Content uploaded by Christer Johansson
Author content
All content in this area was uploaded by Christer Johansson on May 29, 2020
Content may be subject to copyright.
Prototypes and Recognition of Self in Depictions of Christ
Carla-Sophie Lembke1, Per Olav Folgerø2, Alf Edgar Andresen2 & Christer Johansson2
1Institute of Cognitive Science, 2Department of Linguistic, Literary and Aesthetic Studies
University of Osnabrück, Germany University of Bergen, Norway
Carla.Lembke@gmx.de, Per.Folgero@uib.no, alf@spotstudio.no, Christer.Johansson@uib.no
Abstract
We present a study on prototype effects. We designed an experiment investigating the effect of adapting a prototypical
image towards more human, male or female, prototypes, and additionally investigating the effect of self-recognition in a
manipulated image. Results show that decisions are affected by prototypicality, but we find less evidence that self-
recognition further enhances perceptions of attractiveness. This study has implications for the psychological perception
of faces, and may contribute to the study of Christian imagery.
Keywords: Prototype effects, Self Similarity, Attractiveness, Subjectivity in Face Perception, Experimental Esthetics
1. Introduction
The image of Christ, which is central to both modern and
historical Christianity, has undergone many changes to
evolve from the historically accurate, middle-eastern
carpenter into the modern Hippie Christ that we still see
today.
We will try to see this artistic revolution as a mechanism
of prototype formation, where repeated exposure to a
particular visual category influences our liking or
disliking of what we see. Our theory is that the diversity
of Christ images may reflect the diversity of the believers
by means of artistic adaptation, i.e. painters produced
more of the images that appealed most to believers on
trustworthiness, attractiveness and identification with self.
Recently, Jackson et al. (2018) have shown that American
subjects saw God as being similar to themselves regarding
attractiveness and age. We argue that this egocentric bias
also plays an important role when it comes to the Christ
figure. Therefore, we expect that participants will prefer
the images of Christ in which they may recognize features
of themselves.
In accordance with this, other studies have documented
that mere exposure to a category of stimuli increases the
familiarity and liking of that particular domain of stimuli
(Reber et al., 2004; Chenier & Winkielman, 2009). Mere
exposure has been shown to reduce the identification and
classification latencies for stimuli, meaning that it
increases the processing speed or fluency.
Another effect of repeatedly seeing similar variants on a
theme is that certain forms become prototypical. An
everyday example is our tendency to like new retro
models of cars, e.g., the Volkswagen Beetle. Winkielman
et al. (2006), claim that prototypicality is one of many
fluency-enhancing variables. Moreover, they suggest that
part of the preference for prototypicality stems from a
general mechanism that links fluency and positive values.
When encountering novel faces, we are quick to attribute
different traits to them. These attributions happen as
quickly as 33 milliseconds after exposure to the face
stimulus (Todorov et al., 2009) and the mechanisms
responsible for them are already present and reliable in
children of 3 to 4 years of age (Cogsdill et al, 2014). To
form these impressions, we rely mostly on facial cues,
even when other, more relevant, information is available
to us (Rezlescu et al., 2012; Oliviola et al., 2014).
There has also been evidence for a bias towards our own
facial features when we attribute traits to strangers. The
popular observation that couples tend to look alike
supports the theory that, with increasing exposure to our
face and genetically similar faces over time, we develop
an attraction to faces similar to our own (Hinsz, 1989).
Facial similarity also has a positive effect on perceived
trustworthiness, group cooperation, and voter preferences
in political elections (DeBruine, 2002; DeBruine, 2005;
Krupp et al., 2007; Bailenson et al., 2008). In our
experiment, we test whether adding the subjects’ features
to the image of Christ, will make that image more likable
as well.
This study explores the idea that the image of Christ has
evolved to be more likable by adapting a similarity to the
community of believers, including the female believers,
by ameliorating hurdles to identify with the image. We
hypothesized that this adaption leads to an increase in the
attractiveness of the Christ figure. Furthermore, we
hypothesized that participants would judge images
containing their own image more favorably, even without
being conscious of the presence of their own image. This
would provide additional empirical evidence for the mere
exposure hypothesis, according to which, a participant
would prefer an image containing features that are
familiar to them.
Previous research on the image of Christ has shown that
the Renaissance preference for depicting Christ (as God)
en face, is associated with enhancing positive attributions
such as being harmonious, caring, trustworthy, inclusive
and respected (cf. Folgerø et al 2016a). Furthermore, it
has been shown that people may judge the gender of a
face from facial proportions between the tip of the nose
and the eyebrows (cf. Geniole et al., 2014). For images of
Christ, Folgerø et al (2016b) showed in a priming
experiment that a brief presentation of a word (female or
male) made participants significantly over-represent a
choice of female for images of Christ when primed by the
word for female. Images of young men and women were
less affected when primed by the opposite gender (ibid).
