No role for lightness in the encoding of Black and White: Race-contingent face aftereffects depend on facial morphology, not facial luminance

Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=pvis20

Download by: [58.172.34.59] Date: 20 September 2015, At: 16:34

Visual Cognition

ISSN: 1350-6285 (Print) 1464-0716 (Online) Journal homepage: http://www.tandfonline.com/loi/pvis20

No role for lightness in the encoding of Black andWhite: Race-contingent face aftereffects dependon facial morphology, not facial luminance

O. Scott Gwinn & Kevin R. Brooks

To cite this article: O. Scott Gwinn & Kevin R. Brooks (2015) No role for lightness in theencoding of Black and White: Race-contingent face aftereffects depend on facial morphology,not facial luminance, Visual Cognition, 23:5, 597-611, DOI: 10.1080/13506285.2015.1061085

To link to this article: http://dx.doi.org/10.1080/13506285.2015.1061085

Published online: 04 Aug 2015.

Submit your article to this journal

Article views: 27

View related articles

View Crossmark data

No role for lightness in the encoding of Black andWhite: Race-contingent face aftereffects depend on

facial morphology, not facial luminance

O. Scott Gwinn1,2 and Kevin R. Brooks1,2

1Department of Psychology, Faculty of Human Sciences, MacquarieUniversity, Sydney, NSW, Australia2Perception in Action Research Centre, Department of Cognitive Science,Faculty of Human Sciences, Macquarie University, Sydney, NSW, Australia

(Received 9 March 2015; accepted 6 June 2015)

In the current study we use contingent face aftereffects to examine the contributions madeby morphology and facial luminance cues to the encoding of race. African and Europeanfacial images differed in both morphology and luminance (ML), in morphology alone(M), or in luminance alone (L). Significant aftereffects were found in conditions wheretest stimuli included morphological information (ML and M), but not when it wasabsent (L). Furthermore, the size of the aftereffect for test conditions specifying raceusing both cues (ML) was no greater than when morphological information was thesole cue (M). This suggests that the effects measured here can be accounted for bydifferences in morphology alone. These results indicate that not only are individualreports of perceived racial typicality more significantly influenced by morphologicalshape cues than surface cues, but morphological cues appear to form the basis of theunderlying neural encoding for faces of different races.

Keywords: Face perception; Adaptation; Aftereffects; Race; Visual cues; Encoding.

Models of object perception have traditionally focused on the importance of shapeinformation defined by high contrast edges, with surface reflectance properties gen-erally being considered secondary information (e.g., Biederman 1987; Biederman& Ju, 1988; Marr & Nishihara, 1978). Biederman and Ju (1988) found that

© 2015 Taylor & Francis

Please address all correspondence to O. Scott Gwinn, Department of Psychology, MacquarieUniversity, Sydney, NSW 2109, Australia. E-mail: [email protected]

Thanks to Verena Willenbockel for providing stimulus materials.O. Scott Gwinn was supported by a Macquarie University Research Excellence Scholarship.No potential conflict of interest was reported by the authors.

Visual Cognition, 2015Vol. 23, No. 5, 597–611, http://dx.doi.org/10.1080/13506285.2015.1061085

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

performance for naming basic object categories was unaffected by the exclusion ofreflectance information, with accuracy in conditions comprising simple line draw-ings equal to performance with full colour photographs. Within the area of faceperception, a similar distinction has been drawn between the types of informationcontained within faces, namely, morphology (feature, shape, structure) and skinsurface properties (colour, reflectance, pigmentation, texture, tone) (Russell &Sinha, 2007).

A substantial amount of face recognition research has been devoted to investi-gating the relative importance of shape and reflectance cues (Bruce & Langton,1994; O’Toole, Vetter, & Blanz, 1999; Russell, Sinha, Biederman, & Nederhouser,2006; Russell, Biederman, Nederhouser & Sinha, 2007). In contrast to models ofobject recognition, many authors have concluded that reflectance information isfundamental when identifying an individual face (Bruce & Langton, 1994;O’Toole et al., 1999; Russell et al., 2006). Using highly familiar faces known person-ally to the observers, Russell and Sinha (2007) found that participants were better atrecognizing faces from reflectance information than from shape information.Similarly, Russell et al. (2007) found higher recognition accuracy for imagesvarying in reflectance information when colour images were used, although for grey-scale images the oppositewas true,withvariations in shape informationprovingmoreuseful.

