Content uploaded by Tobias Matthias Schneider
Author content
All content in this area was uploaded by Tobias Matthias Schneider on Apr 17, 2014
Content may be subject to copyright.
Beside variables like facial symmetry (e.g. Grammer and Thornhill 1994; Jones et al
2001; Rhodes et al 2001; Fink et al 2006), averageness (e.g. Carbon et al; 2010) and
skin color (Jones et al 2004 & Fink et al 2006) the perceived human weight proves to
be a reliable predictor for health and fitness (Coetzee, Perrett & Stephen, 2009). They
showed that perceived facial adiposity is used as a cue to health. Unanimously people
make fairly accurate weight judgments using facial cues alone, for instance BMI
explains 43% of the variance in facial weight judgments.
To the authors’ knowledge, there is no study considering the impact of viewing angle
on weight perception. Specifically, are we able to influence weight judgments in 2D
pictures by changing the viewing angle and if so, are we still able to judge the weight
accurately of people of another morphological group (race) due to lack of knowledge
about their body prototypes?
To approach the first question, we let participants judge the weight of persons on
basis of human faces in three viewing conditions in two experiments. The third
experiment investigates the impact of human race on perceived weight.
Figure 2. Results of experiment 1 and 2: Main effect of viewing angle
Losing weight without dieting:
Viewpoint-dependent weight assessment on the basis of faces
Tobias M. Schneider1, Heiko Hecht1 and Claus-Christian Carbon2
1Department of General Experimental Psychology, Johannes Gutenberg-University, Mainz, Germany
2Department of General Psychology and Methodology, University of Bamberg, Germany
Introduction
35
40
45
50
55
60
65
70
75
+30° frontal -30°
Estimated weight
Viewing angle
Means within subjects
Means between subjects
Participants. 31 (18 female) subjects from Germany (Mainz) participated in the
experiment. Mean age was 36.8 years (SD = 17.2, range 18 to 69 years).
Material. The material consisted of three sets each with 48 3D models of human faces
(24 female & 24 male, M = 23.5, SD = 16.0, range 3 to 56 years) in three viewing
conditions (slanted downwards by +30°, face frontal, or slanted upwards by -30°).
Conditions were balanced across the sets in a fixed order. Gender of the stimuli was
arranged blockwise. Each picture was standardized to a width-to-height-ratio of 1024
× 768 pixels.
Procedure. Participants were randomly allocated to one set. The task was to judge the
weight of persons on the basis of 48 human faces. Each trial started with a fixation
cross, followed by a blank screen and the target until a verbal response was made.
Stimuli were presented on a Dell e151FPp 15” TFT-monitor (60 Hz). Eyes-monitor-
distance was 50 cm. Weight judgments were made in kg.
Method
Experiment 1: weight assessment on the basis of faces (within subjects)
0
20
40
60
80
100
120
+30° frontal -30°
Estimated weight
Viewing angle
Method
Results
A repeated measures ANOVA for weight estimation resulted in a significant main
effect for viewing angle, F(2,58) = 278.4 , p < .001 , ηp2 = .906 (see figure 2, blue line
and figure 3). We found an effect for the gender of stimulus, F(1,29) = 590.7 ,
p < .001 , ηp2 = .953.
Figure 1a) Example stimulus +30° 1b) Example stimulus “frontal” 1c) Example stimulus -30°
Experiment 2: weight assessment on the basis of faces (between subjects)
Participants. 30 (15 female) subjects from Germany (Mainz) participated in the
experiment. Mean age was 35.0 years (SD = 15.5, range 19 to 68 years).
Material. The material consisted of three sets each with 48 3D models of human faces
(24 female & 24 male, M = 26.8, SD = 18.0, range 3 to 64 years) with the same three
viewing angles as in experiment 1. Each set consisted of only one viewing angle.
Gender of the stimuli was arranged blockwise. Each picture was standardized to a
width-to-height-ratio of 1024 × 768 pixels.
Procedure. The weight estimation task was the same as in experiment 1.
Results
Figure 3 (experiment 1). Distribution of weight judgments (stimulus
based analysis), F(2,94) = 300.3 , p < .001 , ηp2 = .865.
We also examined the relationship between “+30°” and “-30°” viewing angle together
and “frontal” with a regression analysis (see figure 4). Viewing angles significantly and
positively predicted 96,9 % of the variance in weight judgments, F(2,47) = 696.0, p <
.001, R2 = .969.
