impressions of babyfaced individuals across the …...mcarthur & apatow, 1983-1984;...

Developmental Psychology1992, Vol. 28, No. 6,1143-1152

Copyright 1992 by the American Psychological Association, Inc.

Impressions of Babyfaced Individuals Across the Life Span

Leslie A. ZebrowitzBrandeis University

Joann M. MontepareTufts University

Four questions were addressed concerning perceptions of babyfaced individuals from infancy toolder adulthood: (a) Do perceivers make reliable babyface judgments at each age; (b) does a baby-face have the same effects on trait impressions at each age; (c) are the effects of a babyface indepen-dent of the effects of attractiveness; and (d) what facial maturity features are associated with baby-face ratings, and do these features predict trait impressions? Ratings of portrait photographs re-vealed that perceivers reliably detect variations in babyfaceness across the life span. Facialmeasurements revealed that large eyes, a round face, thin eyebrows, and a small nose bridgecharacterized a babyface. Trait impressions showed a babyface overgeneralization effect at eachage: Babyfaced individuals were perceived to have more childlike traits than their maturefacedpeers, and this effect was independent of attractiveness.

Considerable evidence indicates that appearance has a signif-icant impact on social perceptions and social interactions,which may have important implications for personality develop-ment. Not only is appearance one of the initial qualities thatchildren and adults mention in their descriptions of people(Fiske & Cox, 1979; Livesley & Bromley, 1973), but it also influ-ences the traits they ascribe to others (see Alley, 1988; Bull &Rumsey, 1988; and McArthur, 1982, for reviews of pertinentresearch). Moreover, people's physical appearance influencesthe desire to become acquainted with them (Lyman, Hatelid, &Macurdy, 1981) as well as the nature of their social interactions(e.g., Langlois, 1986; Reis, Nezlek, & Wheeler, 1980; Reis,Wheeler, Spiegel, Kernis, Nezlek, & Perri, 1982; Snyder, Tanke,& Berscheid, 1977). Developmental theorists have argued thatthese effects of appearance on social perceptions and socialinteractions may yield self-fulfilling prophecy effects, wherebypeople with certain appearance qualities develop the traits thatare expected (Adams, 1977; Langlois, 1986; Sorrell & Nowak,1981).

One appearance quality that has been shown to influencesocial perceptions and social interactions is facial attractive-ness. Impressions of attractive people have been characterizedas a positive halo effect, whereby those who are physically at-tractive are perceived more positively than the unattractive on avariety of dimensions (see Adams, 1977; Berscheid & Walster,

Portions of this research were presented at the 42nd annual meetingof the Gerontological Society of America in Minneapolis, Minnesota,on November 1989. This research was supported by National Instituteof Mental Health Grant No. BSR-5-R01-MH42684 to Leslie A. Zebro-witz.

We thank Sheila Brownlow, Liz Tighe, Susan McDonald, and KarenHoffman for their help in collecting data. We also thank KatherineHildebrandt Karraker for sharing with us her slides of infant faces andBarbara Burek for procuring the slides of older adults from the archivesof the Institute for Human Development at the University of Califor-nia, Berkeley.

Correspondence concerning this article should be addressed to Les-lie A. Zebrowitz, Department of Psychology, Brandeis University,Waltham, Massachusetts 02254.

1974, for reviews). Moreover, the positive halo effect for facialattractiveness has been demonstrated for reactions to targetsranging from infants to older adults (e.g, Berkowitz & Frodi,1979; Dion, 1972,1974; Hildebrandt & Fitzgerald, 1983; John-son & Pittenger, 1984; Langlois, 1986; Stephan & Langlois,1984). Finally, facial attractiveness has been shown to have avariety of social interaction consequences (see Bull & Rumsey,1988, for a review of pertinent research).

A second appearance quality that influences social percep-tions and social interactions is facial babyishness. Impressionsof babyfaced people have been characterized as an overgeneral-ization effect, whereby adults with facial structures that etholo-gists identify as babyish (e.g., Guthrie, 1976; Lorenz, 1943) areperceived to have childlike traits. More specifically, adults withthe babyish facial features of large eyes, thin and high eyebrows,a large cranium, a small chin, and a curved rather than anangular face are perceived as more socially dependent, intellec-tually naive, physically weak, honest, and warm than their ma-turefaced peers (Berry & McArthur, 1985; Keating, 1985;McArthur & Apatow, 1983-1984; Zebrowitz-McArthur &Montepare, 1989). A babyface has also been shown to havenumerous social interaction consequences. (See Alley, 1988, forrelevant reviews and also Berry & Zebrowitz-McArthur, 1988;Zebrowitz, Kendall-Tackett, & Fafel, 1991; Zebrowitz &McDonald, 1991; Zebrowitz, Tenenbaum, & Goldstein, 1991.)

Facial attractiveness and babyishness are most likely to havean impact on personality development if perceivers can reliablyidentify these qualities in targets across the life span and if anattractive face or a babyface elicits the same trait expectations atvarious ages. Research has demonstrated that both of theseconditions are met for attractiveness (e.g, Berkowitz & Frodi,1979; Dion, 1972,1974; Hildebrandt & Fitzgerald, 1983; John-son & Pittenger, 1984; Langlois, 1986; Stephan & Langlois,1984). On the other hand, there has been no systematic investi-gation of these questions regarding a babyface, although thereis some evidence for the cross-age generality of the overgeneral-ization effect for perceptions of competence. Specifically, notonly are babyfaced adults perceived to be naive (Berry &McArthur, 1985; McArthur & Apatow, 1983-1984), but also

1143

1144 LESLIE A. ZEBROWITZ AND JOANN M. MONTEPARE

babyfaced infants may be perceived as less competent thanthose with a more mature appearance (Ritter, Casey, & Lang-lois, 1991). Because the effects of infants' babyfaceness, pre-sumed age, and attractiveness were not separated in the Ritteret al. study, the question remains as to whether maturefaced andbabyfaced infants will be perceived to have different competen-cies when they are equally attractive and when their ages areknown to perceivers.

