This document presents a speculative but formally structured and carefully argued model for understanding why psi phenomena often exhibit a "decline effect" or even reverse (psi-missing) over repeated trials. The core idea is that psi results from a multiscale Precedence Principle (loosely a form of "morphic resonance"), which operates both locally (in individual experiments and other situations) and globally (across the entire cosmos, and/or large regions thereof). When a local psi pattern initially gains support, its low algorithmic complexity allows it to flourish. As the pattern proliferates and variants increase, its combined complexity eventually mismatches the broader cosmic resonance, causing suppression or inversion of the effect.

We show how this narrative might find a physics underpinning, via aligning it with a previously-presented theory of the physical foundations of psi, the Occamistic Precedence framework in Causal Set Theory, where each new observation is a causal-graph node whose probability is weighted by its historical frequency and algorithmic complexity.

This suggests that the neural underpinnings of psi phenomena can be modeled within a causal-set framework, where each neural "event" corresponds to adding discrete informational elements whose descriptive complexity governs their likelihood. Local neural templates that match low-complexity global patterns enjoy high insertion probability—forming shallow informational wells—while accumulating divergent variants deepen the well, suppressing or inverting further psi-like activity; analogous mechanisms could be engineered in AI via causal-set-inspired memory structures and complexity-based priors.

We also demonstrate a formal correspondence between an agent’s psi capability—its ability to exploit low-complexity psi correlations—and its universal intelligence as defined by Legg-Hutter (Solomonoff/AIXI). Under a wide class of "psi environments," both psi performance and general intelligence hinge on the agent’s capacity for low-complexity hypothesis generation and compression. We further relate Weaver’s notion of open-ended intelligence to psi capacity, showing that agents which continually seek ever simpler, unifying models naturally maintain resonance with broad cosmic patterns, thereby minimizing psi perversities. Finally, we outline empirical validation strategies spanning neuroscience (e.g. EEG/MEG complexity measures, TMS/tACS modulation) and AI prototyping (e.g. digital causal-set memories, neuromorphic implementations).