journal-entry JR-VE-PE-001

Perceptual Envelope Program Inception

JR-VE-PE-001: Perceptual Envelope Program Inception

Starting question

What new typography and color experiments should Visual Engineering run, especially for dyslexia, color-vision deficiency, and other disabilities, and what direction could differentiate the methodology?

Repository observations

The operational UI briefing had solid general accessibility rules but did not yet express disability-specific mechanisms or experimental pathways.

The deeper research base was stronger than the operational layer:

  • Project Atlas already rejects one global readability score;
  • typography research already models conditional optima and compensation;
  • the color REP already rejects universal contrast, palette, and polarity claims;
  • Itten experiments already examine CVD and aging redundancy.

The missing step was integration.

External evidence checked

  1. 2026 dyslexia-font meta-analysis: no reliable general speed/accuracy benefit.
  2. individualized-font research: large within-reader differences with preserved comprehension.
  3. visual-crowding research: spacing benefit in a measured high-crowding subgroup.
  4. personalized-font recommendation research: prediction from reader/font features is plausible.
  5. September 2026 WCAG 3 draft: contrast model remains under development and explicitly acknowledges size/weight and possible CVD-specific requirements.

Synthesis

The common structure across typography, low vision, CVD, and dyslexia is not diagnosis. It is a constrained perception-and-task system.

This led to two candidate concepts:

  • Perceptual Envelope;
  • Semantic Channel Survivability.

Decisions

  • Do not create a canonical dyslexia mode.
  • Do not create a canonical color-blind palette.
  • Preserve WCAG 2.2 conformance as mandatory baseline.
  • Investigate measured bottlenecks and individualized response surfaces.
  • Separate preference, performance, and comfort.
  • Treat simulation as fault injection.
  • Use affected users for validation.
  • Keep adaptive profiles semantic-neutral and privacy-preserving.

Immediate experiments

  • EX-VE-TYP-001
  • EX-VE-TYP-002
  • EX-VE-COL-006
  • EX-VE-X-002

Negative knowledge captured

The program explicitly records what should not be inferred:

  • "dyslexia font has no average benefit" does not mean typography does not matter;
  • "personalization can help" does not mean automatic adaptation is net beneficial;
  • "CVD simulation finds a collision" does not mean an affected user will make that error;
  • "simulation finds no collision" does not prove accessibility;
  • "high contrast helps some low-vision tasks" does not mean maximum luminance is optimal;
  • "user preference matters" does not mean preference equals performance.