Eye-Catcher or Blind Spot? The Effect of Photographs of Faces on E-Commerce Sites

Abstract

E-commerce still suffers from consumers' lack of trust. Most e-commerce researchers focus on building trust through cues that appeal to rational decision-making. These cues are usually implemented as text (e.g. privacy policies). In print advertising, affective attitudes and trust are commonly built by using photographs of people. On the web, however, users as well as interface design experts believe that photos impair task performance by attracting visual attention. In this study we compared online-shoppers' gaze patterns on pages with a photo of a person to pages with a text box of the same size. Our data does not support the claim that photographs decrease task performance. In fact, users spend more time looking at the text box. We found that the photograph attracts more attention than the text box on a first time view of a particular page. However, participants quickly learn the structure of a page and ignore the photo on subsequent pages. We outline further research, incorporating physiological measurements to infer user cost and affective responses.

Full text

1
Eye-Catcher or Blind Spot?
The Effect of Photographs of Faces on E-Commerce Sites
Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
University College London
Abstract: E-commerce still suffers from consumers’ lack of trust. Most e-commerce
researchers focus on building trust through cues that appeal to rational
decision-making. These cues are usually implemented as text (e.g. privacy
policies). In print advertising, affective attitudes and trust are commonly built
by using photographs of people. On the web, however, users as well as
interface design experts believe that photos impair task performance by
attracting visual attention. In this study we compared online-shoppers' gaze
patterns on pages with a photo of a person to pages with a text box of the same
size. Our data does not support the claim that photographs decrease task
performance. In fact, users spend more time looking at the text box. We found
that the photograph attracts more attention than the text box on a first time
view of a particular page. However, participants quickly learn the structure of
a page and ignore the photo on subsequent pages. We outline further research,
incorporating physiological measurements to infer user cost and affective
responses.
Key words: E-commerce, Eye-tracking, Trust, Photographs, Facial Cues, Re-embedding
1. INTRODUCTION
In e-commerce, unlike in classic media, photographs have only been used
sparingly. Human Computer Interaction (HCI) experts have argued for
several years that photos and graphics should be minimised as they increase
download times and act as distracters when user complete their tasks. They
are said to decrease the usability of on-line services [19, 13]. However,
bandwidth to potential clients is increasing, while ‘lack of consumer trust’ is
still seen as an obstacle in the development of e-commerce [17]. Many

2Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
studies have identified elements that can help to build trust in e-commerce.
They are mainly cognitive cues for trustworthiness and professionalism that
appeal to rational decision-making. Examples include privacy statements,
third-party certification and order tracking systems [4, 6]. However, trust
comprises cognitive and affective dimensions [5]. Thus, we should examine
how the affective impact of photographs, which advertising has relied on for
many years, can be used to signal trustworthiness on-line.
In this paper, we first introduce a conceptual basis for the use of
photographs as trust-builders, drawing on expertise from advertising and
sociology, and discuss results from studies on the effect of photos on
perceived trustworthiness (section 2). We then present an eye-tracking study
where we examined the impact of photographs of people on users’ gaze
behaviour and task performance (sections 3, 4, 5). We will conclude with
practical recommendations for designers of e-commerce sites and with
suggestions for directions of further research on this topic (section 6).
2. BACKGROUND
2.1 Photos in Advertising
When comparing print and poster adverts from the early days of
consumerism to today’s advertising, it is easy to see that the focus shifted
from explaining product attributes to showing attractive people using the
products. Why? The advertising industry has learned to make use of our
impressive capabilities to process and store images, particularly images of
people [12]. This capability is paired with a great weakness: we tend to
interpret images, particularly photographs, in the same way we interpret the
real thing. Whereas we critically ponder text we read, images we look at
directly stimulate affective responses [18, 12].
What is true for pictorial information in general is even more the case for
the processing of faces - be they real or on paper or on a screen. The neural
area that deals with the processing of faces is distinct from the brain region
that processes other information [24]. These human capabilities and
weaknesses have, often intuitively, been widely used in the advertising
industry for many years.
2.2 Re-embedding: Trust in Distant Interaction
It is well established that interactions with organisations that are stretched
over time and space and mediated through technical systems require more

