Article

Information Visualization
1–10
Ó The Author(s) 2017
An eye-tracking study on the effect of Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
infographic structures on viewer’s DOI: 10.1177/1473871617701971
journals.sagepub.com/home/ivi

comprehension and cognitive load

Azam Majooni1, Mona Masood1 and Amir Akhavan2

Abstract
The basic premise of this research is investigating the effect of layout on the comprehension and cognitive
load of the viewers in the information graphics. The term ‘Layout’ refers to the arrangement and organization
of the visual and textual elements in a graphical design. The experiment conducted in this study is designed
based on two stories and each one of these stories is presented with two different layouts. During the experi-
ment, eye-tracking devices are applied to collect the gaze data including the eye movement data and pupil
diameter fluctuation. In the research on the modification of the layouts, contents of each story are narrated
using identical visual and textual elements. The analysis of eye-tracking data provides quantitative evidence
concerning the change of layout in each story and its effect on the comprehension of participants and varia-
tion of their cognitive load. In conclusion, it can be claimed that the comprehension from the zigzag form of
the layout was higher with a less imposed cognitive load.

Keywords
Visualization, visual storytelling, infographic, layout, eye tracking

Introduction the flow of the story. Besides the visual elements, cul-
tural and language background can be very important
In this study, the main objective is to investigate the in the formation of the layout. English-speaking read-
effect of the layout of the Infographics in the formation ers naturally read textual contents from left to right
of the mental model and cognitive load. The most and top to bottom. The results of this study show that
important elements of information graphics are the the subjects of this study tend to move from left to
use of visuals (images, graphs, font size and colour).1 right and from up to bottom, therefore arrangement of
Infographics can be easier to read if effective visuals
layout plays an important role in the design of the
are applied in their structure. The parameters such as
Infographics.5 In order to determine the influence of
balance in the placement of text and graphics are
the layout on the cognitive load, the gaze data of the
essential in making an infographic aesthetically pleas-
participants are recorded using the eye-tracking
ing to look at.2 The visual elements of an Infographic
should be able to narrate the visualized story with a
minimal requirement for additional text explanations.3 1
Centre for Instructional Technology & Multimedia, Universiti
Several studies support that images and graphs are Sains Malaysia, George Town, Malaysia
salient compared to the texts and can capture the 2
School of Computer Sciences, Universiti Sains Malaysia, George
immediate attention of the viewers.4 The proper Town, Malaysia
arrangement of visuals and graphs can provide coher-
Corresponding author:
ence for an infographic design. A coherent layout is Azam Majooni, Centre for Instructional Technology & Multimedia,
very important in guiding the viewers from one section Universiti Sains Malaysia, George Town, 11800 Penang, Malaysia.
of the infographic to the other related section based on Email: [email protected]
2 Information Visualization

