RESEARCH DOSSIER
Color Vision and Color Appearance
Abstract. Color vision is the transformation of wavelength-dependent stimulation into discriminable visual signals, whereas color appearance comprises hue, brightness, lightness, colorfulness, saturation, and chroma under specified viewing conditions. This fundamental research monograph reconstructs the scientific history of color matching, trichromatic receptor theory, opponent processing, and spectral colorimetry; analyzes cone physiology, genetics, retinal and cortical coding, adaptation, chromatic discrimination, color constancy, spatial induction, material perception, and illumination inference; and examines language-sensitive color categorization, individual variation, computational prediction, and artificial vision. Its method is a structured critical narrative synthesis of identifiable original experiments, theoretical papers, technical standards, and scholarly reviews. The evidence supports well-characterized cone-based encoding and opponent mechanisms, alongside context-sensitive and partially stable appearance representations whose precise neural and inferential organization remains under investigation. Distinct psychophysical endpoints and calibration requirements constrain generalization from laboratory matching to natural scenes. The monograph concludes by identifying falsifiable research priorities in observer variability, object-level color, illumination statistics, neural mechanism, linguistic effects, and cross-system comparison. It presents neither an original experiment nor a registered systematic review or quantitative meta-analysis.
Keywords: color vision; color appearance; trichromacy; color matching; cone photoreceptors; opponent-process theory; chromatic discrimination; color adaptation; color constancy; color categories; color naming; memory color; surface color; visual neuroscience; psychophysics; CIE colorimetry; CAM16; natural illumination; Artificial Era
Color Vision and Color Appearance as a Scientific Problem
Color vision is the capacity to distinguish variations in the spectral composition of light through the responses of visual photoreceptors and subsequent neural computations. Color appearance concerns the perceived attributes of a stimulus or surface—hue, brightness, lightness, colorfulness, saturation, and chroma—under a specified viewing context. The distinction is operational: discrimination experiments can show that two samples differ without establishing how either looks, whereas appearance-matching and scaling experiments address how a difference is experienced. A color percept therefore depends jointly on optical stimulation, retinal encoding, neural adaptation, the spatial configuration of a scene, viewing history, and the task by which the observer reports it. The central explanatory demand is to connect these levels without treating a physical wavelength, an RGB code, or a color name as a complete account of appearance. Witzel & Gegenfurtner (2018) Gegenfurtner (2025)
The light reaching the retina after reflection from a surface depends on both the spectral power distribution of the illuminant and the wavelength-dependent reflectance of the material. For a simple matte surface under a single illuminant, the retinal spectral radiance can be approximated by their wavelength-by-wavelength product, modified by viewing geometry and the eye's spectral transmission. A visual system must extract useful information from this mixture. Distinct physical spectra can produce indistinguishable cone responses, while an unchanged surface can generate different retinal spectra under changed lighting. These properties make color matching, chromatic adaptation, and color constancy complementary research problems: the first establishes equivalences in receptor-limited stimulation; the second characterizes history-dependent transformations; the third measures the degree to which surface appearance remains stable when the illumination changes. Maloney & Wandell (1986) Brainard & Wandell (1992) Foster (2011)
This monograph is a structured critical narrative synthesis of original experiments, physiological measurements, computational models, and scholarly reviews. It traces the scientific pathway from color matching to cone photopigments, opponent channels, cortical representations, surface-color perception, and contemporary color appearance models. It evaluates where stimulus control justifies inference about receptors, where perceptual matches justify inference about experience, and where neural and algorithmic findings support explanatory mechanisms. The analysis emphasizes the reproducible structure of major paradigms and the limits of generalizing from calibrated patches to materials in real-world scenes. It does not report new participant data or a preregistered exhaustive systematic review. CIE (2018a) Conway et al. (2010) Witzel & Gegenfurtner (2018)
Concepts, Physical Stimuli, and the Appearance Vocabulary
A monochromatic light has a narrowly concentrated spectrum; a broad-spectrum source contains energy across many wavelengths. Objects usually return wavelength-dependent mixtures of incident illumination, with additional optical effects from scattering, fluorescence, gloss, translucency, and interreflections. A color stimulus should consequently be specified by spectral radiance or by carefully calibrated tristimulus values with explicitly defined observers and viewing conditions. Chromaticity represents colorimetric proportions while omitting an independent luminance scale. Luminance is a standardized photometric quantity weighted by luminous efficiency; brightness is the observer's perceived intensity of light, and lightness is appearance judged relative to a similarly illuminated white or reference. These terms carry distinct measurement implications and should not be substituted for one another. CIE (2018a)
Hue denotes the characteristic by which a color is described as reddish, yellowish, greenish, bluish, or combinations thereof. Colorfulness describes the perceived degree of chromatic content; saturation expresses colorfulness relative to perceived brightness; chroma characterizes colorfulness relative to a similarly illuminated reference white. An observer's judgment of unique red or unique blue concerns a perceptual category boundary, not a single spectral wavelength encoded as such in the retina. A neutral stimulus can appear colored through simultaneous contrast or after adaptation, and a physically identical sample can differ in apparent hue or lightness according to its surround. These phenomena demonstrate why perceptual reports must be anchored to reference fields, adaptation state, and explicit instructions. Hurvich & Jameson (1957) Wandell (1993) Bosten et al. (2015)
Three levels of equivalence govern empirical work. A spectral match reproduces the same spectral distribution at the relevant plane. A metameric match reproduces equivalent responses in the reference observer's color-matching system despite spectral differences; metamers may cease to match for a different observer or under different optical conditions. An appearance match equates specified perceptual attributes and often changes when backgrounds, illumination, spatial structure, or adaptation differ. None of these equivalences automatically transfers across all conditions. The practical scientific lesson is that color-space coordinates represent specified functions of stimulus and observer assumptions, while color appearance requires an explicit model of context. Stockman & Sharpe (2000) Brainard & Wandell (1992) CIE (2018a)
Origins of Experimental Color Theory
Nineteenth-century three-component theories associated with Thomas Young and Hermann von Helmholtz developed the proposition that many color-matching results could arise from a small number of physiological sensitivities. James Clerk Maxwell's experimental color matching helped transform this idea into quantitative stimulus mixing and measurement. The historical achievement was to identify the low-dimensional structure of human trichromatic matches, not to provide a completed account of all perceived color relations. The later CIE 1931 observer functions institutionalized a reproducible standard representation of matches under defined conditions; they remain conventions for standardized colorimetry, with modern procedures described in CIE (2018a). The physiological interpretation emerged through subsequent measurements of cones and visual pigments. Stockman & Sharpe (2000) Nathans et al. (1986a)
Ewald Hering's opponent tradition drew attention to the phenomenological organization of red–green, blue–yellow, and light–dark relations. Red and green commonly function as mutually exclusive components in unique-hue judgments, as do blue and yellow; perceptual complements and afterimages display systematic opponent structure. Hurvich and Jameson developed a quantitative opponent-process theory that linked hue cancellation and psychophysical measurements to opponent response functions Hurvich & Jameson (1957). Trichromatic receptor theory and opponent coding answer questions at different stages of processing: three cone classes constrain the receptor measurements, while subsequent circuits transform those measurements into signals relevant to chromatic differences and appearance. Modern neurophysiology explains how these principles can coexist without making every opponent response a one-to-one representation of a perceptual unique hue. Gegenfurtner (2003)
The major historical transition of the late twentieth century joined photoreceptor genetics, isolated-cone recordings, selective adaptation, and computable scene models. Land and McCann's Retinex account treated spatial comparisons as central to lightness and color under variable illumination Land & McCann (1971). Controlled color adaptation experiments subsequently connected psychophysical change to multiple approximately separable chromatic mechanisms Krauskopf et al. (1982). The molecular identification of photopigment genes established the biological foundations for normal and altered cone sensitivities Nathans et al. (1986a) Nathans et al. (1986b). Across these traditions, explanatory progress came from separating physically controlled input, observer judgments, and neural transformation, while testing how those levels interact rather than assuming an isolated experiment captures the entire perceptual process.
