Making the incredible credible: Afterimages aremodulated by contextual edges more than real stimuli

Georgie Powell $School of Psychology, Cardiff University, Cardiff, UK

Aline Bompas # $CUBRIC/School of Psychology, Cardiff University,

Cardiff, UK

Petroc Sumner # $School of Psychology, Cardiff University, Cardiff, UK

We explored whether color afterimages and faint physical chromatic stimuli are processed equivalently by the visualsystem. Afterimage visibility in classic illusions appears to be particularly influenced by consistent contexts, while realstimulus versions of these illusions are absent in the literature. Using both a matching and a nulling paradigm, we presentconverging evidence that luminance edges enhance the perceived saturation of afterimages more than they do physicalstimuli of similar appearance. We suggest that afterimages violate the response norms associated with real stimuli. Thisleads to the afterimage signal being ambiguous for the visual system, and thus more susceptible to modulation by contextsthat increase or decrease the probability of the signal representing a real object. This could explain why afterimages arerarely experienced in everyday life, where they will be overruled by inconsistent context.

Keywords: color, afterimages, context, luminance edges, ambiguity

Citation: Powell, G., Bompas, A., & Sumner, P. (2012). Making the incredible credible: Afterimages are modulated bycontextual edges more than real stimuli. Journal of Vision, 12(10):17, 1-13, http://www.journalofvision.org/content/12/10/17,doi:10.1167/12.10.17.

Introduction

Negative color afterimages occur when adaptation toa particular hue results in the subsequent illusoryperception of its complementary hue. From this wecould conclude that afterimages are simply negativeimages, or perceptions arising from a shift in relativeactivity of cells early in the visual system (e.g., retinal).According to this view, the independent adaptation ofphotoreceptor cells and subsequent shifts in the activityof opponent processes are mainly responsible for thegeneration of afterimages (Brindley, 1962; Craik, 1940;von Kries, 1970; Zaidi, Ennis, Cao, & Lee, 2012). Theassumption follows that these signals are then pro-cessed by higher level visual areas equivalently tosignals arising from any real stimulus.

However, perceptually, afterimages do not alwaysbehave in the same manner as percepts of real objects.This could suggest that the visual system does notprocess afterimage signals equivalently to those arisingfrom real stimuli. For example, afterimages areperceptually unstable: they can fade in and out ofconscious perception (Comby, 1909; Wade, 1978).Moreover, they are highly dependent on context: insome situations we see them clearly, in others we do

not, even given equivalent adaptation conditions (Daw,1962). Classic afterimage illusions, such as the Spanishcastle illusion (Sadowski, n.d.) and the recent illusionby van Lier, Vergeer, and Anstis (2009), demonstratecompelling effects of contextual modulators on after-image visibility. In these illusions, afterimages embed-ded in a consistent context (usually a luminance edge orcontour) are unequivocally visible, yet without suchcontext they are much less visible, or even invisible.However, variants of these illusions displaying similarvisibility modulation for real physical stimuli arenotably absent in the literature. These observationsled us to ask whether the representations of afterimagesand real stimuli are substantially nonequivalent in thevisual brain, and specifically, whether contextual cues(such as luminance contours) are particularly impor-tant for afterimage visibility.

Previous psychophysical and electrophysiologicalresearch has demonstrated that luminance or chromaticedges are important in facilitating the perception of,and cellular responses to, physical chromatic stimuli tosome extent (Friedman, Zhou, & von der Heydt, 2003).Perceptually, luminance contrasts (contours and ped-estals) facilitate detection and discrimination of phys-ical chromatic stimuli (Chaparro, Stromeyer,Kronauer, & Eskew, 1994; Cole, Stromeyer, & Kro-

Journal of Vision (2012) 12(10):17, 1–13 1http://www.journalofvision.org/content/12/10/17

doi: 10 .1167 /12 .10 .17 ISSN 1534-7362 � 2012 ARVOReceived March 15, 2012; published September 29, 2012

nauer, 1990; Eskew, Stromeyer, & Kronauer, 1994;Gowdy, Stromeyer, & Kronauer, 1999; Gur & Akri,1992; R. Hilz & Cavonius, 1970; R. L. Hilz, Hupp-mann, & Cavonius, 1974; Montag, 1997; Mullen &Losada, 1994). In particular, a flashed suprathresholdluminance pedestal or contour (ring) facilitates detec-tion of a coincident chromatic target (Chaparro et al.,1994; Cole et al., 1990; Eskew et al., 1994). Addition-ally, weak, blurry chromatic signals spread (i.e., fill in/fill out) until they reach a luminance edge (von derHeydt, Friedman, & Zhou, 2003). A demonstration ofthis process can be seen in the watercolor and Boyntonillusions (Mollon, 1995; Pinna, Brelstaff, & Spillmann,2001). At a physiological level, orientation selectivityand heightened responses to edges are common featuresof visual cortex cells (Friedman et al., 2003). There isalso evidence of facilitatory interactions in the primatestriate cortex (V1) between cells sensitive to luminancecontrast and color (Horwitz, Chichilnisky, & Albright,2005).

It is possible that contexts, such as consistentluminance edges, are important cues to disambiguatereal objects from variations in lighting. Most realobjects possess clear luminance edges (Fine, MacLeod,& Boynton, 2003; Hansen & Gegenfurtner, 2009; Zhou& Mel, 2008), whereas this is not consistently the casefor features less significant to awareness, such asvariations in lighting (Kingdom, 2008), reflections,and afterimages. It is known that perceiving a coloredsurface as material rather than a light figment leads toan increase in its perceived saturation, suggesting thatthe visual system actively enhances the perceptions ofobjects (Bloj, Kersten, & Hurlbert, 1999). Althoughthere are many possible means of distinguishing lightfrom materials (see Kingdom, 2008, for a review), onepossibility is that the visual system has learned toacknowledge, or evolved to enhance, faint chromaticsignals when a luminance edge is present and disregardthem when it is not.

We were interested in whether a color afterimagethat is matched to a real stimulus in hue, saturation,luminance, and edge blur and is viewed under fixationconditions is simply another example of a faint,chromatic stimulus. If this is the case, we would expectafterimages and real stimuli to be affected by context toa similar extent. Our general question is whether thebrain treats afterimages and real stimuli as the same orwhether there is an added uncertainty to afterimagesignals that renders them more susceptible to contex-tual modulations.

