Reversed facial images and the mere-exposure hypothesis

Abstract

Tested the mere-exposure hypothesis with 33 female undergraduates, each of whom was accompanied by a close female friend, in a context in which the plausibility of a demand-characteristics explanation was minimized. It was assumed that a person has more frequently been exposed to his or her mirror image than true image, whereas a person's friend has more frequently been exposed to the person's true image than mirror image. According to the mere-exposure hypothesis, the S should prefer her mirror image, whereas the S's friend should prefer the S's true image. A single frontal facial photograph of each S was printed in such a way that one print corresponded to the S's true image and another to her mirror image. In 2 studies, Ss were found reliably to prefer their mirror image over their true image, whereas the reverse tendency characterized preferences of Ss' friends. (PsycINFO Database Record (c) 2012 APA, all rights reserved)

Full text

Journal of Personality and Social Psychology
1977, Vol. 35, No. 8, 597-601
Reversed Facial Images and the Mere-Exposure Hypothesis
Theodore H. Mita, Marshall Dermer, and Jeffrey Knight
University of Wisconsin—Milwaukee
The mere-exposure hypothesis was tested in a new context in which the plausi-
bility of a demand-characteristics explanation was minimized. It was assumed
that Person has more frequently been exposed to her mirror image than true
image, whereas Person's Friend or Lover has more frequently been exposed to
Person's true image than mirror image. According to the mere-exposure hypoth-
esis,
Person should prefer her mirror image, whereas her Friend or Lover should
prefer Person's true image. A single frontal facial photograph of Person was
printed in such a way that one print corresponded to Person's true image and
another to her mirror image. In two studies, Person was found reliably to prefer
her mirror image over her true image, whereas the reverse tendency character-
ized preferences of Person's Friend or Lover.
The "mere-exposure" hypothesis (Zajonc, than other stimuli, suggesting that partici-
1968),
that the mere repeated exposure of an pants' intuition could possibly account for
individual to a stimulus is a sufficient condi- the exposure-affect relationship found in the
tion for the enhancement of stimulus evalu- other studies. Stang suggested that the vari-
ation, has been studied extensively. In the ables investigated be more carefully embedded
basic within-participants design, individuals in plausible cover stories and, at the very
initially view unfamiliar stimuli, such as non- least, that postexperimental inquiries regard-
sense words, ideographs, or facial photographs, ing demand characteristics be made. More-
a varying number of times and subsequently land and Zajonc (1976) recently addressed
rate these materials along either good/bad or the demand issue in a study in which ex-
like/dislike dimensions. Typically, the stimuli posure frequency was a between-partidpants
that are seen more frequently receive more manipulation. Participants appeared to have
favorable evaluations, in accord with the ex- been unaware of the experimental hypothesis
posure hypothesis. as revealed by postexperimental inquiries,
A serious problem in nearly all of the and the mere-exposure hypothesis was cor-
studies involving the experimental manipula- roborated.
tion of exposure frequency has been that of Our intent in the present study was to test
demand characteristics (Saegert, Swap, & Za- the mere-exposure hypothesis so that there
jonc,
1973; Stang, 1974). Stang (1974) re- was virtually no possibility of sensitizing par-
ported that participants' affective ratings of ticipants to the frequency-affect hypothesis,
stimuli increased even when they simply Thus, like Moreland and Zajonc (1976) we
imagined having seen them more frequently wished to render demand explanations im-
plausible. Unlike these investigators, however,
we chose to utilize an entirely new paradigm
to probe the generality of mere-exposure
ecls-
We assumed that individuals are more
This research was partially supported by a Faculty «__»,.
Research Grant from the graduate school of the enecls-
University of Wisconsin—Milwaukee to Marshall We
termer. Special thanks are due Alan Baron for his likely to be exposed repeatedly to their mir-
SomUlnUggeStir
regafntg
"V*"011' ^ ror facial image than to their true facial
sontz Dernier for conducting the testing sessions lv" B
associated with the replication, and Robert B. Zajonc image. We therefore reasoned that it tne
'°r reviewing an earlier draft of this report. mere-exposure hypothesis is true, individuals
Requests for reprints should be sent to Marshall . { j j hotograph fl^
corre-
"enner, Department of Psychology, University of r " ^ . , .<
Wisconsin—Milwaukee, MUwaukee, Wisconsin
S3201.
sponds to their mirror image rather tnan to
597