devices. The recorded gaze data for two different attractiveness.2 These requirements have been
Infographic stimuli are analysed and the results are addressed and labelled as Appeal for attractiveness
discussed. Without loss of generality, in this experi- and Comprehension and Retention for utility by Lankow
ment, a suggested layout with zigzag form demon- et al.3
strated better results. Therefore, it could be assumed There has been an ongoing debate about the appeal
that this form of layout if applied in the real-world or beauty (usually referred as aesthetics) of visualiza-
Infographics should be more comprehensible. The tions between the psychologists, artists, architects,
predictable orientation of the chunks in this model computer scientists and philosophers.12 According to
helps the viewers follow the story of the Infographic Fechner,13 the most central principle of aesthetics is
more steadily with less distraction. ‘the right combination of diversity in unity’. Moreover,
basic adjustable features of aesthetics such as colour,
contrast and clarity;14 simplicity and complexity;15 and
Background balance and scale16 along with novelty17 and surprise18
have been analysed intensively by the researchers to
The term ‘Infographic’ is the abbreviation for the
find the effectiveness (aesthetic impression) of each of
information graphics. The name has become popular
the features in the design of visual artefacts.19
in the past few years. From the novelty perspective,
The appeal of an Infographic presentation is neces-
some believe Infographics have been used from a long
sary for attracting the attention of the viewer; nonethe-
time ago and even consider the prehistoric paintings in
less, the main goal of any visualization is to deliver
the caves as the early Infographics which tell a story
abstract information to the viewer. Therefore, compre-
using graphical elements and illustrations.6,7 However,
hension and the retention of the visualization are the
the demand and application of Infographics have been
key factors in the design of any Infographic.20
growing rapidly in the past few years. The popularity
In order to successfully maintain these factors,
of the terms searched on Google search engine can be
storytelling based on instructional design strategies
estimated by the aid of Google trends;8 these statistics
can be applied. Storytelling is a very successful way for
provide that the term ‘Infographics’ is searched almost
transferring knowledge,7,21 although an Infographic
seven times more frequent during the year 2014 com-
aims to take advantage of this technique, but it is the
pared to the same period in 2011. The main purpose
receiver that perceives the presented visualization.
of Infographics is to represent information and convey
Eventually, the viewer starts developing a mental
meaning by the aid of graphical elements; this task is
model after seeing the external visual and textual ele-
sometimes accompanied by small textual contents.9
ments inside the Infographic.22 In the process of devel-
An Infographic usually contains a story embedded
oping the mental model, the viewer automatically
in it; therefore, it can be assumed as a visual storytell-
creates connections between the mental model and the
ing method.10 The modern automated illustration
external visualization.22
tools have extensively simplified the design process of
According to the scaffolding theory, a temporal
Infographics, but in the crowded and information
framework can improve the understanding of the
overloaded world, attracting and keeping the attention
viewer; additionally, according to the chunking princi-
of the audience is an extremely difficult task. Almost
ple, the cognitive load can be reduced by presenting
40 years ago, Herbert Simon11 was one of the first
the visualizations in small chunks.23,24 In this study,
researchers to directly point out that attention is a
both the scaffolding theory and chunking principle are
scarce resource in an information-rich world, he also
applied to design the experiment. The main goal of
mentioned that: ‘What information consumes is rather
cognitive load theory is to introduce methods to
obvious: it consumes the attention of its recipients’.
enhance the intellectual performance of the learners;25
Similar to any other storytelling medium, such as
therefore, monitoring cognitive load can show the
novels, scripts and motion pictures, visual storytelling
effectiveness of visualizations.26
requires a skilled director and an attractive plot to cap-
ture the attention of the audience.
There exist only a few elaborated guidelines to
Motivation
design a good Infographic and also there is no clear defi-
nition for estimating the efficiency of the design. The goal of designing Infographic is to transfer knowl-
However, some general conceptual requirements edge (mostly conceptual knowledge) to the viewers in
based on the standards stated by the Roman architect a very short time, premised on the assumption that
Vitruvius are presented by Moere and Purchase.2 The there are three main elements in the scenario of
Vitruvius triangle states the basic requirements for Infographics design: message (story), the sender (story
an architectural design: soundness, utility and teller) and receiver (viewer). The designer should go
Majooni et al. 3

Figure 1. Schematic form of the layouts for stories A and B.

