The Cartographic Journal
The Impact of Global/Local Bias on Task-solving in Map-related Tasks Employing
Extrinsic and Intrinsic Visualization of Risk Uncertainty Maps
--Manuscript Draft--
Manuscript Number: CAJ428
Full Title: The Impact of Global/Local Bias on Task-solving in Map-related Tasks Employing
Extrinsic and Intrinsic Visualization of Risk Uncertainty Maps
Article Type: Survey Paper
Keywords: Geovisualization method, avalanche risk, cognitive style, Navon's hierarchical figure,
combined extensive-intensive research design, eye-tracking.
Corresponding Author: Čeněk Šašinka, Assistant professor
Masaryk University
CZECH REPUBLIC
Corresponding Author Secondary
Information:
Corresponding Author's Institution: Masaryk University
Corresponding Author's Secondary
Institution:
First Author: Čeněk Šašinka, Assistant professor
First Author Secondary Information:
Order of Authors: Čeněk Šašinka, Assistant professor
Zdeněk Stachoň, Assistant professor
Petr Kubíček, Associated professor
Sascha Tamm, Associated Professor
Aleš Matas, Assistant professor
Markéta Kukaňová, Mgr.
Pavel Humpolíček, Assistant professor
Order of Authors Secondary Information:
Abstract: The form of visual representation affects both the way in which the visual
representation is processed and the effectiveness of this processing. Different forms of
visual representation may require the employment of different cognitive strategies in
order to solve a particular task; at the same time, the different representations vary as
to the extent to which they correspond with an individual's preferred cognitive style.
The present study employed a Navon-type task to learn about the occurrence of
global/local bias. The research was based on close interdisciplinary cooperation
between the domains of both psychology and cartography. Several different types of
tasks were made involving avalanche hazard maps with intrinsic/extrinsic visual
representations, each of them employing different types of graphic variables
representing the level of avalanche hazard and avalanche hazard uncertainty. The
research sample consisted of two groups of participants, each of which was provided
with a different form of visual representation of identical geographical data, such that
the representations could be regarded as "informationally equivalent". The first phase
of the research consisted of two correlation studies, the first involving subjects with a
high degree of map literacy (students of cartography) (intrinsic method: N = 35;
extrinsic method: N = 37). The second study was performed after the results of the first
study were analyzed. The second group of participants consisted of subjects with a low
expected degree of map literacy (students of psychology; intrinsic method: N = 35;
extrinsic method: N = 27).The first study revealed a statistically significant moderate
correlation between the students' response times in extrinsic visualization tasks and
their response times in a global subtest (r=0.384, p<0.05); likewise, a statistically
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
significant moderate correlation was found between the students' response times in
intrinsic visualization tasks and their response times in the local subtest (r=0.387,
p<0.05). At the same time, no correlation was found between the students'
performance in the local subtest and their performance in extrinsic visualization tasks,
or between their scores in the global subtest and their performance in intrinsic
visualization tasks. The second correlation study did not confirm the results of the first
correlation study (intrinsic visualization/"small figures test": r = 0.221; extrinsic
visualization/"large figures test": r = 0.135). The first phase of the research, where the
data was subjected to statistical analysis, was followed by a comparative eye-tracking
study, whose aim was to provide a more detailed insight into the cognitive strategies
employed when solving map-related tasks. More specifically, the eye-tracking study
was expected to be able to detect possible differences between the cognitive patterns
employed when solving extrinsic- as opposed to intrinsic-visualization tasks. The
results of an exploratory eye-tracking data analysis support the hypothesis of different
strategies of visual information processing being used in reaction to different types of
visualization.
Funding Information: Masaryk University
(MUNI/M/0846/2015)
Not applicable
Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Cognitive strategies
Map
Natureofsolvedproblem
map
literacy
domain
knowledge
level of
cognitive
function
mapcontent
visualization
method
situation
context
task
type
Figure 1. Triarchic structural model of performance when solving map-related task
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Figure 2. Snow avalanche hazard and hazard uncertainty map: left - extrinsic visual
representation; right - intrinsic visual representation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Figure 3. Differences in the encoding of variables between the extrinsic (first column)
and intrinsic (second column) form of visualization; third column shows the difference
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Analysis of performance outputs:
correctness and response time
Exploration of tasks processing
Extrinsic
Intrinsic
CLT
Extrinsic
Intrinsic
CLT
Group with high
map literacy
Group with low
map literacy
Extrinsic
Intrinsic
Correlation study I Correlation study II
Comparative
study
Group with low
map literacy
time
First phase:
confirmative studies
Second phase:
explorative study
Figure 4. Structure of used combined extensive-intensive research design
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
1) Students of Cartography
Extrinsic
Visualization
Intrinsic
Visualization
Instruction
{
{
{
Training task
{
One variable
Two variables
2) Students of Psychology
Extrinsic
Visualization
Intrinsic
Visualization
Instruction
{
{
Training task
One variable
{
{
Two variables
Figure 5. Structure of map-related tasks (students of cartography - top; students of
psychology - bottom)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Figure 6. Sample map-related tasks: Left (intrinsic visualization) - Select the area with a
moderate level of avalanche hazard. Right (extrinsic visualization) - Select the area
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Figure 7. A compound stimulus (left)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Figure 7. an item from the Compound Figures Test adapted into the MuTeP
environment (right).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
2,5
2,0
1,5
Global figuresLocal figures
Figure 8. Reaction times (s) for both subtests of CFT
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Stimulus
Reaktion/
Output
Stimulus
Reaktion/
Output
observed behaviour
by task processing
engaged cognitive
procesess
1)
2)
Figure 9. Process of investigating variables: 1) stimulus -> output; 2) stimulus ->
cognitive processing and associate behavior -> output
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Eye-tracking study
Extrinsic
Visualization
Intrinsic
Visualization
Instruction
{
{
Training task
One variable
{
{
Two variables
{
Principles of
uncertainty
Figure 10. Structure of map-related tasks in the eye-tracking study
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
0
2000
4000
6000
8000
10000
12000
14000
1 2 3 4 5 6 7 8 9 10 11
trial duraƟon:
EXTRINSIC
trial duraƟon:
INTRINSIC
Figure 11. LEFT Trial duration (in ms; left)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
0
1
2
3
4
5
6
7
8
1 2 3 4 5 6 7 8 9 10 11
transiƟon:
EXTRINSIC
transiƟon:
INTRINSIC
Figure 11. (right) between the map and the legend
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
IntrinsicExtrinsic
1
2
3
4
5
6
7
Figure 12. Number of transitions between the legend and the map
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
0
10
20
30
40
50
60
70
80
90
1 2 3 4 5 6 7 8 9 10 11
dwell Ɵme:
MAP
dwell Ɵme:
LEGEND
EXTRINSIC
Figure 13. left Dwell times related to the map and the legend in extrinsic
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
1 2 3 4 5 6 7 8 9 10 11
dwell Ɵme:
MAP
dwell Ɵme:
LEGEND
0
10
20
30
40
50
60
70
80
90
INTRINSIC
Figure 13. (right) visualization tasks (in %)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11
fixaƟon duraƟon:
MAP
fixaƟon duraƟon:
LEGEND
Figure 14. LEFT Fixation durations related to the map and the legend in extrinsic (left)
and intrinsic
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11
fixaƟon duraƟon:
MAP
fixaƟon duraƟon:
LEGEND
Figure 14. (right) visualization tasks (in ms)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
INTRINSICEXTRINSIC
200
250
300
350
400
Legend Map Legend Map
Figure 15. Fixation durations related to the map and the legend in
extrinsic (two boxplots on the left) and intrinsic (two boxplots on
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in
Map-related Tasks Employing Extrinsic and Intrinsic
Visualization of Risk Uncertainty Maps
Čeněk Šašinka1
, Zdeněk Stachoň1
, Petr Kubíček1
, Sascha Tamm3
,
Aleš Matas1
, Markéta Kukaňová1
, Pavel Humpolíček2
1
Center for Experimental Psychology and Cognitive Sciences,
Masaryk University
2
Department of Psychology, Masaryk University
3
Center for Applied Neuroscience, Freie Universität Berlin
The form of visual representation affects both the way in
which the visual representation is processed and the
effectiveness of this processing. Different forms of visual
representation may require the employment of different
cognitive strategies in order to solve a particular task; at the
same time, the different representations vary as to the extent to
which they correspond with an individual’s preferred
cognitive style. The present study employed a Navon-type task
to learn about the occurrence of global/local bias. The research
was based on close interdisciplinary cooperation between the
domains of both psychology and cartography. Several
different types of tasks were made involving avalanche hazard
maps with intrinsic/extrinsic visual representations, each of
them employing different types of graphic variables
representing the level of avalanche hazard and avalanche
hazard uncertainty. The research sample consisted of two
groups of participants, each of which was provided with a
different form of visual representation of identical
geographical data, such that the representations could be
regarded as “informationally equivalent”. The first phase of
the research consisted of two correlation studies, the first
Manuscript - with full author details
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 2
involving subjects with a high degree of map literacy (students
of cartography) (intrinsic method: N = 35; extrinsic method: N
= 37). The second study was performed after the results of the
first study were analyzed. The second group of participants
consisted of subjects with a low expected degree of map
literacy (students of psychology; intrinsic method: N = 35;
extrinsic method: N = 27).The first study revealed a
statistically significant moderate correlation between the
students’ response times in extrinsic visualization tasks and
their response times in a global subtest (r=0.384, p<0.05);
likewise, a statistically significant moderate correlation was
found between the students’ response times in intrinsic
visualization tasks and their response times in the local subtest
(r=0.387, p<0.05). At the same time, no correlation was found
between the students’ performance in the local subtest and
their performance in extrinsic visualization tasks, or between
their scores in the global subtest and their performance in
intrinsic visualization tasks. The second correlation study did
not confirm the results of the first correlation study (intrinsic
visualization/“small figures test”: r = 0.221; extrinsic
visualization/“large figures test”: r = 0.135). The first phase of
the research, where the data was subjected to statistical
analysis, was followed by a comparative eye-tracking study,
whose aim was to provide a more detailed insight into the
cognitive strategies employed when solving map-related tasks.
More specifically, the eye-tracking study was expected to be
able to detect possible differences between the cognitive
patterns employed when solving extrinsic- as opposed to
intrinsic-visualization tasks. The results of an exploratory eyetracking
data analysis support the hypothesis of different
strategies of visual information processing being used in
reaction to different types of visualization.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 3
Keywords: Geovisualization method, avalanche risk,
cognitive style, Navon’s hierarchical figure, combined
extensive-intensive research design, eye-tracking.
1. Introduction
Space and time play a crucial role during hazardous events in
general and natural hazards in particular. Successful decision
making during emergency situations depends on the availability
and relevancy of information presented in the right time and an
understandable way. Such information improves the transparency
and credibility of the decisions taken.
Only “raw” spatial data is currently available for emergency
management support. Emergency decision makers all around the
world have available maps with natural hazards (such as flood,
avalanche, and landslides), vulnerable zones, land use or geology,
and make decisions based on implicit information inferred from
such map sources. Such implications are not straightforward and
may even differ according to the different professional background
of a particular decision maker. Or more generally, the form of
information visualization should be adjusted to the cognitive
characteristics of the users. The responsible persons are able to
process only a limited number of graphics (maps) in case of an
emergency. This situation is even more critical when working
under severe time pressure. The visualization form can
significantly influence the final decision. Zhang and Goodchild
(2002) proved that visual form can improve the communication
about spatial data uncertainty within spatial analysis and spatial
decision support. Uncertainty often possesses spatial patterns.
Uncertainty visualization can thus reveal such patterns and serve
not only for presentation but also for exploration and visual
analysis of spatial data.
The present study is a continuation of an earlier research project on
crisis management (Konečný et al., 2011; Staněk et al. 2010). The
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 4
study focuses on perception and cognitive processing of different
forms of bivariate visual representation, using alternative visual
representations derived from identical avalanche hazard datasets
(Kunz, 2011; Kunz and Hurni, 2011). The employment of different
cartographic bivariate visualizations of the same area and topic
made it possible to create meaningful and complex stimuli that
were visually different and at the same time fulfilled the
requirement of informational equivalence (Larkin and Simon,
1987).
A map functions as a communication channel (Koláčný, 1977),
and when used as a stimulus material in psychological
experiments, it enables the researcher to purposely manipulate the
form of the communicated information. The objective of the study
was to investigate the impact of different types of visualization and
different cognitive styles on the effectiveness of solving maprelated
tasks.
2. Uncertainty visualization and the use of bivariate visualization
MacEachren (1992) suggested the use of Bertin’s graphic variables
to depict uncertainty and even added specialized variables for
depicting uncertainty including crispness, resolution, and
transparency. Gershon (1998) grouped these into intrinsic and
extrinsic visual variables depending on whether the variable is
visually separable from the variable depicting the actual attribute.
While extrinsic variables are separable, intrinsic variables are not.
