Mechanism
Branch dormancy is a state of a planning graph branch in the forecasting engine. A planning graph is a mutable, memory-referenced, directed semantic structure whose root node represents the agent's current verified state and whose branches each represent a distinct hypothetical trajectory. A dormant branch is one that is neither actively evaluated nor pruned: it is preserved in a reduced-resource state for potential future reactivation. Dormancy exists because a speculative trajectory may be non-viable under current conditions yet become viable later, when the agent's verified state, environmental conditions, or policy configuration change.
A dormant branch is stored in a reduced-resource format. The branch's speculative content, its projected outcome, and its evaluation metadata are preserved, but the branch is excluded from active simulation, slope projection, and affective reinforcement evaluation cycles. While dormant, the branch consumes minimal computational resources. Dormancy is therefore distinct from pruning, which removes a branch, and from active evaluation, which spends simulation and slope-projection cycles on a branch in each forecasting execution cycle.
When a Branch Enters Dormancy
A branch enters dormancy when any of three conditions is met. First, the branch has received a classification that renders it currently non-promotable, namely introspective or slope-ineligible, but the pruning manager determines that the branch has potential future value based on its alignment with the agent's long-term intent. Second, the branch's projection window has not yet expired, but the branch's evaluation score has fallen below the active evaluation threshold while remaining above the pruning threshold. Third, the forecasting engine explicitly marks the branch as dormant in response to environmental uncertainty that makes the branch's viability indeterminate.
These conditions reference the branch classification taxonomy used throughout the forecasting engine. A branch is eligible when it has passed slope validation, satisfied policy compatibility, and received positive or neutral affective reinforcement. A branch is introspective when it is slope-eligible and policy-compatible but negatively reinforced. A branch is slope-ineligible when its hypothetical Derived Anchor Hash would break continuity with the agent's trust slope trajectory. Dormancy provides a place to keep a branch that is not currently promotable but is not yet appropriate to discard.
Dormancy Does Not Suspend Temporal Expiration
Each planning graph branch is temporally anchored: it carries a timestamp recording the verified state from which it was generated and a projection window specifying the temporal range over which its speculative projections are considered valid. The pruning manager continues to apply temporal expiration to dormant branches. A dormant branch whose projection window expires is pruned even if it has not been reactivated. Dormancy preserves a branch across changing conditions, but it does not exempt the branch from the temporal-anchoring lifecycle that prevents stale speculation from persisting indefinitely.
Reinterpretation
Branch reinterpretation is the process by which a dormant or active branch is re-evaluated under changed conditions and assigned a new meaning, classification, or projected outcome. Reinterpretation occurs when the agent's verified state or environmental conditions change in a manner that affects the branch's evaluation. A branch that was introspective, that is, negatively reinforced, may be reinterpreted as eligible when the agent's affective state shifts. A branch that was slope-ineligible may be reinterpreted as slope-eligible when the agent's trust slope trajectory changes. A branch that was policy-incompatible may be reinterpreted as policy-compatible when the agent's policy configuration is updated.
Reinterpretation is consistent with the broader principle that branch classification is not permanent. The forecasting execution cycle re-evaluates branch classifications at each iteration, so the planning graph reflects the agent's current cognitive landscape rather than a single past snapshot. Dormancy and reinterpretation together let a branch survive the interval between the moment it was set aside and the moment changed conditions give it a new classification.
Deferred Promotion
Deferred promotion is the mechanism by which a branch that was not eligible for promotion when initially evaluated is retained, in either active or dormant state, and subsequently promoted when conditions change to render it eligible. This is how the forecasting engine implements temporal flexibility in speculative reasoning: the agent is not required to make irrevocable planning decisions at the time of initial evaluation. Instead, the agent may construct speculative branches, evaluate them under current conditions, retain the most promising branches across state changes, and promote them when the conditions for execution are met.
Promotion remains governed. Only a slope-eligible branch may be promoted to execution, and promotion through the governance-validated promotion interface constitutes a self-directed execution commitment: the agent has determined that the branch represents a viable, slope-eligible, policy-compatible, positively reinforced future state. Deferred promotion changes when a branch may be promoted, not the conditions a branch must satisfy to be promotable.
What the Combination Enables
The combination of dormancy, reinterpretation, and deferred promotion enables the forecasting engine to manage planning graph branches over temporal horizons that exceed any single evaluation cycle. This supports long-duration planning in environments characterized by uncertainty, intermittent resource availability, and evolving policy constraints. A branch may be held in dormancy at low computational cost, reinterpreted when the agent's state or environment changes, and promoted only once it becomes eligible, all without forcing the agent to either commit prematurely or discard a trajectory that may prove valuable.
The branch lifecycle is recorded. The pruning manager records pruning events in the agent's lineage as cognitive metadata, recording which branches were pruned, the pruning criterion that triggered removal, and the branch's evaluation state at the time of pruning. When a planning graph is pruned or a planning graph lifecycle terminates, the structure may be archived in a cognitive history store rather than deleted, preserving the branch structure, classification labels, affective reinforcement tags, slope projections, and promotion or pruning outcomes for later forensic reconstruction. Dormancy thus operates within a lifecycle whose transitions remain auditable.
Disclosure Scope
Branch dormancy, reinterpretation, and deferred promotion, comprising the preservation of a planning graph branch in a reduced-resource state that excludes it from active simulation, slope projection, and affective reinforcement cycles; the entry conditions tied to non-promotable classification, an evaluation score between the active evaluation threshold and the pruning threshold, or explicit marking under environmental uncertainty; the continued application of temporal expiration to dormant branches; the re-evaluation and reclassification of a dormant or active branch under changed verified state, affective state, trust slope trajectory, or policy configuration; and the retention and later promotion of a branch through the governance-validated promotion interface once it becomes eligible, 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 embodiments across deployment domains in which planning graph branches are managed over temporal horizons exceeding a single evaluation cycle, provided dormancy preserves a branch's speculative content and evaluation metadata, reinterpretation reassigns classification under changed conditions, and deferred promotion remains constrained to slope-eligible, policy-compatible, positively reinforced branches.