through a sequence of steps, which includes ‘gather- The answer to these statements is investigated using
ing, processing, pictorial rendering, analysing and the comprehension questions and analysing the gaze
interpreting data’.27 Because of the nature of the data including the fixations, saccades, the number of
Infographics, the designer attempts to map the story to the blinks and the pupil dilation of the subjects.
be attractive and clear so that the viewer can develop
the intended mental model of the story with the least
effort.7 Card et al.28 have presented a visual mapping Experiment
model that includes four sequential steps. At the first The experiment includes two stories (A and B) which
and second stages, the designer possesses original data compare two types of layouts (1 and 2) for each story.
and prepares data tables. In the third stage, the data The designs are based on chunking principle, scaffold-
tables are transformed/mapped into visualizations. ing theory and Cards method28 manipulating two dif-
Finally, in the last stage, the designed visual structure ferent layouts. The schematic shape of the layouts for
is transformed to view. The process of mapping the A-1, A-2, B-1 and B-2 are presented in Figure 1.
data tables into visualization can be carried out by sev- As it can be seen from Figure 1, in the story A, the
eral visualization techniques.29 In this study, the main layouts 1 and 2 compare the Stairs (Steps) with the
concern is to overview the layout issue of the info- Zigzag organization, also in the story B the layouts 1
graphic. Several studies have discussed issues about and 2 compare Vertical columns with Horizontal rows
the layout, and automated design of layouts,30 but to organization of the chunks.
the best of our knowledge, there has not been any In order to reduce the possible bias caused by exter-
study investigating the layout of visualization using nal variable, the effort has been to identify effective
eye-tracking techniques and analysing the cognitive variables and remove the sources of bias. The variables
load yet. There are two main research questions in this related to the subjects such as age, gender, educational
study: (1) What is the effect of layout in the formation and cultural background are addressed by inviting par-
of the mental model? (2) Which type of suggested lay- ticipants from almost the same age, cultural and edu-
outs is effective in decreasing the cognitive load? cational background and an equal number of male and
4 Information Visualization

female participants from the same ethnics. The partici- divided into a sequence of small chunks of data. In the
pants are put into two groups with different tasks each second step, the chunks are translated into graphical
one containing both types of stimuli. Similarly, in the elements accompanied with small textual data (story
design of the tasks, the duration of the task, the applied elements). In the last step, each chunk is organized by
colours and symbols, font size and other graphical ele- composing the layout nodes. The steps used for the
ments are kept constant. design of the stimuli are similar to those suggested by
Card et al.28
Participants
In this experiment, 23 local undergraduate (second
Procedure
year) students from the same field of study (School of Subjects were randomly assigned to one of the two
Housing, Building and Planning, Universiti Sains groups I and II. These groups each contained two sep-
Malaysia) and nationality participated. Three of the arate tasks (A and B) recorded by eye-tracking device.
participants had poor eye-tracking calibration in the There was a 1-min break between the first and the sec-
final results; therefore, their results were eliminated ond task. The instruction and the steps of the experi-
from the final experiment results and gaze data of the ment were briefly described to the subject to achieve
remaining 20 participants (10 females, 10 males) were better calibration on the eye-tracking device. After
used for the analysis. The age range of the participants each task in the experiment, each participant was
was between 22 and 23 years with standard deviation asked to answer questions related to the comprehen-
of less than 1 year. All the participants were randomly sion of the Infographic. The participants are assured
put into two groups of experiment with similar content that the results of the questions are anonymous and
but different designs (each group containing n = 10 will not be distributed. The process of the experiment
participants). was straightforward and no other interaction was made
with the subject. The assigned exposure duration for
each stimulus was 1 min, the amount was equal for all
Apparatus the participants. According to the initial pilot study,
In order to record the gaze data, in this experiment, 1 min was sufficient for the tasks.
remote eye-tracking device (SMI, ‘iView X RED’,
50 Hz) is used. In order to achieve the best results,
the device was calibrated and the calibration is vali-
Results
dated before each task and the level of the eye-tracking In order to explore the comprehension of the subjects,
cameras were adjusted based on the height of the sub- scores of the answers to the questions are calculated.
jects. At the final stage, the results of the subjects with The comprehension scores for each group are pre-
lower calibration score were ignored. The recorded sented in Table 1.
gaze data are then transferred for the further analysis. The analysis of the gaze data and score of the ques-
tions for each task can be used to study the effect of
layouts on the comprehension and cognitive load of
Material the subjects. To analyse gaze data and compare the
Two stories are visualized to design the Infographics variations of the cognitive load, a threshold method is
and are used as stimuli in this experiment. Each story applied to study the sequence of saccades. In the sec-
is presented with two different layouts (2 stories 3 2 ond analysis, the overall increase in the cognitive load
layouts = 4 stimulus). The graphical elements and col- for each stimulus is calculated using the CLS
ours used for each story are identical and only their (Cognitive Load Score) calculation methods.31
layout differ. The Infographics contain general infor-
mation from two different general areas. In the first
set, layout A-1 is based on a single Stairs (Steps) path Comprehension
from left to right and top to bottom, whereas layout In order to calculate the comprehension score for the
A-2 is based on a single Zigzag path from top to bot- tasks, immediately after each task, the subjects are
tom with the same story. In the second set, layout B-1 asked to answer a set of conceptual questions about
has divided the page into three major vertical columns, the viewed Infographics. Because of the short period
whereas layout B-2 has divided the page into three hor- of the tasks and existence of no time span between the
izontal rows. The schematic form of the layouts is pre- tasks and the questions, the long-term memory is not
sented in Figure 1. involved and only the comprehension is estimated.
In order to apply the layout, initially, the arbitrarily The final results presented in Table 1 demonstrate the
selected pieces of textual information are outlined and scores of the subjects in respect to every stimulus.
Majooni et al. 5

