell from \(\approx 6.5\%\) to \(\approx 0.55\%\) of full-scale intensity. An ordinary least-squares fit of \(\log_{10} \Delta\) versus \(\log_{10} B\) yielded $$ \Delta(B) \propto B^{-\alpha_{\text{img}}}, \qquad \alpha_{\text{img}} \approx 0.50 \pm 0.01, $$ with \(R^2 \approx 0.998\). Within this regime the camera behaves like a textbook averaging detector: SNR grows as \(\sqrt{B}\) and the minimal detectable amplitude shrinks as \(B^{-1/2}\).

Real-patch variant.
A stricter variant replaces the synthetic i.i.d. patches with actual non-overlapping patches from the real \(1024 \times 1024\) crop and repeats the same matched-filter procedure for a small set of spatial frequencies (for example, 2, 4, and 8 cycles across the patch). In this case the statistic \(\hat A\) probes a mixture of sensor noise and scene structure (edges, textures, aligned/anti-aligned patterns). The resulting 5σ thresholds become less smoothly monotone in \(B\), and simple power-law fits can yield small or even sign-flipped apparent exponents. This is expected: once rich scene statistics are included, the RFH probe is no longer isolating a pure “sensor + noise” regime; it is partly measuring the structure of the environment relative to the chosen pattern.

Interpretation.
The i.i.d. camera-noise case realizes RFH in an incoherent averaging regime: increasing spatial bandwidth by adding pixels in the patch lowers the detection threshold as \(1/\sqrt{B}\), consistent with the Regime A class (\(\alpha = 1/2\)) in §H.4. Together with the coherent LIGO regime (§H.3), where \(\alpha_{\text{GW}} \approx 1\), this provides a cross-domain pair of real-world RFH probes—time-series interferometry and spatial imaging—with distinct but stable exponents set primarily by the observer architecture and coherence class. The real-patch variant highlights a limitation: RFH exponents inferred from mixed sensor+scene regimes must be read as properties of the observer–environment pair, not of the instrument alone.

---

H.6 RobotCar Radar Odometry: Coherent Navigation Regime¶

Context.
Automotive and robotics platforms increasingly rely on FMCW radar odometry fused with inertial navigation. This provides a natural RFH testbed in an everyday setting: bandwidth is the number of consecutive radar steps coherently integrated, and discreteness is the relative yaw-change error compared to a higher-rate inertial (INS) reference.

Setup.
Using a Navtech CTS350-X FMCW radar traversal with ≈2200 steps (≈9.4 minutes at ≈4 Hz), and a synchronized INS pose stream, we define for each window of \(K\) consecutive radar increments: $$ \Delta\theta_\mathrm{radar}^{(i)}(K) = \sum_{j=i}^{i+K-1} \Delta\theta_j,\qquad \Delta\theta_\mathrm{INS}^{(i)}(K) = \theta_\mathrm{INS}(t_\text{end}) - \theta_\mathrm{INS}(t_\text{start}), $$ with \(t_\text{start}\) and \(t_\text{end}\) the timestamps of the first and last radar step in the window. The effective bandwidth is proportional to the number of integrated steps, $$ B \propto K,\qquad K \in {4, 8, 16, 32, 64, 128, 256}, $$ corresponding to integration times from \(\mathcal{O}(1\ \mathrm{s})\) to \(\mathcal{O}(1\ \mathrm{min})\).

The discreteness proxy is the dimensionless RMS relative yaw-change error, $$ \left.\frac{\Delta f}{f}\right|{B(K)} = \frac{\sqrt{\big\langle(\Delta\theta\mathrm{radar}^{(i)}(K) - \Delta\theta_\mathrm{INS}^{(i)}(K))^2\big\rangle_i}} {\sqrt{\big\langle(\Delta\theta_\mathrm{INS}^{(i)}(K))^2\big\rangle_i}}, $$ averaged over non-overlapping windows \(i\).

Result.
As \(K\) grows from 4 to 256, \(\Delta f/f\) drops from \(\mathcal{O}(1)\) to \(\sim 10^{-2}\). A log–log fit of the RFH form $$ \log_{10}!\left(\frac{\Delta f}{f}\right) = -\alpha \log_{10} B + \mathrm{const.},\quad B \propto K, $$ over this range yields $$ \alpha_\mathrm{RobotCar} \approx 0.99, $$ consistent with a coherent, phase-like RFH regime: doubling the integration time roughly halves the relative yaw-change error. This mirrors the LIGO behaviour but in a terrestrial navigation stack.

---

H.7 ECG Beat Averaging: Incoherent Physiological Regime¶

Context.
Electrocardiogram (ECG) analysis often relies on beat-averaged templates to reduce noise and highlight morphology. This provides a physiological analogue of the camera patch-averaging regime: bandwidth is the number of beats averaged, and discreteness is the relative error in reconstructing the morphology template.

Setup.
Using a standard record from the MIT–BIH arrhythmia database (e.g., record 100 at 360 Hz), R-peaks are detected with a simple amplitude + local-maximum rule and refractory period, yielding \(\gtrsim 2000\) beats. For each beat \(b\) at index \(n_b\), we extract a fixed-length beat-centred segment \(\mathbf{s}_b \in \mathbb{R}^L\) spanning \(L\) samples around the R-peak (e.g. 0.6 s total, 216 samples).

The global morphology template is $$ \mathbf{s}\ast = \frac{1}{N\mathrm{beats}} \sum_{b=1}^{N_\mathrm{beats}} \mathbf{s}b, $$ with RMS amplitude $$ f = \sqrt{\frac{1}{L}\sum. $$ Effective bandwidth is taken as the number of beats averaged, $$ B \propto N, $$ and for each }^{L} s_{\ast,j}^2\(N\) we repeatedly draw \(N\) beats, form their mean waveform \(\mathbf{s}_{(N)}\), and compute the RMS deviation $$ \delta(N) = \sqrt{\frac{1}{L}\sum_{j=1}^{L} (s_{(N),j} - s_{\ast,j})^2}. $$ The discreteness proxy is the dimensionless morphology error $$ \frac{\Delta f}{f}(N) = \frac{\langle \delta(N)\rangle}{f}, \quad B \propto N, $$ where \(\langle\cdot\rangle\) averages over draws.