The Receptor Problem: Trichromacy and Spectral Integration
The human daylight color system depends primarily on three cone classes commonly called S, M, and L, distinguished by relatively short-, middle-, and long-wavelength-sensitive spectral response functions. Their sensitivities overlap broadly; an L cone is not exclusively a detector of red light, and an M cone is not exclusively a detector of green objects. Each photoreceptor pools absorbed photons over wavelength. For a stimulus spectral distribution E(λ) and cone sensitivity s_i(λ), the expected quantum catch is proportional to the integral of E(λ)s_i(λ) over wavelength, with optical and adaptation factors specified by the experimental apparatus. This integration discards much of the source spectrum before subsequent neural processing. Baylor et al. (1987) Stockman & Sharpe (2000)
Because three cone responses summarize a high-dimensional spectrum, many distinct spectra produce the same three-dimensional receptor signal for a given observer. This metamerism grounds practical color reproduction but also imposes an inverse problem: no unique spectral reconstruction can generally be inferred from cone excitations alone. The geometry of color matches is additionally influenced by pre-receptoral filtering from the lens and macular pigment, retinal location, adaptation, and individual variation in pigment genes. Genotype-informed psychophysical measurements refined estimates of L- and M-cone fundamentals, emphasizing that a standard observer function is a population convention rather than a precise copy of every person's visual apparatus. Stockman & Sharpe (2000) Stockman et al. (1993) CIE (2018a)
The receptor account is supported by converging evidence from in vivo measurements, primate single-cell recordings, molecular genetics, and psychophysical isolation. Baylor, Nunn, and Schnapf characterized the spectral sensitivity of primate cones directly Baylor et al. (1987); Stockman and Sharpe inferred human cone sensitivities using known genotypes Stockman & Sharpe (2000); Nathans, Thomas, and Hogness identified the genes encoding human photopigments Nathans et al. (1986a). Each technique addresses a different inferential link: cellular absorption, behavioral response, and molecular specification. Their agreement makes trichromacy among the most securely established components of visual science, while contextual color appearance remains an additional problem at later stages of the system.
Color Matching, Standard Observers, and Measurement Boundaries
A color-matching experiment asks an observer to adjust a mixture of primary lights until a test and comparison agree under controlled viewing. A three-primary representation of normal trichromatic matches provides a basis for tristimulus coordinates, even when an experimental match requires extrapolation or a negative primary coefficient in a particular algebraic system. Linear transformations of tristimulus coordinates can preserve metameric equivalence while changing the numerical geometry. The CIE colorimetric framework documents reference observer functions, standard illuminants, chromaticity coordinates, color differences, and conventions for specified fields and viewing conditions CIE (2018a). Standardization secures reproducibility across laboratories and devices only when the physical measurement conditions are actually respected.
The spatial and temporal boundaries of matching matter. A small foveal field differs from a larger field that stimulates different retinal regions; age-dependent lens absorption and macular pigment alter the spectral stimulus incident upon cones. Monitor RGB values without measured spectral output and gamma behavior are insufficient as physical stimuli. Light-emitting diodes with sharp spectral peaks may produce color matches that differ from broader-band light sources for observers with varying cone fundamentals. Measurement should therefore report spectroradiometric calibration, luminance, field size, background, viewing distance, exposure duration, adaptation protocol, and observer screening. CIE (2018a) Stockman & Sharpe (2000)
The principal conceptual boundary is between colorimetry and a model of color appearance. A coordinate difference in a CIE space is a standardized description of stimulus differences under particular observer assumptions; its perceptual significance depends on the intended metric, region of color space, display range, adaptation, surround, and task. Appearance models build additional transformations to predict correlates such as hue, brightness, chroma, and colorfulness in changes of viewing conditions Li et al. (2017). A rigorous study first establishes whether stimuli are colorimetrically comparable and then tests whether a perceptual prediction survives contextual changes. This prevents a match under one laboratory arrangement from becoming an unwarranted claim of invariant appearance.
Cone Mosaics, Adaptation, and Retinal Variability
The retinal mosaic contains spatially distributed cones whose relative abundances and spectral sensitivities vary among observers. A signal recorded from one cone indicates photopigment absorption under the prevailing adaptation state; it carries no intrinsic label identifying the reflectance of an external surface. Local comparisons across different cone classes and neighboring positions make spectral and spatial contrasts available for further processing. Retinal wiring transforms photoreceptor signals through bipolar, horizontal, amacrine, and ganglion-cell circuits before transmission through the optic nerve. Because the cones' response spectra overlap and their numbers differ across retina, color discrimination must be understood as a problem of sampling and signal comparison under noise and adaptation rather than a direct mapping from a wavelength to an elementary perceived color. Baylor et al. (1987) Conway et al. (2010)
The appearance of a fixed chromatic stimulus can drift with recent illumination because gain control adjusts sensitivity to the current range of stimulation. Adaptation can be local, chromatically selective, temporal, or jointly dependent on image structure. Its laboratory expression may include reduced sensitivity, biased appearance judgments, afterimages, or slower recovery. Cone adaptation participates in these effects, but chromatic aftereffects that depend on orientation demonstrate that additional mechanisms operate downstream of early receptors. McCollough's classic experiment found a color aftereffect contingent on the orientation of gratings used during adaptation McCollough (1965). An aftereffect is therefore evidence about the joint conditions of the adapting and test stimuli, with physiological localization requiring separate experiments.
Observer variability is not merely measurement error to be averaged away. Variations in opsin genes, lens density, macular pigment, age, exposure to light, and ocular disease can shift effective sensitivities. The genetics of L/M pigment variation has been described at the molecular level Nathans et al. (1986b) Neitz & Neitz (2011); the resulting discrimination phenotype depends on both photopigment properties and downstream interpretation of signals. Researchers must report screening and select exclusion rules according to the scientific target: a study of standard trichromacy, a study of natural variation, and a study of deficient color vision legitimately require different sampling strategies. Statistical treatment of between-observer variation belongs in the model rather than in a final comment about uncontrolled nuisance effects.
Chromatic Opponency and Cardinal Directions
Color-opponent coding begins with differences and sums of cone signals. An idealized postreceptoral description includes an approximately luminance-related combination of L and M cone excitations, an approximately L-minus-M chromatic dimension, and a dimension comparing S-cone input with the sum of L and M signals. These are useful coordinates for visual computation and selective stimulation, but biological circuits exhibit nonlinearities, adaptation, spatial filtering, and diverse weights. MacLeod and Boynton constructed a chromaticity representation in cone-excitation coordinates for isoluminant stimuli MacLeod & Boynton (1979). The value of such coordinates lies in specifying experimental directions in receptor space with greater precision than everyday red, green, yellow, and blue labels.
Krauskopf, Williams, and Heeley found that adaptation to temporally modulated chromatic fields elevated thresholds preferentially for changes along the adapting color direction, with separate sensitivity to luminance and cardinal chromatic variation Krauskopf et al. (1982). The selective threshold elevations supported the proposal of functionally distinguishable postreceptoral mechanisms. Their experimental claim was about direction-dependent sensitivity within a specified color space, measured with controlled temporal modulation. It does not imply that every perceived hue corresponds to a separate isolated retinal channel or that cardinal axes coincide exactly with the psychologically judged unique hues. The latter relationships are empirical questions about transformations between detection mechanisms and color appearance.
The opponent tradition also predicts changes in perceptual balance through chromatic adaptation and hue cancellation. Hurvich and Jameson's theory connected psychophysical cancellation settings to red–green and yellow–blue opponent response functions Hurvich & Jameson (1957). A contemporary synthesis preserves that explanatory structure while acknowledging distinct levels: cone-opponent retinal mechanisms, cardinal detection directions, cortical preferences, and the organization of conscious color judgments need not have identical tuning functions. Gegenfurtner's physiological review documents increasing diversity in chromatic selectivity from retina and lateral geniculate nucleus into cortex Gegenfurtner (2003). Claims of a universal isolated 'red neuron' or a fixed neural code for every basic color ignore that progression.
