Ng, S.Y., Zebrowitz, L.A., & Franklin, R.G. Jr (2014). Age differences in the differentiation of trait impressions from faces. Journals of Gerontology Series, B: Psychological Sciences and Social Sciences,
doi:10.1093/geronb/gbu113
Age Differences in the Differentiation of Trait
Impressions From Faces
Stacey Y. Ng, Leslie A. Zebrowitz, and Robert G. Franklin, Jr.
Department of Psychology, Brandeis University, Waltham, Massachusetts.
Objectives. We investigated whether evidence that older adults (OA) show less differentiation of visual stimuli than
younger adults (YA) extends to trait impressions from faces and effects of face age. We also examined whether age differences in mood, vision, or cognition-mediated differentiation differences. Finally, we investigated whether age differences
in trait differentiation mediated differences in impression positivity.
Results. OA showed less differentiated trait ratings than YA. Measures of mood, vision, and cognition did not mediate
these rater age differences. Hostility was differentiated more for younger than older faces, while health was differentiated
more for older faces, but only by OA. Age differences in differentiation mediated age differences in impression positivity.
Discussion. Less differentiation of trait impressions from faces in OA is consistent with previous evidence for less
differentiation in face and emotion recognition. Results indicated that that age-related dedifferentiation does not reflect
narrow changes in visual function. They also provide a novel explanation for OA positivity effects.
Key Words: Aging—Differentiation— Face perception—Positivity bias— Trait impressions.
P
eople readily form first impressions of others’ psychological qualities from their facial appearance.
Indeed, after only 39 ms of exposure to a face, people’s
impressions are no different than after prolonged exposure
(Bar, Neta, & Linz, 2006; Willis & Todorov, 2006). While
trait impressions from faces show strong agreement across
raters of various ages and cultural backgrounds (Zebrowitz,
Franklin, Hillman, & Boc, 2013; Zebrowitz, Montepare,
& Lee, 1993; Zebrowitz, et al., 2012), some differences
exist. In particular, older adults (OA) form more positive trait impressions than do younger adults (YA) (Castle
et al., 2012; Ruffman, Sullivan, & Edge, 2006; Zebrowitz,
Franklin, Hillman, & Boc et al., 2013). This and other positivity biases have been attributed to age-related changes in
motivation. According to the socioemotional selectivity theory (Carstensen, 1995; Carstensen & Mikels, 2005), OA are
more motivated than YA to maintain a positive mood due
to a shorter future time perspective, which yields less processing of negative stimuli or greater processing of positive
stimuli (Carstensen, Isaacowitz, & Charles, 1999; Murphy
& Isaacowitz, 2008; Isaacowitz et al., 2007; Kellough &
Knight, 2012). This study tested an alternative hypothesis
based on the dedifferentiation hypothesis of cognitive aging
which posits that cognitive processes become less distinct
in older adulthood (Balinsky, 1941; Baltes, Cornelius,
Spiro, Nesselroade, & Willis, 1980; Salthouse, Hancock,
Meinz, & Hambrick, 1996). Specifically, we predicted that
OA would show less differentiation in their trait ratings than
do YA—that is OA would perceive faces as less different
from one another—and that these age differences would
contribute to more positive trait ratings by OA.
Rater Age and Differentiation
The hypothesis that OA will show less differentiation in
trait impressions of faces is supported by evidence that OA
show a reduced ability to perceive other differences among
visual stimuli. For example, accurate detection or recall of
stimuli by OA requires greater levels of visual contrast and
greater differences in their orientation or pattern (Betts,
Sekuler, & Bennett, 2007; Toner, Pirogovsky, Kirwan, &
Gilbert, 2009). Most pertinent to this study, OA showed
reduced accuracy in face recognition and higher false recognition of new faces (Bartlett, Leslie, Tubbs, & Fulton,
1989). They also showed reduced ability to perceive differences in faces that are revealed in changed expressions,
reversed photos, or changed poses (Bartlett & Leslie, 1986),
and they required greater differences between face pairs in
order to discriminate faces (Goh, Suzuki, & Park, 2010).
In addition, OA have shown poorer recognition of emotional expressions (Ruffman, Halberstadt, & Murray, 2009;
Ruffman, Henry, Livingstone, & Phillips, 2008), including many intensities of sadness, fear, and anger (Ortega
& Phillips, 2008), as well as smaller differences in their
behavioral responses to enjoyment versus non-enjoyment
smiles (Slessor, Miles, Bull, & Phillips, 2010). Given that
OA are less able to perceive differences in the identity or
© The Author 2014. Published by Oxford University Press on behalf of The Gerontological Society of America.
All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
Received April 17, 2014; Accepted August 5, 2014
Decision Editor: Myra Fernandes, PhD
Page 1 of 10
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Method. We used a differentiation index adapted from previous work on stereotyping to assess OA and YA likelihood
of assigning different faces to different levels on trait scales. We computed scores for ratings of older and younger faces’
competence, health, hostility, and untrustworthiness.
Page 2 of 10
NG ET AL.
Face Age and Differentiation
Face age, like perceiver age, may influence the differentiation of trait impressions. Memory for positive first impressions is enhanced for self-similar others (Leshikar, Park, &
Gutchess, 2014), and the age similarity of faces moderates
age differences in face perception. More specifically, the
overall advantage of younger raters is attenuated or reversed
for older faces as shown in face recognition (Anastasi &
Rhodes, 2005; Fulton & Bartlett, 1991), age recognition
(Voelkle, Ebner, Lindenberger, & Riediger, 2011), emotion
recognition (Malatesta, Izard, Culver, & Nicolich, 1987),
and identifying perpetrators of crimes in lineups (Perfect &
Harris, 2003; Wright & Stroud, 2002). It is possible that such
an own-age bias also will be shown in the differentiation of
trait impressions from faces. Main effects for face age may
also exist. For example, both OA and YA may show greater
differentiation in trait impressions most relevant to the age of
the face. Since older people are more likely to actually vary
in health, differentiation of this trait may be greater for older
than younger faces. Since younger people are more likely
to show criminal behavior (Steffensmeier, Allan, Harer, &
Streifel, 1989), differentiation of traits such as hostile and
untrustworthy may be greater for younger than older faces.