Result.
Over \(N \in \{1,2,4,8,16,32,64,128,256,512,1024\}\), \(\Delta f/f\) decays slowly: increasing \(N\) by a factor of 16 reduces \(\Delta f/f\) only by a factor of a few. A power-law fit $$ \log_{10}!\left(\frac{\Delta f}{f}\right) = -\alpha \log_{10} B + \mathrm{const.}, $$ over this range yields $$ \alpha_\mathrm{ECG} \approx 0.5\text{–}0.55, $$ consistent with the incoherent averaging regime where uncertainties fall as \(1/\sqrt{B}\). This parallels the camera RAW patch-averaging behaviour, but now in a physiological time-series: beat-averaged morphology improves only with the square root of the number of beats.

---

H.8 Pulsar Timing: Super-Coherent Spin Estimation¶

Context.
Millisecond pulsar timing provides an extreme radio analogue of coherent measurements. Long-term phase-coherent fits to time-of-arrival (TOA) data yield exquisitely precise spin-frequency estimates, making this an ideal RFH probe in a super-coherent regime.

Setup.
Using public NANOGrav 12.5-year data for PSR J1713+0747, we take:

• a timing model (.par) with spin frequency \(F_0 \approx 218.8118437865\ \mathrm{Hz}\) and reference epoch \(\mathrm{PEPOCH} = 55662\ \mathrm{MJD}\),
• a wideband TOA file (.tim) with 1012 barycentric arrival times \(\{t_i\}\) over \(\sim 12.5\) years.

Each TOA is converted to a time offset \(t_i'\) relative to PEPOCH and assigned an integer pulse label \(n_i\) using \(F_0\). For a baseline \(T\) (in years) centred on PEPOCH, we retain only TOAs with \(|t_i - \mathrm{PEPOCH}| \le T/2\) and treat $$ B \propto T, $$ which is also proportional to the number of TOAs \(N_\mathrm{TOA}\) in that window. In practice \(T \in \{0.5, 1, 2, 4, 8, 12\}\) years, corresponding to \(\{40, 74, 128, 252, 552, 953\}\) TOAs.

For each \(T\), we fit a linear timing model $$ n_i \approx \phi_0 + F_0^{(\mathrm{fit})}(T)\, t_i' $$ on the retained TOAs and extract the best-fit frequency \(F_0^{(\mathrm{fit})}(T)\) and its standard error \(\Delta F_0(T)\). The discreteness proxy is the relative spin-frequency uncertainty $$ \left.\frac{\Delta f}{f}\right|_{B(T)} = \frac{\Delta F_0(T)}{F_0}. $$

Over the baselines above this yields, for example, $$ \frac{\Delta f}{f} \sim 5.4\times 10^{-11}\ (T=0.5~\mathrm{yr}),\quad 1.8\times 10^{-11}\ (1~\mathrm{yr}),\quad 4.1\times 10^{-13}\ (12~\mathrm{yr}), $$ with a smooth monotonic decrease in between.

Result.
Fitting $$ \log_{10}!\left(\frac{\Delta f}{f}\right) = -\alpha \log_{10} B + \mathrm{const.},\qquad B \propto T, $$ to the six \((T,\Delta f/f)\) points gives $$ \alpha_\mathrm{pulsar} \approx 1.5. $$ In this simple spin-frequency fit, the relative frequency uncertainty improves roughly as \(\Delta f/f \propto T^{-1.5}\), significantly steeper than both the incoherent \(1/\sqrt{B}\) regime (\(\alpha \approx 1/2\)) and the LIGO/RobotCar-style \(\alpha \approx 1\) behaviour. This is consistent with long-term super-coherent phase integration on a very stable clock. At longer baselines, or with more complete timing models including red noise (spin noise, propagation, or a stochastic gravitational-wave background), one expects this simple scaling to break down, making high-precision pulsar timing a natural stress-test for RFH in the presence of long-timescale correlations.

---

H.8b Paleomagnetic Excursions: RFH Pilot (Laschamp + Mono Lake)¶

Context.
Geomagnetic excursions are transient departures of Earth's magnetic field from the axial dipole state, recorded in sediment magnetization. Resolution varies by sedimentation rate and sampling density. This provides a natural RFH testbed in a geophysical domain: bandwidth is the effective temporal resolution of the paleomagnetic record, and discreteness is the apparent duration of the excursion as resolved by that record.

We tested RFH using sedimentary and lacustrine records with a standardized directional definition (excursion interval via VGP latitude threshold). We define: $$ B = 1000 / \text{smoothing time (yr)}, \qquad \Delta = \text{apparent directional duration (yr)}. $$ Fits are performed in log–log space.

Data.
Records were compiled from the GGF100k database (Panovska et al., 2018) and related literature, filtered to those covering the target excursion with explicit resolution metadata.

Methods (uncertainty-aware).

• OLS on center values.
• ODR / errors-in-variables using uncertainty in both axes.
• Interval Monte Carlo (20,000 draws from low/high bounds).
• Leave-one-out sensitivity and Cook's distance.
• Basin block-bootstrap for non-independence.

Results.

Event	N	ODR \(\alpha\) [95% CI]	MC \(\alpha\) [95% CI]	OLS \(\alpha\) [95% CI]	Status
Laschamp	15	0.248 [0.010, 0.486]	0.146 [0.076, 0.220]	0.151 [-0.131, 0.433]	Pilot
Mono Lake	8	0.746 [0.201, 1.290]	0.526 [0.230, 0.869]	0.635 [-0.081, 1.351]	Pilot

Interpretation.
Both events show positive bandwidth-discreteness scaling, consistent with RFH directionally. For Laschamp, interval Monte Carlo places \(\alpha\) mostly below the predeclared \([0.3,0.7]\) band, while ODR overlaps that band only partially. Mono Lake remains compatible with a broad incoherent-to-coherent transition but with wide uncertainty. Under predeclared regime-call rules (95% CI fully contained within a target band and stable under sensitivity checks), these pilots do not yet support a firm regime label.