Chromatic Discrimination and Detection Thresholds
Threshold measurements ask when a chromatic variation can be detected or differentiated under specified conditions. A two-alternative forced-choice task can reduce dependence on explicit yes–no response criteria, while a yes–no detection task requires attention to both sensitivity and decision bias. Psychometric functions relate stimulus distance to response probability; their estimates depend on stimulus sampling, guessing rate, lapse rate, and fitted slope. A just-noticeable difference is consequently a task-conditioned threshold estimate rather than an intrinsic fixed separation between any two named colors. The direction and location of a difference in color space, stimulus area, luminance, noise, and surrounding scene all affect its measured detectability. Krauskopf et al. (1982) CIE (2018a)
A demanding study controls isoluminance explicitly. A pair of stimuli with equal standard photometric luminance may not be individually equiluminant for all observers, and spatial chromatic variation can generate luminance artifacts on a monitor with uncorrected display characteristics. Inferences about color-specific processing therefore require appropriate luminance controls, counterbalancing, and sensitivity checks. Adaptation protocols also change the operating point of the mechanisms under study. Findings that a threshold is elevated after exposure along one chromatic direction are evidence of selective sensitivity loss in that protocol, not proof of a single anatomical locus. Krauskopf et al. (1982) MacLeod & Boynton (1979)
Discrimination thresholds and perceived color differences should be analyzed together where possible, but they are not identical constructs. Fine discrimination can occur between two stimuli assigned the same basic name; two stimuli carrying different names can still be difficult to discriminate. Standardized color-difference formulas organize industrial tolerances, whereas laboratory psychophysics estimates perceptual performance for an actual observer population. Studies should specify whether they test an objective forced-choice accuracy, a subjective degree of difference, a color-category boundary, or a cross-context appearance match. Comparisons across laboratories should preserve these endpoint distinctions and report uncertainty intervals rather than transferring a threshold without its measurement context. CIE (2018a) Witzel & Gegenfurtner (2018)
From Retina to Visual Cortex
Cone and retinal-opponent signals enter the thalamic lateral geniculate nucleus and then reach primary visual cortex, where chromatic information is represented together with spatial, orientation, contrast, and other visual features. Early visual cortical neurons can respond to chromatic contrasts with selectivities beyond those predicted by the canonical cardinal axes. The cortex contains distributed populations whose chromatic tuning depends on the stimuli, the visual field, and the task. This makes a purely retinal account insufficient for explaining color appearance across complex objects and scenes. Gegenfurtner's review distinguishes the well-characterized early channels from richer response patterns in cortical processing Gegenfurtner (2003).
Conway's review of color signals through dorsal and ventral pathways emphasizes color-relevant activity beyond primary visual cortex Conway (2014). This evidence modifies a simplistic anatomical division between a dedicated color module and a wholly color-blind processing pathway. Color signals contribute to object recognition, material interpretation, attention, and action, while neural recording and imaging measure different aspects of those functions. A cortical region's sensitivity to color variation does not establish its exclusive responsibility for hue experience; causal evidence from focal lesions, stimulation, and targeted behavior is required to distinguish necessity from covariation. Perceptual judgments also recruit recurrent processing and task-dependent feedback, so anatomical pathways need to be interpreted in relation to experimental demands.
Multivoxel neuroimaging measures distributed responses rather than single-neuron hue codes. Brouwer and Heeger demonstrated decoding and reconstruction of color information from patterns of activity in human visual cortex Brouwer & Heeger (2009). Such representational decoding shows that color-related information is available within a population under particular measurement and modeling assumptions; it does not by itself determine whether the decoded pattern is the mechanism producing a specific subjective appearance. The appropriate inference combines psychophysical adaptation, neural selectivity, causal perturbations where feasible, and computational prediction across stimuli. Advances across these techniques have helped move color science from isolated cone signals toward natural object appearance Conway et al. (2010) Gegenfurtner (2025).
Genetic Variation, Dichromacy, and Altered Color Vision
Congenital red–green color vision deficiencies commonly arise from variation in the X-linked long- and middle-wavelength opsin gene array. Altered pigment spectra, absent functional cone classes, and atypical gene arrangements can change the discriminability of some spectral variations. Distinguishing protan and deutan pathways requires appropriate measurements rather than interpreting all red–green difficulty as one homogeneous phenotype. Genetic analyses in the 1980s provided direct evidence connecting variation in photopigment genes to inherited color-vision differences Nathans et al. (1986b). Later synthesis integrates genetic, retinal, and psychophysical diversity in normal and defective color vision Neitz & Neitz (2011).
Screening procedures test particular characteristics. Pseudoisochromatic plates can efficiently detect common red–green problems but do not fully estimate every observer's thresholds or color appearance. Anomaloscope matching and carefully designed threshold tests provide complementary information about photopigment-related matching ranges and discrimination. Illumination spectra, instruction comprehension, age, and ocular status matter for interpreting results. A change on a screening test is not automatically evidence that a person perceives the world in a particular way, and one can have anomalous trichromacy with a pattern of difficulties distinct from dichromacy. Research describing 'color blindness' should therefore specify the affected mechanisms, experimental endpoint, and observation conditions. Neitz & Neitz (2011) CIE (2018a)
Gene replacement in adult dichromatic monkeys provides an important test of neural plasticity. Mancuso and colleagues introduced an additional photopigment class and reported new behavioral evidence of red–green chromatic discrimination in adult animals Mancuso et al. (2009). The experiment challenges the proposition that useful postreceptoral processing of an added cone signal necessarily requires normal early-life trichromatic experience. It remains a primate intervention with its own vector, species, behavioral measures, and safety context. It should not be converted into a claim of established routine correction of human congenital color deficiency. The underlying finding bears on the adaptability of receptor-to-behavior mapping, while clinical applicability requires independently verified human evidence, longer follow-up, and regulatory assessment.
Luminance, Chromatic Channels, and Visual Tasks
Color processing operates alongside the detection of edges, textures, and brightness contrast. A chromatic boundary can separate surfaces with little luminance contrast; a luminance boundary can remain visible when chromatic differences are negligible. Their relative contributions depend on spatial frequency, motion, viewing distance, illumination, and the task. An experiment that isolates chromatic signals must establish how its luminance artifacts were controlled and how its display was calibrated. A natural scene, by contrast, often combines spectral and luminance cues that jointly support stable object and material recognition. The challenge is to model these interactions rather than presuppose that natural perception is a sum of independent laboratory features. Gegenfurtner (2003) Witzel & Gegenfurtner (2018)
Adaptation of luminance sensitivity differs from the selective effects of cardinal chromatic modulation in the classic experiments by Krauskopf and colleagues Krauskopf et al. (1982). Their results motivate functional separation under specific stimulus conditions. Yet cortical receptive fields mix color and other spatial attributes in a diversity of response patterns Conway (2014). A task emphasizing fine chromatic discrimination may rely heavily on particular postreceptoral directions, whereas sorting colored objects in a cluttered setting can depend on spatial grouping, attention, and acquired category knowledge. Research should therefore state the proximal dependent variable and the conditions needed to move from a detection result to a broader psychological account.
An integrated model must also preserve the difference between the visual detection of light and the many non-image-forming effects of illumination. The CIE's framework for intrinsically photosensitive retinal ganglion cell–influenced responses addresses quantities relevant to biological responses to light CIE (2018b). These functions participate in entrainment and pupil and other physiological responses, but they cannot be substituted for cone-based chromatic discrimination or hue appearance. Spectral lighting interventions may alter both classes of response, making careful stimulus specification and separate outcomes essential. This separation has practical importance for visual ergonomics, lighting research, and claims about psychological effects of colored illumination.
The Psychophysics of Hue, Lightness, and Perceived Colorfulness
Color appearance research measures attributes that become meaningful through visual comparison. Hue matching compares the quality of chromatic experience, lightness matching estimates perceived surface reflectance relative to reference conditions, and brightness matching estimates the apparent intensity of light. Reports of saturation and colorfulness add relational dimensions that may covary with adaptation level and viewing surround. A complete experiment specifies instructions, response device, background, geometry, and the temporal sequence of comparison. Without those conditions, a verbal claim that a sample is 'more intense' is ambiguous between luminance, brightness, chroma, colorfulness, and emotional salience. The CIE documents the measurement conventions behind relevant colorimetry and appearance correlates CIE (2018a).
The visual system transforms cone excitations into contextual judgments. Unique-hue settings, for example, identify perceptual judgments of red without perceived yellow or blue components, or analogous settings for other categories. They need not line up exactly with isolated cone-opponent cardinal directions. A neutral setting represents an observer's current criterion for achromaticity and changes with adaptation, field, surround, and experimental method. Bosten, Beer, and MacLeod measured unique-white settings and found informative variation across directions in color space Bosten et al. (2015). This evidence cautions against identifying white perception with a single universal null point of cone opponency, while supporting further investigation into environmental calibration.
The distinction between stimulus difference and perceptual difference matters whenever perceptual color spaces are used for scientific inference. Formulas that map spectral or tristimulus quantities to distances organize measurements, but a perceptual metric should be evaluated against observer judgments across relevant materials and backgrounds. An approximate uniformity claim applies to the conditions and tasks used for its validation. Evaluating a digital design, a clinical color-screening image, and a research chromatic threshold task with the same unqualified RGB difference would erase essential distinctions among colorimetric signal, discrimination probability, and contextual appearance. Li et al. (2017) CIE (2018a)
Chromatic Adaptation and the Von Kries Tradition
Chromatic adaptation reflects the change in visual responses and appearance after exposure to altered spectral stimulation. The traditional von Kries account models adaptation as a change in the relative gains assigned to cone signals. This formulation is mathematically tractable and predicts a class of transformations between receptor coordinates under different illuminants. The theory acquires empirical substance when the change in appearance is measured by matching an object presented in one context to an object in another. Such asymmetric color matches distinguish stimulus equivalence from perceived equivalence across illumination. Brainard & Wandell (1992)
Brainard and Wandell used controlled simulated illuminants and object displays to study asymmetric matches and found that, for the tested conditions, the mappings were well approximated by diagonal linear transformations of cone coordinates Brainard & Wandell (1992). This observation supports gain-change approximations for a substantial experimental domain. Its scope is the set of stimuli, simulated surfaces, visual contexts, and observers tested. It does not establish that all color constancy can be generated by three independent gain factors without spatial, inferential, or categorical contributions. An adequate model must predict matched appearance over both incremental changes and substantial rearrangements of illumination and scene structure.
