Chromatic visual evoked potentials: A review of physiology, methods and clinical applications

Objective assessment of the visual system can be performed electrophysiologically using the visual evoked potential (VEP). In many clinical circumstances, this is performed using high contrast achromatic patterns or diffuse flash stimuli. These methods are clinically valuable but they may only assess a subset of possible physiological circuitries within the visual system, particularly those involved in achromatic (luminance) processing. The use of chromatic VEPs (cVEPs) in addition to standard VEPs can inform us of the function or dysfunction of chromatic pathways. The chromatic VEP has been well studied in human health and disease. Yet, to date our knowledge of their underlying mechanisms and applications remains limited. This likely reflects a heterogeneity in the methodology, analysis and conclusions of different works, which leads to ambiguity in their clinical use. This review sought to identify the primary methodologies employed for recording cVEPs. Furthermore cVEP maturation and application in understanding the function of the chromatic system under healthy and diseased conditions are reviewed. We first briefly describe the physiology of normal colour vision, before describing the methodologies and historical developments which have led to our understanding of cVEPs. We thereafter describe the expected maturation of the cVEP, followed by reviewing their application in several disorders: congenital colour vision deficiencies, retinal disease, glaucoma, optic nerve and neurological disorders, diabetes, amblyopia and dyslexia. We finalise the review with recommendations for testing and future directions.


Introduction
Visual evoked potentials (VEP) are cortical responses elicited by visual stimulation that can be recorded non-invasively by electrodes placed on the observer's scalp.Stimuli that are used to produce VEPs largely include pattern reversal (prVEP), pattern onset-offset (poVEP) and spatially homogeneous flashes (fVEP), which are all incorporated into international standards for clinical VEP testing (Odom et al., 2016).The applications of clinical VEPs are to assess, characterise and provide objective (i.e.independent of any subjective observer reports) information of normal and abnormal visual pathway function up to the level of the visual cortices.This is particularly important for characterising function and is useful in individuals who are unable to complete reliable subjective or psychophysical tests.As such, VEP responses may reflect many aspects of visual processing and visual perception, such as motion, texture, contrast and colour.This review is concerned with VEPs elicited by stimuli that contain chromatic (colour) information and that are related to visual processing within the chromatic visual pathways and to colour vision.Standard clinical VEPs are typically elicited by achromatic, high-contrast checkerboards or diffuse flashes (typically with a broadband spectrum and of white chromaticity) which generally do not, at least not selectively, assess mechanisms involved in colour vision.However, colour vision may be affected in many conditions where the retina, optic nerve, visual pathway or cortex are affected.The consequences of these conditions may not be captured by the standard clinical VEP.As such, providing an understanding of the mechanisms underlying chromatic function objectively through VEP recordings is important.The chromatic VEP (cVEP) can be an aid to support a diagnosis of genetic or acquired colour vision impairments, or to characterise the the DKL space is that the L-and M-cones are assumed to have also equal strengths in the achromatic and blue-yellow axes.This may not be true.Generally, the luminance pathway is L-cone dominated and there may be substantial inter-individual differences.Third, although the physiological pathways on the one hand and the cardinal axes and psychophysical colour opponent systems on the other hand are here and elsewhere treated as equivalent, they are not identical (Conway et al., 2023).These considerations probably should be taken into account when using the DKL space to quantify the stimuli.
The cardinal systems are likely the inputs for further cortical colour processing that may influence VEP outcomes.In the present review, we limit ourselves to the influence of the main subcortical pathways on VEP data.For the influence of more central cortical processes on VEPs, we refer to the review by Shapley (2024).

Luminance system (magnocellular pathway)
In the luminance system, a percept of the intensity of a stimulus is obtained, independently of its wavelength content.Under certain circumstances it may be able to detect colour changes, but in those cases it is not able to discriminate between the directions of colour change [e.g. from red to green or from green to red (Lee et al., 1998a)].The luminance system has several properties such as transitivity (if the A and B are isoluminant and if A and C are isoluminant then B and C are also isoluminant), proportionality (i.e., if two colours are isoluminant, then they are still isoluminant when their luminances are multiplied by the same factor) and additivity (the luminance of A + B equals the luminance of A plus the luminance of B) (Kaiser et al., 1990).These properties probably prompted the CIE (Commission Internationale de L' Éclairage) to use data from this system to formally quantify the luminance of a stimulus.The spectral sensitivity of the luminance system is described by the V λ (the spectral luminosity function).Thus, although the response of the luminance system depends on the wavelength of the stimulus, it does not carry information about its wavelength content (because wavelength and intensity of the stimulus can be exchanged such that the response remains unaltered) and therefore does not bear colour information.There are different psychophysical tasks such as heterochromatic flicker photometry, heterochromatic modulation sensitivity, minimal motion or minimally distinct border which can assess this system (Lennie et al., 1993;Wagner and Boynton, 1972).Assuming that eye movements play an important role in the minimally distinct border task, all psychophysical tasks that have a V λ -like spectral sensitivity are characterised by a modulation between differently coloured stimuli at a relatively high temporal frequency.Luminance with a V λ spectral sensitivity is obtained by an additive interaction of the outputs of the L-and M-cones with an averaged ratio of about 2:1.This ratio reflects the ratio of L-to M-cone numbers in the retinae of normal trichromats (Brainard et al., 2000).The psychophysical tasks further show that, although the photopic spectral sensitivity can be described on average by V λ , there is considerable inter-individual variability, mainly caused by genetic variability in the cone fundamentals, inter-individual differences in the macular pigment optical density and particularly in the ratios of L-to M-cone numbers.This variability has consequences for creating isoluminant stimuli, i.e. stimuli where luminance is not modulated, because this variability may lead to luminance intrusion.It is advisable to determine the individual isoluminance conditions using one of the psychophysical tasks mentioned above.Since the isoluminance point may also depend on spatial location, it should ideally be determined at the location where the experimental stimuli are presented.ERGs to high temporal frequency stimuli (>30 Hz) show a V λ -like spectral sensitivity.In addition, the ERGs show inter-individual variability that is correlated with the psychophysically found variability and with ratio of L-to M-cone numbers (Brainard et al., 2000;Jacobs and Neitz, 1993;Kremers et al., 2000).Thus, the high frequency ERGs probably reflect the activity of the luminance system.The physiological basis of the luminance system is the magnocellular (MC) pathway that contains parasol retinal ganglion cells (RGCs), receiving additive input from the L-and the M-cones through diffuse bipolar cells.In the lateral geniculate nucleus (LGN) the MC layers are ventrally located.The receptive fields of the MC (parasol) RGCs contain a centre and a generally larger antagonistic surround that both receive additive input from several L-and M-cones.The strengths of the L-cone and M-cone driven signals reflect their packing density; on average the L-cone driven responses are about twice as strong as the M-cone driven signals ([+2 L+M] for on-centre cells or [-2 L-M] for off-centre cells) with an average distribution of L:M around 2:1, but there may be a large inter-individual variability.As the receptive fields receive input from several cones, they are relatively large (Kremers and Lee, 1998;Lee et al., 1998a).They have a high temporal resolution and high contrast sensitivity.Because of the high contrast sensitivity they have a relatively high spatial resolution that is similar to those of PC-cells in the central 10 • (Crook et al., 1988).In addition, they can respond to small spatial displacements so that their responses may underlie hyperacuity (Lee et al., 1995).In macaques and the marmosets, MC-cells have been found to have lower spatial resolution than PC-cells.MC cells are generally "colour blind"; they may not respond to isoluminant colour changes or their responses are similar for both directions, being attuned for luminance detection (i.e. the response to red to green change is similar to the reversed change) (Derrington and Lennie, 1984;White et al., 2001).Luminance and brightness are often used as synonyms but they are not identical.Luminance is a physical measure, brightness is the corresponding percept that can be measured with psychophysical tasks using static stimuli or stimuli with low temporal frequencies [such as detection of increment thresholds with stimuli that are large and presented with relatively long durations (Lennie et al., 1993))].The spectral sensitivity of brightness is not identical with the V λ function (King- Smith and Carden, 1976), as cone opponent mechanisms are thought to play a role (Sperling and Harwerth, 1971).Furthermore, the abovementioned properties of the luminance system (transitivity, proportionality and additivity) mostly do not apply to brightness.In brightness, two chromatic systems may be involved.It has been proposed that the brightness signal may come from ipRGCs because brightness estimation for luminance equated stimuli increased with increasing melanopsin excitation (Zele et al., 2018a).However, the alternative explanation, that other mechanisms with increased sensitivity to short wavelengths relative to the V λ (as is the case for instance with the King-Smith and Carden type of spectral sensitivity) may play a role in brightness perception, was not excluded by these experiments.cVEPs may be a way to solve this issue because ipRGCs constitute a minority of RGCs and to a large part do not project to the primary visual cortex.As a result, their activity may have little influence on VEP responses whereas cone-opponent mechanisms are likely a major source of cVEP responses (see below).

Red-green colour system (parvocellular pathway)
In the red-green chromatic system, L-and M-cone driven signals are processed antagonistically.The parvocellular (PC) pathway is considered to be the physiological basis of this colour system.PC (midget) RGCs receive input from midget bipolar cells.PC ganglion and LGN cells generally have a centre-surround structure.Unlike the luminance system, the signals from L-and M-cones have about equal strength (Smith et al., 1992).Thus, the L/M-ratio is about unity.This was also found psychophysically with flicker detection thresholds that are mediated by the red-green system (Kremers et al., 2000).Given that the number of L-cones is generally greater than the number of M-cones, and that there is a large inter-individual variability in these numbers, there apparently is a compensatory mechanism which normalizes the cone signals so that they have equal strengths.It is worth mentioning that ERG responses to intermediate temporal frequency (between about 8 and 16 Hz) stimuli containing red-green chromaticity changes also show L/M opponency and L/M ratios close to unity, indicating that these ERGs reflect activity of the red-green colour system (Brainard et al., 2000;Kremers et al., 2000Kremers et al., , 2010;;Parry et al., 2012).The receptive field centres of foveal PC RGCs receive input from one cone (either L-or M-) whilst those of the peripheral cells receive input from several cones.The surrounds receive the input from either the opponent cone type or from a mixture of both (Field et al., 2010).There is some debate if the surrounds receive random input from the L-and M-cones or if there is some degree of cone selectivity (Lee et al., 1998b).The L-/M-signal ratio is about one ([+L-M] for L-on and M-off centre cells or [+M-L] for M-on and L-off centre cells) (Smith et al., 1992).This ratio may reflect the compensation mechanism mentioned above and would be an argument for at least some cone selectivity in the receptive field surrounds.The small receptive fields have the consequence that the PC cells have a high spatial resolution.However, their temporal resolution is lower than of MC-cells (Derrington et al., 1984;Kremers et al., 1997;Solomon et al., 1999).The PC RGCs make synapses in the dorsal layers of the LGN with neurons projecting to the cortex.Red-green opponency in the PC pathway, in evolutionary terms, is relatively young and evolved when the primordial L-/M-photoreceptor divided into separate L-and M-cones and thus when trichromacy appeared.The PC pathway processes L-and M-cone signals in very similar manners in Old-and New World monkeys, although trichromacy evolved independently in the two groups, indicating that the L-/M-opponency directly emerged without many additional changes.Possibly, the small receptive fields of the PC cells would result in a cone imbalance in receptive field centres and surrounds (and thus cone opponency) as soon as they received input from two spectrally different cone types (Kremers et al., 1999).The process that normalizes L-and M-cone inputs to a ratio of about unity may be a subsequent adaptation necessary for robust colour vision that is independent of inter-individual differences in the number of cones.

Blue-yellow colour system (koniocellular pathway)
The blue-yellow system is evolutionarily older than the red-green system and generally present in all mammals.In the blue-yellow system, S-cone signals are antagonistically processed against an addition of L-and M-cone signals (or one of them in dichromats).The physiological basis for the blue-yellow chromatic system is a subgroup of cells belonging to the koniocellular (KC) pathway.The KC pathway is a structurally and functionally heterogeneous group of cells.Of interest for colour processing are the cells that receive excitatory input from the S-cones and inhibitory input from the L-and M-cones (+S -[L+M]).The O.R. Marmoy et al.S-cone signals are transmitted via blue cone bipolar cells whereas diffuse bipolar cells transmit the information from the L-and M-cones (Grunert and Martin, 2020).It therefore can be assumed that the relative weighting of L-and M-cones in this pathways is similar, if not identical, to their ratio in the luminance pathway.They transmit the information to small field bistratified RGCs (Dacey and Lee, 1994).As the name already suggests, these cells receive their input in two distinct tiers in the inner plexiform layer; one in the outer part where they contact the diffuse bipolar cells and an inner part where they are connected with the blue cone bipolar cells (Martin, 1998).The antagonistic fields in these cells have about similar sizes and they thus do not have centre-surround structure.These cells have relatively large receptive fields and a low spatial resolution (White et al., 2001).Their temporal resolution is similar to those of PC-cells (Yeh et al., 1995).The small field bistratified neurons project to the LGN cells located between or surrounding the MC and PC layers.Cells with inhibitory S-cone input are less abundant and are probably of another type.