Sensitivity.
Leave-one-out on Laschamp is narrow (\(\alpha \in [0.081, 0.195]\)), indicating no single-record sign flip. Influence diagnostics highlight leverage records (ODP_1061B_1062, p226, Black_Sea_Stack), and basin bootstrap widens CIs, consistent with non-independence across regional stacks.

Reproducibility.

• Data: analysis/rfh_paleomag/data/paleomag_records.csv
• Runner: analysis/rfh_paleomag/run_rfh_paleomag.py
• Output: analysis/rfh_paleomag/out/20260206T202954Z_rfh_paleomag/

---

H.8c Economic Time-Series RFH Pilot (BTCUSDT + ETHUSDT, 2025)¶

Context.
Financial time series are not precision-physics instruments, but they are high-throughput observer pipelines with explicit bandwidth choices (sampling and aggregation). This makes them useful as a cross-domain RFH stress test: does apparent discreteness scale predictably with measurement bandwidth in a noisy, non-laboratory domain?

We tested two high-liquidity instruments (BTCUSDT, ETHUSDT) using full-year spot data from Binance (2025, 1-minute klines). The goal is not to claim a universal market exponent, but to test whether RFH-style scaling survives in an economic signal environment.

Setup and definitions.

• Base data: monthly Binance spot 1m kline CSVs, Jan-Dec 2025.
• Effective bandwidth: $$ B(\Delta t) = \frac{1440}{\Delta t}, $$ where \(\Delta t\) is aggregation interval in minutes.
• Intervals tested: $$ \Delta t \in {1,3,5,10,15,30,60,120,240}\ \text{minutes}. $$

We used two discreteness proxies, both defined as distortion relative to the 1-minute baseline:

1. Realized-volatility distortion $$ \Delta_{\text{RV}}(\Delta t)=\operatorname{median}{d}\left|\log\frac{\mathrm{RV}\right|, $$ where }}{\mathrm{RV}_{d,1m}\(\mathrm{RV}_{d,\Delta t}=\sum r_{d,\Delta t}^2\) on day \(d\).

2. Tail-distortion (99th percentile) $$ \Delta_{\text{tail}}(\Delta t)=\operatorname{median}{d}\left|\log\frac{Q99\right|, $$ where }}{Q99_{d,1m}\(Q99\) is the 99th percentile of \(|r|\) on day \(d\).

Fits use: $$ \log_{10}\Delta = -\alpha\log_{10}B + c. $$ Because \(\Delta(1m)=0\) by construction, the log fit uses the remaining 8 bandwidth points.

Methods and robustness checks.

• OLS fit in log-log space for each metric/instrument pair.
• Day-level bootstrap (3000 resamples) for \(\alpha\) 95% CI.
• Leave-one-interval sensitivity (drop one \(\Delta t\) at a time).
• Cross-instrument replication (BTCUSDT vs ETHUSDT) under identical pipeline/config.

Results (bootstrap CI primary).

Symbol	Metric	\(\hat\alpha\)	95% CI (bootstrap)	\(R^2\)	Regime call
BTCUSDT	\(\Delta_{\text{RV}}\)	0.521	[0.478, 0.550]	0.996	sub_incoherent \([0.3,0.7]\)
BTCUSDT	\(\Delta_{\text{tail}}\)	0.284	[0.275, 0.292]	0.888	below predeclared band
ETHUSDT	\(\Delta_{\text{RV}}\)	0.521	[0.486, 0.564]	0.993	sub_incoherent \([0.3,0.7]\)
ETHUSDT	\(\Delta_{\text{tail}}\)	0.282	[0.275, 0.292]	0.882	below predeclared band

Sensitivity and stability.

• RV metric leave-one-interval \(\alpha\) ranges:
• BTCUSDT: \([0.502, 0.531]\)
• ETHUSDT: \([0.491, 0.532]\)
• Tail metric leave-one-interval \(\alpha\) ranges:
• BTCUSDT: \([0.233, 0.333]\)
• ETHUSDT: \([0.228, 0.333]\)

This indicates the RV result is stable and reproducible across both instruments, while the tail metric is weaker and remains outside predeclared regime bands.

The RV exponent is nearly identical across instruments (\(\alpha_{\text{RV}}=0.521\) for both BTCUSDT and ETHUSDT in this run), suggesting the effect is not purely one-asset idiosyncratic in this sample, while shared venue/time confounders remain.

Interpretation.

This pilot supports a new-domain RFH statement in limited form:

• A robust positive scaling relation exists for RV distortion (\(\alpha\approx 0.52\)), consistent with a sub-incoherent regime that sits close to the classical incoherent \(\alpha\approx 0.5\) boundary.
• Tail distortion scales more weakly (\(\alpha\approx 0.28\)), suggesting saturation/heavy-tail effects and no firm regime-label call under predeclared criteria.

So this is a useful RFH portability check in economics, but it is complementary to controlled physics probes (not a replacement for them).

Limitations.

• Single venue (Binance spot), two symbols, one calendar year.
• No explicit order-book/quote microstructure controls.
• No causal intervention; this is an observational scaling test.
• Regime segmentation (volatility states, session effects, event windows) is not yet modeled.

Reproducibility.