This suggests that not only had the Renaissance image of
Christ adapted towards a more Italian / European portrait,
but also the painters may have included some female
features, adding a more universal androgynous appeal.
2. Method
2.1 Participants
17 students (8 male, 9 female) were initially recruited for
the study. Due to the nature of the morphing procedure we
used, one female participant was excluded from the study,
so there were a total of 16 students, 8 male, and 8 female.
All participants gave their informed consent to participate
in the study and to have their picture taken and used for
publication.
Only 12 participants (aged 18 to 65; mean 26.4 CI[18.2;
34.6]) participated in the final experiment. Thus only 12
images matched the 12 participants for self: 6 male and 6
female.
2.2 Stimuli
We used Sqirls Morph, which uses Beier & Neely’s
(1992) algorithm to morph pictures.
We first chose three renaissance depictions, and one
Eastern depiction from the 6th century A.D. of the Holy
Face and we produced a “Christ prototype” by morphing
them (Figure 1).
Furthermore, we created a female and male prototype by
combining the pictures of eight female participants and
eight male participants, respectively. The male and female
prototypes were also combined to produce a human
prototype (Figure 2).
We then morphed each picture (the individual pictures
and the prototypes) with our Christ prototype to create the
stimuli used in the experiment. The 16 individualized
Christ images consisted of 80% Christ and 20% the image
of the participant (Figure 3).
Figure 1: Creation of the Christ prototype. First four
canonical images of Christ are morphed pairwise, and
then the pairs are morphed.
Figure 2: Prototypes created from participants. All created
by pairwise morphing. Upper row: Female, Human and
Male prototypes. The lower row shows the effect of
adding the Christ prototype.
Figure 3: Individual participants morphed with the Christ
prototype. The alternating rows show first male, then
female participants. The morphed pictures consist of 80%
Christ prototype and 20% individual picture.
3. Design and Procedure
The experiment has two phases. In the first phase, all
participants had their picture taken by a professional
photographer in a standardized setting. They sat at an
equal distance to the photographer in front of a uniform
gray wall directly facing the photographer. The second
phase took place six weeks after all pictures had been
taken. We created a balanced Round Robin tournament in
SuperLab, where an individualized picture (20% from an
individual and 80% from the Christ prototype) was
presented next to one of the prototypes. All pairs were
presented in a different random sequence for each subject.
The side of the screen on which the prototype was
presented was also randomized (left or right). All
combinations were presented exhaustively.
Participants were asked to select the image they found
most attractive of the two, as in a “Hot or Not” task. All
participants were asked to make their responses quickly
while remaining accurate. Reaction times were collected,
and difficult choices were expected to show increased
reaction times.
The experiment is prepared for a follow-up using a
“Visual World” paradigm, where eye-tracking is used to
detect which of the images receive the longest focused
attention. Eye-tracking was not available in our lab at the
time of our experiment.
4. Results
Four participants did not take part in the final task. That
left us with data from 12 participants (aged 18 to 65; mean
26.4 CI[18.2; 34.6]), and gender balanced. Responses that
were faster than 300ms were excluded because it would
be impossible to process both images and take a decision
within that time.
Visualization is performed by assoc from the R vcd
package (cf. Meyer et al. 2003). Prototypes competed
against 12 individualized images and the original four
images of Christ (cf. Figure 1), each presented one time
on the left and one time on the right side.
The female and Christ prototypes won significantly more
competitions than any of the other images (Figure 4). The
female prototype also shows the fastest reaction times.
Subjects did not show evidence of self-recognition in
preference (Figure 5) or decision times. The differences
are as in Figure 4. Self tends to win more over the human
and male prototypes. In the debriefing after the study,
only one participant claimed to have recognized
themselves in the images.
Reaction time data was analyzed using a mixed effects
model (Kuznetsova et al. 2017) using two fixed factors:
the prototype and the choice (for prototype or person).
