Meditation moves the signs but not the field: dissociation between marginal shape and joint structure in peripheral physiology
Zenodo (CERN European Organization for Nuclear Research) September 7, 2026 DOI: 10.5281/zenodo.22584488 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Observational cohort Peer reviewed |
|---|---|
| Sample size | 15 |
| Population | Non-expert practitioners in the WESAD dataset |
| Intervention | Guided meditation |
| Topics | Meditation |
| Key findings | Guided meditation altered the distributional shape of individual peripheral signs (four of five with p < 0.05) but did not reorganize the statistical dependency among them; spectral integration (p = 0.72) and O-information (p = 0.60) were null, and the field discriminated conditions at chance (AUC = 0.58). |
Abstract
Meditation research has measured isolated physiological channels for decades, but the set of peripheral signs has rarely been treated as a field — a totality whose inter-channel dependency structure is itself the observable. We tested whether a guided meditation session moves the integrated peripheral field, or only the individual signs, in the public WESAD dataset (15 subjects; within-subject baseline and guided meditation; electrocardiography, electromyography, electrodermal activity, temperature, and respiration recorded simultaneously from chest-mounted sensors). Per 30-second epoch, we computed two observables of the five-channel totality: the von Neumann entropy of the correlation matrix (spectral integration) and non-parametric O-information (redundancy-synergy balance), each sign tested separately under its own ruler. The result is a dissociation between marginal shape and joint structure. Individual signs moved with the meditative axis robustly (four of five signs with p < 0.05; respiration d_z = +1.16, p = 0.0004; electrodermal activity d_z = +0.79; temperature d_z = -0.52; electrocardiography d_z = +0.35), and respiration alone discriminated the conditions with high accuracy (AUC = 0.89). But the integrated field did not move: spectral integration (p = 0.72, d_z = -0.20) and O-information (p = 0.60, d_z = -0.25) were robust nulls, and the field discriminated no better than chance (AUC = 0.58). Phase surrogates confirmed that the measured integration is second-order linear structure, with no non-linear excess in either condition. Short guided meditation in non-expert practitioners alters the distributional shape of each peripheral sign without reorganizing the statistical dependency among them. We propose that field-level integration tracks level transitions of consciousness rather than attentional axes of wakefulness, and discuss the measurement implications for integrative physiology.