Definitionv1
Feedback loop: a closed-loop system mechanism that observes
Feedback loop: a closed-loop system mechanism that observes its own output, compares it to a reference standard, and adjusts future behavior based on the comparison, enabling systems to learn and adapt
Why This Is a Definition
This definition establishes the precise semantic boundary of 'feedback loop' by identifying its genus (closed-loop system mechanism) and differentia (observation, comparison, adjustment process). It distinguishes it from mere observation or reaction, and provides the operational characteristics that make it a learning mechanism rather than just a monitoring tool.
Source Lessons
L-0461
Feedback loops are how systems learn
Any system that cannot observe its own output cannot improve.
L-0462
The feedback loop has four parts
Action observation evaluation and adjustment form the basic feedback cycle.
L-0479
Feedback loop hygiene
Regularly check that your feedback loops are still connected to meaningful outcomes.
L-0560
Monitoring is the feedback loop for your agents
Monitoring completes the feedback loop — observation enables adjustment enables improvement.
L-0466
Negative feedback loops stabilize
Self-correcting loops maintain balance by countering deviations.
Connections
Defines (63)
AxiomExponential Information DecayAxiomExtended Cognition ThesisAxiomDirected Attention as Depletable ResourceAxiomPerception as Predictive ConstructionAxiomHindsight Bias and Calibration NecessityAxiomTwo-Level Metacognitive ArchitectureAxiomExpertise Transforms Perceptual ChunkingAxiomComplementary Learning Systems ArchitectureAxiomConversational Memory Asymmetry From Production PlanningAxiomUltradian and Circadian Cognitive RhythmsAxiomAttention as Gate to Conscious PerceptionAxiomNeural Plasticity Enables Lifelong Automatic LearningAxiomPatterns Exist in Hierarchical Logical LevelsAxiomPerceptual Plasticity Through TrainingAxiomSystematic Overconfidence TaxonomyAxiomEmotion as Systematic Cognitive ModulatorAxiomGlucose-Cognition Dependency ThresholdAxiomBias Blind Spot AsymmetryAxiomBrain as Hierarchical Prediction MachineAxiomCognition Operates Through Dual Processing SystemsAxiomLooping Effects of Human ClassificationAxiomAutomatic Pattern PerceptionAxiomHierarchical Chunking Expands CapacityAxiomDunbar's Number Limits Stable RelationshipsAxiomBasic-Level Category PrivilegeAxiomExperience Segments into Nested Hierarchical EventsAxiomPiagetian Equilibration Through Schema DynamicsAxiomYou necessarily trust your own cognitive faculties as aAxiomReference class forecasting (using base rates from similarAxiomWhen a habit forms, neural activity spikes at the cue andPrincipleApply the same tags to notes from different domains whenPrincipleInstrument systems with the minimum number of metrics thatPrincipleConsolidate all agent status information onto a singlePrincipleBenchmark efficiency, accuracy, and quality dimensionsPrincipleAdjust systems when environmental context shifts even ifPrincipleWhen designing cognitive agents, examine the full patternPrincipleMake context switching costs visible through deliberatePrincipleBlock your measured peak attention hours on your calendar asPrincipleUse the 'five whys' technique on any significant energyPrincipleIdentify your information pipeline bottleneck by testingPrincipleMake the desired transition between behavioral links thePrincipleFor complex cognitive tasks that resist starting, designPrincipleTo make a replacement behavior competitive with anPrincipleConduct functional analysis before attempting extinction byPrincipleDesign experiments to produce intelligent failures—small,PrincipleWeight emotional data more heavily in domains where you havePrincipleWhen you experience an emotion in a relationship butPrincipleConnect contribution to direct impact visibility by creatingPrincipleFor each value you pursue, ask 'why do I want this?'PrincipleImplement structural interventions rather than relying onPrincipleReview accumulated notes in batches spanning weeks or monthsPrincipleBefore applying expertise developed in one domain to aPrincipleCognitive offloading must become an automatic daily habitPrincipleConduct blameless post-mortems that ask what the teamPrincipleWhen learning fails repeatedly despite effort, tracePrincipleDesign every information artifact with explicit compressionPrincipleWhen multiple valid hierarchies exist for the same data,PrincipleImplement continuous small schema updates based on newPrincipleWhen learning effort fails repeatedly, question your schemaPrincipleDirect change effort toward what you control (judgments,PrincipleBuild feedback loops into agent systems through regularPrincipleTreat major changes in goals, team composition, or systemPrincipleSpecify every delegation with five components: concrete