Mechanism

The affective state field is introduced as a seventh structural field of the semantic agent schema, alongside the intent field, context block, memory field, policy reference field, mutation descriptor field, and lineage field. It is a deterministic, policy-bounded data structure that encodes valence-weighted feedback derived from prior execution outcomes and environmental observations. It does not encode emotion in the phenomenological or subjective sense. It encodes a structured modulation vector that influences how the agent weighs alternatives, tolerates ambiguity, persists under partial failure, and escalates under constraint pressure.

The cross-primitive significance of the field is that it serves as an input to other cognitive operations. The affective state field modulates the confidence computation described in Chapter 5 and the forecasting operations described in Chapter 4, and in the broader architecture it also modulates the integrity engine and discovery traversal. Cumulative execution experience, encoded as affective state, modulates the agent's willingness to execute, encoded as confidence, and the agent's speculative planning behavior, encoded as planning graph dynamics. Because affect is a structural field rather than metadata, an annotation, or an external signal, this modulation participates in the same governance, lineage tracking, and policy enforcement that apply to every other agent field.

The Named Control Fields

The affective state field is organized as a structured modulation layer comprising a plurality of named control fields. Each named control field corresponds to a measurable modulation axis with defined semantics, value ranges, update rules, and governance bounds. In an embodiment the named control fields comprise at least: an uncertainty sensitivity field, encoding responsiveness to epistemic uncertainty; an ambiguity tolerance field, encoding capacity to operate where multiple interpretations are plausible and unresolved; a novelty appetite field, encoding disposition toward previously unobserved patterns, entities, or execution paths; a persistence-under-partial-failure field, encoding the tendency to continue a line of execution when intermediate results indicate partial failure; an escalation-under-time-pressure field, encoding the tendency to escalate, request external input, or delegate when operating under temporal constraints; a risk sensitivity field, encoding the weighting of potential negative outcomes relative to positive ones; and a cooperation disposition field, encoding the tendency to favor collaborative execution strategies over independent execution.

Each named control field is represented as a tuple comprising a current magnitude value within a defined range, a decay rate governing how rapidly the field returns toward a baseline in the absence of reinforcing stimuli, a policy-defined ceiling and floor bounding the permissible range, and a timestamp recording the most recent update. These named fields are the inputs the downstream primitives read. The article therefore describes a specific, enumerated set of dimensions, not an open-ended affect schema, and the cross-primitive coupling is defined in terms of which named fields modulate which downstream parameter.

Affect as Input to Confidence

With respect to confidence computation, the agent's affective state modulates the rate at which confidence decays and recovers. In an embodiment the confidence decay rate is multiplied by a factor derived from the agent's uncertainty sensitivity and risk sensitivity: when these fields are elevated, confidence decays faster, so the agent transitions more quickly from executing mode to non-executing cognitive mode as described in Chapter 5. When uncertainty sensitivity and risk sensitivity are suppressed, confidence decays more slowly, permitting the agent to sustain execution through periods of moderate uncertainty. The confidence recovery rate is similarly modulated: elevated persistence-under-partial-failure increases the recovery rate, permitting the agent to return to executing mode more rapidly after a confidence interruption.

This produces a defined dynamic. An agent that has accumulated negative execution experience, reflected in elevated uncertainty sensitivity and risk sensitivity, becomes progressively more cautious, because confidence decays faster and recovers more slowly. An agent that has accumulated positive execution experience becomes progressively more willing to execute. These dynamics are bounded by policy to prevent runaway confidence, meaning overconfidence leading to execution under unsafe conditions, and to prevent confidence collapse, meaning excessive caution preventing any execution.

Affect as Input to Forecasting

With respect to forecasting, the agent's affective state modulates planning graph construction parameters. The forecasting engine described in Chapter 4 reads the agent's current affective state when initializing a new planning graph generation cycle. The agent's novelty appetite modulates the branching factor of the planning graph, that is, the number of speculative alternatives explored at each forecasting step. The agent's risk sensitivity modulates the pruning criteria, that is, how aggressively low-probability or high-risk branches are discarded. The agent's persistence-under-partial-failure modulates the depth of the planning graph, that is, how many steps into the future the forecasting engine projects before terminating exploration.

These modulations cause the agent's speculative planning to reflect its accumulated experience: more conservative forecasts when experience warrants caution, more expansive forecasts when experience warrants exploration. The same named control fields therefore appear at two distinct downstream sites, confidence and forecasting, each reading the fields relevant to its own parameters rather than a generic affect scalar.

What Affective State Modulates Generally

The cross-primitive coupling extends to a set of enumerated deliberation and execution parameters. Each target is a defined computational parameter whose value the affective state field adjusts within governance bounds; the field does not create new capabilities, authorize new actions, or bypass policy. The modulation targets comprise at least: promotion thresholds, the minimum score required for a candidate to advance from one evaluation stage to the next, raised by elevated risk sensitivity or uncertainty sensitivity; search breadth, the number of candidate alternatives explored at each decision point, widened by elevated novelty appetite or ambiguity tolerance; branch growth rates during forecasting; decay rates for unpromoted candidates, slowed by elevated persistence-under-partial-failure; escalation thresholds, lowered by elevated uncertainty and risk sensitivity; persistence parameters, the number of retry or reformulation cycles before declaring failure; delegation routing preferences, modulated by cooperation disposition; and mutation acceptance thresholds, the stringency of the validation gate applied to mutations proposed by an external inference engine.