Rater Age and Positivity
In addition to age differences in behavioral and neural
differentiation of visual stimuli, there are also differences
in responses to positive and negative stimuli. OA show
decreased memory and attention to negative stimuli and/or
increased memory and attention to positive stimuli, including faces (Carstensen & Mikels, 2005; Charles, Mather, &
Carstensen, 2003; Isaacowitz, Wadlinger, Goren, & Wilson,
2006). OA also give more positive trait and affect ratings to
faces than do YA (Castle et al., 2012; Ebner, 2008; Kellough
& Knight, 2012; Zebrowitz et al., 2013). Some evidence
also exists for an OA positivity effect at the neural level.
(Cassidy, Leshikar, Shih, Aizenman, & Gutchess, 2013;
Castle et al., 2012).
Whereas OA positivity effects have been attributed to
age-related changes in the motivation to avoid negative
stimuli, the present study tested the hypothesis that the more
positive trait impressions shown by OA are mediated by
dedifferentiation. More specifically, we examined whether
a tendency for OA to make fewer distinctions among faces
when judging their traits—for example to judge faces as
more similar to one another on dimensions such as trustworthiness or health—can account for the previously documented greater positivity of OA impressions (Zebrowitz
et al., 2013). Evidence for a link between differentiation
and positivity is provided by research demonstrating an
informational negativity effect: negative stimuli are more
cognitively elaborated than positive stimuli. For example,
evaluations of negative stimulus persons produce more
complex descriptions involving a mixture of both positive
and negative terms than do descriptions of positively valued
persons (Peeters & Czapinski, 1990). Also, compared with
negative trait adjectives, positive trait adjectives are more
similar in meaning and more strongly correlated—that is
less differentiated (Claeys & Timmers, 1993). (for reviews,
see Taylor, 1991; Rozin & Royzman, 2001).
Summary and Hypotheses
In summary, OA show decreased differentiation of faces
as evidenced by a lesser ability to recognize subtle variations in facial stimuli as well as deficits in recognizing
specific emotions and identities, effects that are paralleled
by age-related neural dedifferentiation. We extended these
findings to the differentiation of trait impressions from
faces, examining effects for both older and younger faces as
well as implications for the OA positivity effect. We tested
the following hypotheses
1. Previous evidence for poorer differentiation of
faces among OA would extend to less differentiated trait
impressions.
2. Previous evidence for an own-age bias in face recognition would extend to more differentiated impressions of
own-age faces.
3. Differentiation of impressions of health would be
greater for older than younger faces, while differentiation
of impressions of trustworthiness and hostility would be
greater for younger than older faces.
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
emotion expressions of faces, they may also be less likely
to perceive differences in their traits
Although this study focuses on behavioral indices of differentiation, there also is evidence for less neural differentiation in OA than YA, which may underlie the behavioral
effects and provides evidence that they do not merely reflect
age differences in response biases. Most relevant to the present research are age-related declines in neural indicators of
face differentiation. When OA and YA were asked to make
same-different judgments to face pairs that were identical,
moderate in similarity, or different, both OA and YA demonstrated a greater decrease in neural responsiveness, or
adaptation, in the fusiform face area in the identical as compared to the different face condition. However, OA, but not
YA, also showed more adaptation in the moderately similar
than the different face condition, indicating less neural sensitivity to subtle facial differences among OA (Goh, Suzuki,
& Park, 2010). Research also has found OA dedifferentiation in the ventral visual cortex using multi-voxel pattern
analysis (Carp, Park, Polk, & Park, 2011). These deficits in
neural differentiation of visual stimuli are only minimally
explained by changes in the anatomy of the visual pathway
(Spear, 1993). Instead, OA visual deficits may be accounted
for by neurological changes in the ventral visual cortex that
responds to faces and other categories of stimuli (Kanwisher,
McDermott, & Chun, 1997; Haxby et al., 2001).
AGE DIFFERENCES IN THE DIFFERENTIATION OF TRAIT IMPRESSIONS
4. Given evidence that positive judgments tend to be less
differentiated than negative ones, the less differentiated trait
impressions shown by OA would mediate their tendency to
report more positive impressions.
5. Finally, we conducted exploratory analyses to determine whether any observed age differences in trait differentiation could be explained by differences in mood, vision,
or cognitive function.
Method
Facial Stimuli
Three sets of facial images were used to assess the
generalizability of our results. One set was taken from a
Boston sample of individuals who participated in the study
of Midlife Development in the United States (MIDUS)
funded by the John D. and Catherine T. MacArthur
Foundation (Lachman, 1997). The images used were
color frontal facial photographs of people displaying neutral expressions and wearing a grey cape to mask their
clothing. A total of 69 YAs (41 men) between the ages
of 25 and 39 (M = 32.7, SD = 4.2), and 68 OAs (44 men)
between the ages of 60 and 74 (M = 66.1, SD = 4.1) were
included. Images were displayed at an approximate size
of 7.5 × 9.5 cm. Two additional sets of grayscale images
were taken from the Intergenerational Studies (IGS)
archive, a longitudinal study of representative samples
of individuals who were born in Berkeley, California, in
the late 1920s or attended school in Oakland, California,
in the 1930s (Eichorn, 1981). The images included two
photographs of 148 individuals (74 men). One photo was
taken when the person was between ages 17–18 and a
second when the person was between ages 52–62, for a
total of 296 images. To reduce the face rating task to a
reasonable length, the IGS images were divided into two
sets, each of which was rated by different participants.
Each set contained equal numbers of older and younger
male and female faces. In addition, the images in each set
were matched within demographic group on previously
collected attractiveness ratings by YA (for more information about the photographs and the ratings, see Zebrowitz,
Olson, & Hoffman, 1993).
Dependent Measures
Trait ratings.—The ratings used to calculate age differences in trait impression differentiation were taken from a
previous study investigating age differences in agreement
and positivity of trait impressions (Zebrowitz, et al., 2013).
