Content uploaded by Esa Rantanen
Author content
All content in this area was uploaded by Esa Rantanen on Oct 23, 2017
Content may be subject to copyright.
Eye contact and video-mediated communication: A review
Leanne S. Bohannon
a,
⇑
, Andrew M. Herbert
b
, Jeff B. Pelz
c
, Esa M. Rantanen
d,1
a
Rochester Institute of Technology, 18 Lomb Memorial Drive, Rochester, NY 14623, United States
b
Rochester Institute of Technology, Department of Psychology, 2323 Eastman Hall , 18 Lomb Memorial Drive, Rochester, NY 14623, United States
c
Rochester Institute of Technology, Carlson Center for Imaging Science, 3112 Carlson Hall, 54 Lomb Memorial Drive, Rochester, NY 14623, United States
d
Rochester Institute of Technology, Department of Psychology, 2353 Eastman Hall, 18 Lomb Memorial Drive, Rochester, NY 14623, United States
article info
Article history:
Available online xxxx
Keywords:
Video-conferencing
Eye contact
Communication
Eye-tracking
Display
Laptop
abstract
A relatively new form of human communication, video-conferencing has become more popular as video
technology improves and with increasing demands for real-time communication across greater distances.
The full effects of video-conferencing on human communication are still being explored. Video-confer-
encing is presumed to be a somewhat richer form of communication than email and telephone, but
not quite as informative as face-to-face communication. This review explores research into the influence
of eye contact on communication and how video-conferencing mediates both verbal and non-verbal
interactions. Facilitation of eye contact is a challenge that must be addressed so that video-conferencing
can approach the rich interactions of face-to-face communication.
Ó2012 Elsevier B.V. All rights reserved.
1. Introduction
In the history of human communication, video-conferencing is a
relatively new means of correspondence dating from the introduc-
tion of the AT&T Picturephone at the New York World’s Fair in
1964 [1]. However, the technology has advanced in leaps and
bounds in the last 10 years, making videoconferencing available
to anyone who can afford a modern laptop computer or a smart-
phone. In comparison to email, instant messaging and other elec-
tronically mediated communication, video-conferencing did not
become commonplace until very recently. It is now standard for
laptop computers and tablets to come equipped with a small cam-
era mounted above the monitor, Apple’s iPhones feature an appli-
cation called FaceTime, Skype (video-conferencing software) has
been purchased by Microsoft, and Cisco (a company that special-
izes in remote communication) has introduced
umi, a video-con-
ferencing system designed for home use. The recent revolution
could also be driven in part by an increasingly global economy
[2], family migration patterns, faster network connections, and
substantial improvements in video-conferencing technology such
as higher-resolution images and larger screens. Regardless of the
causes for its growing popularity, like other technology-assisted
communication media, video-conferencing reveals important as-
pects of human communication that should be considered in the
development of new applications afforded by ever-advancing
technology.
Along a continuum of communication media and their ability to
manage rich information, or information that most reduces uncer-
tainty and equivocation, face-to-face communication is the richest
medium that provides immediate feedback and conveys many cues
in the form of natural language [3]. Video-conferencing falls below
face-to-face communication but above the telephone in terms of
information richness. The importance of the view of the interlocu-
tor’s face is exemplified by the finding that including a still photo-
graph to electronic mail can enhance the intended message and
even prompt compliance with it [4]. Human communication is
complex and based upon a combination of verbal and non-verbal
cues to exchange information. Eye contact is arguably one of the
most important non-verbal cues in communication, and we will
examine its role in video-conferencing systems.
The unique properties of video-conferencing can impact our
visual and verbal communication along with our perceptions of
one another. Users of such systems also show substantial
preference for video-conferencing that facilitates eye contact [5].
Understanding these effects can assist in system design and in
mitigating obstacles to communication in further iterations of
the technology. Ensuring natural (or the best approximation
possible) eye contact in video-conferencing can increase the
effectiveness of this mode of communication.
2. Eye contact in communication
Gaze can be defined as directed looking at any object, person, or
direction. Eye contact is gaze directed at another’s eyes, and
mutual eye contact occurs when two people make eye contact
simultaneously. When two people are in casual conversation, eye
0141-9382/$ - see front matter Ó2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.displa.2012.10.009
⇑
Corresponding author. Tel.: +1 585 210 8513.
E-mail addresses: Leanne.S.Bohannon@gmail.com (L.S. Bohannon), esa.rantanen
@rit.edu (E.M. Rantanen).
1
Tel.: +1 585 475 4412.
Displays xxx (2012) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Displays
journal homepage: www.elsevier.com/locate/displa
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
contact occurs about 61% of the time, about half of which is mutual
eye contact [6]. People make eye contact 75% of the time while
listening and 41% of the time while speaking [6]. However, these
figures may vary based on the context of the conversation. While
it seems clear that people spend more total time looking at the
other while listening than speaking, the length of the gazes also
vary. In an experiment with conversational dyads, participants
made long gazes at their partner and looked away only briefly
when listening, but made shorter gazes of equal length both at
and away from their conversational partner while speaking [7].
Also, there is evidence that conversational partners tend to spend
approximately the same amount of time looking at each other,
adjusting their gaze based on the other’s gaze [7]. The impact of
gaze, and more specifically eye contact, on our social interactions
as humans is significant. While we rely on a number of nonverbal
cues to communicate information to others, eye contact seems to
stand out distinctly from the rest.
2.1. Biological and developmental considerations
There is evidence that humans become sensitive to gaze direc-
tion by around 5 months of age [8], and infants as soon as 48 h
after birth prefer faces that are gazing at them to faces that are
looking away [9]. For example, infants looked significantly longer
at photographs of faces whose eyes were looking straight ahead
over faces with averted eyes. Furthermore, calm 9–12-week-old in-
fants preferred a familiar face to a stranger’s face when the familiar
face was previously associated with both eye contact and sucrose
delivery [10].
The amygdala, a brain structure related to processing emotional
stimuli, is thought to be a key neural component of gaze detection;
a patient with an amygdala lesion gazed more at people’s mouths
and less at their eyes during conversation as compared to subjects
without an amygdala lesion [11]. Other cortical regions related to
visual processing have also been shown to respond differentially
for gaze direction. Area V5 of the visual cortex (part of the dorsal
visual pathway involved in motion perception and coding actions)
is more active when viewing images of faces whose eyes are
animated to make eye contact as opposed to faces whose eyes
are animated away from eye contact [12]. These results suggest
that eye contact is an integral (or ‘‘hard-wired’’) part of human
interaction.
