FIELD STUDY 008 / THE PERCEPTION SERIESINDEPENDENT RESEARCH EXHIBIT · 2026
THE CHROMATIC OBSERVATORY

SAME PIXELS.
DIFFERENT WORLD.

The colour may not have changed. Its company has. Put one colour in two different surroundings, then watch your visual judgement negotiate with context.

No sign-in · No tracking · No claim to read your mind

01 / LOOK02 / QUESTION03 / VERIFY04 / LEARN
01 — EXPERIMENTILLUSTRATIVE · NOT A CLINICAL OR PSYCHOPHYSICAL TEST

THE CONTEXTUAL COLOUR COMPARATOR

WHAT CHANGES
WHEN NOTHING CHANGES?

The two centre swatches below are programmed to use exactly the same hex colour. Only their backgrounds and neighbouring patterns differ. Try to make one look brighter, duller or differently tinted without touching the other.

LIVE LABORATORY / LOCAL-ONLY
CHAMBER ADARK SURROUND
IDENTICAL TARGET#9971C9
CHAMBER BLIGHT SURROUND
IDENTICAL TARGET#9971C9
THE TARGET PIXELS MATCH.

Both central swatches use #9971C9. Any perceived difference is not a change to their specified RGB colour.

REARRANGE THE UNIVERSE.

CHANGE THE SURROUNDINGS, NOT THE EVIDENCE.
01 / START WITH A SCENE
02 / COLOUR IN THE MIDDLE

Changes both target swatches together. The target always remains identical in both chambers.

RGB DIFFERENCE BETWEEN TARGETS0 / 0 / 0
03 / COLOUR AROUND IT
#211935
#F2E8CE
04 / BACKGROUND STRUCTURE

Patterns add edges and spatial cues. They do not turn this demo into a measurement of visual processing.

ILLUMINANCE-STYLE LUMINANCE COMPARISON / MATH, NOT MIND-READING
CHAMBER A—
CHAMBER B—

These computed ratios compare each target colour with the surrounding flat colour using the WCAG sRGB luminance formula. They do not quantify how different the targets look to you.

THIS IS THE WHOLE POINT

YOUR EYES RECEIVE LIGHT.
YOUR MIND INTERPRETS A SCENE.

THOSE ARE RELATED TASKS. THEY ARE NOT IDENTICAL TASKS.
02 — THE SCIENCERESEARCH-BASED EXPLANATIONS · CITED BELOW

A FIELD GUIDE TO SEEING

THE COLOUR
ISN'T ALONE.

Colour is not a small label stuck to an object. What we perceive depends on the light reaching our eyes, how our visual system adapts, the relationship between neighbouring regions, and assumptions we may form about a scene.

A hex code is a specification, not a sensation.

On an ordinary sRGB display, a hex triplet such as #9971C9 specifies red, green and blue channel values for a pixel. It does not specify what the colour must look like in every environment. The display's calibration, the room's illumination, reflections on the screen, neighbouring pixels and the viewer's visual system all matter. Our interactive exhibit holds the central swatch's specified sRGB value constant while you alter its surroundings. That is a useful controlled demonstration: we know what the code is doing, even when we cannot prescribe what you must experience.

To understand the distinction, consider a musical note. Its frequency can be measured while its perceived quality changes with harmony, room acoustics and loudness. Colour isn't literally sound, but the analogy helps: physical descriptions and subjective experience are not interchangeable. Vision science studies precisely how measurable stimulation relates to reported appearance. Shevell and Kingdom's review of colour in complex scenes emphasizes that the appearance of a region cannot be understood from that region's light in isolation. [1]

Different viewers may report different effects, and even one viewer's judgement may vary with adaptation. This is not automatically an error. It is a reminder that the visual system is actively processing information rather than passively preserving an RGB spreadsheet. A striking result in this exhibit does not imply that your brain has been tricked by a designer; the mechanisms normally support stable and useful perception in complicated scenes.

The neighbours have a vote.

