We perceive the egocentric position and velocity of ourselves by many
senses, such as vision, proprioception and vestibular sense. Normally
the information by these senses is in harmony. However, there are situ
ations in which the information is inharmonious. When we watch a wide-
screen monitor or we ride in an accelerating or turning vehicle, visua
l information conflicts with proprioceptive and vestibular information
. Since human egocentric perception has been studied for each separate
sense in the previous research, it is not clear how the integration a
mong these senses contributes to perception of human orientation. In E
xperiment 1, we investigated the contribution of visual, proprioceptiv
e and vestibular information in isolation and in combination to percep
tion of direction of forward self-motion. An observer rode in small ve
hicle (vestibular information) or walked (proprioceptive information)
through a narrow corridor. Many vertical bars were hung from a ceiling
of the corridor. When the observer moved forward, she/he viewed expan
ding optic flow of the bars (visual information) through video cameras
and a helmet-mounted stereoscopic display. By fastening the observer
and/or the cameras at different angles, consistent or conflicting info
rmation about forward direction was given through each sense. It was f
ound that when there was conflicting information about the direction o
f forward motion, (a) vestibular information was more dominant than vi
sual cue, (b) visual and proprioceptive information was linearly added
with weighting, and (c) visual information was dominant for backward
motion. In Experiment 2, we investigated sensory integration when we m
oved forward with linear acceleration. Direction of the acceleration w
as either forward or sideways, namely corresponding to a ride in an ac
celerating or turning vehicle. We developed a new method to measure se
nsation of self-motion more objectively by using the three-dimensional
position-sensor system. Positions of observer's head, shoulder, waist
and ankle were measured to find body tilt accompanied by sensation of
self-motion. It was found that the body tilted towards the opposite d
irection of the self-acceleration and the angle of body tilt was in go
od agreement with the subjective amount of the accelerating sensation.
The body tilt was even induced by solely visual information. This imp
lied that visual information contributed to perception of self-acceler
ation as well as self-motion.