# Protocol deviations, disclosed 2026-10-29

1. The frozen saved-array analysis protocol specified reporting temporal integrals, but the executed `analyze.py` reported maxima and means only. After the run, `integrals.py` computed trapezoidal integrals from the saved, hashed `comparison.npz`; its script, inputs and outputs are included. This post-run computation did not rerun Flyvis or the spiking circuit and did not alter the original protocol or output.
2. The frozen rectified-bridge protocol says to report quiet and grey independently, but its explicit condition list contains grey and A/B input-block trials, not a separate quiet trial. The runner executed the explicit 30 conditions. The input-block conditions had zero DNp15 and DNa02 output, while the grey condition had one whole-network spike and zero DNp15 output on each seed. A standalone quiet condition for the rectified bridge was not run. Do not treat input-block or grey as evidence of a separately executed quiet condition.

The original frozen protocol and runner hashes remain unchanged. These deviations were identified during independent post-run review.