Critical systems do not look calm, and they do not look random. They flicker. They produce bursts on many scales. This paper shows that the same “many-scales” pattern appears in a non-trained cognitive runtime when one analyzes its firing-variability logs. Technically, a 29,177-tick trajectory is analyzed using two standard criticality signatures: (i) spectral scaling, where the power spectral density (PSD) follows an approximate 1/fβ form, and (ii) avalanche scaling, where burst sizes and durations follow heavy-tailed laws. In the mid-run window (t ∈ 8264, 16504), the firing-variance PSD slope is β ≈ 1.04, consistent with 1/f-like scaling. Using a quantile threshold (q = 0.75) to define bursts, the avalanche size exponent is αS ≈ 1.46 and the duration exponent is αT ≈ 1.72 (tail counts ≥ 67). A shuffle surrogate destroys both signatures (PSD flattens; long durations vanish), and a threshold sweep shows that the exponents remain in a narrow band for moderate quantiles but drift at extreme thresholds. All claims are bounded to the included artifacts and should be treated as an instrumented measurement report.
Justin Lietz (Sat,) studied this question.