• Runner: analysis/rfh_econ/run_rfh_econ.py
• Data folder: analysis/rfh_econ/data/
• BTC output: analysis/rfh_econ/out/20260206T211049Z_rfh_econ_btcusdt_2025/
• ETH output: analysis/rfh_econ/out/20260206T212616Z_rfh_econ_ethusdt_2025/

---

H.B1 Bioelectric Morphogenesis I: Gap-Junction Connectivity and Head Identity (Emmons-Bell / Levin)¶

These H.Bx entries are order-of-magnitude RFH hints in morphogenetic control systems. Bandwidth and discreteness are inferred from coarse bioelectric connectivity and morphology proxies, not from full multi-level sweeps, so they should be read as conceptual bridges rather than precise log–log fits.

Context.
Planarian flatworms regenerate heads with high anatomical fidelity. Emmons-Bell et al. showed that transient gap-junction blockade with octanol in Girardia dorotocephala can make decapitated worms regenerate heads with the morphology of other species (D. japonica, S. mediterranea, P. felina), all from the same genome. They used morphometric analysis (Procrustes distances) to quantify head shape differences, and voltage imaging to show that octanol increases the number of bioelectric “domains” (isopotential regions) during regeneration.

CCT treats the head-regenerating tissue as a bandwidth-limited observer–controller: a bioelectric network maintaining a target head attractor under injury. In this view, gap-junction connectivity defines the size of the tissue-level “self” (cognitive light cone) over which morphology is controlled.

Setup.

• Bandwidth \(B\): effective gap-junction connectivity.
  Emmons-Bell et al. used DiBAC voltage imaging to count isopotential regions across the head. Octanol-treated “pseudo” worms show more such regions (more fragmented connectivity) than wild-type controls, and this number relaxes back toward wild-type as the morphology slowly remodels to the canonical G. dorotocephala head. We take $$ B \propto \frac{1}{N_\text{domains}}, $$ so fewer domains → better electrical percolation → higher effective bandwidth.

• Discreteness proxy \(\Delta f/f\): morphological head-shape error.
  The paper reports Procrustes distances between group-mean head shapes for each condition (WT G. dorotocephala vs pseudo D. japonica, pseudo S. mediterranea, pseudo P. felina, etc.). For each condition \(c\), define $$ \frac{\Delta f}{f}(c) = d_\text{Proc}(\text{shape}_c,\ \text{WT-GD}), $$ a dimensionless shape error (0 = perfect WT; larger = more distorted). Representative values are on the order of:

• WT GD vs WT GD regeneration: distance \(\sim 0.1\) (baseline error),
• GD vs pseudo-D. japonica: \(\sim 0.19\),
• GD vs pseudo-S. mediterranea: \(\sim 0.21\),

• GD vs pseudo-P. felina: \(\sim 0.38\).

• Confounders \(Z\).
  Species background (fixed to G. dorotocephala), amputation plane, imaging pipeline, octanol vs hexanol control, soak schedule, and time since amputation.

Result.
Comparing:

• High-bandwidth condition (WT GD) – intact gap junctions, few isopotential regions, WT head after 10 days: $$ (\Delta f/f)_\text{high} \approx 0.10. $$
• Low-bandwidth condition (octanol pseudo-DJ) – GJ blocked, more isopotential regions, DJ-like head: $$ (\Delta f/f)_\text{low} \approx 0.19. $$

Morphology error nearly doubles when gap-junction connectivity is transiently disrupted: $$ \frac{(\Delta f/f)\text{low}}{(\Delta f/f)\text{high}} \approx 1.9. $$

Let \(B_\text{high}\) and \(B_\text{low}\) be effective bandwidths inferred from domain counts and \(r = B_\text{low}/B_\text{high}<1\). An RFH-style law $$ \frac{\Delta f}{f} \propto B^{-\alpha} $$ implies $$ 1.9 \approx r^{-\alpha} \quad\Rightarrow\quad \alpha = \frac{\ln 1.9}{\ln(1/r)}. $$

For plausible connectivity drops (e.g. octanol doubling or quadrupling the domain count so \(r\sim 0.5\text{–}0.25\)), the implied \(\alpha\) lies in roughly the 0.5–1 band, similar to other RFH probes.

Interpretation.
Even with sparse points and proxy bandwidth, this system behaves like an RFH domain:

• Increased gap-junction bandwidth (few domains) → tissue-scale “self” spans more cells → head shape remains pinned to the WT attractor (low \(\Delta f/f\)).
• Reduced bandwidth (more domains) → the cognitive light cone fragments → the same genome samples other species’ head attractors with larger shape error.

This is a controller-type RFH probe: the bandwidth-limited entity is a morphogenetic policy rather than a passive sensor, but it still obeys “more bandwidth → less discretization error” in a way compatible with the RFH form used elsewhere in this appendix.

---

H.B2 Bioelectric Morphogenesis II: Planarian Head and Organ Scaling (Beane)¶

Context.
Beane et al. showed that membrane-voltage–dependent signaling controls head size and internal organ scaling during planarian regeneration. Knocking down the H,K-ATPase (hyperpolarizing the tissue) produces worms with shrunken heads and enlarged pharynges, while blastema size remains similar.

CCT reads this as a bandwidth-limited scaling controller: a bioelectric network that must coordinate growth across the body axis to maintain correct size ratios.

Setup.

• Bandwidth \(B\): H,K-ATPase-mediated bioelectric coherence.
  The H,K-ATPase participates in establishing a long-range \(V_\text{mem}\) pattern. In control worms, this pattern supports coordinated scaling; in H,K-ATPase RNAi worms, it is disrupted. We treat effective bandwidth as a decreasing function of this coherence: $$ B_\text{high} \propto \text{intact pump–mediated pattern},\quad B_\text{low} \propto \text{RNAi-disrupted pattern},\quad B_\text{low}<B_\text{high}. $$ (A more detailed RFH analysis would tie \(B\) to imaging-based measures of pattern smoothness or information rate in the \(V_\text{mem}\) field.)