Cortical processing of colours
Chromatic visual information initially reaches the primary visual cortex (V1) from the retina via the LGN.V1 contains several coloursensitive neurons that play an important role in colour perception (Shapley and Hawken, 2011).The cortical processing of colour is complex and likely involves multiple cortical regions, including and wider than V1 alone (Bartels and Zeki, 2000;Johnson and Mullen, 2016;Winawer et al., 2010).After V1-V2, processing then occurs via ventral (occipito-temporal) and dorsal (occipito-parietal) pathways, the first also named 'what' stream due its role in identification of visual objects and the latter also named the 'where' stream due to its role in localisation of visual objects in space.The ventral pathway contains areas where faces, object shapes, object colour, body parts etc. are preferentially processed.Cortical neurons are not exclusively colour sensitive or colour selective.Their responses can also be influenced by spatial properties of the stimulus (Johnson and Mullen, 2016).In addition, within an area the responses to chromatic stimuli are far more complex than the cone opponency that describes the responses in retino-geniculate pathways.They display response properties that indicate inputs from more than one subcortical pathway.As a result, the colour selectivity may not be along the cardinal axes.For a comprehensive review of cortical colour vision processing and the cVEP the reader is referred to the review by Shapley (2024).

Review methods
This review aimed to evaluate the literature exploring cVEP methodology, physiology and clinical applications.As such, the search strategy was performed systematically whereby the authors reviewed Medline (1967-present), Embase (1974-present), CINAHL, PsycINFO and Web of Science Core Collection databases for studies between 1967 and September 2021.The search terms used by the authors were ((Chromatic* OR Colo*r) AND ("VEP" OR "VECP" or "VER" or "Visual Evoked potential" or "visual evoked cortical potential" or "visual evoked response" or "Evoked").A further manual search through references of eligible works was performed to retrieve any published items which were not identified within the primary search.The resultant records had duplicates removed before the record title and abstracts were screened.Of these, the full-text articles were read in full by the authors and further screened for eligibility.Studies included were assessed in accordance with the OCEBM levels of evidence (Howick et al., 2011), including full-text articles, theses, or chapter texts which used cVEPs experimentally to understand the underlying physiology of colour vision or applied these within disorders of vision.The full review process is illustrated in Fig. 1.Evidence of levels 1-3 were critically reviewed by the authors, Fig. 1.PRISMA systematic search process for this cVEP review.
O.R. Marmoy et al. with level 4 (case reports, series, or editorials/letters) only reviewed fully if appearing of particular scientific merit as assessed by the authors.Only articles available in English language were included.Animal studies or recordings of retinal potentials in isolation were excluded, unless useful in understanding basic underpinning scientific methods.
The subsequent systematic review comprised of allocating studies according to relevance (1) techniques to elicit the cVEP (2) those performed in healthy participants to understand underlying physiology (3) clinical applications in disease.

Methodological approaches in recording cVEPs
The methodology used to record cVEPs can vary according to the clinical purpose of the recording.Similar to conventional achromatic pattern and flash VEPs, cVEPs strongly depend on the spatial, temporal and spectral properties of stimuli (Regan, 1970(Regan, , 1973(Regan, , 1974)).Through this extensive review we categorise the major considerations according to their respective spatial properties, temporal properties, chromaticity, and recording or analysis techniques.Most methods involving cVEPs have attempted to isolate respective retino-geniculate PC or KC pathways, sometimes complemented by achromatic stimuli for the MC pathway.A response biased or selective for the K-pathway would generally need an S-cone mediated stimulus along the tritan confusion line, often utilised in the form of a B-Y stimulus.Conversely, for the PC pathway this would require an L-and M-cone mediated stimulus, often utilised in the form of a R-G stimulus.The stimulus and recording parameters used to record the cVEP most likely reflect summed activity from cortical neurons in areas V1-V2, as such this review generally describes the properties of cVEPs relating to the visual apparatus up to this cortical level.Whilst higher or other association cortices may contribute to colour vision or perception, it is not currently clear from the reviewed evidence that this has a significant influence on the cVEP.

Stimulus field size
The stimulus field size is an important consideration for cVEP testing.It can have a direct impact on the amplitude and/or peak-time of responses, alongside affecting the underlying mechanisms generating the response.Whilst many studies have used spatially homogenous fields with Ganzfeld stimulation, much of the more recent cVEP works utilised structured stimuli where field size was often determined by the visual display unit.Accordingly, with structured stimuli it has been demonstrated that the cVEP to blue or green gratings is largest around 8-9 • , with amplitude showing a linear increase with field size, although it is acknowledged that amplitude alone may not be an appropriate measure of chromatic response selectivity (Fig. 2).Amplitude for field sizes beyond 7 • blue cVEPs appear to saturate (Korth et al., 1994).Whilst the cVEP response is largely dominated by the central field and fovea (Gerth et al., 2003), increasing the field size can lead to undesired luminance intrusion, because of inclusion of peri-or paramacular stimulation.A small chromatic stimulus that is isoluminant for the observer would achieve high B/Y pathway selectivity, as both chromatic aberration and the potential effects of macular pigment (spatially localised to central macula) would be better controlled.For B-Y gratings, 3 • is most chromatically selective (e.g., Kulikowski et al., 1996Kulikowski et al., , 1997a;;Robson et al., 2006), and larger fields possible for R/G, as longer wavelengths are less absorbed by MP and less affected by chromatic aberration.These regions have a different spectral sensitivity compared to the macular region due to the macular pigment that only covers the central part of the retina, alongside chromatic aberration which can be more intrusive with larger field sizes and spatial frequencies (Kulikowski et al., 1997b).As a result, stimuli that are isoluminant for the macular region may not be isoluminant for the perimacular region which can affect the cVEP mechanism or waveform, meaning that smaller fields are required for better neural selectivity.As macular pigment mainly absorbs short wavelengths, B-Y stimuli are particularly affected by this luminance intrusion (Kulikowski et al., 1996).
Fig. 2 suggest that, when testing with supra-threshold contrasts, a cVEP stimulus subtending 5-10 • may achieve a reasonable response selectivity for post-retinal R-G/B-Y colour systems, although smaller fields (i.e.3-6 • ) are likely the most selective for chromatic pathways, particularly when performing B-Y cVEPs.The influence of macular pigment optical density is correlated with cVEPs elicited by B-Y and tritan stimuli, suggesting that field size must be limited when using B-Y cVEPs to minimise luminance intrusion (Robson et al., 2006).Whilst some studies have demonstrated that larger fields up to 21 • can produce similar cVEPs (Pompe et al., 2012), these responses are likely to be more confounded by chromatic aberration and luminance intrusion, so are best reserved for patients with unstable fixation (i.e.children) with an understanding that these larger field cVEPs may be less selective to the underlying neural generators.Certainly, in any studies specifically examining contrast sensitivity or near-threshold responses, smaller stimulus fields are a necessity to minimise these confounding effects.
Another source of luminance intrusion may be the above-mentioned inter-individual variability in isoluminance.One method commonly used to account for this in cVEP studies is to perform heterochromatic flicker photometry prior to testing.This technique can be used to determine individual isoluminance points, which can then be used to specify the luminance properties of stimuli when performing cVEPs.This method may limit potential luminance intrusion if performed using the same field size as the cVEP stimulus, as if a larger field is used this cannot control for effects of chromatic aberration or eccentricity dependent changes in isoluminance.Nevertheless, cVEPs have been demonstrated to reflect similar post-receptoral mechanisms using a large field stimulus in several studies, for example between 7 • and 21 • fields (Madrid and Crognale, 2000;Pompe et al., 2012Pompe et al., , 2014)), but subtle intrusions of chromatic aberration or luminance cannot be ruled out, especially as the studies using larger fields are in younger participants who can show a Fig. 2. Influence of stimulus field size on the cVEP.Data extracted and modified from Korth and Nguyen (1997) and Pompe et al. (2014).Korth et al.'s data were corrected for log field area and extracted to provide an angular subtense against response amplitude to onset-offset blue (triangular data points, solid blue line) and green (circle data points, dashed green line) gratings presented on a yellow background.Of note, the 450 nm gratings produced a major N1 negativity whereas 550 nm gratings produced a major P1 positivity, which Data are presented as mean ±1SD.The data from Pompe et al. (2014) were extracted and the N1 (empty diamonds, solid lines) or P1 (filled squares, dashed lines) components of the paediatric cVEP to R-G and B-Y stimulation to 7 • and 21 • field size were plotted.A potential loss of chromatic response selectivity of cVEPs to large stimulus fields is discussed in the main text.morphology of major positivity (which also occurs with luminance contamination in cVEPs) rather than negative morphology typically seen in adults.Given the potential problems associated with luminance intrusion and chromatic aberration, some authors have developed specific stimulus paradigms accounting for retinal eccentricity dependent changes in isoluminance.This includes development of pan-isoluminant stimuli where the stimuli are altered in an eccentricity dependent manner, taking account for changes in macular pigment density (Parry and Robson, 2012).Other authors have used multiple gabor stimuli to correct for spatial changes in luminance sensitivity which may account for inhomogeneities in isoluminance which are not solely eccentricity dependent (Skiba et al., 2014).Furthermore a recent study suggested the use of a pseudoisochromatic stimuli, comprising of a mosaic stimulus of different spatial and luminance noise where the artefacts between target and mosaic field are masked, to eliminate the requirement of heterochromatic flicker photometry prior to performing cVEP (Salomao et al., 2019), although the properties of this stimulus are different from other cVEP methods which may have an impact on the underlying physiological mechanisms.Overall, stimulus field size is an important consideration in performing cVEPs, particularly for near-threshold stimuli.These studies suggest that a stimulus field of <10 • provides a reasonable neuronal selectivity and the least confounding effects from luminance or chromatic aberration, but the studies highlight that for stimuli along the tritan confusion line (i.e.B-Y stimuli) field sizes down to 3 • are required for the greatest neuronal selectivity.Larger field sizes have been shown to elicit larger and earlier cVEPs which are beneficial for testing children (Madrid and Crognale, 2000;Pompe et al., 2012Pompe et al., , 2014)), but this may be at the expense of response selectivity due to contamination by luminance intrusion or effects of chromatic aberration.

Spatially homogenous versus structured stimuli
In many initial studies investigating chromatic mechanisms, spatially homogenous stimuli (i.e.Ganzfeld bowl or Maxwellian view systems) were used to elicit cVEPs as this was technically easier to achieve and provided insights into chromatic processing of the entire visual field.The earliest studies of human cVEPs utilised chromatic flashes that varied in wavelength, but maintained similar luminance (Armington, 1964(Armington, , 1966;;Cigánek and Ingvar, 1969;Shipley, 1965;1968).These studies supported the notion of a trichromatic receptor system with maximal sensitivities for long, middle and short wavelengths corresponding to the three cone receptor types.Further work using isoluminant colour stimuli demonstrated that the spectral sensitivity profiles were consistent with evidence of R-G and B-Y opponent colour mechanisms in cVEPs (Yamanaka et al., 1973).While most later and more recent publications have used structured stimuli (i.e.bars, gratings or checks), full-field stimuli are still useful for assessment of post-receptoral pathways and identification of colour deficiency, such as the heterochromatic flicker VEP described below.
With the advent of new technologies for display systems, the number of publications on cVEPs elicited by structured stimuli increased.cVEPs can show marked differences depending on the type of structured stimuli, for example bars, sinusoidal gratings or checkerboards.It was demonstrated in the seminal works by Murray et al. (1987) that cVEPs elicited by the same fundamental spatial frequency produce waveforms of different polarity for checkerboards compared to gratings.This work was crucial in the development of subject-dependent isoluminant onset-offset cVEPs, which were later adopted by other centres.Similar techniques later showed that the choice of sine wave versus square wave gratings will affect the response, with the latter providing larger signals (Kulikowski et al., 1997a).However, given that square wave stimuli contain higher spatial frequencies in addition to the fundamental frequency, this may undesirably increase contribution of MC pathways and risk of chromatic abberation.Similarly, it has been shown that the response amplitude is correlated to the number of edges presented in a checkerboard (Kelly, 1974), with edge contrast strongly influencing cVEP amplitude (Regan, 1973), particularly to higher spatial frequencies.Nevertheless, those who have maintained the use of square-wave gratings have shown that the influence of chromatic aberration is small with square-wave gratings, provided that field size, spatial frequency and isoluminance are well controlled (Kulikowski et al., 1997a).The orientation of grating stimuli has been reported to influence response amplitude, whereby vertically oriented heterochromatic stimuli produce smaller amplitude responses compared to horizontally oriented stimuli (May et al., 1974), whereas peak-time tends to be more similar for horizontal and vertical gratings, but significantly increased for oblique (diagonal) gratings (Rabin et al., 1994).