598T. MITA,
M.
DERMER, AND
J.
KNIGHT
Table
1
Preferences
of
Targets
and
Friends
for True
and Mirror Prints (Study
1)
Friend's
preference
True print
Mirror print
Column total
Target's
pref-
erence
True
print
5
7
12
Mirror
print
IS
6
21
Row total
20
13
33
their true image. Analogously,
we
reasoned
that
a
close friend would have been exposed
more often
to the
true than
to the
mirror
image
and
would therefore prefer
a
photo-
graph that corresponds
to the
true image.
Our
methodology simply involved photographing
individuals, printing mirror
and
true images,
and assessing
the
preferences
of the
photo-
graphic subjects
and
their close friends
for
the
two
prints.
Study
1
Method
Participants. Thirty-seven women enrolled in un-
dergraduate psychology courses at the University of
Wisconsin—Milwaukee earned experimental credit
for participation in a "self-perception" study and
were designated targets. These women were each
asked to bring a "close female friend living in the
Milwaukee area" to the study, and the women in
this latter group were designated friends. Four tar-
gets did not comply with this request; thus 33 dyads
were studied
Procedure. In the first session, each target sat for
two frontal facial exposures. Only one of the nega-
tives was printed so that the two resulting (25 X 20
cm, black and white, glossy) photographs were mir-
ror images of each other The print corresponding to
how the target appears to herself when looking in
the mirror was designated the mirror print; the re-
maining print, corresponding to how she generally
appears to others, was designated the true print.
Target and friend arrived together for the second
session but were tested individually. The target was
tested first, while her friend waited in another room,
and then the situation was immediately reversed to
minimize dyadic communication.
Participants were seated in a well-lit room at a
table upon which the photographs were displayed.
Left and right positioning of the photographs was
randomized for each participant, who was asked to
view both photographs and to indicate which photo-
graph she "like(d) better." The experimenter did not
explain what was meant by the phrase, "like better,'
but merely repeated the instruction if elaboration
was requested. In the event a participant felt that
the photographs were equally likeable, she was asked
to express a preference even though it might reflect
a minute difference.
After the participant had responded, she was
asked to state her reasons for making her choice in
order to ascertain how well she had understood the
instructions and to assess demand characteristics
She was then asked what she thought the purpose
of the experiment was. All responses were recorded
by the experimenter. Participants were not informed
about the purpose of the study until the preferences
of all dyads had been assessed.
Results
and
Discussion
The joint prediction that targets would
pre
fer
the
mirror print
and
friends would favor
the true print
was
confirmed.
As
depicted
in
Table
1, IS of the 33 (.45)
dyads performed
in this fashion
(p <
.01,
binomial [33,1/4]).!
The column
and row
marginals reflect,
re-
spectively, targets'
and
friends' individual
preferences
for the two
prints. Although
the
proportion
of
targets preferring
the
mirror
print
was .64 (21 of 33),
this statistic
did
not differ from chance
at
conventional levels
of significance
(p <
.08, binomial
[33, 1/2])
When, however,
the
data
of the 4
targets who
did
not
bring friends were added,
the
propor-
tion
of
times
the
mirror print
was
preferred
rose
to .68 and
reliably differed from chance
(p
< .02,
binomial
[37,
1/2]).
Although
the
proportion
of
friends preferring
the
true print
was
.61
(20/33),
in the
predicted direction,
this statistic
did not
differ from chance
(p
<
.15,
binomial
[33,
1/2]).
While these results were encouraging,
w:
thought
it
best
to
replicate
the
study, given
the novelty
of the
paradigm.
In
particular,
the friend's preference appeared weaker,
and
we sought
a
more powerful testing procedure
We therefore revised
our
procedure
in tv
ways.
First,
we
reasoned that
if any
individ-
ual spends
a
great deal
of
time viewing
'
woman's face,
it is her
lover. Therefore,
*•
i All p values are one-tailed. For this binoini'
test there were 33 participant pairs, and the prob-
ability of "success" was 1/4. In this case, succes-'
was the target's preferring her mirror print and hfl
friend's preferring the true print.