2.2. Cultural, age, and sex differences
Eye contact during conversation varies from culture to culture,
however. Arabs, Latin Americans, and Southern Europeans make
more eye contact during conversation than Asians and Northern
Europeans [6]. The English prefer to stand farther away and have
unwavering eye contact while being attentive to what another is
saying while Americans tend to stand closer and occasionally
glance away from, and between, the two eyes of the person speak-
ing [13].
In Japanese culture eye contact is considered rude, and people
are taught to look at a person’s Adam’s apple instead of the eyes
[14]. The level of eye-contact avoidance in the Japanese culture
also depends on social rank – eye contact with a superior should
most definitely be avoided [14]. The different effects of these cul-
tural norms in communication are unknown. However, it is inter-
esting to note that the lack of eye contact in one culture can be just
as important a component of communication as the presence of
eye contact in another.
In conversation, females make more eye contact while talking
than do males [15]. The amount of gaze during conversation in-
creases from between the ranges 4–6 and 6–9 years of age. Gaze
decreases around 10–12 years of age, perhaps due to an increase
in self-consciousness during those years of development, and in-
creases again in adulthood [15].
2.3. Attention and memory
Gaze can be a powerful director of attention. To illustrate this,
one study presented participants wearing an eye-tracker with a
screen for 1500 ms displaying a fixation point between two
squares [16]. Then the fixation point changed color. The color
was a cue to the participant to look either to the left or the right.
The fixation point disappeared after 150 ms, and then a photo-
graph of a person’s face appeared after either 50, 100, or 150 ms.
The eyes of the face were looking either to the left, right, or center.
The face stayed on the screen for 150 ms. In the trials with the
50 ms interval there were more antisaccades (saccades looking in
the opposite direction of the color cue) when the eye direction
was incongruent with the color cue. That is, the gaze direction of
the face overrode the color cue instructions [16]. This is evidence
that our attention tends to follow the gaze of another. Similarly,
participants were faster to respond to a target when it was accom-
panied by a schematic face gazing in the same direction [17], pos-
sibly indicating that the tendency to follow another’s gaze is
reflexive in nature. Many illusions and card tricks work in part
by misdirecting the gaze of viewers. Participants viewing a trick
while being eye-tracked follow the gaze of the person performing
the trick, thereby missing critical information that would reveal
the trick [18]. The timing of gaze direction effects is consistent with
prioritized visual processing of this information.
Memory, too, is affected by gaze direction. When presented
with a prerecorded sales pitch for a soap product, participants
remembered more about the product in the condition where the
salesman made eye contact with the camera (the participant) than
when he made no eye contact [19].
2.4. Impression formation
Gaze direction can influence the likeability and attractiveness of
a person as perceived by another. When participants in a study
[20,21] viewed female faces that made a gaze shift toward them,
they rated the faces to be more likeable than when the eyes of
the face made an animated shift away from the participant. Males
rated female faces that made eye contact as more attractive than
female faces whose eyes looked away [21]. Couples that scored
higher on a questionnaire test measuring level of romantic love en-
gaged in more eye contact than couples that had a lower romantic
love score [22].
It also appears that believed eye contact, or what we are told
about the level of our own or another’s eye contact can affect our
perception of that person [22]. Kleinke et al. [22] had male–female
dyads engage in a conversation. After the conversation, the exper-
imenter told the participants that the amount of gaze of one of
them was either less, about the same as, or more than that of most
people. Partners alleged to have made less than average eye con-
tact were rated as less attentive. When males were told they gazed
less than average at their female partner, they rated her more
favorably. Conversely, females who were told they gazed at their
male partner more than average rated him more positively. When
males were told their female partner gazed an average amount,
they rated her as more relaxed. Conversely, when females were
told their male partner gazed an average amount they rated him
as less relaxed. Males who were told their female partners had less
than average gaze rated her as least attractive while females who
were told their male partner had more than average gaze rated
him least attractive. Both males and females who were told their
partner had more than average gaze rated him or her as being
more sincere [22]. Thus, regardless of the actual gaze of another
2L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
person, the amount of gaze we think someone is directing at us af-
fects our impression of him or her.
2.5. Other impacts of eye contact
Eye contact also plays an important role in other social interac-
tions. For example, the presence or absence of eye contact can af-
fect the perceived emotion expressed by the face. When presented
with photographs of neutral faces, half of which gazed straight
ahead (made eye contact) and half of which had averted gaze, par-
ticipants were more likely to judge neutral faces that made eye
contact as expressing anger or joy, whereas neutral faces with
averted gaze were more likely to be rated fearful or sad [23]. When
participants were presented with composite faces expressing both
fear and anger, if the gaze was averted the face was more likely to
be rated as fearful. When the composite faces were gazing at the
participant, the face was equally likely to be rated as either fear
or anger. Anger was rated as more intense when eye contact was
present and fear was rated as more intense when gaze was averted
[23].
In one unique experiment, an ‘‘invader’’ either stared or did not
stare as he tried to ‘‘invade’’ the pew of local church-goers [24].If
the invader stared, the worshipper was likely to slide down the
pew and make room for the person. However, in infrequent cases
when the worshipper stared back at the invader, he or she typically
did not make room for the invader. If the invader did not stare, the
worshipper was less likely to move. It was uncertain if staring in
this instance had to do with communicating aggression, or if its ab-
sence maintained a certain anonymity that made it easier for
church-goers to ignore the request for a seat [24], but in any case
underscores the importance of eye contact as nonverbal
communication.
Eye-contact plays a role in perceived trust. US Customs inspec-
tors were more likely to want to search a passenger in a mock
interview when the passenger diverted his gaze frequently [25].
This was true despite the fact that gaze avoidance was not corre-
lated with lying [25,26]. Eye contact impacts other social interac-
tions such as impression formation, emotion and compliance and
these affects will be examined later in this paper in the context
of video-conferencing.
3. Eye contact in video-conferencing
Feeling as if the person with whom you are conversing is look-
ing at your eyes is important; when given the option between
using a video-conferencing system that enables eye contact and
one that does not, people overwhelmingly prefer the system that
enables eye contact [5]. The geometry of video-conferencing influ-
ences the perception of eye contact. Current video-conferencing
technology is designed such that a camera sits atop a screen. The
conversation partner’s image appears on the screen and the camera
captures one’s face (and vice versa). For each conversation partner
it appears as if the other person is looking downward – but each
partner is looking right at his or her partner’s eyes on the screen.