Spatial frequency
Spatial contrast sensitivity functions (CSFs) of cVEPs show a large influence of spatial frequency, which has been relatively well studied.An important consideration for performing chromatic CSF measurements is that these are highly dependent on type of stimulus and the modality of presentation, for example transient or steady-state (see next section) and reversal or onset-offset modes.
The spatial frequency characteristics of cVEPs relate well to the respsective characteristics of R-G (PC) or B-Y (KC) systems.The typical adult cVEP to onset-offset stimulation shows a prominent negativity at about 120ms at spatial frequencies below 6c/deg, with maximum amplitude around 2c/deg (Murray et al., 1987;Porciatti and Sartucci, 1999).However, at spatial frequencies beyond 8c/deg the onset-offset R-G cVEP responses develop a positive deflection, likely due to luminance intrusion.B-Y onset-offset cVEPs show slight differences compared to those elicited by R-G stimulation.Furthermore, no verified cVEPs have been recorded beyond 7c/deg, when controlled as a pure chromatic stimulus (Kulikowski et al., 1997a).The B-Y cVEP amplitude also peaks between 1-2c/deg but they display sharper amplitude decreases to high spatial frequencies than those elicited to R-G stimuli (Du et al., 1991;Klingaman and Moskowitz-Cook, 1979;Korth and Koca, 1993;Rabin et al., 1994).In addition, B-Y cVEPs have longer peak-times than R-G cVEPs, reported to be around 13ms or 25-30ms later (Rabin et al., 1994;Robson and Kulikowski, 1998) or even up to 55ms for tritan stimuli compared with R-G cVEPs (Robson and Kulikowski, 1998).This difference may be related to the fact that R-G onset cVEPs exhibit an additional early negative morphology that may reflect a transient type MC response to isoluminant R-G borders (Lee et al., 1989;Schiller and Colby, 1983), resulting in 'apparently' faster R-G processing (i.e. a chromatic 'texture channel') (Kulikowski et al., 1997b;Lee et al., 1989).Although lower spatial frequencies are preferred for maximal B-Y cVEPs, if the spatial frequency becomes too low (and thus the period of one modulation too large) then may theoretically increase a local luminance activation and thus be less chromatically selective (Kulikowski et al., 1997a).Therefore, B-Y stimulus spatial frequency should be maintained between 1 and 4c/deg to maintain response selectivity (Barboni et al., 2013;Kulikowski et al., 1997a;Robson et al., 2006).
Regan was one of the first to demonstrate that at fixed spatial frequencies, there is good agreement between cVEP contrast for R-G gratings when comparing to heterochromatic flicker photometry (1973), with some amplitude saturation observed at high contrast also reported (Barboni et al., 2013;Regan, 1973).Later studies have shown a positive linear relationship between the logarithm of the pooled cone contrast (defined as the root mean square of all cone contrasts: = ) and the amplitude of the major negative component (N1) with onset-offset stimulation (Gomes et al., 2010;Souza et al., 2008).Later components of the cVEP tend to saturate at high contrast levels (Fig. 3).The non-linear relationship has been suggested to originate at the cortical level as a form of chromatic gain control (Nunez et al., 2017).It is likely that responses to contrast increment, which produce the typical cVEP onset negativity, are generated by the sustained/tonic neurons, whereas to contrast decrement stimuli reflect more transient activity (Robson and Kulikowski, 2012).These mechanisms likely have different contrast sensitivities and therefore the results cannot be directly compared.
Overall, the preferred spatial configuration of structured cVEP stimuli is likely to be a horizontal grating of ~2c/deg spatial frequency, which may be a sinusoidal or square wave as long as control is maintained for field size and spatial frequency to enable relative comparability between R-G and B-Y systems.Evidently, increasing field size increases the risk of introducing chromatic aberration and differing spectral sensitivity due to macular pigmentation.These studies suggest that a field size of 5-10 • is required, or less, to minimise these confounding effects to achieve an appropriate balance of response amplitude and limitation of the above confounding effects.Whilst larger fields have been applied up to 21 • , these have been studied in patients with poorer fixation (i.e.nystagmus or in children), with heterochromatic flicker photometry performed prior to testing in attempt to minimise these potential confounding effects.

Onset-offset versus reversal stimuli
The cVEP depends strongly on whether the stimulus is presented in an onset-offset mode or as a reversal, likely related to the underlying neuronal basis.The onset of coarse isoluminant chromatic gratings elicits cVEPs that are larger and of opposite polarity to chromatic reversal cVEPs, achromatic onset-offset, and achromatic reversal VEPs (Kulikowski et al., 1997a;Murray et al., 1987;Rabin et al., 1994;Robson andKulikowski, 1998, 2001;Souza et al., 2008;Suttle and Harding, 1999;Tobimatsu et al., 1995).Such responses correlate well with psychophysiological measurements (Kulikowski et al., 2002).
While reversal stimulation mainly activates phasic neurons, particularly at high temporal frequencies, onset-offset stimulation with substantial onset and offset times likely elicit substantial tonic cell activation (Kulikowski and Kato, 1979;McKeefry et al., 1996;Murray et al., 1987).Tonic RGCs mainly project to PC and KC layers of the LGN, whereas phasic cells mainly project to MC layers, therefore stimulation of tonic cells in onset-offset mode are more advantageous in recording cVEPs (McKeefry et al., 1996;Tobimatsu et al., 1995).This is particularly important for evaluation of B-Y cVEPs as S-cones provide exclusive input to the tonic pathways (Gouras, 1968).Furthermore, the reversal cVEP has been suggested to be more susceptible to achromatic intrusion (Kulikowski et al., 1997a), perhaps as phasic cells are more responsive to high frequency reversals.On this basis, the majority of clinical studies have focused upon cVEPs using an onset-offset mode of stimulation, although some studies used reversal stimuli (often to high temporal frequencies) to determine isoluminance and individual colour vision characteristics (see temporal properties section).

Transient versus steady state responses
As with conventional achromatic VEPs, the temporal frequency of the stimulus presentation determines if transient or steady-state responses are elicited.With transient responses, all response components to a stepwise change in the stimulus are recorded.The temporal frequency is therefore low (i.e. with low temporal frequency reversal or onset-offset stimuli).Steady state responses are elicited by high temporal frequency stimuli (mainly reversal).The responses to different step-changes in the stimulus merge and result in a repetitive waveform dominated by the fundamental and a limited number of higher harmonics.Transient cVEP responses have generally been more regularly used in clinical research, although steady-state responses have been utilised and beneficial for those wishing to perform a Fourier analysis (to describe the response in the frequency domain) or multiple contrast or luminance steps.
Transient cVEPs to onset-offset gratings typically utilise a long recording epoch of >500ms, ideally 1000ms, with an onset duration ranging between 200 and 500ms.Responses are largest with onset durations around 50ms due to summation of onset-and offset-components, with longer durations causing a decrease in amplitude due to separation of onset-and offset-responses (Porciatti and Sartucci, 1999).It is highlighted that a short duration, although optimal for maximising VEPs to achromatic stimuli, may not be optimal for high chromatic selectivity.A long stimulus onset period (e.g.200-500ms) can increase the negativity of the cVEP likely reflecting a more chromatic-specific response.Other authors have similarly demonstrated that longer integration times are required to optimise chromatic-specific VEP components, particularly for Tritan stimuluation (Robson et al., 2003).The higher amplitude responses elicited by short onset durations may be influenced by achromatic or transcient response intrusions, reducing chromatic selectivity.This increased sensitivity has similarly been demonstrated when chromatic aberrations are uncontrolled with high spatial frequency (Kulikowski et al., 1997a), thus this increased amplitude and/or sensitivity could reflect reduced response selectivity to a particular colour mechanism.The amplitudes of onset-offset cVEPs decrease for stimulus frequencies above 2p/sec (Murray et al., 1987).Therefore, responses are commonly elicited by 1-2 Hz stimuli.The onset response is most typically studied, as offset responses are diminished at photometric isoluminance, presumably due to the poorer responses of chromatic cortical cells to offset stimuli (Suttle and Harding, 1999).
Temporal contrast sensitivity functions for reversal cVEPs tend to show a band-pass shape (Skiba et al., 2014), with the N1 peak amplitude for S-and L-M stimuli at around 4 Hz (Crognale et al., 1997;Kulikowski et al., 1997a).Steady state reversal responses are elicited by high temporal frequency (typically >12 Hz) stimuli.Steady-state cVEPs to reversal R-G stimuli demonstrate a low-pass function with a decrease in amplitude to higher spatial frequencies (Arakawa et al., 1999;Barboni et al., 2013).High frequency reversal stimulation has been used to objectively determine isoluminance (Kulikowski et al., 1994;Robson and Parry, 2008), and also to estimate macular pigment optical density.Heterochromatic flicker stimuli are also used in psychophysical experiments to determine subjective isoluminance (with the heterochromatic flicker photometry -HFP-method).The temporal frequency of the stimuli is generally above 30 Hz.The luminance (with fixed contrasts) or the contrast (with fixed luminances) of the one of the two colours, are modified until a minimal flicker percept is reached, where, by definition, the two stimuli are subjectively isoluminant.In photopic light levels, this task only depends on activity of the L-and M-cones.The exact condition for subjective isoluminance depends on the ratio of L-and Mcone numbers in the retina and on the cone fundamentals (that is their spectral sensitivities at the cornea where, apart from cone absorption spectra, preretinal transmission is considered; see the section on the luminance system).Both can vary in trichromatic subjects.As a result, the condition for subjective isoluminance varies amongst trichromatic subjects (Kremers et al., 1999).A standard trichromatic observer with a spectral sensitivity that is described by the photopic V λ is employed to determine physical isoluminance.The L-/M-ratio of this standard observer is about 2:1.This method has been used in a range of conditions, such as congenital colour vision defects and retinal disorders (see section) which can show alterations in the point of isoluminance.
Overall, transient responses are most effectively elicited by onsetoffset stimulation using a temporal frequency of 1-2 Hz and a minimum onset duration of 200ms (so that the responses to onset and offset do not merge).Steady-state heterochromatic flicker VEPs will most probably introduce contributions from the luminance system when using stimulation frequencies above 30 Hz, which are similar to recording of the heterochromatic flicker ERG (Kremers and Bhatt, 2016).Steady state responses elicited by lower temporal frequencies (e. g. 12 Hz) are more selective for chromatic systems, mostly when the stimulus is made isoluminant (by minimizing the response at high temporal frequencies or by using a psychophysical isoluminance setting).

Spectral properties 4.3.1. Spectral sensitivity
Measuring spectral sensitivity of luminance and chromatic systems can be assessed after a steady state of light adaptation is reached.The cVEP is typically measured to stimuli with a variety of wavelengths with equal strength.Alternatively, the responses at each wavelength are measured with different stimulus strengths, while the strength for a predefined threshold response is determined.By utilising this method, the inverse of the stimulus strength at threshold defines the sensitivity for that spectral wavelength.By repeating this procedure for several wavelengths, sensitivity can be plotted as a function of stimulus wavelength to obtain spectral sensitivity curves of the observer.Whilst this is theoretically easy to perform, it may be fraught with difficulty as the response may depend on several factors such as the mode of stimulation, dominant temporal frequency, and state of adaptation.Nevertheless, studies examining the spectral sensitivity using cVEPs show a close relationship with psychophysical assessment of spectral sensitivity (Moskowitz-Cook, 1979).
It was demonstrated that the cVEP may be used to model the V λ function in healthy individuals, indicating that the VEPs reflected activity of the luminance (MC) system (Fig. 4).Earlier studies used fullfield chromatic flashes alone, before later works introduced isoluminant or differently coloured backgrounds surrounding a monochromatic stimulus.They showed that the cVEP had similar V λ as psychophysical tests to similar stimuli (Johnsen et al., 1995).Mostly, the use of coloured flashes without specific consideration of background luminance did not reveal the response of a specific chromatic post-receptoral pathway (Willeford et al., 2016).Further developments of spectral sensitivity measurements using cVEPs came following better understanding of human photopigments, photoreceptors and post-receptoral pathways.Accordingly, control in cVEP techniques was achieved through use of adapting backgrounds that selectively desensitized particular photoreceptors (Estevez et al., 1975;Klistorner et al., 1998).It proved possible to measure the spectral sensitivity of each respective cone photoreceptor class using monochromatic stimuli of different wavelengths on an adapting background that desensitized photoreceptors not of interest (Frederiksen et al., 1993b).With different desensitising backgrounds, the responses of distinct photoreceptor types can be obtained.However, a direct comparison of the responses obtained with different backgrounds is not permitted because the different states of adaptation also result in different modes of operation of the O.R. Marmoy et al. neuronal circuits that underly the generation of the cVEP.
Conversely, with heterochromatic flicker or silent substitution cVEP recordings the state of adaptation can be fairly constant so results are more comparable and controlled for isolating individual cone receptor classes (Kremers, 2003).Further details underpinning the techniques of heterochromatic flicker and silent substitution methods are described elsewhere (Cao et al., 2011;Donner and Rushton, 1959;Estévez and Spekreijse, 1982;Kremers et al., 2010;Kuchenbecker et al., 2014;Maguire et al., 2018).
CVEPs to flashes can also differ when measured under photopic vs. scotopic conditions, including alterations with the duration of dark adaptation.The responses to long and short wavelength flashes in recordings of cVEPs reflect response bias originating in the rods and cones and they can be used to describe their differences in adaptation dynamics after a change from high photopic to scotopic conditions (i.e.dark adaptation curves) through repeated cVEP measurements at different times after the start of dark adaptation (Klingaman, 1976).