Eye-Catcher or Blind Spot?3
trust than face-to-face interaction [8, 16]. E-Commerce is no exception, as
the persistent ‘lack of consumer trust’ shows [17]. This type of interaction
has been described as dis-embedded interaction [8]. Trust in dis-embedded
interaction can be built through personal trust in representatives, which is
usually based on periods of face-to-face interaction. This process is called
re-embedding [8]. Ordering books on-line and collecting them in the local
bookshop, a service offered by many booksellers with physical outlets, is an
example of re-embedding in e-commerce. Based on concepts of mediated
interaction such as social presence [26], parasocial interaction [11] and
telepresence [15], we have introduced the notion of virtual re-embedding
[22]. We argue that even mediated socio-emotional cues that are normally
present in face-to-face interaction signal trustworthiness. An example of a
rich implementation of virtual re-embedding would be a personal customer
service agent that can be contacted via chat or video-link from an e-
commerce site.
2.3 Photos on the Web
As a first step to investigate the idea of virtual re-embedding,
Riegelsberger & Sasse [23] and Steinbrück, Schaumburg, Duda & Krüger
[27] conducted studies on the effect of photographs of people on an e-
commerce site. Steinbrück et al. found that a photograph significantly
increased the perceived trustworthiness of an on-line banking site. This
result corresponds with findings from Fogg et al. [7], who demonstrated that
photos of authors can increase the credibility of articles in on-line
magazines.
The findings of the qualitative study by Riegelsberger & Sasse [23],
which looked at the effect of photos of people on the on-line shopping site
Amazon.de, are more controversial. Reactions of participants ranged from
welcoming their presence to expressions of increased mistrust, as they
perceived the photographs as an attempt at trust manipulation. We
categorised the participants based on their reactions in four distinct groups:
relationship seeking, function seeking, lack of benefits, lack of trust.
Relationship-seeking shoppers and non-shoppers that lack benefits displayed
overall positive reactions to the photographs. In the eye-tracking study
reported in this paper, we concentrate on the disapproval of the group we
termed function seekers. Member of this group called the photographs a
‘nuisance’, as they ‘cluttered the screen without serving any purpose’. Their
reactions support the previously mentioned notion, expressed by many HCI
experts, that photographs decrease the usability of an on-line service.
Function seekers claimed that finding links and selecting products on an e-
commerce web page is more difficult in the presence of photographs. Thus,

4Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
we assume that an increased difficulty in completing a product selection or
search task will result in an increased task completion time. Hence, our first
hypothesis is:
H1: Task performance will be lower in the presence of a non task-
related photograph than in the presence of non task-related text of the
same size.
The participants’ claims are plausible, as many eye-tracking studies on
print advertising have shown that photos, particularly faces (and here
particularly eyes) get the first fixations when a page is viewed [12].
However, the Stanford Poynter project that researched on-line news reading
does not confirm these results for on-line environments. On the contrary,
their results show that users first focus on text, headlines in particular, and
tend to ignore graphics and photographs [14]. The notion of banner
blindness, introduced by Benway [2], further advocates caution when
applying knowledge on eye movement from off-line reading to an on-line
context. Benway found that users of web sites largely ignore banner ads that
are designed to attract visual attention through imagery and animation [2].
However, these on-line studies have not specifically looked at photographs
of people and faces, as they are being used in print advertising. Thus, based
on the results of studies on print media and based on the claims of
participants in our earlier qualitative study, we assume that photos of faces
attract more visual attention than text.
H2: Users spend more time looking at a non task-related photograph
of a face on an e-commerce page than on non task-related text of the
same size.
Variance in the effects of pictorial stimuli across different media might
be due to differences in tasks that are performed. Pagendarm & Schaumburg
demonstrated that the type of task that is performed by the user has a
significant influence on recall and recognition of peripheral elements, such
as banner ads [20]. They compared a goal-directed search task to aimless
browsing, where participants were asked to just have a look at the web page
to get a general impression. Based on studies by Benway [2] and Bachofer
[1] they assume that goal-directed search relies on schemata that involve top-
down information processing. Top-down information processing modes will
result in less attention given to non task-related elements. Schaumburg &
Pagendarm assume that ‘aimless browsing’ is governed by bottom-up
information processing, where visual attention is not guided by schemata but
by external cues. Indeed, recall and recognition of peripheral banner ads
were significantly higher in the ‘aimless browsing’ condition. Based on these
results, we assume that the differences in information processing will also
result in differences in gaze behaviour. Thus, we hypothesize:
H3: When users engage in a non goal-directed task, they will spend
more time looking at photographs of faces than on text of the same size.