This discrepancy between the camera’s view of the head and where
the person is gazing is a form of parallax. One would need to look
directly into the camera for there to be eye contact. This is not
what people do, because the intention is to look at the conversa-
tion partner’s face, not the camera.
3.1. Cone of gaze
Early research on gaze perception focused on finding the
threshold for what is considered eye contact. That is, how far can
gaze deviate from the eyes and still be considered eye contact?
Gamer and Hecht [27] coined the term ‘‘cone of gaze’’ to refer to
the ‘‘range of considerable width wherein a person feels looked
at’’ (p. 706).
Many studies examining gaze perception used an experimental
setup involving pairs of people, one called the looker and the other
called the observer [27–32] (see Table 1 for a summary). The looker
is a confederate to the experiment and looks at different points on
a scale invisible to the observer. The observer then makes some
type of judgment about where the looker is looking. These gaze
detection studies used different metrics for measuring the just
noticeable deviation in gaze; some studies use degrees of visual an-
gle and others use centimeters. When converted to degrees of vi-
sual angle the just noticeable deviation in gaze varies between
0.73°and 9.3°. This large variation between results is not surpris-
ing given the methodological differences in the studies. Thus, it
is perhaps not useful to compare results directly to one another
but instead to look for patterns; the important information to take
from research on gaze direction perception is that (1) people are
very sensitive to deviations in gaze from eye contact (2) except,
perhaps to downward gaze and (3) perception of gaze direction
is influenced by head orientation. All three of these points are rel-
evant to video-conferencing.
3.2. Gaze perception during video-conferencing
Some studies on gaze perception focus specifically on eye con-
tact perception and video-conferencing. These studies explore the
role of video and parallax on gaze perception.
Chen [32] setup a ‘‘recording studio’’ in which a camera was
placed behind a hole in the middle of a projector screen. Points
were drawn 1°apart on the screen along a horizontal, vertical,
and diagonal orientations indicating targets for the looker. The
downward direction covered 15°and the other seven lines of
points covered a range of 5°each to form a star-like pattern. The
observer looked at videos of the looker gazing at the different
points in random order. It was the observer’s task to indicate
whether the looker was or was not looking at him.
Observers were very sensitive to eye contact when the
looker gazed up, left, or to the right but less sensitive when the
looker gazed below the screen as shown in Table 1. When
the looker was looking down, observers were much less sensitive
to eye contact; the looker could look as far as 10°away before
the observer reported eye contact was lost. One key difference
between Chen’s [32] study and a video-conferencing system is that
participants in Chen’s study made judgments about images of faces
that were taken looking directly into the camera as opposed to
from an elevated angle as is the case with video conferencing.
Chen [32] suggested that horizontal eye contact sensitivity to
gaze deviation might be accurate because, as the eye rotates hori-
zontally, the sclera becomes more visible. But as the eye rotates
downward it becomes difficult to know if a person is making direct
eye contact because less or no sclera can be seen. Chen [32] sug-
gests we perceive downward gaze as great as 10°to be eye contact
and recommends that in a video-conferencing system the eyes dis-
played on the screen should be located approximately 5°below the
camera to create the illusion of eye contact.
To examine gaze direction perception under video-mediated
conditions, Gale and Monk [33] had a looker and observer either
sit across from each other with the targets between them, or a vi-
deo camera was placed where the observer’s head would have
been. In both conditions both the looker and the observer were
seated at eye-level with one another. Targets were marked on a
vertical line that extended upward to the observer and on a hori-
zontal arc such that every point on the arc was the same distance
from the looker. Targets were marked such that the looker had to
rotate his or her head 2°to look from one to the other. In one
L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx 3
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
Table 1
A summary of studies examining gaze perception sensitivity.
Study Apparatus and stimuli Angles Distance Task Gaze sensitivity
Gibson and Pick
(1963)
Targets marked on wall behind O A 6 cm horizontal span with targets
separated by 2.9°
200 cm O responded ‘‘yes’’ or ‘‘no’’ to the question of
whether or not L was looking directly at him
The just noticeable deviation was 2.58°or
about the distance from O’s nose to the edge of
her face
Cline (1967) Transparent target board through
which O viewed a mirror which
reflected L
Targets along a vertical and
horizontal axis separated by 2°
122 cm O pointed to where she thought L was looking.
She also said when she thought L was looking
at her
The threshold at which O still perceived L to
be looking at him was:
0.73°horizontal and 1.27°vertical for
experiment 1
1.34°horizontal for experiment 2 and
2.48°when O’s head was rotated 30°right
or left
Anstis et al. (1969) Horizontal scale placed between L
and O 6 cm above eye level. Scale
rotated for experiment 2
Targets along a horizontal axis
spanning 0°,5°,10°,15°,20°to the
left and right
84 cm O indicated on a scale where L was looking by
pointing with a rod
Overestimated where L was looking by
between 50% and 87%
When L’s head rotated 30°left of O, O
estimated gaze to be 7.7°to the right. When
rotated right, estimation was 5.44°to the left
When TV was rotated to O’s left, estimation
was 1.81°to the left. When rotated right then
4.56°to the right
When scale was vertical, estimates were
true to scale
Chen (2002) L’s gaze positions were recorded by a
camera through a hole in a projector
screen which had targets drawn on it.
L was prerecorded on video tape
Targets were separated by 1°
increments downward to 15°. There
were seven other lines radiating from
the center that spanned 5°each and
formed a star-like pattern
480 cm (O and L
each sat 240 cm
from camera/
screen)
O indicated if L was looking at her or not When L gazed up, down, left, or right, L
could look at most 1°away from camera before
O perceived L was no longer looking and her
L could look 10°downward from the center
of the camera before O noticed L was no longer
looking at her
Gamer and Hecht
(2007)
Moveable eyes in a virtual and real
looker eyes
Eyes of both the virtual and real head
moved 1°at a time with the aid of
computer software or an electronic
pointer device, respectively. Eyes
could move as far as 10°to the right
or left
100 cm and
500 cm
O moved L’s eyes either toward or away from
him until he thought L was looking directly at
him
The cone of gaze was 9.3°at a distance of
100 cm and 8.2°at a distance of 500 cm for the
virtual head
The cone was 8.2°at 100 cm and 3.9°at
500 cm for the real head
When L’s head was rotated to right or left
10°the gaze was perceived to be shifted more
in the direction of the head rotation than it
actually was. The effect was greater for the
100 cm distance than the 500 cm distance
4L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
condition, observers were allowed to see the eye and head move-
ments of the looker and in other conditions they were not. The loo-
ker first looked at the extreme points and center on the scale to
‘‘calibrate’’ the observer. The observer then made estimations
about where the looker was looking.