Each participant rated faces from one of the three sets on
four 7-point scales with endpoints labeled (1) not at all
(healthy, competent, untrustworthy, aggressive or hostile)
to (7) very (healthy, competent, untrustworthy, aggressive
or hostile). MIDUS faces were rated on “aggressive” which
was replaced by “hostile” for participants rating IGS faces,
because ratings of the aggressiveness of the MIDUS faces
showed lower agreement among OA than YA. This suggested that OA may have varied in their interpretation of
the term, some construing it as “assertive” and others as
“hostile” The term hostile yielded comparable agreement
in impressions formed by OA and YA, and we therefore
included ratings of hostile in the differentiation analyses,
but not ratings of aggressiveness. Significant age differences in positivity were found for impressions of hostility,
untrustworthiness, and health, with OA rating faces as less
hostile, less untrustworthy, and more healthy, but not for
impressions of competence. (Appearance ratings of attractiveness and babyfaceness were made after the trait ratings,
but will not be discussed here as they are not relevant to our
hypotheses regarding trait differentiation.)
Differentiation scores.—Trait differentiation scores were
computed using the probability of differentiation (PD), where,
i is the level on the rating scale (1–7), and P is the proportion of ratings at the nth level (Linville, Salovey, & Fischer,
1986). This index captures the probability that a perceiver will
assign two randomly chosen individuals to different levels on
a rating scale. The more levels on a rating scale of a trait that
a perceiver uses when rating a series of faces, the higher the
differentiation score. This measure was initially developed to
investigate stereotype category differentiation (the ability to
perceive many types within a given category), and has also
been used to investigate differentiation of impressions of ownrace versus other-race faces to determine whether other-race
faces tend to “all look alike” (Linville et al., 1986; Zebrowitz,
Montepare, & Lee, 1993). We calculated each rater’s differentiation scores for older male, younger male, older female,
and younger female faces. These scores were calculated separately to control for the possibility that participants in one age
group could have higher differentiation scores simply because
their trait impressions showed greater differentiation among
faces that varied in sex or age. The final differentiation scores
for each participant averaged scores for older male and female
faces to create a score for all old faces, and scores for younger
male and female faces to create a score for all young faces.
One OA female participant was dropped from analyses on
healthy and untrustworthy ratings because her ratings did not
vary across faces.
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Participants
Forty-eight young adults (23 males) and 48 OA (24
males) participated in the study. YA were recruited from a
local university and were ages 18–22 (M = 18.8, SD = 1.0).
They completed the study for course credit or payment of
$15. OA, ages 65–86 (M = 76.3, SD = 6.4), were recruited
from the local community through fliers, paid $25 for completing the study and screened through the Mini-Mental
State Examination (Folstein et al., 1975) all scoring above
26 out of 30 (M = 28.9, SD = 1.2). Participants in each age
group viewed one of three sets of facial stimuli.
Page 3 of 10
Page 4 of 10
NG ET AL.
4 seconds (MIDUS faces) or 3 s (IGS faces) after which
the trait rating scale appeared and stayed on screen until
the participant made a rating. (IGS faces were rated after
data collection had been completed for MIDUS faces.
We reduced the exposure time both because MIDUS participants reported that the exposure duration was long and
also because there were more faces in the IGS sets, which
meant that reducing the exposure time kept the total rating duration more comparable across participants assigned
to each of the three face sets.) Participants were randomly
assigned to one of two trait orders: healthy, untrustworthy,
competent, and hostile or hostile, competent, untrustworthy, and healthy. Faces were shown in one of four orders,
counterbalancing age and sex of face, and participants were
asked to rate each face in comparison to the other faces of
that age/sex grouping. This procedure was employed so that
variations in trait impressions would be less likely to reflect
sex or age stereotypes, as we were interested in whether the
trait differentiation of OA was responsive to variations in
facial appearance within demographic groups. Participants
rated all faces on one trait rating scale before moving on to
the next scale. Following the trait ratings, participants completed the remaining control measures.
Procedure
Participants came into the laboratory and gave informed
consent. Next, they completed a computerized version of
the Positive and Negative Affect Schedule (PANAS). They
then rated either MIDUS faces or one of the two sets of
IGS faces, using MediaLab software (Empirisoft, New
York City, NY). Faces were shown individually for either
Overview of Analyses
Differentiation scores for each trait impression scale
were analyzed using 2 (rater age: OA vs. YA) × 2 (face age:
older vs. younger) × 3 (face set: MIDUS, IGS set 1, IGS
set 2) mixed-factor ANOVAs, where face age was a withinsubjects factor and rater age and face set were betweensubjects factors. Since face set effects have no theoretical
Results
Table 1. Older and Younger Adults’ Scores on Control Measures.
Older adults
Measure
PANAS (PA) (Watson, Clark, & Tellegen, 1988)
PANAS (NA) (Watson, Clark, & Tellegen, 1988)
Snellen Visual Acuity (denominator)
Mars Letter Contrast Sensitivity
(Mars Perceptrix, Chappaqua, NY)
Ishihara’s Test for Color Deficiency (Ishihara, 2010)
Benton Facial Recognition Test (Benton, Van Allen,
Hamsher, & Levin, 1983)
Timed Pattern Comparison Test (Salthouse, 1993)
Shipley Vocabulary Test (Shipley, 1946).
Wisconsin Card Sorting Test (the Berg Card Sort
Task (BCST validated by Piper et al., 2012; |
http://pebl.sourceforge.net/battery.html)
Letter-Number Sequencing Test
(Wechsler, 1997)
Educationa
Younger adults
M
SD
M
SD
t-value
p-value
2.90
1.73
34.69
1.56
.70
.67
14.27
.19
2.78
1.92
19.75
1.73
.84
.66
5.47
.06
.79
1.37
6.77
8.87
.434
.173
<.001
<.001
13.46
41.35
1.87
9.31
13.48
42.21
1.65
8.58
28.17
35.48
29.03
5.16
3.37
8.92
42.56
31.77
36.40
5.85
3.89
5.56
12.79
4.99
3.59
<.001
<.001
.001
8.19
1.52
12.56
2.80
5.49
<.001
5
2–7
3
.06
.468
.954
.641
2–3
Note. Medians and range are reported.
