Computational Efficiency η and Neural Dissipation ||Δ|| as Universal Markers of Phenomenal State: Evidence from Five Independent EEG Datasets

Abstract We introduce η (computational efficiency) and ||Δ|| (dissipation norm) as scalar EEG metrics derived from the Effective Power framework (Eff = Amp × Sγ; Waterman 2026a). η = ||Ψ|| / ||A|| measures the fraction of total oscillatory energy converted into organised spatial patterns. ||Δ|| = ||A − Ψ|| measures the energy expenditure that fails to produce organised output. We validate these metrics across five independent EEG datasets spanning two fundamentally different classes of phenomenal change: (1) intra-conscious transitions — reversal learning (ds004295, N=22), anagram insight (Oh et al. 2020, N=30), and meditation mind-wandering (ds001787, N=24); and (2) consciousness-level reductions — propofol sedation in two independent datasets (Bajwa et al. 2024, N=21; Chennu et al. 2016, N=20). η is significantly reduced across all five datasets and all comparisons (d=−0.492 to −4.145; all p<0.05). ||Δ|| is significantly elevated in all five (d=+0.410 to +1.619). Across both sedation datasets, η dramatically outperforms Lempel-Ziv Complexity (LZC) as a discriminator of sedation state: AUC 0.988 vs 0.755 (Bajwa); AUC 1.000 vs 0.612 (Chennu). LZC shows paradoxical behaviour under propofol — increasing rather than decreasing with sedation depth — due to delta-amplitude artefact. η is immune to this confound because it measures spatial pattern organisation rather than temporal complexity. In the Chennu dataset, η decreases monotonically in all 20 subjects (Spearman r<0 in 20/20; mean r=−0.680), and correctly differentiates subjects who become behaviourally unresponsive (drowsy) from those who remain responsive, at baseline (p=0.0002) and recovery (p=0.006). We propose η and ||Δ|| as a new class of EEG marker — "phenomenal state efficiency metrics" — sensitive to both intra-conscious transitions and consciousness-level gradients, complementing rather than competing with existing temporal complexity measures.