Table 1. Average gaze data for each stimuli.

Stimuli Blinks Pupil Average Average overall Average Efficiency Score (%)
(per minute) dilation (%) saccades fixation (ms) fixation
(mm) count

A-1 0.3 33.33 36.70 236 187 0.57 76

A-2 0.2 31.3 44.33 209 210 0.95 95
B-1 0.238 32.5 42.80 212 189 0.54 60
B-2 0.3 34.9 38.35 255 173 0.34 51

Figure 2. Sample of a saccade symbolic representation.

Saccade patterns salient position of the infographic. In addition, the sub-

When the visual attention is directed to another loca- jects tend to explore the whole infographic in the first
tion, the eyes move fast to position the fovea to the 2–5 s to create an overall understanding of the whole
point of attention to achieve higher resolution image story. In this analysis, we categorize the saccades for
from the attention point. This fast movement of the each stimulus into three groups (short, medium and
eye (relocation of gaze) is called as saccade.32 long saccades) according to their length. This grouping
Monitoring saccades can be very helpful in providing is related to the distance between the chunks inside
the scan path of the stimuli. As mentioned in the ear- every infographic layout. The medium size saccades
lier sections, the information used to design the info- are the ones which are in the range of the distance
graphic were initially outlined and each item has been between two neighbour chunks in the stimuli. The sac-
translated into one graphical element and accompa- cades longer than and shorter than the medium-sized
nied abstract textual contents (chunks). The chunks saccades for each stimulus are set as long and short
are placed such that subject sees one chunk at a time. saccades, respectively. The symbols (S, M and L) are
Accordingly, the fast eye movements (saccades) assigned to ‘short, medium and long’ saccades, respec-
between each fixation show the path in which the sub- tively. This method is similar to threshold filtering in
ject goes through the chunks in the infographic. the symbolic dynamics. Figure 2 demonstrates a sam-
However, the subject starts this path from the most ple of a saccade symbolic representation.
6 Information Visualization

Figure 3. Density map of the gaze data for story A and story B layouts.