The ability to see eye and head movements did not affect gaze
direction perception ability. Gaze estimation was accurate to with-
in about a target point (2°) for both live and video-mediated esti-
mating conditions. Similar to Chen [32], gaze estimation error
was smaller for horizontal than for vertical directions. In a second
experiment the calibration was eliminated and observers made
estimations about directions for both horizontal and vertical – gaze
direction perception was also very accurate for the combination of
directions [33].
Grayson and Monk [34] evaluated the perception of horizontal
gaze direction with desktop video-conferencing systems. In one
condition the camera was placed directly above the image of the
looker and in the other it was offset. Additionally, the size of the
looker’s face on the screen was varied. The image was either large
(352 288 pixels) or medium (176 144 pixels). The looker ran-
domly looked (moving both eyes and head) at one of six points
(43 mm apart) along a horizontal line that extended from either
side of the image of the person on the screen. The observer esti-
mated which point the looker was gazing at and the looker told
the observer what the correct answer was.
Image size had no effect on the perception of gaze direction
[34]. When the camera was positioned over the face of the partic-
ipant, gaze estimation was quite accurate (84%) but when the cam-
era was not place directly above the person being looked at, gaze
estimation dropped (67% accuracy). When the camera is positioned
directly over the image of the face, gaze direction perception accu-
racy decreases the further the gaze deviates from the centered
‘‘looking at me’’ position. Estimations showed little variation when
the camera was offset.
In a second experiment, Grayson and Monk [34] examined gaze
perception with a medium (176 144 pixels) and a small (88 72
pixels) image of the person on the screen. Also, in half of the trials
the looker held a card with the fixation point on it over the image
of the observer whenever he made a fixation – this was to elimi-
nate the possibility that subtle communication could reveal gaze
location when the looker was gazing at the image of the observer.
Accurate detection of the eye contact position was 87% for the
medium image and 68% for the small picture. Discrimination be-
tween points decreased for points further from the center. Making
the observer invisible to the looker did not influence judgments of
gaze direction.
Unlike most of the previous gaze research detailed earlier, these
two studies [33,34] did not focus on thresholds of when perceived
eye contact was lost. Instead, they focused on estimation of where
someone was looking. When there is no parallax, estimation of
gaze direction made from video seems to be quite accurate [33]
but once parallax is introduced as in desktop video-conferencing,
gaze direction perception accuracy decreases as the gaze direction
deviates from a centered position, directly under the camera [34].
The varying effects of image size are consistent with the idea that
small images lead to a poor user experience, but that camera posi-
tion matters more than image size.
4. Effects of eye contact on video-mediated communication
4.1. Communication efficacy
Some researchers have used the map task as a tool to analyzing
the effects of video-conferencing and eye contact on communica-
tion. The map task was originally suggested by Anderson et al.
[35] as a way to teach and evaluate language. It is a useful tool
to study communication because the map task requires informa-
tion transfer and reception through spoken dialogue. In the map
task, each participant is provided a map with landmarks scattered
about it. The two maps contain some similar landmarks but are
slightly different. One person is designated to be the instruction gi-
ver, and the other person is designated to be the instruction recei-
ver. The instruction giver’s map contains a trail. The goal of the task
is for the giver to verbally articulate the trail’s path to the receiver
who must replicate the trail on his own map. Error in the final re-
sult can then be measured and provide one measure of success.
Boyle et al. [36] evaluated performance and communication
efficacy on the map task in two conditions: co-present audio-only
and face-to-face. Performance was measured by placing the
instruction giver’s map on top of the follower’s map and calculat-
ing the difference in the paths (deviation) in square centimeters.
While performance did not differ between the two conditions,
communication patterns did.
Conversation efficiency was measured by number of turns,
number of words, number of words per turn, number of interrup-
tions (and interruption rate), and number of back channel re-
sponses (and back channel rate). Conversation pairs who could
not see each other took more turns and used more words than
those who could see each other. Participants who could not see
each other took shorter turns, that is, they spoke fewer words
per turn, than did participants who could see each other. Partici-
pants who could not see each other while completing the map
mask interrupted each other twice as often as participants who
could see each other. Conversations in which participants could
not see each other were longer (as measured in number of words)
than conversations in which participants could not see each other,
and the interruption rate (percentage of turns containing interrup-
tions was higher for the no-visibility condition. A back channel re-
sponse was defined as ‘‘a turn which consisted of an ‘uhuh’ or
‘mhm’’’ ([36], p. 10). Back channel responses were greater in both
number and rate for conversations that enabled visibility than con-
versations that did not. Thus, the end product (the path) did not
differ as a function of eye contact, but conversational efficiency
was increased when the pair could see each other.
Boyle et al. [36] also examined the instruction follower’s gaze
toward the instruction giver. Conversations were videotaped and
the transcribed dialogue was marked every time it was estimated
the follower looked up and at the giver. When discussing a feature
that differed between the two map versions, the instruction fol-
lower gazed at the giver more often than when discussing a feature
that was the same. Other research examining gaze and turn taking
found that people tend to look more at their conversational partner
when they are ready to stop talking and let their partner begin
speaking [7]. Participants in Boyle et al.’s [36] experiment may
have looked at their partner more when discussing incongruent
map features because they were trying to elicit verbal explanation
from their partner regarding more details about the features.
According to Boyle et al.’s [36] measures of conversation effi-
ciency, the ability to view one’s conversation partner significantly
increases the amount of information being conveyed in a shorter
amount of time and with fewer words. It can be inferred that inter-
ruptions result from some sort of confusion about either whose turn
it is or about the task. Back channel responses can be interpreted as
a means of communicating understanding and agreement.
Doherty-Sneddon et al. [37] distinguished between the surface
features and pragmatic function of dialogue by exploring dialogue
and gaze under the context of five communication media – co-
present audio, remote audio, face-to-face, video-mediated that
enabled eye contact and video-mediated that did not enable eye
contact. Like Boyle et al. [36], Doherty-Sneddon et al. [37] used
the map mask as a cooperative task for their participants.