Another logical step is to describe how these variables including
possible additions or modification, might be logically matched
with different components of data uncertainty (Buttenfield 1991,
MacEachren 1992, Leitner and Buttenfield 2000). MacEachren
(1992), for instance, stated that the graphical variables size and
color value are most appropriate for depicting uncertainty in
numerical information, while color hue, shape, and perhaps
orientation can be used for uncertainty in nominal information.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 5
More recently MacEachren et al (2012) focused on discrete
symbols that could be used to signify the uncertainty of individual
items within information graphics, maps. The experiments
examine relative effectiveness of a set of uncertainty representation
solutions—differing in the visual variable leveraged and level of
symbol iconicity. Trau and Hurni (2007) and Kunz (2011)
theoretically analyzed the suitability of visual variables and
visualization techniques for uncertainty depictions in hazard
prediction maps.
Most applied uncertainty visualizations in the field of natural
hazards are simplistic univariate representations where hazard
related data are displayed in one map and inherent uncertainties are
depicted in a second map display (Kunz 2011). Kunz was among
the first to propose the use of bivariate depiction of studied
phenomenon and its uncertainty, applied this approach in the
dynamic environment and even presented brief feedback from
expert users.
Capabilities and limitations of bivariate map types based on the
combination of visual variables and symbol dimensionalities
(point, line, and polygon) were studied and tested also by Elmer
(2013). Using the selective attention theory, he empirically tested
eight bivariate map types for map reading tasks recording their
accuracy and response time. These tests also included the
combination of size/value (Choropleth/Graduated symbols) and
value/hue-saturation (Bivariate Choropleth) which are relevant for
our study. Both aforementioned combinations performed above
average for accuracy and response time and were also rated
positively by users as an appropriate combination to read and
understand the information on the map. However, the author
himself concluded that the study only revealed significant
differences in perceived combinations and further research is
needed in order to understand different mental strategies of users
and identify their cognitive behavior.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 6
Cognitive cartography encompasses the application of cognitive
theories and methods to understanding maps and mapping and the
application of maps to understanding cognition (Montello 2002).
Different ways of displaying the same spatial information can
dramatically affect problem-solving performance. Spatial cognition
research uses distinct concepts of informational and computational
equivalence of representations (Simon 1978). Alternative
cartographic visualizations follow the premise of information
equivalency. The possibility to visualize (code) the same spatial
information in an alternative way which is informationally
equivalent offers valuable input material for comparison of
cognitive processes used for decoding. Such a comparative
principle enables better understanding of the human cognitive
apparatus. Differences in representations deal not only with
various visual forms but also with different operations necessary
for their decoding and interpretation.
Cognitive assumptions are closely connected with visual variables
used for uncertainty visualization. Both methods differ in the type
of visualization method and visual variables (intrinsic and
extrinsic). The perception of variables has a close connection with
the theory of pre-attentive perception (Treisman and Gelade 1980,
Wolfe et al 1989). The extrinsic graphical variable size is generally
considered to be a pre-attentive feature. Such features are
appropriate for the determination of presence or absence or
particular elements or boundary detection. On the other hand
intrinsic graphical variable saturation has not been confirmed as
pre-attentive.
However, the situation is rather different when using a
combination of both uncertainty portraying variables with a main
attribute variable. The intrinsic visualization method combines
color hue for the main attribute value and saturation for its
uncertainty. The resulting map legend is comprised of 9 categories
and constitutes a higher potential cognitive load for users. The
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 7
extrinsic visualization method combines color saturation for the
main attribute value and proportional circle size for uncertainty.
The map legend is comprised of only 6 categories.
3. Cognitive Style in Map-Related Tasks
Kozhevnikov (2007) defines cognitive style as heuristics (i.e.
strategy derived from experience with similar problems used for
processing external information). An individual’s cognitive style
can be detected at all levels of perception, from the elementary and
highly automated ones to those that are complex and conscious.
The concept is used to refer to the way individuals think, perceive
and orient themselves in the environment. According to Brigham et
al. (2007), cognitive style is a pervasive bipolar dimension that is
stable over time and can be studied using psychometric techniques.
He further states that a cognitive style may be value-differentiated,
meaning that it describes differences concerning value rather than
quality. However, cognitive style is far from a well-defined
construct, both with respect to its content and application to
different levels of the personality system. A vast range of different
interpretations of cognitive styles exist (see Witkin, 1967; Rayner,
2000; Kirton, 1989; Pask, 1976). Riding and Cheema (1991)
surveyed more than 30 conceptions of cognitive style, concluding
that each of the investigated conceptions pertains to one of two
principal dimensions. The verbal-imagery dimension encompasses
an individual’s preference for representing information in
words/verbal associations, or in mental pictures. The wholist–
analytic dimension is characterized as an individual’s preference
for processing information either in integrated wholes or in
discrete parts. Given the nature of the tasks used in the study and
the differences between intrinsic and extrinsic visual
representations (see Fig. 2), it was the wholist–analytic dimension
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 8
that was of greater importance for the present study. According to
Graff (2003), wholist–analytic cognitive style can be defined as a
tendency to process information either as an integrated whole or in
discrete parts of that whole. The wholist-analytic cognitive
dimension is based on the conception of global/local bias (see Dale
and Arnell, 2014) related to whether visual information is
perceived at a broad (global) level, or at a more focused (local)
level, with more attention being paid to partial characteristics of
objects and phenomena and to their analytical processing. Rezaei
and Katz (2004) add that globally-oriented individuals consider
phenomena in a broader perspective and context. Analyticallyoriented
individuals, on the other hand, view each situation as an
aggregate of discrete elements, typically preferring to focus on one
or two elements at a time at the expense of other elements/aspects.
Graff (2003) further states that analytically-oriented individuals are
better at apprehending concepts in parts, but may experience
difficulty integrating such concepts into complete, consistent
wholes, while globally-oriented individuals view concepts as
wholes, but are unable to separate individual aspects of the
concepts into discrete parts. According to Kozhevnikov (2007), the
analytical cognitive style tends to be characterized as convergent,
differentiated, sequential, reflective and deductive, whereas the
“global” style has been described as divergent, intuitive, impulsive,
inductive, and creative. Globality is often discussed in connection
with e.g. attentional breadth in selective attention (Dale and Arnell,
2014) and rapid scene categorization (Brand and Johnson, 2014).
The level of an individual’s performance in map-related tasks is
a result of the interaction of three variables: a) user characteristics;
b) task type and situational context; c) map-related
characteristics/type of visual representation (Fig. 1). Wehrend and
Lewis (1990) constructed a comprehensive catalogue of maprelated
operations, including identification (identifying visual
characteristics of features on the map), localization (determining
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 9
the absolute or relative position) or categorization (placing in
specifically defined divisions in a classification; this may be done
by color, shape or size). Map-related tasks can be ranked
depending on their relative difficulty, from relatively easy
operations of “finding the shortest path” (e.g. Jones, 1997) to
highly complex tasks (Smith Mason et al., 2016) of “planning
military operations” (Hofmann et al., 2015), which require several
concepts to be explored simultaneously, compared and integrated.
Working with a map always needs to be viewed as mental
manipulation with semantically rich material rather than a mere
visual search and processing of visual stimuli (MacEachren and
Taylor, 1994, Montello, 2009, Roth et al., 2011). The highest-level
performance in fulfilling the task can be expected if the cognitive
style of the user matches the nature of the task (Hammond, 1996)
and the form of visual representation used. For instance, at a task
requiring analytical thinking, an analytically-oriented individual
can be expected to perform better than a globally-oriented
individual with the same degree of cartographic literacy (Hojnik
and Hus, 2013) and domain knowledge (Alexander, Kulikowich
and Schulze,1994). Additionally, a form of visualization allowing
for sequential analytical processing will result in a better
performance than a visualization requiring simultaneous and
intuitive processing. The user’s performance will be the best if all
three of the areas (i.e. type of task, type of cognitive style and form
of visualization) are in consonance.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 10
Figure 1. Triarchic structural model of performance when solving
map-related task
From the point of view of experimental psychology, maps are
highly valuable as stimulus material (Olson, 1979) in that they
constitute complex external representations with variables
amenable to accurate control and change of value. By representing
identical content (data), different forms of visualization are
informationally equivalent, enabling computational equivalence to
be studied (Larkin and Simon, 1987). Any differences in
performance and in the way an individual works with different
maps can be viewed as directly linked to the difference in visual
presentation of the same content. In the present study, avalanche
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 11
hazard maps were used with intrinsic vs. extrinsic visual
representations (Fig. 2, right and left, respectively).
Figure 2. Snow avalanche hazard and hazard uncertainty map: left
– extrinsic visual representation; right – intrinsic visual
representation (adapted from Kunz, 2011)
The difference between the extrinsic and intrinsic representation
exists in the way the two thematic layer variables (snow avalanche
hazard and snow avalanche hazard uncertainty) are represented. In
extrinsic representation, the avalanche hazard level is expressed in
terms of variation in the saturation of the color blue, while hazard
uncertainty is represented by the size of the dots. In intrinsic
representation, the same map base is used, with the avalanche
hazard expressed in terms of hue, and hazard uncertainty expressed
in terms of variation in the saturation of the hue. Both types of
visual representations use a combination of two graphic variables.
However, while the extrinsic type uses two different modalities
(hue and size) for the presentation of two phenomena, intrinsic
representation is expressed in terms of variation of two properties
(hue and saturation) of a single modality (color). There is a clear
difference in categorization between the two types of
representations. In the intrinsic visual representation, 9 (3 x 3)
categories are encoded explicitly; the extrinsic visual
representation, on the other hand, has separate categories for each
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 12
of the two phenomena it represents (3 for avalanche hazard level
and 3 for hazard uncertainty), with only 6 categories being
encoded explicitly (although there are a total of nine combinations
as well). Even though color properties (hue and saturation in this
particular case) tend to be regarded as two independent variables in
the field of cartography (Bertin, 1973), in the psychology of
perception color properties are viewed as interacting with each
other (D’Zmura, 1991, Itti and Koch, 2000; Lindsey et al, 2010).
For instance, a desaturated blue on light (white) background will
be less luminous and thus more salient than a fully saturated
yellow (e.g. Nothdurft, 2000; see Fig. 3). The phenomenon can be
applied to map reading as well.
Figure 3. Differences in the encoding of variables between the
extrinsic (first column) and intrinsic (second column) form of
visualization; third column shows the difference in salience
between a fully saturated yellow (1) and partially desaturated blue
(6) on white and black backgrounds
The objective of the presented study was to investigate the
relationship between different forms of bivariate visual
representation and an individual’s cognitive style as reflected by
their performance in Navon’s test of hierarchical figures (see
section 4.2.1). We hypothesize (1) a link between global
processing efficiency and extrinsic visualization abilities, and (2) a
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 13
link between local processing efficiency and intrinsic visualization
abilities. We thus consider that a variation of both hue and
saturation using a single color modality (intrinsic) would refer to a
more local approach, whereas two types of independent graphic
hues and size variables (extrinsic) will be considered a more global
approach. In other words, the elaboration of two parameters (hue
and saturation) of one modality - color (intrinsic) would be linked
to local processing abilities, while considering two different
parameters, hue and size (extrinsic), simultaneously would be
linked to global processing abilities.
4. Method
The study uses a mixed (confirmatory-exploratory) research
design, with the aim to combine extensive research (for data
collection and subsequent statistical analyses) with a more in-depth
exploratory data analysis (EDA; see Andrienko and Andrienko,
2005). The advantages of the mixed research design, along with its
theoretical grounding, have been described by Šterba et al. (2014),
who also provide a sample study. The authors use the term “mixedresearch
design”, which, however, tends to be viewed as referring
to a combination of qualitative and quantitative methods,
particularly in the context of social sciences and constructivist
approaches (see Leech and Onwuegbuzie, 2009; Creswell, 2003).
In order to distinguish between the two concepts we propose the
term “combined extensive-intensive research design”. A combined
mixed extensive-intensive research works primarily with objective
data, combining a confirmatory stage of the research with an
exploratory stage. The first (confirmatory) phase of our research
comprised two correlation studies. The first consisted of the
collection of data on subjects with a high degree of map literacy
(students of cartography), while the second focused on subjects
with a low degree of map literacy (students of psychology). After
the confirmatory phase, an exploratory phase followed which was
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 14
represented by a comparative eye-tracking study. The aim of the
comparative study was to reveal possible differences in tasksolving
strategies depending on the type of visualization employed
by usage of an exploratory data analysis. The structure of the
research is shown in detail in Fig.4. While extensive methods are
concerned only with the effect of stimulus on behavioral outputs in
the sense of speed or correctness intensive methods concentrate on
the process alone, as in, what happens between stimulus and
reaction? And this is also the purpose of the eye-tracking
comparative study which can deeper illuminate results of the
previous phase of the study.
Figure 4. Structure of used combined extensive-intensive research
design
5. Correlation Studies I and II
The objective of the first phase of the research was to investigate
the relationship between an individual’s cognitive style and the
type of visualization used. By way of the first phase, two
correlation studies were conducted. The design of the second study
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 15
was adjusted based on the results of the first study. More
specifically, several items were added to the subtest involving a
one variable. The second study also included tasks focused on the
degree of understanding of the concept of visualization of snow
avalanche hazard and hazard uncertainty. These more complex
tasks are not part of the present article, which focuses primarily on
lower cognitive processes (e.g. visual search as reflecting an
individual’s cognitive style).
