The morphological characteristics that mark a baby’s face
are large eyes relative to the rest of the face, fine, high
eyebrows, light skin and hair color, red lips that are pro-
portionally larger, a small, wide nose with a concave
bridge, and a small chin. The facial features are also
placed lower on the face [11, 42]. Across cultures, these
same facial configurations characterize babyfaced adults
[9, 42]. Even a single babyish facial feature can effect
trait impressions [11]. For instance, faces with large eyes
are considered warmer, weaker, more honest, credulous,
and submissive [43, 44].
Other significant age related differences in faces concern
developmental changes in craniofacial profile shape. Of
particular note are differences in the relative size of the
brain capsule and the slant of the forehead in relation to
the chin. The infantile cranium is proportionally much
larger than the fully mature cranium, and the infantile
forehead protrudes whereas the adult forehead recedes.
Another important characteristic is a dramatic increase in
jaw size. Todd and Mark [45] have modeled facial
maturity by using a cardioidal strain transformation. Ap-
plied to standard profile shapes, this transform has been
shown to accurately approximate real growth. Further-
more, studies on the trait attributions of profiles that vary
in the degree of cardioidal strain applied are consistent
with findings on facial maturity [11, 46]. As craniofacial
profile maturity decreases, so do perceived alertness, reli-
ability, intelligence, and strength [19]. Moreover, infan-
tile profile shapes are more lovable, less threatening [19],
and elicit stronger desires to nurture and protect [46].
Yet another way to model the trait impressions of the face
is to focus directly on those features that give rise to vari-
ous trait impressions. Early investigations into the per-
son perception of the face relied almost exclusively on
measuring the relative distances of important facial key
points. A major problem with using this approach to
model the trait impressions of the face is deciding a priori
which key points are most responsible for the elicitation
of specific trait impressions [12, p. 2]. Processing faces
using holistic techniques, such as an autoassociative
neural network [47] or, equivalently, prinicipal compo-
nent analysis (PCA) [48], offer a better alternative as they
allow the classifier system to discover the relevant fea-
tures a posteriori. Although neural networks have not
Page 5
been put to task in modeling the trait impressions of the
face, they have proven successful in modeling face per-
ception in terms of gender [49], age [50], and facial
expression [51]. This author is currently investigating the
feasibility of applying this approach in modeling the trait
impressions of the face. Preliminary results suggest that a
PCA model may be capable not only of classifying faces
along specific trait dimensions but also of generating
novel faces with a high probability of producing a
specific physical personality.
Reservations
Guided by the psychological literature on person percep-
tion, ways in which the physical personality of agents
with faces can be modeled have been presented, along
with a discussion of some of the advantages in doing so.
There are, of course, a number of concerns that need to
be addressed. It is possible, for instance, that altering the
physical personality of an agent could lead to confusion
and provoke adverse reactions from users [27]. There are
also societal implications in modeling many of the
stereotypes associated with physical personality. To
some degree, exploiting the trait impressions of various
physical characteristics would perpetuate these stereo-
types. Nonetheless, as Reeves and Nass have argued,
“It’s easier to process what is expected than what is new
or counter-stereotypical.” [27, p. 169] As with human
beings, an agent’s physical form will be searched for
clues regarding its psychological personality. The ques-
tion becomes whether the physical personalities of agents
will take advantage of these impressions to enhance hu-
man interactions and honestly advertise the nature of their
attitudes, drives, desires, and intentions.
From:
64.233.161.104/search?q=cache:xBUn91bnTo0J:www.di.uniba.it/intint/sonthofen/brahnam-f.pdf+cranium+profile+shapes+male+&hl=en