human interactions and personal space in collaborative virtual environments
TRANSCRIPT
ORIGINAL ARTICLE
Human interactions and personal space in collaborativevirtual environments
Nasser Nassiri • Norman Powell • David Moore
Received: 14 October 2009 / Accepted: 20 August 2010 / Published online: 7 September 2010
� Springer-Verlag London Limited 2010
Abstract As humans start to spend more time in col-
laborative virtual environments (CVEs) it becomes
important to study their interactions in such environments.
One aspect of such interactions is personal space. To begin
to address this, we have conducted empirical investigations
in a non immersive virtual environment: an experiment to
investigate the influence on personal space of avatar gen-
der, and an observational study to further explore the
existence of personal space. Experimental results give
some evidence to suggest that avatar gender has an influ-
ence on personal space although the participants did not
register high personal space invasion anxiety, contrary to
what one might expect from personal space invasion in the
physical world. The observational study suggests that
personal space does exist in CVEs, as the users tend to
maintain, in a similar way to the physical world, a distance
when they are interacting with each other. Our studies
provide an improved understanding of personal space in
CVEs and the results can be used to further enhance the
usability of these environments.
Keywords Collaborative virtual environment � Anxiety �Personal space
1 Introduction
The study of personal space is referred to as ‘‘proxemics’’,
and was founded by Hall (1959). Personal space refers to
an invisible ‘‘bubble’’ that people carry around themselves
which expands and contracts depending on a number of
factors such as gender (Adler and Iverson 1974; Aiello
1987), culture (Vaksman and Ellyson 1979), age (Hayduk
1983), and relationship (Allegier and Byrne 1973). Per-
sonal space seems to be an important factor in interactions
in the physical world as it functions as a comfort zone
during interaction (Dosey and Meisels 1969; Knapp 1978;
Hall 1959) and it tends to indicate the type of relationship
between interacting individuals (Hall 1963). Sommer
(2002) found that personal space has been used in the
design of offices, stores, banks and other building types.
Sommer also notes that the American space agency
(NASA) used personal space research to enhance living in
the space station. Further, Argyle and Dean’s (1965)
equilibrium theory of an inverse relationship between
mutual gaze and interpersonal distance has been re-visited
by Bailenson et al. (2001, 2003) in a virtual setting. Their
studies suggest that consistent gazing leads recipients of
the gazing to seek more interpersonal distance, especially
in the case for female participants.
A personal space ‘‘invasion’’ is said to occur when
someone trespasses into another person’s self-boundaries
or personal space (Sommer 1969). Felipe and Sommer
N. Nassiri (&)
Department of Information Technology,
Higher Colleges of Technology, Dubai Women’s College,
P.O box 16062, Dubai, United Arab Emirates
e-mail: [email protected]
N. Powell
Centre for Excellence in Enquiry-Based Learning (CEEBL),
The University of Manchester,
P.O. Box 88, Manchester M60 1QD, UK
e-mail: [email protected]
D. Moore
Innovation North: Faculty of Information and Technology,
Leeds Metropolitan University,
Leeds LS6 3QS, UK
e-mail: [email protected]
123
Virtual Reality (2010) 14:229–240
DOI 10.1007/s10055-010-0169-3
(1966) found that tension levels increase hugely when
personal space is invaded. They suggest that the responses
to personal space invasion fall into two categories: block-
ing tactics (e.g. gaze averting) and anxiety reduction
responses (e.g. hair-pulling and foot-tapping). Similarly,
Hayduk (1981) found a linear relationship between intru-
sion of personal space and discomfort, and Sawada (2003)
found a significant change of the heart rate of participants
while they were approached by a stranger.
Our research is concerned with how, if at all, such
proxemics phenomena operate in virtual environments. A
virtual environment (VE) is a software system that creates
the illusion of a world that does not exist in reality. VE can
be broadly categorized into immersive and non immersive.
An immersive VE is an environment where the user is fully
immersed in it such that all his sensory input comes from
the VE. This requires the tracking of user head and body
movement through sensors. A non immersive virtual
environment uses a conventional computer monitor to
display the 3D graphics of the virtual world on standard
computers, requiring no special graphic hardware cards.
Users of these environments typically interact with the
virtual world using the keyboard and a mouse and some-
times microphone. Other input devices, such as joysticks
and 3D mice, may also be used. A Collaborative Virtual
Environment (CVE) is a special case of a virtual environ-
ment system which allows for multiple simultaneous users
and enables them to communicate with each other via their
‘‘avatars’’; the emphasis may be more on collaboration
between users than on simulation.
Increasingly, interactions are taking place in such non
immersive virtual environments. CVEs are being used to
support training (Oliveira et al. 2000), education (Corbit
and DeVarco 2000; Johnson et al. 1999) and community
activities (Lea et al. 1997), and it has been argued that CVE
technology offers a potentially powerful training tool for
people with autism (Cobb et al. 2000; Fabri and Moore
2005). An important issue therefore is whether personal
space retains its importance in such environments. This
paper reports on an experiment and an observational study
that we have carried out to investigate this issue.