5.1. Participants
In the first correlation study, the research sample consisted of
students of geography in the 1st to 3rd year of their studies. A total
of 73 volunteers aged 19 to 27 years were tested. For the extrinsic
visualization task, there were 37 subjects (19 male, 18 female),
while 35 subjects (19 male, 16 female) were given a task
employing intrinsic visualization. In the second correlation study, a
total of 62 volunteers aged 19 to 55 years were tested, all students
of psychology in the 1st to 3rd year of their studies. For the
intrinsic visualization task, there were 35 subjects (8 male, 27
female), most of them aged 19 to 25 years, with one male outlier
aged 37 years; 27 subjects (3 male, 24 female) were given the
extrinsic visualization task. Most of the subjects were 19 to 28
years, with one male outlier aged 55 years. None of the subjects
had any previous experience of participation in cartographic
visualization testing.
5.2. Aparatus
The test was created and administered in MuTeP (Multivariate
Testing Program; see Kubíček et al., 2014, Kubíček et al., 2016,
Štěrba et al., 2015). After every data collection a group discussion
followed. Participants were asked about their subjective experience
by elaboration of the test battery.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 16
5.3. Stimulus Material - Map-related Tasks
In the correlation studies, avalanche maps were used showing both
the level of the snow avalanche hazard (expressed in terms of load
on the snowpack) and the level of avalanche hazard uncertainty.
Both variables were expressed by means of a three-level scale
(Low – Moderate – High). The map consisted of a base layer and a
thematic layer, with the latter being explained by the legend of the
map. In the experimental tasks, the base layer played only a very
marginal role. The subjects’ performance depended on their ability
to comprehend the instructions, decode the legend, and perform a
visual search to locate the four numbered target areas and decide
which of them meets the criteria given. There was always only one
correct answer. The visual scene consisted of a map, a map legend,
task instructions located at the top of the computer screen and
numbered “buttons”. The instructions were always presented
separately on a blank screen. No time limit for the instructions was
given so that there would be enough time for the participants to
comprehend the instructions. The overall test structure was as
follows:
1. Training – 2 training tasks
a. identify the level of variable A (avalanche hazard)
b. identify the level of variable B (avalanche hazard
uncertainty)
2. One variable tasks
a. identify the level of variable A (avalanche hazard) –
low, moderate, and high respectively.
b. identify the level of variable B (avalanche hazard
uncertainty) – low, moderate, and high respectively.
3. Two variable tasks
a. identify the level of variable A (avalanche hazard)
and variable B (avalanche hazard uncertainty) –
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 17
combination of low, moderate, and high
respectively).
Both tests were made up of one- and two-variable tasks. In the
first correlation study, the participants were given five test tasks
(see Fig. 5 top line) – two for one variable and three for two
variables. In the second correlation study (performed on subjects
with a low degree of map literacy) the batch of one-variable tasks
included four items instead of two (see Fig. 5 bottom line).
Figure 5. Structure of map-related tasks (students of cartography –
top; students of psychology – bottom)
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 18
Visualizations used in training were similar to the rest of the test.
Sample instructions of one variable task are as follows: “Select the
area with a moderate level of avalanche hazard” (see Fig. 6 left).
The subjects completed each task by clicking on the respective
“button”. Instructions of two variables task were formulated in a
following way: “Select the area with a high level of avalanche
hazard accompanied by a low level of avalanche hazard
uncertainty” (see Fig. 6 right).
Figure 6. Sample map-related tasks: Left (intrinsic visualization) –
Select the area with a moderate level of avalanche hazard. Right
(extrinsic visualization) – Select the area with a high level of
avalanche hazard accompanied by a low level of avalanche hazard
uncertainty.
5.4. Stimulus Material - Compound Figures Test
The second part of the test battery consisted of a psychological test
of cognitive style. It contained a variation of Navon’s Hierarchical
Figures Test (Navon, 1977) – a compound figures test (see Fig. 7,
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 19
left) which was designed for the purposes of the present study
using MuTeP. Navon’s test (Navon, 1977) is one of the most
frequently used methods for the measurement of the wholistanalytic
dimension of cognitive processing (e.g. Brand and
Johnson, 2014; Duchaine et al., 2007; Yovel et al., 2005, Milne
and Szczerbinski, 2009). It enabled us to measure a person’s ability
to direct attention either to the local level of the visual stimulus
material or to the global level. Bouvet et al. (2011) used tasks
based on the Hierarchical Figure test and found evidence that
participants exhibit a similar processing style across modalities
with respect to both vision and audition modalities. Thus these
findings support the theory that Navons Hierarchical Figures can
be understood and used more generally as a measure of wholisticanalytic
cognitive style and not only in the narrower way, as an
indicator of a visuo-attentional global/local style. The compound
figures test included instructions, several training items and 32 test
items. The stimulus material consisted of a set of large, single-digit
numbers compounded of small numbers. There were two subtests,
local and global, each comprising 16 items. The participants were
instructed to identify the small numbers in the first subtest and the
large numbers in the second subtest by clicking on one of four
available buttons as quickly as possible (see Fig. 7, left). The
overall response time of the respondents was recorded
electronically; the average response time per item was calculated.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 20
Figure 7. A compound stimulus (left); an item from the Compound
Figures Test adapted into the MuTeP environment (right).
5.5. Results
Two subjects with a high error rate for CFT were excluded from
further analyses. One student of cartography erred 5 times in the
local subtest and 6 times in the global subtest. One student of
psychology erred 12 times in the first subtest. The number of
mistakes made by the other participants was insignificant; most of
them achieved 100% accuracy, others erred less than three times.
Reaction times of the outlying errors were included in the analyses.
The overall score for the local and global subtests were counted
based on 15 items from each subtest. The first item of each subtest
was excluded from the analysis. In line with previous findings, the
internal consistency (Cronbach’s alpha) for both subtests was
found to be high (α = 0.805 and α = 0.864, respectively).
Differences in reaction times between the two subtests (Fig. 8)
were found to be in agreement with previous findings as well. A
paired t-test revealed faster processing of global figures by 15%
(N=133; mean reaction time for the local subtest s=1.9; SD=0.2;
mean reaction time for the global subtest s=1.6; SD=0.17; df=132;
p=0.01). The results showed a positive correlation between the two
subtests (r=0.437, p<0.01).
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 21
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 22
Figure 8. Reaction times (s) for both subtests of CFT
The error rate of students of cartography was 7.2% for intrinsic
visualization and 9.7% for extrinsic visualization. Students of
psychology exhibited a significantly higher error rate in both
subtests: 13.8% for intrinsic visualization and 20.0% for extrinsic
visualization. The overall error rate was counted from the
individual error rates of all the subjects who completed the
cartographic part of the research, including outliers, which were,
however, excluded from further analyses. Outlying subjects in the
research sample of students of cartography were those who erred
two and more times (out of the total of five items (i.e. those with
an error rate >40%; intrinsic visualization – 1 participant; extrinsic
visualization – 4 participants). As regards the research sample
consisting of students of psychology, 4 subjects were excluded
(extrinsic visualization) who erred three and more times (out of 7;
error rate >40%). Although the above outliers were excluded from
further analyses, all reaction times of remaining participants were
(as in CFT) included (see Tab. 1 and 2).
Students of
Cartography
Intrinsic
visualization (n=34)
Extrinsic visualization
(n=33 )
local
subtest
global
subtest
local
subtest
global
subtest
One Variable .400**
.180 -.021 .375*
Two variables .168 -.114 .304*
.302*
Overall Score .387* .050 .176 .384*
** Correlation is significant at the 0.01 level; * Correlation is significant at the 0.05 level
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 23
Table 1. Students of Cartography: Relationship between Maprelated
Task Performance (response time) and CFT Results
(Pearson’s Correlation; one-tailed)
Students of
Psychology
Intrinsic
visualization (n=35)
Extrinsic visualization
(n=23)
local
subtest
global
subtest
local
subtest
global
subtest
One Variable .211 .222 .100 .291
Two Variables .106 .230 -.079 -.140
Overall score .221 .266 .027 .135
** Correlation is significant at the 0.01 level; * Correlation is significant at the 0.05 level
Table 2. Students of Psychology: Relationship between Maprelated
Task Performance (response time) and CFT Results
(Pearson’s Correlation; one-tailed)
5.6. Interpretation of the Results
The results of the CFT employing a variation of Navon’s
hierarchical figures test confirmed the effect of global precedence;
the average response time in the global subtest was lower by 15%
compared with the local subtest. In addition, the results showed a
positive correlation between the two subtests, which is in line with
the expectation that an individual’s processing capacity in both
tasks will be driven by the same psychomotor speed (Spirduso,
1980). After the exclusion of subjects with high error rates (either
in CFT or in map-related tasks) the subjects’ results in CFT and
map-related tasks were tested for possible correlations. The
analysis was conducted by means of comparison of overall
response times in map-related tasks with response times in CFT.
The results of the first correlation study (on subjects with a high
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 24
degree of map literacy) revealed a positive correlation between
scores obtained in the local subtest and response times in maprelated
tasks employing intrinsic visualization; at the same time, a
positive correlation was found to exist between global-subtest
response times and response times in map-related tasks employing
extrinsic visualization. The fact that no cross-correlation was found
between either the local subtest and extrinsic visualization or
between the global subtest and intrinsic visualization (although the
subjects’ performance in map-related tasks was also driven by the
same psychomotor speed) renders the above findings even more
significant. In the second correlation study (on students of
psychology), no significant relationship between the variables was
found. We suppose that the absence of any correlation might have
been caused by low levels of map literacy, which was reflected in
comparatively high error rates.
6. Comparative Study: Eye-tracking
The reason for replicating the correlation studies using an eyetracking
system was to obtain additional information which would
provide deeper insight into the process of solving map-related
tasks and related cognitive strategies. With the decrease in cost of
the technology, eye-tracking became a more widely utilized
method, not one only used in cartographic studies (see Alaçam and
Dalci, 2009; Çöltekin et al., 2009; Popelka and Brychtová, 2013;
Popelka and Dědková, 2014; Ooms et al., 2012). The aim of the
eye-tracking study was to investigate the impact of different types
of cartographic visualization on task-processing. While the
confirmatory phase of the research focused on the speed and
“correctness” of reaction to visual stimuli, the exploratory phase
consisted of investigating the very process of task-solving (see Fig.
9).
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 25
Figure 9. Process of investigating variables: 1) stimulus -> output;
2) stimulus -> cognitive processing and associate behavior ->
output
6.1. Participants
A total of 14 subjects aged 21 to 35 years volunteered to
participate in the study; they were either university students or had
already completed their university studies. None of the participants
received any education in cartography or related fields. The
participants were divided into two groups of seven.
6.2. Apparatus
The device used in the study was EyeLink 1000, a remote eyetracking
system manufactured by SR Research. The original test
battery presented in MuTeP was adapted to the environment of
Experiment Builder, a software tool by SR Research for creating
eye-tracking experiments.
Eye-tracking data were saved in a format enabling an exploratory
data analysis to be carried out in EyeLink Data Viewer, software
that can be used for viewing, filtering and processing eye-tracking
data recorded with an EyeLink eye-tracker. The application offers
a range of eye-tracking metrics. In addition, it enables the analysis
of transitions between defined areas of interest (AOIs). In the
analysis, visualization tools such as an attention map and video
analysis were also used. After every single data collection an
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 26
individual inquiry followed. Participants were asked about their
subjective experience by elaboration of a test battery,
6.3. Stimulus Material
The original test structure was slightly modified (see fig. 10) and
several new training tasks were added to the original test battery.
A detailed graphical explanation of the principles of avalanche risk
and uncertainty was added in order to properly introduce both the
risk mapping and its uncertainty. These should enhance the
understanding of the experiment by non-experts.
Figure 10. Structure of map-related tasks in the eye-tracking study
6.4. Results and Interpretation
The exploratory data analysis was performed at both the subject
level and group level (extrinsic vs. intrinsic visualization). One
participant from the extrinsic visualization group was excluded
from the analysis due to a heavy data dropout. Several areas of
interest were defined, with key AOIs being set around the legend
and the map. The following measures were analyzed: trial
duration, number of transitions between the legend and the map,
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 27
the number of fixations on the legend and on the map, dwell time1
on the legend andon the map, and fixation durations.
Fig. 11 (graph on the left) shows average trial durations for the
eleven tasks used in the study. The first seven tasks employ only
one variable (level of avalanche hazard or avalanche hazard
uncertainty), while the rest of the tasks inquire about both
variables. The area represented by the map was identical in all the
tasks. The graph on the right in Fig. 11 shows the number of
transitions between the map and the legend for each task (in both
directions). Fig. 12 (box plots) shows the average number of
transitions and their spread for both extrinsic and intrinsic
visualization tasks. As is evident from the graphs, the subjects’
performance across the individual tasks was much more
homogeneous when working with extrinsic visualization; the
subjects’ performance at intrinsic visualization tasks, on the other
hand, appears to have been more sensitive to the nature of the task.
The higher variability might have been caused by greater
differences in the difficulty of discriminating between different
levels of brightness (caused by the interaction between hue and
saturation). The differences in brightness (see Fig. 2 right and Fig.
3 second and third columns) make it e.g. easier to distinguish
between low uncertainty/low avalanche hazard and low
uncertainty/moderate avalanche hazard than between high
uncertainty/low hazard and high uncertainty/moderate hazard.