Doherty-Sneddon et al. [37] analyzed the conversations between
L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx 5
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
participants by use of Conversational Games Analysis. The Conver-
sational Games Analysis technique, developed by Kowtko et al.
[38], breaks the content of conversation into games that may con-
tain several moves. Portions of conversations are given the analogy
of a game because there are ‘‘rules’’ or conventions that are known
by both parties.
In their first experiment, Doherty-Sneddon et al. [37] compared
the dialogue of participants who did the paper and pencil map task
under audio-only conditions to those who did the task face-to-face.
In the audio-only condition, participant’s dialogues contained 16%
more conversational games than in the face-to-face condition.
When participants could not see each other they had a greater need
to confirm and check their understanding of one another; audio-
only conditions contained significantly more conversational games
that checked and confirmed information. Videos of the face-to-face
conversation were coded for instances of gaze between partici-
pants. Doherty-Sneddon et al. [37] considered gaze to have
occurred when they perceived one partner had looked at the face
of the other. Participants tended to look at the other’s face at points
in the conversation when they were verbally eliciting feedback.
Performance on the map task did not differ between conditions.
In their second experiment, Doherty-Sneddon et al. [37] com-
pared dialogue structure and gaze between remote-audio (accom-
plished using microphones and speakers) and video-mediated
communication that either enabled (by use of a video-tunnel,
which uses mirrors to create the illusion of eye-contact) or pre-
vented eye contact. Eye contact was prevented in one condition
by raising the camera so that the participants were ‘‘now looking
slightly down on his or her partner’’ ([37], p. 116). In contrast to
Boyle et al. [36] who found that audio-only conversation required
more words and turns, Doherty-Sneddon et al. [37] reported that
there was no difference in number of words or turns between re-
mote audio-only conversation and video-mediated conversation
that did not enable eye contact. In fact, video-mediated conversa-
tion that enabled eye contact had significantly more words and
turns than conversations in the no eye contact and audio-only
conditions. Doherty-Sneddon et al. [37] explained this unexpected
result by pointing out that the video-tunnel allowing eye contact
may have been a novel experience for participants that fascinated
them, resulting in more conversation. Another possibility is that
the greater number of words recorded in the eye-contact condition
could be explained by an increase in attention required by looking
at another’s face; some research suggests that increased eye
contact during conversation increases the number of spoken
hesitations. Repetitions, fillers like ‘‘er’’ and ‘‘um,’’ false starts,
and other remarks that may be irrelevant to semantic communica-
tion such as ‘‘I mean’’ and ‘‘sort of’’ were almost double when
participants looked constantly at an interviewer as compared to
when they were allowed to avert their gaze [39]. In both the
video-mediated conditions, instruction givers initiated confirmed
information significantly less than instruction givers in the
audio-only condition. However, when video-conferencing was
compared to remote audio-only conversations, there was no
reduction of the number times instruction followers checked infor-
mation. It appears as if, like face-to-face communication, video-
mediated communication increases feedback to instruction givers
when compared to audio-only communication. However, unlike
face-to-face communication (in experiment 1), video-mediated
communication does not reduce the need for instruction followers
to check their knowledge of the instructions when compared to
audio-only conversations. These findings suggest that, while
video-mediated communication is an improvement over audio-
only communication, it is not comparable to face-to-face commu-
nication in its ability to facilitate efficient communication.
Doherty-Sneddon et al. [37] analyzed gaze between the
conditions and found that participants looked at each other
significantly more in the video-conferencing condition that
enabled eye contact than in the video-conferencing condition that
did not enable eye contact. They also gazed more in the video-
conferencing condition that enabled eye contact than in the
face-to-face condition of the first experiment.
4.2. Trust
Besides establishing mutual knowledge, other important social
interactions take place during communication and can be im-
pacted by video-conferencing. Building trust, impression formation
and establishing social status are all affected by different forms of
mediated communication.
To determine the effect of eye contact in video-mediated com-
munication on trust, Bekkering [40] created a scenario in which
participants indicated the trustworthiness of a message delivered
by people in three video-conferencing conditions. Participants
were to imagine a scenario in which they had just been denied a
job. The person at human resources informed them that the reason
they did not get the job might have had something to do with a
problem with letters of references. The participants were to imag-
ine that they called the two people they had asked to write letters
of references to check if letters had been submitted. Participants
then listened to messages from the two people - one familiar per-
son (their professor) and one unfamiliar person (a minister from a
neighboring town) confirming that they did indeed send letters.
The messages came in the form of an email, a voicemail, a video
in which the person was making eye contact (the camera was
eye level with the face), a video in which the camera was placed
above the person’s face, and a video in which the camera was
placed to the side of the person’s face. Each participant listened
to the messages in all their forms (the script was the same for each)
and indicated how much they trusted that the person had sent in
their letter of reference.
Whether or not the person delivering the message was familiar
to the participant did not affect perceived trust. When compared to
the video that enabled eye contact (center camera), videos that
prevented eye contact (side and top camera) resulted in lower
perceived trust scores [40]. Trust scores did not differ between
voicemail and centered video. Trust was significantly better for
the centered video than for email. In summary, video-
communication that prevents eye contact can reduce trust when
compared to video-communication that does. Voicemail enabled
just as much trust as the video that created eye contact, perhaps
because lack of eye contact cannot be perceived in audio-only
communication.
When participating in games that require group cooperation,
participants trust each other more when communicating face-
to-face as compared to computer-mediated media [41]. Partici-
pants, who played the Daytrader social dilemma game in groups
of three, across four media conditions (face-to-face, video-confer-
encing, three-way phone conference, and text chat), had to make
an investment choice that was either individual in which they were
guaranteed two tokens for themselves or as a group in which all
players were guaranteed three tokens so long as they all invested.
Every five rounds of the game a bonus of 90 tokens was given to
whoever had the most tokens at that point, and participants had
the opportunity to discuss and agree upon a strategy for the next
five rounds. If, for the five rounds preceding the bonus, all players
cooperated and invested as a group, the bonus was split three ways.
