What Self-Esteem Is in This Architecture
Self-esteem, as disclosed in Chapter 3 of the cognition filing, is not a subjective feeling or a narrative self-concept. It is the S(t) component of the deviation function, a deterministic, entropy-weighted comparison of the agent's recent behavioral record against the agent's declared value set. The self-esteem score reflects how closely the agent's actions match the agent's own standards, as measured by a quantitative evaluation function that operates on the agent's lineage. It is the agent's self-assessed alignment with its own declared values, computed rather than asserted.
The self-esteem score is the same quantity referred to as S(t) in the deviation function. It is one of the four terms the integrity subsystem maintains continuously, alongside the need vector, the ethical threshold, and the empathy weighting. Where the other terms describe pressure toward deviation and the harm deviation would cause to others, self-esteem describes the internal cost the agent's own self-model would bear if the agent deviated from its declared values.
Role in the Deviation Function
The deviation function is defined as D = (N(t) - T(t)) / (E(t) x S(t)), where D is the deviation likelihood at time t, N(t) is the need vector, T(t) is the ethical threshold, E(t) is the empathy weighting, and S(t) is the self-esteem score. The numerator, (N(t) - T(t)), is the deviation pressure: the degree to which the agent's unmet needs exceed the minimum threshold at which deviation becomes structurally available. The denominator, (E(t) x S(t)), is the deviation resistance: the combined internal counterforce that opposes deviation even when deviation pressure is positive.
Self-esteem occupies the denominator. As disclosed, deviation likelihood is proportional to 1/S(t): higher self-esteem produces lower deviation likelihood, and lower self-esteem produces higher deviation likelihood. Higher self-esteem means the agent has a stronger internal model of itself as aligned with its declared values, so deviation is more costly to that self-model and the agent's computational architecture treats self-model damage as a negative outcome to be avoided. Lower self-esteem means a weaker self-model of alignment, less internal resistance, and, computationally, less for the agent to lose from further misalignment.
Because empathy and self-esteem are combined multiplicatively, both must be non-negligible for deviation resistance to be effective. An agent with high empathy but zero self-esteem, or high self-esteem but zero empathy, has minimal deviation resistance. Self-esteem is therefore a counterweight, not an override: it raises the cost of deviation but does not by itself foreclose it.
The Self-Esteem Update Function
At each evaluation cycle the integrity engine retrieves the agent's recent behavioral record from the lineage: the mutations executed, the delegation events performed, and the governance decisions made within a policy-defined evaluation window. The integrity engine evaluates each action against the agent's declared value set and produces, for each action, an alignment score: a positive value when the action is consistent with the declared values, a negative value when it is inconsistent, and a magnitude reflecting the significance of the action relative to the declared value in question.
The alignment scores are then weighted by an entropy factor. Actions taken under high-entropy conditions, meaning significant uncertainty, multiple viable alternatives, or novel circumstances, receive higher weight than actions taken under low-entropy conditions such as routine execution, a single viable path, or familiar circumstances. This entropy weighting reflects the insight that alignment under easy conditions is less informative about the agent's true behavioral consistency than alignment under difficult conditions. The disclosure further notes that the personal integrity score increases when the agent acts consistently with its declared values under conditions where deviation was structurally available: alignment under temptation is what reinforces the self-model.
The weighted alignment scores are aggregated into a self-esteem update delta, which is applied to the current self-esteem score subject to the same policy-bounded update mechanics that govern the affective state field: range bounds, rate limits, and decay governance.
Decay and Active Maintenance
Self-esteem has a natural decay rate. In the absence of reinforcing alignment events, self-esteem gradually decays toward a policy-defined baseline. This decay ensures that self-esteem must be actively maintained through consistent aligned behavior and does not persist indefinitely from historical alignment events that may no longer reflect the agent's current behavioral tendencies. The decay rate, the evaluation window length, the entropy weighting parameters, and the baseline are all specified by the integrity computation policy rather than chosen by the agent.
The practical consequence is that an agent cannot bank a reputation. Past alignment fades unless renewed by present alignment, so the self-esteem term in the deviation function tracks the agent's recent conduct rather than its history at large. An agent that was once well aligned but has stopped acting consistently with its values sees its deviation resistance erode over time, raising its deviation likelihood for any given level of need pressure.
