Mechanism
The affective state field is a seventh structural field added to the semantic agent schema. It encodes a deterministic, policy-bounded modulation vector that shapes how the agent deliberates: how it weighs alternatives, tolerates ambiguity, persists under partial failure, and escalates under constraint pressure. The disclosure maintains a strict separation of concerns between this modulation layer and the governance infrastructure. The affective state field modulates how the agent thinks; it does not determine whether the agent is permitted to act. Execution admissibility is decided by the governance gate, and the affective state field is not an input to that gate.
Stated in the spec's own terms, the affective state field cannot create authority the agent does not possess, cannot bypass policy constraints, cannot override trust slope validation, cannot validate truth claims, and cannot authorize execution that governance has denied. Even if the affective state modulation produces maximal confidence disposition and minimal risk sensitivity, the governance gate independently determines whether the proposed action satisfies all policy requirements. The separation is structural, not advisory: the gate evaluates policy compliance, trust slope validation, and cryptographic provenance independently of the agent's affective state.
The Four Dimensions of Separation
The disclosure enumerates four specific dimensions in which the separation operates. The first is authority: affective state cannot grant permissions. An agent with elevated cooperation disposition and low risk sensitivity still cannot delegate to an agent outside its policy-defined delegation scope, and an agent with elevated novelty appetite still cannot access semantic neighborhoods excluded by its policy reference field.
The second is truth validation: affective state cannot validate or invalidate factual claims. An agent with suppressed uncertainty sensitivity does not thereby treat uncertain information as verified. The agent's epistemic state, what it knows and with what confidence, is maintained independently of its dispositional state. Affective state modulates how the agent responds to uncertainty, not whether uncertainty exists.
The third is policy compliance: affective state cannot relax policy bounds. Even where the agent's current modulation state would benefit from broader exploration or more aggressive execution strategies, policy-imposed ceilings on field values, execution scope limitations, and governance requirements remain inviolable. The fourth is trust slope validation: the agent's trust slope, the cryptographic lineage trajectory that establishes its provenance and behavioral continuity, is computed and validated independently of affective state, and an agent of any affective configuration must still satisfy trust slope continuity requirements before execution is permitted.
Affect Modulates Inputs, Not Admissibility
Where affect acts at all, it acts on the quantitative parameters that shape how existing authorized processes execute, not on the admissibility decision itself. The disclosure states that the affective state field does not create new capabilities, authorize new actions, or bypass policy constraints; it modulates enumerated deliberation parameters within governance bounds. These targets include promotion thresholds, search breadth, branch growth rates, decay rates for unpromoted candidates, escalation thresholds, persistence parameters, delegation routing preferences, and mutation acceptance thresholds.
Each of these is a parameter the agent applies while preparing and ranking candidates, upstream of the gate that decides admissibility. Elevated risk sensitivity or elevated uncertainty sensitivity may raise a promotion threshold so that stronger evidence is required before a candidate advances; suppressed risk sensitivity may lower it. In every case the affective field changes the shape of deliberation, not the standard against which an action is finally judged admissible.
Modulating Sensitivity Without Lowering the Floor
The clearest illustration of the boundary is affect-modulated trust slope validation. The disclosure extends trust slope validation so that a validating agent's uncertainty sensitivity and risk sensitivity modulate the strictness of the continuity criteria. Elevated uncertainty sensitivity causes the validating agent to require a longer historical trajectory and tighter deviation bounds before accepting a delegate's trust slope as continuous; elevated risk sensitivity causes recent deviations to be weighted more heavily.
This modulation does not override the structural requirements of trust slope validation. The cryptographic lineage must still be verifiable, the temporal ordering must still be consistent, and the policy compliance record must still satisfy minimum thresholds. Affect adjusts only how close to those minimum thresholds the validation will accept. As the disclosure puts it, the minimum acceptable trust slope criteria cannot fall below the policy-defined floor regardless of the agent's affective state. A cautious agent demands stronger provenance evidence than an exploratory one, but neither can go beneath the policy floor.
Policy Bounds Enforced on Every Update
The separation is reinforced by the way affective updates are themselves governed. Every update to the affective state field is a policy-bounded mutation. The policy reference field specifies, for each named control field, range bounds, rate limits, an admissible trigger set, update authority, and decay governance. The affective state update function enforces these policy bounds as hard constraints during every update cycle.
Because affect itself is held inside a policy-defined operating envelope, it cannot be driven into a configuration that would let it exert undue upstream influence. If a computed update would exceed a ceiling it is clamped to the ceiling; if it would fall below a floor it is clamped to the floor; the magnitude of change per cycle is capped by the rate limit. Observations outside a field's admissible trigger set are ignored, which prevents spurious or adversarial environmental signals from modulating fields they should not affect. The disclosure states that no single observation or sequence of observations can drive the affective state outside its policy-defined operating envelope.
Auditability of the Boundary
The separation is made observable through the lineage record. Every mutation to the affective state field is recorded in the agent's lineage, subject to policy validation, and auditable by governance infrastructure. The governance gate records whether the agent's affective state was within policy bounds at the time of execution as a binary compliance record, without recording the specific field values in any externally accessible log.
Lineage entries for affective mutations record the observation type, the update direction on each affected field, and the policy compliance status, but not the absolute field values or the raw observations. Because each update is deterministic and each observation is recorded, the agent's affective state at any historical point is reconstructable by replaying the deterministic update function over the recorded observation sequence. This permits an auditor to verify that the agent's affective evolution followed policy-compliant paths and to reconstruct the modulation configuration at the time of a disputed decision, confirming that affect stayed within bounds rather than reaching across into the admissibility decision.
Composition With the Cognitive Architecture
The separation composes with the broader architecture without weakening it. With respect to confidence, affect modulates the rate at which confidence decays and recovers, so an agent with accumulated negative experience becomes progressively more cautious and one with accumulated positive experience becomes progressively more willing to execute. These dynamics are bounded by policy to prevent both runaway confidence and confidence collapse, and the confidence governor, not the affective field, still gates execution readiness. With respect to forecasting, affect modulates planning graph branching, pruning, and depth, shaping speculation without authorizing any action it produces.
The same pattern holds at the volatility boundary. When a volatility condition is detected, the agent is routed to an emotional quarantine state in which its operational scope is reduced: promotion thresholds rise to their policy-defined maxima, delegation authority is suspended, mutation acceptance thresholds are elevated, and execution is routed through an additional validation layer. Quarantine does not suppress the agent's affective state; the agent continues to process observations and update its field. It restricts operational scope until the affective state stabilizes. Across confidence, forecasting, and quarantine, affect enriches deliberation while the governance boundary that protects safety properties remains intact.
Disclosure Scope
The separation of concerns between the affective state modulation layer and the governance infrastructure, the four enumerated dimensions of that separation (authority, truth validation, policy compliance, and trust slope validation), the structural exclusion of the affective state field as an input to the governance gate, the modulation of enumerated deliberation parameters within governance bounds, the policy floor that affect-modulated trust slope validation cannot fall below, the policy-bounded update mechanism with range bounds, rate limits, admissible triggers, update authority, and decay governance, and the lineage-based auditability and deterministic forensic reconstruction of affective state are disclosed in the cognition filing (U.S. Application No. 19/647,395 and its international counterpart) in the affective state chapter. This article describes that disclosed mechanism. The scope extends to embodiments in which the affective state field is realized as a scalar valence, a multi-dimensional vector, or a structured record of named control fields, provided that in each case affect modulates deliberation dynamics and is never exposed as an input to the admissibility determination of the governance gate.