1
dwell time refers to the sum duration of fixations falling within a
particular AOI; fixation duration refers to the duration of
individual fixations.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 28
Figure 11. Trial duration (in ms; left) and number of transitions
(right) between the map and the legend
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 29
Figure 12. Number of transitions between the legend and the map
The following analyses focus on the differences between
extrinsic and intrinsic visualization as reflected by the differences
in dwell times and fixation durations (related to the legend and the
map). Graphs in Fig. 13 show the percentages of map/legend dwell
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 30
time for all the tasks. For extrinsic visualization tasks, the legend
dwell time ranges from 10 to 30% (of the total dwell time); visual
search time (map dwell time) ranges from 60 to 80%. Intrinsic
visualization, on the other hand, exhibits a far greater variability
across the individual tasks; for the legend, the dwell time ranges
from below 10% in one task to over 50% in another. In the latter
case, the time needed for decoding the legend even exceeds the
time needed to locate the target in the map.
Figure 13. Dwell times related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in %)
Fig. 14 presents a comparison of map and legend fixation
durations in intrinsic (right) and extrinsic (left) visualization tasks.
Fig. 14 shows the average fixation durations and spreads. As with
dwell times, the data show lower variability across the individual
tasks for extrinsic visualization: the curves of fixation durations are
nearly identical (see, for instance, the data concerning the second
task, with fixation durations slightly above 250 ms for both the
legend and the map). Intrinsic visualization, however, led to a
different situation during the same (second) task: legend-related
fixations were on average more than 100 ms longer than maprelated
fixations. From Fig. 15 it can be seen that extrinsic
visualization caused average fixation durations to last
approximately 250 ms for the map and the legend alike. In the case
of intrinsic visualization, the same value was true for map-related
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 31
average fixation durations; the average duration of legend-related
fixations, however, far exceeded 300 ms.
Figure 14. Fixation durations related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in ms)
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 32
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 33
Figure 15. Fixation durations related to the map and the legend in
extrinsic (two boxplots on the left) and intrinsic (two boxplots on
the right) visualization tasks (in ms)
7. Discussion and conclusion
The primary objective of the study was to investigate the
relationship between the participants’ cognitive style and their
performance at map-related tasks employing different forms of
visualization. First, a Compound Figures Test (a variation of
Navon’s Hierarchical Figures Test) was administered using
MuTeP in order to discriminate between analytically- and globallyoriented
individuals. The test was already been successfully
employed in previous studies (Kubíček et al., 2016, Horváth,
2012). The results of the test confirmed the effect of global
precedence (Poirel, Pineau and Mellet, 2008). Subsequently, two
independent correlation studies were conducted in order to
establish the relationship between the participants’ performance at
global and local subtests of the CFT and their performance at maprelated
tasks employing different forms of cartographic
visualization (intrinsic and extrinsic). Each correlation study used a
research sample with a different level of map literacy. The first
research sample with a high degree of map literacy consisted of
students of cartography, while subjects with low levels of map
literacy were represented by students of psychology. In contrast
with the first study, the second correlation study showed no
significant correlation between the participants’ performance at
CFT and their performance at map-related tasks. We suppose that
the absence of any correlation might have been caused by low
levels of map literacy and the resulting subjective difficulty of the
map-related tasks. In a post-experimental inquiry, the students of
psychology more often reported that they had difficulties
understanding the notion of “avalanche risk uncertainty”. The
above was reflected in comparatively high error rates displayed by
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 34
the students of psychology. We believe that comprehension
problems acted as an intervening variable and might have been a
source of significant “noise” in the collected data. The studies
focused on the investigation of low-level cognitive processes,
particularly the visual decoding of the legend and subsequent
visual search for target areas in the map. It is, however, important
to realize that a map represents a complex communication channel
and its thematic layer (if present) triggers high-level cognitive
processes which may override the underlying low-level cognitive
processes. Thus, the amount of time needed for solving the maprelated
tasks did not primarily depend on the type of visualization
used or the speed of visual processing, but rather on the degree and
speed of understanding the instructions. According to Booth
(2006), tasks involving low-level cognitive processes are less
prone to invoking between-task differences in the subjects’
performance than tasks involving high-level cognitive processes,
where local/global processing styles are overridden by
executive/strategic processes. Thus, if a medium-strong correlation
is found between a simple, selective attention task (such as CFT)
and a complex map-related task, it can be viewed as indicating that
identical cognitive processes are at play in both cases.
In the first correlation study, which worked with a group of
participants with high levels of map literacy, significant
correlations were established between the participants’ cognitive
style and the type of visualization used in map-related tasks. A
statistically significant positive correlation (r= .387; p<0.05) was
established between scores obtained in the local subtest and
response times in map-related tasks employing intrinsic
visualization. No relationship was found between intrinsic
visualization and scores obtained in the global subtest. Similarly,
while a positive correlation (r= .384; p<0.05) was found to exist
between global-subtest response times and response times in maprelated
tasks employing extrinsic visualization, no relationship was
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 35
found between extrinsic visualization and the local subtest. Thus, it
can be concluded that subjects with high levels of map literacy
exhibit differences in their performance at map-related tasks
depending on their cognitive style and on the type of visualization
employed. Analytically-oriented individuals were better at tasks
involving intrinsic visualization, while globally-oriented
individuals performed better at tasks employing extrinsic
visualization. The fact that no cross-correlation was found between
either the local subtest and extrinsic visualization or between the
global subtest and intrinsic visualization renders the findings even
more significant. Following the two correlation studies in the first
phase, a comparative eye-tracking study was conducted in the
second phase with the aim of obtaining deeper insight into the
cognitive processes which were at play during task-solving; it was
not designed to test pre-defined hypotheses or compare types of
visualization with respect to their effectiveness.
The results of the eye-tracking part of our research indicate a
high dependency of performance at intrinsic visualization tasks on
the nature of the tasks (i.e. on the values of target variables). An
example of the above is represented by the high across-task
variability in the number of map/legend transitions. At the same
time, intrinsic visualization tasks exhibit a far greater number of
transitions than extrinsic visualization tasks, requiring more
“checking look-backs” at the legend. Another analyzed parameter
was represented by the map/legend dwell time ratio. For extrinsic
visualization tasks, the legend dwell time was relatively short,
ranging from 10 to 30% (of the total dwell time), while visual
search time (map dwell time) ranged from 60 to 80%. Intrinsic
visualization, on the other hand, exhibited a far greater variability
across the individual tasks. For the legend, the dwell time ranges
from below 10% in one task to over 50% in another. Time spent on
visual search amounted to more than 70% of the total dwell time in
some tasks and less than 40% in others. In some cases, the time
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 36
needed for decoding the legend even exceeded the time needed to
locate the target in the map. In addition, intrinsic visualization
resulted in longer fixation durations and greater fixation duration
variability across the individual tasks. The results indicate that in
intrinsic tasks, decoding the legend may in some cases require
significant cognitive effort (see Longo and Barrett, 2010). An
instance of the above may be represented by the established acrosstask
variability in the difficulty of legend decoding in intrinsic
visualization tasks. The findings regarding intrinsic and extrinsic
visualization support the hypothesis that the different types of
visualization are not computationally equivalent and that each of
them requires a different cognitive strategy. It can be assumed,
further, that an individual’s performance will be the best if his/her
cognitive style and the preferred cognitive strategy are in
consonance with the type of task given (see Fig. 1). We believe,
based on the results of the first correlation study, that analyticallyoriented
individuals were better able to decode the legend (i.e.
discriminate between the employed variables), which resulted in
shorter task times.
The results of the first correlation study and of the comparative
eye-tracking study confirmed the hypothesis that the process of
solving map-related tasks is, to a large extent, dependent on the
type of visualization used in the task. The correlation study
established a difference between the participants in their maprelated
performance depending on their cognitive style, indicating
that different cognitive strategies might have been at play.
Globally-oriented individuals were better at extrinsic visualization
tasks, where two variables were represented by two different
modalities which needed to be processed simultaneously; intrinsic
visualization tasks, on the other hand, allowing for sequential
processing of two variables represented by a single modality
(color), were “easier” for locally- (analytically-)oriented
individuals. The eye-tracking study enabled to record the
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 37
participants’ eye movements during the experiments and to learn
about possible differences in task-solving in a more direct way.
We are currently working on a replicative study with a modified
research design so that the findings of the present study can be
fully confirmed. The new design will avoid the term “uncertainty”;
care will be taken to ensure that only concepts familiar to
individuals with low levels of map literacy are used. In addition, a
more in-depth analysis of the participants’ performance in maprelated
tasks will be carried out. A new testing system will be used,
enabling the interconnection of SW Hypothesis (Štěrba et al.,
2015) with SW Ogama (Voßkühler et al., 2008) and the eyetracking
system Eye Tribe or SMI RED250mobile (Popelka et al.,
2016): The system uses a presentation interface which allows for
extensive and intensive data collection to be conducted
simultaneously.
Acknowledgement
This research was supported by funding from the project of
Masaryk University under the grant agreement No.
MUNI/M/0846/2015, which is called ‘Effect of cartographic
visualization methods on the success rate of the practical and
learning spatial tasks’.
References
Alaçam, Ö. and Dalci, M. 2009. A usability study of WebMaps with eye tracking
tool: the effects of iconic representation of information. In: J. A. Jacko, ed.
Human-Computer interaction. New trends. Berlin, Heidelberg, New York:
Springer, pp.12-21.
Alexander, P. A. Kulikowich, J. M. and Schulze, S.K. 1994. The influence of
topic knowledge, domain knowledge, and interest on the comprehension of
scientific exposition. Learning and Individual Differences, 6(4), pp.379–397.
Bertin, J. 1973. Sémiologie graphique. Paris: La Haye.
Booth, R. D. L. 2006. Local-global processing and cognitive style in autism
spectrum disorders and typical development. [pdf] PhD. Institute of Psychiatry,
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 38
King's College London. Available at:
<https://kclpure.kcl.ac.uk/portal/files/2935437/433168.pdf>[Accessed 12 June
2016].
Bouvet, L. Rousseta, S. Valdoisa, S. and Donnadieu, S. 2011. Global
precedence effect in audition and vision: Evidence for similar cognitive styles
across modalities. Acta Psychologica, 138, pp.329–335.
Brand, J. and Johnson, A. P. 2014. Attention to local and global levels of
hierarchical Navon figures affects rapid scene categorization. Frontiers in
Psychology, [e-journal] 5, 1274. Available at: <doi:
10.3389/fpsyg.2014.01274>[Accessed 12 June 2016].
Buttenfield, B. 2000. Mapping ecological uncertainty. In: C. T. Hunsaker et al.,
eds. Spatial uncertainty in ecology. New York: Springer.
Çöltekin, A. Heil, B. Garlandi, S. and Fabrikant, S. I. 2009. Evaluating the
effectiveness of interactive map interface designs: a case study integrating
usability metrics with eye-movement analysis. Cartography and Geographic
Information Science, 36(1), pp.5-17.
Creswell, J. W. 2008. Research Design: Qualitative, Quantitative, and Mixed
Methods Approaches. Los Angeles, London, New Delhi, Singapore: Sage
Publications, Inc.
Dale G. and Arnell, K. M. 2014. Lost in the Forest, Stuck in the Trees:
Dispositional Global/Local Bias Is Resistant to Exposure to High and Low Spatial
Frequencies. PLoS ONE, [e-journal] 9(7): e98625. Available at:
<doi:10.1371/journal.pone.0098625>[Accessed 10 July 2016].
Duchaine, B. Yovel, G. and Nakayama, K. 2007. No global processing deficit in
the Navon task in 14 developmental prosopagnosics. Social Cognitive and
Affective Neuroscince, 2(2), pp.104–113.
D’Zmura, M. 1991. Color in visual search. Vision Research, 31, pp.951–966.
Elmer, M., E. 2013. Symbol Considerations for Bivariate Thematic Maps.
Proceedings of ICC Conference, [pdf] Dresden, 25-30 August 2013. Available
at: <
http://icaci.org/files/documents/ICC_proceedings/ICC2013/_extendedAbstract/
278_proceeding.pdf>[Accessed 10 April 2016].
Gershon, N. 1998. Visualization of an imperfect world. Computer Graphics and
Applications, 18, pp.43–45.
Graff, M. 2003. Learning from web-based instructional systems and cognitive
style. British Journal of Educational Technology, 34(4), pp.407–418.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 39
Hammond, K. R. 1996. Human judgment and social policy: Irreducible
uncertainty, inevitable error, unavoidable injustice. New York: Oxford University
Press.
Hofmann, A. Hošková-Mayerová, Š. Talhofer, V. and Kovařík, V. 2015. Creation
of models for calculation of coefficients of terrain passability. Quality and
Quantity, 49(4), pp.1679-1691.
Horváth, J. 2012. Funkce kognitivního stylu při čtení mapy v kontextu krizového
řízení. (Role of cognitive style with respect to map reading in context of crisis
management). Diploma Thesis. Brno: Masaryk University.
Hus, V. and Hojnik, T. 2013. Comparative Analysis of Cartographic Literacy in
the Selected Curricula at the Primary Level. Creative Education, 4(12), pp.757-
761.
Itti, L. and Koch, C. 2000. A saliency-based search mechanism for overt and
covert shifts of visual attention. Vision Research, 40, pp.1489–1506.