The present study had four goals. The first goal was to deter-mine whether perceivers make reliable judgments about thefacial babyishness of male and female targets across the lifespan, including infants, preschoolers, young children, adoles-cents, young adults, and older adults. The second goal was tosystematically investigate whether the babyface overgeneraliza-tion effect holds true for targets across the life span. The thirdgoal was to determine whether the effects on impressions of ababyface are independent of the effects of attractiveness which,as noted earlier, has also been shown to influence impressions.A final goal was to identify the facial features associated with ababyface for male and female targets across the life span and todetermine whether these features influence trait impressions.

Method

Subjects

Three hundred and forty two students who were enrolled in an intro-ductory psychology class participated for partial course credit. Approx-imately equal numbers of men and women were randomly assigned torate targets representing one of six age groups: infants (N = 32), pre-schoolers (AT = 48), fifth graders (iV = 48), eighth graders (N = 48),young adults (iV= 118),1 and older adults (N = 48). Subjects assigned tothe infant group rated both male and female targets, whereas subjectsassigned to the other age groups rated either male or female targets. Foreach group of targets, subjects viewed one of two orders of faces andcompleted ratings in one of two orders.

Facial Stimuli

The facial stimuli consisted of black-and-white slides depicting thefaces of White male and female targets at each of the six age levels.Slides within each age level were selected with the goal of achieving arange of babyface and attractiveness values, and they consequently donot represent a random sample of faces at each age. The slides of infantswere drawn from a set of faces used in research by Hildebrandt andFitzgerald (1979). They consisted of 12 boys and 12 girls ranging in agefrom 5 to 9 months. The slides of children were made from yearbookphotographs of children attending a private school and included 24boys and 24 girls representing each of three grade levels: preschool,fifth grade, and eighth grade. The slides of the young adults were madefrom yearbook photographs of 20 men who were college seniors and 20women who were high school seniors. The slides of the young adultmen were those used by Berry and McArthur (1985). The slides of theolder adults were obtained from the Berkeley Growth and Guidancestudy archives of the Institute of Human Development at the Univer-sity of California, Berkeley, and included 24 men and 24 women be-tween 52 and 54 years of age.

Subjects who rated infants, children, or older adults were told a spe-cific age or school grade for each target group so that judgments abouttargets within each group would not reflect variations in their pre-sumed ages. Infants were described as 6 months old; preschoolers weredescribed as 3'/2 years old; the school grade was given for fifth andeighth graders; and older adults were described as 60 years old. Al-though the age of the young adults had not been specified when thefaces were rated for previous investigations (Berry & McArthur, 1985;Zebrowitz, Tenenbaum, & Goldstein, 1991), subjects in these studiesestimated the targets' ages so that presumed age could be statisticallycontrolled. The real ages of infants were also statistically controlled(see footnote 5).

Two slide orders were generated for each group of targets. Male andfemale infant faces were presented in blocks such that all of the malefaces were presented before the female faces in one order, whereas thesequence of male and female faces was reversed in the other order. Forthe remaining target groups, one order of slides was a random se-quence and the other was the reverse of the first with one exception:The second order of the eighth grade target faces consisted of a randomorder different from the first order.

Dependent Measures

Impression ratings. Targets were rated on scales reflecting dimen-sions that had been influenced by variations in facial babyishness inpast research with young adult faces (e.g., Berry & McArthur, 1986).Two types of rating scales were used: behavioral scales and bipolar traitscales. For the behavioral scales, subjects indicated on 7-point scaleswith endpoints labelled definitely no and definitely yes the extent towhich the targets were likely to perform certain behaviors. The bipolartrait scales consisted of 7-point scales along which subjects indicatedthe extent to which the target possessed a particular trait as opposed toits polar opposite.

A detailed description of the rating scales appears in Table 1, whereit can be seen that some scales were varied slightly across the differentage groups so that they were age appropriate. Also, young adult targetswere rated only on trait scales, because these data had been collected inprevious research that did not use the behavioral scales. The scaleswere grouped into five a priori composites paralleling the dimensionsused in past research on impressions of babyfaced adults (e.g., Berry &McArthur, 1986): social autonomy, physical weakness, naivete,warmth, and honesty. The last dimension was not represented for in-fant targets, because it was assumed that they would be rated uni-formly high on the deceitful/straightforward scale. For each composite,the behaviors described in Table 1 were scored in a positive direction,as were the second poles of the trait scales.

In addition to rating the targets' traits and likely behaviors, subjectsrated how maturefaced/babyfaced and how unattractive/attractiveeach target was on 7-point bipolar scales. Subjects who rated the youngadults—for whom ages were not supplied—also estimated the age, inyears, of each target. Ratings of the targets on all scales could rangebetween 1 and 7 and were coded so that higher scores representedgreater social autonomy, physical weakness, naivete, warmth, honesty,babyfaceness, or attractiveness.

Facial characteristics. Five age-related facial characteristics were

1 Eighty of these subjects had previously rated targets' naivete,warmth, and honesty in research by Berry and McArthur (1985). Theremaining 38 subjects rated physical weakness and social autonomy,which had not been included in the previous research.