Eye-Catcher or Blind Spot?5
3. METHOD
3.1 Participants
The study was conducted with 40 students at the Department of
Computer Science at University College London. They were paid £5 for
their participation. The mean age of the participants was 22 years, 13
participants were female. The participants were very experienced Internet
users; 38 of them stated that they used the Internet for at least one hour per
day.
3.2 Materials
We used pages from the on-line shop of a well-known British
supermarket chain. The pages chosen were ‘aisle-pages’ containing product
listing from different categories. Figure 1 shows examples of the pages used.
Their structure was identical, but the products listed differed.
Figure 1. Examples of stimulus pages.
For each page, we created a version with a photograph of a customer
service agent (photo) and one with a box containing a text that detailed the
services of the supermarket’s bank (text). Both photo and text were of the
same size (160 x 135 pixels); subsequently we refer to this area on the screen
as region of interest.
3.3 Procedure
We calibrated the eye-tracker with several test-screens. Instructions were
presented to the participants on the screen. Once they had read and

6Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
understood the instructions, we displayed the web page. On completion of
their task they clicked the mouse. The web pages were incorporated as static
screenshots, due to technological constraints of the eye-tracking software.
In order to compare the effects of different tasks we asked participants to
perform 8 different tasks on the on-line shopping pages. Table 1 gives an
overview on the tasks:
Table 1. Participants' tasks.
1. Select favourite product
2. Select favourite product
3. Find ‘product search’ function
4. Find ‘change delivery time’ function
5. Find ‘help’ section
6. Find customer service telephone number
7. Find recipe ideas
8. ‘Just look at the page and get an impression’
Tasks and conditions (photo / text) were counterbalanced. For every
subject the conditions altered with every task. We duplicated the first task in
order to allow us to make a meaningful comparison between first time view
and subsequent views of the same page structure. Hence for task 1 and 2, we
had 4 stimulus pages, allowing us to counterbalance the products they listed
(fruit and deli products) and the photo / text conditions.
After completing the experiments we conducted brief qualitative
interviews with the participants to find out, whether they had experienced
any difficulties in completing their tasks and whether they could recall or
recognise specific interface elements.
3.4 Measurements
The core dimensions that are evaluated in the field of Human Computer
Interaction (HCI) are task performance, user cost (comprising physiological
and mental cost) and user satisfaction [25]. Task performance has received
most attention in the past. However, when we evaluate e-commerce systems
that are targeted at consumers, user satisfaction as well as user cost are
equally important, as they will influence their decision as to where to shop
online.
The trust users hold in an e-commerce site will impact user satisfaction,
and can also minimise user cost. In the study we report here, we explored
potentially adverse effects of elements that aim to build trust, i.e.
photographs of people. One of the dependent variables measured is thus the
time taken to complete the task.

Eye-Catcher or Blind Spot?7
We used eye-tracking to explore the effects of photographs on the
atomic level of users’ visual attention. The aim was to gain a better
understanding of potential effects on task performance and user cost. Thus
we used the eye-tracker to measure whether the photos had been looked at
and how much time was spent looking at them. We used the LC Eye Gaze
system (www.eyegaze.com), an infrared based remote tracking system that
allowed us to sample the position of the participants’ gaze with a frequency
of 50 Hz. We report this measure as sample count, where each sample count
equates 0.02 seconds of visual attention given to a region of interest. We
further analysed fixations. Fixations where detected by looking for
sequences of gazepoint measurements that remain relatively constant. If a
new gazepoint lies within a circular region around the running average of an
on-going fixation, the fixation is extended to include the new gazepoint. The
radius of the acceptance circle was set to 6.35mm.
4. RESULTS
4.1 H1: Task Performance
As figure 2 shows, there is no significant difference in the total time
taken to complete a task between the photo and the text condition. There is
thus no evidence that the presence of a photo compared to the presence of
non task-related text has an influence on the time taken to complete a task.
We cannot reject the null hypothesis, thus there is no evidence that
photographs decrease task performance.
0
5
10
15
20
25
30
Seconds
Select
favourite
product
Select
favourite
product
Find
'Product
Search'
Find
'Delivery
Time'
Find
'Help'
Find
'Service
Tel.'
Find
'Recipe
Ideas'
'Just look
at the
page'
F
igure 2: Total time taken (white: text / grey: photo).

8Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
4.2 H2: Time spent in Region
Again, over most tasks, there are no significant differences between the
photo and the text condition. Participants do not significantly spend more
time on photographs than on text. To the contrary, for most tasks,
participants spend more time looking at the text than at the photograph.
Exceptions are task one and two, which both contain first time views of the
pages (see section 4.4).
4.3 H3: Task Type
Only for task 8 (‘Just look at the page’), the time spent looking at the
region of interest differed significantly: Participants spent more time looking
at the region in the text condition than in the photo condition. After
excluding participants with a tracking accuracy1 of below 95% and removing
outliers, we had 16 valid measurements for task 8. Table 2 gives an overview
on the results for this task.
Table 2. Results for task 8: ‘Just have a look at the page’.
Text Photo p N
Absolute Time .76 .17 .015 16 (1 outlier)
Fixation Count 2.75 .75 .048 16 (1 outlier)
Relative Time 3.6 % .5 % .002 16 (1 outlier)
In the text condition, participants spent on average .76 sec on the region,
whereas in the photo condition the average was .17 sec. The differences
become more evident when comparing relative times, i.e. the time spent
looking at the region of interest relative to the time taken to complete the
whole task. Thus, the data falsifies H3: For a non goal-directed task,
participants spend more time looking at text rather than at photos.
4.4 Attraction of Region
One concern with above results is the extent to which time in a region is
biased by the content of the region. As previously mentioned, it is known
that face processing is fast and dealt with by specialised brain structures
[20]. Thus, using time in region might be an unfair comparison between the
two conditions as it is biased against H2 and H3. Thus, as a secondary
1 Tracking accuracy is the number of registered samples divided by the total number of
samples that could have been taken during task completion. Tracking accuracy is
decreased by participants’ head movements or gazes away from the screen.

Eye-Catcher or Blind Spot?9
measure we compared the conditions on the grounds of whether the region
had been looked at or not. This measures the attraction of the region alone
and is unbiased by differences in processing time.
This analysis revealed a significant difference between the first and
second page view. When they look at the page the first time, significantly
more participants look at the region when the photo is present compared to
when text is displayed (Pearson Chi-Square, p=.027). Looking at the second
view of the pages, where participants again selected a favourite product,
there are no significant differences, between conditions – the majority of
participants ignore the region, be it text or photo (see table 3).
Table 3. Number of participants who viewed the region of interest.
1st view TEXT PHOTO TOTAL
Region NOT viewed 14 7 21
Region viewed 6 13 19
2nd view
TEXT
PHOTO
TOTAL
Region NOT viewed 15 16 31
Region viewed 5 4 9
4.5 Qualitative Interviews
Most of the participants who reported problems finding items because of
the pages being too ‘cluttered’ referred to text as the reason for this. They
said that there was either too much text they had to skim through, or that it
was too small. When asked about interface elements they could recall or
recognise, most participants mentioned a banner ad for a product range that
was covered by TV advertising during the period of the experiments. Both
the photo and the text we tested were recognised by only half the participants
in the post-experimental interviews. Comments on the photographs made by
participants who recognised the photo in the post-experimental interviews
were generally positive.
5. DISCUSSION
5.1 Task Performance
The data does not allow us to conclude that the variance in time taken to
complete a task is related to the presence of text or photos. Thus, the claims
of participants in the previous qualitative study [23] are not corroborated.

10Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
Our results confirm those by Burke & Hornof, who found that animated
banners do not decrease task performance in a visual search task [3].
However, they report that subjective workload, measured with the NASA
Task Load Index (TLX), was increased. They conclude that in the presence
of animations, task performance is held constant by users, albeit at the
premium of increased mental cost. Based on our results and the claims of
participants in our earlier study, we hypothesize that non task-related
photographs have a similar effect.
The photos do not affect task performance, but they might affect user
cost, as users have to counteract this stimulus (bottom-up) driven guidance
of attention to keep task performance high.
5.2 Time in Regions & Visual Attraction
5.2.1 Effect of Photos
When carrying out a product selection or search task, users do not spend
significantly more time looking at non-related photos than at non-related
text. As said before, this measure of time spent in a region is biased towards
text, as it takes longer to process. Thus, we have used the measure of
attraction – whether a region had been viewed all. Again, across tasks, there
were no significant differences between the attraction of photo and text.
Thus, our findings should raise concerns about directly transferring results
from classic media (e.g. print) to on-line environments.
However, the first time the participants viewed a product selection page,
the majority of them looked at region in the photo condition, whereas the
majority did not look at the region in the text condition. Thus, for first-time
viewing, results from off-line magazine reading translate to on-line
shopping. Photos of faces do attract visual attention. They can thus be used
as a modest means of virtual re-embedding.
5.2.2 Effect of Task
We expected to activate schemata that are similar to off-line magazine
reading by asking participants to 'just look at the site' without asking them to
perform a search or product selection task (task 8).
Comparing the attraction of the regions did not yield any significant
results for this task. However, participants spent significantly less time on
the region in the photo condition than in the text condition. We explain this
effect by the fact that this task did not put time pressure on the participants;

Eye-Catcher or Blind Spot?11
hence they could take the time to read the text, which they previously might
just have skimmed when they were engaged in their search tasks.
On a methodological level, this result supports Pagendarm &
Schaumburg’s [20] claim that we must distinguish between different task
types and processing modes when researching web site usage on such an
atomic level. The task effects which mode of information processing is
dominant: stimulus-driven or top-down.
5.2.3 Effect of Repetition
Another interesting finding from the measure of attraction is that for
subsequent views of the same page structure the region was largely ignored,
independent of whether photo or text was displayed. We explain this effect
through the participants’ quick learning of the structure of the page. They
learned where to expect task-related information. Unlike magazine-reading,
where readers also look at photographs and adverts that are clearly
identifiable by their position and are clearly not related to content they are
reading, on-line shopping is more goal-focused.
6. CONCLUSIONS
6.1 Substantive Findings
This study is the first to look into the effect of photographs of people on
visual attention on an on-line shopping site. Evidence from related studies on
the effect of photos on visual attention in other media is mixed (section 2.1
and 2.3): Studies conducted in classic media clearly state that photographs,
particularly of faces, attract visual attention. The best-known study for on-
line media, the Poynter project, does not support these findings for on-line
news reading.
Our most important result is that photos do attract more visual attention
than text on a first time view of a page, when users perform a product-
selection task. Steinbrück et al. [27] found that one-time exposure can
already have a positive effect on attributed trustworthiness. Thus, experience
from advertising can be used to give e-commerce sites affective attributes
and signal trustworthiness.
However, our results suggest caution if a photo is used repeatedly on
several pages of the shopping process. Firstly, it will largely be ignored and,
secondly, ignoring it might come at a cost to the user. They might perceive
the pages as ‘cluttered’, as participants in our previous study stated.

12Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
Furthermore, we found that on average, users spend more time looking at
non-task related text than on non task-related photos (significant for task 8).
Also, based on our post-experimental interviews, they suffer more from non
task-related text than from photos, when performing search tasks. This
finding should advocate caution when trying to communicate trustworthiness
through more traditional, cognitive approaches such as privacy statements
and third party assurances, as they all rely on text. When trying to make it
easier for users to perform their tasks on an e-commerce site, text should be
as carefully placed as photos or graphics.
Finally, our study showed that users are able to learn the structure of a
particular page surprisingly fast. This finding further emphasises the need for
a consistent structure of pages across a site, as it has been advocated by
many HCI experts [19, 13].
6.2 Methodological Findings
We found that the time spent looking at an interface element, or the
number of fixations on this element, is a problematic measure when inferring
on its impact on the user’s information processing and – ultimately – task
performance, user cost and user satisfaction. We compared photos and text,
elements that are known to be processed in very different ways. Thus, we
advocate the use of the percentage of users who looked at a particular region
of interest, or location of first and subsequent fixations, to measure the initial
attraction of an interface element.
We have shown that results from on-line news-reading are not directly
applicable to on-line shopping. Thus, results from eye-tracking studies from
one domain of on-line services should not be generalised to other domains.
Furthermore, even within the domain of e-commerce, our study showed that
the task that is performed by the user on a particular page has an impact on
gaze behaviour. Thus, for further studies it is advisable to differentiate the
sub-tasks users perform on an e-commerce site and to evaluate them
separately.
6.3 Limitations
Before we introduce thoughts for further research, we want to point out
the limitations of this study. Even though we used a remote eye-tracking
system that is non-invasive and that allows participants to sit and move
rather comfortably, the strictly controlled nature of the experimental study
lead to a rather low ecological validity of the participants’ tasks. They only
interacted with static pages in a repetitive fashion, rather than exploring a
whole site by following links. Nonetheless, we believe that the results give