As it can be seen from Figure 2, there exist more that the upper portion of the Infographics is more
than five repetition of type ‘S’ saccade between every salient.
couple of type ‘M’ saccade. By investigating the area of
interest (AOIs), it can be presumed that the subjects
relocate their gaze from one chunk to another chunk Cognitive load analysis
with type ‘M’ saccade whereas they tend to explore the
whole infographic in the beginning and at the end of There exist two other features besides the fixations
the experiment using type ‘L’ saccades. and saccades that can be recorded easily using the eye-
tracking devices. The number of the blinks and pupils
dilation during the experiment are two measures that
Density map can help estimation of variation of the cognitive
A density map (also called as heatmap) diagram pro- load.34–37
vides information about the number and duration of
eye fixations of a subject on stimuli acquired from an
eye-tracking device. In the density map diagram, the Pupil dilation. Klingner et al.36 have investigated the
longer fixations are represented by red and yellow col- effect of an increase in cognitive workload on the
ours, whereas green and blue colours represent less pupillary dilation. They have concluded that subtle
attention and shorter fixation time.33 Figure 3 demon- changes in the pupil diameter can be used to measure
strates density map diagram for story A and story B the cognitive workload. In this study, in order to com-
layouts of all subjects. As it can be seen from this fig- pare the overall cognitive load between two different
ure, the fixations are overlapped on the layout of the types of layouts for each story, the numerical integral
Infographics. According to the density map diagrams, dilation of the pupil diameter for each eye is calcu-
it can be concluded that majority of the fixations are lated. Figure 4 shows a sample of papillary dilation on
in the upper half of the page, which confirms the claim infographic type A-1.
Majooni et al. 7

Figure 4. Sample of papillary dilation on infographic type A-1.

Blinks. Blinks rate can also be a good measure of the dependent on the score of the questions, pupil dila-
cognitive load and fatigue.36,38–40 Based on the study tion, blinks rate and number of long saccades and fixa-
conducted by Recarte et al.,41 the number of blinks tions. The classification is operated using the Weka
can even be more effective in estimating the increase software, which is a collection of neural network algo-
in the cognitive load caused by the visual searching in rithms for data mining tasks.42 In this study, the
comparison to the pupil dilation. They claim that the Zigzag layout A-2 had the lowest amount of blinks rate
cognitive load estimation based on pupil dilation and pupillary dilation and highest amount of correct
method cannot differentiate between the mentally and answers and accordingly achieves highest efficiency
visually challenging tasks. However, in this study, simi- rate. In contrast, the Horizontal rows layout B-2 with
lar to the method suggested by Majooni et al.,31 the the lowest comprehension and highest pupillary dila-
blinks rates and the pupil dilations are recorded and tion has the lowest efficiency score.
stored in a database (along with the saccades, fixations
and the gaze location) for all the subjects.
Analysis of AOIs. The designed Infographics are based
on chunks order and their relations in a sequential for-
Efficiency. The efficiency measure is calculated using mat; therefore, in this experiment, the sequences of
the classification of the number correct answers to the the fixations on AOIs are investigated. In order to
questionnaire, the blinks rate, the saccades length, the complete this task, an indexed AOI is assigned to each
number of fixation and the average of fixation dura- one of the chunks with the order intended by the
tion. The normalized value of the Efficiency is a mea- designer. The average visit sequence of the AOIs for
sure between zero and one, when it is closer to one it each stimulus is then calculated based on the number
means the design has been more efficient from both and latency of the visit. For each AOI, the average visit
cognitive load and comprehension point of view. In order and average visit latency are assigned and then
the CLS method, five distinct and independent para- again a second sequence is generated based on these
meters (saccades length, fixations length, blinks rate, latencies and visit order. In the last step, the similarity
pupil dilation and the score of the questions) are effec- of the sequence of visited AOIs versus the intended
tive in classification of the value of the efficiency score. order of chunk’s sequence is estimated using classifica-
Each one of these parameters separately affects the tion of similar clusters method. Table 2 provides the
efficiency score. The maximum and minimum for average similarity of the AOIs index order versus
each parameter is first identified and then the classifi- chunks order. According to the similarity results, the
cation is applied to avoid the presence of any bias. The AOI sequence of the layouts A-2 is slightly more simi-
method can easily be applied to any type of info- lar to the intended chunk order compared to the AOI
graphic or visualization. The value is equally sequence of layout A-1. In the case of sequence
8 Information Visualization