There may be a number of features of afterimagerepresentations that distinguish them from responses toreal stimuli, thus making afterimage signals moreambiguous. Importantly, we are not suggesting thatafterimage signals and chromatic responses to realobjects are not generated by the same cellular

populations. Rather, we suggest that the nature oftheir responses—the temporal profile and distributionof signal strengths across different brain areas—maynot be identical in all respects (see general discussionfor elaboration). The visual system may then be facedwith a dilemma over whether the afterimage signalshould be perceived or suppressed. A surroundingcontext that is consistent with an ambiguous signalwould raise the probability that it represents a realobject, and thus raise the likelihood that it is perceived.If the signals underlying afterimages are by their naturemore ambiguous than those for weak real stimuli, thenwe should expect that afterimage visibility will benefitmore from a consistent context compared with aphysical stimulus of similar appearance.

We designed a series of complementary experimentsto directly compare the enhancement effect of lumi-nance edges on both color afterimages and supra-threshold physical stimuli. We used a simultaneouscomparison task in which the chromatic contrast(saturation) of a physical stimulus was adjusted tomatch that of an afterimage (Experiment 1) and anulling task in which the afterimage was nulled by aphysical stimulus of complementary hue (Experiment2). We then explored whether the effect of the contouron afterimages and physical stimuli was found for anysurrounding edge, even if it was blurry, by substitutingthe contour for a sharp or blurred luminance pedestal(Experiment 3). We report converging evidence thatsharp luminance edges (contours and pedestals) en-hance afterimages to a greater extent than they dophysical stimuli of similar appearance.

Method and materials

Observers

For Experiment 1, eight observers (seven naive, oneauthor; five males, three female) participated in boththe afterimage and physical stimulus comparison tasks.A total of four observers (three naive, one author; threemales, one female) participated in Experiments 2 and 3.All had normal color vision and normal or corrected-to-normal visual acuity.

Apparatus

Stimuli were presented on a 21-inch Sony GDM-F520 Trinitron monitor (Sony) at 100 Hz, controlled bya Cambridge Research Systems (CRS) ViSaGe (Cam-bridge Research Systems) and a PC running Matlab(Manufacturer). Stimuli were viewed binocularly at adistance of 72 cm while the observer’s head was

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 2

maintained by a chin rest. Manual responses were madewith a CRS CB6 button box. Eye movements wererecorded by a CRS high-speed video eye trackersampling at 250 Hz.

Experiment 1: Afterimage and physicalstimulus comparison task

Experiment 1 involved two stages. In the first stage,we measured the effect of luminance contours on theperceived saturation of afterimages by asking observersto compare the saturation of afterimages, framed ornot by a contour, with physical comparison stimuli thatvaried in saturation. In the second stage, we measuredthe effect of the luminance contour on the perceivedsaturation of a physical reference stimulus that wassimilar to the afterimage in hue, luminance, andcontrast. In other words, we substituted a real faintstimulus in place of the afterimage and based theproperties of the physical stimulus on the afterimagematching results from stage one.

Pilot studies

Before beginning Experiment 1, we conducted twopilot studies to ensure that our physical comparisonpatches were as similar in appearance to the afterim-ages as possible. One pilot calculated the appropriateamount of edge blur to introduce to the physicalpatches as a precaution against a chromatic edgeovershadowing any effect of the luminance contour.The edge blur associated with afterimage perceptsarises from fixational jitter during adaptation shiftingthe edges of the adaptation region. The second pilotperceptually equated the hue and luminance of physicalpatches with the afterimage, since in the mainexperiments only the chromatic contrast would bemodulated.

Stage 1 stimuli and procedure: Afterimagemeasurement

The chromaticity of the stimuli were originallycalculated in MacLeod and Boynton color space(MacLeod & Boynton, 1979), but are reported in CIEchromaticity coordinates (x,y) and luminance incandelas per square meter (Y) (Smith & Guild, 1931)for convenience. The adapting stimuli were green (x ¼0.252, y¼ 0.487, Y¼ 28.8) or pink (x¼ 0.303, y¼ 0.171,Y¼ 28.8) 38 diameter circles, presented 38 to the left orright of center. The physical comparison patchessubtended approximately 38 (see explanation of edgeblur below). One observer completed a pilot with astaircase to equate the hue and luminance of thecomparison patches with the afterimages for a range of

saturation levels (0, 10, 20, 30, and 40% of the adaptingstimulus saturation). A second pilot was conducted tomodel the degree of edge blur predicted by fixationaljitter during adaptation. One observer adapted to thesame green/pink circles described above for 1.5 s across360 trials, and eye movements were sampled every 4ms. The retinal position (derived from the eye trackingdata) of the stimulus at each 4-ms point duringadaptation was then simulated, and these positionswere translated into predicted stimulus edge blurduring adaptation. This edge blur profile was thenused to draw the comparison patches.

Example trials are shown in Figure 1. All phases ofthe experiment were conducted on a grey background(x¼ 0.308, y¼ 0.316, Y¼ 28.8). During each trial of themain experiment, 8 observers fixated a black 0.158diameter central dot and were adapted for 1.5 s oneither the left or the right of fixation. Immediatelyfollowing adaptation, one of five comparison patcheswas presented for 700 ms on the opposite side toadaptation. In half the trials, a 38 diameter contour (x¼0.308, y¼ 0.316, Y¼ 22.25) was presented opposite thecomparison stimulus (to frame the afterimage). Ob-servers were required to respond whether the left orright patch was ‘‘more saturated.’’ To reduce carryoveradaptation, the comparison patches were followed by a600-ms animated mask consisting of multiple 38 circlesrandomly changing position and chromaticity at 100Hz. There were a total of 40 trial types within a 2(adaptation hue) · 2 (contour presence) · 2 (testpresentation side) · 5 (comparison patch saturation)within-subjects design. Observers received 10 repeti-tions of each trial type, totaling 400 trials, presented ina random order. Data were collapsed across testpresentation side, resulting in 20 observations at eachcomparison patch level for each adapting color.Individual observer psychometric functions were fittedfor each condition. From these we extracted the pointof subjective equality (PSE), which represents thesaturation level of the comparison patch perceived asequal to the afterimage.

Stage 2 stimuli and procedure: Physical stimulusmeasurement

The physical stimuli comparison task was identicalto the afterimage task, but without the adaptationphase. The physical reference stimulus was a green orpink 38 diameter circular patch (with the same edgeblur as the comparison patches) presented 38 to the leftor right of center. The saturation value of the physicalreference stimulus was set to the saturation level thatmatched the afterimage without a contour from stage 1.This was derived from the PSE between afterimage andphysical stimulus in the no contour condition. Thesevalues were set individually for the green and pink

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 3

afterimages and for each observer. The remainingdesign was the same as the afterimage stage.