Temporal adaptation interacts with spatial context. The same local cone response can support different perceived colors when neighboring regions change, and a previously adapted observer may set a different match from an unadapted one. Color appearance research therefore benefits from factorial designs that independently vary adaptation time, illuminant spectrum, surrounding reflectance, and task instructions. Reported gains should be distinguished from the downstream mechanisms producing contextual effects. Wandell's analysis of illumination and spatial pattern describes color appearance through successive transformations combining coordinate changes and response scaling Wandell (1993), illustrating why a simple receptor-gain model may be a component of a larger processing account.
Simultaneous Contrast, Induction, and Spatial Organization
A target patch does not have a context-free appearance. Surrounding colors change apparent hue, saturation, lightness, and brightness through processes collectively described as induction and simultaneous contrast. A gray region embedded in a reddish surround can acquire a different chromatic appearance than the same spectral target in a neutral surround. To investigate the mechanism, researchers vary spatial frequency, target size, contour arrangement, adaptation field, and luminance while measuring target settings. Contrasts attributed to local interaction must be distinguished from effects of changing the inferred illumination or the expected material identity of the target. Wandell (1993)
Spatial comparison mechanisms may help interpret surfaces under complex illumination. Land and McCann proposed that color and lightness judgments emerge from comparisons across spatial patterns, placing the spatial structure of scenes at the center of theoretical explanation Land & McCann (1971). Such a theory is valuable in specifying algorithmic operations and experimental predictions. Its success in explaining selected patch patterns should be assessed independently from whether it captures the appearance of natural scenes with three-dimensional shapes, gloss, mixed light, transparency, and familiar materials. A retina-wide operation cannot simply be assumed to supply every scene-level perceptual inference.
When a study reports an induction magnitude, the result requires a precisely identified reference and response measure. A shift in achromatic adjustment is not the same measurement as a shift in categorical color naming, and a changed luminance-match setting is different again. Selective adaptation and chromatic spatial contrast can interact with general visual attention and task demands. Any attempt to reconstruct an 'intrinsic' color of the patch from a single reported RGB value conflates externally measured stimulation with observer-dependent inference. A credible theoretical account predicts how appearance varies when neighborhood, segmentation cues, and illuminant evidence are independently manipulated Foster (2011) Witzel & Gegenfurtner (2018).
Color Constancy Across Changes of Illumination
Color constancy refers to the relative stability of perceived surface color when the spectral composition or intensity of the light source changes. It is a graded empirical phenomenon rather than a guarantee of perfect identity. Researchers can ask observers to match the appearance of the same object under distinct illuminants, judge whether it is the same material, classify colors into categories, or adjust a reference patch until an object appears achromatic. These tasks produce different operational measures. Foster reviews the experimental literature and emphasizes that the outcome depends on the task, available scene information, surfaces, and magnitude of illumination change Foster (2011).
A central ambiguity is that the retinal spectrum mixes reflectance and illumination. Maloney and Wandell proposed computational recovery of surface spectral reflectance under restricted assumptions about the dimensionality of lights and reflectances Maloney & Wandell (1986). Such constraints make aspects of the inverse problem solvable, while unqualified exact recovery from arbitrary spectra remains impossible. Psychophysical evidence of successful surface matching does not imply that human observers literally recover a complete reflectance spectrum. More modest representations—relative cone excitations, ratios among surfaces, estimated illuminant properties, and material categories—can support behavior under many viewing situations.
Brainard's study of achromatic loci in nearly natural images examined how color constancy generalizes beyond highly simplified stimuli Brainard (1998). Naturalistic scenes bring extra informative regularities and additional uncontrolled complexity. Reflectances and illuminants in the environment are statistically structured, and objects persist across changes in viewpoint and light. A research design should separate the effect of adding realistic context from changes in image contrast, surface layout, or display dynamic range. It should also measure individual differences and report the specific constancy index or response transformation used. A stronger appearance theory predicts new scenes rather than fitting a single set of color matches.
Categorical and relational constancy provide complementary evidence. Olkkonen, Witzel, Hansen, and Gegenfurtner examined categorical color constancy with real surfaces Olkkonen et al. (2010). Categories can remain stable while fine-grained perceptual matches change, so categorical success cannot establish exact perceptual invariance. Foster and later reviews emphasize the need to separate signal-level information from tasks involving objects and material interpretation Foster (2011) Witzel & Gegenfurtner (2018). The resulting account is layered: receptor encoding defines the available information, adaptation transforms it, spatial relations constrain plausible illuminants, and behavioral goals determine which degree of stability is functionally sufficient.
Surface Color, Materials, and Object Identity
Objects appear colored in ways that depend on their physical materials and visual contexts. A matte painted tile, a polished metallic surface, a translucent fruit, and a fluorescent sign may share approximate tristimulus coordinates at one viewing point while presenting different characteristic variations across position, illumination, and motion. Their appearances carry clues about surface reflectance, gloss, scattering, and illumination. Research that seeks to explain object-color perception must therefore consider distributions of colors across an object's image rather than restricting analysis to the value of one isolated pixel. Witzel and Gegenfurtner synthesize evidence relating color appearance to material and object recognition Witzel & Gegenfurtner (2018).
The same apparent object color can emerge from distinct physical combinations of surface, light, and viewing geometry. Recognizing a familiar object across changes of illumination may be aided by prior experience, shape, texture, location, and scene coherence. Importantly, those cues can be experimentally separated: an unfamiliar material avoids some memory effects, a randomized illuminant challenges learned assumptions, and a physically rendered or measured hyperspectral scene constrains the incoming signal more directly than conventional RGB. A computational model should state which properties it estimates and how performance is evaluated, whether by surface identification, appearance match, or physical reflectance recovery. Maloney & Wandell (1986) Hurlbert & Yu (2025)
Gegenfurtner's 2025 review proposes an object-level perspective in which distributions of color coordinates across an object contribute to the characterization of its appearance Gegenfurtner (2025). This is a synthesis of behavioral, neural, and computational evidence rather than a replacement for spectrally calibrated experiments. Object-scale models gain explanatory value when they predict effects of changing material boundaries, three-dimensional illumination, and local color distributions independently of trivial average differences. The move toward object-centered perception also clarifies the research boundary of this dossier: general visual object recognition is a neighboring domain; color-specific evidence and appearance judgments define the present scientific object.
Memory Colors and Prior Knowledge
Perceived color can be influenced by expectations about familiar objects. Hansen, Olkkonen, Walter, and Gegenfurtner asked observers to adjust the colors of familiar fruit images until they appeared achromatic. Neutral settings were displaced in a direction opposing the object's typical color, suggesting that stored object-color knowledge influenced appearance judgments Hansen et al. (2006). The experimental result concerns a well-defined matching task involving familiar shapes, with the important possibility of response and decision contributions evaluated through controls. It provides evidence for high-level context in color appearance but does not imply unlimited top-down alteration of early sensory responses.
Memory-color experiments face a general inferential challenge. An observer may visually experience a changed color, adopt a different criterion for what counts as neutral, or anticipate what a familiar object 'should' look like while carrying out the task. Different response methods and manipulations of familiarity help discriminate these possibilities. A stronger design compares known versus unfamiliar object forms, rotates or changes semantic identity while preserving color statistics, and controls visual shape and response instructions. Convergent evidence from adaptation and perceptual discrimination may strengthen claims about sensory modulation relative to report bias. The target is a mechanism-specific explanation of the interaction of knowledge and appearance, not a declaration that language or memory overrides the retina.
The influence of object knowledge is compatible with stability in some perceptual dimensions and change in others. An observer can maintain accurate forced-choice discrimination while shifting neutral-point settings, or maintain a category label while experiencing changes in saturation. The scientific synthesis therefore treats object identity as an experimentally manipulable source of expectations, with distinct potential effects on encoding, attention, response criterion, and categorical judgment. Reviews of object color and constancy provide the broader context in which these tests belong Witzel & Gegenfurtner (2018) Gegenfurtner (2025).