BABYFACES ACROSS THE LIFE SPAN 1145

Table 1Dependent Measures and Reliability Coefficients Within the Six Age Groups

Infant Preschool 5th grade 8th grade Young adult Older adultComposites/individual

impression scales

Matureface/babyfaceNot at all attractive/very

attractiveSocial autonomy"

Dependent/independent1"

Submissive/dominating0

Not homesick at campNot cry when parents

leave the roomNot give in to friendsNot listen to parents

when they say "Don'ttouch"

Naivete*Sophisticated/naived

Not know right fromwrong like an adult

Not know it is naughtyto do certain things

Not able to followcomplicatedinstructions

Fooled into thinking atoy is gone if hiddenbehind your back

Physical weakness*Physically strong/

physically weakNot strong enough to

take a toy from a babyof the same age

Warmth*Not at all affectionate/

very affectionate'Enjoy being hugged

Honesty'

M

.87

.86

.94

.83

.84

.90

.91

.95

.77

.79

.83

.95

.89

.89

.90

F

.91

.75

.90

.67

.90

.85

.92

.92

.89

.75

.83

.89

.84

.91

.90

M

.86

.87

.90

.66

.79

.81

.57

.89

.65

.74

.76

.87

.90

.91

.83

.79

F

.93

.83

.96

.79

.66

.89

.87

.92

.86

.62

.83

.95

.90

.87

.84

.85

M

.96

.90

.98

.92

.88

.93

.91

.86

.91

.74

.75

.96

.90

.94

.94

.91

F

.96

.93

.95

.88

.83

.90

.64

.90

.90

.83

.87

.94

.92

.86

.91

.80

M

.91

.84

.96

.86

.89

.87

.82

.72

.79

.57

.74

.90

.94

.84

.85

.81

F

.93

.89

.94

.82

.62

.78

.76

.87

.88

.83

.83

.89

.92

.86

.89

.82

M

.98

.96

.83

.61

.91

.96

.96

.79

.98

.97

F

.87

.95

.95

.79

.88

.83

.84

.74

.72

M

.92

.90

.92

.79

.92

.85

.52

.81

.77

.95

.96

.93

.91

.80

F

.91

.91

.92

.83

.82

.89

.94

.88

.90

.74

.96

.93

.91

.77

Note. M = male; F = female. * Cronbach alpha coefficients for composite measures. b Scales for infants were very dependent/not very depen-dent. c Scales for infants were very submissive/not very submissive. d Scales for infants were not very naive/very naive and for adults were shrewd/naive. ' Scales for young adult men were cold/warm and cruel/kind. 'Scales for young adults were dishonest/honest; scales for all others weredeceitful/straightforward.

determined from facial measurements made independently by twojudges who projected each slide onto a wall and measured the facialfeatures depicted in Figure 1 to the nearest '/is of an inch. The specificfacial characteristics included eye area, eyebrow height, eye separa-tion, chin area, and cranium height. All of these features had beenshown in previous research to predict babyface ratings (Berry &McArthur, 1985; Zebrowitz-McArthur & Montepare, 1989).

In addition to the foregoing linear facial measures, three qualitativeratings were made of other facial features that had been shown to beassociated with babyface ratings. Specifically, two independent judgesrated the following facial characteristics of each target on 7-point bipo-lar scales: thin eyebrows/thick eyebrows, small nosebridge/large nose-bridge, and round face/angular face. Faces were coded as round to theextent that face length and width appeared equal, and the jaw wasrounded. They were coded as angular to the extent that face lengthappeared greater than width, and the jaw was angular. A neutral rating

was given to faces that appeared equal in length and width and had anangular jaw or to faces that had greater length than width and arounded jaw. Ratings of smiling were also obtained for use in the dataanalyses as a control variable. Specifically, the two judges rated theamount of smiling displayed by each target on 7-point scales withendpoints labeled little smile/big smile.

Procedure

Subjects were informed that the study dealt with how people's physi-cal appearance influences perceivers' impressions of them and thattheir task was to rate faces of different people on a series of scales. Eachface was presented for 8 s with a 2-s interslide interval, during whichtime subjects made their ratings on one scale. After the first rating ofall of the faces was completed, subjects viewed the set of target faces asecond time and completed the second rating. This procedure wasrepeated until all of the ratings had been made.


4.

10

Figure 1. Facial measurements. (The methods for calculating physiognomic measures were as follows:Eyesize was the product of eye height [4; distance between top and bottom of eye at center] and eye width[5; distance between inner and outer corners of the eye]. Eyebrow height was the distance from the pupilcenter to the eyebrow [3]. Eye separation was the distance between the inner corners of each eye [6]. Chinarea was the product of the width at the point halfway between the bottom lip and the base of the chin [9]and the-distance from the lower lip center to the base of the chin [10]. Cranium height was the distancefrom the middle of the pupil to the top of the head minus half of hair length [2 ]. To correct for variations inface size, all vertical measures were divided by face length [1 ], horizontal measures [5 ] and [6 ] were dividedby face width [7], and horizontal measure [9] was divided by face width [8].)

Two orders of rating scales were generated for each target group. Forinfant targets, half of the subjects rated the targets on a random orderof the behavioral and trait scales, and this rating order was reversed forthe remaining subjects. For the child and older adult targets, half of thesubjects rated the targets on a random order of the set of trait scalesfollowed by a random order of the set of the behavioral scales. Therandom order of scales within each set was reversed for the remainingsubjects, but the trait scales still preceded the behavioral scales. For theyoung adult targets, for whom only trait scales were used, half of thesubjects rated the targets on a random sequence of the scales, and thisrating order was reversed for the remaining subjects. For every targetgroup, ratings of attractiveness and babyfaceness were made after alltrait and behavioral ratings had been completed, with half of the sub-jects making babyface ratings first and half of the subjects rating attrac-tiveness first. These ratings were made last so that subjects would notbe explicitly aware of variations along these appearance dimensionswhen making their impression ratings.

Results

Data Base

Facial measures. Pearson correlation coefficients betweenthe two judges' measurements of facial features revealed high

reliability except for the size of the infants' nose bridge. The lowreliability for this measure (.33 for boys and .44 for girls) re-flected a tendency for the judges' ratings to be clustered at thefloor of the scale owing to the absence of a noticeable nosebridge in most infants. The average reliability across all targetage and sex groups for each facial measure was .80 for eye area,.84 for eyebrow height, .83 for eye separation, .80 for eyebrowthickness, .87 for cranium height, .84 for chin area, .82 for facialroundness, .74 for smiling, and .84 for nosebridge size, exclud-ing infants. The reliable facial measurements and ratings madeby the two judges were averaged to yield a single score for eachtarget for use in subsequent analyses.

