lex-cognitive-echo

Residual cognitive activation modeling for the LegionIO cognitive architecture. Recent thoughts leave echoes that decay over time but continue to influence current processing through priming and interference effects.

Cognitive echo models the afterimage of prior thinking — the background activation that persists after an active thought has passed. An echo is created when a cognitive event (thought, emotion, decision, observation, prediction, error, success, or failure) finishes processing. It starts at a configurable intensity, decays each cycle, and gets pruned once it falls below the silent threshold (0.05).

Echoes serve two roles: priming (echoes above 0.3 bias new retrieval and processing toward their domain) and interference (echoes above 0.4 are strong enough to distort new processing). The extension also tracks chamber resonance — the ratio of interfering echoes to total echoes — as an early signal that an echo chamber is forming.

client = Legion::Extensions::CognitiveEcho::Client.new

# Register a new echo from a just-processed event
result = client.create_echo(
  content: 'recursive planning loop detected',
  echo_type: :thought,
  domain: :planning,
  intensity: 0.8
)
echo_id = result[:echo][:id]

# Reinforce an echo (e.g., it was encountered again)
client.reinforce_echo(echo_id: echo_id, amount: 0.15)

# Periodic maintenance — decay all echoes, prune silent ones
client.decay_all

# Query by role
client.active_echoes
client.priming_echoes       # intensity > 0.3 — influencing current processing
client.interfering_echoes   # intensity > 0.4 — strong enough to distort

# Domain-focused queries
client.echoes_by_domain(domain: :planning)
client.priming_effect(domain: :planning)  # aggregate priming for a domain

# Status
client.echo_status

bundle install
bundle exec rspec
bundle exec rubocop

MIT