Named Control Field Modulation Architecture
by Nick Clark | Published March 27, 2026
Each affect dimension represented as a tuple of current magnitude, decay rate, policy-defined ceiling and floor, and timestamp, independently readable, writable, and auditable.
What It Is
Each dimension of the affective state is represented as a named control field: a tuple of current magnitude, decay rate, policy-defined ceiling and floor, and timestamp. These fields are independently readable, writable, and auditable. The architecture treats each affect dimension as a discrete, governable control parameter rather than an opaque emotional score.
Named control fields include uncertainty sensitivity, ambiguity tolerance, novelty appetite, persistence, escalation, risk sensitivity, and cooperation disposition. Each can be queried, updated, and constrained independently, enabling fine-grained governance over how affect modulates agent behavior.
Why It Matters
Monolithic emotional scores collapse distinct psychological dimensions into a single number, making it impossible to determine whether an agent is cautious because of high uncertainty or because of low cooperation disposition. Named fields preserve the dimensionality of affect, enabling precise diagnosis and targeted governance.
When affect dimensions are independently addressable, policies can constrain specific dimensions without affecting others. A governance rule might cap escalation tendency without limiting novelty appetite. A diagnostic system can identify which specific dimension is driving anomalous behavior.
How It Works Structurally
Each named control field maintains its own update pipeline. When an event triggers an affective update, the update function evaluates which dimensions are affected and computes new values for each independently. The ceiling and floor values are defined in the agent's policy reference and enforced at every update.
Audit trails record per-dimension changes rather than aggregate state changes, enabling forensic reconstruction of how each affect dimension evolved over the agent's operational history. Decay rates are per-dimension, allowing fast-decaying dimensions like escalation to return to baseline quickly while slow-decaying dimensions like cooperation disposition retain accumulated adjustments.
What It Enables
Named control fields enable affect-aware governance that operates at the correct granularity. Therapeutic agents can target specific affect dimensions for intervention. Companion AI systems can model relational dynamics through cooperation disposition and novelty appetite independently.
Cross-primitive coupling becomes precise: the confidence governor can weight uncertainty sensitivity more heavily than novelty appetite when computing execution authorization, while the forecasting engine might weight novelty appetite more heavily when selecting exploration strategies.