Participants who played the game in the text chat condition
made significantly fewer tokens from their investments than did
groups in the other three conditions. Group investments started
high and remained high for the face-to-face condition [41].As
the game progressed, group investments increased for the audio
and video conditions. Group investments started low and stayed
6L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
low for the text chat condition. If the amount of group investment
is interpreted as trust, establishing trust was delayed in the audio
and video conditions but gradually reached the same level as in
face-to-face condition. Sometimes after each discussion, group
members would gradually deviate from group cooperation. Coop-
eration was usually strongest right after a discussion. This measure
of defection was used as an indicator of the fragility of trust [41].
Defection for the computer-mediated conditions was significantly
greater than for the face-to-face condition. Bos et al. [41] inter-
preted these results to mean that trust is both more delayed and
more fragile in computer-mediated-communication modes than
in face-to-face communication.
Also using the Daytrader dilemma, Nguyen and Canny [42]
examined trust under three media conditions: face-to-face,
non-directional video-conferencing, and directional video-
conferencing. Directional video-conferencing was defined as
video-conferencing that enables accurate cues about gaze and
gesture direction and was achieved by Nguyen and Canny’s [42]
MultiView video-conferencing system. This system differs from
traditional video-conferencing systems in that a camera is placed
in front of each member of the group giving them an accurate per-
spective of the other group members they are viewing on the
screen.
The amount of cooperative investments made by groups did not
differ between the face-to-face condition and the directional
video-conferencing condition, and both of these were significantly
higher than for non-directional video-conferencing. There was no
difference in delay of trust between the three conditions. However,
trust was significantly more fragile in the non-directional
video-conferencing system than in both the directional video-
conferencing system and the face-to-face conditions [42].
Additionally, participants’ self-reported trust was significantly
lower in the non-directional video-conferencing system than in
both the directional video-conferencing system and the face-
to-face conditions. The results of this study indicate that a
video-conferencing system that affords more eye contact than
the traditional video-conferencing system will create group trust
levels similar to those seen in face-to-face group meetings [42].
4.3. Impression formation
In a study particular to video-conferencing, Fullwood [43] used a
unique interaction task to examine the role of communication med-
ia on impression formation. Same-sex participant pairs who were
strangers to each other communicated either face-to-face or via
video-conferencing. One partner in each pair was the ‘‘thought
transmitter’’ and the other was the ‘‘mind-reader.’’ The thought
transmitter was given a card with a numbered list of 10 colors on
it. The mind-reader was shown at list for 10 s that contained all pos-
sible colors he would be exposed to. The thought transmitters had
to attempt to transmit each color to the mind-reader in the order
that they appeared on the page by thinking about each color for
5 s. After each 5 s, the mind-reader had to guess what color the
transmitter had been thinking of. The mind reader was not given
feedback as to whether they had identified the colors correctly.
The participants in the pair then swapped roles and afterwards
filled out questionnaires about their partners which assessed their
feelings about how much they liked the other participant, how
intelligent they thought he or she was, and how good at reading
minds they thought he or she was. The participants in the
face-to-face condition liked their partners more than in the video-
conferencing condition and rated them as more intelligent [43].
When groups of participants had discussions about a common
strategy to survive on the moon (the Lost on the Moon Task)
conditions enabling selective gaze had greater effects on behavior
than other conditions [44]. Participants were more easily
persuaded by a confederate to incorrectly rank order survival
equipment when they were communicating over a video-confer-
encing setup that did not facilitate selective gaze compared to a
video-conferencing setup that did enable selective gaze and via
face-to-face communication [44]. Furthermore, when performing
a murder mystery-solving task, groups of participants were more
likely to communicate unshared information over video-
conferencing than face-to-face [44].
A number of factors are likely responsible for the impact of
video-conferencing on impression formation over face-to-face
interactions. The quality and speed of the video and audio signal
can mitigate the transmission of subtle non-verbal communication
[45]. Furthermore, people may be more self-conscious when talk-
ing over video-conferencing (intensified by being able to see their
own image on camera), and may therefore behave differently than
they normally do [45]. Additionally, eye contact perception is not
accurate in video-conferencing, and the presence or absence of
eye contact can affect the impression of personality and emotions
[45] [23].
5. Solutions to parallax in video-conferencing
The role of eye contact in human communication and the
impact of the now ‘‘traditional’’ video-conferencing arrangement
on perceived eye contact underline the importance of creating
the optimal perception of eye contact during video-conferencing.
Solutions to the lack of eye contact in video-conferencing are
varied and involve hardware and software approaches. Perhaps
one of the earliest solutions was the design of the video-tunnel
in Doherty-Sneddon, et al. [37]. This comprises a half-silvered
mirror placed at a 45°angle from a screen that is lying flat on
the surface in front of the person communicating [46]. The image
of the screen (and thus the other person’s face) is reflected onto
the mirror. A camera (or in some cases multiple cameras) is placed
behind the mirror so that the person is both looking at the image of
the eyes of the other and at the same time looking directly into the
camera. This creates the appearance of eye contact. Such a setup
can be limiting in terms of screen size and sometimes the person
is required to place his head in a chin rest to maintain eye contact
[46]. A modern-day attempt to adapt the video-tunnel to today’s
existing technology involves a miniature video tunnel that fits over
small monitor-mounted web cameras [47].
Similar in concept to the video-tunnel is embedding a camera in
the display. Tapia et al. [48] embedded a camera (1.5 mm in diam-
eter) onto a surface and projected the other’s face onto that surface.
This enables the speaker to look at both the image of the other
person and the camera, creating eye contact. This is not a viable solu-
tion for current laptops and monitors in commercial production.
Ott et al. [49] used a combination of multiple cameras and
software to create eye contact. A camera was placed both above
and below the screen; the top camera tilted slightly downward
and the bottom camera tilted slightly upward. Together these
cameras captured the same person from two different viewpoints.
Corresponding pixels were merged and a combination image was
used to create a single image in which it appeared the person
was looking straight into the camera. However, it was not possible
to capture the entire face resulting in some missing pieces in the
final image [49].
While many alternate video-conferencing systems have been
developed, the most prevalent video-mediated communication
system remains the version where one camera is placed above
the screen. Such systems are cost-efficient, portable, and compati-
ble with many types of computer systems. Most laptops and
tablets are now made with a small camera placed directly above
the center of the display. Solutions that aim to improve eye contact
L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx 7
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
perception under the constraints of this prevalent video-
conferencing system are thereby most likely to be adopted by
the majority of current users.