• Discreteness proxy \(\Delta f/f\): head and pharynx scaling error.
  Beane et al. report head size and pharynx area as a percentage of total worm area at fixed days post-amputation. Controls show a characteristic head size and pharynx fraction; H,K-ATPase RNAi worms show shrunken heads and oversized pharynges. Define per-feature relative errors: $$ e_\text{head} = \frac{|H_\text{RNAi} - H_\text{WT}|}{H_\text{WT}},\quad e_\text{ph} = \frac{|P_\text{RNAi} - P_\text{WT}|}{P_\text{WT}}, $$ and an aggregate morphology error $$ \frac{\Delta f}{f} = \sqrt{\tfrac12(e_\text{head}^2 + e_\text{ph}^2)}. $$

Using the published percentages as representative values (WT head fraction \(\sim 20\%\), RNAi head \(\sim 7.5\%\); WT pharynx \(\sim 11\%\), RNAi pharynx \(\sim 16\%\)), the aggregate error is on the order of \(\Delta f/f \approx 0.5\text{–}0.6\) for H,K-ATPase RNAi worms, while controls cluster close to zero by construction.

• Confounders \(Z\).
  Genetic background, wound geometry, days post-amputation, pharmacology beyond the H,K-ATPase perturbation, and measurement noise in area estimates.

Result.
Under H,K-ATPase knockdown:

• The fraction of worms with an obvious “shrunken head” phenotype is reported at >90% in some conditions, versus ~0 in controls.
• The aggregate scaling error \(\Delta f/f\) jumps from near-baseline (controls) to ≈0.5–0.6 (RNAi), i.e. at least a several-fold increase in relative size distortion.

Treating RFH in the simple form $$ \frac{\Delta f}{f} \propto B^{-\alpha}, $$ this behaviour is consistent with a positive RFH exponent: reducing scaling bandwidth via pump disruption sharply increases the discreteness/quantization of size control. For plausible bandwidth ratios \(r = B_\text{low}/B_\text{high}\) in the 0.25–0.5 range, the implied \(\alpha\) again sits comfortably in the “RFH band” (\(\sim 0.5\text{–}1+\)) seen in other platforms, though the small number of bandwidth levels precludes a precise fit.

Interpretation.
Beane’s data realize an RFH probe on proportional morphogenesis:

• The H,K-ATPase-dependent bioelectric pattern plays the role of a finite-bandwidth controller distributing scale information along the worm.
• When that channel is intact (high \(B\)), head and organ sizes are computed as smoothly varying functions of body size; \(\Delta f/f\) stays small.
• When the channel is degraded (low \(B\)), the system snaps to coarse, mis-scaled attractors (tiny heads, oversized pharynges), increasing \(\Delta f/f\).

This complements H.B1: instead of “which head attractor do you land in?”, the RFH lens here measures “how finely can you control size ratios?” under a bandwidth constraint.

---

H.B3 Bioelectric Morphogenesis III: Xenopus Tail Regeneration Fidelity (Adams / Tseng)¶

Context.
Adams, Masi & Levin showed that H\(^+\) V-ATPase activity and associated \(V_\text{mem}\) changes are necessary and sufficient for tadpole tail regeneration in Xenopus laevis. Blocking the proton pump with concanamycin severely impairs regeneration; mis-staged tails that would normally be refractory can be forced to regenerate by inducing appropriate bioelectric changes. Tseng et al. extended this to transient Na\(^+\) currents, showing that activating specific Na\(_\text{v}\) channels can restore regeneration in refractory tails, again via bioelectric patterning.

These experiments provide a natural RFH testbed where the outcome is a scalar “regeneration quality” index rather than continuous shape.

Setup.

• Bandwidth \(B\): bioelectric control bandwidth in the tail bud.
  In the Adams assay, concanamycin inhibits the H\(^+\) V-ATPase, flattening or mis-patterning the \(V_\text{mem}\) field needed to drive growth. We treat bandwidth as proportional to the capacity of the bioelectric network to support the correct spatiotemporal \(V_\text{mem}\) pattern: $$ B_\text{high} \propto \text{normal V-ATPase activity},\quad B_\text{low} \propto \text{concanamycin-blocked pump}, $$ with analogous roles for Na\(_\text{v}\)-mediated patterns in Tseng’s work.

• Discreteness proxy \(\Delta f/f\): normalized regeneration failure.
  Adams et al. define a regeneration index (RI) on a fixed scale (e.g. 0–300) that scores tail regrowth quality. In controls, tails amputated at a permissive stage achieve high RI (near full tails); with V-ATPase inhibition the RI drops sharply, corresponding to truncated or absent tails. Define $$ f = \frac{\text{RI}}{\text{RI}_\text{max}},\quad \frac{\Delta f}{f} = 1 - f, $$ i.e. the dimensionless missing fraction of full regeneration.

Representative values from Adams’ figures (illustrative; precise numbers require digitization) are on the order of:

• Control tails: RI \(\sim 200+\) on a 0–300 scale → \(f_\text{high}\approx 0.7\text{–}0.8\), \((\Delta f/f)_\text{high}\approx 0.2\text{–}0.3\).
• Concanamycin-treated: RI \(\sim 50\) → \(f_\text{low}\approx 0.15\text{–}0.2\), \((\Delta f/f)_\text{low}\approx 0.8\text{–}0.85\).

Thus $$ \frac{(\Delta f/f)\text{low}}{(\Delta f/f)\text{high}} \sim 3, $$ i.e. regeneration error triples when the pump-mediated bioelectric bandwidth is strongly reduced.

• Confounders \(Z\).
  Developmental stage at amputation, temperature, drug dose and exposure schedule, systemic health, and (for Tseng) the genetic construct or drug used to induce Na\(^+\) currents.

Result.
Assuming the RFH form $$ \frac{\Delta f}{f} \propto B^{-\alpha}, $$ and letting \(r = B_\text{low}/B_\text{high}<1\), the Adams data imply: $$ \frac{(\Delta f/f)\text{low}}{(\Delta f/f)\text{high}} \sim 3 \approx r^{-\alpha} \quad\Rightarrow\quad \alpha = \frac{\ln 3}{\ln(1/r)}. $$

For reasonable bandwidth drops:

• If concanamycin halves effective bandwidth (\(r\approx 0.5\)), then \(\alpha\approx 1.6\).
• If it cuts bandwidth to a quarter (\(r\approx 0.25\)), then \(\alpha\approx 0.8\).

