cognitive-load-heatmap-prediction
Skill4 starsChanged 3 months agoMITView on GitHub
Predict visual attention paths and audit cognitive friction using eye-tracking simulation, Fitts's Law motor effort, Hick-Hyman Decision Time, and Visual Saliency Ratio. Use when cognitive load, attention heatmap, eye tracking, visual friction, CTA hierarchy, decision overload, or first 500ms attention.
In this repository46 other files it gives its agents.
- accessibility-reviewSkill
- accessibility-runtime-auditSkill
- adaptive-component-behaviorSkill
- ai-feature-ux-patternsSkill
- behavioral-pricing-psychologySkill
- business-model-readingSkill
- business-model-thinkingSkill
- component-style-guideSkill
- cross-functional-reviewSkill
- dark-mode-theming-systemSkill
- dashboard-layout-patternsSkill
- dashboard-scaffolding-contractSkill
- design-drift-detectorSkill
- design-system-builderSkill
- design-to-code-handoffSkill
- design-tokensSkill
- error-boundary-resilience-designSkill
- flow-based-functional-testingSkill
- growth-impact-reviewSkill
- i18n-localization-strategySkill
- information-architectureSkill
- interactive-element-auditSkill
- kinematic-motion-choreographySkill
- micro-interaction-motion-designSkill
- module-registry-syncSkill
- mvp-scopingSkill
- perceived-performance-loadingSkill
- prd-traceability-matrixSkill
- product-discoverySkill
- product-strategySkill
- project-health-diagnosticsSkill
- qa-feedback-loop-orchestratorSkill
- responsive-breakpoint-strategySkill
- responsive-qa-auditSkill
- route-integrity-checkerSkill
- security-privacy-reviewSkill
- stakeholder-requirement-mappingSkill
- touch-gesture-interactionSkill
- typography-systemSkill
- usability-heuristics-checkSkill
- user-flow-mappingSkill
- user-research-synthesisSkill
- ux-chaos-monkeySkill
- visual-hierarchy-reviewSkill
- visual-responsive-regression-testingSkill
- visual-style-extractorSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 94
If Primary CTA is placed in the top 30% of a mobile screen, automatically flag as Thumb-Hostile ($ID$ penalty $+1.5$ bits).
- Distinctive: line 146
Before marking cognitive audit complete, verify ALL items:
- Distinctive: line 162
After executing this skill, produce:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
🧠 Cognitive Load & Visual Heatmap Prediction Engine
Thesis
Visual attention is mathematically predictable. The interface must guide the user's eye to the primary value action within the first 500ms before decision fatigue occurs.
An interface with a Cognitive Friction Index (CFI) above 50 will consistently underperform regardless of visual quality. This engine quantifies that risk before a single pixel ships.
When to Activate
Activate this skill automatically when the user mentions:
"cognitive load","attention heatmap","eye tracking","where does the user look""CTA hierarchy","decision overload","too many buttons","which button is primary""first impression","above the fold","500ms","visual weight""landing page conversion","form is too complex","user gets confused"
5-Step Protocol
─── STEP 1: Map the First-500ms Eye-Tracking Scanning Pattern
Identify the dominant scan pattern based on layout type:
| Layout Type | Dominant Pattern | Primary Fixation Zone |
| Landing Page (1 CTA) | Z-Pattern | Top-left → Top-right → Diagonal → Bottom-right CTA |
| Dashboard / Data-Dense | F-Pattern | Top horizontal → Left-aligned scan → Spot fixation |
| Form / Input Flow | Single-Column Spine | Center top → Sequential downward march |
| E-Commerce Card Grid | Gutenberg Diagram | Top-left (Primary Optical Area) → Bottom-right (Terminal Area) |
| Article / Long-form | F-Pattern | First 2–3 lines read fully → Left-rail scan only |
Action: Confirm the layout type and predict the primary focal zone. Report the first fixation coordinates as a named region (e.g., Primary Optical Area, Terminal Action Zone, Dead Zone).
─── STEP 2: Hick's Law — Decision Branch Audit
Apply the Hick-Hyman Law to quantify decision reaction time:
$$RT = b \cdot \log_2(n + 1)$$
- $RT$ = User reaction time (seconds)
- $b$ = Empirical constant ≈ $0.155$ seconds per bit (for trained users)
- $n$ = Number of equally-weighted choices visible simultaneously
| $n$ (Choices) | $RT$ | Verdict |
| 1 | 0.155s | ✅ Optimal |
| 2 | 0.247s | ✅ Acceptable |
| 3–5 | 0.40–0.52s | ⚠️ Monitor |
| 6–9 | 0.56–0.65s | ❌ Cognitive Overload — Progressive Disclosure Required |
| ≥ 10 | > 0.70s | 🚨 Redesign Mandatory |
Action: Count all equally-weighted interactive actions visible in one unscrolled viewport. If $n > 5$, mandatory resolution:
- Promote the single highest-value action to Primary CTA with strong visual hierarchy.
- Demote remaining actions to secondary text links or a collapsed overflow menu.
- Apply 2-Step Progressive Disclosure for advanced/destructive actions.