One such solution is software that alters the face image so as to
create the illusion of eye contact [50]. The software replaces the
eyes with computer-generated eyes and adjusts the head pose by
mapping the face image onto a head model. The software is able
to detect where the person is looking and adjust the appearance
of the gaze accordingly. The disadvantage of the software is that
head movements of more than 30°cause distortion.
However, simply by understanding human perception of gaze
during video-conferencing, as was summarized in Section 4of this
paper, much simpler solutions are possible. One algorithm takes
advantage of the large margin of error when perceiving gaze in a
downward direction by locating the eyes of conversational part-
ners and always displaying the image of the eyes close to the top
of the display where the camera is located [51]. Without the use
of extra equipment or computer programs, the average user can
improve his or her video-conferencing experience by ensuring that
his or her own image is close to the top of the camera’s field of
view, the image of the conversational partner is as close to the
camera as possible, and by sitting back from the display until the
perception of eye contact is achieved.
6. Conclusion
Much of the literature on gaze direction sensitivity emphasizes
differences among samples (based on gender, culture, role, etc.).
Perhaps the similarities should be emphasized. In general, allowing
eye contact makes communication richer and more efficient at the
same time. Trust is improved, with clear implications for how
business transactions may improve with eye contact facilitated.
There are very expensive solutions available for ‘life-like’ video-
conferencing, but the vast majority of users are employing Skype
or analogous free solutions on their existing computer systems.
Based on the review of the literature, it seems clear that users
can improve outcomes with a few simple measures: (1) by keeping
the interlocutors’ image as near the camera as possible; (2) by
sitting back a bit to improve the impression of mutual eye contact;
and, (3) smaller views of interlocutors should be avoided. The
ordering of these three measures reflects their relative impact from
most to least important given the literature reviewed. Video-
conferencing software that facilitates such adjustments may
provide better user experience. Furthermore, users can make com-
munication via video-conferencing more like talking in-person by
implementing some other simple methods - making sure hands
and arms are in view of the camera assure that more non-verbal
cues are communicated, and initially making face-to-face contact,
or engaging in an ice-breaker activity builds report between
communicants [45].
Research indicates that video-conferencing has an impact on
verbal and non-verbal communication, also affecting social
onstructs such as trust and impression formation. As video-
conferencing becomes a more pervasive form of sharing informa-
tion and relationships, technology needs to evolve to approach
the effectiveness of face-to-face communication. Research on
human sensitivity to gaze perception suggests the mounting of a
camera above the screen is a good solution. Users appear to adapt
to this viewing arrangement rapidly, but there is much greater
satisfaction with systems such as directional video-conferencing
or technical solutions such as the video tunnel or mounting a
camera in the screen. Designers of video-conferencing systems
need to be sensitive to the limitations of current systems, and
research on gaze detection and eye contact needs to keep pace with
technological innovations.
References
[1] AT&T, 1970: Picturephone, 2010. <http://www.corp.att.com/attlabs/
reputation/timeline/70picture.html> (retrieved 28.10.10).
[2] L. Wolgemuth, Knights of the virtual table: video conferencing grows up
thanks to ‘telepresence’ systems, US News and World Report 144 (2008) 48–
50.
[3] R.L. Daft, R.H. Lengel, Organizational information requirements, media
richness, and structural design, Management Science 32 (1986) 554–571.
[4] N. Guéguen, C. Jacob, Social presence reinforcement and computer-mediated
communication: the effect of the solicitor’s photography on compliance to a
survey request made by e-mail, CyberPyschology and Behavior 5 (2002) 139–
142.
[5] S.H. McNelley. The Significance of Eye Contact While Desktop Video
Conferencing, Ph.D. Dissertation, United States International University,
United States – California, 2001.
[6] M. Argyle, M. Cook, Gaze and Mutual Gaze, Cambridge University Press,
London, 1976.
[7] A. Kendon, Some functions of gaze-direction in social interaction, Acta
Psychologica 26 (1967) 22–63.
[8] A.J. Caron, R. Caron, J. Roberts, R. Brooks, Infant sensitivity to deviations in
dynamic facial–vocal displays: the role of eye regard, Developmental
Psychology 33 (1997) 802–813.
[9] T. Farroni, E. Menon, M.H. Johnson, Factors influencing newborns’ preference
for faces with eye contact, Journal of Experimental Child Psychology 95 (2006)
298–308.
[10] E.M. Blass, C.A. Camp, The ontogeny of face recognition: eye contact and sweet
taste induce face preference in 9- and 12-week-old human infants,
Developmental Psychology 37 (2001) 762–774.
[11] M.L. Spezio, P.S. Huang, F. Castelli, R. Adolphs, Amygdala damage impairs eye
contact during conversations with real people, The Journal of Neuroscience 27
(2007) 3994–3997.
[12] S. Watanabe, R. Kakigi, K. Miki, A. Puce, Human MT/V5 activity on viewing eye
gaze changes in others: a magnetoencephalographic study, Brain Research
1092 (2006) 152–160.
[13] E.T. Hall, The Hidden Dimension, Anchor Books Doubleday & Company, Inc.,
Garden City, NY, 1969.
[14] J.C. Nixon, J.F. West, Intercultural preparation for managers going to Japan,
Mid-American Journal of Business 10 (1995) 57–64.
[15] M.H. Levine, B. Sutton-Smith, Effects of age, sex, and task on visual behavior
during dyadic interaction, Developmental Psychology 9 (1973) 400–405.
[16] P. Ricciardelli, E. Bricolo, S.M. Aglioti, L.C. Chelazzi, My eyes want to look where
your eyes are looking: exploring the tendency to imitate another individual’s
gaze, NeuroReport 13 (2002) 2259–2264.
[17] C. Friesen, C. Moore, A. Kingston, Does gaze direction really trigger a reflexive
shift of spatial attention?, Brain and Cognition 57 (2004) 66–69
[18] G. Kuhn, B.W. Tatler, G.G. Cole, You look where I look! Effect of gaze cues on
overt and covert attention in misdirection, Visual Cognition 17 (2009) 925–
944.
[19] C. Fullwood, G. Doherty-Sneddon, Effect of gazing at the camera during a video
link on recall, Applied Ergonomics 37 (2006) 167–175.
[20] M. Mason, E. Tatkow, C. Macrae, The look of love: gaze shifts and person
perception, Psychological Science 16 (2005) 236–239.
[21] Z. Rubin, Measurement of romantic love, Journal of Personality and Social
Psychology 16 (1970) 265–273.