In either case, \(\alpha\) is firmly > 0 and plausibly in the same broad band (≈0.5–1.5) seen in other RFH probes in this appendix. Tseng’s Na\(_\text{v}\) induction experiments add further points where increasing bioelectric bandwidth in refractory tails dramatically improves RI, consistent with the same monotone relationship.

Interpretation.
The Xenopus tail system realizes an RFH-style “all-or-something” morphogenetic controller:

• A finite-bandwidth bioelectric network in the tail bud must integrate injury signals and global state to decide whether to regrow a tail.
• When bandwidth is high enough (normal V-ATPase / appropriate Na\(^+\) currents), regeneration proceeds and RI is high (low \(\Delta f/f\)).
• When bandwidth collapses (drugs or genetic blocks), the controller fails; RI plummets and \(\Delta f/f\) jumps toward 1.

This is less finely graded than the planarian shape examples, but it shows that coarse “regrow or not” decisions in vertebrate appendage regeneration also behave like a bandwidth-limited RFH domain.

---

H.B4 Bioelectric Gap Junction Network: Quantitative RFH Sweep¶

Context.
The qualitative RFH hints in §H.B1–B3 rely on two-point comparisons (high vs low bandwidth). To obtain a quantitative RFH exponent, we ran a synthetic gap junction network simulation with systematically varied coupling strength.

Model.
A 1D chain of \(N = 100\) cells, each with membrane potential \(V_i\), coupled via gap junctions: $$ \frac{dV_i}{dt} = g_{\text{gap}} (V_{i-1} + V_{i+1} - 2V_i) + g_{\text{leak}} (V_{\text{rest}} - V_i) + \eta_i(t), $$ where \(g_{\text{gap}}\) is the gap junction conductance (our bandwidth proxy), \(g_{\text{leak}} = 0.1\) is the leak conductance restoring toward \(V_{\text{rest}} = -70\) mV, and \(\eta_i(t)\) is Gaussian noise with amplitude 0.01.

Setup.
- Bandwidth \(B\): Gap junction conductance \(g_{\text{gap}} \in \{0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, 5.0\}\).
- Discreteness proxy \(\Delta\): Coefficient of variation of steady-state voltage, $$ \Delta = \frac{\sigma_V}{|\bar{V}|}, $$ i.e., the normalized heterogeneity of the voltage pattern. Lower \(\Delta\) = more uniform = better pattern control. - Trials: 10 independent runs per \(g_{\text{gap}}\) value, averaged.

Result.
A log–log regression of \(\log_{10} \Delta\) versus \(\log_{10} B\) yielded: $$ \Delta(B) \propto B^{-\alpha_{\text{bio}}}, \qquad \alpha_{\text{bio}} = 0.35 \pm 0.02, $$ with \(R^2 = 0.98\). The relationship is highly linear in log–log space, confirming the RFH power-law form.

Interpretation.
This is a sub-incoherent regime (\(\alpha < 0.5\)), distinct from the standard incoherent averaging (\(\alpha = 0.5\)) seen in ECG and camera noise. The lower exponent suggests that:

1. Correlated noise: Cell-to-cell noise is not fully independent; gap junction coupling propagates fluctuations, reducing the benefit of more coupling.
2. Saturation effects: At high \(g_{\text{gap}}\), the system approaches a fully synchronized state where additional coupling has diminishing returns.
3. Bistable dynamics: Morphogenetic systems may have multiple attractors, preventing smooth \(1/\sqrt{B}\) improvement.

This result quantitatively extends the qualitative hints in §H.B1–B3 and provides the first fitted RFH exponent for a bioelectric morphogenesis model.

Code availability.
The simulation script is available at tools/run_bio_efh_synthetic.py. Results are saved to bio_efh_synthetic_results/.

---

Part II — Exploratory Extensions and Phase 3+ Signposts¶

H.8d Toy Hierarchy Generation and Under-Determination¶

This section collects an exploratory toy result previously carried in Appendix C. It is kept here because it is a useful Layer-3 signpost, but it is not part of the operational identification spine used for fitting and falsification.

Context.
Open Problem 0 asks: Can CCT-style rule-space dynamics necessarily produce Standard-Model-like structure (e.g., three generations, specific gauge symmetries), or can we prove it cannot?

This toy theorem provides a constructive existence result for generation-like hierarchies arising from information-geometric stability selection in rule-space, while simultaneously proving fundamental under-determination: the same CCT machinery admits infinitely many stable hierarchies, so additional physical input is required to single out our observed structure.

Model: Hierarchical Rule-Space with Stability Filtering

Consider a rule-space manifold \(\mathcal{R}\) parameterized by \(d\) coordinates \(R = (R_1, \dots, R_d)\), where each coordinate controls a "coupling strength" or "interaction parameter" in a toy dynamical system (e.g., effective masses, coupling constants, or decay rates).

Let the informational potential be: $$ S(R) = \sum_{i=1}^{d} \left[\frac{1}{2}a_i R_i^2 + \frac{1}{4}b_i R_i^4\right] + \sum_{i<j} c_{ij} R_i R_j, $$ with \(a_i > 0\), \(b_i > 0\) (double-well structure in each coordinate), and interaction terms \(c_{ij}\) that couple the wells.

This yields an information metric: $$ g_{ij}(R) = \partial_i \partial_j S(R) = (a_i + 3b_i R_i^2)\delta_{ij} + c_{ij}(1 - \delta_{ij}). $$

Stability selection via renormalization-group flow analog.
Define a gradient flow in rule-space: $$ \dot{R}_i = -\eta \, g^{ij}(R) \, \partial_j S(R) + \sqrt{2D} \, \xi_i(t), $$ where \(\eta\) is adaptation rate, \(D\) is diffusion/noise amplitude, and \(\xi_i(t)\) is white noise.