Experiments 2 and 3: Nulling task

In Experiment 2, we used a nulling task in which aphysical stimulus of complementary hue was added tothe afterimage until grey was perceived. The saturationof the physical stimulus (with and without a contour)was adjusted until the afterimage percept was nulled(the point of perceived grey; PPG). If afterimages areenhanced by luminance contours to a greater extentthan physical stimuli of similar appearance, we wouldpredict that more saturated physical stimuli arerequired to null the afterimages when both are framedby a contour. If the contour were to enhance bothafterimage and physical nulling stimulus by the sameamount, then these enhancements would cancel and thecontour would produce no shift in the PPG (only anincrease in slope might be expected due to enhancedcolor discrimination). We also repeated Experiment 2with luminance afterimages to examine whether theeffect was limited to color afterimages. All conditionsof Experiment 2 were conducted in central vision whilestimuli were presented peripherally in Experiment 1,which allowed us to investigate whether any effect ofthe contour differed with eccentricity.

Experiment 3 followed the same nulling procedure asExperiment 2 but employed a luminance pedestal inplace of a contour. Our hypothesis was that any

luminance edge would produce afterimage enhance-ment, whether contour or pedestal. It was important toconfirm this because although the afterimage illusionsthat motivated our research used luminance contours,most previous reports of chromatic facilitation ofphysical stimuli by luminance edges have used pedes-tals. Further, to examine the extent to which any effectof a pedestal was due to its edge rather than to thepresence of a luminance component in the chromaticpatch, we tested both sharp- and blurry-edged pedes-tals. Previous research has suggested that facilitation ofphysical chromatic stimuli by luminance signals isconsiderably more powerful with sharp, as opposed tograded, luminance differences. For example, discrimi-nation of chromatic gratings is enhanced by superim-posed square-wave luminance to a greater extent thansine-wave luminance gratings (Gowdy et al., 1999).

General apparatus and stimuli description are thesame as Experiment 1, unless stated otherwise. InExperiment 2, observers fixated a central dot and wereadapted for 1.5 s to a central 38 diameter green or pinkcircle. Immediately following adaption, one of theseven nulling patches was presented for 700 ms. Thenulling patches were drawn with edge blur profiledetermined in the pilot study for Experiment 1 andconsisted of seven saturation vectors taken from thetwo adaptation hues: 10, 20, and 30% of pink andgreen, and one 0% (grey). The nulling patch wasframed by a grey 38 diameter contour in half the trials.Observers were required to manually respond whetherthe patch appeared ‘‘pinkish’’ or ‘‘greenish’’ by pressing

Figure 1. Example trials from Experiment 1. (a) Afterimage task, no contour condition. Observers fixate centrally and are then adapted to a

pink or green circle on the left or right. One of five comparison stimuli is then presented on the opposite side to adaptation. Observers

were required to indicate whether the left or right patch was ‘‘more saturated.’’ Trials ended with a mask to reduce carryover adaptation.

(b) Afterimage task, contour condition. Procedure is identical but a contour is presented on the adaptation side during the test. (c) Physical

stimuli task, no contour condition. Procedure is identical to afterimage task but without an adaptation phase and with a physical chromatic

reference stimulus in place of the afterimage. (d) Physical stimuli task, contour condition.

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 4

the appropriate response button. The nulling patches

were followed by a 600-ms mask (described in

Experiment 1). There were a total of 28 trial types

within a 2 (adaption hue) · 2 (contour presence) · 7

(nulling patch hue/saturation) within-subjects design.Observers received 10 repetitions of each trial type,totaling 280 trials.

For Experiment 3, the task, stimuli, and procedurewere as described for Experiment 2. The singledifference was that the contour was replaced by eitherby a sharp-edged luminance pedestal or a blurry-edgedluminance pedestal that followed the blur profile usedto draw the chromatic edges. There were a total of 42trial types within a 2 (adaption hue) · 3 (pedestal type/presence) · 7 (nulling patch hue/saturation) within-subjects design. Observers received 10 repetitions ofeach trial type, totaling 420 trials. Example trials forExperiments 2 and 3 are shown in Figure 2.

Results

Experiment 1: Afterimage and physicalstimulus comparison task

Figure 3 shows results from Experiment 1 anddisplays an interaction between contour and stimulustype, such that the difference between the perceivedsaturation in the contour and no contour conditionswas larger for afterimages than for physical stimuli(F[1, 7] ¼ 11.66, p , 0.01). No main effect of stimuluscolor was found and color did not interact with thestimulus type or contour presence.

Figure 2. Example trials from Experiment 2 and 3. (a) Experiment 2, no contour condition. Observers centrally fixate and are then adapted

to a green or pink circle. One of seven nulling patches is then presented and observers are required to indicate whether the circle appears

‘‘pinkish’’ or ‘‘greenish.’’ Trial ends with a cycling mask to reduce carryover adaptation. (b) Experiment 2, contour condition. Procedure is

identical but the nulling patch is surrounded by a luminance contour. (c) Experiment 3, sharp pedestal condition. Procedure is identical to

Experiment 2, but a sharp pedestal is superimposed on the nulling patch. (d) Experiment 3, gradient pedestal condition. Same as above,

but with a blurry pedestal.

Figure 3. Results of Experiment 1. Mean point of perceived equality

(units are percentage of adapting stimulus saturation) of afterim-

ages (black) and real stimuli (white) in the contour and no contour

conditions, across 8 observers. The significant interaction is clearly

shown, whereby the contour enhanced the perceived saturation of

afterimages more than physical stimuli of similar appearance. Error

bars show, for each condition, the standard error of the differences

from each participant’s mean (i.e., they are derived from the portion

of the variance that is relevant for within-subject tests by excluding

the irrelevant main effect of subject). The mean for each point is

derived from 20 data readings per observer.

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 5

These results indicate that even though the physicalstimuli were similar to the afterimages in hue,luminance, degree of edge blur, and saturation, theluminance contour increased the perceived saturationof the afterimages significantly more than it increasedthe perceived saturation of the physical stimuli.