REVERSED FACIAL IMAGES
AND
MERE EXPOSURE599
recruited only women
who
believed they were
dating individuals
who
were
in
love with
them. Secondly,
in
order better
to
assess
pref-
erences,
we
asked participants
to
choose which
of
the two
prints they preferred over
a
series
of five trials. Finally,
it
should
be
noted that
in Study
1 the
experimenter knew which
photographs were
the
mirror prints.
In
order
to reduce
the
possibility
of
experimenter bias
in
the
replication,
the
experimenter
was not
told which were
the
mirror prints.
Study
2
Method
Participants. Thirty-eight women enrolled
in sum-
mer session courses
at the
University
of
Wisconsin—
Milwaukee, each
of
whom
had
indicated that
she
was dating
or
living with
a
person
who she
believed
was
in
love with
her,
were designated targets
As
before, they were invited
to
participate
in a "self-
perception" study
and
were asked
to
bring their
lovers—who were men—to
the
second session.
Ten
targets
did not
comply with
the
latter request,
pri-
marily because their lovers were
out of
town.
Procedure.
In the
first session, each target
sat for
three frontal facial exposures.
As
before, only
one
of
the
negatives
was
printed
to
generate
a
mirror
and
a
true print.
The testing procedure
was
essentially identical
to
that used
in
Study
1,
with
the
exception that
the
two prints were presented
for
judgment over five
trials.
Participants were asked
to
turn away from
the viewing table while
the
next photographs were
being displayed.
The
same
two
prints were,
of
course, presented across trials with left
and
right
positioning randomized.
After
the
five trials
had
been completed, partici-
pants were interviewed
as in the
first study, with
their responses
(in
this study) tape recorded.
Par-
ticipants were again
not
informed about
the
purpose
of
the
study until
all
dyads
had
been tested.
Table
2
Preferences
of
Targets
and
Lovers
for
True
and Mirror Prints
on
First Trial (Study
2)
Lover's
preference
True print
Mirror print
Column total
Target's
pref-
erence
True
print
5
3
8
Mirror
print
12
8
20
Row total
17
11
28
Table
3
Modal Preferences (Across Five Trials)
of
Targets
and
Lovers
for
True
and
Mirror
Prints (Study
2)
Lover's
preference
True print
Mirror print
Column total
Target's
pref-
erence
True Mirror
print print
5
14
3
6
8
20
Row total
19
9
28
Results
and
Discussion
Participants' responses
on the
first trial
con-
stitute
a
nearly exact replication
of the
first
study. Furthermore, less than
45
seconds
usually elapsed
(in
both studies) between
the
time
at
which
a
participant received instruc-
tions
and the
time
at
which
she
indicated
her
first preference. This duration would hardly
seem adequate
for a
participant
to
formulate
the experimental hypothesis
and
devise
an
appropriate response,
as per a
demand inter-
pretation. Preferences observed
on the
first
trial
are,
therefore,
of
considerable interest.
The responses
for the
first trial,
for the 28
women
who
brought
a
lover
to the
second
session,
are
presented
in
Table
2. The
joint
prediction
was
again corroborated
for 12 of
the
28 (.43)
dyads
(p < .025,
binomial
[28,
1/4]). Of the 28
targets represented
in
Table
2,
20 (.71)
preferred
the
mirror
to the
true
print
(p < .02,
binomial
[28, 1/2]).
This
preference
was
diminished somewhat,
26 of
38 (.68), when
the
preferences
of the 10
women
who did not
bring their lovers were
included
(p < .018,
binomial
[38, 1/2]). Al-
though
17 of the 28
lovers
(.61)
preferred
the true print, this preference
was not
reliable
(p<
.17,
binomial
[28, 1/2]).
In summary,
the
pattern
of
results
for
preferences
on the
first trial appears virtually
identical
to
that observed
in the
first study.
The joint
and
target preference predictions
were reliably corroborated, whereas only
a
tendency consistent with
the
exposure
hy-
pothesis
was
noted
for the
friend/lover
pre-
diction.