Cronbach's coefficients computed across subjects' babyfaceand attractiveness ratings for male and female targets at eachage revealed high reliability (see Table 1). Thus, facial babyish-ness and facial attractiveness reliably differentiate individualsof both sexes and all ages. Having demonstrated acceptableinterrater reliability, mean babyface and attractiveness ratingsfor each face were computed across subjects. These means andstandard deviations are presented in Table 2.

To determine whether the variability in babyface ratings pro-vided equivalent potential for the prediction of trait impres-


Table 2Means and Standard Deviations for Babyface andAttractiveness Ratings Within the Six Age Groups

Male Female

Age group

InfantPreschool5th grade8th gradeYoung adultOlder adult

InfantPreschoolSth grade8th gradeYoung adultOlder adult

M SD

Babyface ratings

4.624.154.454.293.933.65

0.740.811.320.980.971.00

FM1(6,2O«) = 3.22

Attractiveness ratings

4.024.073.883.753.543.37

0.770.760.810.740.760.76

FmJ,6, 20) = 1.20

M

4.214.104.003.863.543.54FmJii

4.054.093.723.813.973.11F^t

1,20)

i,20)

SD

0.961.211.171.030.731.04

= 2.77

0.750.900.960.810.970.86

= 1.65

Note. For infants, n = 12; for young adults, n = 20; for all others, n = 24.For all Fma values, p > .05." df= the harmonic mean of the cell «s.

sions across the different age groups, we performed tests ofheterogeneity of variance. Hartley's test revealed that the aver-age variance in babyface ratings did not differ significantlyacross the six age groups for male or for female targets, bothps > .05 (see Table 2).2

Impression measures. Standardized Cronbach's coefficientscomputed across subjects' impressions ratings for male and fe-male targets at each age revealed high reliability (see Table 1).Having demonstrated acceptable interrater reliability, meanimpression ratings for each face were computed across subjects.Next, composite scores for social autonomy, physical weakness,naivete, warmth, and honesty were computed by averagingacross mean impression ratings included in the five a priorigroupings of the rating scales. As shown in Table 1, the alphacoefficients calculated for each composite were highly signifi-cant.3

Impact of a Babyface on Impressions

To assess whether previously documented effects of a baby-face held true for male and female targets at different age levels,we computed Pearson correlation coefficients between baby-face ratings and the composite impression scores within eachtarget group. These coefficients appear in Table 3 and providestrong evidence for cross-age and cross-gender generality in im-pressions elicited by a babyface. With few exceptions, baby-faced male and female targets in all age groups were perceivedas less socially autonomous, physically weaker, more naive, andwarmer than their more mature-looking peers. A babyface wasalso positively correlated with perceived honesty, although this

effect did not always achieve significance. Table 3 also includescorrelations between babyface and attractiveness ratings, aswell as correlations between attractiveness and trait compositesfor the interested reader.4

Although the correlations shown in Table 3 indicate that fa-cial babyishness and attractiveness were significantly corre-lated only in the infant age group, we performed a series offorced-entry regression analyses to establish that the effects of ababyface on impressions were independent of any linear attrac-tiveness effects. Degree of smiling was also entered as a predic-tor in these analyses along with the babyface and attractivenessratings, because smiling was not held constant in the facialphotographs and it has been shown to affect social perceptions(Cunningham, 1986; Keating, Mazur, & Segall, 1981).

The resultant partial correlations between babyface and traitratings controlling for the linear effects of attractiveness andsmiling are presented in Table 3. A babyface continued to pre-dict impressions of low social autonomy, high physical weak-ness, high naivete, high warmth, and high honesty.5 An excep-tion to the general tendency for the effects of a babyface to holdup in the partial correlations was obtained for ratings of infants'

2 Note that the means in Table 2 reveal a linear decrease in babyfaceratings with increasing age for both male and female targets, linearFs(l, 122) = 6.60 and 6.66, respectively, both ps < .01, as well as margin-ally higher babyface ratings for female targets, F(\, 244) = 3.26, p<.07.There was also a linear decrease in attractiveness ratings with increas-ing age for male and female targets, linear Fs(l, 122) = 13.07 and 10.59,respectively, both ps < .001, but no sex differences in attractiveness,F < 1 .

3 A factor analysis revealed that the five composites could be re-duced to two: power, comprised of high social autonomy, low naivete,and high physical strength; and social goodness, comprised of highwarmth and honesty. However, all five composites were retained fortwo reasons: (a) They parallel the five rating dimensions used in pastresearch with adult targets; and (b) although high social autonomy, lownaivete, and high physical strength can be statistically grouped, eachmay relate differently to the appearance variables. For example, pastresearch has shown that attractiveness creates impressions of highcompetence (i.e., low naivete), but there is no evidence that it createsimpressions of physical strength.

4 Although various relationships have been hypothesized betweenfacial babyishness and attractiveness in adults, past research has notyielded consistent effects, and the present findings are therefore not atvariance with the existing literature (see Berry, 1991; Cunningham,1986,1990; Keating, 1985; McArthur & Apatow, 1983-1984).

5 It should be recalled that subjects were told a specific age or gradelevel for all targets except young adults, so that babyface effects wouldnot be confounded with variations in presumed age. To assess theeffects of presumed age for young adults, the regressions were re-peated, adding subjects' mean age estimates for each target as a predic-tor. Although presumed age occasionally predicted impressions, thebabyface effect lost significance in only one instance, as a predictor ofmen's physical weakness. Because infants' actual ages were known tothe researchers, and because small variations in age may have contrib-uted to impression effects, the regression analyses were also repeatedfor infants, controlling for actual age. The predictive power of babyface-ness remained highly significant for all impressions, and actual ageemerged as a significant predictor only for perceived warmth, withwhich it was negatively related. Thus the effects of babyfaceness can-not be accounted for by variations in presumed or actual age.