Experiments 2 and 3: Nulling task

Figure 4a plots the PPG across four observers forExperiment 2, which represents the amount of physicalstimulus saturation required to null the afterimage. Forall observers, the PPG in the contour condition isshifted further away from physical grey than in the nocontour condition (t[3] ¼ 6.51, p , 0.01). This findingconfirms that more physical stimulus saturation isrequired to null an afterimage that is framed by acontour. We found no consistent increase in slope inthe contour condition, providing no evidence that thecontour improves discrimination of the combinedstimulus (the afterimage and the null). A steeper slopein the contour condition is predicted by previousresearch reporting that luminance edges improvedetection and discrimination of real chromatic stimuli(Chaparro et al., 1994; Cole et al., 1990; Eskew et al.,1994; Gowdy et al., 1999; Gur & Akri, 1992; R. Hilz &Cavonius, 1970; R. L. Hilz et al., 1974; Montag, 1997;Mullen & Losada, 1994). These findings have also beenreplicated by us in a pilot study, results of which areshown in supplementary information.

The contour effect on afterimages is not specific tochromatic afterimages. Figure 4b shows the results of aluminance afterimage nulling experiment (procedureidentical to Experiment 2, but observers adapted tolight, monochromatic patches and the resultant after-images were framed by a light contour, dark contour,or no contour). As shown in Figure 4b, both light anddark contours tended to shift the PPG relative to the nocontour condition, indicating that the afterimageenhancement exceeded that for the physical nullingpatches.

Figure 4c shows results from Experiment 3, in whichsharp and blurred pedestals were presented instead of acontour. The mean shift in PPG across observers isgreater in the sharp pedestal condition compared withthe no pedestal and blurry pedestal conditions (F[2, 6]¼8.88, p , 0.05). Posthoc tests revealed that theperceived saturation of the afterimages in the sharppedestal condition was significantly greater than that inboth the blurry pedestal and no pedestal conditions (p¼0.041 and p ¼ 0.023, respectively). In contrast, theblurry pedestal condition was not different to the nopedestal condition (p ¼ 0.554). This suggests that thebenefit in the sharp pedestal condition depends on thepresence of a sharp edge rather than the mere presence

Figure 4. Results from Experiments 2 and 3. Grey lines represent

individual observers; black is the mean across observers.

Calculation of error bars is the same as described for Experiment

1. There are 20 data readings for each individual observer point.

(a) Experiment 2. Point of perceived grey (PPG; in percentage of

adapting stimulus saturation) in the no contour and contour

conditions. Across all observers, more physical saturation is

required to null the afterimage in the contour condition compared

with the no contour condition. (b) Nulling procedure replicated with

luminance afterimages. Individual data from four observers in the

no contour, light contour, and dark contour conditions. The PPG,

or the point at which nulling patch is judged perceptually equal in

luminance to background, is plotted in log units from the

background. Dark contours consistently shifted the null point

compared with the no contour condition. Light contours shifted the

PPG in a similar direction for three out of four observers. (c) Mean

physical saturation required to null the afterimage (PPG; units are

percentage of adapting stimulus saturation) across sharp, blurry,

and no pedestal conditions across four observers. Figure

illustrates that a sharp pedestal increases the perceived

saturation of the afterimage compared with the no pedestal

condition, whereas a blurry pedestal does not increase the

saturation of afterimage above the no pedestal condition.

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 6

of a luminance increase congruent with the colorgradient. For all but one observer the psychometricslope in the sharp pedestal condition was steeper thatthe no pedestal or blurry conditions. This suggests thatthe discrimination of the combined stimulus (thenulling stimulus and the afterimage) was improved bya superimposed sharp pedestal, a finding consistentwith previous findings for the enhancement of realchromatic stimuli. Further, the importance of edges forchromatic perception is demonstrated in examples ofchromatic spreading and ‘‘filling in,’’ such as theBoynton and watercolor illusions (Mollon, 1995; Pinnaet al., 2001). Why we observed improved discriminationin the sharp pedestal condition here, but not with theluminance contours in Experiment 2, is unknown.

Taken together, the results of Experiments 1 through3 suggest that the contextual modulators of afterimagevisibility seen in recent compelling afterimage illusions(Sadowski, n.d.; van Lier et al., 2009) is not observed tothe same degree for real, faint chromatic stimuli.

Thus, although previous research has demonstratedthat real chromatic stimuli are enhanced by luminanceedges, our findings indicate that afterimages areenhanced by luminance edges more than physicalpatches. This finding implies that there may besomething different about afterimage representationsthat results in particularly powerful modulations bycontextual cues (in this case, luminance contours).

Control experiments

We conducted two further experiments to control forother differences between the afterimages and physicalstimuli that could have driven the different degrees ofcontour enhancement found in Experiments 1 through 3.

Control 1 (Experiment 4): Were the edges of thephysical stimuli blurry enough?

The larger contour effect for afterimages comparedwith real stimuli could be due to an underestimation ofafterimage edge blur. If the edges of the physical stimuliare less blurry than those of the afterimages, we mightexpect to find less modulation by the contour becausesharper chromatic edges may themselves contribute tothe contour/edge effect. The edge blur profile used todraw the physical patches was based on eye movementjitter from 1 observer, and analysis of the jitter fromother observers revealed some variance with thisstandard (Experiment 1: 3 lower, 5 higher thanstandard; Experiment 2: 3 higher, 1 lower thanstandard). Although no correlation was apparentbetween this jitter and the difference between theafterimage and physical stimulus contour effects, thesecorrelational analyses had very low power. Therefore,we tested the effect of edge blur in a further comparisonexperiment, identical to the physical stimulus condi-tions described for Experiment 1.

The edge blur of the reference stimulus wasmodulated from increased blur to sharp edged (meth-ods in supplementary information) and presented withand without a luminance contour. The results showedthat the contour effect did not increase as the edges ofthe patches became increasingly blurry (Figure 5).Specifically, there was no interaction between thecontour effect and degree of blur when the referencestimuli were physically equally saturated across edgeblur conditions (Figure 5a). This was also the casewhen the perceived saturation levels of the referencewere equated across levels of edge blur (Figure 5b).Perceived saturation was equated to counteract changesin saturation due solely to edge blur, as we wanted to besure that this did not interact with the contour effect.These findings indicate that even if the blurriness of our

Figure 5. Results of Experiment 4. Perceived saturation of the reference from the physical saturation, across edge blur and contour

conditions, when the patches were (a) physically equally saturated and (b) perceptually equally saturated. Physical saturation for the

reference was 20. Both illustrate that although the contour increases the perceived saturation of the reference, it does not do so in a way

that interacts with the blurriness of the reference. Error bars are calculated in the same manner as Experiment 1 (some are smaller than

the marker size). There are 20 data readings for each point.

