Coupled Information-Action for Controlled Lucid & Group Dreaming: A Three-Dial Framework with Pre-Sleep Conditioning and Semantic Anchors

We propose a quantitative framework for gently steering dream phenomenology toward lucidity and group motif convergence without external storytelling or strong stimulation. The framework couples (i) biophysical sensitivity (temperature, magnetic field alignment, spectral matching), (ii) interface timing and dose of micro-cues during REM, and (iii) shared semantic anchors (a small lexicon of simple motifs; primary anchor: hands). We formalize a conserved "information-action budget," define an effective cue information Icue that factorizes into physical, interface, and semantic gains, and derive logistic predictors for lucidity probability and motif convergence. We add two modulators: Ψpred (prospective/predictive utility of the dream) and a super-agent vector U (values/discipline shaping in-dream choices). We specify a practical pre-sleep conditioning protocol (10-min Lights Dark + "hands-first" training), safety constraints, and A/B study designs with objective metrics. The model yields falsifiable predictions and a compact path to replication.