Interestingly, there was a significant and strong relationship between +30° and the
“frontal” condition, t(47) = 6.9, β = .772, p < .001, while the relationship between
“frontal” and “-30°” was not significant, t(47) = 2.0, β = .217, p = .058, n.s.
ANOVA for weight estimation resulted in a
significant main effect for viewing angle,
F(2,27) = 9.1 , p = .001 , ηp2 = .403) (see
figure 2, orange line and figure 5).
Figure 4. Interrelation between the “+30°” + “-30°” and the “frontal” condition
0
20
40
60
80
100
+30° frontal -30°
Estimated weight
Viewing Angle
Figure 5 (experiment 2). Distribution of weight judgments (stimulus
based analysis), F(2,94) = 97.5 , p < .001 , ηp2 = .675.
Experiment 1 and 2 demonstrated that weight judgments are susceptible to the
manipulation of viewing angle. Faces seen from -30° yielded the highest judgments of
weight and +30° viewing angles produced the lowest.
Beside the significant relationship between +30° and “frontal”, the relationship
between -30° and “frontal” was not significant, what indicates the use of different
cognitive processes.
In the third experiment we found an interesting cultural effect. Whereas German
observers judged all faces equally well, Japanese observers had trouble with
European faces but not with Asian faces.
Discussion
Carbon, C. C., Gruter, T., Gruter, M., Weber, J. E., & Lueschow, A. (2010). Dissociation of facial attractiveness and distinctiveness processing in congenital prosopagnosia.
Visual Cognition, 18(5), 641-654.
Coetzee, V., Perrett, D. I. & Stephen, I.D. (2009). Facial adiposity: a cue to health? Perception, 38, 1700-1711.
Fink, B., Neave, N., Manning, J. T., Grammer, K. (2006). Facial symmetry and judgments of attractiveness, health and personality. Personality and Individual Differences, 41,
491-499.
Grammer, K. & Thornhill, R. (1994). Human (homo sapiens) facial attractiveness and sexual selection: the role of symmetry and averageness. Journal of Comparative
Psychology, 108, 233-242.
Jones, B. C., Little, A. C, Penton-Voak, I. S., Tiddenman, B. P., Burt, D. M., Perret, D. I. (2001). Facial symmetry and judgments of apparent health: support for a ‘good genes’
explanation of the attractiveness-symmetry relationship. Evolution and Human Behavior, 22, 417-429.
Jones, B. C., Little, A. C., Feinberg, D. R., Penton-Voak, I. S., Tiddenman, B. P. & Perett D. I. (2004). The relationship between shape symmetry and perceived skin condition
in male facial attractiveness. Evolution and Human Behavior, 25, 24-30.
Rhodes, G., Zebrowitz, L. A., Clark, A., Kalick, S. M., Hightower, A. & McKay, R. (2001). Do facial averageness and symmetry signal health? Evolution and Human Behavior,
22, 31-46.
References
Regression standardized predicted value
Estimated weight in the frontal condition
Method
Experiment 3: Intercultural differences relating to weight perception
Participants. The Japanese sample consisted of 146 (100 female) psychology students
from Ritsumeikan University (Kyoto). Mean age was 19.1 years (SD = 2.0, range 18 to
36 years). The German sample consisted of 125 (106 female) psychology students
from the Johannes Gutenberg-University (Mainz). Mean Age was 22.1 years (SD = 3.9,
range 18 to 42 years).
Material. The material consisted of a printed questionnaire with 12 (6 Japanese & 6
German) frontal pictures of human faces (equal ratio for gender). Mean age was 24
years (SD = 3.9). The stimuli were printed in 4 different random orders.
Procedure. Participants were randomly allocated to one of the orders. The task was to
judge the weight of persons on the basis of 12 human faces. Weight judgments were
made in kg.
Results
An independent samples t-test for weight estimation resulted in a significant effect for
the ethnicity of the observers, t(269) = 23.6, p < .001. We also found an effect for the
observer’s average relative estimation error (AREE), t(269) = 23.8, p < .001. AREE for
the ethnicity of stimulus was also significant, t(270) = 35.3, p < .001 (see figure 5).
-6.30
6.70
-20.00
-7.30
Observer
German Japanese
Stimulus (face)
German Japanese AREE
.20
-13.50
Figure 5. Distribution of the average relative estimation error in %.
Negative numbers indicate underestimation.
*
Contact: toschnei@students.uni-mainz.de
-13.20 -.30
*