Spectral content effect on the cVEP
Whilst any visible wavelength may theoretically be used for stimulation to elicit a cVEP, it has been demonstrated widely that the most specific responses are elicited by isoluminant colour stimuli of low chromatic contrast to stimulate colour-opponent neurons (Kulikowski et al., 1996(Kulikowski et al., , 1997a;;Murray et al., 1987;Petry et al., 1982;Porciatti and Sartucci, 1999).Deviations from isoluminance alter the morphology in a way to make this predominantly positive, likely reflecting a shift of the underlying response generating mechanism from chromatic pathways to luminance pathways (Kulikowski et al., 1996(Kulikowski et al., , 1997a;;Murray et al., 1987).Most conventional cVEPs to isoluminant stimuli use wavelengths corresponding to respective L-, M-and S-cone fundamental peaks for red, green and blue stimuli respectively.Yellow as an isoluminant contrast colour to blue can vary (e.g.unique yellow at a wavelength between 575 and 580 nm, but also chosen at the maximum of the V λ at 555 nm).Rabin et al. (1994) studied cVEPs across the DKL (Derrington, Krauskopf and Lennie) colour space using isoluminant stimuli.The results showed that S-cone mediated cVEPs have longer peak-times.Importantly, these authors found that psychophysical and single-unit recordings showed similar spectral properties with cVEPs to both Sand L/M-axis stimuli.The spectral content has similarly been demonstrated elsewhere to be most selective for the P or K systems using R-G or B-Y/tritan stimuli, respectively (Boon et al., 2007;Kulikowski et al., 1996Kulikowski et al., , 1997aKulikowski et al., , 2002;;McKeefry et al., 1996;Murray et al., 1987;Pompe et al., 2006;Porciatti and Sartucci, 1999;Robson et al., 2006).It should also be noted that the requirements for B-Y and R-G may differ when attempting to achieve high response selectivity, due to the B-Y cVEPs being heavily influenced by macular pigment meaning much smaller field sizes are required to maintain selective pathway stimulation, which is less problematic for R-G stimuli.

Relationship with psychophysics
Earlier cVEP studies employed counterphase modulation of bars between isoluminant colours or between colours and a black background.Alternating red and green bars were shown to elicit larger cVEP responses than those elicited by counterphase modulation between black and either green or red, perhaps as the latter introduced larger luminance activity (Perry et al., 1972).Riggs and Sternheim (1969) followed by May and Siegfried (1970) utilised monochromatic gratings to elicit the cVEP, where there was good qualitative agreement between cVEP magnitude according to wavelength across the spectrum compared with psychophysical flicker photometry and brightness matching.Interestingly, they also found that the cVEP was more sensitive to small shifts in wavelength compared with the ERG, suggesting a form of post-retinal processing involved in wavelength discrimination.May and Siegfried (1970) found a correlation between psychophysical measurements using incremental detection thresholds and the cVEP stimulus.However these authors noted discrepancies at the short and long-wavelength spectra, possibly owed to differences in background luminance and/or stimulus size, alongside variation in methodological approach.Subsequent studies have demonstrated a close relationship between cVEPs and psychophysically measured colour vision (Boon et al., 2007;Gerling et al., 1997;Klingaman and Moskowitz-Cook, 1979;Korth et al., 1994;Kulikowski et al., 1997a;Rabin et al., 2016;Robson et al., 2006;Robson and Kulikowski, 1998;Tekavcic-Pompe et al., 2010).4.3.3.1.Heterochromatic flicker cVEPs.Heterochromatic flicker VEPs using high temporal frequencies can be used to assess contrast sensitivity and/or spectral sensitivity of post-receptoral pathways because the responses can display a minimum at stimulus conditions where the involved pathways also have a minimal response.These studies have disclosed a close relationship with psychophysical assessment of the luminous spectral sensitivity (described by the V λ or V' λ ) obtained with heterochromatic flicker photometry and other luminance-based tasks mentioned above.In heterochromatic flicker photometry (and the other photometric tasks) different colours are modulated in counterphase at high temporal frequencies.At isoluminance the percept of flicker is minimal.In cVEPs the response goes through a minimum at the point of perceptual isoluminance.This has been used in studies of cVEP to objectively determine individual isoluminance points (Boon et al., 2005).Isoluminance points vary amongst different observers (Murray et al., 1987;Porciatti and Sartucci, 1999;Thompson and Drasdo, 1992) that, as mentioned above, reflects individual variability in cone densities and cone fundamentals.This has been demonstrated for cVEPs in normal participants compared to protanopes who showed different isoluminance points (Bach and Gerling, 1992), which are further described in the section on genetic colour vision defects.Importantly, heterochromatic flicker testing prior to cVEPs to determine isoluminance should contain the same spatial field size as the cVEP stimulus, otherwise this will not reflect the true isoluminance and lead to potential confounding effects.It is particularly highlighted that with larger fields and B-Y stimulus modulation, the isoluminance point may be inaccurate due to spatial differences in macular pigment optical density over a larger field, which can have a significant effect on determining isoluminance.Therefore, for B-Y stimuli a small field and subsequent small field stimulus (i.e. 3 • ) are required for chromatic selectivity when exploring B-Y cVEPs and heterochromatic flicker photometry.

Neural origins of the cVEP
Analysis methods are also similar to those employed to analyse clinical VEPs.The amplitude is typically measured from baseline to the major negativity (N1) and peak-time at the minimum of the negative trough of the cVEP (Fig. 5a).As is later described, in children the major component is an early positive peak (Fig. 5b).In addition to the N1, some other components can be described.The basis of cVEPs demonstrate that different post-receptoral channels are processing in parallel but primarily reflect activity at V1, at least for early portions of the transient response (Klee et al., 2017).This has also been suggested from fMRI data showing similar activation patterns to chromatic stimuli (Pitzalis et al., 2018).If an early positivity is observed before the N1 trough this likely reflects a luminance intrusion to the signal (Kulikowski et al., 1997a).The waveform after N1 is more complex.They may reflect feedback and feedforward mechanisms and may depend on the type of stimulus presented such as its chromaticity, luminance, spatial and temporal frequency (Nunez et al., 2018;Pitzalis et al., 2018).This is particularly relevant when considering that combined colour and luminance elicited cVEPs are not simply a sum of the response to selective chromatic and luminance stimulation, but instead may reflect interactions between them, possibly through V1-V3 cortex activation (Martinovic and Andersen, 2018;Nunez et al., 2017).It has been demonstrated through fMRI that alongside V1, the early portion of N1 may have origins in V8/VO, although the stimuli in this study were not exclusively from isoluminant colour on-off stimuli to stimulate colour-opponent neurons (Pitzalis et al., 2018) and typical electrode montages would not theoretically identify a signal from V8/VO.Furthermore, mondrian stimulus cVEPs, not widely used clinically, suggest that they involve cortical regions V1, V2 and V4 (Buchner et al., 1994).The cortical basis of colour vision is described in the review by Shapley (2024).
Overall, whilst it is accepted that the early major negativity (N1), or positivity in young persons, reflects V1 cortical activation, it is possible that N1 reflects activity of multiple distinct neuronal populations (Nunez et al., 2021).Later components after this major response are likely reflective or influenced by extra-striate cortices, but are highly dependent on stimulus parameters used.

Recording and analysis
Electrode montages used to record the cVEP are comparable to the conventional (achromatic) clinical VEP, these include the use of an active electrode over the occiput (Oz) with a reference electrode located over the anterior scalp (Fz) (Odom et al., 2016).
As individuals may have different spectral luminosity functions, when using chromatic stimuli it is often essential to perform heterochromatic flicker photometry prior to testing in order to determine the individual point of isoluminance.This can be performed through fast counterphase reversal (typically >30cyc/sec) of the stimulus (R-G or B-Y) chromaticity and altering the luminance ratio of R-G or B-Y, whilst maintaining a steady mean luminance.This would thereafter be expressed as a fraction, for example if R-G are perceptually isoluminant this would provide a value of r = 0.5 (R/R+G).The point of minimal flicker perception should be used as the isoluminant point for cVEP stimulus settings.
As has been described, the typical onset-offset VEP elicited to R-G or B-Y isoluminant gratings is dominated by a major negativity (N1) (Fig. 5).It is fairly well agreed that the N1 component of the cVEP is reflective of the respective (i.e.PC or KC) pathway with R-G or B-Y which are well controlled (Murray et al., 1987;Porciatti and Sartucci, 1999;Rabin et al., 1994).Whilst at high chromatic contrasts (>90%) a response is observed between 100 and 150ms after stimulus onset, this early peak-time may reflect achromatic intrusion (Porciatti and Sartucci, 1999), with low chromatic contrast responses otherwise being more validated to chromatic pathways, which have later peak-time (Kulikowski et al., 1996(Kulikowski et al., , 1997a)).Whilst controlled chromatic, luminance, temporal and spatial profiles can be achieved, there has been debate regarding whether there is complete isolation of an individual post-receptoral pathway in cVEPs (Brigell et al., 1996).Nevertheless, R-G or B-Y responses are commonly abolished or significantly reduced in individuals with congenital colour vision deficiency (see congenital colour vision deficiency section).Therefore, it is considered that with adequate reference data accounting for age (see section "Maturation of cVEPs") and sex (Dion et al., 2013), cVEPs may be utilised in clinical practice for evaluation of colour vision defects when appropriate control of stimulus parameters are achieved.Whilst cVEP recordings are likely to be similar to achromatic VEPs whereby they are vulnerable to degradation with poor compliance or attention, it has been suggested that distraction stimuli have minimal effect on cVEP waveforms (Highsmith and Crognale, 2010).
Those cVEPs elicited to high temporal frequency (i.e. to heterochromatic flicker or steady-state reversal; see below), are often analysed in the frequency domain using Fourier analysis, to identify the amplitude at stimulus frequency and/or at higher harmonics of this frequency.The Fourier analysis also provides information about the timing through the phases.This technique can provide a wealth of information with generally good signal-to-noise-ratio (because, instead of maxima and minima within defined time windows the complete recording time is considered in Fourier analysis), but is not easily achieved clinically through commercial systems at the time of this review.The reader is referred to other works which describe or utilise this technique (Bach and Meigen, 1999;Gur and Zeevi, 1980;Hamilton et al., 2021).

Maturation of chromatic VEPs
Several parts of the human visual system are physiologically and structurally immature at birth.The human retina has reduced sensitivity at birth, likely due to immature photoreceptor outer segments, alongside possible influences from wave-guiding properties of the inner segment and immaturity of the inner retina (Reese, 2011;Teller, 1997).Furthermore, RGCs and the cortical pathways undergo a complex process of growth, pruning and neurodevelopmental change (e.g.myelination) which influences how these signals are transmitted to and within the cortex.In spite of these immaturities, development of chromatic vision can be evident by the second or third month of age (Bornstein, 1978;Peeples and Teller, 1977;Teller, 1997).The cVEP can reflect these developmental changes that can be utilised in assessment of colour vision pathways in infants and children.