Table 3

Zero-Order and Partial Correlations ofBabyface and Attractiveness Ratings With Impression Scores Within the Six Age Groups

Impression/age group

Social autonomyInfantPreschool5th grade8th gradeYoung adultOlder adult

NaiveteInfantPreschool5th grade8th gradeYoung adultOlder adult

Physical weaknessInfantPreschool5th grade8th gradeYoung adultOlder adult

WarmthInfantPreschool5th grade8th gradeYoung adultOlder adult

HonestyInfantPreschool5th grade8th gradeYoung adultOlder adult

AttractivenessInfantPreschool5 th grade8th gradeYoung adultOlder adult

Zero-order correlation

Babyface

M

-.78**- . 6 1 * *-.84***-.94***-.78***-.57**

.85***

.65***

.86***

.64***

.81***

.70***

.60

.50*

.93***

.92***

.81***

.17

.82***

.46*

.78***

.74***

.78***

.55**

—.27.68***.46*.70***.40*

.81***

.08

.17- .12

.38t

.12

F

- .66*-.78***-.86***-.83***-.50*- .81***

.81***

.89***

.90***

.82***

.54*

.82***

•55f.79***.86***.72***.46*.62***

.70*

.52**

.73***

.44*

.27

.76***

—.62***.35.26.36.69***

,49t- .29- .20- .22

.06

.07

Attractive

M

-.52.19.07.15

-.21.08

.46-.52**- .23-.62***

.16- .08

.44- .06-.01-.09

.19- .20

.86***

.42*

.56**

.23

.56**

.59**

—.23-36f.36t.60**.49*

F

- .59*.17.28.21

- .11- .05

•55f- .48*-.54**-.49*

.17- .05

.60*- .30- .12

.06

.34- .03

.66*

.24

.31

.21.52*.55**

—.17.31.07.25.40*

Partial

Babyface*

M

-.80**-.66***-.88***-.94***-.76***-.68***

.95*"*

.82***

.94***

.74***

.81***

.74***

.55f

.51*

.94***

.91***

.80***

.34

.39

.58**94**..84***.72***.56**

—.26.67***.55**.62**.43*

F

- .56f-.85***-.86***-.80***- .49*- . 6 1 * *

.75**

.91***

.96***

.81***

.53*

.65***

.35

.85***

.87***

.72***

.46f,36f

.50

.64***

.85***

.58**

.34

.71***

—.71***.44*.25.37.49*

correlation

Attractive1"

M

.29

.34

.20

.07

.13

.02

-.77**-.76***-.64***-.70***- .26- .14

- .06- .13-.31

.06-.21-.08

.60f

.58**

.44*

.57**

.48*

.62**

—.22.31.49*.50*.51*

F

- .45-.21

.20

.02

.07

.17

.34- .49*-.87***-.54**

.16-.35

.44- .06

.16

.33

.34-.19

.46

.48*

.68***

.31

.57*

.77***

—.47*,40t.14.26.40f

Note. For infants, n = 12; for young adults, n = 20; for all others, n = 24. " These partial correlations control for linear effects of attractiveness andsmiling. b These partial correlations control for linear effects of babyfaceness and smiling.ip<A0. *p<.05. **p<.01. ***/><.001.

warmth. Whereas the zero-order correlations revealed a posi-tive relationship between a babyface and the perceived warmthof infants, this relationship was not significant in the partialcorrelations because of the positive correlation between attrac-tiveness and babyfaceness. The regression analyses also re-vealed some independent effects on impressions of attractive-ness when the linear effects of babyfaceness and smiling werecontrolled. Specifically, attractiveness tended to predict im-pressions of high warmth, high honesty, and low naivete (seeTable 3).

Structural Components of a Babyface

Having demonstrated that a babyface influences impres-sions across the life span, we attempted to identify the facialfeatures that gave rise to the perception of a babyface. Because ababyface is a configural quality that cannot be captured by anysingle structural feature, we performed regression analyses todetermine the predictive power of a constellation of features.Specifically we entered the eight facial structure measures aspredictors of babyface ratings in hierarchical regression analy-


ses within target sex using the combined data for preschoolersthrough older adults. Infants were omitted from these analyses,because one of the facial predictors, nosebridge size, had notbeen reliable for that group. In these analyses, targets' agegroup was entered on the first step as a control variable, and thefacial measures were entered as predictors using a stepwise pro-cedure on the second step.

The subset of significant predictors that emerged for malefaces included a round as opposed to an angular face, large eyes,and thin eyebrows, which together accounted for 32% of theadjusted variance in babyface ratings over and above that ex-plained by age group alone, F(4,111) = 16.01, p < .0001. Thesubset of significant predictors for female faces included a smallnose bridge and a round face, accounting for 25% of the ad-justed variance in babyface ratings over and above that ex-plained by age group alone, F(3,112) = 15.24, p < .0001. Step-wise regressions performed within each target age and sexgrouping yielded results quite similar to those from the cross-age regressions. In particular, a rounder face was a significantpredictor for male targets in every age group except eighthgraders, and a smaller nosebridge was a significant predictor forfemale targets in every age group except older adults. We incor-porated the physiognomic predictors of babyface ratings intostructural babyface composites for use in subsequent analysesby converting their values to z scores within their respectivetarget age and sex group and then computing the average of thetransformed scores.

Because the measure of nosebridge size was unreliable forinfants, a separate regression analysis was performed for thesetargets in which the seven facial measures excluding nosebridgewere entered as predictors of babyface ratings within each tar-get sex. A single predictor emerged for male infants—largeeyes—accounting for 42% of the adjusted variance in babyfaceratings, F(l, 10) = 8.86, p < .05, and eye size was used as astructural babyface index for male infants. The regressionequation was not significant for female infants, and conse-quently the one facial component that was most highly corre-lated with babyface ratings—a large cranium, r(10) = .55, p <. 10—was used as a structural babyface index for female infants.These structural babyface composites were more highly corre-lated with infant babyface ratings than the composites derivedfrom the other age groups with nosebridge deleted. The averageintercorrelation between babyface ratings and the structuralbabyface composite scores across all age groups was .60 for bothmale and female targets (see Table 4).

