This replication extends the initial four-model CPA-coherence study to an extended GPT-4o dataset comprising 90 runs (30 per constraint level). Using identical prompts and fixed parameters, we confirm that increasing informational constraint systematically reduces both the mean surprisal and mean token entropy, replicating the predicted coherence-under-constraint signature with high statistical confidence. The experiment additionally reveals a strong correlation between mean surprisal and mean entropy (r = 0. 957, p << 0. 001), suggesting a shared latent mechanism governing both uncertainty measures. Finally, the aggregate surprisal curve fits the Vogel-Fulcher-Tammann (VFT) equation with epsilon < 10^-5, mirroring glass-transition behavior observed in physical systems. An independent code and data audit confirmed the reproducibility of these findings, validated the VFT fit with an RMSE of 0. 02085, and verified data integrity across all 240, 313 token observations. All codes, data, and figures are provided for replication and review. Failure to reproduce this constraint-linked uncertainty reduction would falsify the CPA signature in this domain.
Debra S. Gavant (Tue,) studied this question.