The role of shape and reflectance cues has also received considerable attentionwithin the context of sex classification (Hill, Bruce & Akamatsu, 1995; Nestor &Tarr, 2008; Russell, 2003; Russell, 2009). In line with results from the aforemen-tioned recognition studies, Hill et al. (1995) found sex to be judged more accuratelywhen defined by skin cues rather than shape. Further, Russell (2009) found thatfacial images with androgynous morphologies could be made to appear femaleby increasing the contrast of the face, or to appear male by decreasing the contrast.There is also evidence that the ability to recognize faces from shape cues interactswith the sex of the face, with male faces defined by shape alone recognized betterthan female faces (Bruce et al., 1991; Russell & Sinha, 2007).

The racial labels “Black” and “White”, principally used to refer to those ofAfrican and of European descent, respectively, contain explicit reference to thereflectance properties or “albedo” of the skin, suggesting a central role for skin-based cues to the perception of race. In accordance with this suggestion, peoplegenerally report that they rely on such skin cues when making racial judgments(Brown, Dane, & Durham, 1998). However, more explicit tests of racial categor-ization suggest that, counterintuitively, skin tone may actually play a relativelyminor role. While Hill et al. (1995) showed classifications of sex to be primarilybased on skin cues, racial judgments of Asian and European faces were found tobe based more substantially on shape cues. Similarly, Brooks and Gwinn (2010)found that although simultaneous contrast effects could alter the perceived light-ness of a face, this did not affect judgements of the racial typicality of Black andWhite faces. It was concluded that this may reflect a relatively small contribution

598 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

of skin tone to perceptions of race, which is instead dominated by morphologicalcues. In a later study, Willenbockel, Fiset, and Tanaka (2011) created a set ofmorphed facial stimuli in which shape and luminance cues were independentlyvaried to be more typically African or European. Results from this study demon-strated that categorical race judgments are primarily influenced by shape cues,with luminance cues making only a secondary contribution to perceptions ofrace. Similar results were presented by Stepanova, Strube, and Yablonsky(2013), where Russian observers were asked to rate synthetic faces on howRussian or non-Russian they appeared. Faces with an African morphology wereconsistently rated as non-Russian regardless of the lightness of their skin tone,indicating the substantial influence of morphology as a cue. Faces with a Europeanmorphology and lighter skin tone were rated as more Russian than faces with thesame morphology but darker skin tone, indicating some influence of skin tone.However, faces with European morphologies were consistently rated as moreRussian than African morphologies overall, suggesting a dominance of morpho-logical cues.

While these studies are certainly informative, each involves explicit categoriz-ation or ratings of typicality, raising the possibility that the results could be influ-enced by demand characteristics, social or political issues, or other non-perceptualfactors. The aim of the current study is to provide a deeper examination of the roleof shape and skin cues in face processing across race using a novel approach: race-contingent face distortion aftereffects. This technique is able to reveal the implicitcategorical perceptual encoding of faces while minimizing other non-perceptualfactors.

Perceptual aftereffects are characterized by changes in the appearance of a “test”stimulus following an extended period of exposure to an “adaptation” stimulus. Forexample, adapting to downward motion results in a stationary scene appearing todrift upwards: the well-known Waterfall Illusion (Addams, 1834). This is thoughtto result from the retuning of cells sensitive to motion (Barlow & Hill, 1963).Similarly, prolonged viewing of coloured stimuli can lead to a biased colourpercept in a neutral (e.g., white) test stimulus (Krauskopf, Williams, & Heeley,1982). McCollough (1965) found that opposing colour aftereffects could be simul-taneously induced using bars of different orientations. Following adaptation to avertical orange grating alternating with a horizontal blue grating, participants experi-enced different colour percepts when observing either vertical or horizontal testpatterns. This indicates the presence of dissociable populations of cells sensitiveto orientation, each showing a different colour aftereffect. Since this seminalpaper, studies have shown not only that aftereffects can be seen after adapting toimages of distorted faces (O’leary & McMahon, 1991; Webster & MacLin,1999), but that opposing aftereffects can be simultaneously induced contingentupon the sex (Jaquet & Rhodes, 2008; Little, Debruine, & Jones, 2005), age(Little, DeBruine, Jones, &Waitt, 2008) and, of more relevance to the current inves-tigation, race (Gwinn & Brooks, 2013; Jaquet, Rhodes, & Hayward, 2007; Jaquet,

ENCODING VISUAL CUES TO RACE 599

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

Rhodes, & Hayward, 2008) of the facial stimuli. Jaquet et al. (2008) adapted partici-pants to contracted Asian and expanded European faces (and vice versa in anothercondition). Subsequently, slightly contracted Asian test faces were judged as morenormal whereas preferred European faces were slightly expanded. As with otherresults of visual adaptation, face aftereffects are understood to reflect the retuningof cells, with contingent aftereffects indicating opposing adaptation of dissociablepopulations (Clifford & Rhodes, 2005; Jaquet & Rhodes, 2008; Webster &MacLeod, 2011). These effects show substantial invariance to changes in retinalposition and are thought to originate in the higher levels of the visual system andareas of the brain (Leopold, O’Toole, Vetter, & Blanz, 2001; Rhodes, Jeffery,Watson, & Clifford, 2003; Watson & Clifford, 2003; Webster & MacLin, 1999).As such it is generally accepted that these effects reveal the mechanisms involvedin face perception specifically, rather than more general properties of vision(Susilo, McKone, & Edwards, 2010; Webster & MacLeod, 2011).