Participants and test items (marked for first or second
trial) were used as random factors (explanations for
random variance). Furthermore, we used different
intercepts for prototypes by each participant and for
choice by each item. The reaction times were transformed
using a natural logarithm transformation that improves
skewed data (long decision times are thought to signal
close decisions, but the analysis demands normal
distribution). One similar well-known transform is
acoustic energy into the decibel scale, which mirrors our
perception of sound volume. We investigated some
models that included interaction between choice and
prototype, but this interaction was not significant and was
thus excluded for better model-convergence. The Mixed
Effects analysis of the reaction times for decisions (Figure
6) shows a significant effect for choice. When the
decision is for a prototype the decision is faster (F(1,35.6)
= 5.3; p = 0.027). There were also differences between
prototypes (F(6,15.2) = 3.9; p = 0.015), most notable PW
is faster. We could not confirm any interaction between
participant gender and choice (i.e., male subjects
seemingly had a larger, but not significant, prototype
effect).
Figure 4: Prototypes are: PH (Human), PHX (Human with
Christ), PJ (Jesus Christ), PM (Man), PMX (Man with
Christ), PW (Woman), PWX (Woman with Christ).
Differences are significant. Red marks cells with lower
than expected frequencies, blue are higher than expected.
χ2
(6)=154.4, p<0.001, Φc=0.096
PH
PHX
PJ
PM
PMX
PW
PWX
PE
241
254
138
236
271
151
227
PR
220
214
335
227
190
319
237
Table 1: Frequency of choice for person (PE) or prototype
(PR), competition for each prototype.
Figure 5: Same graph restricted to choices between self
(=person) and prototype. χ2
(6)=16.5, p=0.011, Φc=0.140.
PH
PHX
PJ
PM
PMX
PW
PWX
PE
16
12
5
15
14
7
13
PR
8
12
18
9
10
17
11
Table 2: Table 1 restricted for choices between self (PE)
and a prototype (PR)
-5.7
-4.0
-2.0
0.0
2.0
4.0
5.4
Pearson
residuals:
p-value =
< 2.22e-16
prototype
choice
proto person
PH PHX PJ PM PMX PW PWX
-1.9
0.0
1.8
Pearson
residuals:
p-value =
0.01146
B
A
proto person
PH PHX PJ PM PMX PW PWX
Figure 6: Response times (natural logarithms). Prototypes
are generally faster when the decision is for the prototype,
with exceptions for the Jesus Christ prototype (PJ) and the
woman prototype (PW).
A model test of the residuals shows an excellent fit to a
normal distribution up to +2 quantiles, but the larger
residuals give room for improvement (Figure 7).
Figure 7: Model test of residuals shows a good fit.
Contrary to our hypothesis, we did not find any significant
tendency for participants to rate their own disguised
image as more attractive. Instead, there was an
insignificant tendency in the opposite direction. One
interpretation is that we simply did not have enough
statistical power since four of our initial 16 participants
were not able to participate in the final task. Alternatively,
our subjects might have judged their own image as
slightly less attractive than the ratings from others.
5. Ethical Considerations & Discussion
Our work poses many ethical questions that we would like
to briefly examine.
In general, as we are working with personal images,
special care must be taken that the subjects stay
anonymous. It is every researcher’s responsibility to
inform all participants of what exactly will happen with
their images after the study itself has come to an end.
Explicit consent was gathered from all participants, also
regarding the use of images. Special care should
nevertheless be taken when publishing the data. In our
case, we decided to only publish the images of the
morphed face stimuli in order to keep the subjects’
anonymity intact. The debriefing of our subjects verified
the validity of this method as only one of our subjects
reported that they had recognized themselves in any of the
presented pictures.
Since we are using images of Christ, we realized that this
might be sensitive in a religious setting. However, the
image of Christ is widespread and familiar to all of our
subjects. We discussed this in the debriefing with our
participants, with no negative reactions. Participants were
generally positive about the underlying theme of finding
something holy in everyone.
This study is thus limited, and therefore generalization of
our findings may be less than absolute. Our small subject
pool may not be representative outside of our local student
population. Minorities are difficult to represent fairly
within a study limited to a dozen subjects. An option to
increase subject diversity is to partner with other
researchers, taking particular caution to safely sharing the
images in order to protect the subjects’ interests.
As with any other field of science, there are distinct
ethical concerns that arise when we research human
attributes. It is essential for all researchers to identify
these ethical issues to the best of their abilities.
In a small study, it is essential to limit the number of
variables in order to have better control of variance. Many
factors affect attractiveness. Our study was open to
everyone, and thus we could not balance all possible
features. Skin tone is one feature that has been linked to
attractiveness, and a lighter skin tone is often reported as
more attractive (Vera Cruz 2018). In our study, the
participants were all similar in skin tone, which was toned
down further by the morphing process. Similarly, blue
eyes became a tone of brown after morphing. It is
conceivable that both skin tone and eye color may affect
ratings of attractiveness. In our experience, when we
observed art interpretations of Christ from the relevant
period it is obvious that Christ has a lighter skin tone and
bluer eyes in Northern Europe than in Southern Europe,
which may be interpreted as an adaptation to the local
populations. In a larger study, the relative importance of
features can be estimated. Symmetry and androgyny may
be more important factors than skin tone and eye color.