2 Personal space in the physical world and in CVEs
Human beings have an awareness of an appropriate dis-
tance to be kept between themselves during various kinds
of interaction. The regulation of this distance is referred to
as ‘‘proxemics behaviour’’ and has been studied within
social psychology. Hall (1966) argued from observational
evidence that individuals during interaction use one of four
personal space zones. These zones are intimate (0–45 cm),
personal (45 cm–1.20 m), social (1.20–3.60 m), and public
(3.60 m onward); the first three of these zones are illus-
trated in Figs. 1, 2 and 3.
According to Hall (1966), the particular zone that people
use depends on several factors which can be grouped into
two different categories, environmental and situational.
The environmental factors concern aspects of the envi-
ronment layout such as room size and shape, location of
the room (Cochran et al. 1984), room height (Cochran
and Urbanczyk 1984) and room illumination (Adams and
Zuckerman 1991).
The situational factors that influence personal space are
considered to be gender, age, culture, and relationship. For
gender, several researchers have demonstrated that,
Fig. 1 The intimate distance (0.3 m)
Fig. 2 The personal distance (1.20 m)
230 Virtual Reality (2010) 14:229–240
123
compared to men, women maintain less space between
themselves (Adler and Iverson 1974; Aiello 1987; Gifford
1996; Klinge 1999). For age, Hayduk (1983) found a linear
relationship between age and interpersonal distances, the
younger a person is, the smaller personal space he/she
needs. For culture, Vaksman and Ellyson (1979) have
suggested that different cultures tend to have different sizes
of personal space bubbles. For example, Middle Eastern
peoples tend to accept closer distances than people from
Britain (Hall 1959). Concerning relationship, Allegier and
Byrne (1973) found that interpersonal distance depends
upon the individuals’ relationship; if they like each other or
they are friends then they tend to interact at a closer dis-
tance. Unlike the physical world, few researchers have
studied personal space in virtual environments. The results
from these researchers however do offer preliminary evi-
dence to suggest that personal space exists in both non
immersive and immersive environments.
Concerning non immersive virtual environments,
Becker and Mark (1998) conducted a study comparing the
social conventions for greeting, communication, and per-
sonal space. They found evidence to suggest that personal
space exists and that CVE users keep a certain distance
during interactions. Krikorian and colleagues (2000) also
showed that the notion of personal space exists in CVEs.
They found that a certain distance is kept and that inva-
sions of such distance produce anxiety with attempts to
re-establish the preferred distance. Jeffrey (1998) spent
approximately 25 h in ActiveWorlds, a non immersive
CVE, and recording observations of interactions between
inhabitants of this virtual environment. He noticed that
communicating avatars maintain a physical distance
between themselves. Jeffery also noticed that individuals
tend to show feelings of discomfort and anxiety in their
reactions to violation of this personal space. Similarly,
Taylor (2002) found that users often move their avatars if
their personal space has been invaded.
These results are also borne out in immersive virtual
environments: Bailenson and colleagues (2005) found that
the concept of personal space exists in an immersive virtual
environment and they used it to measure the sense of being
with someone else in the virtual environment (i.e. co-
presence). Steed and colleagues (2005), in their study about
interaction in CVE, noticed that participants usually try to
avoid walking through each other, although there were
occasional collisions with objects and their partner. How-
ever, this situation (i.e. collisions) was often quickly rec-
tified, by reversing or stepping out of the way.
The limited evidence to date suggests, then, that per-
sonal space does exist in CVEs. Little research has thus far
addressed the issue of what influences this personal space.
One exception to this is a study by Yee and colleagues
(2007), which uses an automatic script to capture: the
interpersonal distance; gender; direction of gaze; location
indoors or outdoors; and whether or not they are talking, of
hundreds of dyads of avatars interacting in Second Life, a
non immersive CVE. To begin to address the issue of what
factors influence personal space in virtual environments,
we carried out an experiment concerned with the possible
influence of avatar gender on personal space.
3 The avatar gender experiment
Several researchers have argued that proxemics behaviour
in the physical world differs for men and women. Gifford
(1987) found that males interacting with other males
require the largest interpersonal distance, followed by
females interacting with other females, and finally males
interacting with females. Similarly, Hewitt and Henley
(1987) found that men allow women to invade their per-
sonal space to the highest degree, followed by women
allowing other women, then men allowing men, and finally
women allowing men to invade their space the least.
Whilst gender clearly influences personal space in the
physical world, its influence on personal space in CVEs is
of interest. Given this, an experiment was conducted to
investigate the effect of avatar gender on personal space in
a CVE. In this experiment, participants of both genders had
their avatars’ personal space ‘‘invaded’’ by another avatar
(of either the same or the opposite gender), and reported
their anxiety levels through the use of a post experiment
Fig. 3 The social distance (3.40 m)
Virtual Reality (2010) 14:229–240 231
123
questionnaire. The questionnaire was designed to assess
personal space invasion anxiety level (PSIAL) in the CVEs
(Nassiri et al. 2005); PSIAL is defined as the degree of
anxiety generated from an invasion of someone’s personal
space. The questionnaire has been constructed to address
issues corresponding to the factors seen as underlying
anxiety in the physical world, namely threat (Abbey and
Harnish 1995), discomfort (Patterson et al. 1971), and
flirtation and attraction (Abbey and Harnish 1995). Each
factor (i.e. threat, discomfort, and flirtation) was allocated
six questions. A fuller justification of the questionnaire is
given elsewhere (Nassiri 2006) and a copy is available
from the corresponding author.