Table 2. Average similarity of the AOI’s sequences versus besides using the fixation data for identification of
chunks order. attention, we classified the saccade data into three
categories and compared the eye movement beha-
Stimuli Story A (%) Story B (%)
viours of the subjects using the saccades classification
Layout 1 69 64 (Figure 2) and the pupil dilation (Figure 4).
Layout 2 72 41
Layout and cognitive load
In order to study the variation in the cognitive load
similarity, story B-1 has better similarity score com- and also calculate the average workload for each sti-
pared to B-2. The analysis of order of AOIs is based mulus, the average number of the blinks and the sum
on the calculation of Bayesian similarity of clusters of dilation of the pupils for each stimulus is calcu-
method and does not necessarily prove the efficiency lated for all the subjects. In order to calculate the
of any of the layouts and it can only be addressed as workload, the results are linearly normalized between
an insight. zero and one and workload score is calculated based
on the number of blinks and integral of pupil dilation
for each subject and stimuli. In addition, results pro-
Discussion vide that the average saccade length for the subjects
Comprehension viewing stimuli A-2 in the first 30 s has been slightly
longer than Stimuli A-1. This difference could be
The answer to the first research question of this study caused by the distances difference between chunks in
can be investigated by comprehension score of the each stimulus.
subjects, meaning that the layouts can be significantly The analysis of the recorded parameters for each
effective in conveying information. With the study of participant on each stimuli can be used to calculate
the scan path of each of the subjects, we noticed that overall efficiency index for each subject. The average
the subjects tend to move from one chunk to the other of overall efficiency index of each layout for all subjects
chunk usually following the path provided by the lay- is calculated and presented in Table 1. According to
out. The observation demonstrates that, during the the results, it can be concluded that the layout A-2
experiment, the subjects tend to go back a few (two or which is based on a Zigzag path has significantly higher
three) chunks and start over (revisit the chunks). This efficiency index compared to layout A-1. In addition,
revisit of AOIs can either be interpreted as the diffi- the layout B-2 (three vertical columns) has been more
culty of understanding43 or steps of making the con- efficient compared to top-down layout B-1 (three hori-
nection between the previous and current chunk zontal rows).
(inductive reasoning). The analytic study of pupil dila-
tion shows that this eye movement behaviour was
almost the same for all the subjects. Therefore, it can Conclusion
be claimed that it is related to the visual memory and Instructional design strategies, such as scaffolding the-
the process of mental model formation rather than the ory, if applied in the design of Infographics can be
difficulty of the information. effective in decreasing the cognitive load and reducing
the initial confusion. In this study, two contexts are
presented in the form of Infographics. The two con-
Saccades texts are initially split into small chunks (following the
According to Treisman and Gormican,44 human eye chunking principle) and then the chunks are presented
can process simple visual objects in a time period as on four different layouts (two for each context). All of
short as 40–50 ms. In this study, the number of revi- the layouts are organized according to the scaffolding
sits and dwell time for each AOI has been calculated. theory. The main goal of this study is to investigate the
The average dwell time for each AOI has been in the effect of the layout on the comprehension and cogni-
range of 80–250 ms. However, the dwell time was not tive load of the subjects. In order to investigate the
proportional to the number of words and elements in effect of each layout on comprehension of the partici-
each chunk. In some cases, two similar chunks with pants, a set of brief conceptual questions are asked
almost exactly the same amount of content had an about the Infographics from the subjects. In addition,
extremely different average dwell time. In this case, it to identify the effect of each layout on the cognitive
can be claimed that the subject was no more looking load of subjects, their gaze data (fixations, saccades,
at the AOI and actually was busy processing the infor- blink rate and pupil dilation) were recorded. The gaze
mation and building the mental model. In this study, data for the participants were analysed using simple
Majooni et al. 9