Generation hierarchy emergence:

1. Local minima as "generations."
   The multi-well structure of \(S(R)\) creates multiple locally stable equilibria \(\{R^{(1)}, R^{(2)}, R^{(3)}, \dots\}\). When \(d \ge 3\) and the coupling matrix \(c_{ij}\) has suitable structure, one can construct \(S(R)\) with exactly three primary basins of comparable depth but differing curvature.

2. Energy hierarchy via curvature.
   Let basin \(k\) have characteristic curvature eigenvalues \(\lambda_k^{(\min)}\) and \(\lambda_k^{(\max)}\). Define an effective "mass scale" for excitations in that basin: $$ m_k \propto \sqrt{\lambda_k^{(\text{avg})}}, $$ where \(\lambda_k^{(\text{avg})}\) is the geometric mean of metric eigenvalues at the local minimum.

3. Information-geometric hierarchy theorem (constructive):
   For appropriately chosen \((a_i, b_i, c_{ij})\) with \(d = 9\) (three families × three types), there exists a configuration such that:

4. \(S(R)\) has exactly three stable basins under the flow,
5. The basins have nested curvature scales: \(\lambda_1 : \lambda_2 : \lambda_3 \approx 1 : m_\mu^2/m_e^2 : m_\tau^2/m_e^2\),
6. Transitions between basins require energy jumps proportional to the basin separation in the metric distance: $$ \Delta E_{jk} \propto \int_{\text{geodesic}(j \to k)} \sqrt{g_{ij} \, \mathrm{d}R^i \mathrm{d}R^j}. $$

Proof sketch (existence).
Numerically construct the following: - Set \(d = 9\) (three "generations" × three "types": e-like, μ-like, τ-like). - Choose \(a_i\) to create three basins with centers at \(R^{(1)} \approx (1, 0, 0, \dots)\), \(R^{(2)} \approx (0, 1, 0, \dots)\), \(R^{(3)} \approx (0, 0, 1, \dots)\) (one-hot encoding). - Set \(b_i\) and \(c_{ij}\) to enforce: - Basin 1 has low curvature (light "mass"), - Basin 2 has medium curvature (\(\approx 200 \times\) larger), - Basin 3 has high curvature (\(\approx 3500 \times\) larger), matching rough electron–muon–tau mass ratios. - Verify numerically that gradient flow from random initial conditions falls into one of the three basins with probabilities determined by basin volume in the \(S(R)\)-weighted measure.

This establishes that generation-like hierarchies CAN emerge from CCT-style information-metric selection.

Anti-uniqueness theorem (under-determination).

However, we can also prove the following:

Claim.
For any integer \(n \ge 2\) and any set of \(n\) ratios \(\{\lambda_1, \lambda_2, \dots, \lambda_n\}\), there exists a rule-space potential \(S_n(R)\) with dimension \(d_n\) and coupling matrix \(c_{ij}^{(n)}\) such that: 1. \(S_n(R)\) has exactly \(n\) stable basins, 2. The basins have curvature ratios matching \(\{\lambda_1, \dots, \lambda_n\}\), 3. The gradient flow converges to the \(n\)-basin structure under the same noise and adaptation parameters \((\eta, D)\).

In particular: - \(n = 2\) yields a "two-generation" hierarchy, - \(n = 4\) yields a "four-generation" hierarchy, - \(n = 12\) produces structure resembling a 12-fold particle zoo.

Proof sketch (non-uniqueness).
For each \(n\), set \(d_n = 3n\) (arbitrary but sufficient). Construct \(S_n(R)\) as: $$ S_n(R) = \sum_{k=1}^{n} \left[\sum_{i \in G_k} \frac{1}{2}a_k R_i^2 + \frac{1}{4}b_k R_i^4\right] + \text{cross-coupling}, $$ where \(G_k\) partitions the coordinates into \(n\) groups of size 3. Set the curvature parameters: $$ a_k + 3b_k (R^{(k)})^2 = \lambda_k, $$ at the basin minima. By choosing appropriate inter-group couplings \(c_{ij}\), the basins remain isolated under typical noise levels \(D\).

Numerical implementation:

def construct_n_generation_potential(n, lambda_ratios):
    d = 3 * n
    a = np.ones(d)
    b = np.zeros(d)
    for k in range(n):
        idx = slice(3*k, 3*(k+1))
        # Set local curvature at basin k to match lambda_ratios[k]
        a[idx] = lambda_ratios[k] / 2
        b[idx] = lambda_ratios[k] / 4
    # Construct cross-coupling to isolate basins
    c = construct_isolation_matrix(n, d)
    return a, b, c

In simulations with \(n \in \{2, 3, 4, 5\}\), each configuration produces the expected number of stable generations, confirming that CCT does not uniquely determine \(n = 3\).

Implications for Open Problem 0.

This baby theorem establishes two complementary results:

1. Constructive possibility (positive direction):
   Rule-space dynamics with information-metric stability selection can produce hierarchical generation-like structure. The specific number of generations and their mass scales arise as attractors in rule-space geometry, not as arbitrary inputs. This shows CCT is at least compatible with generation hierarchies.

2. Fundamental under-determination (negative direction):
   The same CCT machinery admits infinitely many \((n, \{\lambda_k\})\) configurations. Additional constraints beyond the core CCT axioms (continuous rule-space, information metric \(g_{ij} = \partial_i \partial_j S\), feedback flow \(\dot{R} = F(R, I)\)) are required to select \(n = 3\) and the observed mass ratios.