The experiment involved 40 participants (20 males and
20 females), each of whom had their avatar’s personal
space ‘‘invaded’’ by the avatar of a further participant—a
‘‘confederate’’—who was acting under instructions from
the researchers. The participants were Lebanese students
from the same university in Lebanon (i.e. American Uni-
versity of Beirut—AUB) and were all between 20 and
23 years of age. The experiment also involved 10 con-
federates (5 males and 5 females) which were also drawn
from the same university but were unknown to the partic-
ipants. The participants were divided into eight groups of
five participants each; five of these groups consisted only
of male participants, the other four only of female partic-
ipants. Each member of the group was embodied into the
virtual environment by using either a male avatar or female
avatar depending on the participant’s actual gender.
The confederates were also divided into two groups of
five; one group consisted only of males, the other only of
females. Each member of the confederate group was
assigned one of the avatars, depicted in Fig. 6, based on the
confederate’s actual gender. The confederate and partici-
pant avatars chosen wore typical American University of
Beirut (AUB) student attire, which is the attire mainly used
in the participants’ cultural context. The genders of the
avatars are clearly apparent by their appearance and would
be instantly recognised by the participants.
The invasions took place in a non immersive virtual
environment built specifically for the experiment by the
authors, using ActiveWorlds (www.activeworlds.com).
Activeworlds is a VE that is used by a large number of
users who can communicate through text messages. The
environment does not support facial expression or eye gaze,
and there is no collision detection mechanism. Figure 4
shows a room in this environment and Figs. 5 and 6 show the
avatars adopted by the participants and confederates,
respectively. To standardize the behaviour between the
confederates, a script and action plan were devised that all
the confederates followed when they interacted with the
participants’ avatars. While the confederate avatar and the
participant avatar were touring the house, the confederate’s
avatar maintained a ‘‘social distance’’ from the participant
avatar. Then the confederate asked the participant to read a
sign in the virtual house. When the participant avatar was
reading the sign, the confederate moved his/her avatar in
front of the participant avatar, initially to the ‘‘personal
zone’’ and then to the ‘‘intimate zone’’ of the participant, and
stayed in that position for exactly 10 s.
The literature review in Sect. 2 showed that there are
several factors that influence personal space in the physical
world such as age, culture, and relationship. These factors
may influence the outcome of the experiment, confounding
the results with the influence of gender, the factor under
investigation. Consequently, the following measures were
taken to control these factors in the design and implemen-
tation of the experiment. The relationship variable (i.e. the
pre-knowledge between the participants and confederate)
was controlled by ensuring that they met for the first time in
Fig. 4 The virtual environment used for study 1
Fig. 5 Male and female avatars used by participants
232 Virtual Reality (2010) 14:229–240
123
the collaborative virtual environment. This was accom-
plished by selecting the confederates and the participants
from two different departments at a local university.
Therefore, the expectation was that the experimental par-
ticipants had no prior relationships with each other. In
addition, the confederates and participants were unaware of
each others’ identities during the experiment.
To control the age and culture variables, the partici-
pants, drawn from Lebanese students, were chosen to be
from the same culture and with the same age range
(between 20 and 23 years). Regarding the culture, even
though there were inevitably cultural differences, the prior
knowledge of the group by the experimenter suggested that
such differences were minimal. The underlying user gender
variable was controlled by providing the participants with
avatars which were the same as their own gender.
The descriptive statistics of anxiety of each gender
invasion group are shown in Table 1 where, for example,
‘M-m’ indicates a male avatar invading another male
avatar’s space, and ‘M-w’ indicates a male avatar invading
a female avatar’s space. In this table, the anxiety means of
all the groups are either negative or near zero—an
indication that the participants in general did not register
anxiety when invaded, and indeed unexpectedly tended to
be positive about it.
The analysis of variance (ANOVA) for the PSIAL vari-
able (Table 2) shows that there is a statistically significant
difference in anxiety level among the groups of differ-
ent gender combinations of invasions (F(3,36) = 4.599,
p = 0.008, i.e. p \ 0.01). However, although the ANOVA
test shows that there is a difference in PSIAL among the
group means; it does not indicate which groups are signifi-
cantly different from each other. Therefore, LSD (Least
Significant Difference) post-hoc comparisons were also used
and the results are presented in Table 3. The table shows that
the adjacent groups in the order of their PSIAL mean (M-m,
W-w, W-m, and M-w) are not significantly different from
each other. However, groups that is more than one position
away in the order of their mean PSIAL are significantly
different from each other. Specifically, the PSIAL of the
group M-m is significantly higher than the PSIAL of
the groups W-m (p \ 0.05) and M-w (p \ 0.01), and also the
PSIAL of the group W-w (p \ 0.05) is significantly higher
than the PSIAL of the group M-w (p \ 0.05; Fig. 7).
The fact that the 95% CI for three of the four groups
includes zero, suggests that we cannot differentiate their
response from a neutral response. The fourth group, man
invades woman, is significantly negative (i.e. the 95% CI
does not cross zero) suggesting a positive emotional
response to invasion.
Summarising the results of this experiment: (a) avatar
gender combination had an influence on personal space in
the CVEs, (b) the ranking of avatar gender combination
groups had a striking difference from that observed for
personal space invasion in physical environments (Hewitt
and Henley 1987), and (c) the participants in general did
not register high anxiety when their personal space was
invaded, as might be expected from personal space inva-
sion in physical environments which tends to cause anxiety
and discomfort (cf. Section 2).