Table 4 reports the correlations between trait impressionsand the structural babyface composite for each target group.Male targets who were independently determined to have thebabyish features of relatively large eyes, thin eyebrows, andround faces (or, in the case of infants, large eyes) are perceived asless socially autonomous, physically weaker, more naive,warmer, and more honest than those with smaller eyes, thickereyebrows, and more angular faces. This pattern of results repli-cated across all age groups, although the correlations were notsignificant for preschool boys and, in some cases, older men,which are the two age groups for which the male structuralbabyface composite was most weakly correlated with babyfaceratings. Table 4 also reveals that female targets who were inde-

pendently determined to have the babyish features of relativelysmall nosebridges and round faces (or, for infants, a large cra-nium) were also perceived as less autonomous, weaker, andmore naive, though not consistently warmer or more honestthan those with larger nosebridges and more angular faces.This pattern of results replicated across all age groups, althoughthe correlations were not significant for infants or older adultwomen.6

Discussion

The present findings reveal that perceivers do reliably detectvariations in facial babyishness across the life span. There arebabyfaced babies, babyfaced children, babyfaced adolescents,babyfaced young adults, and babyfaced older adults of bothsexes. The present study also determined that babyface ratingshave an objective referent in structural features that generalizeacross the life span. Specifically, larger eyes, thinner eyebrows,and a rounder face are characteristic of more babyfaced malesfrom infancy through older adulthood, whereas a smaller nose-bridge and a rounder face are characteristic of more babyfacedfemales.

The fact that somewhat different facial features emerged asthe best predictors of a babyface for males and females mayreflect idiosyncracies in the faces of the targets used in thisresearch. Evidence to indicate that the same facial features maypredict babyface ratings for certain male and female faces isprovided by the finding that eye size, a significant predictor formale infants as well as for male targets in the cross-age regres-sions, did emerge as a significant predictor for preschool andfifth grade girls in within-age regressions. Also, nose bridgesize, a significant predictor only for female targets in the cross-age regressions, emerged as a significant predictor for fifth andeighth grade boys in within-age regressions. Thus, there is someevidence to suggest that the components of a babyface do gener-alize across targets of both sexes.

Although there are similarities in the determinants of a baby-face across sex, it is also possible that the most salient determi-nants differ somewhat for male and female faces. In particular,the finding that larger eyes and thinner eyebrows were morestrongly associated with babyface ratings for men could reflectthe fact that women in our culture often pluck their eyebrowsand apply makeup to enlarge the appearance of their eyes,making large eyes and thin eyebrows less salient markers of ababyface for women. Consistent with this suggestion, femaletargets in the present study had significantly thinner eyebrowsand larger eyes than male targets, and both of these effects were

6 We also performed regression analyses to determine which, if any,of the measured features predicted attractiveness ratings. For malepreschoolers through older adult males, the only significant predictorwas large eyes, which accounted for 3% of the adjusted variance. Forfemale preschoolers and adults, the only significant predictor was higheyebrows, which accounted for 6% of the adjusted variance. The pre-diction of attractiveness was no better in within-age regression analy-ses than in these cross-age analyses, and it was no better in separateanalyses of the ratings of male and female subjects than in the analysesfor all subjects combined.


Table 4Zero-Order Correlations Between the Structural Babyface Composites and Impressionand Perceived Babyface Scores Within the Six Age Groups

Impression

Social autonomyMF

Physical weaknessMF

NaiveteMF

WarmthMF

HonestyMF

BabyfaceMF

Infant

-.81***-.44

.81***

.43

.66**

.52t

.58**

.39

—

.68*•55t

Preschool

-•39f-.47*

.33f

.42*

.25

.58**

.01

.17

.15•34f

.44*

.67***

5thgrade

-.69***-.42*

.73***

.42*

.77***

.53**

.45*

.26

.54**

.06

.76***

.61**

8thgrade

-.64***-.69***

.59**

.47*

.17

.66***

.62***

.12

.50*

.02

.67***

.64***

Youngadult

-.48*-.76***

•41t.62**

.56**

.70***

.53*

.14

.46*

.40t

.72***

.55*

Olderadult

-.61**-.28

.45*

.18

.48*

.27

.09

.31

.37t

.35f

.32

.59**

Note. For male targets, the structural babyface composite is large eyes for infants, and large eyes, thineyebrows, and round face for all other age groups. For female targets, the structural babyface composite is alarge cranium for infants, and a small nosebridge and a round face for all other age groups. For infants, n =12; for young adults, n = 20; for all others, n = 24.

0. *p<.05. **p<.01. ***p<.001.

significant only for adults. There were no significant sex differ-ences in any of the other features that were included in thebabyface composites.

Perceivers not only agreed as to who was baby faced, but theyalso shared common expectations concerning the traits pos-sessed by these targets: The babyface overgeneralization effectheld true for male and female targets across the life span. Withthe linear effects of attractiveness and smiling held constant,more babyfaced male and female targets from infancy to olderadulthood were perceived as less socially autonomous andmore naive than their more maturefaced peers. Specifically, themore babyfaced targets were perceived as more dependent,more submissive, more naive and easily fooled, less likely toknow right from wrong, and less able to follow complicatedinstructions. More babyfaced individuals were also perceivedas physically weaker, albeit only at a marginal level of signifi-cance for infants and older adults.

Babyfaced male and female targets not only were perceivedas more childlike in the domains of social, intellectual, andphysical power, but they were also perceived as warmer andmore honest. Specifically, they were rated as more affectionateand desirous of being hugged, kinder, warmer, more straight-forward, and more honest than their maturefaced peers, al-though the last effect cannot be generalized to infants becausetheir perceived honesty was not assessed. Also, it should benoted that the effects of a babyface and attractiveness on im-pressions of infants' warmth were not independent of eachother. Finally, the strong correlations between trait ratings andthe structural babyface composite demonstrate that the corre-lations between trait and babyface ratings do not merely reflectcarryover effects from subjects' trait ratings to their babyface

ratings. Rather, these trait ratings are evoked by a specific con-stellation of infantile facial features that can be measured objec-tively.

