research-document RR-VE-THM-2026-F4B2
Theme Perception Experiment Matrix
Theme Perception Experiment Matrix
Status: active Related: #13
Goal
Turn the theme catalog into a reusable perceptual evidence set, not a palette gallery.
Independent outcomes
Never combine these into one "quality" score:
- readability/discriminability
- figure-ground salience
- visual hierarchy
- pair/palette harmony
- aesthetic preference
- activation association
- valence association
- semantic/status discrimination
- visual fatigue/attention competition
- context fit
Current matrix
| Theme | Experimental dimension |
|---|---|
| THM-0001 Quiet Fjord | light + cool + restrained |
| THM-0002 Ember Paper | light + warm + energetic |
| THM-0003 Night Ember | dark + warm + salient accent |
| THM-0004 Deep Current | dark + cool |
| THM-0005 Heather Field | near-hue / muted harmony |
| THM-0006 Signal Neutral | achromatic + single accent control |
| THM-0007 Verdant Signal | green / natural cool |
| THM-0008 Cobalt Citrus | blue-orange dual accent |
| THM-0009 Plum Gold | dark purple-gold association |
| THM-0010 Mono Graphite | achromatic control |
| THM-0011 Coral Tide | warm-cool complementary candidate |
| THM-0012 Electric Night | dark + multiple high-chroma accents |
Evaluation gates
Gate 1 — computed
- semantic text contrast
- non-text boundary contrast
- lightness range
- chroma range
- hue separation
- duplicate/near-duplicate distance
Gate 2 — rendered specimen
Render identical reference surfaces for every theme:
- prose + headings + links
- primary/secondary/destructive actions
- form controls, focus, disabled
- cards and elevation
- table
- status messages
- small data visualization
- dense dashboard
- print/document sample when applicable
Gate 3 — perceptual review
Record outcomes independently:
- hierarchy clarity
- foreground salience
- harmony
- attentional competition
- semantic confusion
- fatigue concerns
Gate 4 — contextual evidence
Where a theme is intended for a real product/context, record target population and task. Do not generalize a contextual result to all applications.
Pairwise experiments
Prefer controlled pairs that alter one dimension:
- warm vs cool with similar lightness/chroma
- high vs low chroma with similar hue/lightness
- light vs dark mode using comparable semantic relationships
- analogous vs separated hue while holding lightness contrast
- one accent vs two competing accents
- subtle vs functional 3:1 boundaries
- chromatic vs achromatic control
Status policy
draft: hypothesis onlyevaluated: computed checks plus rendered review completeaccepted: useful for a stated context with evidencedeprecated: retained for history but no longer recommended for that context
Acceptance is contextual. There is no globally "best" theme.
Next expansion
After THM-0012, prioritize:
- controlled one-variable variants of existing themes
- status palettes for success/warning/error/info
- color-vision-deficiency robustness
- data-series palettes
- print mappings
- reduced-chroma/fatigue variants
- high-contrast functional boundary variants
Do not add more decorative combinations until these controlled variants begin producing comparative evidence.