4 An observational study
The evidence from the experiment reported above suggests
that, contrary to expectations, the participants did not
Fig. 6 Male and female confederate avatars
Table 1 Descriptive statistics
of anxiety and gender
combination
N Mean Std. dev Std. error 95% Confidence interval for mean Min Max
Lower bound Upper bound
M-m 10 2.10 8.36 2.64 -3.88 8.08 -11 17
W-w 10 -0.90 7.81 2.47 -6.49 4.69 -14 13
W-m 10 -6.50 10.80 3.42 -14.23 1.23 -25 8
M-w 10 -10.50 5.36 1.69 -14.33 -6.67 -21 -4
Virtual Reality (2010) 14:229–240 233
123
register high personal space invasion anxiety. This raises
an important question about the very existence of personal
space in CVEs. We investigated this question by adopting a
different approach: an observational study. The study
involved observing the interpersonal distance adopted by
CVE users in their normal use of the CVE. The adoption of
this approach i.e. observational rather than experimental
approach was based on the view of Schroeder (2002) that
since human interaction in CVE is a new research area it
should be investigated by both experimental and observa-
tional methods.
The observation carried out as part of the study spanned
a period of 4 weeks for approximately 3 h per day, a total
of around 80 h (Nassiri 2006). Results of the observation
showed that the majority of the observed participants
maintained a social distance during interactions. Figures 8
and 9 illustrate two observed encounters and are typical of
the distance that was kept between avatars, irrespective of
whether the avatars are of a different or the same gender,
during conversation.
It was also observed that CVE users rarely invaded the
personal space of each other. Further, the users did not go
though each other when they were moving in the envi-
ronment, even though ActiveWorlds allows this in its
interaction model. They were observed to go around other
avatars instead of going though them. This may suggest
that CVEs users may be respecting the personal space of
others, or it may be that they are being careful to protect
their own personal space, or a combination of both. Thus,
the observations give some evidence that personal space
has been transferred to collaborative virtual environments.
During the observational study, some inhabitants of the
environment had their (presumed) personal spaces invaded
by the researcher. These inhabitants were then invited to
take part in an informal interview with the researcher.
Eighteen such interviews were conducted. The rationale for
this was that it would generate data about reactions to a
personal space invasion, and such data would not be
available if CVE users do not invade the personal space of
others during the time of the observation.
The interviews aimed to address the underlying reasons
behind any reaction to an invasion of personal space (why,
for example, the user moved his avatar away from the
confederate avatar). The interviews were conducted in an
informal manner, and the precise nature of the questions
asked varied according to what was seen as appropriate to
the specific circumstances of each user. Full transcripts are
available from the corresponding author.
The majority of the interviewed users reacted to their
personal space being invaded, through either verbal or non-
Table 2 The ANOVA for the anxiety level
Sum of squares df Mean square F Sig.
Between groups 953.100 3 317.700 4.599 0.008
Within groups 2,486.800 36 69.078
Total 3,439.900 39
Table 3 The LSD Post Hoc test
(I)
group
(J)
group
Mean
difference
(I-J)
Std.
error
Sig. 95% Confidence interval
Lower bound Upper bound
M-m W-w 3.00 3.72 0.425 -4.54 10.54
W-m 8.60a 3.72 0.027 1.06 16.14
M-w 12.60b 3.72 0.002 5.06 20.14
W-w W-m 5.60 3.72 0.141 -1.94 13.14
M-w 9.60* 3.72 0.014 2.06 17.14
W-m M-w 4.00 3.72 0.289 -3.54 11.54
a The mean difference is significant at the .05 levelb The mean difference is significant at the .01 level
Fig. 7 Anxiety means
measured by the PSIAL
questionnaire and 95%
confidence intervals (CI) for
gender combinations
234 Virtual Reality (2010) 14:229–240
123
verbal behaviours: moving back, vanishing from the scene,
displaying aggression, offence or nervousness, and giving
verbal advice concerning where to stand in CVEs.
Participants were asked whether they felt the degree of
realism of the CVE would affect their response to PSI.
Most of the participants in the interviews suggested that
their experience of personal space invasion would not be
influenced by the level of realism of the CVE in general.
They suggested that irrespective of the environment and
avatar representation, they still prefer their personal spaces
not to be invaded. One interviewee expressed clearly that
the person who is invading is the human behind the avatar,
‘‘in general it’s rude to stand close no matter what the
environment and avatar look like’’. Another user went
further and said explicitly about the degree of realism of
the avatar:’’no matter what the avatar looks like, personal
space is still there, your pixilated representation could be
anything, a person, a box, a car… just maintain your
distance as you would in reality. This is somewhat at odds
with results of Bailenson et al. (2003) who found that
realism was a factor in determining the size of personal
space bubbles. A possible explanation of the difference is
that Bailenson et al.’s study was carried out in an immer-
sive environment, in contrast to the non-immersive envi-
ronment of our study.
5 Discussion
We have outlined an experiment, an observational study,
and user interviews concerned with personal space in CVE.
The experimental results suggest that the gender being
portrayed by an avatar does seem to influence personal
space in the CVE. Interestingly, however, the results sug-
gest a different order of gender combinations from those
found in the physical world by Hewitt and Henley (1987).