Illumination as a Perceived Property
The color of a surface and the color of its illumination are coupled through the reflected spectrum. Yet observers can sometimes describe a scene as lit warmly, illuminated by daylight, or tinted by an artificial source, indicating that illumination itself can function as a perceptual estimate. This creates a two-way scientific question: how does perceived light color affect surface-color appearance, and what evidence about surfaces allows an observer to judge illuminant color? A model that treats the light source merely as a nuisance parameter may overlook perceptual representations of illumination as an object of judgment in its own right. Hurlbert & Yu (2025)
Modern experiments can render controlled spectral changes in realistic imagery and ask observers to identify plausible scenes, match object colors, or directly compare lights. The 2025 review by Hurlbert and Yu describes how natural illumination measurements and the growing capacity to create novel light spectra have sharpened this research problem Hurlbert & Yu (2025). The appearance of illumination interacts with color constancy, prior expectations, visual adaptation, and contextual relationships. The same physical light manipulation can change both apparent surface color and the judged source of that change. Mechanistic explanation benefits from recording both sets of outcomes rather than inferring one from the other.
Research on visible chromatic appearance should additionally distinguish conscious judgments about illumination from the biological effects of light exposure measured in chronobiology and photobiology. The CIE ipRGC-focused metrology standard quantifies stimulation relevant to several non-image-forming pathways CIE (2018b). Such physiological responses can be triggered by the same spectral manipulation used to study color constancy, but they require different models, exposure metrics, and endpoints. This separation becomes especially important in claims about lighting that allegedly improves behavior or well-being: visual appearance, circadian timing, alertness, and emotional evaluation should be measured and reported as distinct constructs.
Color Categories and the Organization of Naming
Color categories divide a continuous space of discriminable visual differences into useful regions for recognition and communication. Languages vary in the number and boundaries of their conventional terms; individual speakers also vary in naming consistency. The psychological problem is how a categorical system emerges from constraints imposed by sensory discrimination, environmental statistics, learned communicative practices, and task demands. A named category is a linguistic convention over perceptions, while a discrimination threshold is an operational property of a task. They can influence each other but must be separately measured. Regier et al. (2007) Witzel & Gegenfurtner (2018)
Regier, Kay, and Khetarpal modeled color-naming systems as partitions of color space and investigated whether attested naming patterns reflect communicatively effective divisions Regier et al. (2007). Such models combine a formal objective with assumptions about perceptual distance and possible category systems. Their predictions are evidence about the structure of color-language conventions, not evidence that every community shares identical boundaries or that a particular language is uniquely optimal. Cross-cultural research also must account for sampling, natural lighting, materials, display calibration, language proficiency, and culturally relevant use of color words. The relationship between visual biology and lexicon is therefore a problem of constrained variation, not an argument that either one determines all other levels.
A later experimental program measures how word boundaries affect search and response time when colorimetric differences are controlled. The key issue is whether category alignment adds predictive value beyond physical distance, familiarity, and response mapping. The task design matters: naming a color explicitly recruits lexical retrieval; speeded odd-one-out discrimination makes different demands; simultaneous visual search introduces competition and attentional selection. A category effect observed under one protocol cannot establish a permanent alteration in elementary photoreceptor sensitivity. More recent cross-lingual work continues to examine category effects in visual search and the interaction of long-established and newly acquired lexical boundaries Wakui et al. (2025).
Language Effects on Color Discrimination
Winawer and colleagues compared Russian and English speakers using a speeded discrimination task for shades of blue. Russian distinguishes lighter and darker blues with common basic terms, and Russian participants showed an advantage for some pairs that crossed the lexical boundary compared with pairs within a category. Importantly, verbal interference disrupted the effect under the tested conditions Winawer et al. (2007). The experiment gives direct evidence that linguistic resources can alter response performance on color tasks. Its interpretation should remain tied to the operational endpoints: accuracy, reaction time, category placement, and the role of a concurrent verbal task. It does not demonstrate that the retinal responses to those stimuli had changed.
A cross-linguistic difference may reflect several pathways. Words can shift attentional selection toward diagnostic differences; category memory can change how reference stimuli are retained; task strategies can recruit overt or covert naming; and learned perceptual organization may yield longer-lasting effects. Distinguishing these mechanisms requires manipulations that target verbal access, perceptual exposure, retention intervals, and response criterion separately. Linguistic effects may be larger near established category boundaries and for difficult rather than easy discriminations, as the original Russian-blue experiment reported Winawer et al. (2007). Methodological triangulation should reproduce the main effect across laboratories, languages, and nonlinguistic tasks before drawing general conclusions about language's influence on color perception.
A newer literature explores category formation both behaviorally and with artificial network models. Research on artificial visual systems has shown that category-like structure can arise partly from visual training objectives while also reflecting linguistic training in multimodal systems Akbarinia (2025). These models supply testable analogies and computational hypotheses; they do not identify the cognitive process in human observers merely because a network predicts part of a behavioral dataset. Separately, studies of color concepts among congenitally blind individuals and language statistics support the role of verbal knowledge in acquiring color associations that extend beyond direct visual experience Liu et al. (2025). The ability to know linguistic associations about color and the ability to discriminate color visually are distinct psychological outcomes.
Individual Differences, Development, and Environmental Exposure
Variation in color vision begins with receptor genetics and ocular optics and extends to attentional strategies, experience, environmental statistics, and learned naming systems. Individuals can differ in cone ratios, lens transmission, macular pigment, chromatic discrimination, neutral settings, and the consistency of their verbal categories. The sensitivity of a given measurement to these sources varies. Genotype-informed cone fundamentals clarify early biological variation Stockman & Sharpe (2000), and molecular work explains important inherited color-vision differences Neitz & Neitz (2011). Researchers should therefore avoid treating variation across adults as a single dimension of general visual ability.
Developmental claims require particular care because an observed shift across age can reflect changes in receptor function, attention, language, response training, or the conditions under which stimuli reach the retina. The development of a color lexicon is neither identical to the maturation of cone-based discrimination nor a simple chronological transformation from one stage into another. An appropriate longitudinal design tracks the same observers with age-suitable tasks, calibrated stimuli, and stable definitions of the dependent variables. Cross-sectional differences alone cannot isolate developmental mechanisms from cohort, educational, or environmental effects. Earlier environmental explanations linking ocular exposure to color naming illustrate how optical and social variables can be jointly hypothesized but require careful causal testing Lindsey & Brown (2002).
Acquired changes in color performance may arise from ocular, retinal, neural, or environmental factors and should be described by verified functional measures. A screening test intended for inherited red–green deficiencies has different validity from an instrument for acquired chromatic loss or a laboratory appearance match. Population prevalence depends on how the phenotype and denominator are defined; broad percentages should not be transferred across ancestry, age, or diagnostic conventions. This dossier treats the sensory and psychophysical organization of color experience; diagnosis and clinical management belong in specialized clinical research under current professional standards. Neitz & Neitz (2011)
Color Appearance Models and Engineering Standards
Color appearance models use colorimetric inputs, adaptation assumptions, viewing-surround information, and nonlinear transformations to predict perceptual correlates under specified conditions. Their task is predictive: if the same observer sees a display in a dim surround and a print under a bright illuminant, a successful model predicts aspects of the difference in appearance. Models are evaluated against matching and scaling datasets rather than being established solely by algebraic elegance. The CIE's Colorimetry, 4th Edition describes standard observers, illuminants, tristimulus values, color differences, and additional appearance-model material CIE (2018a). Its institutional status makes it a central measurement source but not an empirical guarantee for every ecological viewing situation.
CAM16 and its associated chromatic adaptation transform and approximately uniform color space are examples of engineered prediction systems developed to improve stability and applicability relative to earlier model families. Li and colleagues introduced CAM16, CAT16, and CAM16-UCS in their 2017 technical treatment Li et al. (2017). The correct scientific reading is that the models offer a particular calibrated mapping with tested predictive aims. Their success depends on the validation conditions and their input assumptions. A CIE model, a psychophysical detection model, and a neural model may use different intermediate representations without being logically contradictory, because they aim at different endpoints.
A color-appearance implementation should explicitly identify the illuminant white point, adaptation conditions, luminance scaling, surround, and expected observer. When spectra are available, integration over the appropriate sensitivity functions precedes colorimetric transformation. When only image RGB values are available, device characterization and a transfer function are required to derive a meaningful stimulus estimate. Tests of generalization should reserve data from unseen illumination conditions and material sets, compare perceptual-prediction errors against simple baselines, and report where the model fails. Successful numerical agreement for standard patches does not settle more complex perceptual questions such as material classification, illumination perception, or memory-color effects. CIE (2018a) Hurlbert & Yu (2025)
Natural Statistics, Computational Inference, and Color Constancy Models
The physical environment contains regularities that can help constrain color estimation. Natural illuminants, reflectance spectra, and scene layouts are not arbitrary combinations selected independently at every pixel. Maloney and Wandell used constraints on spectra to describe a computational method for estimating surface reflectance under unknown lighting Maloney & Wandell (1986); a related analysis examined the adequacy of low-dimensional reflectance representations Maloney (1986). These models show how assumptions can reduce the ambiguity created by trichromatic encoding. The strength of an inferred color therefore depends on both input information and the appropriateness of priors or structural constraints.