In the current study we used race-contingent aftereffects to infer the degree towhich the neural populations encoding faces of different races are sensitive tomorphological and facial luminance cues. Using stimuli from Willenbockelet al. (2011), opposing aftereffects of expansion and contraction were inducedusing European and African facial images. The extent to which these aftereffectstransfer to faces whose races are defined by morphology and luminance (ML),morphology only (M) or luminance only (L) was then assessed. The aftereffectmagnitude was then used to infer which cues the relevant cells are sensitive to.Based on previous studies’ conclusion that perceptions of race are primarily influ-enced by morphological shape cues, we expected to find larger contingent after-effects for faces whose race is specified by morphology.

METHOD

Participants

One of the authors and seven Caucasian undergraduate students enrolled in Psy-chology at Macquarie University took part in the study for course credit. Thesample had a mean age of 19.5 (SD = 1.52) and comprised four males. Under-graduate participants were naïve to the purposes and hypotheses of theexperiment.

Stimuli

The facial images used in this study were identical to those used by Willenbockelet al. (2011). This is a set of facial images in which morphology and luminancevary independently to be either more typical of European or African faces. Adetailed description can be found in the original study. For the purposes of thecurrent study, six image pairs, each comprising one European and one African

600 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

face, were selected. These represent images defined by morphology and lumi-nance (ML). These images were not manipulated in terms of their pixel valuesand so do not exceed the range of normal photographic variation. For each ofthe six pairs, four more images were selected that contained: (1) the Europeanimage’s morphology with luminance averaged across both faces, (2) the Africanimage’s morphology with luminance averaged across both faces, (3) an averagemorphology of both faces with European luminance and (4) an average mor-phology of both faces with African luminance. Images described in (1) and (2)are defined by morphology only (M), while those described in (3) and (4) aredefined by luminance only (L) (see Figure 1). This resulted in a total of 18image pairs (six sets for each of the ML, M and L conditions). Naïve participantswere not familiar with the images prior to testing.

Adapting stimuliUsing the “spherize” function in Adobe Photoshop CS5, two distorted versionswere created for ML images in which the internal features were expanded or con-tracted by 50% around the midpoint of the face (see Figure 2). This is a commonmanipulation used when studying face aftereffects (e.g., Gwinn & Brooks, 2013;Jaquet et al., 2008). When viewed from a distance of 65 cm, images subtended avisual angle of 4.24 degrees.

Test stimuliThirteen versions of each of the images in the ML, M and L conditions werecreated to serve as test images. These involved distortions ranging from -30%(contraction) to +30% (expansion) in 5% steps. Test images were 75% the sizeof the adapting images and subtended a visual angle of 3.18 degrees.

Procedure

The size of race-contingent aftereffects was measured using a double interleavedstaircase to establish points of subjective normality (PSN). This estimates thepoint at which an aftereffect is nulled, following the same procedure as Gwinn

Figure 1. Examples of images where race is defined by morphology and luminance (left), by morphologyonly (middle) and by luminance only (right).

ENCODING VISUAL CUES TO RACE 601

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

and Brooks (2013). The experiment was run in three blocks, each comprising aseparate test condition (ML, M or L). Each block was separated by a 5 minbreak and the order of block presentation was randomized across participants.The whole experiment took approximately one hour to complete.

Adaptation phaseEach block began with a 2 min adaptation sequence in which participants pas-sively viewed the facial images. Half of the participants viewed three differentEuropean faces expanded by 50% and three different African images contractedby 50% (E+/A-). The other half of the participants saw expanded African and con-tracted European faces (E-/A+). Each participant viewed only one adaptation con-dition. Images were presented individually and sequentially for 2 s at a time on acontinual loop. Participants also saw 6 s of “top-up” adaptation in between eachtest trial. This consisted of three European and three African faces each shown for1 s. The presentation order was pseudo-randomized ensuring that the same facewas never presented twice in a row. Faces used during adaptation were notused for testing to ensure any observed aftereffects are category-contingent andnot identity-contingent.