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 7

physical stimuli was underestimated for some observersin Experiments 1 through 3, this is unlikely to explainwhy we found increased contour modulation of theafterimages relative to physical stimuli because increas-ing edge blur does not lead to an associated increase incontour modulation.

Control 2 (Experiment 5): Temporal order of contourand physical stimulus onsets

Another possible explanation for why afterimageswere enhanced to a greater extent by edges comparedwith the physical stimuli is that adaptation signals arepresent in the visual system prior to the presentation ofthe contour or the physical stimuli. Indeed, asluminance signals tend to reach the visual cortex 10to 30 ms before chromatic signals (Bompas & Sumner,2008; Maunsell & Gibson, 1992; Nowak, Munk,Girard, & Bullier, 1995; Schmolesky et al., 1998) andthe adapted color signal is present before this, theafterimage will benefit from any contour enhancementprior to the physical stimulus. In order to control forthis difference, 3 observers repeated the afterimage andphysical stimulus comparison task (Experiment 1).However, we varied the onset of the contour so thatit was presented 0, 20, 40, or 60 ms after adaptation orphysical stimulus onset. If the physical stimulus

suffered from arriving in the cortex after the contour,we might predict its contour effect to increase for acontour delay of 20 to 40 ms. Against that prediction,results revealed that the contour effect remains fairlystable between contour onsets of 0 to 60 ms for bothafterimages and physical stimuli, suggesting thatcontour onset does not notably modulate contourenhancement effects (Figure 6). Further, the contoureffect is greater for afterimages than physical stimuli,thus replicating the results of Experiment 1.

General discussion

We investigated whether negative color afterimagesare treated identically to real physical stimuli. Wetheorized that there are differences between theadaptation response that results in afterimage perceptsand signals from physical stimuli, and that thesedifferences may lead to a different degree of luminanceedge modulation of afterimages compared with realstimuli. Our theory was motivated by noting theabsence of physical stimulus versions of illusionsdemonstrating compelling contextual modulations ofafterimage visibility.

Afterimages are particularly enhanced bycontext

We report converging evidence from both a com-parison and nulling paradigm that luminance edges(contours or pedestals) enhance the visibility of colorafterimages to a larger extent than they do for physicalstimuli of similar appearance. In the comparisonexperiment (Experiment 1) the physical stimuli weresimilar to the afterimages in appearance (hue, lumi-nance, and saturation), yet were not enhanced by thecontour to the same degree. The nulling paradigm(Experiments 2 and 3) revealed that more physicalstimulus saturation was required to null the afterimageswhen framed by a luminance edge. This result excludesthe possibility that the edge enhanced both theafterimage and the physical nulling patch equally,which could have resulted in an overall increase indiscrimination when the contour was present, but notthe observed shift in the null point. This is because thecontour would enhance discrimination of the combinedsignal of the nulling patch and the afterimage. Ourfindings are not constrained to chromatic stimulibecause we replicated the results of Experiment 2 usingluminance afterimages.

A control study (Experiment 4) revealed that thedifference in contour enhancement between afterimagesin physical stimuli was not due to an underestimationof the edge blur used to draw the physical patches

Figure 6. Results of Experiment 5. Perceived saturation of

afterimages (black) and physical stimuli (white), with and without

a contour, across four contour onset times. The no contour and 0-

ms onset points are a replication of Experiment 1 and show that

both afterimages and physical stimuli were perceived as equally

saturated when not framed by a contour (as expected from the

experimental design), but afterimages were perceived as more

saturated when framed by a contour than were physical stimuli.

However, the pattern is similar across different contour latencies,

showing that contour onset time did not interact with the contour

effect or the type of stimulus (physical or afterimage). Error bars

are calculated in the same manner as Experiment 1. There are 20

data readings for each point.

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 8

(fixational jitter during adaptation will blur afterimageedges). We found that increasing or decreasing theblurriness of the physical patches did not lead to adecrease or increase in the size of the contour effect. Afurther control study (Experiment 5) revealed that ourresults do not arise from temporal differences in theonset of afterimage versus real signals. Contourenhancement was greater for the afterimages comparedwith physical stimuli, thus replicating the results ofExperiment 1. However, delaying presentation of thecontour (0–60 ms) after presentation of the physicalstimulus did not impact the degree of contourmodulation.

Afterimages as ambiguous stimuli

Our finding that luminance edges enhance afterim-ages to a greater extent than they enhance physicalstimuli suggests that afterimage representations areprocessed to some extent differently than signals arisingfrom real-world objects. One possible reason for thiseffect is that afterimages are ambiguous, like otherphenomena such as binocular rivalry and ambiguousfigures. For example, the tendency of afterimages tofade in and out of conscious awareness (Wade, 1978),especially when not supported by a consistent context,could be analogous to the perceptual oscillationspresent during binocular rivalry. Increased ambiguity,whatever its source, is likely to result in greatersusceptibility to contextual modulation. Moreover, ineveryday viewing, such context would be powerfulenough to suppress the perception of afterimages mostof the time. Support for these suppression effects isfound in the experiments of Daw (1962) and in theillusion by van Lier et al. (2009), where afterimagepercepts are inhibited when the context was inconsis-tent with the afterimage (see also our ‘‘Welsh Castle’’demo in supplemental information, where the afterim-age is suppressed when the context is presented upsidedown). This might explain why we do not oftenperceive afterimages in everyday life despite the easewith which they can be evoked in demonstrations andillusions.

Although we are not arguing that the underlyingcellular populations differ between afterimages andperceptions of real chromatic objects, there are anumber of plausible reasons (which we will discussbelow) for why the pattern of activity in these cells maybe importantly different between afterimages and realstimuli. If the perceptual system is attuned to thesedifferences, the afterimage signals will present uncer-tainty—in some ways the signal will be like that of areal stimulus, but in some ways it will not be. In thissituation, perception will be particularly influenced byany disambiguating cues, such as luminance edges, that

increase the likelihood of the signal representing a realobject.

