Interactive Causal Memory Theory: Emergent Noise, Shock, and Record Formation in Multi-Agent Systems

This paper presents Interactive Causal Memory Theory (ICMT), a framework in which noise,shock, and irreversible record formation emerge endogenously from interactions among finitecapacity agents. Noise is not treated as exogenous stochasticity, but as the accumulation of detectedyet policy-misaligned records generated by interaction under finite update speed. We formulate aclosed dynamical loop linking interaction, perception, inference, choice, and record update, andidentify shock events as bursts in misaligned record accumulation that precede regime transitions.The framework applies uniformly to artificial intelligence systems, social dynamics, and measurement processes, and provides a concrete dynamical instantiation of collapse-as-record principlesdeveloped in companion foundational work.This research is grounded in the framework of informationism, consistent with Wheeler’s “It fromBit” concept (1990) and the informational ontology (Floridi, 2011), which regards information asthe fundamental constituent of reality