In simultaneous colour contrast, a target can appear to shift when the adjacent region changes. A grey patch may seem lighter against a black field than against a white field. A neutral-looking target surrounded by a strong chromatic colour can seem displaced toward a contrasting hue. This general family of context effects is often called chromatic induction when it concerns colour attributes. It does not have a single simple rule that predicts every display, patch size or viewing condition.

Some mechanisms operate on relative signals in spatially organized visual pathways. Receptive fields, opponency and interactions between neighbouring regions give the system ways to encode contrast, not merely independent absolute measurements. Research by Krauskopf, Zaidi and Mandler tested mechanisms of simultaneous colour induction and reported results consistent with interaction in higher-level chromatic mechanisms. Their work is one source among many, not a licence to reduce all perception to a single neuron or universal contrast formula. [3]

That is why the experiment offers flat fields, rings and grid patterns. Changing boundaries and nearby structure can change what spatial relations are available to vision. The pattern controls are designed to provoke curiosity, not to reproduce the laboratory apparatus or isolate the causal role of any neural circuit. If nothing dramatic happens for you, that is still a useful observation. Try altering distance, saturation or room lighting, then record what actually changed rather than forcing the expected result.

Seeing a surface means estimating a world.

Imagine a white shirt seen at dawn, under a warm lamp, and beneath a blue skylight. The spectral distribution of the illumination varies, so the light reaching your eyes changes even though the fabric has not been repainted. Yet the shirt often looks approximately white across these settings. That helpful stability is called colour constancy. It is often imperfect, depends on context, and has generated substantial experimental and theoretical research. Foster's review surveys its complexity rather than presenting it as a solved computational trick. [2]

Visual adaptation adds another layer. After extended exposure to a particular distribution of light or colour, responses can shift as the system recalibrates to prevailing conditions. Webster describes visual adaptation as a widespread property of sensory coding, with mechanisms operating across different timescales. [7] A quick on-screen comparison cannot cleanly separate the effects of adaptation, immediate contrast, expectation and image layout. It can, however, encourage careful questions about these possibilities.

One influential way of thinking about constancy treats visual perception as an estimation problem: the retinal signal entangles surface reflectance with the illumination falling on it. The system uses contextual information to infer what belongs to the material and what belongs to the light. Delahunt and Brainard investigated constancy under changes in reflected illumination, a reminder that real scenes challenge simplified models. [4] The important word is estimate, not magically recover: different conditions can support more than one interpretation.

Is the effect local contrast, or a hypothesis about light?

Scientists have developed multiple accounts of contextual appearance. One family emphasizes relatively local spatial interactions: neighbouring regions alter responses to a target, and patterns of contrast propagate through visual processing. Another emphasizes scene interpretation: boundaries, shadows, illuminants and material relationships influence what a region is judged to be. These are not mutually exclusive explanations for all phenomena. Different mechanisms may operate together, or become more prominent in different experimental settings. The current page cannot decide which explanation dominates any visitor's experience.

The distinction becomes especially visible with images that contain shadows and perspective. Edward Adelson's 1995 checker-shadow illusion shows two regions of the same image intensity that appear to have different lightness because they occupy different interpreted contexts. This is principally a lightness demonstration, not proof that any two arbitrary colours will always diverge. [5] Even the famous 2015 photograph of a dress provoked competing interpretations: some analyses emphasize assumptions about illumination, while others test spatial filtering and image statistics. [6] [10]

That disagreement is productive. A strong explanation should propose observable predictions, survive carefully designed controls and specify its boundary conditions. Our comparator tells you the exact target colours and shows the relative luminance contrast of each target against its flat background. It does not claim to reveal the internal algorithm your visual system used. A physical proof of pixel equality and a theory of perceived colour are two different kinds of evidence.

03 — CONTEXT IN THE WILDTWO DOCUMENTED EXAMPLES
CASE / 1995
A = B?

THE CHECKER-SHADOW

MIT vision scientist Edward Adelson's checker-shadow demonstration places equal-intensity patches inside a scene that implies different lighting. What looks like a mistake in measuring grey can be understood as part of a visual system trying to interpret surfaces and shadows.

