The reproducible dimensionality of the human intracranial evoked response grows without detectable ceiling up to 480 trials

How many reproducible dimensions does the evoked response of a human brain contain? We recently introduced R-dim — the number of principal response directions that replicate across independent halves of trials, estimated fully out-of-sample — as the quantity that perturbational complexity indices track (doi:10.5281/zenodo.22100546, doi:10.5281/zenodo.22100826). In simulated recurrent networks, the growth of R-dim with trial count decelerates toward system-specific levels; convergence to a ceiling is suggested by those data but not demonstrated (doi:10.5281/zenodo.22168190, constraint C5). In humans probed with TMS, R-dim grows approximately linearly in ln(n) up to the 120 trials that standard protocols provide (positive within-subject trend in 27 of 30 subjects, sign test p = 4 × 10⁻⁶; exp31, the single primary criterion of a publicly sealed protocol, archived here). Here we ask whether a human ceiling becomes visible when far more trials are available, using the open COGITATE intracranial-EEG dataset (visual stimulation, up to 720 trials per subject; doi:10.17617/1.6qpg-gz82). The analysis plan, exclusion rules, two hypothesis tests and their four-cell interpretation matrix were preregistered with a public timestamp before any data file was opened; an interim analysis at 15 subjects and the final analysis of the complete dataset (32 usable subjects of 38; 58–194 intracranial channels; 245–698 valid trials) gave the same verdict. R-dim kept growing to the largest trial counts measured: the sealed deceleration test was negative (high-n increments, median +0.58, not significantly smaller than low-n increments, median +0.72; p = 0.32) while top-range growth was robustly positive (p = 1.8 × 10⁻⁵) — the preregistered verdict cell reads "no detectable ceiling up to n_max" (240–480 trials per subject). The verdict is unchanged under a 1–30 Hz analysis band, holds separately within the n_max = 240 and n_max = 480 subgroups, and is unchanged when the three subjects inspected during the pilot phase are also excluded. (v3) An audit of the estimator on synthetic data with this study's geometry, archived over eight seeds, shows it does not manufacture this pattern: under a response-free null R-dim sits on a positive floor of 0.40 ± 0.15 (mean ± sd over the trial grid and seeds) that does not grow with trials — the null's own increments are +0.03 ± 0.16 at the top of the grid and -0.05 ± 0.29 at the bottom, both within one standard deviation of zero — which is the property this result requires. On a 3-component synthetic response the estimator saturates by n = 30 but at 3.27, overshooting the true dimensionality by ~9%. The circular-shift surrogate that v1 presented as a second null is withdrawn as a control: it grows 2.5× across the grid, so it is not a null for the growth question. On a 25-component response with 1/k amplitudes R-dim grows steadily (1.93 → 6.04 at n = 480, recovering ~24% of the true dimension). That last, structure-matched control reproduces the human curves closely, supporting the economical reading — offered explicitly as calibration-based conjecture, not measurement — that the human evoked response contains many weak reproducible components, plausibly placing its true dimensionality in the tens, far beyond what current trial counts reveal. A dissociation reinforces this: the level of R-dim scales with electrode coverage (Spearman ρ = 0.58 with channel count), but the growth at high trial counts does not (ρ = 0.03) — coverage determines how much of the response you can see; trials keep revealing more of it regardless. Absolute levels carry the estimator's positive floor and are per-recording; every claim of this paper is a claim about growth and ordering with trial count, not about level. Together with the TMS results, trial-scaling growth of R-dim is now documented in 62 brains across two modalities and two independent datasets, in each case under a publicly sealed protocol: for the iEEG, sealed before any data file was opened; for the TMS-EEG, sealed before the growth analysis was run, on recordings the program had already used in earlier protocols. Files: full preprint (HTML), figure 1 (per-subject curves and increment scatter), figure 2 (four-condition estimator audit, mean ± sd over eight seeds), and a reproducible code archive containing all analysis scripts, the sealed preregistrations, per-subject numerical results, the archived eight-seed audit run (auditoria_exp32_v2.py / .jsonl), the exp31 script and results, and the verification script that recomputes every number claimed in the paper from those files. No human subject data are redistributed.