Test phaseFollowing the initial adaptation phase, participants viewed test images. In eachML, M and L condition, participants viewed one face with a typically Europeanmorphology and/or luminance and one with a typically African morphologyand/or luminance. For each participant, the faces used in the ML, M and L con-ditions were from the same image set. Across participants, the same test imageswere seen following both E+/A- adaptation and E-/A+ adaptation.

For each test image, participants were required to rate whether the faceappeared expanded or contracted. Test images remained on the screen for aminimum of 2 s before ratings could be provided. After this time they remainedon the screen until a rating was recorded. If a participant rated an image as

Figure 2. Examples of images used during the adaptation phase: expanded European and contractedAfrican (left), and contracted European and expanded African (right). In all cases, race is specified byboth luminance and morphology.

602 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

appearing expanded, the next time that face was seen a more contracted versionwas presented, and vice versa. This continued until a reversal was obtained (par-ticipants rating the face as appearing contracted) at which point a more expandedversion was presented. Beginning with a 50% adjustment in expansion/contrac-tion (e.g., going from -30% contraction to +20% expansion), this step size wasreduced by 15% after each reversal so the minimum 5% step size was reachedafter the third reversal. Each staircase progressed until it reached a maximum ofeight reversals, at which point the average distortion level of a participant’s lastfour reversals was calculated. This calculation represents an estimate of thePSN indicating the level of expansion/contraction an observer perceives asbeing undistorted. All participants reached eight reversals within the maximum28 trials. Two interleaves were run for each face in each condition, with one start-ing at fully expanded and the other fully contracted. The order of interleave pres-entation was randomized, and PSN estimates from the two interleaves wereaveraged to reach a single PSN value for each image (six in total) and participant.Data collection continued until two image pairs in each of the ML, M or L testconditions had been rated by two observers following E+/A- adaptation andanother two observers following E-/A+ adaptation.

RESULTS

Mean PSNs for images in the ML, M and L conditions following adaptation canbe seen in Figure 3. Higher values represent a perception of normality forexpanded stimuli while lower values indicate that contracted faces appear undis-torted. In the ML plot, it can be seen that PSNs for the same test images differbetween adaptation conditions. After adapting to expanded European and con-tracted African images (E+/A-), participants recorded high PSNs for Europeantest images and low values for African images. The reversal of this pattern ofresults (lower PSNs for Europeans, higher PSNs for Africans) in the opposingadaptation condition (E-/A+) indicates the presence of a race-contingent afteref-fect. A difference in PSNs between adaptation conditions for the same testsimages can also be seen in the M condition. The same does not appear to betrue for L images whose PSNs, although negative, are similar for both adaptationconditions.

Data for each of the ML, M and L conditions were formally analysed using 2 × 2ANOVAs with the within subjects variable of Face Race (European vs. African) andbetween subjects variable Adaptation Condition (E+/A- vs. E-/A+). Main effects ofFace Race and Adaptation Condition were found to be not significant (p > .05) ineach of the ML, M and L conditions and have no bearing on the research question.Results from the ANOVAs revealed significant Face Race*Adaptation Conditioninteractions in both the ML (F(1,6) = 9.1, p = .023, h p2 = .603) and M conditions(F(1,6) = 8.3, p = .027, h p2 = .583), confirming the presence of race-contingent

ENCODING VISUAL CUES TO RACE 603

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

Figure 3. Mean levels of expansion (positive values) and contraction (negative values) observers perceiveas being undistorted following adaptation to expanded and contracted images. ML shows data for test imagescontaining both morphological and luminance cues, M shows data for images containing morphological cuesonly, and L shows data for images containing luminance cues only. Error bars show standard error of themean.

604 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

aftereffects. No interaction was found in the L condition (F(1,6) = .01, p = .916, h p2

= .002).To further examine the size of the contingent aftereffects found in the ML and

M conditions, the PSN for the image congruent with the race of the contractedadaptor was subtracted from the PSN for the image congruent with the expandedadaptor (Size = PSNexpanded - PSNcontracted) for each participant. As no sig-nificant aftereffect was observed in the L condition, these data were not analysedfurther. In Figure 4 it can be seen that the mean size of the aftereffect is similar inboth ML and M conditions. A paired samples t-test confirmed that the size of theaftereffects in each condition was not significantly different (t(7) = .54, p = .609,d = .27, two-tailed).