Spectral maturation of chromatic VEPs
Early VEP studies of infant spectral sensitivity were conducted by Dobson (1976) who employed different monochromatic flash stimuli in two-month-old infants.It was found that peak times of flash cVEP components decreased with increasing age.Infants showed elevated responses at 15-18 weeks old to short-wavelength stimuli relative to responses elicited by long wavelengths compared to adults.This was similarly demonstrated by Moskowitz-Cook (1979) who found larger responses to short-wavelength stimuli in infants aged 19-22 weeks compared with adults.No significant differences between infants and adults were found to medium-to-long wavelengths.Other studies have explored the effect of heterochromatic flicker VEPs in infants and demonstrated some enhancement of short-wavelength sensitivity, but otherwise similar spectral sensitivity compared with adults (Bieber et al., 1995).It is unlikely that these early observations reflected changes in macular pigment or lens aging, as the rate of change or development of macular pigment and lens would not occur so rapidly in the first months of life (Bernstein et al., 2013).It is possible that the observed alterations in the spectral sensitivity reflect differences in foveal and perifoveal sensitivity for short wavelengths, because the best responses were found for cVEPs elicited by stimuli with large check sizes as can be expected from responses originating from the KC pathway.However, these early studies did not isolate cone responses, therefore rod intrusion cannot be excluded.
Later studies with stricter isolation of photoreceptor classes found inverse findings.Volbrecht and Werner (1987) found that infant S-cone-driven cVEPs (at 4-6 months of age) are in fact reduced in amplitude by approximately 1 log unit relative to adults.Furthermore, Knoblauch et al. (1998) elaborated on these findings using silent substitution stimuli, to 'silence' contributions of photoreceptors which were not of interest.They demonstrated that functional M-and L-cones can produce VEPs at four weeks of age, but with amplitudes that are around a quarter of those found in eight-week-old infants, whereas S-cone mediated VEPs were more attenuated.Whilst reduced short wavelength sensitivity has been inferred from sweep cVEP work (Suttle et al., 2002) it is likely that the temporal and spatial modulation techniques used in those studies did not allow sufficient stimulation of the KC system which has lower spatial and temporal resolution.These conclusions suggesting a reduced short-wavelength sensitivity in infancy are supported by other authors who have demonstrated that the B-Y cVEP develops slower than the R-G cVEP in infancy (Crognale, 2002).It is known in older age that sensitivity of the B-Y system decreases with increasing age, likely owing to yellowing of the lens and increased macular pigment density.

Maturation of transient cVEPs
The largest age-span of studied subjects has been performed by the group led by Crognale (Crognale, 2002;Crognale et al., 2001) who has reported R-G and B-Y modulated onset-offset grating cVEPs in subjects between 1 week and 93 years of age, including serially measured infants during their first year of life.It has was demonstrated here that cVEPs to R-G and B-Y onset stimuli show complex changes in the first year of life, adopting a positive-negative waveform, distinct from the adult-like negative-positive waveform, although an early negativity can sometimes be observed (Fig. 6).Madrid and Crognale (2000) explored the positive-negative morphology differences of the maturing cVEP using transient pattern onset-offset and reversal stimuli.They demonstrated that the onset cVEP between 2 and 6 years is dominated by a single positive peak at ~160ms followed by a broad large negativity.After this age, around pubertal age, the earlier negativity (~150ms) becomes more prominent at 6-14 years of age and then dominates the cVEP from 14 years onward as observed in the adult cVEP as the N1 component.These results are in agreement with other findings that a positive-dominant waveform shifts to a negative-dominant waveform at around pubertal age with no further morphology change thereafter (Crognale, 2002;Page and Crognale, 2005).The differences between paediatric and adult cVEP waveforms may be due to immaturities in the pattern of cortical activation or the number of activated neuronal populations (Boon et al., 2009).Assuming that the earliest negativity reflects chromatic pathway activity, we may infer that chromatic pathways mature slower than achromatic pathways.The maturation speed for L-M and S-cone axis stimuli are similar when elicited to 0.5c/deg sinusoidal gratings (Crognale, 2002;Madrid and Crognale, 2000).Pompe et al. (2006) later demonstrated the N1 as observable in 52% of children under 10 years of age, whereas the positivity could be recorded in all children.Pompe et al. (2012) also showed in younger children (2-6 years old) that cVEPs to R-G and B-Y stimulation are earlier to 7 • and 21 • field sizes, without clear age-dependent effects in amplitude.However, it is cannot be excluded that the dominant positivity observed in paediatric cVEP does not reflect or contain neurons from the achromatic luminance system, especially since two of these studies did not define stimuli based on individual isoluminance, and/or utilised large field stimuli in which retinal inhomogeneities in isoluminance are likely to have a larger effect (Crognale, 2002;Page and Crognale, 2005).The majority of cVEPs measurements in children mentioned above were performed binocularly.Pompe et al. (2006) demonstrated that the positive components in cVEPs are earlier and larger with binocular than with monocular stimulation.
Whilst in adult aging the general morphology remains mostly constant, the peak-time of the N1 increases with age by approximately 8-9.6ms per decade for R-G modulated gratings with spatial frequencies between 0.5 and 1c/deg (Crognale, 2002;Crognale et al., 2001;Page and Crognale, 2005).In addition, amplitudes decrease with age, perhaps more so for the B-Y than the R-G cVEPs, which likely reflects yellowing of the lens with age or other age related effects, although due to response variability this cannot be accurately quantified.

Considerations for isoluminance in young persons
As discussed in the methodology section, determination of individual isoluminance is an important factor in performing cVEPs.Photometric isoluminance (e.g.radiometric isoluminance measurement) does not Fig. 6.Example chromatic visual evoked potential waveform changes with age to red-green stimulation.The age of each child is indicated next to the corresponding waveform.The positive wave at 7 years of age (bottom plot) and the negative (N1) wave in adult at about 26 years of age (upper plot) are marked.The negative N1 wave can be observed from 9 years on.Reproduced with permission from Pompe et al. (2006).
account for individual variations in isoluminance (e.g.minimum flicker perception in heterochromatic flicker photometry), which may change due to alterations in the cone fundamentals with age.Therefore, the assessment of individual isoluminance points is recommended prior to testing.Fiorentini et al. (1996) demonstrated that isoluminance with red-green gratings in cVEP studies was achieved with more/less red luminance in young adults compared with 'elderly' adults.Pompe et al. (2006) studied 60 children age 7-19 years to red-green modulating onset-offset stimuli.These authors pragmatically performed heterochromatic flicker photometry (HFP) by displaying R-G sinusoidal gratings at 12.5 Hz prior to recording cVEPs, whereby the contrast of one of the colours was adjusted so that no flicker was perceived.The cVEP to R-G stimuli were performed at this point of isoluminance, including at 10% above and below this level to control for age-related alterations in cone fundamentals.In contrast with the findings of Fiorentini et al. (1996), they found no significant alterations compared with adults.
Consideration of isoluminance is important for paediatric cVEP recordings, as the presence of a positive peak in adults has been attributed to luminance (MC pathway) intrusion (Robson et al., 2006).However, whilst luminance intrusion cannot be excluded, it has been demonstrated that the morphology difference in the paediatric cVEP remains evident even following HFP to determine isoluminance, suggesting little luminance contribution when using large and small fields (Pompe et al., 2006).Whilst HFP can be a useful technique to maximise response selectivity, this can be challenging in children.Boon et al. (2007) found HFP was unsuccessful in the majority of their young participants.

Table 1
Properties and conclusions of studies of congenital colour vision deficiencies and retinal pathology.Moreover, the stimuli used to determine isoluminance should match that of the cVEP.In those children where HFP may not be possible, the point of photometric isoluminance has often been used as it is comparable to the majority of paediatric individual isoluminance conditions.Boon et al. (2007) aimed to explore the relationship between cVEPs and psychophysical development of L-M sensitivity in children school age and older; they found that cVEPs have higher sensitivity to chromatic stimuli than psychophysical measures, in which this difference could be hypothesised reflect luminance intrusion into the cVEP signal given that HFP was not possible in their participants.Morrone et al. (1990;1993) and Kelly et al. (1997) measured the chromatic contrast sensitivity function (cCSF) in infants using the cVEP.They showed that in the first 6-9 weeks of age, there was a diminished cVEP CSF to R-G chromatic stimuli when compared to those of adults.In older infants, the cCSF was found to be similar to adults.It was found that luminance and chromatic CSF develop independently at differing rates, with cVEPs developing more slowly.Such differences in luminance versus chromatic pathway function suggests that the maturation effect is in the post-receptoral pathways rather than in the photoreceptors.Kelly et al. (1997) further demonstrated that chromatic CSFs as a function of spatial frequency are low pass for all ages from 2 months to adulthood, but that the overall sensitivity was lower at younger age.Page and Crognale (2005) found that onset cVEP isoluminant thresholds increased with age for L-M and S-stimuli.Importantly, they did not find these differences with achromatic stimuli, suggesting a difference in aging of chromatic versus achromatic pathways (Allen et al., 1993).

Clinical applications
The cVEP is a technically demanding technique.Whilst many efforts have attempted to demonstrate their diagnostic utility in clinical conditions, this review highlights that there has not been systematic comparison of cVEPs to achromatic (i.e.ISCEV standard) VEPs to entirely determine their additional value relative to conventional VEP techniques.Nevertheless, cVEPs have shown a variety of clinical benefits alone in conditions affecting the visual system which are discussed below (summarised in Table 2).

Congenital colour vision deficiencies
cVEP recordings were performed with the aim of establishing a noninvasive assay for congenital colour vision deficiencies (CVDs).The stimuli used can be broadly categorised into two different types of stimuli (see Table 1): (1) heterochromatic (mainly red-green) modulation and (2) silent substitution/cone isolating stimuli.Most recordings were performed with spatial gratings, checkerboards or multifocal stimuli.Full-field stimuli were also used sporadically.Whilst stimuli for these studies were varied in terms of modality and type, these seemed to have little influence on the general results concerning the detection of colour vision deficiencies.However, as we will discuss in this section the stimulus characteristics play an important role in interpreting the results.

Heterochromatic stimuli
Most subjects with congenital (hereditary) colour vision deficiencies (CVDs) either have L-or M-cones with substantially different fundamentals (in anomalous trichromats) or one of the two is absent and thus the L-/M-ratio is 0 (in protanopes) or infinite (in deuteranopes).As a result, the stimulus condition for subjective isoluminance deviate when compared with those from normal trichromats.Thus, the point of subjective isoluminance may indicate the presence of a CVD and also the type of CVD.However, some trichromats have relatively large L-/M-ratios (of up to 10, indicating that they have many more L-than Mcones; nevertheless, they have normal colour vision).The R/(R+G) setting for subjective isoluminance of these observers may therefore be difficult to distinguish from those of deuteranomalous trichromats and of deuteranopes.The genes for L-and M-cone pigments are on the Xchromosome.As a result, males are more often affected than females.Hereditary tritanopia (absence of S-cones) occurs less often, probably because the gene for S-cone pigment is on chromosome 7 instead of the X-chromosome.On the other hand, acquired tritanopia occurs more often (Simunovic, 2016;Tekavcic-Pompe and Tekavcic, 2008).
The heterochromatic flicker paradigm to obtain individual isoluminance conditions has also been used in ERG recordings.At the conditions for subjective isoluminance, the ERGs are found to display a minimal response when stimuli with relatively high temporal frequencies (>25cyc/sec) are used (Aher et al., 2018;Kremers and Bhatt, 2016;Kremers et al., 2000).Accordingly, the minimal response criterion is used to determine a point of individual isoluminance.As a result, the conditions of minimal ERG response were found to be different between normal trichromats and subjects with CVDs, although ERGs can provide additional information.The psychophysical isoluminance points and those obtained with ERGs correlate quite well (Jacobs and Neitz, 1993;Kremers et al., 2000).At intermediate temporal frequencies (between 8 and 14cyc/sec), the ERG responses were found to be mainly determined by the red-green chromatic content of the stimulus, strongly suggesting that the responses reflect activity of the R-G (PC) pathway.CVD subjects that lack a functional R-G chromatic pathway can show this in the ERG at intermediate temporal frequencies and provide valuable additional information in detecting the presence of a CVD.Indeed, at the intermediate frequencies the responses from dichromats (deuteranopes and protanopes) were consistently different from those of trichromats, showing that they can be used to establish the presence of a CVD (Aher et al., 2018;Kremers et al., 2010;Parry et al., 2012).
Several studies investigated if cVEP recordings with heterochromatic stimuli can also be used to determine the presence of a CVD and the data can possibly interpreted in the same manner as the ERGs.Indeed, cVEPs were found to be very useful to determine the presence and the type of colour vision deficiency.In most of these cVEP measurements, relatively high temporal (reversal) frequencies (more than 15 Hz or more than 15 cyc/sec; see Table 1) were used (Bach and Gerling, 1992;Gerling et al., 1997;Hoeve et al., 1996;Zheng et al., 2021).These measurements similarly used a minimum response criterion, showing that a response minimum was found at R/(R+G) values (i.e. the fraction of red contrast normalized to the sum of red and green contrast when red and green are modulated in counterphase) that also resulted in subjective isoluminance, suggesting that it is possible to determine the presence and the type of colour vision deficiency.The R/(R+G) values are expected to be larger than 0.5 for protanopes and protanomalous trichromats (they need stronger red modulation to counteract green modulation at subjective isoluminance) whereas they will be smaller than 0.5 for deuteranopes and anomalous trichromats (where stronger green modulation is required to reach subjective isoluminance).However, since these R/(R+G) values for isoluminance are also smaller than 0.5 for normal trichromats with high L-/M-ratios it still remains to be determined if the results of deuteranopes and trichromats with large L-M-ratios can be distinguished.
In one study reporting cVEP measurements with children, lower temporal frequencies (1 Hz onset-offset) were used (Tekavcic-Pompe et al., 2010).The measurements were performed with a fixed R/(R+G) ratio of 0.5 (which is photometric isoluminance).Normal trichromats showed a substantial response to this stimulus, despite the fact that the stimulus was close to isoluminance for them, suggesting that, similar to the ERG data, the response originated in the red-green chromatic (PC) pathway.The responses in dichromats and anomalous trichromats were substantially smaller than those measured in age matched normal trichromats indicating that, again similar to ERGs, lower frequency stimuli may be very useful to determine if a colour vision deficiency is present.

