Views
of the Uncanny Valley
A Humanoids
2005 Workshop
Tsukuba, Japan, 5 December 2005
www.theuncannyvalley.org Call
for Participants
The 2005 IEEE-RAS International Conference on Humanoid Robots
will hold workshops on Intelligence Dynamics, Cognitive Architectures
for Humanoids, and Views of the Uncanny Valley in addition
to the invited talks "Neuroscience and Humanoid Robots" and
"Humanoid Robots Step Up to Perform Human Tasks." The theme
of the main conference is "Humanoid Robots that Interact with
Humans and Society," and there will be 30 oral and 48 poster
presentations in the main conference. Topics of interest range
from anthropomorphism in robotics to human robot interaction
and the interplay between robotics software, hardware, and
theory and the social and neurosciences. Registration information
is available at www.humanoidrobots.org/humanoids2005.
This announcement provides more detailed information on the
workshop Views of the Uncanny Valley.
Theme
and goals
The term "bukimi no tani" or "uncanny valley" was coined 35 years ago by Dr. Masahiro Mori, and it stands
today as one of the most commonly known design considerations
of humanoid robots and synthetic characters. In this workshop
we bring together researchers in robotics, visual perception,
and neuroscience to discuss the uncanny valley from these
different perspectives. The goal will be to examine how the
perception of human and robot motions are transformed into
an appreciation of the events being observed. In addition,
we will try to direct our findings to a discussion of potential
theoretical bases of human-humanoid interaction and to obtaining
a method for accurate navigation around the uncanny valley.
Organization
The speakers in the workshop reflect three complementary approaches to understanding the uncanny valley: perceptual and cognitive frameworks, robot design and experimental investigations, and brain responses to robots and simulated characters.
Perceptual and cognitive frameworks
Frank E. Pollick: The existence of the uncanny valley presupposes several basic properties of visual cognition. We will first outline these properties and then show evidence for their existence and how they delineate the scope of the uncanny valley.
Christopher H. Ramey: The uncanny valley should not be regarded as unique to humanoid robotics. Uncanny valleys can arise in any cognitive act that links qualitatively different categories by quantitative metrics that call into question the originally differentiated categories. This effect can be especially pronounced when one of those categories is one's self or one's humanity. From a phenomenological standpoint, humanlike robots may force one to confront one's own being by creating intermediate conceptualizations that are neither human nor robot.
Robot design and experimental investigations
David Hanson: Although the uncanny exists, the inherent, unavoidable dip (or valley) may be an illusion. Extremely abstract robots can be uncanny if the aesthetic is off, as can cosmetically atypical humans. Thus, the uncanny occupies a continuum ranging from the abstract to the real, although norms of acceptability may narrow as one approaches human likeness. However, if the aesthetic is right, any level of realism or abstraction can be appealing. If so, then avoiding or creating an uncanny effect just depends on the quality of the aesthetic design, regardless of the level of realism.
Hiroshi Ishiguro: The uncanny valley is a complex phenomenon that is influenced by many factors including the interplay between appearance and behavior, habituation, and the age and personal history of the perceiver. This talk presents an analysis of experiments in human-android interaction that reveal how babies, children, and adults perceive and respond to androids in an age-dependent manner.
Karl F. MacDorman: Robots that lie in the uncanny valley may act as a subconscious reminder of our own mortality. We explore this hypothesis by reproducing seminal experiments from terror management theory, substituting an uncanny robot for the reminder of death in the control group.
Brain responses to robots and simulated characters
Luc
Berthouze: In an fMRI study on the affective appraisal of postures, we measured
cortical activity of subjects evaluating the expressiveness of a
faceless computer graphic avatar depicting affective postures. The
study revealed the importance of neural structures such as face
fusiform gyrus, anterior cingulate (cognitive area), and possibly the
mirror system. Using this study as the starting point of a more
targeted analysis should provide insights on the neural correlates of
the uncanny valley effect.
Thierry
Chaminade
and Jessica
K. Hodgins: We use social cognitive neuroscience paradigms
to investigate the effect of changes in appearance and rendering
style for robots and animated characters. In the first study,
a motor interference paradigm derived from the study of the
mirror system between action perception and execution in humans
is used to investigate human-humanoid robot interaction and
human-animated character interaction. Results show that in contrast
to an industrial robot, the actions of humanoid robots do interfere
with human action. A second study investigates the effect of
the appearance of an animated figure on the perception of biological
motion. Neuroimaging studies, in particular fMRI, are used to
define the neural bases of this effect. Christian
Keysers: In the last ten years, it has become evident
that when humans observe the actions and sensations of others,
these are transformed in the brain of the observer into the
observer's own actions and sensations through a series of
mechanisms called "shared circuits" or mirror circuits. We
will review this literature, suggesting an essential mechanism
for understanding others. In particular, we will show that
the perception of robots performing actions appears to be
processed in much the same way as the perception of other
humans. These findings open the way for a neuroscience of
human-robot social interactions.
Target
participants
Robotics engineers and computer scientists with an interest
in artificial intelligence, machine learning, pattern recognition,
and control, especially those whose target platform includes
humanoid robots; psychologists and sociologists who are concerned
with real-time embodied communication or social development;
cognitive scientists who are concerned with the relationship
between brain processes and social dynamics; social and comparative
biologists; and philosophers. |