Computational explanations should distinguish recovering physical variables from predicting observer responses. A system can estimate reflectance accurately but predict perceived appearance poorly when adaptation or context is mismodeled. Conversely, a machine can reproduce human color category choices without reconstructing the true illuminant spectrum. Studies of color constancy benefit from reporting performance against physical ground truth as well as observer matches when both are available. A comparison of algorithms should hold viewing and evaluation conditions constant and address scene distributions outside the training sample. Foster (2011) Gegenfurtner (2025)
A 2024 study on color expectations across illumination changes examined naturalistic versus artificial spectral manipulations and found evidence that observers use relational structure and prior expectations about surface changes Karimipour & Witzel (2024). These findings are valuable because they test perceptual strategies under richer scene statistics than isolated patches. They are also task- and stimulus-dependent: plausibility judgments about candidate rendered scenes do not establish exact recovery of material spectra. A mature research program combines controlled perturbations of scene statistics, perceptual matches, physically measurable spectra, and models that predict new material and illumination combinations without readjusting to each test scene.
Artificial Color Vision and Cross-System Comparisons
Artificial visual systems can serve two distinct purposes in color science. They can be engineering tools for estimating illumination, identifying surfaces, or performing color correction; they can also serve as computational probes for hypotheses about how task demands shape representations. An artificial model's RGB classification accuracy is not equivalent to human cone-based discrimination, because its sensors, spectral preprocessing, training objectives, and loss functions may differ. A proper comparison presents the same physical or explicitly matched spectral conditions where possible and measures corresponding behavior using compatible metrics. The geometry of an embedding or a categorical boundary supplies a candidate computational explanation but does not by itself establish a shared visual experience. Akbarinia (2025) Gegenfurtner (2025)
Research using networks trained for object recognition and multimodal tasks has explored how color categories appear in learned representations Akbarinia (2025). Such studies can manipulate training objectives more systematically than human developmental history permits, yielding mechanistic hypotheses about the role of classification, semantics, and three-dimensional scene tasks. Their conclusions are limited by architecture, training distribution, and the empirical human dataset chosen as a target. Convergence with human judgments is strongest evidence for functional similarity on the tested task, while divergences can generate new experimental predictions. This domain belongs to cross-system functional comparison within Artificial Era research, with all claims about subjective appearance reserved for evidence capable of testing them.
The material importance of cross-system comparison grows in digital environments where image-generation systems, displays, cameras, and human viewers interact. A generated picture can preserve color-category labels and still fail in appearance under a changed display white point; a camera can produce metrically stable scene values while a viewer sees an altered material; a color-corrected image can optimize reproduction for the standard observer while giving some users a poorer match. Scientific cooperation between human psychophysics and engineered systems requires explicit target quantities, calibrated stimuli, and uncertainty estimates. It is precisely this operational discipline that allows psychological theory and artificial computation to inform one another without converting analogy into an untested identity claim.
Preference, Affect, and the Limits of Color Psychology
The visual appearance of colors can contribute to preference and affective evaluation, but psychological effects depend on objects, experience, social meaning, task, and culture. Palmer and Schloss's ecological valence theory proposes that preferences for colors may be partly related to affective associations with objects that characteristically display them Palmer & Schloss (2010). This hypothesis is testable with preference ratings and object associations. It should not be generalized into a universal table in which every occurrence of one hue produces the same emotion or behavioral outcome. Preference is a distinct dependent variable from wavelength discrimination, hue matching, visual attention, and biological effects of light exposure.
Experimental claims about color and behavior require controlled illumination, comparable luminance and saturation, randomized presentation, culturally appropriate stimuli, and prespecified outcomes. A laboratory reaction-time effect may reflect contrast, visibility, semantic association, or task instructions rather than an autonomous emotional action of 'red' or 'blue.' The evidence base for preferences and associations should be organized around particular hypotheses and designs, with careful distinction between correlational familiarity and experimental effects. Perceptual foundations establish what visual signal observers receive; they do not automatically determine the symbolic or emotional meaning assigned to that signal. Palmer & Schloss (2010) Witzel & Gegenfurtner (2018)
Applied design decisions therefore benefit from combining appearance measurements with empirical user testing. A color-coded warning, visual interface, or data graphic should be evaluated for contrast and discriminability among target observers, including people with color vision deficiencies, and for comprehension of the intended meaning. Accessible design is an outcome of actual task performance under expected viewing conditions, not merely conformity to an attractive palette. This distinction gives color science a direct practical role in communication, safety, and human factors while keeping speculative claims about emotional control outside the established explanatory core. CIE (2018a) Neitz & Neitz (2011)
Experimental Methods and Standards of Inference
The choice of experimental paradigm determines the kind of claim that color research can support. Color matching establishes equivalences between test and comparison stimuli for a specified observer and condition. Chromatic threshold experiments estimate discriminability with probabilistic responses, while hue and lightness adjustments estimate particular appearance correlates. Illumination-change experiments examine the consistency of object judgments or perceptual matches across contexts. Physiological recordings characterize photoreceptor or neural responses, neuroimaging assesses distributed activity, and molecular assays identify components of the pigment machinery. These instruments provide distinct and complementary evidence. A coherent scientific explanation predicts the relationships among them instead of inferring subjective appearance from a single instrument. CIE (2018a) Conway et al. (2010) Foster (2011)
A well-specified color experiment begins with calibration. For light sources and reflective objects, measured spectral power distributions and spectral reflectances are preferable to uncharacterized device RGB settings. In display work, measurements should encompass primaries, transfer functions, viewing geometry, luminance, black level, spatial and temporal uniformity, and possible adaptation effects. The retinal stimulus depends on pupil size, pre-receptoral filtering, and field location. Observer recruitment should describe the screening methods and include repeated measurements sufficient to characterize intra-observer variation. Stimulus randomization, counterbalancing, masking of experimental expectations where feasible, and a preregistered primary comparison reduce avoidable inferential latitude. The selected apparatus must allow accurate rendering of the chosen chromatic directions rather than relying on a post hoc claim that two nominal digital values were perceptually comparable. Stockman & Sharpe (2000) CIE (2018a)
Data analysis must separate discrimination, judgment criteria, and perceptual matching. In a forced-choice task, a fitted psychometric curve can report threshold and slope, with uncertainty derived using appropriate observer-level sampling or hierarchical models. In an asymmetric match, the relevant outcome may be a transformation of cone coordinates, a perceptual difference, or a constancy index that needs an exact formula. Behavioral reports must be checked for repeatability, changes caused by instructions, and asymmetries between comparison directions. Neural or computational correlations are stronger when they predict held-out stimulus sets, but prediction alone does not establish a unique causal mechanism. The use of quantitative metrics should therefore be accompanied by their operational meaning and by realistic estimates of measurement uncertainty. Brainard & Wandell (1992) Brouwer & Heeger (2009) Foster (2011)
The strongest research designs test explicit alternatives. To evaluate color constancy, one can manipulate the illuminant independently of the surface spectral reflectance and compare a cone-gain model with a model incorporating spatial relations or priors. To test memory-color effects, one can preserve low-level image statistics while changing the recognition of a familiar object Hansen et al. (2006). To test language, one can compare within- and across-category discrimination under verbal interference Winawer et al. (2007). Each manipulation narrows the class of plausible explanations while retaining limits to generalization. Such comparative designs yield more informative conclusions than studies that demonstrate an attractive effect without specifying what would have counted against the preferred mechanism.
Evidence Quality, Replication, and Generalizability
The strongest claims in the field concern the existence of broadly overlapping S-, M-, and L-cone classes and the central role of their integrated signals in photopic color matching. These claims have independent physiological, molecular, and psychophysical support Baylor et al. (1987) Nathans et al. (1986a) Stockman & Sharpe (2000). The evidence for opponent coding is also extensive, although specific channel coordinates and their relation to unique-hue appearance depend on the organism, neural stage, stimulus, and response task Hurvich & Jameson (1957) Krauskopf et al. (1982) Gegenfurtner (2003). Color constancy has a substantial behavioral literature with reproducible partial stability under illumination changes; the size and mechanism of the effect vary considerably with the paradigm, scene content, and chosen constancy metric Foster (2011).