Jones, Ch. 1997. Geographical Information Systems and Computer
Cartography. London and New York: Routlege.
Kirton, M. J. 1989. Adaptors and innovators: styles of creativity and problemsolving.
London: Routledge.
Koláčný, A. 1977. Cartographic Information – A Fundamental Concept and Term
in Modern Cartography. Cartographica: The International Journal for
Geographic Information and Geovisualization, 1(14), pp.39-45.
Konečný, M. et al. 2011. Dynamická geovizualizace v krizovém managementu.
Brno: Masarykova univerzita.
Kozhevnikov, M. 2007. Cognitive styles in the context of modern psychology:
Toward an integrated framework. Psychological Bulletin, 133(3), pp.464-481.
Kubíček, P. Šašinka, Č. and Stachoň, Z. 2014. Vybrané kognitivní aspekty
vizualizace polohové nejistoty v geografických datech. Geografie - Sborník
České geografické společnosti, Česká geografická společnost, 119(1), pp.67-90.
Kubíček, P. et al. 2016. Cartographic Design and Usability of Visual Variables
for Linear Features. The Cartographic Journal, pp.1-11.
Kunz, M. and Hurni, L. 2011. How to enhance cartographic visualizations of
natural hazards assessment results. The Cartographic Journa, 48(1), pp.60-71.
Kunz, M. 2011. Interactive visualizations of natural hazards data and associated
uncertainties. PhD. ETH Curych.
Larkin, J. H. and Simon, A. H. 1987. Why a Diagram is (Sometimes) Worth Ten
Thousand Words. Cognitive science, 11, pp.65-99.
Leitner, L. and Buttenfield, B. 2000. Guidelines for the display of attribute
certainty. Cartography and Geographic Information Science, 27, pp.3–14.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 40
Leech, N. L. and Onwuegbuzie, A. J. 2009. A typology of mixed methods
research designs. Quality & Quantity, 43(2), pp.265-275.
Longo, L. and Barrett, S. 2010. A Computational Analysis of Cognitive Effort.
In: N. T. Nguyen, M. T. Le and J. Świątek, eds. Intelligent Information and
Database Systems. Volume 5991 of the series Lecture Notes in Computer
Science. New York: Springer, pp 65-74.
Maceachren, A. 1992. Visualizing uncertain information. Cartographic
Perspectives, 13, pp.10–19.
MacEachren, A. M. and Taylor, D. R. F. 1994. Visualization in Modern
Cartography. Oxford: Pergamon.
Maceachren, A. et al. 2005. Visualising geospatial information uncertainty.
What we know and what we need to know. Cartography and Geographic
Information Science, 32, pp.139–160.
MacEachren A. M. et al. 2012. Visual semiotics & uncertainty visualization: An
empirical study. Transactions on Visualization & Computer Graphics. 18(2)
pp.2496-2505.
Milne, E. and Szczerbinski, M. 2009. Global and local perceptual style, fieldindependence,
and central coherence: An attempt at concept
validation. Advances in Cognitive Psychology, 5, pp.1-26.
Montello, D. R. 2009. Cognitive Research in GIScience: Recent Achievements
and Future Prospects. Geography Compass, 3(5), pp.1824–1840.
Navon, D. 1977. Forest Before Trees: The Precedence of Global Features in
Visual Perception. Cognitive Psychology, 9(3), pp.353–383.
Nothdurft, H. C. 2000. Salience from feature contrast: Additivity across
dimensions. Vision Research, 40, pp.1183–1201.
Olson, J. M. 1979. Cognitive cartographic experimentation. Cartographica: The
International Journal for Geographic Information and Geovisualization, 16(1),
pp.34-44.
Ooms, K. et al. 2012. Investigating the effectiveness of an Efficient Label
Placement Method Using Eye Movement Data. The Cartographic Journal, 49 (3),
pp.234-246.
Pask. G. 1976. Styles and strategies of learning. British journal of educational
psychology, 46, pp.128-148.
Poirel, N. Pineau, A. and Mellet, E. 2008. What does the nature of the stimuli
tell us about the Global Precedence Effect?. Acta Psychologica, 127, pp.1–11.
Popelka, S. and Brychtová, A. 2013. Eye-tracking study on different perception
of 2D and 3D terrain visualisation. Cartographic Journal, 50(3), pp.240-246.
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 41
Popelka, S. and Dědková, P. 2014. Extinct Village 3D visualization and its
Evaluation with Eye-Movement Recording. In: B. Murgante et al., eds.
Computational Science and Its Applications - ICCSA 2014, Part 1. New York:
Springer, pp. 786-795.
Popelka, S. Stachoň, Z. Šašinka, Č. and Doležalová, J. 2016. Eyetribe Tracker
Data Accuracy Evaluation and Its Interconnection with Hypothesis Software for
Cartographic Purposes. Computational Intelligence and Neuroscience, [ejournal]
Available at: <doi:10.1155/2016/9172506>[Accessed 30 August
2016].
Rayner, S. G. 2000. Reconstructing style differences in thinking and learning:
Profiling learning performance. In: R. J. Riding and S. G. Rayner, eds.
International Perspectives on Individual Differences. Volume 1 Cognitive Styles,
Stamford: Ablex, pp.115-177.
Rezaei, A. R. and Katz, L. 2004. Evaluation of the Reliability and Validity of the
Cognitive Styles Analysis Inventory and Recommendations for Improvement.
Personality and Individual Differences, 36(6), pp.1317-1327.
Roth, R. E. et al. 2011. Card sorting for cartographic research and practice.
Cartography and Geographic Information Science, 38(2), pp.89-99.
Smith Mason, J. et al. 2016. Special issue introduction: Approaching spatial
uncertainty visualization to support reasoning and decision making. Spatial
Cognition & Computation, 16(2), pp.97-105.
Staněk, K. Friedmannová, L. Kubíček, P. and Konečný, M. 2010. Selected issues
of cartographic communication optimization for emergency centers.
International Journal of Digital Earth, 3(4), pp.316-339.
Spirduso, W. W. 1980. Physical Fitness, Aging, and Psychomotor Speed: A
Review. Journal of Gerontology, 35(6), pp.850-865.
Štěrba, Z. et al. 2014. Mixed Research Design in Cartography: A Combination
of Qualitative and Quantitative Approaches. Kartographische Nachrichten,
64(5), pp.262-269.
Štěrba, Z. et al. 2015. Selected Issues of Experimental Testing in Cartography.
Brno: Masaryk University.
Trau, J. and Hurni, L. 2007. Possibilities of incorporating and visualizing
uncertainty in natural hazard prediction. In: Proceedings of the 23rd
International Cartographic Conference ICC. [pdf] Moscow, Russia, 4-10 August
2007, Available at: <
http://www.mountaincartography.org/publications/papers/ica_cmc_sessions/5_
Moscow_Session_Mountain_Carto/moscow_hurni_trau.pdf>[Accessed 17 May
2016].
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 42
Voßkühler, A. Nordmeier, V. Kuchinke, L. and Jacobs, A.M. 2008. OGAMA OpenGazeAndMouseAnalyzer:
Open source software designed to analyze eye
and mouse movements in slideshow study designs. Behavior Research
Methods, 40(4), pp.1150-1162.
Witkin, H. A. Goodenough, D. R. and Karp, S.A. 1967. Stability of cognitive
style from childhood to young adulthood. Journal of Personality and Social
Psychology, 7(3), pp.291-300.
Wehrend, S. and Lewis, C. 1990. A Problem-oriented Classification of
Visualization Techniques. In: A. Kaufman, ed. Proceedings of the 1st conference
on Visualization '90. San Francisco, 23-26 October1990, IEEE Computer Society
Press, Los Alamitos, pp.139-143.
Yovel, I. Revelle, W. and Mineka, S. 2005. Who Sees Trees Before Forest? The
Obsessive-Compulsive Style of Visual Attention. Psychological Science, 16(2),
pp.123-129.
Zhang, J. and Goodchild, M. 2002. Uncertainty in geographical information.
London: Taylor and Francis.
List of figures
Figure 1. Triarchic structural model of performance when solving
map-related task (p.
Figure 2. Snow avalanche hazard and hazard uncertainty map: left
– extrinsic visual representation; right – intrinsic visual
representation (adapted from Kunz, 2011)
Figure 3. Differences in the encoding of variables between the
extrinsic (first column) and intrinsic (second column) form of
visualization; third column shows the difference in salience
between a fully saturated yellow (1) and partially desaturated blue
(6) on white and black backgrounds
Figure 4. Structure of used combined extensive-intensive research
design
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 43
Figure 5. Structure of map-related tasks (students of cartography –
top; students of psychology – bottom)
Figure 6. Sample map-related tasks: Left (intrinsic visualization) –
Select the area with a moderate level of avalanche hazard. Right
(extrinsic visualization) – Select the area with a high level of
avalanche hazard accompanied by a low level of avalanche hazard
uncertainty.
Figure 7. A compound stimulus (left); an item from the Compound
Figures Test adapted into the MuTeP environment (right).
Figure 8. Reaction times (s) for both subtests of CFT
Figure 9. Process of investigating variables: 1) stimulus -> output;
2) stimulus -> cognitive processing and associate behavior ->
output
Figure 10. Structure of map-related tasks in the eye-tracking study
Figure 11. Trial duration (in ms; left) and number of transitions
(right) between the map and the legend
Figure 12. Number of transitions between the legend and the map
Figure 13. Dwell times related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in %)
Figure 14. Fixation durations related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in ms)
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 44
Figure 15. Fixation durations related to the map and the legend in
extrinsic (two boxplots on the left) and intrinsic (two boxplots on
the right) visualization tasks (in ms)
The Impact of Global/Local Bias on Task-solving in
Map-related Tasks Employing Extrinsic and Intrinsic
Visualization of Risk Uncertainty Maps
The form of visual representation affects both the way in
which the visual representation is processed and the
effectiveness of this processing. Different forms of visual
representation may require the employment of different
cognitive strategies in order to solve a particular task; at the
same time, the different representations vary as to the extent to
which they correspond with an individual’s preferred
cognitive style. The present study employed a Navon-type task
to learn about the occurrence of global/local bias. The research
was based on close interdisciplinary cooperation between the
domains of both psychology and cartography. Several
different types of tasks were made involving avalanche hazard
maps with intrinsic/extrinsic visual representations, each of
them employing different types of graphic variables
representing the level of avalanche hazard and avalanche
hazard uncertainty. The research sample consisted of two
groups of participants, each of which was provided with a
different form of visual representation of identical
geographical data, such that the representations could be
regarded as “informationally equivalent”. The first phase of
the research consisted of two correlation studies, the first
involving subjects with a high degree of map literacy (students
of cartography) (intrinsic method: N = 35; extrinsic method: N
= 37). The second study was performed after the results of the
first study were analyzed. The second group of participants
consisted of subjects with a low expected degree of map
literacy (students of psychology; intrinsic method: N = 35;
extrinsic method: N = 27).The first study revealed a
statistically significant moderate correlation between the
students’ response times in extrinsic visualization tasks and
Manuscript - anonymous
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 2
their response times in a global subtest (r=0.384, p<0.05);
likewise, a statistically significant moderate correlation was
found between the students’ response times in intrinsic
visualization tasks and their response times in the local subtest
(r=0.387, p<0.05). At the same time, no correlation was found
between the students’ performance in the local subtest and
their performance in extrinsic visualization tasks, or between
their scores in the global subtest and their performance in
intrinsic visualization tasks. The second correlation study did
not confirm the results of the first correlation study (intrinsic
visualization/“small figures test”: r = 0.221; extrinsic
visualization/“large figures test”: r = 0.135). The first phase of
the research, where the data was subjected to statistical
analysis, was followed by a comparative eye-tracking study,
whose aim was to provide a more detailed insight into the
cognitive strategies employed when solving map-related tasks.
More specifically, the eye-tracking study was expected to be
able to detect possible differences between the cognitive
patterns employed when solving extrinsic- as opposed to
intrinsic-visualization tasks. The results of an exploratory eyetracking
data analysis support the hypothesis of different
strategies of visual information processing being used in
reaction to different types of visualization.
Keywords: Geovisualization method, avalanche risk,
cognitive style, Navon’s hierarchical figure, combined
extensive-intensive research design, eye-tracking.
1. Introduction
Space and time play a crucial role during hazardous events in
general and natural hazards in particular. Successful decision
making during emergency situations depends on the availability
and relevancy of information presented in the right time and an
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 3
understandable way. Such information improves the transparency
and credibility of the decisions taken.
Only “raw” spatial data is currently available for emergency
management support. Emergency decision makers all around the
world have available maps with natural hazards (such as flood,
avalanche, and landslides), vulnerable zones, land use or geology,
and make decisions based on implicit information inferred from
such map sources. Such implications are not straightforward and
may even differ according to the different professional background
of a particular decision maker. Or more generally, the form of
information visualization should be adjusted to the cognitive
characteristics of the users. The responsible persons are able to
process only a limited number of graphics (maps) in case of an
emergency. This situation is even more critical when working
under severe time pressure. The visualization form can
significantly influence the final decision. Zhang and Goodchild
(2002) proved that visual form can improve the communication
about spatial data uncertainty within spatial analysis and spatial
decision support. Uncertainty often possesses spatial patterns.