[22] C.L. Kleinke, A.A. Bustos, F.B. Meeker, Effects of self-attributed and other-
attributed gaze on interpersonal evaluations between males a females, Journal
of Experimental Social Psychology 9 (1973) 154–163.
[23] R.B. Adams, R.E. Kleck, Effects of direct and averted gaze on the perception of
facially communicated emotion, Emotion 5 (2005) 3–11.
[24] D.E. Campbell, G.E. Lancioni, The effects of staring and pew invasion in church
settings, Journal of Social Psychology 108 (1979) 19–24.
[25] R.E. Kraut, D. Poe, Behavioral roots of person perception: the deception of
judgments of customs inspectors and laymen, Journal of Personality and Social
Psychology 39 (1980) 784–798.
[26] Communication. Merriam-Webster.com. (n.d.). <http://www.merriam-
webster.com/dictionary/communication> (retrieved 02.08.12).
[27] M. Gamer, H. Hecht, Are you looking at me? Measuring the cone of gaze,
Journal of Experimental Psychology 33 (2007) 705–715.
[28] S.R.H. Langton, H. Honeyman, E. Tessler, The influence of head contour and
nose angle on the perception of eye-gaze direction, Perception and
Psychophysics 66 (2004) 752–771.
[29] J.J. Gibson, A. Pick, Perception of another person’s looking behavior, American
Journal of Psychology 76 (1963) 386–394.
[30] M.G. Cline, The perception of where a person is looking, American Journal of
Psychology 80 (1967) 41–51.
[31] S.M. Anstis, J.W. Mayhew, T. Morely, The perception of where a face or
television ‘‘portrait’’ is looking, American Journal of Psychology 82 (1969)
474–489.
[32] M. Chen, Leveraging the asymmetric sensitivity of eye contact for
videoconference, in: L. Terveen (Ed.), Proceedings of the AMC CHI 2002
Conference on Human Factors in Computing Systems, ACM Press, New York,
2002, pp. 49–56.
[33] C. Gale, A.F. Monk, Where am I looking? The accuracy of video-mediated gaze
awareness, Perception and Psychophysics 62 (2000) 586–595.
8L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
[34] D.M. Grayson, A.F. Monk, Are you looking at me? Eye contact and desktop
video conferencing, ACM Transactions on Computer–Human Interaction 10
(2003) 221–242.
[35] A. Anderson, G. Brown, R. Shillcock, G. Yule, Teaching Talk; Strategies
for Production and Assessment, Cambridge University Press, New York,
1984.
[36] E.A. Boyle, A.H. Anderson, A. Newlands, The effects of visibility on dialogue and
performance in a cooperative problem solving task, Language and Speech 37
(1994) 1–20.
[37] G. Doherty-Sneddon, A. Anderson, C. O’Malley, S. Langton, S. Garrond, V. Bruce,
Face-to-face and video-mediated communication: a comparison of dialogue
structure and task performance, Journal of Experimental Psychology: Applied
3 (1997) 105–125.
[38] J.C. Kowtko, S.D. Isard, G.M. Doherty, Conversational Games Within Dialogue.
Research Paper Published on the Human Communication Research Center,
1993. <http://www.hcrc.ed.ac.uk/maptask>.
[39] G.W. Beattie, A further investigation of the cognitive interference hypothesis of
gaze patterns during conversation, British Journal of Social Psychology 20
(1981) 243–248.
[40] J.E. Bekkering, Visual angle in videoconferencing: the issue of trust, Ph.D.
dissertation, Mississippi State University, United States – Mississippi, 2004.
[41] N. Bos, J. Olson, D. Gergle, G. Olson, Z. Wright, Effects of four computer-
mediated communications channels on trust development, in: Proceedings of
CHI 2002, ACM Press, New York, 2002, pp. 135–140.
[42] D. Nguyen, J. Canny, MultiView: improving trust in group video conferencing
through spatial faithfulness, in: CHI 2007 Proceedings (Trust & Engagement),
2007, pp. 1465–1474.
[43] C. Fullwood, The effect of mediation on impression formation: a comparison of
face-to-face and video-mediated conditions, Applied Ergonomics 38 (2007)
267–273.
[44] P.J. Werkhoven, J.M. Schraagen, P.A.J. Punte, Seeing is believing:
communication performance under isotropic teleconferencing conditions,
Displays 22 (2001) 37–149.
[45] C. Fullwood, M. Finn, Video-mediated communication and impression
formation: an integrative review, in: Adam C. Rayler (Ed.), Videoconferencing:
Technology, Impact and Applications, Nova Science, 2010, pp. 35–55.
[46] R. Vertegaal, I. Weevers, C. Sohn, C. Cheung, GAZE-2: Conveying Eye Contact in
Group Video Conferencing Using Eye-Controlled Camera Direction, CHI 2003,
ACM Press, 2003, pp. 521–528.
[47] See Eye2Eye: Teleprompter for Web Cameras, 2006. <http://www.bodelin.
com/se2e/>.
[48] E.M. Tapia, S.S. Intille, J. Rebula, S. Stoddard, Concept and Partial Prototype
Video: Ubiquitous Video Communication with the Perception of Eye Contact,
Ubicomp 2003 Video, Program, 2003, 2 p. (un-numbered).
[49] M. Ott, J.P. Lewis, I. Cox, Teleconferencing eye contract using a virtual camera,
in: S. Ashlund, K. Mullet, A. Henderson, E. Hollnagel, T. White (Eds.), INTERACT
‘93 and CHI ‘93 Conference Companion on Human Factors in Computing
Systems (Amsterdam, The Netherlands, April 24-29, 1993), CHI ’93, ACM, New
York, NY, 1993, pp. 109–110.
[50] J. Gemmell, K. Toyama, C.L. Zitnick, T. Kang, S. Seitz, Gaze Awareness for Video-
Conferencing: A Software Approach. IEEE MultiMedia 7, 4 (October, 2000),
2000, pp. 26–35.
[51] J.W. McGowen, US Patent No. US8203595. Washington DC: US Patent and
Trademark Office, 2012. <http://www.google.com/patents/US20100295920>.
L.S. Bohannon et al. / Displays xxx (2012) xxx–xxx 9
Please cite this article in press as: L.S. Bohannon et al., Eye contact and video-mediated communication: A review, Displays (2012), http://dx.doi.org/
10.1016/j.displa.2012.10.009