Although the foregoing effects of a babyface on social per-ceptions held true with the linear effects of attractiveness con-trolled, the question remains as to the joint effects of variousdegrees of facial babyishness and attractiveness. A recent studyby Berry (1991) simultaneously manipulated three levels of fa-cial babyishness and attractiveness in young adult faces andfound that the perception of highly babyfaced men and womenas less powerful than their least babyfaced peers held true at alllevels of attractiveness. Similarly, the perception of highly baby-faced men as more sincere than those low in facial babyishnessheld true at all levels of attractiveness. However, the perceptionof highly babyfaced women as more sincere did not hold truefor women low in attractiveness. Given this evidence of an in-teractive effect of facial babyishness and attractiveness, it wouldbe interesting to investigate their joint effects on impressionsacross the life span as Berry (1991) did at a single age level.

In addition to providing evidence for cross-age generality ofthe babyface overgeneralization effect, the present results reaf-firm the cross-age generality of the attractiveness halo effect.With babyface ratings and smiling held constant, more attrac-tive male and female targets from preschoolers to older adultstended to be perceived as warmer and more honest than theirless attractive peers. In contrast to these findings, attractivenesshad no effect on perceptions of the targets' social autonomy orphysical weakness. This could reflect the fact that these traitsdo not have the strong evaluative connotations of warmth andhonesty and consequently do not engage the attractiveness haloeffect. On the other hand, being physically strong, dominant,


and independent are certainly positive traits in our society, par-ticularly for men, which suggests that the attractiveness halomay be more specific than previously assumed. Finally, attrac-tiveness, like a babyface, had an independent effect on percep-tions of naivete, albeit in the opposite direction. Whereas ababyface increased impressions of naivete, an attractive facediminished such impressions, which is consistent with past evi-dence for an attractiveness halo effect on judgments of intellec-tual competence.

The pattern of results on the naivete measures for infantsclarifies the contribution of facial maturity and attractivenessto the findings of Ritter et al. (1991). These authors found that6-month-old infants who were both relatively young looking(i.e., babyfaced) and attractive were rated less positively on spe-cific measures of cognitive skills, such as uncovering a hiddentoy, but more positively on global measures of cognitive compe-tency. To explain these findings, Ritter et al. suggested thatratings of specific skills were influenced by infants' facial matu-rity, whereas global ratings were influenced by their attractive-ness. The partial correlations in the present study support thisargument: Infant babyfaceness decreased the positivity of com-petence (naivete) ratings, whereas infant attractiveness in-creased the positivity of these ratings, at least for male infants.Thus, attractiveness has the same halo effect on the perceivedcompetence of infants as on the perceived competence of indi-viduals from other age groups so long as facial babyishness iscontrolled.

The present findings clearly demonstrate that facial babyish-ness, like attractiveness, has the potential to influence personal-ity development across the life span. Individuals of all ages arereliably differentiated along this dimension. Moreover, a baby-face elicits the same strong trait expectations, regardless of thetarget's age, and these expectations may result in their beingtreated in ways that elicit confirmatory behavior. Indeed, re-cent research reveals that parents assign less intellectually de-manding tasks to babyfaced children and also respond differ-ently to their misdeeds (Zebrowitz, Kendall-Tackett, & Fafel,1991). Whereas the existent cross-sectional research revealsstrong effects of facial babyishness on social perceptions andsocial interactions, longitudinal research investigating the sta-bility of facial babyishness across the life span is needed tofurther clarify its likely impact on personality development.

It is noteworthy that Zebrowitz, Kendall-Tackett, and Fafel(1991) selected babyfaced and maturefaced children on thebasis of the ratings provided by undergraduates in the presentstudy, and they found that parents' facial maturity ratingsshowed strong agreement with those of the undergraduates.This demonstrates that the impressions documented in thepresent study generalize beyond the population of college stu-dents. Additional research also indicates that impressions ofbabyfaced individuals generalize across perceiver's race and age(McArthur & Berry, 1987; Montepare & Zebrowitz-McArthur,1989). Finally, although the faces investigated in the presentstudy were selected to represent a range of values on the baby-face dimension, there is reason to believe that the results wouldgeneralize to a more random sample of faces. More specifically,a recent investigation revealed a significant impact of litigantfacial maturity on adjudications in small claims court, whereover 1,000 faces were represented (Zebrowitz & McDonald,

1991). Thus, the effects of a babyface across the life span thatwere documented in the present study should hold true across awide range of perceivers and targets.

References

Adams, G. R. (1977). Physical attractiveness research. Toward a devel-opmental social psychology of beauty. Human Development, 20,217-239.

Alley, T. R. (Ed.). (1988). Social and applied aspects of perceiving faces.Hillsdale, NJ: Erlbaum.

Berkowitz, L., & Frodi, A. (1979). Reactions to a child's mistakes asaffected by her/his looks and speech. Social Psychology Quarterly,42, 420-425.

Berry, D. S. (1991). Attractive faces are not all created equal: Jointeffects of facial babyishness and attractiveness on social perception.Personality and Social Psychology Bulletin, 17, 523-531.

Berry, D. S., & McArthur, L. Z. (1985). Some components and conse-quences of a babyface. Journal of Personality and Social Psychology,48, 312-323.

Berry, D. S., & McArthur, L. Z. (1986). Perceiving character in faces:The impact of age-related craniofacial changes on social perception.Psychological Bulletin, 100, 3-18.

Berry, D. S., & Zebrowitz-McArthur, L. (1988). What's in a face? Facialmaturity and the attribution of legal responsibility. Personality andSocial Psychology Bulletin, 14, 23-33.

Berscheid, E., & Walster, E. (1974). Physical attractiveness. In L. Ber-kowitz (Ed.), Advances in experimental social psychology (Vol. 7, pp.158-216). San Diego, CA: Academic Press.

