phase: 4W
experiment_root: /data/digital-drosophila/experiments/phase4w-late-kc-feedback
model_root: /data/digital-drosophila/models/male-cns-lif/late-kc-audit-v1
runtime_root: /data/digital-drosophila/models/male-cns-lif/deterministic-runtime-v2
phase4p_root: /data/digital-drosophila/experiments/phase4p-deterministic-runtime
prior_root: /data/digital-drosophila/experiments/phase4v-csus-semantics
prior_model: /data/digital-drosophila/models/male-cns-lif/csus-audit-v1
watch_neuron_idx: [59832, 78836, 82312]
odor_indices: [0, 1, 2, 3, 4, 5]
seed: 131101
condition: offset
late_start_ms: 500
repeat_odor_index: 0
ancestry:
  max_hops: 512
  rule: greatest positive edge-weight times synaptic-decay since latest eligible presynaptic spike; edge-id tie break
  time_requirement: source actual spike + frozen delay <= target actual spike
  interpretation: time-supported dominant-input ancestry, not proof of necessity or unique causal predecessor
isolated_replay:
  modes: [base, omit_PN_olfactory, omit_KC, omit_APL, omit_MBON, omit_DAN, omit_other, omit_APL_overlay]
  context: fixed recorded other-source arrivals and applied APL overlay; conditional replay, not closed-loop brain intervention
  update_math: exact frozen float32 state update; no new edges or parameter changes
scope:
  diagnostic_only: true
  learning: false
  candidate: false
  reserve_test_odors: untouched
  no_frozen_artifact_writes: true