What it supports: scene context can strongly affect reported lightness. What it does not support: a universal numerical prediction for every colour contrast effect.

SEE THE PRIMARY DEMONSTRATION ↗
CASE / 2015
BLUE/GOLD

THE DRESS DISAGREEMENT

In a widely discussed ambiguous photograph, people reported markedly different colour categories for the same garment image. Lafer-Sousa, Hermann and Conway documented individual differences; subsequent work examined roles for illumination inference and spatial information.

What it supports: the same image can permit divergent appearances. What it does not support: personality typing, moral judgement or claims about someone's intelligence.

READ THE ORIGINAL PAPER ↗
!

WHAT THIS DOESN'T PROVE.

Nothing here measures your colour vision, diagnoses a condition, estimates a personal susceptibility to illusion, or establishes an effect size. An on-screen swatch is not a calibrated spectroradiometer. Device gamut, ambient conditions, colour management, adaptation, accessibility settings and individual variation can all affect appearance. The luminance ratios shown by the tool are a well-defined calculation for specified sRGB colour pairs; they are not perceptual scores. The page is an educational demonstration with evidence-informed explanations, not an original peer-reviewed experiment.

There is also no scientific basis for turning a colour judgement into a verdict about character. Someone who sees a different result is not necessarily inattentive or wrong. Where observations diverge, the useful next question is which conditions changed and which explanation would predict the difference.

04 — FIELDWORKTAKE THE EXPERIMENT INTO THE WORLD

A STUDY WITHOUT A SCOREBOARD

TRY TO DISAGREE
WITH YOUR EYES.

Give the same target to two contexts. Predict what you will see, compare what you actually notice, and make the uncertainty visible. You do not have to experience a particular illusion for the activity to be worthwhile.

DOWNLOAD THE FIELD GUIDE

A real printable PDF: experiment worksheet, explanation, limitations, references and discussion questions.

  1. 01

    PREDICT FIRST.

    Write whether the two targets will look identical, different in lightness, or different in hue before changing a setting.

  2. 02

    HOLD THE TARGET STEADY.

    Keep the centre hex value fixed. Change only one surround at a time. Use the bridge to expose the shared target colour.

  3. 03

    CHANGE YOUR EXPLANATION.

    Compare a local-contrast account with a scene-lighting account. Which observation would make one explanation less plausible?

  4. 04

    DOCUMENT THE LIMITS.

    Record screen type, room conditions, viewing distance and whether the effect was strong, subtle or absent. A single observation is not a population finding.

Context matters in interfaces, too.

Website designers often choose colours using a palette in isolation and later discover that a button looks too weak, a warning dominates the entire page, or a caption becomes hard to read against its actual background. Surrounding structure and lightness can alter perceived emphasis. But designers should not mistake aesthetic drama for accessibility. A calculated contrast ratio is a useful part of evaluating text legibility; meaningful labels, focus states, text alternatives and testing with diverse users matter too.

The W3C's WCAG 2.2 explanation of minimum text contrast describes a 4.5:1 ratio for ordinary text and 3:1 for large text, with defined exceptions. These thresholds concern accessibility requirements, not the magnitude of a colour illusion. [8] Our lab's contrast display deliberately makes this distinction visible. If a pair of backgrounds produces a striking experience, that does not automatically mean it is a good interface combination.

Colour words are not the whole visual world.

When people describe what they see, they frequently use categories such as blue, purple, green or grey. Those categories help us communicate, yet the underlying appearance varies continuously and is influenced by context. Cross-language studies have debated both regularities and differences in colour naming. Regier, Kay and Khetarpal model colour naming as efficient partitioning of perceptual space, rather than a trivial consequence of arbitrary labels alone. [9]

That literature should not be used to claim that speakers of different languages inhabit entirely separate physical colour universes. It raises more careful questions: how do people carve up colour space for communication, and how do perceptual constraints and cultural conventions interact? The present experiment concerns the visual experience of constant swatches across changing backgrounds. Language may shape how visitors report the result, but this tool does not measure that influence.