The rank order of anxiety level amongst participants in our
CVE experiment from highest degree of anxiety to lowest
degree of anxiety was as follows: (1) male invading male,
(2) female invading female, (3) female invading male, (4)
male invading female. In particular, unlike the physical
world, in which a male invasion of a female’s personal
space is highest in terms of anxiety level, in our CVE
environment it has turned out to be the lowest. Establishing
the factors behind this apparent discrepancy between the
physical and the virtual would require further empirical
investigation. In the meantime, a plausible speculation is
that the key factor is the diminished risk of physical harm
in the virtual encounters (Bailenson et al. 2003). This may
also explain why our results show that participants overall
do not register anxiety, and in some cases report nega-
tive anxiety. However, our results are in line with both
results in physical environments and Yee and colleagues
results in virtual environments, that male dyads have a
greater interpersonal distance than female dyads, which
in turn have a greater interpersonal space than mixed
dyads.
Results from our observational study, however, suggest
that despite this apparent lack of anxiety caused by PSI,
personal space in CVEs does exist: individuals maintain, in
a similar way to the physical world, a distance when they
interact. The observation data suggests that the majority
of CVE users, irrespective of their gender, maintained a
certain distance during interpersonal communications
(typically 3ft or 0.91 m range). It was also observed that
CVE users did not go through the each others’ avatars; they
went around each other instead. These results parallel
findings from researchers in the physical world (Hayduk
1983) and also are in line with other researchers in CVEs
(Jeffrey and Mark 1988) who found that individuals in both
worlds maintain a personal distance between themselves
during interactions.
Further, the data from the interviews suggests that most
of the users, whether they are novice or expert, did not
like their personal space to be invaded. Their reactions to
Fig. 8 The distance observed between male visitors’ avatars
Fig. 9 The distance observed between citizen male and female
avatars
Virtual Reality (2010) 14:229–240 235
123
personal space invasion were varied and expressed non-
verbally and verbally. Seven participants moved their
avatars back. This finding parallels results of many
researchers who have studied personal space in the physi-
cal world (Hayduk 1981; Worchel and Teddlie 1976). In
addition, five participants vanished from the world after the
invasion. It may well be that the ‘‘vanishing event’’ from
the world was brought about by the invasion, a similar
reaction to ‘‘flight from the scene’’ as a response to per-
sonal space invasion in the physical world (Felipe and
Sommer 1966). Three participants expressed verbally dif-
ferent responses to personal space invasion which can be
classified as aggression, offence, and nervousness. Three
participants however appeared not to be concerned by
personal space invasion. For example, one user said ‘‘np
[no problem] for me standing too close. it’s just VR and
nothing gona happened u can’t hurt me’’. Overall, though,
the majority of the interviewed users reacted to their per-
sonal space being invaded through either verbal or non-
verbal behaviours.
However, caution is needed when considering our
results, for at least three reasons. First, because our data
collection method involved a post-experiment question-
naire, participants were not able to express their feeling at
the time of the invasion, rather they reported it after the
experiment, and as such they may have forgotten the exact
nature of the feeling when it occurred. A second reason for
caution concerns the features of the specific CVE we used
in our study. ActiveWorlds has no collision detection
facility. Indeed, ‘‘collisions’’ did occur during the experi-
ment. When this situation occurred, the confederate avatar
went back and forth until the intimate view was achieved,
and this might require several moves. It could be that this
situation reduced the degree of immersion felt by the par-
ticipants in the CVE and therefore lead to an absence of
anxiety.
Similarly, ActiveWorlds users can communicate only
through text messages—voice messages are not supported.
This might also affect the results as the participant might
be busy with typing at the keyboard when the confederate’s
avatar was invading his/her avatar personal space and thus
may not notice such an invasion, especially if the partici-
pant’s touch typing skill is poor. The switching of attention
between the screen and the keyboard may also reduce the
participants’ degree of immersion in the environment.
Again, the interaction model of ActiveWorlds does not
support facial expression, nor voice messages which are
also essential factors that transmit many signals in the
physical world. There are also some methodological rea-
sons for treating the results with caution. First, the instru-
ment used to measure anxiety in the experiment was a
questionnaire designed by the authors; independent vali-
dation of the questionnaire would strengthen the results.
Another reason for caution is that the participants in the
experiment were all Lebanese and results might therefore
be culturally specific. A further reason for caution is that
some of the CVE users interviewed in the observational
study vanished from the world; although it can be assumed
intuitively that this was a reaction to personal space inva-
sion, these people did not come back and hence it was
impossible to confirm this assumption. Nevertheless,
despite these concerns, the overall results provide an
improved understanding of personal space in CVEs and can
be used to suggest lessons for CVE users and designers.
A lesson for CVE users concerns a protocol of behaviour
that they should follow during their stay in the CVE.
Humans in the physical world seem to know that, for
example, they should not stand too close to other people.
Intuitively, it can be suggested that it is unlikely that people
will have the same level of knowledge concerning appro-
priate personal space in a virtual world. The research
reported in this paper may therefore be able to help
inhabitants of CVE by suggesting some explicit protocols
that users should seek to observe when in the CVE—a kind
of ‘‘Highway Code’’ for CVE use. The proposed behavioural
code list with regards to personal space is as follows (see
Nassiri 2006 for more discussion):
1. Keep a distance between you and others during
interaction. The distance should be approximately
equal to the length of your avatar’s arm. This distance
should be larger when you are approaching an avatar
with the same gender as you.
