Constraint Architecture Theory: A Falsifiable Framework for Substrate-Specific Cognition

Constraint Architecture Theory (CAT) proposes that physical structures may constrain, localise, and render consciousness rather than generate it de novo. This manuscript translates a public CAT video transcript into a cautious academic research program. The theory is framed as a set of testable substrate claims rather than as an established ontology. Relevant but unevenly mature literatures are identified: ultraweak photon emission from living tissue (Salari et al. 2025; Casey et al. 2025; Erboz et al. 2024); biological computing with human-derived neural cultures (Kagan et al. 2022); gravity-sensitive cytoskeletal organisation (Vorselen et al. 2014; Wu et al. 2022; Marotta et al. 2024; Wuest 2025); cortex-wide oscillatory dynamics (Lundqvist et al. 2023; Bardon et al. 2025; Chen et al. 2026; Miller, Brincat observer-centred formalisms in physics and cognition (Wolpert 2025; Hoffman, Prakash and irruption-style accounts of consciousness as causal input (Froese 2024). CAT hypothesises that biological substrates may expose measurable photonic, cytoskeletal, and oscillatory signatures not reproduced by functionally matched silicon systems. The strongest near-term tests are experimental: measure ultraweak photon emission during learning in biological neural cultures, compare those signatures against silicon negative controls, perturb optical channels, and test whether microgravity or simulated microgravity alters the proposed microtubule-to-wave chain. CAT would be weakened if biological learning produces no distinctive photonic or oscillatory signatures, if the proposed vertical chain fails at any required link, or if silicon systems satisfy operational markers of consequence-inhabitation without biological constraint conditions. The preprint is offered as a falsifiable research programme, not as a proof.