Table 2
Summary of major studies found in the review: clinical question, stimulus characteristics and are described.Studies whereby isoluminance was photometrically determined or not described (e.g.heterochromatic flicker or other psychophysical determinance of isolumance) are asterisked next to 'isoluminant'.High temporal frequency stimuli may then be used to determine the type of CVD by determining the individual point of isoluminance, for example r > 0.5 for protanomalous trichromats and protanopes or r < 0.5 for deuteranomalous trichromats and deuteranopes.Because most trichromats are L-cone dominated, the difference between normal and protanomalous isoluminance is generally larger than the difference between normal and deuteranomalous isoluminance.
In a few experiments with heterochromatic stimuli, m-sequences were used (Sutter, 2001).In one study, the spatial presentation was a hexagonal multifocal configuration (Klistorner et al., 1998).In two other experiments, the heterochromatic stimuli were gratings (Martins et al., 2019;Risuenho et al., 2015).Furthermore, instead of red-green counterphase modulation the two stimuli were presented alternately.This type of stimulation is superficially very similar but results in a dissociation of luminance and chromatic signals, with luminance modulating at twice the frequency of the chromatic modulation (Lee et al., 2011;Parry et al., 2012).The results of these measurements were less conclusive, also because the different response kernels needed to be considered.

Cone isolating stimuli
The cVEP measurements to cone isolating stimuli were generally performed at lower temporal frequencies (less than 4rev/sec) (Barboni et al., 2017(Barboni et al., , 2019;;Crognale et al., 1993;Rabin et al., 2016).It should be highlighted to the reader that whilst the stimuli used in some of these studies may be described as 'cone isolating', there are post-receptoral connections from M-and L-cones to both MC and PC cells.Therefore, whilst the best efforts may be 'isolate' one chromatic pathway, it is possible that cVEPs to cone isolating stimuli contain contents from both chromatic and luminance pathways.Whilst achromatic intrusion may be more evident to transient grating cVEPs through alteration in waveform polarity or morphology, those to 'cone isolating' stimuli may be less selective.Most of these stimuli were presented in onset-offset mode.One exception are the studies of Barboni et al. (2017Barboni et al. ( & 2019) ) who used reversals with sawtooth temporal profiles at 4 Hz.It was generally found that the cVEP responses in dichromats to stimuli that isolated the respective absent cone were strongly reduced and/or delayed relative to those of trichromats.Interestingly, however, the responses also to the functional cone type were reduced.In psychophysical studies, the sensitivities to stimulation of the present cone type at low temporal frequencies were also reduced in CVD subjects when compared with normal subjects (Huchzermeyer and Kremers, 2016).As for the heterochromatic stimuli, this can be interpreted that the cVEP responses to low temporal frequency stimuli are mainly determined by the red-green chromatic (PC) pathway in trichromats.In dichromats, this system loses its cone opponency resulting in a decreased sensitivity at low temporal frequencies (i.e. also when stimulating the present cone type).
In conclusion, most experimental data indicate that, similar to ERGs, cVEPs can be used to determine if a congenital CVD is present.Furthermore, the type of CVD can principally be determined.However, similar as with psychophysical and ERG data, the temporal frequency of the stimulus should be considered.At low temporal frequencies, cVEP responses to stimuli containing a chromatic component are generally reduced, possibly because of a post-receptoral loss of cone opponency.This has the advantage that the presence of CVD can be easily established.However, for determining the type of CVD, higher temporal frequencies may be preferable.
ERGs to low temporal frequency stimuli often contain higher harmonics, i.e. response components at multiples of the stimulus frequency.These components have characteristics that are similar to the fundamental components of ERG responses to high temporal frequencies (Aher et al., 2018).As a result, the ERGs to low temporal frequency in principle can also be used to establish the presence of a CVD (by using the fundamental components) and the type of CVD (by using the higher harmonics).Possibly, similar high frequency components can be identified in cVEPs to low temporal frequency stimuli.This remains to be established experimentally ideally with larger cohorts of subjects.Although ERGs may give cleaner data, cVEPs may still be interesting for measuring subjects that cannot cooperate easily such as children.
Most cVEP studies using high temporal frequencies in CVD do not consider the response phase or timing, although they potentially can give useful additional information: At high temporal frequencies and at the response minima, ERG change their phases by 180 • .When pattern reversals are used, the cVEP response is mainly at the reversal frequency and thus at twice the stimulus frequency.A 180 • phase shift at response minimum is therefore expected not be present in the pattern reversal VEP.Possibly, asymmetric reversals (i.e. using different times between the two reversals) may be useful for obtaining additional information about the VEP response timing.Again additional experiments are necessary to establish if response phases and timing can give additional information.
The comparison between cVEP and ERG data may also provide information if and how contributions of the luminance and the R-G chromatic pathways to visual activity differ in retinal and cortical locations, thereby possibly providing useful insights about information processing in the intermediate loci.Barboni et al. (2019) performed ERG and VEP responses to identical cone isolating stimuli within the same subjects.The stimuli had sawtooth temporal profiles to separate responses to increments (On-) and decrements (Off-).As mentioned above, it was found that responses to the present cone type were substantially reduced in the cVEPs of dichromats compared to trichromats.The ERGs to these stimuli were of similar amplitude or even larger in dichromats in comparison with those in trichromats.This indicates that the red-green chromatic pathway has a larger contribution in the VEP than in the ERG.This may be an interesting consideration for future experiments.Till et al. (2005) Exposition to organic solvents Are different visual pathways affected by organic solvent exposure?
Isoluminant* red-green, blue-yellow and achromatic stimuli Responses to red-green stimulus were demonstrated to be affected, whereas to blue-yellow and achromatic ones were not.

Retinal pathology
The involvement of chromatic mechanisms in retinal pathology has also been evaluated using cVEP responses, although these have been limited to achromatopsia (ACH), cone dystrophies (CD), retinitis pigmentosa (RP) and central serous chorioretinopathy (CSCR).
ACH and CD frequently impair central visual acuity and colour vision significantly (Michaelides et al., 2004), and are often associated with photophobia.The spectrum of colour vision abnormalities can be wide but often severely affects all chromatic systems, but can be gene and condition specific to the type of CD.It has been demonstrated that cVEPs can reflect residual central cone function more reliably than standard full-field ERGs in children with CD and ACH (Kelly et al., 2003).Whilst this may be unsurprising given that full-field ERGs are a generalised retinal response and VEPs are a test of macular function, this nevertheless suggests that cVEPs may be useful in such disorders to monitor residual cone function and perhaps delineate the phenotype of the chromatic disturbance in such cases.Their value in addition to achromatic (e.g.ISCEV Standard) VEPs is yet to be determined.
RP is a group of rod-cone retinal dystrophies in which patients typically exhibit night blindness, peripheral visual field loss, electroretinogram (ERG) abnormalities associated with rod and cone driven signals (affecting the rod system first), alongside bone-spicule-like pigmentation and narrowing of the retinal vessels.Dyschromatopsia is common in RP, and can often relate to the extent of macular cone dysfunction.cVEPs have been used to study chromatic contrast processing in RP patients, whereby both MC and PC pathways are reduced in RP, and to a similar extent as the responses to achromatic contrast VEPs (Alexander et al., 2005).
CSCR is characterised by an idiopathic retinal detachment at the posterior pole of the eye caused by leakage of fluid from the choroid into the subretinal space through a barrier defect in the retinal pigment epithelium.Patients often report a central scotoma, metamorphopsia, decreased visual acuity and dyschromatopsia.VEP responses are delayed in CSCR patients with greater delay for chromatic than for achromatic stimuli, which is consistent with the notion that the central cones in particular are more severely affected in CSCR (Wu et al., 2011).
In summary, despite the known retinal pathology affecting the mechanisms of chromatic visual processing, cVEPs have been used to study chromatic responses in various retinal pathologies, mostly comparing them with achromatic responses dominated by the MC pathway.

Glaucoma
Glaucoma leads to progressive damage of RGCs and their axons, which is the main cause of optic neuropathy and vision loss in this condition.Acquired colour vision deficiency is well described in glaucoma patients and glaucoma suspects using behavioural tests, which have shown a predominantly early deficit to blue colours (Pacheco--Cutillas et al., 1999;Sample et al., 1986).cVEPs can be used to ascertain differences or the extent of dysfunction within specific chromatic and achromatic responses, primarily to differentiate between possible dysfunction of the PC, MC or KC pathway in different subtypes of glaucoma as well as in glaucoma suspects.Original hypotheses suspected a specific damage to the MC stream based on anatomical work that large ganglion cell axons were preferentially affected in glaucoma (Quigley et al., 1987(Quigley et al., , 1988)).However, this hypothesis has since been largely disproven and subsequent work has shown similar dysfunction of both the MC and PC pathways in glaucoma (reviewed in Bach & Hoffman ( 2008)).There have been a number of studies investigating cVEP responses in patients with glaucoma which supports this pathophysiological basis (Accornero et al., 2008;Horn et al., 2000;Korth and Koca, 1993;Korth et al., 1994;Mierdel and Marre, 1978).The study by Greenstein and co-workers has shown that both the MC and PC pathways are affected in patients with open angle glaucoma, whereas only the MC pathway appears to be affected in glaucoma suspects similar to that observed behaviorally (Greenstein et al., 1998).In this study, sweep VEPs were recorded to sinusoidally modulated isoluminant R-G checks upon a steady background, or upon a luminance modulating background.However, another study from Porciatti et al. has demonstrated that glaucoma damage in open angle glaucoma is not selective to the MC pathway (Porciatti et al., 1997).The study has suggested that isoluminant colour-contrast VEP stimuli may be of diagnostic value in grading glaucoma induced dysfunction.In this study, cVEPs were recorded to reversals of isoluminant R-G gratings and of yellow-black (i.e. luminance) gratings.A study from Horn and co-workers has shown that blue-yellow isoluminant VEP stimulation has 90% sensitivity in differentiating patients with visual field losses from healthy controls, whereas the sensitivity of responses to red-green stimuli is about 70% and of those to achromatic stimuli only 30% (Horn et al., 2000).The B-Y onset-offset VEP response was demonstrated to be affected in a pre-perimetric group of open angle glaucoma patients (i.e.patients that display glaucomatous damage of the optic disc without visual field losses) (Dussan Molinos et al., 2022;Horn et al., 2002).Similar findings were observed in another study where phakic and pseudophakic glaucoma patients were included and results were compared to healthy controls, whereby the B-Y cVEP latencies were predominantly affected in glaucoma patients (Fuest et al., 2015).
In summary, although the distinction between the different chromatic pathway dysfunction in glaucoma patients is still not entirely consistent, it appears that the B-Y VEP is altered earliest and most markedly.This is evident even before visual field damage can be observed, suggesting that the B-Y based KC pathway may provide early indication of neuronal damage in these patients.
6.6.Optic nerve and neurological disorders 6.6.1.Demyelinating disease Demyelinating diseases (DD) can affect the visual system through disturbed conduction of action potentials along the optic nerve fibres, with optic neuritis (ON) often being the first manifestation of the disease.Classical episodes of acute ON consist of reduced visual acuity, positive relative afferent pupillary defect, central visual field scotoma and altered colour vision.During an acute episode of ON, colour vision abnormalities are often diffuse and affect both R-G and B-Y systems, although after resolution it has been reported that R-G deficits more commonly persist relative to B-Y (Flanagan and Zele, 2004;Katz, 1995).Achromatic pattern reversal VEPs are routinely used to diagnose and monitor patients with ON.In acute disease these often show attenuated and prolonged responses.After the acute phase, amplitude may improve but in the majority of adults, peak-time delay often persists despite improvement in visual acuity (Halliday et al., 1972;Marmoy and Viswanathan, 2021).It is known that DD primarily affects PC and KC systems (Frederiksen et al., 1993a).Studies of cVEPs in adults (Porciatti and Sartucci, 1996;Sartucci et al., 2001) and children (Tekavcic-Pompe et al., 2020) with DD have shown a high prevalence of abnormalities, confirming vulnerability of these chromatic visual pathways.It has been demonstrated that the peak-time of cVEP responses are particularly altered in patients with and without a previous history of ON (Tekavcic-Pompe et al., 2020).However, R-G and B-Y mechanisms seem to be affected in the majority of young DD patients but to a larger extent in those patients with one or more episodes of ON, suggesting higher sensitivity to detect visual pathway dysfunction in DD patients than achromatic stimuli or PERGs (Tekavcic-Pompe et al., 2014).It is noted that the higher sensitivity of cVEPs reported in this study may reflect that the stimulus was selective for its respective chromatic pathway, whereas achromatic VEPs may contain activity from both M and P neurons (depending on their temporal, spatial and contrast properties) meaning their abnormality may be less marked relative to a selective pathway response.Other studies have suggested altered spectral sensitivity measured through the cVEP which may be consequent on these abnormalities of chromatic visual pathways (Frederiksen et al., 1993a).
In summary, cVEPs reflecting activity of PC and KC pathway activity are particularly impaired in patients with DD, which is sometimes to a greater extent than to achromatic stimuli, perhaps as cVEPs assess more selective function of respective retino-geniculate pathways.