Individual studies must be read at the scale of their causal leverage. Direct recording of cone absorption establishes photoreceptor sensitivity under the experiment; it does not establish how natural objects appear. Randomized adaptation or verbal-interference interventions support an influence of the manipulated condition on the specified behavioral endpoint, but do not automatically isolate the entire neural pathway mediating the change. Observational comparisons of populations and language communities can identify structured associations while remaining vulnerable to correlated environmental and cultural differences. The primate gene-transfer experiment Mancuso et al. (2009) demonstrates a causal manipulation of receptor complement in a specific animal setting; its scope is not a human clinical efficacy claim.
A narrative synthesis can compare theoretical predictions and distinguish designs, yet its source selection is not an exhaustive census of the literature. The present account relies on traceable original studies and influential reviews chosen to represent receptor physiology, colorimetry, adaptation, neural responses, natural scenes, and cognitive context. It makes no claim to a registered systematic search, formal bias rating across every paper, or statistical aggregation of effect sizes. Numerical sample sizes, confidence intervals, and precise effect estimates are intentionally omitted where they have not been checked in the primary full reports. Future release-level scientific review should address publication selection, unpublished replications, cross-laboratory robustness, and the evolving status of each model. The central evidential discipline is to distinguish a well-supported processing constraint from a theory-specific explanation of that constraint.
Generalization is a recurring problem because research stimuli are often smaller, simpler, and more tightly controlled than visual scenes. Patch matching may preserve clean colorimetric inference while excluding material properties, scene segmentation, and naturalistic illuminants. Immersive or real-object experiments restore these factors but may introduce difficult calibration and participant-behavior variability. Models and studies should be evaluated through predictions for novel observers, lighting spectra, reflectance sets, and contextual arrangements. Disagreement across paradigms is informative when it identifies which function was measured, rather than being dismissed as noise. Olkkonen et al. (2010) Hurlbert & Yu (2025) Gegenfurtner (2025)
Competing Explanations and Unresolved Scientific Problems
Color appearance can be modeled as a sequence of response transformations, a set of spatial comparisons, an estimation process constrained by priors, a population code distributed across cortical circuits, or some combination of these operations. These approaches often target different explananda. Cone-gain transformations predict classes of adaptation matches Brainard & Wandell (1992), whereas spatial and scene-based theories seek to explain why surfaces can appear stable in complex images Land & McCann (1971) Foster (2011). Computational inverse models use physical constraints to estimate plausible reflectances Maloney & Wandell (1986). A meaningful theoretical comparison specifies common stimuli and perceptual outputs on which the models offer distinct predictions. Formal compatibility at one level does not establish a mechanism's necessity at another.
One unresolved question concerns the neural bridge between early chromatic channels and the perceptual organization of hue. Physiological opponent axes can be specified precisely in cone coordinates, but the positions of perceptual unique hues and achromatic loci need additional transformations and are variable across observers Bosten et al. (2015) Gegenfurtner (2003). Another concerns how cortical populations represent surfaces, illuminants, and object identity. Pattern decoding from neural activity establishes accessible information, while a causal model must explain when that information changes experience or behavior Brouwer & Heeger (2009). Research combining neural perturbation, adaptation, and matched behavioral outcomes is needed to connect these inferential steps.
The stability of perceived surface color remains a graded phenomenon. An observer's performance on material identification may be excellent while exact perceptual settings differ across illumination. The statistical structure of realistic scenes may make some constancy strategies effective and others misleading; unusual LED spectra or artificial rendering can stress assumptions that work with daylight and familiar materials Foster (2011) Hurlbert & Yu (2025). Future experimental programs should jointly measure perceived illuminant color, appearance matches, spectral data, and scene-level object judgments. They should also compare predictions under independent light and reflectance manipulations, with the metrics stated before examining results.
The relation between categorization, language, and perception raises further questions about when effects reside in attention, visual memory, decision criterion, or genuinely changed appearance. The Russian-blue verbal-interference results Winawer et al. (2007) make language-sensitive task performance a credible empirical phenomenon, but they leave room for multiple explanatory mechanisms. Color naming systems also reflect perceptual and communicative constraints Regier et al. (2007). Large cross-lingual replications using carefully equated lighting, colorimetric coordinates, and multiple task types can isolate which parts of the effect travel across observers, languages, and developmental stages. Artificial networks provide experimental model systems for testing computational hypotheses but require independent human validation Akbarinia (2025).
Scientific Applications and Responsible Translation
Color science informs displays, imaging, lighting, industrial quality control, medical screening, navigation, and information design. Each application has its own success criterion. A display engineer may need a standardized colorimetric match; a safety researcher may need reliable discrimination under rapid attention shifts; a clinician may need diagnostic sensitivity for a specified phenotype; a designer may need comprehension of a sign across color-vision variation; and a researcher of ecological perception may need accurate surface judgments under real-world illumination. The CIE's measurement conventions provide a common physical reference across many of these tasks CIE (2018a), while appearance studies identify conditions under which similar physical signals yield different perceived outcomes.
Accessibility decisions should explicitly account for varying color vision. A colored distinction in a figure or warning should be accompanied by shape, position, text, or another independent cue when loss of the chromatic distinction could defeat the task. Evaluations should recruit the target observer population or simulate optical conditions only as a preliminary engineering check. Simulation alone is not a substitute for behavioral evidence. Color-difference thresholds can help select candidate stimuli, but recognition and comprehension under expected lighting, screen settings, and attention demands constitute the final criterion. The genetic and psychophysical diversity documented in color-vision research reinforces this approach Neitz & Neitz (2011) Stockman & Sharpe (2000).
The distinction between chromatic appearance and the biological impact of light also matters in workplace and health claims. Two lights with similar conventional color appearance can differ in spectral characteristics relevant to non-image-forming retinal responses, while two colored lights can differ in contrast, brightness, glare, or visual comfort. Claims about sleep, mood, alertness, productivity, or clinical improvement require separate validated outcomes and controlled exposure protocols, with appropriate clinical evidence if health interventions are proposed CIE (2018b). This monograph's evidence does not support universal emotional prescriptions based on hue names. Psychological preference, cultural meaning, and sensory processing warrant methodologically distinct assessments Palmer & Schloss (2010).
Artificial image-generation and machine-vision systems create new contexts in which perceptual predictions can be tested. A system's internal chromatic embedding can be evaluated against calibrated human discrimination and appearance matches, and its errors can motivate experiments on visual processing. Such comparisons should report the imaging pipeline and training distributions and preserve the difference between successful functional simulation and evidence of equivalent first-person color experience. This research perspective connects color vision to the Artificial Era through explicit cross-system tests rather than an unexamined transfer of psychological attributes. Akbarinia (2025) Gegenfurtner (2025)
Conclusion
Color vision rests on well-established constraints of spectral sampling by three broadly overlapping cone classes, followed by neural computations that compare, adapt, and reorganize their signals. Color appearance emerges from those signals in relation to adaptation state, spatial structure, surfaces, illumination, visual tasks, and acquired knowledge. These explanatory levels correspond to measurable constructs: photoreceptor excitation, postreceptoral selectivity, behavioral discrimination, contextual matching, cortical population responses, and category judgments. The evidence for color appearance is strongest when findings from these levels converge while preserving the differences among their experimental endpoints. Baylor et al. (1987) Stockman & Sharpe (2000) Gegenfurtner (2003) Foster (2011)
The history of color science provides a model of cumulative scientific explanation. Trichromatic matching established the dimensionality of receptor-limited equivalence; opponent theory captured structured differences and aftereffects; molecular and physiological studies identified photopigments and early mechanisms; and more recent psychophysics, neuroimaging, and computational work examined the perceptual stability of colored objects across changes in the scene. Color constancy remains conditional, with behavior depending on what an observer is asked to compare or recognize. Language and memory can influence measured judgments without licensing an unrestricted claim that conceptual categories determine early sensory encoding. Hurvich & Jameson (1957) Brainard & Wandell (1992) Winawer et al. (2007) Gegenfurtner (2025)
A mature research account consequently demands calibrated spectral stimuli, explicit observer and task definitions, transparent model assumptions, and tests that separate competing explanations. The outstanding questions concern how neural codes become stable object-color experiences, how natural-scene priors and adaptation interact, how language and learned knowledge influence appearance, and how computational models generalize beyond their original laboratory contexts. These questions link perception, biology, cognition, language, psychophysics, and Artificial Era research within one coherent empirical program. Color Vision and Color Appearance is defined here as the systematic psychological study of wavelength-sensitive encoding, discrimination, and context-dependent phenomenal attributes of visually perceived colors, examined through reproducible measurement and critical scientific synthesis.