linear classification method. Finally, the results suggest 14. Pachalska M. The psychology of art and the evolution of
which layouts perform better in decreasing workload the conscious brain. J Nerv Ment Dis 2006; 194(8): 632–
and increasing comprehension. In addition, from the 634.
results of the experiments, in can be concluded that 15. Birkhoff GD. Aesthetic measure. Cambridge, MA and
using natural left-right (up-down) eye movement of London: Harvard University Press, 1933.
16. Cawthon N and Moere AV. The effect of aesthetic on the
the viewers could be useful in the layout design and
usability of data visualization. In: Proceedings of the 11th
improvement of the comprehension of the viewers international conference on information visualization, IV’07,
from the Infographics and visualizations. Zurich, 4–6 July 2007, pp. 637–645. New York: IEEE.
17. Hekkert P, Snelders D and van Wieringen PCW. ‘Most
Funding advanced, yet acceptable’: typicality and novelty as joint
predictors of aesthetic preference in industrial design.
This research was supported by University Sains Brit J Psychol 2003; 94(1): 111–124.
Malaysia, Centre of Instructional Technology and 18. Ludden GD, Schifferstein HN and Hekkert P. Surprise
Multimedia (Grant No. 1001/PMEDIA/817072). as a design strategy. Des Issues 2008; 24(2): 28–38.
19. Moshagen M and Thielsch MT. Facets of visual aes-
References thetics. Int J Hum-Comput St 2010; 68(10): 689–709.
20. Graves A. Creation of visualizations based on linked
1. Harris RL. Information graphics: a comprehensive illu- data. In: Proceedings of the 3rd international conference on
strated reference. Atlanta, GA: Management Graphics, web intelligence, mining and semantics – WIMS’13,
1996. Madrid, 12–14 June 2013.
2. Moere AV and Purchase H. On the role of design in 21. Wojtkowski W and Wojtkowski WG. Storytelling: its role
information visualization. Inform Visual 2011; 10(4): in information visualization. In: European systems science
356–371. congress, Crete, Greece, 16–19 October 2002, pp. 1–2.
3. Lankow J, Ritchie J and Crooks R. Infographics: the power Emerald Group Publishing Limited. Available at: http://
of visual storytelling. Hoboken, NJ: John Wiley & Sons, www.afscet.asso.fr/resSystemica/Crete02/Wojtkowski.pdf
2012. 22. Schnotz W and Kürschner C. External and internal
4. Larkin JH and Simon HA. Why a diagram is (some- representations in the acquisition and use of knowledge:
times) worth ten thousand words. Cognitive Sci 1987; visualization effects on mental model construction. Instr
11(1): 65–100. Sci 2007; 36(3): 175–190.
5. Faraday P. Visually critiquing web pages. In: Proceedings 23. Baker RM. Examples of scaffolding and chunking in
of the 6th conference on human factors the web, Austin, TX, online and blended learning environments. Soc Sci Res
19 June 2000. Netw. Epub ahead of print 2010. DOI: 10.2139/
6. Mol L. The potential role for infographics in science commu- ssrn.1608133.
nication. PhD Thesis, Athena Institute, Vrije Universiteit 24. Young JQ, Van Merrienboer J, Durning S, et al. Cogni-
Amsterdam, Amsterdam, 2011. tive load theory: implications for medical education:
7. Bertschi S, Bresciani S, Crawford T, et al. What is AMEE guide no. 86. Med Teach 2014; 36(5): 371–384.
knowledge visualization? Eight reflections on an evolving 25. Sweller J, Van Merrienboer JJ and Paas FG. Cognitive
discipline. In: Marchese FT and Bannisi E (eds) Knowl- architecture and instructional design. Educ Psychol Rev
edge visualization currents. London: Springer, 2013, pp. 1998; 10(3): 251–296.
13–32. 26. Huang W, Eades P and Hong SH. Measuring effective-
8. Pulizzi J. The rise of storytelling as the new marketing. ness of graph visualizations: a cognitive load perspective.
Publish Res Q 2012; 28(2): 116–123. Inform Visual 2009; 8(3): 139–152.
9. Huang W, Parsons P and Sedig K. Handbook of human 27. Gotel OC, Marchese FT and Morris SJ. On require-
centric visualization. New York: Springer, 2014. ments visualization. In: Proceedings of the second interna-
10. Gershon N and Page W. What storytelling can do for tional workshop on requirements engineering visualization,
information visualization. Commun ACM 2001; 44(8): 2007 (REV 2007), Washington, DC, 15–19 October
31–37. 2007, p. 11. Washington, DC: IEEE Computer Society.
11. Simon HA. Designing organizations for an information- 28. Card SK, Mackinlay JD and Shneiderman B. Readings in
rich world. In: Greenberger M (ed.) Computers, commu- information visualization: using vision to think. San Fran-
nication, and the public interest, vol. 37. Baltimore, MD: cisco, CA: Morgan Kaufmann Publishers, 1999.
The Johns Hopkins University Press, 1971, pp. 40–41. 29. Qin C, Zhou C and Pei T. Taxonomy of visualization
12. Lau A and Vande Moere A. Towards a model of infor- techniques and systems: concerns between users and
mation aesthetics in information visualization. In: Pro- developers are different. In: Proceedings of the Asia GIS
ceedings of the 11th international conference on information conference, Wuhan, China, 16–18 October 2003, pp.
visualization, IV’07, Zurich, 4–6 July 2007, pp. 87–92. 1–14.
New York: IEEE. 30. Nesbitt K and Friedrich C. Applying Gestalt principles
13. Fechner GT. Vorschule der aesthetik, vol. 1. Wiesbaden: to animated visualizations of network data. In:
Breitkopf & Härtel, 1876.
10 Information Visualization