Parkinsons disease
Parkinsons disease (PD) is one of the most common neurodegenerative disorders.Visual dysfunction in PD patients includes abnormal contrast sensitivity, difficulties in motion perception, reduced visual acuity and impaired colour vision.Colour vision deficiency reported in PD has been varied, although some have suggested this predominantly affects the tritan axis (Birch et al., 1998;Haug et al., 1995).The idea of using cVEPs to demonstrate selective chromatic pathway dysfunction is not new (Bodis-Wollner and Yahr, 1978).A study by Büttner and co-workers has demonstrated the presence of a correlation between severity of motor symptoms and cVEP latency in PD patients, although the study has found better sensitivity of achromatic than cVEP responses (Buttner et al., 1996).On the other hand, a study by Sartucci and Porciatti (2006) has demonstrated that cVEP responses to B-Y stimuli have the highest sensitivity, suggesting greatest vulnerability of the KC pathway in PD consistent with psychophysical studies demonstrating tritan axis abnormalities.The results of the work of Barbato et al. (1994) indicate that cVEP responses also correlate with L-DOPA therapy efficacy in PD.
In summary, cVEP have been used in PD to study possible impairment of chromatic mechanisms and L-DOPA therapy efficiency.Both chromatic and achromatic mechanisms are impaired in PD patients, the cVEPs elicited by B-Y stimuli to a greater extent than those to R-G stimuli.

Diabetes mellitus
Diabetes mellitus (DM) is a relatively common endocrinologic disorder in which visual deficits occur due to vascular pathology, metabolic imbalance and delayed neural conduction along the visual pathway.Colour vision was affected in about half of the patients included in the Early Treatment of Diabetic Retinopathy Study (ETDRS), with a blueyellow deficiency being the predominant abnormality (Barton et al., 2004).Chromatic and achromatic VEPs have been used in DM patients to assess the function of MC, PC and KC visual pathways, as they differ in their susceptibility to the disease.A study by Schneck et al. reported an increased susceptibility of the KC pathway to acute changes in blood glucose levels which affects cVEPs to B-Y stimuli (Schneck et al., 1997).Another study included patients with type 1 and type 2 DM with good metabolic control and no clinical features of diabetic retinopathy, but with signs and symptoms of peripheral neuropathy; a correlation between the delay of the cVEP response to R-G stimuli and neuropathy was demonstrated, whereas the VEP delays to achromatic and B-Y stimuli were not correlated with neuropathy.However, all VEPs were affected to a greater extent in DM type 2 compared to type 1 (Gregori et al., 2006).
In summary, cVEPs have been studied in DM patients in an attempt to detect preclinical visual pathway dysfunction and to find possible correlation between central and peripheral nervous system dysfunction.R-G mechanisms were shown to be affected to a greater extent then B-Y, especially in patients with DM and peripheral neuropathy, however MC, PC and and KC mechanisms were shown to be affected in poorly controlled DM.

Amblyopia
Amblyopia is defined as reduced vision in one or both eyes caused by abnormal visual development in early life.The main site of dysfunction in amblyopia is post-retinal, with deficits occurring in the primary visual cortex and extra-striate visual areas (Seignette and Levelt, 2018).The idea to study the cVEP in patients with amblyopia arose after the suggestion that MC mechanisms are relatively spared in amblyopia compared to PC mechanisms.The Australian group led by Mei-Ying Boon has shown that cVEP responses differ between amblyopic and non-amblyopic children (Boon et al., 2016).The cVEP responses in this study were variable in amblyopes and this variability may reflect initial abnormal adaptations and also recovery as they were undergoing treatment.However, the study was not longitudinal and the definite effect of treatment on cVEP responses needs to be further studied (M.Y.Boon; personal communication).
In summary, cVEP responses differ between amblyopic and nonamblyopic children, suggesting that chromatic processing mechanisms are affected in amblyopia, but the exact mechanisms and clinical utility of cVEPs in amblyopia remain to be investigated.

Dyslexia
Dyslexia is a general term for a deficit in the acquisition of adequate reading skills and occurs despite the absence of neurological, cognitive, sensory and social disabilities in children with normal intelligence and normal educational opportunities.It is known that MC processing is primarily affected in dyslexic children, but there is also evidence of PC involvement, that can be assessed with the cVEP.A study by Bonfiglio et al. demonstrated delayed VEPs to both achromatic and isoluminant R-G and B-Y chromatic stimuli and was the first to demonstrate PC visual pathway involvement (Bonfiglio et al., 2017).
Patients with dyslexia also show abnormalities in cVEP responses, but may be less specific for the underlying mechanism of dysfunction.Therefore, these works support the notion that the site of dysfunction may arise at a post-retinal level, perhaps within cortical visual processing.

Other clinical conditions
Sporadically, there are further studies on cVEP with various possible applications in clinical conditions.A study by Fujita et al. has demonstrated PC visual pathway dysfunction in autism spectrum disorder by using cVEP to R-G isoluminant stimuli as well as achromatic VEPs.It was found that the PC visual pathway more affected than the MC pathway in people with autism (Fujita et al., 2011).In another study, similar recordings for investigating PC and MC visual pathway function were utilised in adult patients with Trisomy 21 syndrome comparing the responses with those obtained from a control group.The cVEP responses of patients with Trisomy 21 syndrome showed different characteristics both in the morphology and latency/amplitude of the response as well as high inter-individual variability compared to the control group (Suttle and Lloyd, 2005).
Another interesting clinical application is the study of cVEP responses in children prenatally exposed to organic solvents.It was found that R-G cVEP responses were significantly affected in these children, whereas B-Y and achromatic VEPs were unaffected (Till et al., 2005), suggestive of a vulnerability of the PC pathway.
In summary, cVEPs may be useful in other clinical conditions, such as autism, Down syndrome and organic solvents exposure, but intriguingly the applications of cVEP are still being explored.

Recommendations for clinical testing
The methodology used to record the cVEP are evidently complex, therefore a major aim of this review was to explore the literature on cVEPs to provide some guidance on their recording in practice or for future research studies.Whilst there has been some heterogeneity in methods, we below suggest some major practical considerations for recording the cVEP.These are not exhaustive, and we hope that those attempting to record cVEPs are mindful of the potential issues of stimulus and recording parameters raised throughout this review.Accordingly, we propose two methods in which those wishing to record cVEP O.R. Marmoy et al. may utilise in clinical practice: (1) transient cVEPs (2) heterochromatic flicker VEPs.

Technical considerations & patient preparation
The recording of a cVEP should follow the specifications of electrodes, placement, patient preparation nd signal analysis described for international Standard VEPs (Odom et al., 2016).If full-field or diffuse unstructured stimuli are used, pupil dilation may be required.In the investigation of known or reported colour vision defects, it is helpful to perform psychophysical measurements of colour vision beforehand to guide the testing approach for cVEPs.Recording of the response may be monocular or binocular and likely determined by the clinical question and patient cooperation.
It is most pertinent prior to any cVEP recordings for the patient to perform heterochromatic flicker photometry in order to determine the patient's individual point of isoluminance.This may be performed using the same stimulus parameters as used for the cVEP (specifically, the same stimulus size), but by altering the relative ratio of R-G or B-Y luminance or contrast at a high temporal frequency (>30 reversals) until the patient experiences minimal perception of a flicker in the stimulus.This can then be noted as a the chromatic contrast ratio (i.e.R/[R+G]) and used to guide the point of isoluminance in the cVEP recording.Of note, heterochromatic flicker photometry is a subjective test so may not be easily possible in younger children or those who are poorly cooperative.In such circumstances, the point of photometric isoluminance (i.e. a ratio of r = 0.5 or b = 0.5) may be used, but caution is advised to observe for any potential luminance intrusion in the waveform morphology (see interpretation and reporting section).

Stimulus field and spatial frequency
Stimulus field is an important parameter for the cVEP and cannot be easily recommended without consideration of its clinical utility.The cVEP is sensitive to changes in macular pigment, particularly when using B-Y stimuli, therefore the size of the stimulus field may reflect the pigment heterogeneity at the retina and alter the relative point of individual isoluminance.This review has highlighted that small field sizes (≤10 • ) reduce these effects, however given that a stimulus field of 10 • may still introduce chromatic aberration with a B-Y stimulus, a practical balance is required to enable a sufficient field size for fixation and robust waveform amplitudes, alongside maintaining adequate response selectivity.We recommend that for cooperative observers, a field size of ≤10 • is likely a compromise between reducing achromatic intrusions and enabling sufficient field size for fixation and cVEP amplitude (Fig. 2), however it should be considered, particuarly in good observers (and when utilising B-Y stimuli), smaller field sizes (e.g.<5 • , optimally 3 • ) are more chromatically selective.It should be an exception that, in children or those unable to cooperate with fixation requirements of small fields, that larger field sizes up to 21 • may be used.In such circumstances, it should be acknowledged that the resultant cVEP is less selective during interpretation due to potential achromatic intrusion, therefore interpreted with great caution.If possible, hetrochromatic flicker testing would be performed prior to testing in all of these patients.
The stimulus itself should be circular in shape to attempt to further avoid potential luminance artefacts.For both R-G and B-Y stimuli a spatial frequency of 1-2cyc/deg appears to be the most robust spatial frequency and matching the spatial frequency for both chromatic stimuli allows some direct comparison between post-retinal systems (Fig. 7).Stimuli should be horizontal gratings with equal size of both colour elements with an equal number of isoluminant gratings.There is negligible difference between sinusoidal or square-wave gratings, with the latter producing larger amplitude responses.

Luminance
Ambient room lighting is appropriate for recording the cVEP, although darkened room lighting may allow better qualification of stimulus luminance and spectra as to not be disturbed by room lighting.Mean stimulus luminance has been highly variable across studies, but it is suggested that stimuli between 30 and 60 cd/m 2 are suitable for clinical recordings and the mean luminance should be maintained in the photopic range.However, retinal illuminance rather than luminance per se is of importance.Therefore pupil size has to be taken into account.

Spectral characteristics
The majority of sources in this review have opted to use colour isoluminant stimuli.Whilst KC pathway mediated cVEPs may be elicited from S-cone driven stimuli modulated along the tritan confusion line, and PC pathway mediated by those modulating M-and L-cones, to more selectively stimulate each respective parallel pathway colour opponent stimuli have been used in the majority of the reviewed papers.These stimuli are chosen to maximise and bias responses toward each respective parallel pathway.It is challenging to designate specific colours or wavelength contents to which cVEPs should be recorded in these means as this may likely, least in part, be determined by laboratory equipment or manufacturer specifications.
If cone contrasts need to be maximised then it is advisable to use light sources with narrow band spectral outputs because they are able to separate the excitations of the different cone types.Of note, to obtain adequate visual stimulators for cVEPs may be challenging.Cathode-raytube devices provide high achromatic contrast, temporal characteristics and synchronisation, but the spectral outputs of the differently coloured sources are often broadband.Other monitor devices may be better but they should be adequately controlled for luminance artefacts (i.e. in LCD monitors) or timing jitter (i.e.OLED monitors).Modern solutions are also potential options, such as the Digital Light Processing laser projectors which have recently been assessed (Marmoy and Thompson, 2023) and also applied in recording of cVEPs (Dussan Molinos et al., 2022).Alternatively there are exciting developments in technology, particularly for development of multi-primary sources to enable up to six chromatic channels (Cao et al., 2015;Franke et al., 2019).It is likely that these new devices will need further validation and experimental interrogation but are likely to provide more precise spectral control in cVEP stimuli in the future.
The peak wavelengths of the light sources also influence the maximal possible cone contrasts and optimization algorithms can be used to determine the peak wavelengths for maximal cone contrasts (Here additional considerations may also play a role.For instance, it may be important that not all cone contrasts but only of a subgroup of cones should be maximised.Furthermore, the mean luminance and mean chromaticity of the stimulus influence the maximal cone output; therefore luminance and chromaticity requirements should be considered).Generally, an even distribution of the peak wavelengths across the visual spectrum gives sufficient cone signal strengths.
Stimuli may also be expressed in the Macleod-Boynton and Derrington-Krauskopf-Lennie spaces.Here it is important to consider that the L-M axis maximises cone opponency but may not result in a isoluminant stimulus because the ratio of L-to M-cone inputs to the luminance system is generally larger than unity and may vary between different individuals (necessitating the above describes assessment of the individual isoluminance point).It is emphasised for these reasons that heterochromatic flicker photometry is performed for each individual patient prior to testing to account for differences in L-or M-cone populations.
All visual stimuli should be appropriately calibrated with a calibrated spectroradiometer to verify the spectral properties of the stimulus.Similarly, a photometer should be used to calibrate the photopic luminance of the stimulus screen.

