This subtopic covers how data from learner interactions can be fed back into the system to adjust content, pace, or support.
From the Learning Engineering Toolkit
The Learning Engineering Toolkit roots feedback loops in engineering control theory, suggesting that the field transfers to teaching and training fairly well after a few unfamiliar terms are sorted out [LET-04]. An open-loop control system operates without any feedback, so it can only deliver dependable results when the underlying system is highly accurate and the acceptable error range is wide; a standard classroom, in which all learners do identical work, fits this pattern [LET-04]. A closed-loop system, by contrast, uses a sensor to read the output and feed it back to a controller, which compares that reading against the target value and modifies the input to close the gap [LET-04].
Applied to education, the book swaps in learning vocabulary so that the learner sits inside the system: benchmarks or standards play the role of the target, teaching is the input, acquired skills are the output, and testing functions as the sensor, with formative feedback tuning the instruction [LET-04]. Computing that gap is a subtraction, which the chapter calls negative feedback and regards as desirable, while positive feedback magnifies error and drives a system out of control, much like the screech of a microphone aimed at its own loudspeaker [LET-04].
According to the chapter, good learning relies on several feedback loops nested inside one another: an inner loop responds to the learner during a task, while an outer loop responds between tasks, for instance when deciding what to assign next [LET-04]. Each subsystem, it notes, can carry its own control loops on top of the loop governing the whole [LET-04]. The book maintains that feedback should be frequent, quick, and detailed, and observes that speeding up and multiplying feedback can even make up for an imperfect learning theory or weak assessments [LET-04]. Since an excess of feedback can overwhelm learners, engineers add filters and dampers to avoid over-correction, in the same way a skilled tutor holds back and lets a learner notice and fix a mistake instead of jumping in at once [LET-04]. The chapter also cautions that outcomes are highly sensitive to the chosen metric: feedback loops work so well that measuring the wrong quantity will optimize the wrong skill, a pitfall it links to instruction that merely drills for the exam [LET-04].
Sources from the Learning Engineering Toolkit
- [LET-04]Avron Barr, Brandt Dargue, Jim Goodell & Brandt Redd (2023). Chapter 4: Learning Engineering is Engineering. In Jim Goodell & Janet Kolodner, Learning Engineering Toolkit (pp. 125–151). Routledge / Taylor & Francis. doi:10.4324/9781003276579
Further Reading
Source: wrgr/lecommons — curated by the learning engineering community. Confidence:
medium— lecommons-curated; not yet independently expert-validated in this context. To validate or challenge any item: use the Edit button on this page. Upgrading confidence frommedium→highrequires expert sign-off.
Landmark Academic Papers
-
Design principles for virtual humans in educational technology environments — Scotty D. Craig, Nicolaus L. Schroeder (2018). International Journal of Artificial Intelligence in Education · doi:10.1007/s40593-017-0148-y · ~150 citations · tier: contemporary
Synthesized human-factors engineering principles for designing pedagogical agents. Showed that voice, appearance, and persona of virtual instructors systematically affect learner cognitive load and perceived credibility — establishing that LE must integrate psychological design alongside algorithmic design. Source: lecommons/landscape/data/papers.json · ID: LE-LS-AP-009 · confidence: medium · expert-validated: false
-
The voice quality of pedagogical agent impacts learning and agent perceptions — Ryan F. Siegle, Scotty D. Craig (2024). Journal of Computer Assisted Learning · doi:10.1111/jcal.12997 · ~20 citations · tier: contemporary
Empirically demonstrated that voice quality (prosody, naturalness, warmth) of a pedagogical agent significantly affects both learning outcomes and learner perception. Illustrates the meticulous human factors engineering required in modern AI-driven learning systems. Source: lecommons/landscape/data/papers.json · ID: LE-LS-AP-010 · confidence: medium · expert-validated: false
Key People
- Scotty D. Craig (profile), Arizona State University (active 2000s–present) — Human factors researcher; virtual humans and pedagogical agents specialist
Systematically investigated design principles for virtual humans and pedagogical agents in learning environments Source: lecommons/landscape/data/people.json · ID: LE-LS-PP-014 · confidence: medium · expert-validated: false
Programs & Initiatives
-
Human Factors and Ergonomics Society (CO) · link
Professional society for human factors and ergonomics. Publishes Human Factors journal and organizes the annual HFES conference. The Training Systems technical group directly addresses LE-relevant human factors in instructional systems. Source: lecommons/site/src/data/programs_people_registry.json · ID: LE-PP-133 · confidence: medium · expert-validated: false
-
INCOSE Systems Engineering Body of Knowledge (SEBoK) (CO) · link
Open wiki-based body of knowledge for systems engineering. Includes the Human Systems Integration knowledge area. The SE process framework (requirements, architecture, integration, verification) is the engineering process backbone that LE adapts for learning system development. Source: lecommons/site/src/data/programs_people_registry.json · ID: LE-PP-134 · confidence: medium · expert-validated: false
Lecommons enrichment applied 2026-04-17. All items pending expert validation. See wrgr/lecommons for source data and curation methodology.