DISCUSSION

The aim of the current study was to examine the contributions made by morpho-logical shape cues and facial luminance cues to the encoding of race. Race-con-tingent aftereffects of expansion/contraction were simultaneously induced usingAfrican and European images with morphologies and facial luminance valuestypical of each race. The degree to which these aftereffects transferred toimages containing both morphological and luminance cues (ML), morphologycues only (M) and luminance cues only (L) was assessed. Significant contingentaftereffects were found in both ML and M conditions, with more expanded imagesappearing undistorted after adapting to expanded adaptors and more contractedimages appearing undistorted after adapting to contacted adaptors. The magni-tudes of the aftereffects measured in each of the ML and M conditions were

Figure 4. Mean size of the contingent aftereffect for images in the ML and M conditions. Error bars showstandard error of the mean.

ENCODING VISUAL CUES TO RACE 605

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

also found to not be significantly different from each other. No effect was foundfor images in the L condition.

In all conditions, adapting stimuli contained both morphological and lumi-nance cues. As such we can assume equal adaptation of those cells sensitive toAfrican faces and those sensitive to European faces in all conditions. The signifi-cant aftereffects for ML and M conditions indicates that these faces were largelyencoded by the adapted cells that are sensitive to racial cues, with the priorexposure resulting in the test faces appearing more expanded or contracted. Thelack of a significant contingent aftereffect in the L condition suggests that theseimages were not encoded proportionally more by either the European orAfrican sensitive cells. Interestingly the size of the aftereffects in the ML andM conditions were equal, indicating that both image types were being encodedby the relevant cells to the same extent. This shows that the contingency of theeffects measured here can be solely attributed to the racial typicality of thefaces’ morphology, with the exclusion of luminance variations in the M conditionnot affecting the degree to which images were encoded as European or African.Inspection of the data in the L condition in Figure 3 shows negative PSNs wererecorded for both African and European faces in both adaptation conditions,which may seem to suggest aftereffects of expansion were occurring. However,as this was a contingent aftereffects experiment, the principal measure is the com-parison of effects for stimuli adapted in opposite directions, and hence baselineratings were not collected prior to adaptation. Because of this we cannot besure that the undistorted images (i.e., neither expanded nor contracted) were per-ceived as normal and the negative PSNs may simply indicate that the imagesappeared somewhat expanded prior to adaptation.

The present findings support a growing body of evidence indicating that per-ceptions of race are primarily based on shape cues (Brooks and Gwinn, 2010;Hill et al., 1995; Willenbockel et al., 2011). There is, however, some evidenceof skin cues contributing to the other-race bias (ORB) (Alley & Schultheis,2001; Balas & Nelson, 2010; Brebner, Krigolson, Handy, Quadflieg, & Turk,2011). The ORB refers to the tendency of observers to be more accurate at recog-nizing faces belonging to their own race compared to other races (see Meissner &Brigham, 2001). Alley and Schultheis (2001) found that ORB-like effects couldbe induced by varying the skin luminance of faces with racially ambiguousmorphologies. Bar-Haim, Saidel, and Yovel (2009) also found that recognitionaccuracy could be affected by changes in skin lightness cues, however theseeffects were not as large as those induced by changes in morphology. Researchinto the ORB suggests that this effect may be largely attributable to social judg-ments of group membership and have less to do with the encoding of the raciallyrelevant visual aspects of a face (Bernstein, Young &Hugenberg, 2007). Bernsteinet al. (2007) found that deficits in recognition accuracy could be induced bysimply labelling faces as belonging to a different university or as having a differ-ent personality type to the observers, suggesting that once a face is categorized as

606 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

“out-group” it is not recognized as easily. It may be that changes in skin cues aresufficient to induce this categorization without affecting the overall racial appear-ance of a face. Additional support for the influence of social categorization in facerecognition is presented by Short and Mondloch (2010). In this study, participantsand facial images (all belonging to the same racial group) were arbitrarily assignedto personality groups with associated colours (backgrounds for images and wristbands for participants). In accordance with Bernstein et al. (2007), participantsrecognized faces assigned to their personality group more accurately than facesin the other group. In a separate task, the potential for these images to induceopposing aftereffects contingent upon personality type was assessed. Participantsadapted to facial images from each group that were expanded/contracted in oppo-site directions and subsequently rated the normality of test images that had alsobeen expanded or contracted. No significant aftereffects were observed, indicatingthat while some manipulations can produce variations in recognition this may notreflect encoding via separable neural populations. However, caution should beexercised when interpreting null results, as a failure to demonstrate a significantaftereffect could be due to any one of many details of the presentation ofimages (e.g., duration, extent of face distortion, etc.). It should be noted thatface adaptation effects contingent on social categories (personality type or occu-pation) have been successfully demonstrated by Little, Debruine, and Jones(2011).