05 — QUESTIONSSIX FREQUENTLY ASKED QUESTIONS

THE QUESTIONS
DON'T RUN OUT.

Are the two middle colours truly identical?

Yes: the page assigns exactly the same sRGB hex value to both target elements, and updates them together. Their pixel appearance on your particular screen could still depend on rendering and display characteristics; code equality is what the demonstration guarantees. Use “Show colour bridge” as a visual check.

Why do I sometimes see little or no difference?

Context effects depend on the colours used, the size and geometry of the surrounds, attention, adaptation, viewing conditions and the observer. The presets are illustrations rather than calibrated stimuli. A weak or absent experience does not establish anything about your health or ability.

Does a higher contrast ratio mean a stronger illusion?

No. The ratio is a mathematical comparison of sRGB relative luminances. It helps describe the difference between a target and its surround, but perceived hue, saturation and lightness depend on more than that one number. The interface intentionally avoids a fake “illusion intensity” score.

Is this about colour blindness?

No. This is a general educational exhibit about contextual colour appearance, not a test or screening tool for colour vision deficiency. The explanation is available as text, not through colour discrimination alone. If you have a specific vision concern, this site cannot evaluate it.

Isn't this all just an optical trick?

It is an intentionally simplified demonstration, but the principles are part of ordinary perception. Contrast, adaptation and assumptions about lighting help people interpret changing scenes. “Illusion” should not be confused with evidence that perception is generally unreliable.

Can I cite this as a scientific study?

Cite the primary sources below for empirical claims. You may describe this page as an independently created educational demonstration by Sahil Tayal, dated 2026, but it did not collect participant data, run a validated experiment or publish new research findings.

06 — THE RECEIPTSPRIMARY SOURCES & AUTHORITATIVE EXPLANATIONS

WHAT WE CHECKED.

The page distinguishes published research, interpretation and our illustrative simulation. These links are provided for independent follow-up; some journal articles may require library access.

  1. [01]
    Shevell, S. K., & Kingdom, F. A. A. (2008). Color in complex scenes. Annual Review of Psychology, 59, 143–166.
    SOURCE ↗
  2. [02]
    Foster, D. H. (2011). Color constancy. Vision Research, 51(7), 674–700.
    SOURCE ↗
  3. [03]
    Krauskopf, J., Zaidi, Q., & Mandler, M. B. (1986). Mechanisms of simultaneous color induction. Journal of the Optical Society of America A, 3(10), 1752–1757.
    SOURCE ↗
  4. [04]
    Delahunt, P. B., & Brainard, D. H. (2004). Color constancy under changes in reflected illumination. Journal of Vision, 4(9), 764–778.
    SOURCE ↗
  5. [05]
    Adelson, E. H. (1995). Checker Shadow Illusion. MIT Perceptual Science Group (original visual demonstration).
    SOURCE ↗
  6. [06]
    Lafer-Sousa, R., Hermann, K. L., & Conway, B. R. (2015). Striking individual differences in color perception uncovered by 'the dress' photograph. Current Biology, 25(13), R545–R546.
    SOURCE ↗
  7. [07]
    Webster, M. A. (2015). Visual adaptation. Annual Review of Vision Science, 1, 547–567.
    SOURCE ↗
  8. [08]
    World Wide Web Consortium (W3C). Understanding Success Criterion 1.4.3: Contrast (Minimum), WCAG 2.2.
    SOURCE ↗
  9. [09]
    Regier, T., Kay, P., & Khetarpal, N. (2007). Color naming reflects optimal partitions of color space. Proceedings of the National Academy of Sciences, 104(4), 1436–1441.
    SOURCE ↗
  10. [10]
    2017 vision research. Spatial filtering, color constancy, and the color-changing dress. Journal of Vision (see bibliographic record for full authorship and issue details).
    SOURCE ↗