Uncertainty visualization can thus reveal such patterns and serve
not only for presentation but also for exploration and visual
analysis of spatial data.
The present study is a continuation of an earlier research project on
crisis management (Konečný et al., 2011; Staněk et al. 2010). The
study focuses on perception and cognitive processing of different
forms of bivariate visual representation, using alternative visual
representations derived from identical avalanche hazard datasets
(Kunz, 2011; Kunz and Hurni, 2011). The employment of different
cartographic bivariate visualizations of the same area and topic
made it possible to create meaningful and complex stimuli that
were visually different and at the same time fulfilled the
requirement of informational equivalence (Larkin and Simon,
1987).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 4
A map functions as a communication channel (Koláčný, 1977),
and when used as a stimulus material in psychological
experiments, it enables the researcher to purposely manipulate the
form of the communicated information. The objective of the study
was to investigate the impact of different types of visualization and
different cognitive styles on the effectiveness of solving maprelated
tasks.
2. Uncertainty visualization and the use of bivariate visualization
MacEachren (1992) suggested the use of Bertin’s graphic variables
to depict uncertainty and even added specialized variables for
depicting uncertainty including crispness, resolution, and
transparency. Gershon (1998) grouped these into intrinsic and
extrinsic visual variables depending on whether the variable is
visually separable from the variable depicting the actual attribute.
While extrinsic variables are separable, intrinsic variables are not.
Another logical step is to describe how these variables including
possible additions or modification, might be logically matched
with different components of data uncertainty (Buttenfield 1991,
MacEachren 1992, Leitner and Buttenfield 2000). MacEachren
(1992), for instance, stated that the graphical variables size and
color value are most appropriate for depicting uncertainty in
numerical information, while color hue, shape, and perhaps
orientation can be used for uncertainty in nominal information.
More recently MacEachren et al (2012) focused on discrete
symbols that could be used to signify the uncertainty of individual
items within information graphics, maps. The experiments
examine relative effectiveness of a set of uncertainty representation
solutions—differing in the visual variable leveraged and level of
symbol iconicity. Trau and Hurni (2007) and Kunz (2011)
theoretically analyzed the suitability of visual variables and
visualization techniques for uncertainty depictions in hazard
prediction maps.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 5
Most applied uncertainty visualizations in the field of natural
hazards are simplistic univariate representations where hazard
related data are displayed in one map and inherent uncertainties are
depicted in a second map display (Kunz 2011). Kunz was among
the first to propose the use of bivariate depiction of studied
phenomenon and its uncertainty, applied this approach in the
dynamic environment and even presented brief feedback from
expert users.
Capabilities and limitations of bivariate map types based on the
combination of visual variables and symbol dimensionalities
(point, line, and polygon) were studied and tested also by Elmer
(2013). Using the selective attention theory, he empirically tested
eight bivariate map types for map reading tasks recording their
accuracy and response time. These tests also included the
combination of size/value (Choropleth/Graduated symbols) and
value/hue-saturation (Bivariate Choropleth) which are relevant for
our study. Both aforementioned combinations performed above
average for accuracy and response time and were also rated
positively by users as an appropriate combination to read and
understand the information on the map. However, the author
himself concluded that the study only revealed significant
differences in perceived combinations and further research is
needed in order to understand different mental strategies of users
and identify their cognitive behavior.
Cognitive cartography encompasses the application of cognitive
theories and methods to understanding maps and mapping and the
application of maps to understanding cognition (Montello 2002).
Different ways of displaying the same spatial information can
dramatically affect problem-solving performance. Spatial cognition
research uses distinct concepts of informational and computational
equivalence of representations (Simon 1978). Alternative
cartographic visualizations follow the premise of information
equivalency. The possibility to visualize (code) the same spatial
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 6
information in an alternative way which is informationally
equivalent offers valuable input material for comparison of
cognitive processes used for decoding. Such a comparative
principle enables better understanding of the human cognitive
apparatus. Differences in representations deal not only with
various visual forms but also with different operations necessary
for their decoding and interpretation.
Cognitive assumptions are closely connected with visual variables
used for uncertainty visualization. Both methods differ in the type
of visualization method and visual variables (intrinsic and
extrinsic). The perception of variables has a close connection with
the theory of pre-attentive perception (Treisman and Gelade 1980,
Wolfe et al 1989). The extrinsic graphical variable size is generally
considered to be a pre-attentive feature. Such features are
appropriate for the determination of presence or absence or
particular elements or boundary detection. On the other hand
intrinsic graphical variable saturation has not been confirmed as
pre-attentive.
However, the situation is rather different when using a
combination of both uncertainty portraying variables with a main
attribute variable. The intrinsic visualization method combines
color hue for the main attribute value and saturation for its
uncertainty. The resulting map legend is comprised of 9 categories
and constitutes a higher potential cognitive load for users. The
extrinsic visualization method combines color saturation for the
main attribute value and proportional circle size for uncertainty.
The map legend is comprised of only 6 categories.
3. Cognitive Style in Map-Related Tasks
Kozhevnikov (2007) defines cognitive style as heuristics (i.e.
strategy derived from experience with similar problems used for
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 7
processing external information). An individual’s cognitive style
can be detected at all levels of perception, from the elementary and
highly automated ones to those that are complex and conscious.
The concept is used to refer to the way individuals think, perceive
and orient themselves in the environment. According to Brigham et
al. (2007), cognitive style is a pervasive bipolar dimension that is
stable over time and can be studied using psychometric techniques.
He further states that a cognitive style may be value-differentiated,
meaning that it describes differences concerning value rather than
quality. However, cognitive style is far from a well-defined
construct, both with respect to its content and application to
different levels of the personality system. A vast range of different
interpretations of cognitive styles exist (see Witkin, 1967; Rayner,
2000; Kirton, 1989; Pask, 1976). Riding and Cheema (1991)
surveyed more than 30 conceptions of cognitive style, concluding
that each of the investigated conceptions pertains to one of two
principal dimensions. The verbal-imagery dimension encompasses
an individual’s preference for representing information in
words/verbal associations, or in mental pictures. The wholist–
analytic dimension is characterized as an individual’s preference
for processing information either in integrated wholes or in
discrete parts. Given the nature of the tasks used in the study and
the differences between intrinsic and extrinsic visual
representations (see Fig. 2), it was the wholist–analytic dimension
that was of greater importance for the present study. According to
Graff (2003), wholist–analytic cognitive style can be defined as a
tendency to process information either as an integrated whole or in
discrete parts of that whole. The wholist-analytic cognitive
dimension is based on the conception of global/local bias (see Dale
and Arnell, 2014) related to whether visual information is
perceived at a broad (global) level, or at a more focused (local)
level, with more attention being paid to partial characteristics of
objects and phenomena and to their analytical processing. Rezaei
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 8
and Katz (2004) add that globally-oriented individuals consider
phenomena in a broader perspective and context. Analyticallyoriented
individuals, on the other hand, view each situation as an
aggregate of discrete elements, typically preferring to focus on one
or two elements at a time at the expense of other elements/aspects.
Graff (2003) further states that analytically-oriented individuals are
better at apprehending concepts in parts, but may experience
difficulty integrating such concepts into complete, consistent
wholes, while globally-oriented individuals view concepts as
wholes, but are unable to separate individual aspects of the
concepts into discrete parts. According to Kozhevnikov (2007), the
analytical cognitive style tends to be characterized as convergent,
differentiated, sequential, reflective and deductive, whereas the
“global” style has been described as divergent, intuitive, impulsive,
inductive, and creative. Globality is often discussed in connection
with e.g. attentional breadth in selective attention (Dale and Arnell,
2014) and rapid scene categorization (Brand and Johnson, 2014).
The level of an individual’s performance in map-related tasks is
a result of the interaction of three variables: a) user characteristics;
b) task type and situational context; c) map-related
characteristics/type of visual representation (Fig. 1). Wehrend and
Lewis (1990) constructed a comprehensive catalogue of maprelated
operations, including identification (identifying visual
characteristics of features on the map), localization (determining
the absolute or relative position) or categorization (placing in
specifically defined divisions in a classification; this may be done
by color, shape or size). Map-related tasks can be ranked
depending on their relative difficulty, from relatively easy
operations of “finding the shortest path” (e.g. Jones, 1997) to
highly complex tasks (Smith Mason et al., 2016) of “planning
military operations” (Hofmann et al., 2015), which require several
concepts to be explored simultaneously, compared and integrated.
Working with a map always needs to be viewed as mental
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 9
manipulation with semantically rich material rather than a mere
visual search and processing of visual stimuli (MacEachren and
Taylor, 1994, Montello, 2009, Roth et al., 2011). The highest-level
performance in fulfilling the task can be expected if the cognitive
style of the user matches the nature of the task (Hammond, 1996)
and the form of visual representation used. For instance, at a task
requiring analytical thinking, an analytically-oriented individual
can be expected to perform better than a globally-oriented
individual with the same degree of cartographic literacy (Hojnik
and Hus, 2013) and domain knowledge (Alexander, Kulikowich
and Schulze,1994). Additionally, a form of visualization allowing
for sequential analytical processing will result in a better
performance than a visualization requiring simultaneous and
intuitive processing. The user’s performance will be the best if all
three of the areas (i.e. type of task, type of cognitive style and form
of visualization) are in consonance.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 10
Figure 1. Triarchic structural model of performance when solving
map-related task
From the point of view of experimental psychology, maps are
highly valuable as stimulus material (Olson, 1979) in that they
constitute complex external representations with variables
amenable to accurate control and change of value. By representing
identical content (data), different forms of visualization are
informationally equivalent, enabling computational equivalence to
be studied (Larkin and Simon, 1987). Any differences in
performance and in the way an individual works with different
maps can be viewed as directly linked to the difference in visual
presentation of the same content. In the present study, avalanche
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 11
hazard maps were used with intrinsic vs. extrinsic visual
representations (Fig. 2, right and left, respectively).
Figure 2. Snow avalanche hazard and hazard uncertainty map: left
– extrinsic visual representation; right – intrinsic visual
representation (adapted from Kunz, 2011)
The difference between the extrinsic and intrinsic representation
exists in the way the two thematic layer variables (snow avalanche
hazard and snow avalanche hazard uncertainty) are represented. In
extrinsic representation, the avalanche hazard level is expressed in
terms of variation in the saturation of the color blue, while hazard
uncertainty is represented by the size of the dots. In intrinsic
representation, the same map base is used, with the avalanche
hazard expressed in terms of hue, and hazard uncertainty expressed
in terms of variation in the saturation of the hue. Both types of
visual representations use a combination of two graphic variables.
However, while the extrinsic type uses two different modalities
(hue and size) for the presentation of two phenomena, intrinsic
representation is expressed in terms of variation of two properties
(hue and saturation) of a single modality (color). There is a clear
difference in categorization between the two types of
representations. In the intrinsic visual representation, 9 (3 x 3)
categories are encoded explicitly; the extrinsic visual
representation, on the other hand, has separate categories for each
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 12
of the two phenomena it represents (3 for avalanche hazard level
and 3 for hazard uncertainty), with only 6 categories being
encoded explicitly (although there are a total of nine combinations
as well). Even though color properties (hue and saturation in this
particular case) tend to be regarded as two independent variables in
the field of cartography (Bertin, 1973), in the psychology of
perception color properties are viewed as interacting with each
other (D’Zmura, 1991, Itti and Koch, 2000; Lindsey et al, 2010).
For instance, a desaturated blue on light (white) background will
be less luminous and thus more salient than a fully saturated
yellow (e.g. Nothdurft, 2000; see Fig. 3). The phenomenon can be
applied to map reading as well.
Figure 3. Differences in the encoding of variables between the
extrinsic (first column) and intrinsic (second column) form of
visualization; third column shows the difference in salience
between a fully saturated yellow (1) and partially desaturated blue
(6) on white and black backgrounds
The objective of the presented study was to investigate the
relationship between different forms of bivariate visual
representation and an individual’s cognitive style as reflected by
their performance in Navon’s test of hierarchical figures (see
section 4.2.1). We hypothesize (1) a link between global
processing efficiency and extrinsic visualization abilities, and (2) a
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 13
link between local processing efficiency and intrinsic visualization
abilities. We thus consider that a variation of both hue and
saturation using a single color modality (intrinsic) would refer to a
more local approach, whereas two types of independent graphic
hues and size variables (extrinsic) will be considered a more global
approach. In other words, the elaboration of two parameters (hue
and saturation) of one modality - color (intrinsic) would be linked
to local processing abilities, while considering two different
parameters, hue and size (extrinsic), simultaneously would be
linked to global processing abilities.
4. Method
The study uses a mixed (confirmatory-exploratory) research
design, with the aim to combine extensive research (for data
collection and subsequent statistical analyses) with a more in-depth
exploratory data analysis (EDA; see Andrienko and Andrienko,
2005). The advantages of the mixed research design, along with its
theoretical grounding, have been described by Šterba et al. (2014),
who also provide a sample study. The authors use the term “mixedresearch
design”, which, however, tends to be viewed as referring
to a combination of qualitative and quantitative methods,
particularly in the context of social sciences and constructivist
approaches (see Leech and Onwuegbuzie, 2009; Creswell, 2003).
