Abstract:
This thesis investigates where it is that people locate themselves in their bodies, as well as
how accurately people can indicate the locations of several of their body parts. It is not well
known, whether there is/are one or more region(s) of their bodies people associate themselves with most. To answer this question, three experimental studies were performed
using several virtual reality (VR) setups where participants pointed directly at themselves with a virtual pointing stick. In the first two studies, participants were also asked, outside of VR, to indicate their self-location on pictures of simple body outlines. In the last two studies,
participants were additionally asked, in VR, to point to several of their body parts. Based on
body part locations as pointed out by the participants in VR, the indicated self-locations could subsequently be interpreted in terms of regions of the participants' perceived bodies, besides in terms of regions of their physical bodies (i.e. based on body part locations measured on their bodies).
In studies of self-location in the body, self-localization has mostly been performed using
outlines of bodies not co-located with the participants' own bodies. Results from these
studies have mainly shown self-localization in the (upper) face region, sometimes combined
with in the upper torso region. Studies of self-location in the body using both explicit and
implicit behavioral measures, have mainly shown self-localization in both the upper face and
the upper torso regions. Across these previous studies findings show a mixed picture, which
has motivated this further study of self-location.
For this thesis, a self-directed, first-person perspective (1PP) pointing paradigm was
developed, which was implemented in several VR setups across the different experiments.
This paradigm was used for self-localization, as well as for body part localization. The
participant was instructed to rotate a pointer with a controller for each trial such that it was
pointing "directly at you", or at one of several of his body parts. The VR setups were used in
the present experiments, mainly because they provide strong experimental control and the
possibility of manipulating sensory cues in ways not otherwise possible (the viewpoint in
study three). Further, they make comparisons possibly between results from in- and outside
of VR (all studies), as well as between different VR setups (study two).
In addition to the VR tasks, a not self-directed, third-person perspective (3PP) body template self-localization pointing task was used, outside of VR. There the participant was
instructed to point "directly at you" with a pen on an A4 print of an outline of a body, under
the assumption that this was a picture of himself.
In the first study participants performed the VR self-localization task using the Oculus Rift
DK2 and the template self-localization task. VR self-localization showed a very strong preference for the upper face. This was not in line with previous behavioral studies, showing
self-localization mainly in both the upper face and the upper torso. Template self-localization
was mostly in the upper torso, followed by in the (upper) face. This was not in line with
previous studies using body outlines, showing self-localization mostly in the (upper) face. The present template results are more in line with the previous behavioral findings (from studies outside of VR), whereas the present VR behavioral findings are more in line with the previous body outline findings. It was concluded that wearing a VR headset might make people more head-focused.
To investigate whether the VR findings from study one were specifically due to the use of
a headset (blocking visual access to the body), or more generally to VR, in study two the VR
pointing paradigm was implemented in both the Oculus Rift and a large-screen immersive
display (LSID), where no headset is worn. Further, VR body part localization was added to the VR self-localization. Both in specific clinical, as well as in healthy populations, systematic
distortions in the perception and representations of one's own body have been found. This
has provided additional motivation for the inclusion of body part localization in studies two
and three for this thesis.
In study two, VR self-localization in terms of the physical body was mostly to all regions of
the body from the upper torso upwards, as well as above the head. Further, participants were able to point reasonably accurately to most of their body parts in the LSID, but much less so in the VR headset. Inaccuracies were particularly large for the body parts near the borders of the body. After rescaling the self-localization pointing to the perceived body, it was mainly to the (upper followed by lower) face, followed by the (upper followed by lower) torso. This looked much more like the results from the previous behavioral studies than it did in terms of the physical body, while the differences between the VR setups had disappeared. The template task largely replicated study one, with pointing being to the upper torso most,
followed by the regions of the face. It was concluded that people mostly localize themselves
in the (upper) face and the (upper) torso. Moreover, that, for the interpretation of where
people locate themselves, when using VR setups, it is important to take into account the
occurring inaccuracies in body part localization.
In study three, an individually scaled and gender-matched self-avatar, animated by the
tracked movements of the participant and seen from 1PP (co-located) and a 3PP mirror-view), was implemented in the HTC Vive to provide rich feedback about the participant's body in a VR headset. Two groups of participants performed the VR self- and body part localization tasks, before and after an avatar adaptation phase where the self-avatar was experienced from either (normal) eye-height, or from chest-height.
The self-avatar as such did not reduce inaccuracies in body part localization. Changing the
viewpoint did alter body part localization, though. Pointing to body parts was overall shifted
upwards (more for the lower body parts) from the pre- to the post-test for the chest-height
group, but not for the eye-height group. The self-avatar as such, nor changing the viewpoint,
changed self-location, though. No evidence was found for experienced self-location being
manipulated towards the viewpoint location. A non-significant trend towards higher self-location was present for the chest-height group on the contrary, which might be due to body parts being perceived higher than normal. It was concluded that experienced body part
locations might be more plastic (influenced by viewpoint) than experienced self-location.
The differences between the self-localization results from the VR and the template tasks
are debated and might be due to the 3PP pointing in the template task resembling pointing
to someone else or even an external object, rather than to oneself.
Taken together, this thesis suggests a differential involvement of multi-sensory information processing in our experienced specific self-location and our ability to locate our body parts. Self-localization seems to be less flexible, possibly because it is strongly grounded in the 'bodily senses', while body part localization appears more adaptable to the manipulation of sensory stimuli, at least in the visual modality.