The Consciousness State Vector — Testing the Semi-Independence of Presence (CII) and Depth (PDS) Across Sleep Stages

Abstract

This study tests the central architectural claim of Hierarchical Recursive Integration Theory version 3 (HRIT v3): that conscious experience requires two semi-independent computations — a controlled hallucination of the world (Presence, indexed by CII) and a controlled confabulation of biological existence (Depth, indexed by PDS) — and that either can vary independently of the other under specific conditions. The test exploits the natural dissociation predicted by the framework during REM sleep. REM is characterised by high cortical integration — complex, desynchronised EEG, active dreaming — which predicts high CII. Simultaneously, REM involves skeletal muscle atonia, which suppresses interoceptive afference from the periphery and predicts low PDS. If CII and PDS are truly semi-independent, REM should show high EEG complexity but attenuated heartbeat-evoked potential amplitude relative to wake — a dissociation not predicted by any single-dimensional consciousness measure. Three pre-registered hypotheses are tested in 16 healthy subjects from the CAP Sleep Database (PhysioNet, physionet.org/content/capslpdb/1.0.0/): (H1) permutation entropy stratifies across sleep stages in the predicted order; (H2) HEP peak-to-peak amplitude in the frontal-central window stratifies differently from permutation entropy, with REM showing lower HEP than wake despite comparable EEG complexity; (H3) within-subject CII-PDS correlation is significantly lower during REM than during NREM, constituting a direct empirical test of semi-independence. A secondary exploratory hypothesis (H4) tests whether RBD patients — who lack REM atonia — show a smaller dissociation than healthy controls. All analysis code will be committed to github.com/hamzasalmahi-lab/csv-dissociation-sleep before any confirmatory analysis is run.