In order to distinguish between the two concepts we propose the
term “combined extensive-intensive research design”. A combined
mixed extensive-intensive research works primarily with objective
data, combining a confirmatory stage of the research with an
exploratory stage. The first (confirmatory) phase of our research
comprised two correlation studies. The first consisted of the
collection of data on subjects with a high degree of map literacy
(students of cartography), while the second focused on subjects
with a low degree of map literacy (students of psychology). After
the confirmatory phase, an exploratory phase followed which was
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 14
represented by a comparative eye-tracking study. The aim of the
comparative study was to reveal possible differences in tasksolving
strategies depending on the type of visualization employed
by usage of an exploratory data analysis. The structure of the
research is shown in detail in Fig.4. While extensive methods are
concerned only with the effect of stimulus on behavioral outputs in
the sense of speed or correctness intensive methods concentrate on
the process alone, as in, what happens between stimulus and
reaction? And this is also the purpose of the eye-tracking
comparative study which can deeper illuminate results of the
previous phase of the study.
Figure 4. Structure of used combined extensive-intensive research
design
5. Correlation Studies I and II
The objective of the first phase of the research was to investigate
the relationship between an individual’s cognitive style and the
type of visualization used. By way of the first phase, two
correlation studies were conducted. The design of the second study
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 15
was adjusted based on the results of the first study. More
specifically, several items were added to the subtest involving a
one variable. The second study also included tasks focused on the
degree of understanding of the concept of visualization of snow
avalanche hazard and hazard uncertainty. These more complex
tasks are not part of the present article, which focuses primarily on
lower cognitive processes (e.g. visual search as reflecting an
individual’s cognitive style).
5.1. Participants
In the first correlation study, the research sample consisted of
students of geography in the 1st to 3rd year of their studies. A total
of 73 volunteers aged 19 to 27 years were tested. For the extrinsic
visualization task, there were 37 subjects (19 male, 18 female),
while 35 subjects (19 male, 16 female) were given a task
employing intrinsic visualization. In the second correlation study, a
total of 62 volunteers aged 19 to 55 years were tested, all students
of psychology in the 1st to 3rd year of their studies. For the
intrinsic visualization task, there were 35 subjects (8 male, 27
female), most of them aged 19 to 25 years, with one male outlier
aged 37 years; 27 subjects (3 male, 24 female) were given the
extrinsic visualization task. Most of the subjects were 19 to 28
years, with one male outlier aged 55 years. None of the subjects
had any previous experience of participation in cartographic
visualization testing.
5.2. Aparatus
The test was created and administered in MuTeP (Multivariate
Testing Program; see Kubíček et al., 2014, Kubíček et al., 2016,
Štěrba et al., 2015). After every data collection a group discussion
followed. Participants were asked about their subjective experience
by elaboration of the test battery.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 16
5.3. Stimulus Material - Map-related Tasks
In the correlation studies, avalanche maps were used showing both
the level of the snow avalanche hazard (expressed in terms of load
on the snowpack) and the level of avalanche hazard uncertainty.
Both variables were expressed by means of a three-level scale
(Low – Moderate – High). The map consisted of a base layer and a
thematic layer, with the latter being explained by the legend of the
map. In the experimental tasks, the base layer played only a very
marginal role. The subjects’ performance depended on their ability
to comprehend the instructions, decode the legend, and perform a
visual search to locate the four numbered target areas and decide
which of them meets the criteria given. There was always only one
correct answer. The visual scene consisted of a map, a map legend,
task instructions located at the top of the computer screen and
numbered “buttons”. The instructions were always presented
separately on a blank screen. No time limit for the instructions was
given so that there would be enough time for the participants to
comprehend the instructions. The overall test structure was as
follows:
1. Training – 2 training tasks
a. identify the level of variable A (avalanche hazard)
b. identify the level of variable B (avalanche hazard
uncertainty)
2. One variable tasks
a. identify the level of variable A (avalanche hazard) –
low, moderate, and high respectively.
b. identify the level of variable B (avalanche hazard
uncertainty) – low, moderate, and high respectively.
3. Two variable tasks
a. identify the level of variable A (avalanche hazard)
and variable B (avalanche hazard uncertainty) –
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 17
combination of low, moderate, and high
respectively).
Both tests were made up of one- and two-variable tasks. In the
first correlation study, the participants were given five test tasks
(see Fig. 5 top line) – two for one variable and three for two
variables. In the second correlation study (performed on subjects
with a low degree of map literacy) the batch of one-variable tasks
included four items instead of two (see Fig. 5 bottom line).
Figure 5. Structure of map-related tasks (students of cartography –
top; students of psychology – bottom)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 18
Visualizations used in training were similar to the rest of the test.
Sample instructions of one variable task are as follows: “Select the
area with a moderate level of avalanche hazard” (see Fig. 6 left).
The subjects completed each task by clicking on the respective
“button”. Instructions of two variables task were formulated in a
following way: “Select the area with a high level of avalanche
hazard accompanied by a low level of avalanche hazard
uncertainty” (see Fig. 6 right).
Figure 6. Sample map-related tasks: Left (intrinsic visualization) –
Select the area with a moderate level of avalanche hazard. Right
(extrinsic visualization) – Select the area with a high level of
avalanche hazard accompanied by a low level of avalanche hazard
uncertainty.
5.4. Stimulus Material - Compound Figures Test
The second part of the test battery consisted of a psychological test
of cognitive style. It contained a variation of Navon’s Hierarchical
Figures Test (Navon, 1977) – a compound figures test (see Fig. 7,
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 19
left) which was designed for the purposes of the present study
using MuTeP. Navon’s test (Navon, 1977) is one of the most
frequently used methods for the measurement of the wholistanalytic
dimension of cognitive processing (e.g. Brand and
Johnson, 2014; Duchaine et al., 2007; Yovel et al., 2005, Milne
and Szczerbinski, 2009). It enabled us to measure a person’s ability
to direct attention either to the local level of the visual stimulus
material or to the global level. Bouvet et al. (2011) used tasks
based on the Hierarchical Figure test and found evidence that
participants exhibit a similar processing style across modalities
with respect to both vision and audition modalities. Thus these
findings support the theory that Navons Hierarchical Figures can
be understood and used more generally as a measure of wholisticanalytic
cognitive style and not only in the narrower way, as an
indicator of a visuo-attentional global/local style. The compound
figures test included instructions, several training items and 32 test
items. The stimulus material consisted of a set of large, single-digit
numbers compounded of small numbers. There were two subtests,
local and global, each comprising 16 items. The participants were
instructed to identify the small numbers in the first subtest and the
large numbers in the second subtest by clicking on one of four
available buttons as quickly as possible (see Fig. 7, left). The
overall response time of the respondents was recorded
electronically; the average response time per item was calculated.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 20
Figure 7. A compound stimulus (left); an item from the Compound
Figures Test adapted into the MuTeP environment (right).
5.5. Results
Two subjects with a high error rate for CFT were excluded from
further analyses. One student of cartography erred 5 times in the
local subtest and 6 times in the global subtest. One student of
psychology erred 12 times in the first subtest. The number of
mistakes made by the other participants was insignificant; most of
them achieved 100% accuracy, others erred less than three times.
Reaction times of the outlying errors were included in the analyses.
The overall score for the local and global subtests were counted
based on 15 items from each subtest. The first item of each subtest
was excluded from the analysis. In line with previous findings, the
internal consistency (Cronbach’s alpha) for both subtests was
found to be high (α = 0.805 and α = 0.864, respectively).
Differences in reaction times between the two subtests (Fig. 8)
were found to be in agreement with previous findings as well. A
paired t-test revealed faster processing of global figures by 15%
(N=133; mean reaction time for the local subtest s=1.9; SD=0.2;
mean reaction time for the global subtest s=1.6; SD=0.17; df=132;
p=0.01). The results showed a positive correlation between the two
subtests (r=0.437, p<0.01).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 21
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 22
Figure 8. Reaction times (s) for both subtests of CFT
The error rate of students of cartography was 7.2% for intrinsic
visualization and 9.7% for extrinsic visualization. Students of
psychology exhibited a significantly higher error rate in both
subtests: 13.8% for intrinsic visualization and 20.0% for extrinsic
visualization. The overall error rate was counted from the
individual error rates of all the subjects who completed the
cartographic part of the research, including outliers, which were,
however, excluded from further analyses. Outlying subjects in the
research sample of students of cartography were those who erred
two and more times (out of the total of five items (i.e. those with
an error rate >40%; intrinsic visualization – 1 participant; extrinsic
visualization – 4 participants). As regards the research sample
consisting of students of psychology, 4 subjects were excluded
(extrinsic visualization) who erred three and more times (out of 7;
error rate >40%). Although the above outliers were excluded from
further analyses, all reaction times of remaining participants were
(as in CFT) included (see Tab. 1 and 2).
Students of
Cartography
Intrinsic
visualization (n=34)
Extrinsic visualization
(n=33 )
local
subtest
global
subtest
local
subtest
global
subtest
One Variable .400**
.180 -.021 .375*
Two variables .168 -.114 .304*
.302*
Overall Score .387* .050 .176 .384*
** Correlation is significant at the 0.01 level; * Correlation is significant at the 0.05 level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 23
Table 1. Students of Cartography: Relationship between Maprelated
Task Performance (response time) and CFT Results
(Pearson’s Correlation; one-tailed)
Students of
Psychology
Intrinsic
visualization (n=35)
Extrinsic visualization
(n=23)
local
subtest
global
subtest
local
subtest
global
subtest
One Variable .211 .222 .100 .291
Two Variables .106 .230 -.079 -.140
Overall score .221 .266 .027 .135
** Correlation is significant at the 0.01 level; * Correlation is significant at the 0.05 level
Table 2. Students of Psychology: Relationship between Maprelated
Task Performance (response time) and CFT Results
(Pearson’s Correlation; one-tailed)
5.6. Interpretation of the Results
The results of the CFT employing a variation of Navon’s
hierarchical figures test confirmed the effect of global precedence;
the average response time in the global subtest was lower by 15%
compared with the local subtest. In addition, the results showed a
positive correlation between the two subtests, which is in line with
the expectation that an individual’s processing capacity in both
tasks will be driven by the same psychomotor speed (Spirduso,
1980). After the exclusion of subjects with high error rates (either
in CFT or in map-related tasks) the subjects’ results in CFT and
map-related tasks were tested for possible correlations. The
analysis was conducted by means of comparison of overall
response times in map-related tasks with response times in CFT.
The results of the first correlation study (on subjects with a high
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 24
degree of map literacy) revealed a positive correlation between
scores obtained in the local subtest and response times in maprelated
tasks employing intrinsic visualization; at the same time, a
positive correlation was found to exist between global-subtest
response times and response times in map-related tasks employing
extrinsic visualization. The fact that no cross-correlation was found
between either the local subtest and extrinsic visualization or
between the global subtest and intrinsic visualization (although the
subjects’ performance in map-related tasks was also driven by the
same psychomotor speed) renders the above findings even more
significant. In the second correlation study (on students of
psychology), no significant relationship between the variables was
found. We suppose that the absence of any correlation might have
been caused by low levels of map literacy, which was reflected in
comparatively high error rates.
6. Comparative Study: Eye-tracking
The reason for replicating the correlation studies using an eyetracking
system was to obtain additional information which would
provide deeper insight into the process of solving map-related
tasks and related cognitive strategies. With the decrease in cost of
the technology, eye-tracking became a more widely utilized
method, not one only used in cartographic studies (see Alaçam and
Dalci, 2009; Çöltekin et al., 2009; Popelka and Brychtová, 2013;
Popelka and Dědková, 2014; Ooms et al., 2012). The aim of the
eye-tracking study was to investigate the impact of different types
of cartographic visualization on task-processing. While the
confirmatory phase of the research focused on the speed and
“correctness” of reaction to visual stimuli, the exploratory phase
consisted of investigating the very process of task-solving (see Fig.
9).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 25
Figure 9. Process of investigating variables: 1) stimulus -> output;
2) stimulus -> cognitive processing and associate behavior ->
output
6.1. Participants
A total of 14 subjects aged 21 to 35 years volunteered to
participate in the study; they were either university students or had
already completed their university studies. None of the participants
received any education in cartography or related fields. The
participants were divided into two groups of seven.
6.2. Apparatus
The device used in the study was EyeLink 1000, a remote eyetracking
system manufactured by SR Research. The original test
battery presented in MuTeP was adapted to the environment of
Experiment Builder, a software tool by SR Research for creating
eye-tracking experiments.
Eye-tracking data were saved in a format enabling an exploratory
data analysis to be carried out in EyeLink Data Viewer, software
that can be used for viewing, filtering and processing eye-tracking
data recorded with an EyeLink eye-tracker. The application offers
a range of eye-tracking metrics. In addition, it enables the analysis
of transitions between defined areas of interest (AOIs). In the
analysis, visualization tools such as an attention map and video
analysis were also used. After every single data collection an
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 26
individual inquiry followed. Participants were asked about their
subjective experience by elaboration of a test battery,
6.3. Stimulus Material
The original test structure was slightly modified (see fig. 10) and
several new training tasks were added to the original test battery.