─── STEP 3: Fitts's Law — Primary CTA Motor Effort Index
Calculate the Index of Difficulty (ID) for the Primary CTA:
$$ID = \log_2!\left(\frac{2D}{W}\right) \text{ (bits)}$$
- $D$ = Distance from the estimated starting cursor/thumb position (pixels)
- $W$ = Width of the target element (pixels)
| $ID$ (bits) | Effort | Verdict |
| ≤ 1.5 | Effortless | ✅ Optimal |
| 1.5–2.5 | Low effort | ✅ Acceptable |
| 2.5–3.5 | Moderate effort | ⚠️ Reduce D or increase W |
| > 3.5 | High friction | ❌ Requires immediate resize / repositioning |
Touch-Specific Fitts's Law — Mobile Thumb Zone Rules:
- Thumb start position on a 375px wide screen:
x = 187px, y = 720px(bottom-center). - Safe reach zone: Bottom 40% of screen height.
- If Primary CTA is placed in the top 30% of a mobile screen, automatically flag as Thumb-Hostile ($ID$ penalty $+1.5$ bits).
─── STEP 4: Visual Saliency & CTA Hierarchy Contrast Weight Ratio
Calculate the Saliency Weight Ratio (SWR) between Primary and Secondary CTAs:
$$SWR = \frac{S_{\text{primary}}}{S_{\text{secondary}}}$$
Where Saliency Score $S$ is computed as:
$$S = \text{ContrastRatio} \times \text{Area}_{px} \times \text{ColorTemperatureBoost}$$
ContrastRatio: WCAG contrast ratio of the element vs its background.Area_px: Button area in pixels (width × height).ColorTemperatureBoost:×1.3for warm hues (red, orange, yellow),×1.0for neutral,×0.9for cool (grey, muted).
Targets:
- $SWR \ge 2.5$ → ✅ Primary CTA dominates correctly
- $SWR < 2.5$ → ❌ Visual competition detected — users will hesitate
Common Anti-Patterns to Flag:
- Two buttons with identical fill color and size ("Ghost Button Wars").
- Primary CTA using
outline/ghostvariant while secondary uses solid fill. - CTA buried below the fold on mobile without sticky placement.
─── STEP 5: Cognitive Friction Index (CFI) — Final Score
Compute the unified Cognitive Friction Index (CFI) from 0 (frictionless) to 100 (maximum overload):
$$CFI = \left(\frac{RT_{norm}}{0.70} \times 30\right) + \left(\frac{ID_{norm}}{3.5} \times 30\right) + \left(\frac{\max(0,, 2.5 - SWR)}{2.5} \times 25\right) + \text{Penalty}_{misc}$$
Where $\text{Penalty}_{misc}$ (max 15 points) includes:
+5if scan pattern primary focal zone is in a Dead Zone (bottom-left).+5if a form has $> 6$ consecutive fields without a visual chunk separator.+5if the page has more than 1 Primary CTA in the same viewport.
| CFI Score | Grade | Action |
| 0–25 | A — Optimal | ✅ Ship immediately |
| 26–40 | B — Good | ✅ Minor polish recommended |
| 41–55 | C — Monitor | ⚠️ Address before launch |
| 56–70 | D — High Friction | ❌ Mandatory UX revision |
| 71–100 | F — Redesign Required | 🚨 Block shipping |
Completion Checklist
Before marking cognitive audit complete, verify ALL items:
- Scan pattern identified and primary focal zone confirmed.
- Hick's Law audit: $n \le 5$ or Progressive Disclosure applied.
- Fitts's Law: Primary CTA $ID \le 2.5$ bits (desktop) and $\le 3.0$ bits (mobile).
- SWR: Primary CTA $SWR \ge 2.5$ vs nearest secondary action.
- CFI Score calculated and documented.
- CFI $\le 40$ (Grade B or better) — or escalate with specific remediation plan.
- No "Ghost Button Wars" (two equally-styled competing actions).
- Mobile: Primary CTA in bottom 60% of screen height.
- Forms: $\le 6$ consecutive fields per visual chunk.
Output Artifacts
After executing this skill, produce:
cognitive-audit.md— Eye-tracking path prediction, all calculated scores (RT, ID, SWR, CFI), and remediation tasks.heatmap-prediction.md— Heatmap zone labels with predicted fixation sequence, saccade paths, and dead zones marked per layout region.
Anti-Patterns Reference
| Anti-Pattern | Cognitive Impact | Mandatory Fix |
| Paradox of Choice ($n > 7$) | Decision paralysis, abandonment | Progressive Disclosure + single promoted CTA |
| Primary CTA Demotion (ghost style) | Eye skips CTA, no conversion | Solid filled Primary, demote all secondaries to text links |
| Form Wall ($> 6$ fields) | Overwhelm, form abandonment | Split into named fieldset sections with progress indicator |
| Dead Zone Anchor (Bottom-left focal point) | CTA never seen | Relocate CTA to Terminal Zone or add thumb-reachable sticky footer |
| Above-Fold Mobile CTA (top 30%) | Thumb inaccessible | Move to bottom-sheet or sticky bottom bar |
| Competing Primaries (2+ solid CTA buttons) | User hesitation, lower CVR | One Primary, one Secondary max — all others as links |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