Bull, R., & Rumsey, N. (1988). The social psychology of facial appear-ance. New York: Springer-Verlag.

Cunningham, M. R. (1986). Measuring the physical in physical attrac-tiveness: Quasi-experiments on the sociobiology of female facialbeauty. Journal of Personality and Social Psychology, 50, 925-935.

Cunningham, M. R. (1990). What do women want? Facialmetric as-sessment of multiple motives in the perception of male facial physi-cal attractiveness. Journal of Personality and Social Psychology, 59,61-72.

Dion, K. K. (1972). Physical attractiveness and evaluation of children'stransgressions. Journal of Personality and Social Psychology, 24,207-213.

Dion, K. K. (1974). Children's physical attractiveness and sex as deter-minants of adult punitiveness. Developmental Psychology, 10, 772-778.

Fiske, S. T., & Cox, M. G. (1979). Person concepts: The effect of targetfamiliarity and descriptive purpose on the process of describingothers. Journal of Personality, 47, 136-161.

Guthrie, R. D. (1976). Body hot spots: The anatomy of human socialorgans and behavior. New "York: Van Nostrand Reinhold.

Hildebrandt, K. A., & Fitzgerald, H. E. (1979). Facial feature determi-nants of perceived infant attractiveness. Infant Behavior and Develop-ment, 2, 329-339.

Hildebrandt, K. A., & Fitzgerald, H. E. (1983). The infant's physicalattractiveness: Its effect on bonding and attachment. Infant MentalHealth Journal, 4,3-12.

Johnson, D. E, & Pittenger, J. B. (1984). Attribution, the attractivenessstereotype and the elderly. Developmental Psychology, 20, 1168-1172.

Keating, C. F. (1985). Human dominance signals: The primate in us. InS. L. Ellyson & J. F. Dovidio (Eds.), Power, dominance, and nonverbalbehavior (pp. 89-108). New York: Springer-Verlag.

Keating, C. F, Mazur, A., & Segall, M. H. (1981). A cross-culturalexploration of physiognomic traits of dominance and happiness.Ethology and Sociobiology, 2, 41-48.


Langlois, J. H. (1986). From the eye of the beholder to behavioral real-ity: The development of social behaviors and social relations as afunction of physical attractiveness. In C. P. Herman, M. P Zanna, &E. T. Higgins (Eds.), Physical appearance, stigma, and social behav-ior: The Ontario symposium, Vol. 3 (pp. 23-51). Hillsdale, NJ: Erl-baum.

Livesiey, W J., & Bromley, D. B. (1973). Person perception in childhoodand adolescence. London: Wiley.

Lorenz, K. (1943). Die angeborenen Formen moglicher Arfahrung[The innate forms of potential experience]. Zietschrift fur Tierpsy-chologie, 5, 234-409.

Lyman, B., Hatelid, D., & Macurdy, C. (1981). Stimulus person cues infirst impression attraction. Perceptual and Motor Skills, 52, 59-66.

McArthur, L. Z. (1982). Judging a book by its cover: A cognitive analy-sis of the relationship between physical appearance and stereotyp-ing. In A. Hastorf & A. Isen (Eds.), Cognitive social psychology (pp.149-211). New York: Elsevier/North-Holland.

McArthur, L. Z., & Apatow, K. (1983-1984). Impressions of babyfacedadults. Social Cognition, 2, 315-342.

McArthur, L. Z., & Berry, D. S. (1987). Cross-cultural agreement inperceptions of babyfaced adults. Journal ofCross-Cultural Psychol-ogy, 18,165-192.

Montepare, J. M., & Zebrowitz-McArthur, L. Z. (1989). Children's per-ceptions of babyfaced adults. Perceptual and Motor Skills, 69, 467-472.

Reis, H. T., Nezlek, J., & Wheeler, L. (1980). Physical attractiveness insocial interaction. Journal of Personality and Social Psychology, 38,604-617.

Reis, H. T., Wheeler, L., Spiegel, N., Kernis, M. H., Nezlek, J., & Perri,M. (1982). Physical attractiveness in social interaction: II. Why doesappearance affect social experience? Journal of Personality and So-cial Psychology, 43, 979-996.

Ritter, J. M., Casey, R. J., & Langlois, J. H. (1991). Adults' responses toinfants varying in appearance of age and attractiveness. Child Devel-opment, 62, 68-82.

Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception andsocial behavior: On the self-fulfilling nature of social stereotypes.Journal of Personality and Social Psychology, 35, 656-666.

Sorrell, G. T, & Nowak, C. A. (1981). The role of physical attractivenessas a contributor to individual development. In R. M. Lerner & N. A.Busch-Rossnagel (Eds.), Individuals as producers of their develop-ment: A life span perspective (pp. 389-446). London: AcademicPress.

Stephan, C, & Langlois, J. H. (1984). Baby beautiful: Adult attributionsof infant competence as a function of infant attractiveness. ChildDevelopment, 55, 576-585.

Zebrowitz, L. A., Kendall-Tackett, K., & Fafel, X (1991). The impact ofchildren's facial maturity on parental expectations and punish-ments. Journal of Experimental Child Psychology, 52, 221-238.

Zebrowitz, L. A., & McDonald, S. (1991). The impact of litigants' baby-facedness and attractiveness on adjudications in small claims courts.Law and Human Behavior, 15, 603-623.

Zebrowitz, L. A., Tenenbaum, D. R., & Goldstein, L. H. (1991). Theimpact of job applicants' facial maturity, sex, and academic achieve-ment on hiring recommendations. Journal of Applied Social Psychol-ogy, 21. 525-548.

Zebrowitz-McArthur, L., & Montepare, J. M. (1989). Contributions ofa babyface and a childlike voice to impressions of moving and talk-ing faces. Journal of Nonverbal Behavior, 13,189-203.

Received June 15,1990Revision received August 9,1991

Accepted October 7,1991 •

impressions of babyfaced individuals across the …...mcarthur & apatow, 1983-1984;...

Documents