Memory as Baseline Deviation: Toward a Relational Predictive Processing Framework

Current theories of memory and cognition operate under incompatible paradigms. Neuroscientific models achieve predictive precision but remain individualistic; phenomenological accounts capture relational being but resist formalization. We resolve this fragmentation by introducing a unified architecture: the Baseline-Deviation Framework. We formalize the principle that Being (Ontology) coupled with Obligation (Deontology) gives rise to Knowing (Epistemology) through temporal resonance, using classical stochastic dynamics. The framework has three core theses: 1. Memory as Probabilistic Computation: Memory is not the retrieval of stored traces but the continuous, stochastic encoding of novelty—events that significantly deviate from an agent's integrated experiential baseline. 2. Relational Constitution: The baseline is not individualistic; it is constitutively shaped by others through a dynamic coupling parameter κ(t), which evolves to create resonant states where mutual prediction becomes highly efficient. 3. The Complete Memory Lifecycle: Encoding, retrieval, and reconsolidation are emergent properties of this baseline-centric architecture, formalized as a three-stage metabolic cycle. We ground this framework in cortical-hippocampal neurobiology, link its components to measurable neural signals (N400, P300b), and validate core predictions via computational simulation. The architecture explains phenomena resistant to trace-based models (false memory, childhood amnesia, memory drift, conversational insight) while generating novel, falsifiable predictions for social cognition, clinical intervention, and epistemic convergence. We provide formal mathematical structure to relational ontology, demonstrating that phenomenological insights can be operationalized without loss of their essential character. Memory is the empirical domain where we validate the formalism; the formalism extends to any coupled cognitive system. Keywords: predictive processing, dynamic coupling, memory encoding, social cognition, relational ontology, inverse-inertia, resonant gate, baseline dynamics, deontology