2. Never get too close to someone else’s avatar unless
you are sure that the other user is happy with such
proximity.
3. Never walk though someone else’s avatar.
4. Do not be offended if someone asks you to move away
from his/her avatar or if he/she moves back if you are
too close.
A lesson for CVE designers is that personal space in the
CVE should be protected in some way. A proportion of the
users involved in the experiment felt anxiety when their
personal space was invaded, and the majority of the
interviewed participants did not like their personal space to
be invaded. There is, then, enough evidence of user con-
cern regarding personal space invasion to warrant protec-
tion. This might seem to suggest that CVE designers should
try to build CVE in such a way that the possibility of
personal space invasion is wholly eliminated. On the other
hand, however, personal space invasion did not appear to
cause anxiety to some CVE users, and was actually wel-
comed by some. Eliminating personal space invasion might
disadvantage these users. A possible solution to this
dilemma is that CVE designers should provide the users
with a mechanism to control or manage personal space
236 Virtual Reality (2010) 14:229–240
123
invasion events, rather than seeking to eliminate them. One
way of effecting such control is that CVE designers design
virtual environments where fine grained and easy avatar
movement is provided to users, so that (a) they can easily,
e.g. via one mouse click, move their avatar away to adopt a
new comfortable interpersonal distance and (b) they do not
inadvertently invade the personal space of other users.
However, although these proposed avatar movement
facilities may reduce personal space invasions, they are
unlikely to completely control it; a determined intruder
may repeatedly invade the personal space of others despite
these avatar movement facilities.
An additional mechanism that allows each user to
competently control his or her personal space might
therefore be needed. Such a mechanism, suggested by
Jeffrey and Mark (1988), might be to allow the users to
surround themselves by ‘‘bubbles’’ to prevent accidental or
deliberate personal space invasion. The argument for such
a mechanism is that (1) many researchers define personal
space in the physical world as an area with invisible
boundaries surrounding individuals which functions as a
comfort zone and (2) various analogies have been used to
describe personal space in the physical world, and one of
these is a bubble. The shape of such a bubble can be
designed in line with the evidence from research con-
cerning personal space in the physical world, e.g. the evi-
dence of Argyle (1988), who identified the shape of the
personal space as almost circular but with more space in
front. Bailenson et al. (2003) found evidence of a similar
personal space bubble in their study of an immersive vir-
tual environment. This would yield a bubble such as that
depicted in Fig. 10.
The proposed bubble raises some interesting issues that
need to be investigated: Should the user be able to activate/
deactivate these bubbles, or are they automatically acti-
vated when the user logs on and stay surrounding the users
until they log out of the CVE? Would it be better to have
the bubble visible or invisible to the users? Who controls
the size of the bubble, should it be automatically set up to
be small or big based on the user’s culture and gender, or is
it better to be of fixed size? Each of these issues will now
be discussed in turn.
5.1 Bubble activation
It is suggested that it is unreasonable that the bubbles be
activated automatically when users log in and stay sur-
rounding them until they log out, because sometimes users
need to come closer to do certain tasks such as reading
small characters on an avatar t-shirt. Further, users who
might be unconcerned about, or even welcome, PSI might
be disadvantaged by permanent bubbles. Therefore, the
recommendation is to let users control the activation/
deactivation of their bubbles themselves. For example, the
bubbles can be activated during an invasion that causes
anxiety and deactivated during an invasion that is
welcomed.
5.2 Bubble perceptibility
Bubbles can in principle be designed to be either visible or
invisible to the users. In either case, the same collision
detection mechanisms could be used. When a collision is
detected, the invading avatar is not able to move any clo-
ser, unless the bubble is removed or its size reduced, see
Fig. 11. However, visible bubbles seem intuitively to look
odd and differ from the physical world. On the other hand,
there is an affordance problem with invisible bubbles in
that users who inadvertently or purposely try to get closer
to other avatars might not know why they cannot move any
closer, e.g. they may think that there is something wrong
with the mouse, with the virtual environment, or with the
internet. A possible solution is to keep the bubble invisible
until an avatar comes to within ‘touching distance’ of the
bubble; at this point the bubble can become visible and stay
visible until the avatar moves away. An alternative
approach would be for a pop-up message to appear when
an avatar is about to go through another user’s invisible
bubble. These approaches can be expected to solve the
affordance issue, though the issue of difference from the
physical world would remain.Fig. 10 An avatar with a bubble
Virtual Reality (2010) 14:229–240 237
123
5.3 Bubble size
It might be suggested that a CVE designer should design
the bubble so that, as in the physical world, its size is
related to the culture of the user. Numerous studies have
examined personal space in different cultural contexts and
found that in the physical world individuals’ culture
influences their interpersonal distances. However, design-
ing the bubble sizes based on culture would require an
investigation into whether culture retains its influence on
the size of personal space in CVEs. An alternative would
be to allow the user to adapt the size of their own bubble,
so that through trial and error they find a distance that they
are comfortable with or continue to change their bubble
size with context.