Candidate additional constraints (open research directions):

To move from this toy result toward a resolution of Open Problem 0, CCT would need to incorporate:

• Topological quantum numbers (e.g., winding numbers, homotopy classes in \(\mathcal{R}\)) that restrict basin count,
• Thermodynamic selection under cosmic boundary conditions (e.g., maximizing entropy production or cosmological stability),
• Bootstrap consistency from cross-scale feedback: requiring that the \(n\)-basin structure at microscopic rule-space reproduces itself under coarse-graining to larger scales,
• Anthropic filtering (controversial but honest): observers capable of formulating CCT may only exist in universes with certain generation counts,
• Explicit connection to observed gauge symmetries: showing that \(SU(3) \times SU(2) \times U(1)\) emerges as the unique stabilizer group of certain rule-space flows (currently not demonstrated).

Status of Open Problem 0 after Baby Theorem 0:

Question	Answer
Can CCT produce generation hierarchies?	Yes (constructive proof in toy model)
Must CCT produce exactly 3 generations?	No (infinitely many \(n\) are consistent)
Can CCT derive Standard Model gauge structure?	Open (no current pathway from \(S(R), g_{ij}\) to \(SU(3) \times SU(2) \times U(1)\))
Is additional physics input required?	Yes (under-determination proven)

Recommendation:
Open Problem 0 should be refined into two sub-problems: - OP0a (Weak): Prove that certain classes of rule-space dynamics necessarily produce some hierarchical generation-like structure (this baby theorem provides a toy instance). - OP0b (Strong): Either derive the specific values \((n=3, \text{observed mass ratios, SM gauge group})\) from CCT axioms plus minimal additional constraints, or prove that CCT cannot achieve such specificity without importing large amounts of empirical data.

Current status: OP0a is provisionally solved in toy form; OP0b remains open.

---

H.8e Cross-Domain Portability Note: Meta-Programmability and AI Scaling¶

This note is kept as a portability analogy rather than part of the theorem burden in Appendix C.

Connection to AI Scaling Laws (Chinchilla).
The meta-programmability bound from Baby Theorem 4 in Appendix C, $$ \mathsf{Prog}_T^\star \sim E^{-1/2}, $$ aligns suggestively with the "compute-optimal" scaling laws observed in modern large language models (Hoffmann et al. 2022, "Chinchilla").

• Empirical observation: For a fixed compute budget \(C\), the optimal model size \(N_{\text{opt}}\) and training data \(D_{\text{opt}}\) both scale as \(C^{0.5}\).
• BT4 analogy: If we map energy \(E \to\) compute \(C\), and interpret "optimal allocation" as the split between architecture and data, BT4 predicts that the marginal benefit of reconfiguration energy decays as \(E^{-1/2}\), implying optimal resource allocation scales as \(\sqrt{E}\).
• Interpretive status: This is a cross-domain analogy, not a derivation. Its value is to suggest that meta-programmability bounds may have recognizable engineering echoes outside physics.

---

H.8f BT8 Extrapolation: Baseline Quantum Regime and Drift-Sensitive Constants¶

This section collects the more speculative extrapolation from Baby Theorem 8. The operational baseline remains in Appendix C; the material here should be read as Phase 3+ interpretation rather than part of the identification spine.

Bold Interpretation: ℏ as Baseline Regime Parameter

The fact that the Heisenberg product \(\Delta x \cdot \Delta p = \hbar\) remains saturated across all squeezing levels does not mean \(\hbar\) is fundamental in the CCT sense. Rather:

1. \(\hbar\) is the feedback-stabilized coupling strength in the current rule-space configuration. It characterizes the "quantum attractor" we inhabit.

2. BT8 establishes the detector for rule-space transitions. If we observe:

3. \(\alpha > 1\) persistently → we have exited the standard quantum regime,
4. \(\Delta x \cdot \Delta p \neq \hbar\) (Heisenberg-product anomaly) → \(\hbar_{\mathrm{eff}}\) has shifted,

5. \(\alpha < 0.5\) under conditions where BT8 predicts \(\alpha \geq 0.5\) → BT8/CCT is falsified.

6. Spacetime curvature is also emergent. Just as \(\hbar\) is read here as a local regime parameter, the effective metric \(g_{\mu\nu}\) is a push-forward from rule-space. Metric engineering would then be a search for rule-space transitions where these effective constants take different values.

7. Validation sequence:

8. Phase 1–2: calibrate within current physics (\(\hbar\), \(c\), \(G\) fixed) → proves methodology,
9. Phase 3+: probe for deviations → tests whether constants are emergent,
10. BT8 serves as baseline, not endpoint.

Constants Hierarchy Prediction

CCT does not claim all constants are equally tunable. Different parameters may have different stability depths in rule-space:

Constant	Predicted Stability	Test Strategy
\(\hbar\)	Deep attractor	Look for \(\Delta x \cdot \Delta p\) anomalies
\(c\)	Deep attractor	Look for effective light-cone deformations
\(\Lambda\) (cosmological)	Shallower?	Cosmological probes
\(\alpha_{\text{fine}}\)	Unknown	Precision spectroscopy

Partial success (some constants vary, others do not) would refine CCT's picture of rule-space topology rather than simply falsifying the framework.

---

H.9 Limitations and Next Steps¶

These empirical hints share several limitations:

• heterogeneous data sources and preprocessing;
• device- and platform-specific optimizations unrelated to RFH;
• lack of systematic control over \(\chi\), architecture, and confounders.
• paleomagnetic excursion inputs are still literature-derived interval estimates (not yet raw per-sample MagIC remeasurement with fully standardized excursion extraction).

They are therefore not used as core evidence for CCT. Instead, they:

• demonstrate how to lift RFH and programmability into existing experimental literatures,
• provide templates for RFH-style fits and reporting, and
• motivate dedicated CCT Labs experiments where bandwidth, energy, noise, and feedback are under explicit, logged control (see cct-lab.md for current experimental phases).

Future work can expand this appendix into a catalogue of well-documented case studies, each backed by a public notebook and dataset, to progressively sharpen or falsify the RFH and programmability heuristics across domains.