Eye-Catcher or Blind Spot?13
some valuable indications both for researchers and practitioners at this early
stage.
6.4 Future Work
Studies beyond this initial one should aim to increase ecological validity
by using real pages as stimulus material and by allowing users to navigate an
e-commerce site.
In a next step we will combine eye-tracking measures with synchronous
recording of physiological measurements such as electrodermal activity
(EDA) and blood volume pulse (BPV). These measures have been shown to
be indicative of perceptual strain and user cost [29]. Initial pilots, however,
have shown that these measures are very sensitive to individual differences,
habituation and noise, thus complicating the analysis. Research by Vyzas &
Picard [28] furthermore suggests that physiological measures can be used to
infer on affective responses and emotions. Furthermore, Hess & Polt [10]
claim that pupil dilation, which is recorded by standard eye-tracking
equipment, is indicative of affect. We expect that concurrent analysis of the
position of eye-gaze and physiological measures as well as pupil dilation
will allow isolating immediate affective responses to interface elements.
Furthermore, the effect of virtual re-embedding, be it through
photographs or richer media, needs to be explored further. Current studies
have relied on qualitative methods and questionnaire tools. We will look into
methods with higher ecological validity, making use of economic trust
games and also drawing in established methods from advertising research
such as Implicit Association Tests [21] or Semantic Differentials [9].
ACKNOWLEDGEMENTS
We gratefully acknowledge the help of Gillian Wilson and Daniel
Bruneau. Jens Riegelsberger is funded by a British Telecom studentship
(WEB164414/CT501045). The research reported here was conducted as part
of the HIGHERVIEW project (www.cs.ucl.ac.uk/research/higherview).
7. REFERENCES
[1] Bachofer, M. (1998). Die Stern Bibliothek: Wie wirkt Werbung im Web? Hamburg:
Gruner & Jahr.
[2] Benway, J. P. (1998). Banner Blindness: The irony of attention grabbing on the World
Wide Web. In Proceedings of the Human Factors and Ergonomics Society 42nd
Annual Meeting, pp. 463-467.