However, we do find dark-hued representations of both
the Mother Virgin and Christ. A dark skin tone in Europe
points at an anti-adaptation for the Black Madonna (cf.
https://en.wikipedia.org/wiki/Black_Madonna), and an
adaptation for the Christos Negros of Central America (cf.
https://en.wikipedia.org/wiki/Cristos_Negros_of_Central_
America_and_Mexico).
1
1
7.25 7.30 7.35 7.40 7.45
choice
mean of log(rt)
2
2
3
3
4
4
5
5
6
6
7
7
person proto
prototype
3
5
7
2
4
6
1
PJ
PMX
PWX
PHX
PM
PW
PH
-2 02
-1 0123
Theoretical Quantiles
Sample Quantiles
A possible hypothesis is that facial anatomy and
symmetry are more important than skin tone in regards to
how people identify with a representation.
We also know from the Thatcher-effect that people
perceive features of a face separately. Yamaguchi et al
(1995) investigated features of the face that affect
perception of gender, and found that eyebrows and outline
of the face were important features. They found a bias
towards own gender in Japanese students, which we have
not detected for Norwegian students in our lab. In our own
research, we found that Renaissance portraits of Christ
had facial proportions (width between eyes vs. length
between eyebrows and tip of nose) that were more typical
of portraits of female subjects.
It is also interesting to note the deep history of morphing
and composite (prototype) effects. Galton (1878) used
early photographic techniques to overlay portraits in order
to form a composite image. Galton describes a physical
procedure for normalizing the pictures by aligning some
fix points such as pupils of the eyes. He notes: “... that the
features of the composites are much better looking than
those of the components.” Thus, he is one of the first to
notice the prototype-effect on beauty, as the composites
get more symmetrical and blemishes are blurred out.
Galton also noticed that individual characteristics could be
hard to perceive across ethnic classes, as we tend to
remember deviances from a familiar composite prototype
formed by experience. In a sense, the prototype of the
other could be just as distant as the individual, with
implications for witness psychology.
6. Conclusions
Both female and Christ prototypes were judged as more
attractive (by winning more competitions). Prototypes
were processed more fluently, as reflected in their reaction
times. The female prototype displayed the fastest decision
times, and was more frequently chosen, which may be
interpreted as easier to process and possibly more
attractive to our participants. Being part of all the
individualized images made decisions for the Christ
prototype harder, but this prototype was more frequently
chosen. The findings support our central hypothesis
concerning the adaption of the image of Christ towards a
cognitively more pleasing image. An advantage for
female features was detected, supporting earlier results on
feminine features in the image of Christ (cf. Folgerø et al.
2016b). In Folgerø et al. (2016b), the stimulus was
restricted to a section of the face between the tip of the
nose and the eyebrows, and yet people showed effects for
correct identification of gender, as well as recognition of
Christ as a female when primed with “woman.”
However, we did not find that images containing features
of self were judged as more attractive. Following DeBruin
(2005), we suggest that the results would have been
different if we had asked the participants to judge
trustworthiness instead of attractiveness. In ongoing data-
collection, we note that a majority of our participants now
claim, in debriefings, to have recognized themselves in a
similar task that includes selecting the face they trust the
most. More research is needed to investigate if trustworth-
iness is more associated with self-similarity.
7. Acknowledgements
The participation of Carla Lembke was made possible by
an EU Erasmus+ Traineeship. Alf Edgar Andresen is a
former colleague and a professional photographer, who
has prepared photographs and morphed images. Without
Alf this work could not have been accomplished. The
work has been performed at the Humanities Lab at the
University of Bergen (https://www.uib.no/en/rg/humlab).
Our colleague Tori Larsen provided detailed feedback on
the text. We would also like to thank four anonymous
reviewers for useful suggestions, which resulted in an
extended discussion on the role of skin tone and ethnicity
and some clarifications that we think have benefitted our
presentation.
Bibliographical References
Bailenson, J., Iyengar, S., Yee, N. & Collins, N. 2008.
Facial similarity between voters and candidates causes
influence, Public Opinion Quarterly, 72(5), 935-961.
https://doi.org/10.1093/poq/nfn064
Beier, T. & Neely, S. 1992. Feature-based image meta-
morphosis, Computer Graphics, 26(2), 35–42.
https://doi.org/10.1145/133994.134003
Chenier, T. & Winkielman, P. 2009. The origins of
aesthetic pleasure: Processing fluency and affect in
judgment, body, and the brain. Ch. 14 in Martin Skov
& Oshin Vartanian (Eds) Neuroaesthetics, Routledge /
Baywood, 275-289.