Proceedings of the sixth international conference on informa- artificial environments. Entertain Comput 2013; 4(1):
tion visualization, London, 10–12 July 2002. 63–69.
31. Majooni A, Masood M and Akhavan A. Scientific visua- 38. Stern JA, Boyer D and Schroeder D. Blink rate: a possi-
lizations based on integrated model of text and picture ble measure of fatigue. Hum Factor 1994; 36(2): 285–
comprehension via eye-tracking. Procedia: Soc Behav Sci 297.
2015; 176: 52–59. 39. Fukuda K, Stern JA, Brown TB, et al. Cognition, blinks,
32. Duchowski A. Eye tracking methodology: theory and prac- eye movements, and pupillary movements during perfor-
tice. London: Springer, 2007. mance of a running memory task. Aviat Space Envir Md
33. Drusch G, Bastien J and Dinet J. From gaze plots to eye 2005; 76(7): 75–85.
fixation patterns using a clustering method based on 40. Benedetto S, Pedrotti M, Minin L, et al. Driver work-
Hausdorff distances. In: Proceedings of the 23rd French load and eye blink duration. Transport Res F: Traf 2011;
speaking conference on human-computer interaction, Sophia 14(3): 199–208.
Antipolis, 24–27 October 2011, p. 1. New York: ACM. 41. Recarte MA, Pérez E, Conchillo A, et al. Mental work-
34. Kahneman D and Beatty J. Pupil diameter and load on load and visual impairment: differences between pupil,
memory. Science 1966; 154(756): 1583–1585. blink, and subjective rating. Spanish J Psychol 2008;
35. Ahern S and Beatty J. Pupillary responses during infor- 11(2): 374–385.
mation processing vary with Scholastic Aptitude Test 42. Hall M, Frank E, Holmes G, et al. The WEKA data
scores. Science 1979; 205(4412): 1289–1292. mining software: an update. ACM SIGKDD 2009;
36. Klingner J, Kumar R and Hanrahan P. Measuring the 11(1): 10–18.
task-evoked pupillary response with a remote eye tracker. 43. Goldberg JH and Kotval XP. Computer interface evalua-
In: Proceedings of the 2008 symposium on eye tracking tion using eye movements: methods and constructs. Int J
research & applications, Savannah, GA, 26–28 March Ind Ergonom 1999; 24(6): 631–645.
2008. New York: ACM. 44. Treisman A and Gormican S. Feature analysis in early
37. Di Stasi LL, Antolı́ A and Cañas JJ. Evaluating mental vision: evidence from search asymmetries. Psychol Rev
workload while interacting with computer-generated 1988; 95(1): 15–48.