References
- Akbarinia, A. (2025). Exploring the categorical nature of colour perception: Insights from artificial networks. Neural Networks, 181, 106758. https://doi.org/10.1016/j.neunet.2024.106758
- Baylor, D. A., Nunn, B. J., & Schnapf, J. L. (1987). Spectral sensitivity of cones of the monkey Macaca fascicularis. The Journal of Physiology, 390, 145–160. https://doi.org/10.1113/jphysiol.1987.sp016691
- Bosten, J. M., Beer, R. D., & MacLeod, D. I. A. (2015). What is white? Journal of Vision, 15(16), 5. https://doi.org/10.1167/15.16.5
- Brainard, D. H. (1998). Color constancy in the nearly natural image. 2. Achromatic loci. Journal of the Optical Society of America A, 15(2), 307–325. https://doi.org/10.1364/JOSAA.15.000307
- Brainard, D. H., & Wandell, B. A. (1992). Asymmetric color matching: How color appearance depends on the illuminant. Journal of the Optical Society of America A, 9(9), 1433–1448. https://doi.org/10.1364/JOSAA.9.001433
- Brouwer, G. J., & Heeger, D. J. (2009). Decoding and reconstructing color from responses in human visual cortex. Journal of Neuroscience, 29(44), 13992–14003. https://doi.org/10.1523/JNEUROSCI.3577-09.2009
- Commission Internationale de l'Éclairage. (2018). CIE system for metrology of optical radiation for ipRGC-influenced responses to light (CIE S 026/E:2018). https://doi.org/10.25039/S026.2018
- Commission Internationale de l'Éclairage. (2018). Colorimetry (4th ed.; CIE 015:2018). https://doi.org/10.25039/TR.015.2018
- Conway, B. R. (2014). Color signals through dorsal and ventral visual pathways. Visual Neuroscience, 31(2), 197–209. https://doi.org/10.1017/S0952523813000382
- Conway, B. R., Chatterjee, S., Field, G. D., Horwitz, G. D., Johnson, E. N., Koida, K., & Mancuso, K. (2010). Advances in color science: From retina to behavior. Journal of Neuroscience, 30(45), 14955–14963. https://doi.org/10.1523/JNEUROSCI.4348-10.2010
- Foster, D. H. (2011). Color constancy. Vision Research, 51(7), 674–700. https://doi.org/10.1016/j.visres.2010.09.006
- Gegenfurtner, K. R. (2003). Cortical mechanisms of colour vision. Nature Reviews Neuroscience, 4(7), 563–572. https://doi.org/10.1038/nrn1138
- Gegenfurtner, K. R. (2025). The Verriest Lecture: Color vision from pixels to objects. Journal of the Optical Society of America A, 42(5), B313–B328. https://doi.org/10.1364/JOSAA.544136
- Hansen, T., Olkkonen, M., Walter, S., & Gegenfurtner, K. R. (2006). Memory modulates color appearance. Nature Neuroscience, 9(11), 1367–1368. https://doi.org/10.1038/nn1794
- Hurlbert, A., & Yu, C. (2025). Seeing the light: Perception and discrimination of illumination color. Annual Review of Vision Science, 11, 267–301. https://doi.org/10.1146/annurev-vision-121423-013755
- Hurvich, L. M., & Jameson, D. (1957). An opponent-process theory of color vision. Psychological Review, 64(6), 384–404. https://doi.org/10.1037/h0041403
- Karimipour, H., & Witzel, C. (2024). Colour expectations across illumination changes. Vision Research, 222, 108451. https://doi.org/10.1016/j.visres.2024.108451
- Krauskopf, J., Williams, D. R., & Heeley, D. W. (1982). Cardinal directions of color space. Vision Research, 22(9), 1123–1131. https://doi.org/10.1016/0042-6989(82)90077-3
- Land, E. H., & McCann, J. J. (1971). Lightness and Retinex theory. Journal of the Optical Society of America, 61(1), 1–11. https://doi.org/10.1364/JOSA.61.000001
- Li, C., Li, Z., Wang, Z., Xu, Y., Luo, M. R., Cui, G., Melgosa, M., Brill, M. H., & Pointer, M. (2017). Comprehensive color solutions: CAM16, CAT16, and CAM16-UCS. Color Research & Application, 42(6), 703–718. https://doi.org/10.1002/col.22131
- Lindsey, D. T., & Brown, A. M. (2002). Color naming and the phototoxic effects of sunlight on the eye. Psychological Science, 13(6), 506–512. https://doi.org/10.1111/1467-9280.00489
- Liu, Q., van Paridon, J., & Lupyan, G. (2025). Learning about color from language. Communications Psychology, 3, Article 60. https://doi.org/10.1038/s44271-025-00230-9
- MacLeod, D. I. A., & Boynton, R. M. (1979). Chromaticity diagram showing cone excitation by stimuli of equal luminance. Journal of the Optical Society of America, 69(8), 1183–1186. https://doi.org/10.1364/JOSA.69.001183
- Maloney, L. T. (1986). Evaluation of linear models of surface spectral reflectance with small numbers of parameters. Journal of the Optical Society of America A, 3(10), 1673–1683. https://doi.org/10.1364/JOSAA.3.001673
- Maloney, L. T., & Wandell, B. A. (1986). Color constancy: A method for recovering surface spectral reflectance. Journal of the Optical Society of America A, 3(1), 29–33. https://doi.org/10.1364/JOSAA.3.000029
- Mancuso, K., Hauswirth, W. W., Li, Q., Connor, T. B., Kuchenbecker, J. A., Mauck, M. C., Neitz, J., & Neitz, M. (2009). Gene therapy for red–green colour blindness in adult primates. Nature, 461, 784–787. https://doi.org/10.1038/nature08401
- McCollough, C. (1965). Color adaptation of edge-detectors in the human visual system. Science, 149(3688), 1115–1116. https://doi.org/10.1126/science.149.3688.1115
- Nathans, J., Piantanida, T. P., Eddy, R. L., Shows, T. B., & Hogness, D. S. (1986). Molecular genetics of inherited variation in human color vision. Science, 232(4747), 203–210. https://doi.org/10.1126/science.3485310
- Nathans, J., Thomas, D., & Hogness, D. S. (1986). Molecular genetics of human color vision: The genes encoding blue, green, and red pigments. Science, 232(4747), 193–202. https://doi.org/10.1126/science.2937147
- Neitz, J., & Neitz, M. (2011). The genetics of normal and defective color vision. Vision Research, 51(7), 633–651. https://doi.org/10.1016/j.visres.2010.12.002
- Olkkonen, M., Witzel, C., Hansen, T., & Gegenfurtner, K. R. (2010). Categorical color constancy for real surfaces. Journal of Vision, 10(9), 16. https://doi.org/10.1167/10.9.16
- Palmer, S. E., & Schloss, K. B. (2010). An ecological valence theory of human color preference. Proceedings of the National Academy of Sciences, 107(19), 8877–8882. https://doi.org/10.1073/pnas.0906172107
- 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. https://doi.org/10.1073/pnas.0610341104
- Stockman, A., & Sharpe, L. T. (2000). The spectral sensitivities of the middle- and long-wavelength-sensitive cones derived from measurements in observers of known genotype. Vision Research, 40(13), 1711–1737. https://doi.org/10.1016/S0042-6989(00)00021-3
- Stockman, A., MacLeod, D. I. A., & Vivien, J. A. (1993). Isolation of the middle- and long-wavelength-sensitive cones in normal trichromats. Journal of the Optical Society of America A, 10(12), 2471–2490. https://doi.org/10.1364/JOSAA.10.002471
- Wakui, E., Mylonas, D., Caparos, S., & Davidoff, J. (2025). Categorical color perception shown in a cross-lingual comparison of visual search. Color Research & Application, 50(4), 301–313. https://doi.org/10.1002/col.22964
- Wandell, B. A. (1993). Color appearance: The effects of illumination and spatial pattern. Proceedings of the National Academy of Sciences, 90(21), 9778–9784. https://doi.org/10.1073/pnas.90.21.9778
- Winawer, J., Witthoft, N., Frank, M. C., Wu, L., Wade, A. R., & Boroditsky, L. (2007). Russian blues reveal effects of language on color discrimination. Proceedings of the National Academy of Sciences, 104(19), 7780–7785. https://doi.org/10.1073/pnas.0701644104
- Witzel, C., & Gegenfurtner, K. R. (2018). Color perception: Objects, constancy, and categories. Annual Review of Vision Science, 4, 475–499. https://doi.org/10.1146/annurev-vision-091517-034231