A potential issue in the current study that warrants further investigation is thepossible influence of illumination. Perceptions of lightness are influenced by acombination of the amount of illumination a surface is receiving and the pro-portion of light that it reflects (Adelson, 2000). In the present study, a lightsource was not defined in the displays, meaning that participants may not havebeen sure whether changes in lightness were the result of varying skin tones orvarying lighting conditions. Within face perception research it remains standardpractice to present images in isolation and this concern applies to much of the lit-erature. While it remains to be seen whether many of the effects reported in theliterature persist under more natural viewing conditions, studies discussed herethat find an influence of reflectance information on identification and sex classifi-cation suggest that presenting facial images in isolation does not automaticallyrender reflectance information unreliable. However, if the current results couldbe replicated using a constant light source, this would strengthen the conclusionthat morphology is the dominant cue to race.

There may also be additional luminance-based cues to race that were notaccounted for. In Figure 1, it can be seen that in the ML condition the Africanface appears to have lighter sclera compared to the European face. Images inthe M condition should only contain morphological cues and yet this differenceremains, while for images in the L condition this difference is not present.Although manipulations of sclera lightness has been shown to influence percep-tions of health, attractiveness and age (Provine, Cabrera, & Nave-Blodgett,

ENCODING VISUAL CUES TO RACE 607

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

2013), we are not aware of any studies showing that similar manipulations caninfluence racial appearance. If sclera lightness were found to be a relevant cueto race, then the inclusion of this cue in the L condition may increase the likeli-hood of finding significant contingent aftereffects for these images.

The current study represents a new experimental approach to examining thecues that influence the perception of race and several aspects of the designinvite further examination in future experiments. We focused on university-agedCaucasian subjects in our investigation, as have many previous studies examiningthe perception of race (e.g., Brooks & Gwinn, 2010; Dunham, Stepanova, Dotsch,& Todorov, 2015; Stepanova et al., 2013; Willenbockel et al., 2011) and it wouldbe beneficial to determine whether the same is true for observers from other ageand racial groups. While our subjects were mainly young adults, childrenappear to rely more heavily on skin colour when making racial judgements(Dunham et al., 2015) and when judging the similarity of faces (Balas, Peissig& Moulson, 2015), raising the possibility that a younger sample may produceeffects that show a more substantial influence of luminance cues. In the currentstudy we used one type of aftereffect (expansion/contraction) and it remains tobe seen whether other face aftereffects are similarly contingent upon morphologi-cal cues to race. It is possible that aftereffects that do not involve shape distortionsmay show less of a contingency on morphological cues. Images were also dis-played under largely optimal conditions (i.e., for relatively long durations andin upright orientations) and observers may rely more on luminance cues when pro-cessing of morphological information is disrupted, such as when faces are inverted(Willenbockel et al., 2011). As stated in the introduction, we were interested inexamining the contributions of differences in luminance contained in photographsof faces of “Black” and “White” subjects and so greyscale images were used. Theuse of colour images containing additional pigmentation cues could see anincrease in the contribution made by reflectance information to perceptions ofrace, however studies that have used colour images still conclude morphologicaldifferences are the primary cue to race (Bar-Haim et al., 2009; Hill et al., 1995), atleast for adult observers (Balas et al., 2015). It is also important to note that theconclusions made here are restricted to perceptions of race. O’Neil and Webster(2011) demonstrated larger age aftereffects following adaptation to textural cuescompared to shape. If the current experiment was repeated using age-contingentas opposed to race-contingent aftereffects, we might expect to find the effects tobe predominantly contingent upon reflectance cues.

Previous studies examining the contributions made by morphological and skintone cues to the encoding of race have involved explicit judgments (e.g., Brooks& Gwinn, 2010; Willenbockel et al., 2011). This approach involves many high-level influences, and opens up the possibility of demand characteristics or otherinfluences beyond the initial stages of perceptual encoding. In the current studywe used adaptation and the observation of contingent aftereffects to examinevisual cues to race. The advantage of this method is its simplicity, not requiring

608 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

any specific mention of race or complex decision-making processes beyond thestage of encoding. The contingent aftereffects are used to infer the neural basis ofracial encoding, without asking the participant to make a racial judgement. Thestrength of this psychophysical method can be seen in the clear results obtainedusing relatively few observers. This represents a new experimental approach toexamining the cues that influence the perception of race and the resultssupport a growing body of evidence indicating that race is largely defined byshape cues.

REFERENCES

Addams, R. (1834). An account of a peculiar optical phenomenon seen after having looked at a movingbody. The London and Edinburgh Philosophical Magazine and Journal of Science, 5, 373–374.

Adelson, E. H. (2000). Lightness perception and lightness illusions. In M. Gazzaniga (Ed.), The NewCognitive Neurosciences, (pp. 339–351). Cambridge, MA: MIT Press.