Temporal frequency
Temporal frequency is largely determined by the type of response wished to be obtained; (1) Transient stimuli are likely to be of most benefit to those exploring post-retinal pathways of the PC or KC systems.These have been perhaps the most studied modality of cVEP used in the sources of this review.As such, their behaviour, properties and stimulus characteristics are most understood.Typically these are recorded to R-G or B-Y stimuli to bias responses toward the PC or KC pathways, respectively.(2) Heterochromatic flicker stimuli are stimuli of high temporal frequency, producing a steady-state response, likely to be of use in determining colour vision deficiency type such as in retinal or congenital colour vision defects.

Transient cVEPs
Transient cVEPs recorded using R-G or B-Y stimuli are best produced to onset-offset mode of stimulation.Typically an onset-duration of 200-500ms should be used with the entire recording epoch of 1000ms used (or sufficient enough to capture the entirety of the onset-offset response waveform).A summary of recording characteristics is provided in Table 3.

Heterochromatic flicker cVEPs
Heterochromatic flicker cVEPs are recorded with high temporal frequency.Their responses are mainly determined by the luminance pathway when stimuli are not isoluminant.At isoluminance, the most effective temporal frequency for performing heterochromatic flicker VEPs has not been determined, but typically these are recorded to >30 reversal stimuli.
For determining acquired colour vision changes it may be good to use a similar strategy as described for subjects with colour vision deficiencies (CVD): The use of a low temporal frequency (4-15 Hz) may be useful to determine if colour vision is abnormal.High temporal frequencies (>30 reversals) can then be used to determine how colour vision has been changed.However, additional data from systematic experiments are needed to determine the procedures and temporal frequencies.A summary of recommended recording characteristics is provided in Table 3.

Interpretation and reporting
As with many clinical electrophysiological tests, it is imperative that responses are compared to a normative reference range to determine their normality relative to a healthy population.Age-related effects in cVEP morphology likely make this comparison challenging in early age, although the adult-like negative dominant waveform is usually evident from the second decade of life.Reference ranges should therefore account for differences in age, sex and binocular versus monocular amplitude recording scenarios.
Prior to any assessment of waveform amplitude and/or peak-time, the validity of the response requires confirmation.It has been demonstrated throughout this review that for adult onset cVEP waveforms, early positive peaks are reflective of luminance intrusion (e.g.Kulikowski et al. (1997a)) and are therefore considered non-specific for any particular colour system.Several methods (particularly for validation of B-Y cVEPs) have been proposed to assess specificity of responses.The fastest and perhaps easiest was proposed by Robson et al. (2003Robson et al. ( , 2006) ) whereby the specificity was quantified by providing an index of negative and positive amplitudes within the early cVEP onset waveform, calculated as [onset negativity/(onset negativity+onset posivity].Those responses with an index of 1 have the highest selectivity, whereas those with an index of 0 have the poorest.Additional methods proposed by Kulikowski et al. (1996) require additional or exploratory testing.The cVEP at isoluminance should be larger for onset than reversal stimuli, therefore any positive peaks suspected to result from achromatic intrusion could be quantified to stimuli of shorter and equal onset-offset periods, thereafter longer onset-periods used to quantify more colour specific amplitudes.Secondly, for B-Y cVEPs a high temporal frequency isoluminant stimulus can be used which should be severely reduced in amplitude, unless there is a luminance component to the stimulus.These methods may be used to provide confidence or validation in the specificity of a response reflecting a chromatic mechanism.For the adult negative-dominant waveform, peak-time and amplitude of the major first negativity (N1) should be recorded, with amplitude measured from baseline to N1.In children with positive-dominant waveforms, this should be specified as a positive-dominant morphology and measured as peak-time of the major positive peak (P1) and amplitude from the baseline or preceding negativity.During reporting, clear reference should be made in regard to the morphology of the waveform.
It can be inferred that a relative dysfunction of a chromatic mechanism, to R-G or B-Y stimulation, will indicate a relative dysfunction of its respective post-receptoral PC or KC pathway, respectively.As such, transient cVEPs may be reported in this manner should they deviate from reference ranges.Age effects should be taken into account alongside its effect on the cone fundamentals.Heterochromatic flicker cVEPs may be interpreted by identification of the response minima, to determine whether this fits the reference data for the colour-normal observer or whether there is a significant deviation to suggest an abnormality.Care should be taken when using R-G stimuli for normal or shifts toward red stimuli as the ratio of L-to M-cones may be very high in certain colour normal subjects, which may be difficult to distinguish from those subjects with deutan abnormalities (Kremers et al., 2000).
It is recommended in clinical and research reports that the peakwavelength or CIE coordinates of respective colour stimuli are defined, including the point of isoluminance if determined through photometric measurements.It would, for example, be useful if reporting studies provided relative cone contrasts of stimuli to provide a relative functional perspective of data.The field size, spatial frequency and mean luminance should also be described for clear ability to replicate experiments.

Future perspectives
This review paper has demonstrated that cVEP can be a useful method in evaluating chromatic mechanisms in health and disease.However, as always the information obtained should be considered in combination with other clinical (for example colour vision tests) and visual electrodiagnostic tests (Standard VEPs to achromatic stimuli, ERGs).
The authors suggest that future studies should focus upon the exact spectral combinations required to record a cVEP.There has been limitation in the exact wavelength combinations which are most efficient or robust to elicit the cVEP, both in healthy participants but also in those with altered colour vision.Furthermore, stimulators used for cVEP recording may produce different spectral contents and characteristics.For example, typical Cathode-ray tube (CRT) monitors tend to display colours of a broad spectra, as such may not be as specific to receptoral systems.Other devices, such as DLP or LED based stimuli, may produce narrowband spectra and as such can be more easily quantified and controlled for cVEP measurements.Accordingly, the specifications for stimulus characteristics of recording a cVEP may be beneficial to ensure there is comparable physiological mechanisms when cVEPs are compared between centres.
For future studies, it will be important to describe stimuli in a physiologically meaningful manner at different levels.The expression at the level of the photoreceptor in terms of cone-or rod-td for the mean excitation and cone-and rod-contrast for the stimulus strength gives a better quantification of the input to the visual system.Currently, sufficient data about cone fundamentals are available to perform this quantification.For future research there will be a need in studying interindividual variability at the photoreceptor level and its influence upon cVEPs.As mentioned above, several sources of variability need to be considered: spectra of fundamentals, number and density of photoreceptors, preretinal absorption (lens and macular pigment) and variability in the influence of retinal eccentricity and retinal locus.Future research should also consider post-receptoral pathways.The fundamental principles of signal processing within the major retinogeniculate pathways are relatively well known.Since these pathways underlie basic aspects of visual perception, the results can be compared with psychophysical data.Describing the cVEP stimuli in terms of the expected responses in the major colour sensitive pathways (i.e. the redgreen sensitive PC pathway and the KC pathway that is sensitive to blueyellow changes) may give a direct quantification of the input to the cortex and comparison with psychophysical data may give input about how the brain processes incoming information.Finally, a comparison of cortical chromatic processing and cVEPs may give an overview about which cortical processes are relevant for the cVEP.This aspect of chromatic visual processing can be further studied by combining cVEP stimulation and functional MRI (fMRI) or other means of cortical activity detection.This information would be useful also in different clinical conditions affecting central visual processing, as well as in studying maturational and developmental changes of central visual processing.
Whilst the effects of differing field size has been fairly well explored, our review has demonstrated that the change in spectral sensitivity and isoluminance with eccentricity can affect the cVEP.This has received rather limited attention, but further work may further apply this technique or examine alternative dartboard or radial stimuli based stimuli to account for spatial inhomogeneities.These would enable more controlled large stimulus fields for applications in those with unstable fixation, such as in paediatric testing.
Also in the temporal domain, stimuli could be developed that are less artificial as those that are currently in use (i.e.flashes, step changes or repetitive stimulation) and that include characteristics that are also found under natural circumstances.Possibly, aspects like attention and motivation play a larger role in VEPs than anticipated.The use of e.g.movies or including tasks with rewards may make the recording of VEPs more attractive when testing patients with lower cooperation levels.Currently, it is possible to describe the complete spatial and temporal structure of the stimulus for an extended time.A cross-correlation between measured response and the stimulus may reveal VEP characteristics that are not available or suppressed in VEP recordings obtained under more laboratory conditions.This would also give the opportunity of disentangle the contribution of different chromatic mechanisms from each other and from the luminance or brightness driven system.
A number of studies have compared the alterations in cVEP relative to psychophysical colour vision testing.However, there has been little study of the benefit of cVEP against commonly used clinical colour vision tests, and whether the cVEP may be an appropriate supplement for these tests.This may provide more accurate and quantifiable data which would be useful for outcome measures in future clinical trials, as well as disease monitoring.
Whilst it seems an attractive prospect to develop reference ranges for interpretation of cVEP data, it is likely that the variability in stimulus type and chromaticity may limit comparability between centres.It is possible that those centres using spectrally, temporally and luminancematched stimuli may be able to use pooled reference data, but at the time of writing the variability in devices producing cVEP stimuli may vary too significantly for widespread comparability of data.The present paper may be used as a reference and a guide for those centres attempting to start with cVEP recording.
One unusual finding from our review was the possibility that the R-G pathway may have a larger contribution in the VEP than the ERG.It would be interesting to evaluate whether this is resultant of enhanced processing of R-G signals within the intracortical pathway (i.e. at the LGN) or at a cortical level.Furthermore, the determination of subjective isoluminance may vary between the ERG and VEP which warrants further study.
It is evident from the present review paper that cVEP has been

Fig. 3 .
Fig. 3. Effect of stimulus contrast on the cVEP.Reproduced with permissions from Porciatti and Sartucci (1999).Panels A and B show the cVEP to pure chromatic contrast (defined as the ratio of R, G, B or Y to total luminance, whereby a luminance contrast of 0.5 would provide 100% chromatic contrast), with panel C showing the response to luminance contrast.Stimuli were 2c/deg onset-offset (300ms onset illustrated by triangles on x-axis, 700ms offset) horizontal sinusoidal gratings within a 14 • field.The chromatic responses to R-G and B-Y show the typical negative configuration with those to luminance (Y-black) showing the typical positive waveform.The peak-time for all stimulus conditions increased when relative contrast (i.e.contrast relative to the other colour) decreased (as indicated by the lines connecting the main response components).This increase in peak time was larger for the responses to R-G and B-Y stimuli than for those to the Y-black luminance stimuli.The numbers to the left of the traces denote the relative contrast.

Fig. 4 .
Fig. 4. Relative spectral sensitivity of cVEPs relative to Wald's V λ constructed from data from Moskowitz-Cook (1979) and Riggs and Sternheim (1969).Wald's V λ is shown as the thick black curve.Moskowitz-Cook data (redrawn) are shown for cVEP responses (dashed red line) and psychophysical measurements (solid red line) showing good agreement.Riggs and Sternheim data (redrawn) are shown (solid blue line).These data show close agreement to Wald's V λ apart from short wavelength where there is an apparent underestimation of spectral sensitivity (~430-510 nm).

Fig. 5 .
Fig. 5. Illustrative waveforms of the transient onset-offset cVEP.Traces in A are those elicited to 2c/deg red-green onset-offset gratings (red trace) and 2c/deg blue-yellow onset-offset gratings (blue trace) subtending a 15 • circular field in a healthy participant.The major negative component N1 is noted.Traces in B are modified from Tekavčič Pompe et al. (2012) for a four year old subject.These were elicited to red-green (red trace) and blue-yellow (blue trace) gradings of 2c/deg subtending a 21 • circular field.The major positive component P1, as typical of paediatric cVEPs, is noted.

Fig. 7 .
Fig. 7. illustrative examples of red-green and blue-yellow square wave grating stimuli.The circular field subtends 10 • with a spatial frequency of 1cyc/deg.

Table 3
proposed stimulus characteristics for performing Chromatic Visual Evoked Potentials (cVEPs).