CHAPTER4 COLOUR AND THE CORTEX: WAVELENGTHPROCESSING IN CORTICALACHROMATOPSIA CHARLESA. HEYWOOD, ROBERTW. KENTRIDGE, AND ALAN COWEY Introduction One of the perceptual consequencesof being able to processwavelength differences in the distribution of spectrallight is the chromatic world in which we live. The privacy of colour experienceis undisputedand is reflectedin the term'quale'coinedby C.I. Lewisin 1929to describesuch qualitative content of mental states.However, notwithstanding the philo- sophicaldebate that such a notion hasfuelled, few would deny that we live in a colourful world-never more so,perhaps, since the l8-year-old William Henry Perkin serendipitously discovereda purple aniline dye while attempting to synthesizequinine from coal tar in the mid-nineteenthcentury, and spawnedthe wealthof dyesand pigmentsthat arecommon- placetoday. What doesit mean to havecolour vision, i.e.what are its defining characteristics,and what advantagesdoes it confer on its possessor?Ascribing colour vision to an organism requiresno more, and no less,than the demonstrationthat two spectrallydifferent stimuli, made equally bright with respectto an animal's spectral sensitivity,are discriminable. Alternatively,colour vision can be demonstratedwhen discrimination remains possible when random fluctuations in brightness are introduced into two stimuli of different spec- tral distributions,i.e. the discrimination must be made on the basisof colour differences alone.Almost all vertebratesand some invertebrateshave colour vision, some in a rudi- mentary form, but differ in the number of photoreceptortypes and their spectralcharac- teristics.Trichromacy, however, is the norm for all Old World monkeys,apes, and people. But what role doescolour play in vision?One way this canbe addressedis by assessingthe variation in colour vision acrossa variety of specieswith respectto the visual environ- ments they inhabit. Good examplesare the demonstration that the spectraltuning of long- and middle-wavelengthretinal conepigments of frugivorousplatyrrhine monkeysis optimal for detecting their dietary fruits concealedagainst a variegatedfoliage (Regan et al., 1998)whereas that of African Old World monkeys and chimpanzeesis appropriatelr' tuned for detectingyoung green shoots among other and lesstasty foliage.This revealstn'o of the selectivepressures that haveoperated on primate vision during the evolutionot dichromacyand trichromacy.One role of colour vision, therefore,appears to be the rapid detectionof a particularobiect colour rvhenluminance differences alone rvould providc CULUUK AI\U TN ambiguousclues to its location. Moreover,a stationary object standsout i': ' '.rck- ground on the basisof, among others,luminance, texture, and chromati. ':.es. Spatialor temporalvariations of luminancewould maskan object'scontours ,, :Jr it invisible to a monochromatic observer,suggesting a further role for colour '. - :he segmentationof the visual scene.Finally, colour variation assistsin the rc'., - ': ,ri objects.For example,there is ampleevidence that colour playsa conspicuousr,' . ...r1 signalling,the identificationof conspecificsor, to usea much citedexample, an ..' . -:: of the ripenessof dietaryfruit from its externalappearance. However,there are attendant problems in the designof a systemwhere reflect:.. - ' differentially reflect light of wavelengthsthat constitute the visible spectrunr spectralcomposition of the illuminant can vary widely from moment to mt':- throughout the day.In the natural environment,much of the former variation dc:'. whether an object is illuminated by direct sunlight or the shorterwavelengths ot .. illumination producedby Rayleighscattering. The latter arisesfrom a combinati,': greatersusceptibility of long-wavelengthlight to atmosphericabsorption and thc :: . attenuationof shorterwavelengths as a resultof scattering,the effectof which incr..,- - ' the sun approachesthe horizon.It haseven been suggested that the principalopf,'::. axis of trichromatic vision, blue-yellowand red-green,is an adaptationto such nat..:. - occurring variation in terrestrial illumination (Shepard,1992) . Yet colours remaiIr ; . ceptuallyconstant, and suchconstancy must be a prerequisiteof a biologicalsystem rr'ir :. i-. fulfils the purported roles assignedto it, namely the facilitation of object detection. segmentationand, most notably,identification. Another clue to the role of colour vision can be derivedfrom behaviouralstudies oi peoplein whom colour vision is depleted,as in casesof retinal colour blindness,or r.'hcr. colour vision hasbeen perhaps entirely deletedas a result of a cerebralaccident. It is n,'',' establishedthat damageto human ventromedialoccipitotemporal cortex .un .J.15..;t,., '..: vision, a condition known ascerebral achromatopsia (Damasio et al., 1980;Kijlmcl. " ' Heywood et al., l99l). What light can the examinationof such patientsshed on thc :'. - siblecosts and benefitsto thosewho possesschromatic vision? One consequenceof brain injury canbe a selectivedisorder of vision.f{1r\r{iii. ii - selectivityof a deficitis only asnarrow or broadas the specificitvof the bch.rvi,,.:r.,.i.,'r that is usedto studyit. Examplescome from visualagnosia (the lossoi lisu.rl l.u'.::, ': :.,- : . - orauditoryrecognitionofobjects,withoutlinguistic,sensor\-,orattentiott.tl l,'.. ..:: :' -.' agnosia,appropriate tasks can further pinpoint the impairn.rentin sontcP.rli(r::. .,\ .: \- .. tiveloss of visualrecognition of living comparedu'ith non-living,itenr:. or litL ..:. ":'i, other patients.Alternatively, the recognitionof facesmav be selectilclv J i.l .:: :'. .: ..- ' prosopagnosia,but evenin the latterdisorder deficits can be confinedto .iit:i.'.:.:.r. i . recognitionoffacialexpression,gender,oridentity(seeCorver'. 199-1.J,';;1'. 1q:. .'.l -- the refinementof behaviouraltasks can more accuratelv def-ine n'hat is lti.l.'.r::.:: ' ':' is equallyinformative about the componentProcesses of a particularI irtt.i.: -.:r.: i certainvisual functions, particularly those that arecomnronlv dc'scri[.cc .i. .' ;' might seemunlikelv that the surprisingdissociations of the sort descril'cJi:: ::'.. : * -- orderprocessing of objectand face rvill be observed. For ntnenronic or liniu:.::- .:':'-." colourprocessing, brain damage can disrupt the abiliti'to nantL'.1 it,..';.::r.i '..'- :' ' - l.t'rintto ir namedcolour, as in colouranomia iOxburv cr,l/.. 1969 . (r!-\::.:.::': :.-- .- but sparethe latter,as in disordersof short-term colour memory (Davidoffand Ostergaard, 1984).The deficit may be confined to a failure to respond correctly when askedto provide the appropriatecolour name when confrontedwith the verballabel of a common object, 'What for example, colour is a banana?'(Kinsbourne and Warrington, 1967;Lszzattiand Davidoff, 1994).However, in none of theseinstances is there any difficulty in telling colours apart and colour vision itself, that is the experienceand discrimination of colours, remainsundisturbed. Nevertheless, our experiencewith monochrome imagesreadily allowsus to imagine a world devoid of colour. This, and the evidencethat colour can be processedpreattentively (Tieisman and Gelade,1980) and may therefore constitute a visual primitive, together with the abundant physiological evidencethat colour is processedrela- tively independently,might encourageus to believethat processingof wavelengthvariation in the visual scenecan be selectivelydestroyed. The descriptionofpatients with cerebral achromatopsia,also called cortical colour blindness,where brain damageappears to have selectivelyabolished colour vision, confirms this belief (seeZeki, 1990;Cowey and He1'wood,1997, for reviews).But perhapsthe very easewith which we can (unlike, for example,the perceptionof shape)imagine a world in which colour hasbeen deleteddis- couragesexamination of the degreeto which other aspectsof wavelengthprocessing may be spared.Thus, colour vision being the phenomenalcounterpart of the processingof wavelengthvariation, the loss of the former may be held to imply an absenceof the latter. Does the absenceof colour vision preclude the use of wavelengthdifferences to determine other objectproperties? It is becoming increasinglyapparent that the preoccupationwith the discrimination paradigm to characterizethe qualities ofthe visual input, by requiring decisionsto be made about a stimulus array,neglects the nature of the responserequirements (seeMilner and Goodale,1996,for review).Conventionally, it is assumedthat whether visual choices in a discriminationtask entail a verbal,pointing, or any other motor responseis unimpor- tant. Recentreports make it is clear that certain motor output systemshave privileged accessto different visual inputs. For example, intact visual pathways can still accomplish visuomotor or other orienting responses,albeit in the absenceof consciousawareness of the properties of the visual stimuli that elicit them. And the clinical condition of blindsight is characterizedby an absenceof acknowledgedvisual awarenesspatients in patients who neverthelessperform well at forced-choiceguessing tasks. Blindsight has been considered 'unconscious an example of vision',but equally likely it is the residual motor capacitiesthat remain, but which do not contribute to the consciouspercept. This view has been used to account for visuomotor abilities in casesof visual form agnosia,notably patient DF, whose impaired discrimination of orientation and shapecoexists with accurateand appropriate visuomotor actsrequiring the coding of thesestimulus attributesfor their successful execution