600T. MITA,
M.
DERMER,
AND J.
KNIGHT
Modal preferences across
the
entire five-
trial sequence
are
presented
in
Table
3.
Once
again,
the
joint prediction that
the
target
would prefer
the
mirror print
and her
lover
would favor
the
true print
was
strongly
con-
firmed. As indicated
in
Table
3, 14 of the 28
dyads
(.50)
performed
in
this fashion
(p <
.003,
binomial
[28, 1/4]). As is
evident from
the marginals,
the
proportion
of
targets
who
preferred
the
mirror print
was .71 (20 of 28),
which
is
significant
(p < .02,
binomial
[28,
1/2]).
This preference
is
diminished slightly,
.68
(26 of 38), but is
reliable
(p < .02, bi-
nomial
[38, 1/2]),
when
the
data
of the 10
targets
who
could
not
bring their lovers
are
included. Finally,
as is
illustrated
in
Table
3,
19
of the 28 (.68)
lovers reliably preferred
the true print
(p < .04,
binomial
[28, 1/2]).
It would appear that
the
multiple assessment
procedure
is the
basis
for our
finally detect-
ing
a
reliable lover-preference effect.
The overall pattern
of
findings across
the
two studies
is
consistent with
the
mere-expo-
sure hypothesis. Furthermore,
it is
important
to indicate that there appears
to be
virtually
no evidence
for a
demand interpretation.
As
noted,
a
demand explanation appears particu-
larly implausible
for
initial preferences.
Al-
though participants
had as
much
as 2
weeks
after they were photographed
to
think about
the study,
it
appears unlikely that they
or
their friends could divine
the
experimental
hypothesis
on the
basis
of the
photographic
session.
In attempting
to
assess demand after
pref-
erences
had
been collected,
we
first asked
participants
to
indicate
the
basis
for
their
selections.
In the
first study,
for
example,
participants mentioned "more natural,"
"bet-
ter head tilt," "better facial angle," "better
eyes,"
"longer hair," "straighter part,"
"snicker,"
and
"looks less mean."
It
almost
appears that participants were sometimes
in-
venting differences, because
the
prints were
made from
the
same negative.
At any
rate,
it
does
not
appear that anyone came
at all
close
to
the
experimental hypothesis
on the
basis
of
the initial inquiry
in
either study.
We were then more direct
and
asked
par-
ticipants
to
state
the
purpose
of the
experi-
ment.
In the
first study, nine participants
said that
the
study
was set up to
compare
how they perceive themselves with
how
their
friends perceive them. Forty participants said
that they
did not
know (this
may
have been
due
to
other demand characteristics,
or it
may have indicated that
the
hypothesis
was
not blatantly apparent).
Two of the
friends
said "visual perception."
Two of the
targets
said "self-perception," which,
of
course,
was
the description
we
provided.
Two
participants
said "comparing self-perception with
the way
the pictures
are."
Perhaps
if the
participants
had been required
to
generate formal hypoth-
eses,
it
would
be
clearer
to
what extent they
truly
did not
know what
the
study
was
about
and
to
what extent they were simply
un-
interested
in
what
was
going
on.
Again, none
of
the
participants
in the
first study seemed
to know what
the
study
was
about.
In
the
second study,
two
participants
no-
ticed that
the
photographs were reversals
of
each other,
but
again none expressed knowl-
edge that
a
particular photograph
was
hypoth-
esized
to be
preferred.
It
would have been,
of
course, interesting
to ask
participants which
photographs corresponded
to
their mirror
images,
but we
were afraid that participants
would convey this information
to
other
par-
ticipants
and
explicitly sensitize them
to the
experimental hypothesis.
Nearly
a
decade
has
elapsed since Zajonc
(1968) published
his
monograph
and re-
kindled interest
in the
exposure hypothesis.
Numerous processes have been hypothesized
to mediate
the
exposure-affect relationship.
Except
for the
experimental-demand hypothe-
sis,
the
present paradigm does
not
permit
an
assessment
of the
validity
of
other explana-
tions,
such
as
classical conditioning, instru-
mental conditioning,
or
response competition.
The main reasons
for
considering
the
results
of this study
as
support
for the
mere-exposure
hypothesis
are (a)
that
the
mirror
and
true
facial photographs
are
almost indistinguish-
able,
as
evidenced
by the
participants' inabil-
ity
to
state cogent reasons
for
their prefer-
ences
(the
difference between them
is
just
barely accessible
to the
participants'
con-
sciousness),
and (b)
that frequency
of
expo-
sure
is the
primary dimension along which
they differ.
In
summary,
the
covariations that

REVERSED FACIAL IMAGES AND MERE EXPOSURE601
we predicted on the basis of the mere-exposure
hypothesis were confirmed and do not appear
consistent with a demand-characteristics in-
terpretation.
References
Moreland, R. L., & Zajonc, R. B. A strong test of
exposure effects. Journal of Experimental Social
Psychology, 1976, 12, 170-179
Saegert, S., Swap, W., & Zajonc, R B. Exposure,
context, and interpersonal attraction. Journal of
Personality and Social Psychology, 1973, 25, 234-
242
Stang, D. J. Intuition as artifact in mere exposure
studies. Journal of Personality and Social Psychol-
ogy, 1974, 30, 647-653.
Zajonc, R. B. Attitudinal effects of mere exposure.
Journal of Personality and Social Psychology
Monograph 1968, 9 (2, Pt. 2).
Received February 4, 1977 •
Announcement for Student Subscribers Outside the United States
Students in psychology programs outside the United States are eligible to subscribe
to the Journal of Personality and Social Psychology at the rate of $20 for 1978.
This is the same rate paid by members of the American Psychological Association.
In order to obtain a Student in Psychology application and order form to subscribe
to JPSP and all other APA journals, write to Box M, American Psychological
Association, 1200 Seventeenth Street, N.W., Washington, D.C. 20036, U.S.A.