Level of Education was coded for highest level attained: 1 = no high school diploma, 2 = high school diploma, 3 = some college, 4 = Bachelor’s degree, 5 = some
graduate work, 6 = Masters degree, 7 = Doctorate degree.
a
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Control measures. Control measures of mood, vision,
and cognition were included to examine whether any of
these measures could explain the predicted age differences
in trait impression differentiation. Participants completed
control measures for assessing mood (PANAS; Watson,
Clark, & Tellegen, 1988), visual acuity (Snellen Eye Chart),
contrast sensitivity (Mars Letter Contrast Sensitivity Test,
Mars Perceptrix, Chappaqua, NY), color vision (Ishihara’s
Tests for Color Deficiency, Ishihara, 2010), facial matching (Benton Facial Recognition Test, Benton, Van Allen,
Hamsher, & Levin, 1983), crystallized intelligence
(Shipley’s Vocabulary Test, Shipley, 1946), and processing speed (Pattern Comparison Test, Salthouse, 1993).
To assess executive functioning, participants who rated
MIDUS faces completed a Letter-Number Sequencing Task
(LNST, Wechsler, 1997) and those who rated the IGS faces
completed a computerized version of the Wisconsin Card
Sort Task [Berg Card Sort Task (BCST); downloaded from
http://pebl.sourceforge.net/battery.html and validated by
Piper et al., 2012]. The significant age differences shown in
Table 1 are consistent with previous studies comparing YAs
with community dwelling OAs. Specifically, OA showed
poorer visual acuity and contrast sensitivity than YA as well
as slower processing speed and poorer executive function,
and OA showed better vocabulary scores than YA.
AGE DIFFERENCES IN THE DIFFERENTIATION OF TRAIT IMPRESSIONS
Page 5 of 10
meaning, we report them only when an effect involving a
variable of theoretical interest (rater age, face age) did not
hold true across all face sets, indicating a qualification to
generalizability. (Note that face set had only two levels for
hostile ratings, which were collected only for IGS faces.)
Differentiation as a Function of Face Age
Raters differentiated the health of older faces more than
that of younger faces, F(1,89) = 9.89 p = .002, ηP2 = .10.
They also differentiated the hostility of younger faces more
than that of older faces, F(1,60) = 17.54, p < .001, ηP2 = .23,
with a similar trend for untrustworthiness, F(1,60) = 2.75,
p = .101, ηP2 =.03 (See Figure 2).
The effect of face age on the differentiation of health
ratings was qualified by a significant face age × rater age
effect, F(1, 89) = 5.55, p = .021, ηP2 = .06. Planned comparisons revealed an own age bias in the differentiation of health
ratings by OA, who showed greater differentiation of older
faces (M = 0.70, SD = .11) than younger faces (M = .65,
SD = .15), p < .001. Differentiation of health ratings by YA
did not vary with face age (younger faces M = .73, SD = .08;
older faces M = .74, SD = .06), p = .58. The significant rater
age × face age × face set triple order interaction for hostility
ratings that qualified the rater age effect also qualified the
face age effect. The overall tendency for hostility ratings
to be more differentiated for younger than older faces was
significant for YA ratings of faces in IGS set 1, p = .021 and
OA ratings of faces in IGS set 2, p < .001, but not for YA
ratings of faces in IGS set 2, p = .32 or OA ratings of faces
in IGS set 1.
Control Variables as Mediators of Rater Age Differences
in Differentiation
The mediation analyses followed guidelines provided by
Kenny (2014): http://davidakenny.net/cm/mediate.htm#BK
together with the bootstrapping approach to obtain significant confidence intervals (Preacher & Hayes, 2004) with
Figure 1. Older and younger adults’ probability of differentiation of trait
impressions from faces. **p < .01, ***p < .001.
Figure 2. Probability of differentiation of trait impressions of older and
younger faces. **p < .01, ***p < .001.
the macro mediation script version 4.2 (2011) for SPSS
designed by Andrew F. Hayes (for more details see Preacher
& Hayes, 2008). According to Kenny (2014), necessary
conditions for mediation are that the causal variable be significantly related to the mediator (Step 2) and that the mediator be significantly related to the outcome variable with the
causal variable controlled (Step 3). He notes that James and
Brett (1984) have argued that Step 3 should be modified by
not controlling for the causal variable, because there would
be no need to control for the causal variable if there were
complete mediation. Because complete mediation does not
always occur, Kenny argues for controlling the causal variable in Step 3. On the other hand, he acknowledges that it
is reasonable to test for mediation even if there is no direct
effect of the causal variable on the outcome (Step 1) .
We considered control variables on which OA and YA
differed as potential mediators (Step 2). These included
visual acuity, contrast sensitivity, executive functioning
(BCST or LNST), processing speed (Pattern Comparison
Task), and vocabulary. Because the measure of executive function differed across participants, we calculated
z scores separately for those who completed the BCST
and those who completed the LNST and combined the
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Differentiation as a Function of Rater Age
As predicted, OA showed significantly less differentiation than YA for all trait impressions: competence,
F(1,90) = 7.79, p = .006, ηP2 = .08; health, F(1,89) = 8.32,
p = .005, ηP2 = .09; hostility, F(1,60) = 7.66, p = .007,
ηP2 = .11; untrustworthiness, F(1,89) =14.19, p < .001,
ηP2 = .14 (see Figure 1). The only qualification by face
set was a significant rater age × face age × face set triple
order interaction for hostility impressions, F(1,60) = 4.86,
p = .03, ηP2 = .08. Simple effects comparisons revealed
that OA hostility ratings showed significantly less differentiation than YA ratings for younger and older faces in IGS
set 2, respective ps = .01 and .004, with a trend in the same
direction for younger faces in IGS set 1, p = .12, but no rater
age difference for older faces in set 1, p =.62.
Page 6 of 10
NG ET AL.
Age Differences in Differentiation as a Mediator of OA
Positivity Effects
Previous research using these faces and raters found that
OA showed more positive impressions of hostility, trustworthiness, and health than did YA, with no age differences
in the positivity of impressions of competence (Zebrowitz
et al., 2013). We tested whether the OA positivity effects
for the first three traits were mediated by age differences
in trait differentiation using the same criteria for assessing
mediation described above. We also tested whether the nonsignificant OA positivity effect for competence ratings was
influenced by age differences in differentiation.