Cogsdill, E., Todorov, A., Spelke, E. & Banaji, M. 2014.
Inferring character from faces: A developmental study.
Psychological science, 25(5). 1132-1139.
https://doi.org/10.1177/0956797614523297
DeBruine, L. 2002. Facial resemblance enhances trust.
Proceedings of the Royal Society B: Biological
Sciences, 269(1498), 1307-1312.
DeBruine, L. 2005. Trustworthy but not lust-worthy:
Context-specific effects of facial resemblance.
Proceedings of the Royal Society B: Biological
Sciences, 272(1566), 919-922.
Folgerø, P., Hodne, L., Johansson, C., Andresen, A.,
Sætren, L.C., Specht, K., Skaar, Ø.O. & Reber, R.
2016a. Effects of Facial Symmetry and Gaze Direction
on Perception of Social Attributes: A Study in
Experimental Art History, Frontiers in Human
Neuroscience, 10, September 2016.
https://doi.org/10.3389/fnhum.2016.00452
Folgerø, P., Johansson, C. & Andresen, A. 2016b.
Transgender Priming in Medieval Europe, XXIV.
Conference of the International Association of
Empirical Aesthetics, Vienna, Austria. August 29 –
September 1. 2016.
Galton, F. 1878. Composite portraits. Journal of the
Anthropological Institute of Great Britain and Ireland,
8, 132–142. http://www.galton.org/essays/1870-
1879/galton-1879-jaigi-composite-portraits.pdf
Geniole, S.N., Molnar, D.S., Carré, J.M. & McCormick,
C.M. 2014. The facial width-to-height ratio shares
stronger links with judgments of aggression than with
judgments of trustworthiness. Journal of Experimental
Psychology: Human Perception and Performance,
40(4), 1526–1541.
https://doi.org/10.1037/a0036732
Hinsz, V. 1989. Facial resemblance in engaged and
married couples, Journal of Social and Personal
Relationships, 6, 223-229.
Jackson, J., Hester, N. & Gray, K. 2018. The faces of God
in America: Revealing religious diversity across people
and politics. PloS one, 13.6 (2018): e0198745.
Krupp, D., Debruine, L. & Barclay, P. 2008. A cue of
kinship promotes cooperation for the public good.
Evolution and Human Behavior, 29(1), 49-55.
Kuznetsova, A., Brockhoff, P. & Christensen, R. 2017.
lmerTest Package: Tests in Linear Mixed Effects
Models. Journal of Statistical Software, 82(13), 1–26.
https://doi.org/10.18637/jss.v082.i13
Meyer, M., Zeileis, A. & Hornik, K. 2003. Visualizing
independence using extended association plots. Proc. of
the 3rd International Workshop on Distributed
Statistical Computing, K. Hornik, F. Leisch, A. Zeileis
(eds.), ISSN 1609-395X. http://www.R-
project.org/conferences/DSC-2003/Proceedings/
Olivola, C., Funk, F. & Todorov, A. 2014. Social
attributions from faces bias human choices. Trends in
Cognitive Sciences,18, 566-570.
Reber, R., Schwarz, N. & Winkielman, P. 2004.
Processing fluency and aesthetic pleasure: Is beauty in
the perceiver's processing experience? Personality and
social psychology review, 8(4), 364-382.
Rezlescu, C., Duchaine, B., Olivola, C. & Chater, N.
2012. Unfakeable facial configurations affect strategic
choices in trust games with or without information
about past behavior. PloS one, 7, e34293
https://doi.org/10.1371/journal.pone.0034293
Todorov, A., Pakrashi, M. & Oosterhof, N. 2009.
Evaluating faces on trustworthiness after minimal time
exposure. Social Cognition, 27(6), 813-833.
Vera Cruz, G. 2018. The impact of face skin tone on
perceived facial attractiveness: A study realized with
an innovative methodology, The Journal of Social
Psychology, 158:5, pp. 580-590.
https://doi.or/10.1080/00224545.2017.1419161
Winkielman, P., Halberstadt, J., Fazendeiro, T., & Catty,
S. 2006. Prototypes are attractive because they are easy
on the mind. Psychological Science, 17. 799-806.
https://doi.org/10.1111/j.1467-9280.2006.01785.x
Yamaguchi, M. K., Hirukawa, T., & Kanazawa, S. 1995.
Judgment of gender through facial parts. Perception,
24, 563–575.