14Jens Riegelsberger, M. Angela Sasse & John D. McCarthy
[3] Burke, M. and Hornof, A. J. (2001). The effect of animated banner advertisements on a
visual search task. University of Oregon Technical Report.
ftp://ftp.cs.uoregon.edu/pub/hornof/CHI2002.pdf .
[4] Cheskin (1999). eCommerce Trust Study.
http://www.cheskin.com/think/studies/ecomtrust.html
[5] Corritore,C.L., Kracher,B., & Wiedenbeck,S. (2001). Trust in the online environment.
In M. J. Smith, G. Salvendy, D. Harris, & R. J. Koubek (Eds.), Evaluation and Interface
Design: Cognitive Engineering, Intelligent Agents and Virtual Reality. Mahwah, NJ:
Lawrence Erlbaum, pp.1548-1552.
[6] Egger, F. N. (2001). Affective Design of E-Commerce User Interfaces: How to
maximise perceived trustworthiness. In Helander, M, Khalid, H. M., and Tham (Eds.).
Proceedings of CAHD: Conference on Affective Human Factors Design. June 27-29,
Singapore, pp. 317-324.
[7] Fogg, B., Marshall, J., Kameda, T., Solomon, J., Rangnekar, A., Boyd, J., and Brown,
B. (2001). Web Credibility Research: A Method for On-line Experiments and Early
Study Results. CHI2001: Extended Abstracts. March, 31 – April, 5. Seattle, WA, US,
pp. 295-296.
[8] Giddens, A. (1990). The consequences of modernity Stanford: Stanford University
Press.
[9] Heise, D. R. (1970). The Semantic Differential and Attitude Research. In G. F.
Summers (Ed.). Attitude Measurement. Chicago: Rand McNally, pp. 235-253.
[10] Hess, E.H., and Polt, J.M. (1960). Pupil size is related to interest value of visual stimuli.
Science, 132, 349-350.
[11] Horton,D., & Wohl,R.R. (1956). Mass communication and para-social interaction:
Observations on intimacy at a distance. Psychiatry, 19, 215-229.
[12] Kroeber-Riel (1996). W., Bildkommunikation Munich: Vahlen.
[13] Krug, S. (2000). Don't make me think Indianapolis: New Riders.
[14] Lewenstein, M., Edwards, G., Tatar, D., and DeVigal, A. (2000). Stanford Poynter
Project. http://www.poynter.org/eyetrack2000/.
[15] Lombard,M., & Ditton,T. (1997). At the Heart of It All: The Concept of Presence.
Journal of Computer Mediated Communication, 3(2).
[16] Luhmann, N. (1979). Trust and Power Cichester: Wiley.
[17] Mitra, A. (2002). Trust, Authenticity and Discursive Power in Cyberspace,
Communications of the ACM, vol. 45, no. 3, pp. 27-29.
[18] Nass, C. and Reeves, B. (2000). Perceptual Bandwidth, Communications of the ACM,
vol. 43, no. 3, pp. 65-70.
[19] Nielsen, J. (2000). Designing Web Usability Indianapolis: New Riders.
[20] Pagendarm, M. and Schaumburg, H. (2001).Why are users banner-blind? The impact of
navigation style on the perception of web banners. Journal of Digital Information, vol.
2, no. 1. http://jodi.ecs.soton.ac.uk/Articles/v02/i01/Pagendarm/
[21] Plessner, H. and Barnse, R. (2002). Attitude Measurement Using the Implicit
Association Text. Experimental Psychology 50(2).
[22] Riegelsberger, J. and Sasse, M. A. (2001). Trustbuilders and trustbusters: The role of
trust cues in interfaces to e-commerce applications. In B. Schmid, K. Stanoevska-
Slabeva, and V. Tschammer. Towards the E-Society: E-commerce, E-Business and E-
Government. 1st IFIP Conference on e-commerce, e-business, e-gonvernment (i3e).
October 3-5, Zurich, pp. 17-30.
[23] Riegelsberger, J. and Sasse, M. A. (2002). Face it: Photographs Don't Make Websites
Trustworthy. CHI2002: Extended Abstracts. April 20-25, Minneapolis, MN, US, pp.
742-743.

Eye-Catcher or Blind Spot?15
[24] Rodman, H. R. (2002). Face Recognition. htp://cognet.mit.edu/MITECS/Entry/rodman.
[25] Shackel,B. (1991). Usability - Context, Framework, Definition, Design and Evaluation.
In B. Shackel & S. Richardson (Eds.), Human Factors for Informatics Usability.
Cambridge: Cambridge University Press, pp. 21-37.
[26] Short, J., Williams, E., and Christie, B. (1976). The Social Psychology of
Telecommunications London: John Wiley & Sons.
[27] Steinbrück, U., Schaumburg, H., Duda, S., and Krüger, T. (2002). A Picture Says More
Than A Thousand Words - Photographs As Trust Builders In E-Commerce Websites.
CHI2002 Extended Abstracts. April 20-25, Minneapolis. MN, US, pp. 748-749.
[28] Vyzas, E. and Picard, R. W. (1999). Offline and Online Recognition of Emotion
Expression from Physiological Data. Third International Conference on Autonomious
Agents, May 1, Seattle, WA, US.
[29] Wilson, G. and Sasse, M. A. (2000). Do Users Always Know What's Good For Them?
Utilising Physiological Responses to Assess Media Quality. Proceedings of HCI 2000:
People and Computers XIV. McDonald, S., Waern, Y., and Cockton, G. September 5-8,
Sunderland, UK, pp. 327-339.