At a general level, our explanation relies on theassumption that it is beneficial for perception todissociate afterimage signals from signals arising fromreal-world objects. Previous studies have shown that acolored surface is perceptually enhanced if it isinterpreted as material (i.e., an object) rather than anillumination (Bloj et al., 1999; Kingdom, 2008).Afterimages could be considered similar to signalarising from illumination given that neither suppliesdirect information about objects. The visual systemmay have prior knowledge (implicitly, in the patternand weights of its connections) of the typical activationprofile associated with responses to real-world objects.If signals arising from adapted cells deviate in any wayfrom the typical profile associated with real objects, thiscould lead to the interpretation that the afterimagesignal is illusory. Thus, perception of the afterimage issuppressed.

More specifically, we assume that luminance edgesand other contextual cues support the interpretationthat an ambiguous signal represents a real object. Thisinterpretation may have been learned based on real-world statistics that most objects are delineated byluminance and chromatic contrast (Fine et al., 2003;Hansen & Gegenfurtner, 2009; Zhou & Mel, 2008),whereas this is not as likely for features less significantto awareness (e.g., illuminations, reflections).

What is the source of afterimage ambiguity?

Although we have not explored directly ways inwhich the pattern of activation that produces anafterimage may be different from that arising from areal stimulus, it is possible to consider some likelycandidates. One important factor that could disambig-uate afterimages from real stimuli in normal viewingconditions is eye movements. Our eyes make frequentmovements, and because afterimages are retinotopicthey will move in exact synchrony with the eyes(Helmholtz, 1962). This characteristic is rarely (if atall) observed for real-world stimuli, even those followedwith ‘‘smooth pursuit’’ eye movements (Kolarik,Margrain, & Freeman, 2010). Indeed, there is evidencethat continual eye movements suppress the perceptionof afterimages (Kennard, Hartmann, Kraft, & Boshes,1970; Matin, 1974). For these reasons, all ourexperiments were conducted with the eyes fixed tominimize the perceptual difference between afterimagesand real stimuli created by eye movements.

It is possible that even though we restricted largereye movements in our experiments, small fixationaljitter may have been sufficient to reveal that theafterimage percept was illusory. This knowledge could

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 9

arise from compensatory mechanisms that stabilize theretinal image based on whole-world motion duringjitter (Murakami & Cavanagh, 1998). These compen-satory mechanisms would lead to a discrepancybetween the afterimage (which is stabilized on theretina but has postcompensation movement) and thebackground (which moves on the retina but isstabilized postcompensation).

The contour could have enhanced the perception ofthe afterimage because it provided a visual transientthat is similar to the edge of a real object, and thistransient would dominate the weaker signal from theafterimage itself. If jitter become too large, however, itis likely that it would dissociate the relative movementof afterimage and contour, and thus the enhancementeffect would be lost.

Apart from eye movements, we also controlled formany other possible perceptual differences between theafterimages and the physical stimuli (e.g., they werematched for hue, luminance, edge blur, and saturation).Controlling for these factors makes it possible toconsider the existence of other critical factors differen-tiating the representation of afterimages from that ofreal stimuli of similar appearance.

First, the temporal profile of an afterimage response islikely to be different from that representing a realstimulus. For example, the change from adaptor toafterimage (i.e., the off response) is unlikely to be exactlythe same as the onset of a new stimulus. This differencecould potentially underlie the experience of a small delaybefore the afterimage percept appears after the adaptingstimulus is turned off (Creed, 1928). Similarly, theexponential recovery from adaptation (McLelland,Baker, Ahmed, & Bair, 2010) is unlikely to exactlymimic the activity profile displayed when viewing a realstimulus. Second, if adaptation is present at multiplecolor-sensitive sites throughout the visual system, cells inthese sites may have different cellular architectures anddifferent susceptibilities to adaptation and their relativerecovery from adaptation could occur at different speeds(Fairchild & Reniff, 1995; Jameson, Hurvich, & Varner,1979; Loomis, 1972; McLelland, Ahmed, & Bair, 2009;Rinner & Gegenfurtner, 2000; Yeh, Lee, & Kremers,1996). This means that the relative firing rates afteradaptation will differ between areas in a way that isinconsistent with how real stimulus signals elicit activitypatterns through the visual system.

How context could modulate chromaticsignals

Perceptual demonstrations of filling in and psycho-physical research both demonstrate that luminanceedges are important for constraining and facilitatingchromatic signals (Chaparro et al., 1994; Cole et al.,

1990; Eskew et al., 1994; Gur & Akri, 1992; R. Hilz &Cavonius, 1970; R. L. Hilz et al., 1974; Montag, 1997;Mullen & Losada, 1994). In our experiment, a sharpedge was critical for enhancement above the merepresence of a luminance difference, as we observed thata blurry-edged pedestal did not enhance the afterimageas much as a sharp contour or pedestal. A luminancesine-wave grating does lower chromatic threshold,though not to the extent of square-wave gratings(Gowdy et al., 1999). That the edge closely frames thechromatic signal is also important, as orthogonal edgesdo not produce the facilitation observed with contig-uous edges (Gowdy et al., 1999). Higher level edgerepresentations, such as stereoscopic-depth edges andillusory contours, also modulate chromatic representa-tions (Montag, 1997). This suggests that as long as theedge makes sense in terms of a higher level context itproduces facilitation. It also seems important forfacilitation that the edges are suprathreshold (Chapar-ro et al., 1994).

The fact that chromatic facilitation is reliant onsharp edges, even if the properties of these edges aresomewhat abstract, suggests that it is mainly a corticalprocess rather than just a low-level luminance interac-tion. This is in line with physiological research thatorientation and form become increasingly important athigher levels of the visual system (Friedman et al.,2003) and that responses of blue–yellow color-oppo-nent neurons in the macaque V1 are facilitated byluminance contrast (Horwitz et al., 2005).

Loci of adaptation

The sites of adaptation that may contribute togenerating color afterimages are worth discussing hereas this has been the subject of much debate. Pastresearch has variously championed the existence ofafterimage-generating adaptation in the photoreceptors(Brindley, 1962; Craik, 1940), retinal ganglion cells(Virsu & Laurinen, 1977; Zaidi et al., 2012), and thecortex (Shevell, St Clair, & Hong, 2008; Shimojo,Kamitani, & Nishida, 2001).