6 Conclusion and further work
The research discussed in this paper has investigated some
aspects of personal space in CVEs through an experiment,
an observational study, and interviews. The key findings
are summarized in Table 4. There are several interestingFig. 11 Collision between bubbled and not bubbled avatars
Table 4 Research results in brief
Investigated
issue
Current study’s results Other researchers’ results in CVE Research results in the physical world
Existence of
personal
space
Personal space exists in CVE
(though not for all users).
Personal space exists (Jeffrey 1998) Personal space exists in the physical
world (Sommer 1969)
The
influence
of gender
on
personal
space
Avatar gender influences
personal space in CVEs
Gender influences the distance avatars
stand apart (Yee et al. 2007)
Gender influences personal space
(Adler and Iverson 1974; Aiello 1987)
Anxiety
ranking
(from
highest to
lowest):
Male invades male Male dyads have greater distance than
female dyads which have greater
distance than mixed dyads (Yee et al.
2007)
Male invades female
Female invades female Male invades male
Female invades male Female invades female
Male invades female Female invades male (Hewitt and Henley
1987)
Reaction to
personal
space
invasion
The experimental study suggests that the
reactions to personal space invasion by
the majority of the participants are not
anxiety.
Anxiety and discomfort. Adjusting
position and leaving the environment
(Krikorian et al. 2000, Jeffrey 1998)
Anxiety and discomfort. Adjusting
position and leaving the scene. (Worchel
and Teddlie 1976, Becker and Mayo
1971; Efran and Cheyne 1974, Felipe
and Sommer 1966)The observational study shows that the
reactions to personal space invasion
vary: adjusting position, moving back,
leaving the environment, aggressive
verbal response, no concerns
Avatar
degree of
realism
Users reported that this would not
influence personal space
Durlach and Slater (2000) found that even
avatars with rather primitive expressive
abilities cause strong emotional
responses in people in a CVE.
Not applicable
238 Virtual Reality (2010) 14:229–240
123
ways in which the work could be carried forward. One
important aspect of further work involves factors that
influence personal space in the physical world and that
have not been investigated in the current research, for
example, culture (Aiello 1987), age (Hayduk 1983), and
relationship (Allegier and Byrne 1973).
The results obtained in this research may to some extent
be peculiar to ActiveWorlds, rather than CVEs in general.
Therefore, further research investigating other CVEs would
be valuable. However, the current obtained results remain
relevant to any CVE which adopts similar interface policies
to ActiveWorlds.
The experiment and the observational study in this
research involved avatars of defined genders. However, the
existence of avatars with indeterminate gender is a possi-
bility and might be a factor affecting personal space in the
CVE. This would therefore be a useful area for empirical
investigation.
The participants in the experimental study in general did
not register anxiety when their personal space was invaded
and this might be due to the fact that they do not identify with
their avatar. Hence, further work investigating personal
space of users who already have established for themselves
specific avatars would be interesting. Empirical work re the
proposed bubbles and ‘‘Behaviour Code’’ is also needed.
Much remains to be done, therefore, but as people spend
increasing amounts of time in virtual worlds, issues such as
proxemics become increasingly important, and it is hoped
that this paper makes a contribution to the research of such
issues.
References
Abbey A, Harnish R (1995) Perception of sexual intent: the role of
gender, alcohol consumption, and rape supportive attitudes. Sex
Roles 32:297–313
Adams L, Zuckerman D (1991) The effect of lighting conditions on
personal space requirements. J Gen Psychol 118:335–340
Adler L, Iverson A (1974) Interpersonal distance as a function of task
difficulty, praise, status orientation, and sex of partner. J Percept
Mot Skills 39:683–692
Aiello J (1987) Human spatial behaviour. In: Stokols D, Altman I
(eds) Handbook of environmental psychology. Wiley-Inter-
science, New York
Allegier A, Byrne D (1973) Attraction towards the opposite sex as a
determinant of physical proximity. J Soc Psychol 90:213–219
Argyle M (1988) Bodily communication, 2nd edn. Methuen & Co.,
London
Argyle M, Dean J (1965) Eye-contact, distance and affiliation.
Sociometry 28:289–304
Bailenson B, Blascovich J, Beall C, Loomis M (2001) Equilibrium
revisited: mutual gaze and personal space in virtual environ-
ments. PRESENCE Teleoperators Virtual Environ 10:583–598
Bailenson B, Blascovich J, Beall C, Loomis M (2003) Interpersonal
distance in immersive virtual environments. Personal Soc
Psychol Bull 29:819–833
Bailenson B, Swinth K, Hoyt C, Persky S, Dimov A, Blascovich J
(2005) The independent and interactive effects of embodied
agent appearance and behavior on self-report, cognitive, and
behavioral markers of copresence in immersive virtual environ-
ments. PRESENCE Teleoperators Virtual Environ 14:379–393
Becker B, Mark G (2002) Social conventions in computer-mediated
communication: a comparison of three online shared virtual
environments. In: Schroeder R (ed) The social life of avatars:
presence and interaction in shared virtual environments.