This enumeration is the substance of the cross-primitive input mechanism. Affect is consumed at each of these sites as a parameter, not as a command. The same accumulated experience that raises a promotion threshold in deliberation also narrows search breadth, slows candidate decay, and tightens the mutation gate, so a single experiential signal produces a coherent shift across multiple subsystems.

Affect Does Not Override Governance

A strict separation of concerns is maintained between the affective modulation layer and the governance infrastructure. The affective state field cannot create authority the agent does not possess, bypass policy constraints, override trust slope validation, validate truth claims, or authorize execution that governance has denied. It modulates how the agent thinks, not whether the agent is permitted to act. The separation is enforced structurally: the governance gate evaluates execution admissibility based on policy compliance, trust slope validation, and cryptographic provenance independently of affective state, and the affective state field is not an input to the governance gate.

This separation operates across specific dimensions. On authority, an agent with elevated cooperation disposition and low risk sensitivity still cannot delegate outside its policy-defined delegation scope. On truth validation, an agent with suppressed uncertainty sensitivity does not thereby treat uncertain information as verified; affect modulates how the agent responds to uncertainty, not whether uncertainty exists. On policy compliance, policy-imposed ceilings on field values and execution scope limitations remain inviolable. On trust slope validation, the agent's cryptographic lineage trajectory is computed and validated independently of affect. The cross-primitive coupling thus stops at the parameters of authorized processes; it never reaches the admissibility decision itself.

Deterministic, Policy-Bounded Updates

Updates to the affective state field are deterministic: given the same agent state, environmental inputs, and policy configuration, the update function produces the same output, ensuring the agent's affective evolution is auditable, reproducible, and governable. The update function operates on structured observations derived from the execution environment, which comprise at least repeated failure patterns, competing objectives, time pressure, novelty exposure, uncertainty levels reported by an inference engine, and execution success patterns. Each dimension of the affective state vector is updated independently according to its own rule, subject to policy bounds, and the update is recorded in the lineage as a state mutation event with the input observations and resulting change preserved for audit.

Every update is a policy-bounded mutation. The policy reference field specifies, for each named control field, range bounds clamping permissible values, rate limits capping the magnitude of change per cycle, admissible triggers restricting which observation types may drive a given field, update authority restricting which entities may initiate updates, and decay governance constraining decay parameters. When a structured observation is received, the update function verifies the observation is an admissible trigger, computes the raw update, clamps it to the rate limit, applies it, clamps the result to the range bounds, and records the complete transaction in lineage. This multi-stage clamping ensures no single observation or sequence can drive the affective state outside its policy-defined operating envelope, which is the precondition for using affect as a stable input to confidence and forecasting.

Affect as Input to Discovery Traversal

The affective state field of a discovery object modulates how that object traverses the adaptive semantic index described in Chapter 10. At each anchor node, the anchor's neighborhood evaluation module produces a candidate transition set, and the discovery object's affective state modulates the scoring and selection of candidates. Elevated uncertainty sensitivity assigns higher scores to conservative transitions, defined as transitions to anchors whose semantic neighborhood has high overlap with previously visited neighborhoods, whose content entropy is low, and whose trust slope history indicates stable behavior. Elevated novelty appetite assigns higher scores to novel transitions, defined as transitions to anchors with low overlap and moderate-to-high content entropy. Elevated risk sensitivity penalizes transitions to anchors with short trust slope histories, recently modified content, or high entropy. Elevated persistence-under-partial-failure permits the object to continue a line of traversal through low-relevance anchors rather than backtracking.

The result is that the same query, submitted through different affective contexts, produces different but equally valid traversal paths, and two traces from the same anchor diverge under conservative versus exploratory configurations. The affective state of the discovery object serves as a traversal parameter that shapes the search process without altering the search infrastructure. This is the same cross-primitive pattern seen in confidence and forecasting, applied to a third downstream primitive: the named control fields are read as inputs to that primitive's own scoring parameters.

Disclosure Scope

The affective state field as a seventh structural field of the semantic agent schema, the named control fields of its modulation layer (including uncertainty sensitivity, ambiguity tolerance, novelty appetite, persistence-under-partial-failure, escalation-under-time-pressure, risk sensitivity, and cooperation disposition), the deterministic and policy-bounded update function operating on structured observations, the strict separation by which affect modulates deliberation parameters but is not an input to the governance gate, and the cross-primitive integration of affect as an input to confidence decay and recovery, to planning graph branching, pruning, and depth in forecasting, and to candidate transition scoring in discovery traversal, are disclosed in the cognition filing (U.S. Application No. 19/647,395 and its international counterpart). This article describes that disclosed mechanism. The scope extends to the further enumerated modulation targets, including promotion thresholds, search breadth, candidate decay rates, escalation thresholds, persistence parameters, delegation routing preferences, and mutation acceptance thresholds, in each case as a policy-bounded parameter of an authorized process rather than a grant of authority.