Training Feedback Loop¶
Organizational learning loop — instantiates Mental Model Mismatch Repair
Turns recurring expectation failures across a population into revised training and monitors whether the same mismatch keeps coming back.
Training Feedback Loop is the standing cycle that keeps a trained population's models from drifting. Rather than repairing one person's mismatch once, it treats recurring expectation failures across a cohort as a signal, routes them into revised examples, scenarios, and instructor explanations, and then watches whether the same mismatch reappears in the next cohort. Its defining move is population-scale and continuous: its unit of attention is not a single wrong prediction but a pattern of them over time, and its product is not a fixed model but a loop — capture the recurring failure, feed the revision, and monitor for recurrence. It presumes the corrected model itself is authored elsewhere; its job is to notice which mismatches keep happening and to close the loop so the training program learns from its own graduates' errors.
Example¶
A hospital rolls out a new electronic health record, and nursing onboarding teaches it. Across the first several cohorts, a pattern emerges: new nurses expect that placing a medication order automatically notifies pharmacy, when in fact it queues until a separate verification step — so doses are delayed and nurses are surprised. The Training Feedback Loop captures this recurring mismatch from post-onboarding error logs and preceptor debriefs, tags it as a repeat rather than a one-off, and routes it into the curriculum: a new scenario in which the order sits unnotified until verification, plus an explicit instructor note. The loop does not stop there. It then monitors the next cohorts for the same delay pattern; when the rate of that specific surprise drops and stays down, the fix is treated as holding, and when a related surprise appears elsewhere in the workflow, the loop opens again. The corrected scenario content comes from the curriculum team; the loop's contribution is knowing which mismatch to fix and confirming it stopped recurring.
How it works¶
- Capture failures as a stream, not incidents. Pull recurring expectation errors from error logs, preceptor debriefs, help-desk themes, and assessment misses across the whole population.
- Detect the pattern. Distinguish a repeating, population-level mismatch from a one-off slip — the same wrong prediction showing up cohort after cohort.
- Route the pattern into revision. Hand the identified recurring mismatch to whoever authors the training content, as a specific target to fix.
- Monitor recurrence. Track whether the same mismatch reappears in later cohorts after the revision, and re-open the loop if it does or if a neighbor surfaces.
Tuning parameters¶
- Capture breadth — which channels feed the loop (logs, debriefs, assessments, support tickets). Wider capture catches more mismatches but adds noise and triage load.
- Recurrence threshold — how many repeats before a mismatch is promoted to a curriculum fix. A low threshold reacts fast but chases flukes; a high one is stable but slow.
- Monitoring window — how long to watch later cohorts before declaring a fix successful. Longer windows catch slow-returning drift but delay closing the loop.
- Loop cadence — how often the captured patterns are reviewed and routed. Faster cadence keeps training current but strains the content team; slower lets mismatches accumulate.
When it helps, and when it misleads¶
Its strength is that it keeps a curriculum honest against a moving target: systems change, populations shift, and a one-time fix goes stale, so the loop is what turns training from a static artifact into a self-correcting process. Its recurrence monitoring is a practical instance of measuring training at the behavior level[n1] — did graduates stop making the mismatch on the job — rather than merely at the satisfaction or quiz-score level.
Its failure mode is measuring reaction instead of transfer: a loop that only tracks "did learners like the module" or "did they pass the quiz" will report success while the mismatch keeps recurring in practice, because agreement with an explanation is not the same as corrected behavior. Its classic misuse is treating the loop as a complaint box — capturing failures endlessly but never confirming that a revision actually drove the recurrence rate down. The guarding discipline is to anchor the loop on the on-the-job recurrence rate of the specific mismatch, close each loop only when that rate falls and stays down, and keep the monitoring window long enough to catch drift that returns after the training's glow fades.
How it implements the components¶
feedback_capture_loop— collects recurring expectation failures across the population as a continuous stream and routes each confirmed pattern into curriculum revision.recurrence_monitor— tracks whether a given mismatch reappears in later cohorts after the fix, re-opening the loop when it does and only declaring success when the on-the-job rate stays down.
It routes recurring failures into revision and watches for their return, but it does not author the corrected model itself (model_revision_path) or run the safe rehearsal that validates it in a fresh case (expectation_validation) — that hands off to its nearest neighbor Simulation-Based Correction and to Documentation Revision. Diagnosing any single case (mismatch_diagnosis) belongs to Incident Mental-Model Review.
Related¶
- Instantiates: Mental Model Mismatch Repair — Training Feedback Loop is the standing recurrence-monitoring cycle that keeps a trained population's models from drifting.
- Consumes: Simulation-Based Correction and Documentation Revision — supply the revised scenarios and content the loop routes recurring mismatches into.
- Sibling mechanisms: Usability Testing · Incident Mental-Model Review · User Journey Diagnostics · Interface Affordance Redesign · Documentation Revision · Simulation-Based Correction · Expectation Audit
Editorial Notes¶
Form Classification¶
Form family: Experiment, Test & Rehearsal
Rationale: Training Feedback Loop is defined in the frozen evidence as: Turns recurring expectation failures across a population into revised training and monitors whether the same mismatch keeps coming back. Its operative deployed or enacted form is therefore Experiment, Test & Rehearsal.
Nearest alternative: Communication, Facilitation & Learning — Communication, Facilitation & Learning can support this mechanism, but the evidence centers the concrete operation described above rather than the alternative family's defining operation.
Review outcome: Adjudicated after independent review; medium confidence.
Origin Attribution¶
Primary origin: Organizational & Management Science
Origin pattern: Convergent development
Present-day reach: Universal
Rationale: Both independent reviews identify organizational management as the historical home of the operation—Turns recurring expectation failures across a population into revised training and monitors whether the same mismatch keeps coming back.. The retained alternates document formative adjacent traditions; the reach field, not the origin field, carries later applicability.
Related originating lineages:
- Education & Pedagogy — Education, assessment, and instructional practice supplies a parallel or contributing lineage for the mechanism's defining operation: turns recurring expectation failures across a population into revised training and monitors whether the same mismatch keeps coming back.
- Systems Thinking & Cybernetics — Systems thinking, feedback control, and cybernetics supplies a parallel or contributing lineage for the mechanism's defining operation: turns recurring expectation failures across a population into revised training and monitors whether the same mismatch keeps coming back.
Review resolution: Both blind reviewers independently place the defining operation—Turns recurring expectation failures across a population into revised training and monitors whether the same mismatch keeps coming back.—in organizational management. Their queued differences are secondary: alternate_origin_disagreement, origin_mode_disagreement, domain_reach_disagreement. Reviewer A contributes no unique alternate; reviewer B contributes ['systems_cybernetics']. I preserve the full evidence-supported union of 2 alternate domain(s), without a numeric cap. origin_mode=convergent reflects the reviewers' evidence about historical construction, while domain_reach=universal separately reflects present-day portability. The affirmative encyclopedia-synthesis finding is preserved, and confidence=medium uses the more conservative reviewer level.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; medium confidence.
Notes¶
[n1] Kirkpatrick's four levels of training evaluation run from reaction (did learners like it) and learning (did they pass) up to behavior (did they act differently on the job) and results. The loop's discipline is to judge itself at the behavior level — recurrence of the real mismatch — not the easier reaction or quiz levels. ↩