As shown in Figure 3, age differences in the differentiation of trait ratings mediated the greater positivity of OA
ratings. Specifically, the direct effect of rater age on the positivity of hostility, health, and untrustworthy impressions
lost significance when the age difference in differentiation
was added to the models as a mediator, and this change was
significant for health, 95% CI [0.04, 0.28], hostility, 95% CI
[−0.68, −0.14], and untrustworthy 95% CI [−0.38, −0.03]
ratings. Testing for mediation of competence ratings despite
no direct effect of age or significant effect of differentiation
at Step 3 revealed that controlling age differences in differentiation did not significantly change the direct effect, 95%
CI [−0.07, 0.15].
Although the preceding results support our hypothesis
that age differences in the differentiation of trait impressions would mediate age differences in the positivity of
trait impressions, the correlational nature of our data
leaves open the possibility that age differences in positivity mediate age differences in differentiation rather than
vice versa. Additional mediation tests revealed that the
criteria for this reverse mediation were satisfied for all
trait ratings except for competence, where rater age failed
to predict the tested mediator, impression positivity, p >
Figure 3. The mediation of trait rating positivity by differentiation scores
for ratings of healthy (1), hostile (2), and untrustworthy (3). Subject age is
coded 1 for younger raters and 2 for older raters. Numbers are unstandardized
beta-coefficients, with the standard errors shown in parentheses. The values
shown beneath the paths from rater age to the trait ratings represent the coefficients after differentiation scores were added to the model. *p < .05, ** p <
.01, ***p < .001.
.50. As shown in Figure 4, the effect of rater age on the
differentiation of hostility impressions lost significance
when the trait rating positivity mediator was added to
the model, a significant change, 95% CI [−0.11, −0.01].
In contrast, the effect of rater age on the differentiation
of healthy and untrustworthy ratings remained significant when trait rating positivity was added to the models, although the direct effects did lose strength, 95%
CI [−0.05, −0.004] for healthy (Figure 4), and 95% CI
[−0.05, −0.001] for untrustworthy (Figure 4). In summary, the same traits showed mediation of age differences
in impression positivity by differentiation scores and the
reverse mediation of age differences in differentiation
scores by impression positivity. However, the latter mediation did not eliminate the significant effect of rater age
on health or untrustworthy differentiation scores, whereas
rater age differences in the positivity of health, hostility,
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
values into a single variable. None of these control variables qualified as mediators for any of the four trait ratings using the criterion that they be significantly related
to differentiation scores with participant age controlled
(Step 3), all ps > .11. [Testing mediation despite a lack
of significance at Step 3 revealed that contrast sensitivity,
vocabulary, and processing speed did not mediate age differences in differentiation of any of the trait impressions,
with the 95% CI including zero for all four traits. On the
other hand, age differences in visual acuity significantly
suppressed the differences in hostility differentiation,
95% CI [.0014, .0621], and age differences in executive
function significantly suppressed the differences in competence differentiation, 95% CI [.0003, .0365] and health
differentiation, 95% CI [.0001, .0266]. Thus age differences in differentiation of these traits would have been
even greater had OA and YA not differed on these control
variables.].
AGE DIFFERENCES IN THE DIFFERENTIATION OF TRAIT IMPRESSIONS
and untrustworthy ratings all lost significance when differentiation was entered as a mediator.
Discussion
These findings add to previous research documenting
less differentiation of faces by OA than YA by investigating age differences in the differentiation of trait impressions
and their contribution to an OA positivity effect. Employing
a measure developed to investigate whether some groups
are stereotyped as highly homogeneous (Linville et al.,
1986), we found that OA impressions of hostility, untrustworthiness, health, and competence from faces showed less
differentiation than did YA impressions. We further found
greater differentiation of younger than older faces’ hostility
and older than younger faces’ health, with the latter effect
significant only for OA, consistent with an own-age bias.
Finally, we found that age differences in differentiation
mediated previously reported tendencies for OA to form
more positive impressions of the hostility, trustworthiness,
and health of the faces, with these OA positivity effects
losing significance when differentiation scores were controlled. Although the reverse mediation effects were also
significant, controlling impression positivity reduced, but
did not eliminate most age differences in differentiation.
OA lesser differentiation of faces on trait dimensions
is consistent with previous behavioral research demonstrating that OAs showed reduced accuracy in face recognition (Bartlett et al., 1989), required larger differences
between faces to discriminate them (Goh et al., 2010), and
also showed poorer recognition of emotional expressions
(Ruffman et al., 2008, 2009). Our evidence for dedifferentiation of trait impressions from faces in OA also is consistent with evidence for less neural differentiation (Carp et al.,
2011; Park et al., 2004). The convergence of our results
with this literature argues against the possibility that dedifferentiated trait impressions in OA merely reflect age differences in the use of rating scales rather than underlying age
differences in trait impressions. Research examining neural
correlates of the behavioral differences would corroborate
this conclusion. For example, one might predict that multivoxel pattern analysis in face processing and reward regions
would show less differentiation of faces varying in trustworthiness for OA than YA. The fact that age differences
in visual acuity and contrast sensitivity did not mediate age
differences in trait impression differentiation also supports
the argument that it is one manifestation of a broader tendency toward cognitive dedifferentiation in aging (Balinsky,
1941; Baltes et al., 1980; Salthouse et al., 1996) rather than
reflecting age-related changes on the anatomy of the visual
pathway (Spear, 1993).
The finding that the health of older faces was differentiated more than that of younger ones, with a reverse effect for
hostility, is consistent with our suggestion that people may
show greater differentiation on trait dimensions most relevant to the faces being judged. Health is arguably an evaluative dimension more relevant to older individuals, while
hostility is more relevant to younger ones, who are more
likely to engage in violent criminal behavior (Steffensmeier,
et al., 1989). However, on this account, only OA perceived
health to be more relevant to older than younger faces, since
differentiation of the two age groups by YA did not differ
significantly. Research that manipulates the relevance of a
trait dimension for older and younger faces would provide
insight into the face age effects and their moderation by
rater age.
