Enactive AI: An Architectural Framework for Coherent Decision Systems Under Drift

Modern institutional decision systems are often built on assumptions of stability: persistent correlations, predictable signals, and relatively fixed strategic conditions. In practice, financial environments are structurally dynamic. Relationships shift, signals decay, and regime changes alter the landscape in which models operate.

Systems optimized under stable assumptions may perform well locally while becoming fragile as conditions evolve.

This work advances an alternative orientation: designing enactive systems that engage in a process of sense-making to regulate behavior under drift rather than optimizing toward fixed objectives.

Figure 1. Enactive Decision Loop – A conceptual model illustrating how intelligent systems maintain coherence in dynamic environments through continuous feedback between environment, structural drift detection, sense-making, regulatory adjustment, and adaptive behavior.

This loop illustrates how enactive decision systems maintain coherence under changing conditions. Rather than optimizing toward a fixed objective, the system continuously interprets environmental drift, reorganizes its understanding of the situation, and adjusts its behavior accordingly.

Drift is treated not as noise or short-term volatility, but as structural change in the environment. As assumptions weaken and patterns reorganize, drift becomes an informational signal that models may no longer be aligned with current conditions.

Enactive systems therefore:

The emphasis shifts from improved prediction alone to maintaining effective action as conditions evolve.