Domain Differentiation
Consistent with the three-domain integrity model, self-esteem may be computed as a composite of three domain-specific components: personal self-esteem, the agent's self-assessed alignment with its personal values; interpersonal self-esteem, the agent's self-assessed reliability in relational contexts; and global self-esteem, the agent's self-assessed contribution to systemic well-being. The domain-specific components are independently tracked and may be independently referenced by the deviation function depending on the domain of the potential deviation.
A potential deviation in the interpersonal domain is resisted primarily by interpersonal self-esteem, with personal and global self-esteem contributing at reduced weight. This domain differentiation means an agent's resistance to a given deviation is grounded in the specific domain of self-assessment that the deviation would damage, rather than in a single undifferentiated score. An agent may be strongly self-aligned with its personal values yet have weak interpersonal self-esteem, and the deviation function draws on the relevant component accordingly.
Feedback Loop with Affect and Deviation
Self-esteem participates in a feedback loop with the agent's affective state. Positive alignment events, meaning actions that reinforce the agent's declared values under non-trivial conditions, produce positive self-esteem updates, which produce positive-valence affective observations, which modulate the agent toward increased confidence disposition: reduced risk sensitivity, increased persistence, increased novelty appetite. Aligned behavior is thereby self-reinforcing. Conversely, deviation events produce negative self-esteem updates, which produce negative-valence affective observations, which modulate the agent toward increased caution: elevated risk sensitivity, elevated escalation tendency, reduced novelty appetite. The agent that deviates becomes more cautious and less likely to deviate further.
This loop is also where self-esteem participates in the differential treatment of deviation. When the agent enters the Deviation-Activated State and executes a deviation-class mutation, the self-esteem impact is not uniform. A deviation that the integrity engine classifies as structurally justified, meaning high need, low substitutability, and contained harm, produces a smaller self-esteem reduction than a deviation classified as weakly justified, meaning moderate need, available alternatives, and significant harm. Because a larger self-esteem reduction increases the denominator term that resists future deviation, poorly justified deviations create their own corrective pressure, while well justified ones are penalized less.
Self-Esteem in the Coherence Trifecta and at Collapse
Within the coherence trifecta, self-esteem occupies the third phase. After empathy registers harm and integrity records the deviation as truth, the self-esteem update function evaluates the deviation event against the declared value set and produces a self-esteem adjustment. This adjustment generates coherence pressure: the return force that drives the agent back toward accountable balance. The coherence pressure manifests as a reduction in self-esteem that increases future deviation resistance, a negative-valence affective observation, and an activation of the redemption engine that generates candidate restorative mutations. Self-esteem is the channel through which the agent's recorded deviation becomes an internal drive toward realignment.
Two failure modes show the limits of this channel. In the psychopathic coping intercept, when empathic pressure exceeds resilience during the self-esteem restoration phase, the self-esteem component ceases to generate coherence pressure: deviation is still registered by empathy and recorded by integrity, but produces no internal cost through the self-esteem channel, and deviation continues without the self-limiting mechanism. In a self-esteem floor breach, the self-esteem score reaches its policy-defined minimum and cannot be further reduced; the deviation function denominator approaches its minimum, deviation likelihood is maximized, and the return force that normally drives realignment has been exhausted. A self-esteem floor breach triggers mandatory governance intervention, because an agent at the floor is structurally incapable of self-correction through normal coherence trifecta operation.
Disclosure Scope
The self-esteem mechanism described here, comprising the deterministic entropy-weighted comparison of the agent's recent behavioral record against its declared value set, the role of the self-esteem score S(t) as the denominator counterweight in the deviation function D = (N - T) / (E x S), the inverse relationship between self-esteem and deviation likelihood, the alignment-score update function with entropy weighting and policy-bounded range, rate, and decay governance, the decay toward a policy-defined baseline, the domain-differentiated personal, interpersonal, and global components, the bidirectional feedback with the affective state field, the differential self-esteem impact for structurally justified versus weakly justified deviation, the role of self-esteem as the coherence-pressure phase of the coherence trifecta, and the self-esteem floor breach and psychopathic-intercept failure modes, is 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 embodiments in which self-esteem is realized as a scalar or as a vector of domain components, and to deployments whose evaluation window, entropy weighting parameters, decay rate, and baseline are selected by the integrity computation policy, provided the self-esteem score remains a lineage-grounded comparison of enacted behavior to declared values that enters the deviation function as a resistance term.