Although the evidence for an own-age bias in the differentiation of trait impressions was limited to OA impressions
of health, this finding adds to a literature that has shown
similar effects in other social judgments, including face
recognition (Anastasi & Rhodes, 2005; Fulton & Bartlett,
1991), age recognition (Voelkle, et al., 2011), emotion recognition (Malatesta, et al., 1987), and identifying perpetrators of crimes in lineups (Perfect & Harris, 2003; Wright &
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Figure 4. The mediation of differentiation scores by trait rating positivity
for healthy (1), hostile (2), and untrustworthy (3). Numbers are unstandardized
beta-coefficients, with the standard errors shown in parentheses. The values
shown beneath the paths from rater age to the trait differentiation scores represent the coefficients after trait rating positivity was added to the model. *p <
.05, ** p < .01, ***p < .001.
Page 7 of 10
Page 8 of 10
NG ET AL.
Conclusions
We found that previous evidence for less perceptual differentiation in OA, as indexed by behavioral measures of
face and emotion recognition as well as neural measures,
extends to less differentiation of trait impressions from
faces. These results could not be explained by age differences in mood, vision, processing speed, vocabulary, or
executive function. We also found that differentiation of
the trait impressions of younger and older faces varied with
the relevance of the trait to faces of that age as well as the
relevance of the faces’ age to the perceiver. Finally, age differences in the differentiation of trait ratings mediated the
tendency for OA to show more positive impressions of faces
than YA. It would be useful for future research to investigate
the neural mechanisms associated with age differences in
trait impression differentiation in order to provide a closer
link between our findings and the dedifferentiation hypothesis of cognitive aging (Balinsky, 1941; Baltes, et al., 1980;
Salthouse et al., 1996).
Funding
This research was supported by NIH grant AG038375 to the first author.
Acknowledgments
We express appreciation to the Institute of Human Development
at the University of California, Berkeley for providing access to the
Intergenerational Studies (IGS) facial photographs employed in this study.
These photographs are now housed at the Henry A. Murray Research
Archive, Harvard University, Cambridge, MA.
Correspondence
Correspondence should be addressed to Leslie A. Zebrowitz, PhD,
Department of Psychology, MS 062, Waltham, MA, 02454. E-mail: zebrowitz@brandeis.edu.
References
Anastasi, J. S. & Rhodes, M. G. (2005). An own-age bias in face recognition for children and older adults. Psychonomic Bulletin & Review,
12, 1043–1047. doi:10.3758/BF03206441
Balinsky, B. (1941). An analysis of the mental factors of various age groups
from nine to sixty. Genetic Psychology Monographs, 23, 191–234.
Baltes, P.B., Cornelius, S.W., Spiro, A., Nesselroade, J.R., & Willis,
S.L. (1980). Integration versus differentiation of fluid-crystallized
intelligence in old age. Developmental Psychology, 16, 625–635.
doi:10.1037/0012-1649.16.6.625
Bar, M., Neta, M., & Linz, H. (2006). Very first impressions. Emotion, 6,
269–278. doi:10.1037/1528-3542.6.2.269
Bartlett, J. C., & Leslie, J. E. (1986). Aging and memory for faces versus single views of faces. Memory & Cognition, 14, 371–381.
doi:10.3758/BF03197012
Bartlett, J. C., Leslie, J. E., Tubbs, A., & Fulton, A. (1989). Aging and
memory for pictures of faces. Psychology and Aging, 4, 276–283.
doi:10.1037/0882-7974.4.3.276
Betts, L. R., Sekuler, A. B., & Bennett, P. J. (2007). The effects of aging
on orientation discrimination. Vision Research, 47, 1769–1780.
doi:10.1016/j.visres.2007.02.016
Benton, A., Van Allen, M., Hamsher, K., & Levin, H. (1983). Test of
facial recognition manual. Iowa City: Benton Laboratory of
Neuropsychology.
Carp, J., Park, J., Polk, T. A., & Park, D. C. (2011). Age differences in
neural distinctiveness revealed by multi-voxel pattern analysis.
NeuroImage, 56, 736–743. doi:10.1016/j.neuroimage.2010.04.267
Carstensen, L. L., Isaacowitz, D. M., & Charles, S. T. (1999). Taking time
seriously. A theory of socioemotional selectivity. The American
Psychologist, 54, 165–181. doi:org/10.1111/j.0963-7214.2005.00348.x
Carstensen, L. L. (1995). Evidence for a life-span theory of socioemotional selectivity. Current Directions in Psychological Science, 4,
151–156. doi:10.1111/1467-8721.ep11512261
Carstensen, L. L. & Mikels, J. A. (2005). At the intersection of emotion
and cognition. Current Directions in Psychological Science, 14,
117–121. doi:10.1111/j.0963-7214.2005.00348.x
Cassidy, B. S., Leshikar, E. D., Shih, J. Y., Aizenman, A., & Gutchess, A.
H. (2013). Valence-based age differences in medial prefrontal activity during impression formation. Social Neuroscience, 8, 462–473.
doi:10.1080/17470919.2013.832373
Castle, E., Eisenberger, N. I., Seeman, T. E., Moons, W. G., Boggero, I.
A., Grinblatt, M. S., & Taylor, S. E. (2012). Neural and behavioral
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Stroud, 2002). Our limited support for an own age bias may
be due to the fact that “own-age” faces were not exactly
matched in age to participants, although they were closer in
age than the “other-age” faces. Thus, greater differentiation
of own-age faces may be shown in other trait impressions
when faces more closely match the age of perceivers.
We examined trait differentiation across three sets of faces
to assess the generalizability of our results. The tendency
for OA to show less differentiation than YA in their impressions of competence, health, and untrustworthiness did not
vary with face set. However, the lesser differentiation of OA
than YA hostility impressions was not significant for faces
in one of the two sets for which this trait was rated. Also, the
greater differentiation of hostility ratings for younger than
older faces was not significant for YA ratings of faces in one
set and for OA ratings of faces in the other set. Although we
have no theoretical explanation for differences across face
sets, it is noteworthy that the face set that failed to show
less differentiation of hostility impressions by OA than YA
in the present study also failed to show greater positivity of
these impressions by OA than YA (Zebrowitz et al., 2013).