Alley, T. R., & Schultheis, J. A. (2001). Is facial skin tone sufficient to produce a cross-racial identi-fication effect? Perceptual and Motor Skills, 92, 1191–1198. doi:10.2466/pms.2001.92.3c.1191

Balas, B., & Nelson, C. A. (2010). The role of face shape and pigmentation in other-race face percep-tion: An electrophysiological study. Neuropsychologia, 48, 498–506. doi:10.1016/j.neuropsychologia.2009.10.007

Balas, B., Peissig, J., & Moulson, M. (2015). Children (but not adults) judge similarity in own-andother-race faces by the color of their skin. Journal of Experimental Child Psychology, 130, 56–66.

Bar-Haim, Y., Saidel, T., & Yovel, G. (2009). The role of skin colour in face recognition. Perception,38, 145–148. doi:10.1068/p6307

Barlow, H. B., & Hill, R. M. (1963). Evidence for a physiological explanation of the waterfall phenom-enon and figural after-effects. Nature, 200, 1345–1347. doi:10.1038/2001345a0

Bernstein, M. J., Young, S. G., & Hugenberg, K. (2007). The cross-category effect mere social categ-orization is sufficient to elicit an own-group bias in face recognition. Psychological Science, 18,706–712. doi:10.1111/j.1467-9280.2007.01964.x

Biederman, I. (1987). Recognition-by-components: S theory of human image understanding.Psychological Review, 94(2), 115–147. doi:10.1037/0033-295X.94.2.115

Biederman, I., & Ju, G. (1988). Surface versus edge-based determinants of visual recognition.Cognitive Psychology, 20(1), 38–64. doi:10.1016/0010-0285(88)90024-2

Brebner, J. L., Krigolson, O., Handy, T. C., Quadflieg, S., & Turk, D. J. (2011). The importance of skincolor and facial structure in perceiving and remembering others: An electrophysiological study.Brain Research, 1388, 123–133. doi:10.1016/j.brainres.2011.02.090

Brooks, K. R., & Gwinn, O. S. (2010). No role for lightness in the perception of black and white?Simultaneous contrast affects perceived skin tone, but not perceived race. Perception, 39, 1142–1145. doi:10.1068/p6703

Brown, T. D., Dane, F. C., & Durham, M. D. (1998). Perception of race and ethnicity. Journal of SocialBehavior and Personality, 13, 295–306.

Bruce, V., Healey, P., Burton, M., Doyle, T., Coombes, A., & Linney, A. (1991). Recognising facialsurfaces. Perception, 20, 755–769. doi:10.1068/p200755

Bruce, V., & Langton, S. (1994). The use of pigmentation and shading information in recognising thesex and identities of faces. Perception, 23, 803–822. doi:10.1068/p230803

Clifford, C. W. G., & Rhodes, G. (2005). Fitting the mind to the world: Adaptation and after-effects inhigh-level vision (Vol. 2). Oxford: Oxford University Press.

ENCODING VISUAL CUES TO RACE 609

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

Dunham, Y., Stepanova, E. V., Dotsch, R., & Todorov, A. (2015). The development of race-basedperceptual categorization: Skin color dominates early category judgments. DevelopmentalScience, 18(3), 469–483.

Gwinn, O. S., & Brooks, K. R. (2013). Race-contingent face aftereffects: A result of perceived racialtypicality, not categorization. Journal of Vision, 13(10), 1–11.

Hill, H., Bruce, V., & Akamatsu, S. (1995). Perceiving the sex and race of faces: The role of shape andcolour. Proceedings of the Royal Society of London. Series B: Biological Sciences, 261, 367–373.doi:10.1098/rspb.1995.0161

Jaquet, E., & Rhodes, G. (2008). Face aftereffects indicate dissociable, but not distinct, coding of maleand female faces. Journal of Experimental Psychology: Human Perception and Performance, 34(1),101–112. doi:10.1037/0096-1523.34.1.101

Jaquet, E., Rhodes, G., & Hayward, W. G. (2007). Opposite aftereffects for Chinese and Caucasianfaces are selective for social category information and not just physical face differences. TheQuarterly Journal of Experimental Psychology, 60, 1457–1467. doi:10.1080/17470210701467870

Jaquet, E., Rhodes, G., & Hayward, W. G. (2008). Race-contingent aftereffects suggest distinct percep-tual norms for different race faces. Visual Cognition, 16, 734–753. doi:10.1080/13506280701350647

Krauskopf, J., Williams, D. R., & Heeley, D. W. (1982). Cardinal directions of color space. VisionResearch, 22, 1123–1131.