Springer-Verlag, London, pp 19–40
Becker F, Mayo C (1971) Delineating personal distance and
territoriality. Environ Behav 3:375–381
Cobb S, Beardon L, Eastgate R, Glover T, Kerr S, Neale H, Parsons S,
Benford S, Hopkins E, Mitchell P, Reynard G, Wilson J (2000)
Applied virtual environments to support learning of social
interaction skills in users with Asperger’s syndrome. Digit
Creativ 13:11–22
Cochran D, Urbanczyk S (1984) The effect of availability of vertical
space on personal space. J Psychol 111:137–140
Cochran D, Hale W, Hissam C (1984) Personal space requirements in
indoor versus outdoor locations. J Psychol 117:121–123
Corbit M, DeVarco B (2000) Two 3-D implementations of CVE
science museums. In: Proceedings of CVE 2000 conference, San
Francisco
Dosey M, Meisels M (1969) Personal space and self-protection. J Pers
Soc Psychol 11:93–97
Durlach N, Slater M (2000) Presence in shared virtual environments
and virtual togetherness. Presence Teleoperators Viertual Envi-
ron 1:214–217
Efran M, Cheyne J (1974) Affective concomitants of the invasion of
shared space: behavioural, physiological and verbal indicators.
J Pers Soc Psychol 29:219–226
Fabri M, Moore D (2005) The use of emotionally expressive avatars
in Collaborative Virtual Environments. In: Proceeding of
symposium on empathic interaction with synthetic characters,
held at artificial intelligence and social behaviour convention
2005 (AISB 2005), Hertfordshire
Felipe N, Sommer R (1966) Invasions of personal space. Soc Probl
14:206–214
Gifford R (1987) Environmental psychology. Needham Heights, MA
Gifford R (1996) Environmental psychology. Allyn and Bacon,
Boston
Hall E (1959) The silent language. Doubleday, New York
Hall E (1963) A system for the notation of proxemics behaviour.
J Am Anthropol 65:1003–1026
Hall E (1966) The hidden dimension. Anchor books, Doubleday and
company, New York
Hayduk A (1981) The permeability of personal space. J Behav Sci
13:274–287
Hayduk A (1983) Personal space: where we now stand. Psychol Bull
94:293–335
Hewitt J, Henley R (1987) Sex differences in reaction to spatial
invasion. Percept Mot Skills 64:809–810
Jeffrey P (1998) Personal space in a virtual community. In:
Proceedings of a conference on human factors in computing
systems. ACM Press, New York, pp 347–348
Jeffrey P, Mark G (1988) Constructing social spaces in virtual
environments: a study of navigation and interaction. Workshop on
personalised and social navigation in information space. Swedish
Institute of Computer Science: SICS, Stockholm, pp 24–38
Johnson A, Moher T, Ohlsson S, Gillingham M (1999) The round
earth project: collaborative VR for conceptual learning. IEEE
Comp Graph Appl 19:60–69
Klinge J (1999) How different areas of personal space are protected: a
look at gender differences. Georgetown University Department
of Psychology
Virtual Reality (2010) 14:229–240 239
123
Knapp M (1978) Nonverbal communication, human interaction, 2nd
edn. Holt, Rinehart and Winston Inc, New York
Krikorian D, Lee J, Chock T, Harms C (2000) Isn’t that spatial?:
distance and communication in a 2-D virtual environment.
J Comp Mediated Commun 4
Lea R, Honda Y, Matsuda K, Hagsand O, Stenius M (1997) Virtual
society: collaboration in 3D spaces on the internet computer
supported cooperative work. J Collaborat Comput 6:227–250
Nassiri N (2006) Personal space invasion anxiety in desktop
collaborative virtual environment, PhD thesis, Leeds Metropol-
itan University, Leeds
Nassiri N, Powell N, Moore D (2005) Avatar gender and personal
space invasion anxiety level in desktop collaborative virtual
environments. Virtual Real J 8:107–117
Oliveira C, Shen X, Georganas N (2000) Collaborative virtual
environment for industrial training and e-commerce. In: Proceed-
ings of the workshop on application of virtual reality technologies
for future telecommunication systems, San Francisco
Patterson M, Mullens S, Romanao J (1971) Compensatory reactions
to spatial intrusion. Sociometry 34:114–121
Sawada Y (2003) Blood pressure and heart rate responses to an
intrusion on personal space. Jpn Psychol 45:115–121
Schroeder R (2002) Social interaction in virtual environments: key
issues, common themes, and a framework for research, chapter 1.
In: Schroeder R (ed) The social life of avatars: presence and
interaction in shared virtual environments. Springer, London
Sommer R (1969) Personal space: the behavioural basis of design.
Prentice-Hall Inc, New Jersey
Sommer R (2002) Personal space in a digital age. In: Bechtel RB,
Churchman A (ed). Wiley, Inc, New York
Steed A, Roberts D, Schroeder D, Heldal I (2005) Interaction between
users of immersion projection technology systems. In: Proceed-
ings of HCI 2005. Las Vegas
Taylor T (2002) Living digitally: embodiment in virtual worlds,
chapter 3. In: Schroeder R (ed) The social life of avatars:
presence and interaction in shared virtual environments.
Springer, London
Vaksman E, Ellyson S (1979) Visual and spatial behaviour among
US and foreign males: a cross-cultural test of equilibrium
theory. The Eastern Psychological Association, Convention,
Philadelphia
Worchel S, Teddlie C (1976) The experience of crowding: a two
factor theory. J Pers Soc Dist 34:30–40
Yee N, Bailenson JN, Urbanek M, Chang F, Merget D (2007) The
unbearable likeness of being digital: the persistence of nonverbal
social norms in online virtual environments. CyberPsychol
Behav 10:115–121
240 Virtual Reality (2010) 14:229–240
123