We can identify three possible sites of adaptation,although our results are consistent with any or all of theseand do not exclusively point to a cortical locus. First,because we found that edges enhance afterimages morethan they do physical stimuli, we could assume someadaptation at the level(s) at which edges enhancechromatic signals. As discussed earlier, this is likely tobe cortical. Second, it is possible that although the edgefirst facilitates chromatic signals in the cortex, feedbacksignals are subsequently relayed back to the lateralgeniculate nucleus (LGN) to amplify and tune itsresponse (Ferster & Miller, 2000). This means thatcortical influences that are responsive to the luminance

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 10

edge, such as fine orientation tuning and attention, couldevoke relative changes in LGN activity. Thus, adaptationresponses in the LGN could be enhanced by edge-drivenamplification by the cortex. Finally, higher level visualareas may be sensitive to temporal differences betweenretinal ganglion cell rebound signals resulting fromadaptation and those arising from real stimuli. Thus,these ganglion signals may be more susceptible to edgeenhancement because they are deemed more ambiguousor unusual by higher levels. As noted above, our resultsdo not distinguish between these suggested adaptationsites, only that at some point the visual system is able todissociate signals arising from adaptation from thoserepresenting real objects.

Summary

We have presented converging evidence that lumi-nance edges enhance afterimages more than they dophysical stimuli of similar appearance. This findingappears to be specific to sharp edges, as a gradedluminance pedestal did not produce the afterimageenhancement found with a sharp pedestal or contour.These results demonstrate that the brain processessignals arising from adapted cells nonequivalently tothose arising from real stimuli. We suggest that becausesignals that generate afterimage percepts fail to perfectlymatch those arising from real stimuli, the visual system isunsure of whether the afterimage represents the presenceof a real object. This would explain why afterimages areinfluenced by contextual cues that reduce their uncer-tainty more than responses triggered by real objects.Conversely, in everyday viewing contextual cues will beunlikely to align with afterimages, and thus ourperception of them is often suppressed.

Acknowledgments

Thanks to Tom Freeman and Chris Miles for usefulcomments on the manuscript. This work was funded bythe BBSRC and Cardiff School of Psychology.

Commercial relationships: none.Corresponding author: Georgie PowellEmail: powellg7@cf.ac.uk.Address: School of Psychology, Cardiff University,Cardiff, UK.

References

Bloj, M. G., Kersten, D., & Hurlbert, A. C. (1999).

Perception of three-dimensional shape influencescolour perception through mutual illumination.Nature, 402(6764), 877–879, doi:10.1038/47245.

Bompas, A., & Sumner, P. (2008). Sensory sluggishnessdissociates saccadic, manual, and perceptual re-sponses: An S-cone study. Journal of Vision, 8(8):10, 1–13, http://www.journalofvision.org/content/8/8/10, doi:10.1167/8.8.10. [PubMed] [Article]

Brindley, G. S. (1962). Two new properties of fovealafter-images and a photochemical hypothesis toexplain them. The Journal of Physiology, 164(1),168–179.

Chaparro, A., Stromeyer, C. F., III, Kronauer, R. E.,& Eskew, R. T., Jr. (1994). Separable red-green andluminance detectors for small flashes. VisionResearch, 34(6), 751–762, doi:10.1016/0042-6989(94)90214-3.

Cole, G. R., Stromeyer, C. F., III, & Kronauer, R. E.(1990). Visual interactions with luminance andchromatic stimuli. Journal of the Optical Societyof America A, 7(1), 128–140.

Comby, J. H. (1909). The intermittence of minimalvisual sensations. I. The fluctuation of the negativeafter-image. Psychological Bulletin, 6(9), 305–307,doi:10.1037/h0063993.

Craik, K. J. W. (1940). Origin of visual after-images.Nature, 145(3674), 1.

Creed, R. S. (1928). On the latency of negative after-images following stimulation of different areas ofthe retina. Journal of Physiology, 66(3), 281–298.

Daw, N. W. (1962). Why after-images are not seen innormal circumstances. Nature, 196(4860), 1143–1145.

Eskew, R. T., Jr, Stromeyer, C. F., III, & Kronauer, R.E. (1994). The time-course of chromatic facilitationby luminance contours. Vision Research, 34(23),3139–3144, doi:10.1016/0042-6989(94)90079-5.

Fairchild, M. D., & Reniff, L. (1995). Time course ofchromatic adaptation for color-appearance judg-ments. Journal of the Optical Society of America A,12(5), 824–833.

Ferster, D., & Miller, K. D. (2000). Neural mechanismsof orientation selectivity in the visual cortex.Annual Review of Neuroscience, 23(1), 441–471,doi:10.1146/annurev.neuro.23.1.441.

Fine, I., MacLeod, D. I. A., & Boynton, G. M. (2003).Surface segmentation based on the luminance andcolor statistics of natural scenes. Journal of theOptical Society of America A, 20(7), 1283–1291.

Friedman, H. S., Zhou, H., & von der Heydt, R.(2003). The coding of uniform colour figures inmonkey visual cortex. The Journal of Physiology,

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 11

548(2), 593–613, doi:10.1111/j.1469-7793.2003.00593.x.

Gowdy, P. D., Stromeyer, C. F., III, & Kronauer, R. E.(1999). Facilitation between the luminance and red-green detection mechanisms: Enhancing contrastdifferences across edges. Vision Research, 39(24),4098–4112, doi:10.1016/s0042-6989(99)00115-7.

Gur, M., & Akri, V. (1992). Isoluminant stimuli maynot expose the full contribution of color to visualfunctioning: Spatial contrast sensitivity measure-ments indicate interaction between color andluminance processing. Vision Research, 32(7),1253–1262, doi:10.1016/0042-6989(92)90220-d.

Hansen, T., & Gegenfurtner, K. R. (2009). Indepen-dence of color and luminance edges in naturalscenes. Visual Neuroscience, 26(01), 35–49, doi:10.1017/S0952523808080796.

Helmholtz, H. (1962). Helmholtz’s treatise on physio-logical optics (Vol. II). New York: Dover.

Hilz, R., & Cavonius, C. R. (1970). Wavelengthdiscrimination measured with square-wave grat-ings. Journal of the Optical Society of America,60(2), 273–277.

Hilz, R. L., Huppmann, G., & Cavonius, C. R. (1974).Influence of luminance contrast on hue discrimina-tion. Journal of the Optical Society of America,64(6), 763–766.

Horwitz, G. D., Chichilnisky, E. J., & Albright, T. D.(2005). Blue-yellow signals are enhanced by spatio-temporal luminance contrast in macaque V1.Journal of Neurophysiology, 93(4), 2263–2278, doi:10.1152/jn.00743.2004.