This result is consistent with the links between impression
differentiation and impression positivity discussed below.
Our finding that age differences in the differentiation
of trait impressions from faces mediated the tendency for
OA to form more positive impressions provides a novel
explanation for OA positivity effects. In contrast to a motivational explanation whereby OA focus on positive information in the service of emotion regulation (Carstensen,
1995; Carstensen & Mikels, 2005), our results indicate that
OA positivity effects can reflect a failure to make fine distinctions among stimuli, consistent with previous evidence
that positive judgments are less differentiated than negative ones (Taylor, 1991; Rozin & Royzman, 2001). At the
same time, reverse mediation effects also showed that age
differences in the positivity of trait impressions mediated
the differences in differentiation. Although the latter effects
tended to show partial rather than full mediation, experimental research is needed to manipulate differentiation or
positivity of impressions in order to demonstrate a causal
influence of one on the other.
AGE DIFFERENCES IN THE DIFFERENTIATION OF TRAIT IMPRESSIONS
Malatesta, C. Z., Izard, C. E., Culver, C., & Nicolich, M. (1987). Emotion
communication skills in young, middle-aged, and older women.
Psychological Aging, 2, 193–203. doi:10.1037/0882-7974.2.2.193
Murphy, N. A. & Isaacowitz, D. M. (2008). Preferences for emotional information in older and younger adults: a meta-analysis of
memory and attention tasks. Psychology and Aging, 23, 263–286.
doi:10.1037/0882-7974.23.2.263
Ortega, V. & Phillips, L. H. (2008). Effects of age and emotional intensity
on the recognition of facial emotion. Experimental Aging Research,
34, 63–79. doi:10.1080/03610730701762047
Park, D. C., Polk, T. A., Park, R., Minear, M., Savage, A., & Smith,
M. R. (2004). Aging reduces neural specialization in the ventral
visual cortex. Proceedings of the National Academy of Sciences
of the United States of America, 101, 13091–13095. doi:10.1073/
pnas.0405148101
Peeters, G. & Czapinski, J. (1990). Positive-negative asymmetry in evaluations: The distinction between affective and informational effects.
In W. Stroebe & M. Hewstone (Eds.), European Review of Social
Psychology, vol. 1 (pp. 33–60). Wiley: Chichester.
Perfect, T. J. & Harris, L. J. (2003). Adult age differences in unconscious
transference: Source confusion or identity blending? Memory &
Cognition, 31, 570–580. doi:10.3758/BF03196098
Piper, J., Li, V., Eiwaz, M., Kobel, Y., Benice, T., Chu, A., & Meuller, S.
(2012). Executive function on the psychology experiment building language tests. Behavior Research Methods, 44, 110–123.
doi:10.3758/s13428-011-0096-6
Preacher, K. J. & Hayes, A. F. (2004). SPSS and SAS procedures for
estimating indirect effects in simple mediation models. Behavior
Research Methods, Instruments, & Computers, 36, 717–731.
doi:10.3758/BF03206553
Preacher, K. J. & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891. doi:10.3758/
BRM.40.3.879
Rozin, P., & Royzman, E. B. (2001). Negativity bias, negativity dominance, and contagion. Personality and Social Psychology Review, 5,
296–320. doi:10.1207/S15327957PSPR0504_2
Ruffman, T., Halberstadt, J., & Murray, J. (2009). Recognition of facial,
auditory, and bodily emotions in older adults Journal of Gerontology
B: Psychological Sciences and Social Sciences. 64B, 696–703.
doi:10.1093/geronb/gbp072
Ruffman, T., Henry, J. D., Livingstone, V., & Phillips, L. H. (2008).
A meta-analytic review of emotion recognition and aging:
Implications for neuropsychological models of aging. Neuroscience
and Biobehavioral Reviews, 32, 863–881. doi:10.1016/j.
neubiorev.2008.01.001
Ruffman, T., Sullivan, S., & Edge, N. (2006). Differences in the way older
and younger adults rate threat in faces but not situations. Journals of
Gerontology, Series B: Psychological Sciences and Social Sciences,
61, 187–194. doi:10.1093/geronb/61.4.P187
Salthouse, T. A., Hancock, H. E., Meinz, E. J., & Hambrick, D.
Z. (1996). Interrelations of age, visual acuity, and cognitive functioning. The Journals of Gerontology. Series B,
Psychological Sciences and Social Sciences, 51, P317–P330.
doi:10.1037/0033-295X.103.3.403
Salthouse, T. A. (1993). Speed and knowledge as determinants of
adult age differences in verbal tasks. Journal of Gerontology, 48,
29–36.
Shipley, W. C. (1946). Institute of living scale. Los Angeles, CA: Western
Psychological Services.
Slessor, G., Miles, L. K., Bull, R., & Phillips, L. H. (2010). Age-related
changes in detecting happiness: Discriminating between enjoyment
and non-enjoyment smiles. Psychology and Aging, 25, 246–250.
doi:10.1037/a0018248
Spear, P. D. (1993). Neural bases of visual deficits during aging. Vision
Research, 33, 2589–2609. doi:10.1016/0042-6989(93)90218-L
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
bases of age differences in perceptions of trust. Proceedings of the
National Academy of Sciences of the United States of America.
doi:10.1073/pnas.1218518109
Charles, S. T., Mather, M., & Carstensen, L. L. (2003). Aging and emotional memory: The forgettable nature of negative images for older
adults. Journal of Experimental Psychology: General, 132, 310–324.
doi:10.1037/0096-3445.132.2.310
Claeys, W., & Timmers, L. (1993). Some instantiations of the informational
negativity effect: positive-negative asymmetry in category breadth and
in the estimated meaning similarity of trait adjectives. European Journal
of Social Psychology, 23, 111–129. doi:10.1002/ejsp.2420230202
Ebner, N. C. (2008). Age of face matters: Age-group differences in ratings
of young and old faces. Behavior Research Methods, 40, 130–136.
doi:10.3758/BRM.40.1.130
Eichorn, D. H. (1981). Present and past in middle life. New York:
Academic Press.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). “Mini-mental
state”. A practical method for grading the cognitive state of patients
for the clinician. Journal of Psychiatric Research, 12, 189–198.
doi:10.1016/0022-3956(75)90026-6
Fulton, A. & Bartlett, J. C. (1991). Young and old faces in young and old
heads: The factor of age in face recognition. Psychology and Aging,
6, 623–630. doi:10.1037/0882-7974.6.4.623
Goh, J. O., Suzuki, A., & Park, D. C. (2010). Reduced neural selectivity increases fMRI adaptation with age during face discrimination.