Leopold, D. A., O’Toole, A. J., Vetter, T., & Blanz, V. (2001). Prototype-referenced shape encodingrevealed by high-level aftereffects. Nature Neuroscience, 4(1), 89–94. doi:10.1038/82947

Little, A. C., DeBruine, L. M., & Jones, B. C. (2005). Sex-contingent face after-effects suggest distinctneural populations code male and female faces. Proceedings of the Royal Society B: BiologicalSciences, 272, 2283–2287.

Little, A. C., DeBruine, L. M., & Jones, B. C. (2011). Category-contingent face adaptation for novelcolour categories: Contingent effects are seen only after social or meaningful labelling. Cognition,118(1), 116–122.

Little, A. C., DeBruine, L. M., Jones, B. C., & Waitt, C. (2008). Category contingent aftereffects forfaces of different races, ages and species. Cognition, 106, 1537–1547.

Marr, D., & Nishihara, H. K. (1978). Representation and recognition of the spatial organization ofthree-dimensional shapes. Proceedings of the Royal Society of London. Series B. BiologicalSciences, 200, 269–294.

McCollough, C. (1965). Color adaptation of edge-detectors in the human visual system. Science, 149,1115–1116.

Meissner, C. A., & Brigham, J. C. (2001). Thirty years of investigating the own-race bias in memory forfaces: A meta-analytic review. Psychology, Public Policy, and Law, 7(1), 3–35.

Nestor, A., & Tarr, M. J. (2008). Gender recognition of human faces using color. PsychologicalScience, 19, 1242–1246.

O’leary, A., & McMahon, M. (1991). Adaptation to form distortion of a familiar shape. Perception &Psychophysics, 49, 328–332.

O’Neil, S. F., & Webster, M. A. (2011). Adaptation and the perception of facial age. Visual Cognition,19, 534–550.

O’Toole, A. J., Vetter, T., & Blanz, V. (1999). Three-dimensional shape and two-dimensional surfacereflectance contributions to face recognition: An application of three-dimensional morphing. VisionResearch, 39, 3145–3155.

Provine, R. R., Cabrera, M. O., & Nave-Blodgett, J. (2013). Red, yellow, and super-white sclera.Human Nature, 24(2), 126–136.

Rhodes, G., Jeffery, L., Watson, T. L., Clifford, C. W. G., & Nakayama, K. (2003). Fitting the mind tothe world face adaptation and attractiveness aftereffects. Psychological Science, 14, 558–566.

Russell, R. (2003). Sex, beauty, and the relative luminance of facial features. Perception, 32, 1093–1107.

610 GWINN AND BROOKS

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015

Russell, R. (2009). A sex difference in facial contrast and its exaggeration by cosmetics. Perception, 38,1211–1219.

Russell, R., Biederman, I., Nederhouser, M., & Sinha, P. (2007). The utility of surface reflectance forthe recognition of upright and inverted faces. Vision Research, 47(2), 157–165.

Russell, R., & Sinha, P. (2007). Real-world face recognition: The importance of surface reflectanceproperties. Perception, 36, 1368–1374.

Russell, R., Sinha, P., Biederman, I., & Nederhouser, M. (2006). Is pigmentation important for facerecognition? Evidence from contrast negation. Perception, 35, 749–759.

Short, L. A., & Mondloch, C. J. (2010). The importance of social factors is a matter of perception.Perception, 39, 1562–1564.

Stepanova, E. V., Strube, M. J., & Yablonsky, G. S. (2013). Now one is Russian, now one is not: Ethnicattitudes and categorization effects in the Russian federation. International Perspectives inPsychology: Research, Practice, Consultation, 2(2), 85–99.

Susilo, T., McKone, E., & Edwards, M. (2010). Solving the upside-down puzzle: Why do upright andinverted face aftereffects look alike? Journal of Vision, 10(13), 1–16.

Watson, T. L., & Clifford, C. W. G. (2003). Pulling faces: An investigation of the face-distortion after-effect. Perception, 32, 1109–1116.

Webster, M. A., & MacLeod, D. I. A. (2011). Visual adaptation and face perception. PhilosophicalTransactions of the Royal Society B: Biological Sciences, 366, 1702–1725.

Webster, M. A., & Maclin, O. H. (1999). Figural aftereffects in the perception of faces. PsychonomicBulletin & Review, 6, 647–653.

Willenbockel, V., Fiset, D., & Tanaka, J. W. (2011). Relative influences of lightness and facial mor-phology on perceived race. Perception, 40, 621–624.

ENCODING VISUAL CUES TO RACE 611

Dow

nloa

ded

by [

58.1

72.3

4.59

] at

16:

34 2

0 Se

ptem

ber

2015