Jameson, D., Hurvich, L. M., & Varner, F. D. (1979).Receptoral and postreceptoral visual processes inrecovery from chromatic adaptation. Proceedingsof the National Academy of Sciences, 76(6), 3034–3038.

Kennard, D. W., Hartmann, R. W., Kraft, D. P., &Boshes, B. (1970). Perceptual suppression ofafterimages. Vision Research, 10(7), 575–585, doi:10.1016/0042-6989(70)90051-9.

Kingdom, F. A. A. (2008). Perceiving light versusmaterial. Vision Research, 48(20), 2090–2105, doi:10.1016/j.visres.2008.03.020.

Kolarik, A., Margrain, T., & Freeman, T. (2010).Precision and accuracy of ocular following: Influ-ence of age and type of eye movement. Experimen-tal Brain Research, 201(2), 271–282, doi:10.1007/s00221-009-2036-6.

Loomis, J. M. (1972). The photopigment bleachinghypothesis of complementary after-images: A

psychophysical test. Vision Research, 12(10),1587–1594, doi:10.1016/0042-6989(72)90031-4.

MacLeod, D. I., & Boynton, R. M. (1979). Chroma-ticity diagram showing cone excitation by stimuli ofequal luminance. Journal of the Optical Society ofAmerica, 69(8), 1183–1186.

Matin, E. (1974). Saccadic suppression: A review andan analysis. Psychological Bulletin, 81(12), 899–917,doi:10.1037/h0037368.

Maunsell, J. H., & Gibson, J. R. (1992). Visualresponse latencies in striate cortex of the macaquemonkey. Journal of Neurophysiology, 68(4), 1332–1344.

McLelland, D., Ahmed, B., & Bair, W. (2009).Responses to static visual images in macaquelateral geniculate nucleus: Implications for adapta-tion, negative afterimages, and visual fading.Journal of Neuroscience, 29(28), 8996–9001, doi:10.1523/jneurosci.0467-09.2009.

McLelland, D., Baker, P. M., Ahmed, B., & Bair, W.(2010). Neuronal responses during and after thepresentation of static visual stimuli in macaqueprimary visual cortex. Journal of Neuroscience,30(38), 12619–12631, doi:10.1523/jneurosci.0815-10.2010.

Mollon, J. D. (1995). Seeing color. Cambridge: Cam-bridge University Press.

Montag, E. D. (1997). Influence of boundary informa-tion on the perception of color. Journal of theOptical Society of America A, 14(5), 997–1006.

Mullen, K. T., & Losada, M. A. (1994). Evidence forseparate pathways for color and luminance detec-tion mechanisms. Journal of the Optical Society ofAmerica A, 11(12), 3136–3151.

Murakami, I., & Cavanagh, P. (1998). A jitter after-effect reveals motion-based stabilization of vision.Nature, 395, 798–801.

Nowak, L. G., Munk, M. H. J., Girard, P., & Bullier, J.(1995). Visual latencies in areas V1 and V2 of themacaque monkey. Visual Neuroscience, 12(02),371–384, doi:10.1017/S095252380000804X.

Pinna, B., Brelstaff, G., & Spillmann, L. (2001). Surfacecolor from boundaries: A new ‘‘watercolor’’ illu-sion. Vision Research, 41(20), 2669–2676, doi:10.1016/s0042-6989(01)00105-5.

Rinner, O., & Gegenfurtner, K. R. (2000). Time courseof chromatic adaptation for color appearance anddiscrimination. Vision Research, 40(14), 1813–1826,doi:10.1016/s0042-6989(00)00050-x.

Sadowski, J. Big Spanish Castle. Internet site: [http://www.johnsadowski.com/big_spanish_castle.php](Accessed August 5, 2011).

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 12

Schmolesky, M. T., Wang, Y., Hanes, D. P., Thomp-son, K. G., Leutgeb, S., Schall, J. D., et al. (1998).Signal timing across the macaque visual system.Journal of Neurophysiology, 79(6), 3272–3278.

Shevell, S. K., St Clair, R., & Hong, S. W. (2008).Misbinding of color to form in afterimages. VisualNeuroscience, 25(03), 355–360, doi:10.1017/S0952523808080085.

Shimojo, S., Kamitani, Y., & Nishida, S. (2001).Afterimage of perceptually filled-in surface. Sci-ence, 293(5535), 1677–1680, doi:10.1126/science.1060161 293/5535/1677 [pii].

Smith, T., & Guild, J. (1931). The C.I.E. colorimetricstandards and their use. Transactions of the OpticalSociety, 33(3), 73.

van Lier, R., Vergeer, M., & Anstis, S. (2009). Filling-inafterimage colors between the lines. Current Biol-ogy, 19(8), R323–R324, doi:10.1016/j.cub.2009.03.010.

Virsu, V., & Laurinen, P. (1977). Long-lasting afterim-ages caused by neural adaptation. Vision Research,17(7), 853–860, doi:10.1016/0042-6989(77)90129-8.

von der Heydt, R., Friedman, H. S., & Zhou, H.(2003). Searching for the neural mechanisms of

color filling-in. In L. Pessoa & P. De Weerd (Eds.),Filling-in: From perceptual completion to corticalreorganization (pp. 106–127). Oxford: OxfordUniversity Press.

von Kries, J. (1970). Chromatic adaptation, Sources ofColor Vision (ed. MacAdam D. L.), Cambridge,Mass: MIT Press. Originally published in: Fest-schrift der Albrecht-Ludwigs-Universitat, 1902,145–148.

Wade, N. J. (1978). Why do patterned afterimagesfluctuate in visibility? Psychological Bulletin, 85(2),338–352, doi:10.1037/0033-2909.85.2.338.

Yeh, T., Lee, B. B., & Kremers, J. (1996). The timecourse of adaptation in macaque retinal ganglioncells. Vision Research, 36(7), 913–931, doi:10.1016/0042-6989(95)00332-0.

Zaidi, Q., Ennis, R., Cao, D., & Lee, B. (2012). Neurallocus of color afterimages. Current Biology, 22(3),220–224, doi:10.1016/j.cub.2011.12.021.

Zhou, C., & Mel, B. W. (2008). Cue combination andcolor edge detection in natural scenes. Journal ofVision, 8(4):4, 1–25, http://www.journalofvision.org/content/8/4/4, doi:10.1167/8.4.4. [PubMed] [Article]

Journal of Vision (2012) 12(10):17, 1–13 Powell, Bompas, & Sumner 13