Cognitive Load Theory — Workshop Design Skill

Core Model

Working memory is finite. Three components:

Rule: IL + EL + GL must not exceed working memory capacity. Every point of EL is a point stolen from GL.

IT Domain: Elevated Intrinsic Load by Default

IT topics have structurally high element interactivity — concepts cannot be learned in isolation because understanding any one requires simultaneously holding several others (Kubernetes: Pod IP + ClusterIP + CNI + DNS + network policy = co-dependent).

Implication: For high-interactivity IT content, even perfect design produces above-average cognitive load. Drive EL to near-zero, because IL cannot be reduced without distorting the domain.

Information Density Ceiling

Rule of 4±1: No segment introduces more than 4 genuinely new concepts simultaneously (Cowan, 2001 — revision of Miller's 7±2, measured under active learning conditions where effective capacity drops to ~4 chunks).

Density Management Checklist

• Count novel concepts: target ≤ 4 for novice, ≤ 6 for expert
• Pre-train vocabulary 48h in advance (90-95% comprehension required for fluent processing)
• Open each segment with a high-level schema diagram
• One activity per concept cluster; do not mix clusters in the same exercise

Audience Expertise Detection

Expertise Reversal Effect

Instruction optimal for novices actively harms experts — and vice versa.

5-Minute Diagnostic at Session Start

Ask learners to label a blank architecture diagram. Missing labels = novice. Correct labels + additions = expert. Wrong labels = misconceptions requiring pre-correction.

Scaffolding Progression — The Fading Model

NOVICE ──────────────────────────────────────── EXPERT

Fully Worked → Completion → Problem + Hints → Independent
  Example      Problem        (optional)       Problem

Fading principle: gradually remove scaffolding — more effective than maintaining full scaffolding or removing suddenly.

IT lab example — "configure a Kubernetes Deployment" for novice:

1. Fully annotated YAML with inline comments (worked example)
2. YAML with specific fields blanked out (completion problem)
3. Requirements brief only; learner writes YAML from scratch (independent)

Multimedia Design Rules

Modality effect: spoken narration + visual diagram outperforms on-screen text + visual diagram for conceptual explanations. No channel collision.

Session Duration and Cognitive Fatigue

Rule: Any segment with high intrinsic load must not exceed 25–30 minutes without a recovery break. The "10-minute attention span" claim is debunked (Wilson & Korn, 2007) — what is supported is that fatigue accumulates nonlinearly and recovery breaks aid schema consolidation.

Recovery options: low-load review activity | discussion / Q&A | genuine break (5+ min)

Cognitive Offloading

Use checklists, reference cards, Miro boards during high-load activities and multi-step procedures where the procedure is not the learning objective.

Do not substitute for encoding: after the activity, retrieve without the reference card — offloading is a performance scaffold during the activity; encoding happens after.

Miro rule: board externalizes the learner's in-progress model, NOT a replica of the instructor's completed model (which creates split-attention).

Bloom's Levels x Cognitive Load

Critical rule: assigning a Create activity to a novice audience is cognitively impossible without prior scaffolding — it produces frustration, not learning.

Pre-Training Heuristics

Deploy when session vocabulary is domain-specific or main activity has high element interactivity.

1. Glossary distributed 48+ hours before
2. Vocabulary warm-up at session open: 3-minute match-the-term exercise
3. High-level architecture overview at session start: skeleton schema to hang detail on

Collaborative Learning

Use pairs for high element interactivity activities where solo processing would exceed individual working memory. The second person's working memory holds elements the first cannot.

Pair composition rule: mixed-expertise pairs outperform same-level for high-load — but the experienced partner must verbalize reasoning, not just produce answers.

Do not use pairs for simple low-interactivity tasks — coordination overhead exceeds benefit.

CLT in Workshop Series (C-02)

Expertise reversal across sessions: Session 1 = fully worked examples; Session 3 = completion exercises; Session 5+ = independent problem-solving only.

Element interactivity ceiling: Do not introduce a new high-interactivity topic in the same session as a foundational concept — participants need one session to consolidate before layering.

Mixed-Audience CLT Mechanics

If experts are disengaged and novices are overwhelmed in the same activity, this is a mixed-audience CLT failure. Fix with one of:

1. Parallel tracks: Two versions — scaffolded (novice) and unscaffolded (expert)
2. Role differentiation: Novices as driver (executing), experts as navigator (reasoning aloud)
3. Fading scaffolding: Start with full worked example; remove scaffold midway when novices have baseline schema

CLT Anti-Patterns in IT Workshop Design

Quick Diagnostic — Is This Activity Cognitively Safe?

1. Element count: > 4 novel concepts for novice? Pre-train or split.
2. Load type audit: List all EL sources. Eliminate each.
3. Bloom match: Does activity level match audience's schema depth? If mismatch, add scaffolding.
4. Duration check: Sustained high-load window under 30 minutes? If not, insert low-load recovery.
5. Expertise reversal risk: Mixed audience? Provide differentiated paths.
6. Offloading appropriateness: Multi-step procedure recall required? Provide reference card; schedule retrieval after.