NeuroImage, 51, 336–344. doi:10.1016/j.neuroimage.2010.01.107
Grady, C. L., Maisog, J. M., Horwitz, B., Ungerleider, L. G., Mentis, M. J.,
Salerno, J. A. … Haxby, J. V. (1994). Age-related changes in cortical
blood flow activation during visual processing of faces and location.
Journal of Neuroscience, 14, 1450–1462.
Haxby, J. V., Govvini, M. I., Furey, M. L., Ishai, A., Schouten, J. L. &
Pietrini, P. (2001). Distributed and overlapping representations of
faces and objects in ventral temporal cortex. Science, 293, 2425–
2430. doi:10.1126/science.1063736
Isaacowitz, D. M., Lockenhoff, C. E., Lane, R. D., Wright, R., Sechrest,
L., Riedel, R., Costa, P. T. (2007). Age differences in recognition
of emotion in lexical stimuli and facial expressions. Psychology and
Aging, 22, 147–159. doi:10.1037/0882-7974.22.1.147
Isaacowitz, D. M., Wadlinger, H. A., Goren, D., & Wilson, H. R.
(2006). Is there an age-related positivity effect in visual attention? A comparison of two methodologies. Emotion, 6, 511–516.
doi:10.1037/1528-3542.6.3.511
Ishihara, S. (2010). Test for colour-blindness. Tokyo, Japan: Kanehara
Trading Company. (Original work published 1917).
James, L. R. & Brett, J. M. (1984). Mediators, moderators, and tests
for mediation. Journal of Applied Psychology, 69, 307–321. doi:
10.1037/0021-9010.69.2.307
Kanwisher, N., McDermott, J., & Chun, M. M. (1997). The fusiform
face area: a module in human extrastriate cortex specialized for
face perception. Journal of Neuroscience, 17, 4302–4311. PMID:
9151747.
Kellough, J. L., & Knight, B.G. (2012). Positivity effects in older adults’
perception of facial emotion: The role of future time perspective, Journal of Gerontology B: Psychological Sciences and Social
Sciences, 67B, 150–158. doi:10.1093/geronb/gbr079
Kenny, D. A. (2014). Mediation. Retrieved from http://davidakenny.net/
cm/mediate.htm#BK
Leshikar, E. D., Park, J. M., & Gutchess, A. H. (2014). Similarity to the
self affects memory for impressions of others in younger and older
adults Journal of Gerontology B: Psychological Sciences and Social
Sciences. Advance online publication. doi:10.1093/geronb/gbt132
Linville, P. W., Salovey, P., & Fischer, G. W. (1986). Stereotyping and
perceived distributions of social characteristics: an application to
ingroup-outgroup perception. In J. F. Dovidio & S. L. Gaertner
(Eds.), Prejudice, discrimination, and racism (pp. 165–208). San
Diego, CA: Academic Press.
Page 9 of 10
Page 10 of 10
NG ET AL.
Willis, J., & Todorov, A. (2006). First impressions: Making up your
mind after a 100-ms exposure to a face. Psychological Science, 17,
592–598. doi:10.1111/j.1467-9280.2006.01750.x
Wright, D. B. & Stroud, J. N. (2002). Age differences in lineup identification accuracy: People are better with their own age. Law and
Human Behavior, 26, 641–654. doi:10.1023/A:1020981501383
Zebrowitz, L. A., Franklin, R. G., Hillman, S., & Boc, H. (2013). Older and
younger adults’ first impressions from faces: similar in agreement, different
in positivity. Psychology and Aging, 28, 202–212. doi:10.1037/a0030927
Zebrowitz, L. A., Montepare, J. M., & Lee, H. K. (1993). They don’t all
look alike: individuated impressions of other racial groups. Journal of
Personality and Social Psychology, 65, 85–101. doi:10.1037/a0030927
Zebrowitz, L. A., Olson, K., & Hoffman, K. (1993). Stability of babyfaceness and attractiveness across the life span. Journal of Personality and
Social Psychology, 64, 453–466. doi:10.1037//0022-3514.64.3.453
Zebrowitz, L. A., Wang, R., Bronstad, P. M., Eisenberg, D., Undurraga, E.,
Reyes-Garcia, V., & Godoy, R. (2012). First impressions from faces
among U.S. and culturally isolated tsimane’ people in the Bolivian
Rainforest. Journal of Cross-Cultural Psychology, 43, 119–134.
doi:10.1177/0022022111411386
Downloaded from http://psychsocgerontology.oxfordjournals.org/ by guest on September 9, 2014
Steffensmeier, D.J., Allan, E. A, Harer, M. D., & Streifel, C. (1989). Age and
the distribution of crime. American Journal of Sociology. 94, 803–831.
doi:10.1086/229069
Taylor, S. E. (1991). Asymmetrical effects of positive and negative events:
the mobilization-minimization hypothesis. Psychological Bulletin,
110, 67–85. doi:10.1037/0033-2909.110.1.67
Toner, C.K., Pirogovsky, E., Kirwan, C.B., & Gilbert, P.E. (2009).
Visual object pattern separation deficits in nondemented older
adults. Learning and Memory, 16, 338–342. doi:10.1101/
lm.1315109
Voelkle, M. C., Ebner, N. C., Lindenberger, U., & Riediger, M. (2012). Let
meguess how old you are: Effects of age, gender, and facial expression on perceptions of age. Psychology and Aging, 27, 265–277.
doi:10.1037/a0025065
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS
scales. Journal of Personality and Social Psychology, 54, 1063–
1070. doi:10.1037/0022-3514.54.6.1063
Wechsler, D. (1997). Wechsler Adult Intelligence Scale-III (WAIS-III)
manual. New York, NY: The Psychological Corporation.