A detailed graphical explanation of the principles of avalanche risk
and uncertainty was added in order to properly introduce both the
risk mapping and its uncertainty. These should enhance the
understanding of the experiment by non-experts.
Figure 10. Structure of map-related tasks in the eye-tracking study
6.4. Results and Interpretation
The exploratory data analysis was performed at both the subject
level and group level (extrinsic vs. intrinsic visualization). One
participant from the extrinsic visualization group was excluded
from the analysis due to a heavy data dropout. Several areas of
interest were defined, with key AOIs being set around the legend
and the map. The following measures were analyzed: trial
duration, number of transitions between the legend and the map,
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 27
the number of fixations on the legend and on the map, dwell time1
on the legend andon the map, and fixation durations.
Fig. 11 (graph on the left) shows average trial durations for the
eleven tasks used in the study. The first seven tasks employ only
one variable (level of avalanche hazard or avalanche hazard
uncertainty), while the rest of the tasks inquire about both
variables. The area represented by the map was identical in all the
tasks. The graph on the right in Fig. 11 shows the number of
transitions between the map and the legend for each task (in both
directions). Fig. 12 (box plots) shows the average number of
transitions and their spread for both extrinsic and intrinsic
visualization tasks. As is evident from the graphs, the subjects’
performance across the individual tasks was much more
homogeneous when working with extrinsic visualization; the
subjects’ performance at intrinsic visualization tasks, on the other
hand, appears to have been more sensitive to the nature of the task.
The higher variability might have been caused by greater
differences in the difficulty of discriminating between different
levels of brightness (caused by the interaction between hue and
saturation). The differences in brightness (see Fig. 2 right and Fig.
3 second and third columns) make it e.g. easier to distinguish
between low uncertainty/low avalanche hazard and low
uncertainty/moderate avalanche hazard than between high
uncertainty/low hazard and high uncertainty/moderate hazard.
1
dwell time refers to the sum duration of fixations falling within a
particular AOI; fixation duration refers to the duration of
individual fixations.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 28
Figure 11. Trial duration (in ms; left) and number of transitions
(right) between the map and the legend
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 29
Figure 12. Number of transitions between the legend and the map
The following analyses focus on the differences between
extrinsic and intrinsic visualization as reflected by the differences
in dwell times and fixation durations (related to the legend and the
map). Graphs in Fig. 13 show the percentages of map/legend dwell
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 30
time for all the tasks. For extrinsic visualization tasks, the legend
dwell time ranges from 10 to 30% (of the total dwell time); visual
search time (map dwell time) ranges from 60 to 80%. Intrinsic
visualization, on the other hand, exhibits a far greater variability
across the individual tasks; for the legend, the dwell time ranges
from below 10% in one task to over 50% in another. In the latter
case, the time needed for decoding the legend even exceeds the
time needed to locate the target in the map.
Figure 13. Dwell times related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in %)
Fig. 14 presents a comparison of map and legend fixation
durations in intrinsic (right) and extrinsic (left) visualization tasks.
Fig. 14 shows the average fixation durations and spreads. As with
dwell times, the data show lower variability across the individual
tasks for extrinsic visualization: the curves of fixation durations are
nearly identical (see, for instance, the data concerning the second
task, with fixation durations slightly above 250 ms for both the
legend and the map). Intrinsic visualization, however, led to a
different situation during the same (second) task: legend-related
fixations were on average more than 100 ms longer than maprelated
fixations. From Fig. 15 it can be seen that extrinsic
visualization caused average fixation durations to last
approximately 250 ms for the map and the legend alike. In the case
of intrinsic visualization, the same value was true for map-related
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 31
average fixation durations; the average duration of legend-related
fixations, however, far exceeded 300 ms.
Figure 14. Fixation durations related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in ms)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 32
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 33
Figure 15. Fixation durations related to the map and the legend in
extrinsic (two boxplots on the left) and intrinsic (two boxplots on
the right) visualization tasks (in ms)
7. Discussion and conclusion
The primary objective of the study was to investigate the
relationship between the participants’ cognitive style and their
performance at map-related tasks employing different forms of
visualization. First, a Compound Figures Test (a variation of
Navon’s Hierarchical Figures Test) was administered using
MuTeP in order to discriminate between analytically- and globallyoriented
individuals. The test was already been successfully
employed in previous studies (Kubíček et al., 2016, Horváth,
2012). The results of the test confirmed the effect of global
precedence (Poirel, Pineau and Mellet, 2008). Subsequently, two
independent correlation studies were conducted in order to
establish the relationship between the participants’ performance at
global and local subtests of the CFT and their performance at maprelated
tasks employing different forms of cartographic
visualization (intrinsic and extrinsic). Each correlation study used a
research sample with a different level of map literacy. The first
research sample with a high degree of map literacy consisted of
students of cartography, while subjects with low levels of map
literacy were represented by students of psychology. In contrast
with the first study, the second correlation study showed no
significant correlation between the participants’ performance at
CFT and their performance at map-related tasks. We suppose that
the absence of any correlation might have been caused by low
levels of map literacy and the resulting subjective difficulty of the
map-related tasks. In a post-experimental inquiry, the students of
psychology more often reported that they had difficulties
understanding the notion of “avalanche risk uncertainty”. The
above was reflected in comparatively high error rates displayed by
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 34
the students of psychology. We believe that comprehension
problems acted as an intervening variable and might have been a
source of significant “noise” in the collected data. The studies
focused on the investigation of low-level cognitive processes,
particularly the visual decoding of the legend and subsequent
visual search for target areas in the map. It is, however, important
to realize that a map represents a complex communication channel
and its thematic layer (if present) triggers high-level cognitive
processes which may override the underlying low-level cognitive
processes. Thus, the amount of time needed for solving the maprelated
tasks did not primarily depend on the type of visualization
used or the speed of visual processing, but rather on the degree and
speed of understanding the instructions. According to Booth
(2006), tasks involving low-level cognitive processes are less
prone to invoking between-task differences in the subjects’
performance than tasks involving high-level cognitive processes,
where local/global processing styles are overridden by
executive/strategic processes. Thus, if a medium-strong correlation
is found between a simple, selective attention task (such as CFT)
and a complex map-related task, it can be viewed as indicating that
identical cognitive processes are at play in both cases.
In the first correlation study, which worked with a group of
participants with high levels of map literacy, significant
correlations were established between the participants’ cognitive
style and the type of visualization used in map-related tasks. A
statistically significant positive correlation (r= .387; p<0.05) was
established between scores obtained in the local subtest and
response times in map-related tasks employing intrinsic
visualization. No relationship was found between intrinsic
visualization and scores obtained in the global subtest. Similarly,
while a positive correlation (r= .384; p<0.05) was found to exist
between global-subtest response times and response times in maprelated
tasks employing extrinsic visualization, no relationship was
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 35
found between extrinsic visualization and the local subtest. Thus, it
can be concluded that subjects with high levels of map literacy
exhibit differences in their performance at map-related tasks
depending on their cognitive style and on the type of visualization
employed. Analytically-oriented individuals were better at tasks
involving intrinsic visualization, while globally-oriented
individuals performed better at tasks employing extrinsic
visualization. The fact that no cross-correlation was found between
either the local subtest and extrinsic visualization or between the
global subtest and intrinsic visualization renders the findings even
more significant. Following the two correlation studies in the first
phase, a comparative eye-tracking study was conducted in the
second phase with the aim of obtaining deeper insight into the
cognitive processes which were at play during task-solving; it was
not designed to test pre-defined hypotheses or compare types of
visualization with respect to their effectiveness.
The results of the eye-tracking part of our research indicate a
high dependency of performance at intrinsic visualization tasks on
the nature of the tasks (i.e. on the values of target variables). An
example of the above is represented by the high across-task
variability in the number of map/legend transitions. At the same
time, intrinsic visualization tasks exhibit a far greater number of
transitions than extrinsic visualization tasks, requiring more
“checking look-backs” at the legend. Another analyzed parameter
was represented by the map/legend dwell time ratio. For extrinsic
visualization tasks, the legend dwell time was relatively short,
ranging from 10 to 30% (of the total dwell time), while visual
search time (map dwell time) ranged from 60 to 80%. Intrinsic
visualization, on the other hand, exhibited a far greater variability
across the individual tasks. For the legend, the dwell time ranges
from below 10% in one task to over 50% in another. Time spent on
visual search amounted to more than 70% of the total dwell time in
some tasks and less than 40% in others. In some cases, the time
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 36
needed for decoding the legend even exceeded the time needed to
locate the target in the map. In addition, intrinsic visualization
resulted in longer fixation durations and greater fixation duration
variability across the individual tasks. The results indicate that in
intrinsic tasks, decoding the legend may in some cases require
significant cognitive effort (see Longo and Barrett, 2010). An
instance of the above may be represented by the established acrosstask
variability in the difficulty of legend decoding in intrinsic
visualization tasks. The findings regarding intrinsic and extrinsic
visualization support the hypothesis that the different types of
visualization are not computationally equivalent and that each of
them requires a different cognitive strategy. It can be assumed,
further, that an individual’s performance will be the best if his/her
cognitive style and the preferred cognitive strategy are in
consonance with the type of task given (see Fig. 1). We believe,
based on the results of the first correlation study, that analyticallyoriented
individuals were better able to decode the legend (i.e.
discriminate between the employed variables), which resulted in
shorter task times.
The results of the first correlation study and of the comparative
eye-tracking study confirmed the hypothesis that the process of
solving map-related tasks is, to a large extent, dependent on the
type of visualization used in the task. The correlation study
established a difference between the participants in their maprelated
performance depending on their cognitive style, indicating
that different cognitive strategies might have been at play.
Globally-oriented individuals were better at extrinsic visualization
tasks, where two variables were represented by two different
modalities which needed to be processed simultaneously; intrinsic
visualization tasks, on the other hand, allowing for sequential
processing of two variables represented by a single modality
(color), were “easier” for locally- (analytically-)oriented
individuals. The eye-tracking study enabled to record the
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 37
participants’ eye movements during the experiments and to learn
about possible differences in task-solving in a more direct way.
We are currently working on a replicative study with a modified
research design so that the findings of the present study can be
fully confirmed. The new design will avoid the term “uncertainty”;
care will be taken to ensure that only concepts familiar to
individuals with low levels of map literacy are used. In addition, a
more in-depth analysis of the participants’ performance in maprelated
tasks will be carried out. A new testing system will be used,
enabling the interconnection of SW Hypothesis (Štěrba et al.,
2015) with SW Ogama (Voßkühler et al., 2008) and the eyetracking
system Eye Tribe or SMI RED250mobile (Popelka et al.,
2016): The system uses a presentation interface which allows for
extensive and intensive data collection to be conducted
simultaneously.
References
Alaçam, Ö. and Dalci, M. 2009. A usability study of WebMaps with eye tracking
tool: the effects of iconic representation of information. In: J. A. Jacko, ed.
Human-Computer interaction. New trends. Berlin, Heidelberg, New York:
Springer, pp.12-21.
Alexander, P. A. Kulikowich, J. M. and Schulze, S.K. 1994. The influence of
topic knowledge, domain knowledge, and interest on the comprehension of
scientific exposition. Learning and Individual Differences, 6(4), pp.379–397.
Bertin, J. 1973. Sémiologie graphique. Paris: La Haye.
Booth, R. D. L. 2006. Local-global processing and cognitive style in autism
spectrum disorders and typical development. [pdf] PhD. Institute of Psychiatry,
King's College London. Available at:
<https://kclpure.kcl.ac.uk/portal/files/2935437/433168.pdf>[Accessed 12 June
2016].
Bouvet, L. Rousseta, S. Valdoisa, S. and Donnadieu, S. 2011. Global
precedence effect in audition and vision: Evidence for similar cognitive styles
across modalities. Acta Psychologica, 138, pp.329–335.
Brand, J. and Johnson, A. P. 2014. Attention to local and global levels of
hierarchical Navon figures affects rapid scene categorization. Frontiers in
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 38
Psychology, [e-journal] 5, 1274. Available at: <doi:
10.3389/fpsyg.2014.01274>[Accessed 12 June 2016].
Buttenfield, B. 2000. Mapping ecological uncertainty. In: C. T. Hunsaker et al.,
eds. Spatial uncertainty in ecology. New York: Springer.
Çöltekin, A. Heil, B. Garlandi, S. and Fabrikant, S. I. 2009. Evaluating the
effectiveness of interactive map interface designs: a case study integrating
usability metrics with eye-movement analysis. Cartography and Geographic
Information Science, 36(1), pp.5-17.
Creswell, J. W. 2008. Research Design: Qualitative, Quantitative, and Mixed
Methods Approaches. Los Angeles, London, New Delhi, Singapore: Sage
Publications, Inc.
Dale G. and Arnell, K. M. 2014. Lost in the Forest, Stuck in the Trees:
Dispositional Global/Local Bias Is Resistant to Exposure to High and Low Spatial
Frequencies. PLoS ONE, [e-journal] 9(7): e98625. Available at:
<doi:10.1371/journal.pone.0098625>[Accessed 10 July 2016].
Duchaine, B. Yovel, G. and Nakayama, K. 2007. No global processing deficit in
the Navon task in 14 developmental prosopagnosics. Social Cognitive and
Affective Neuroscince, 2(2), pp.104–113.
D’Zmura, M. 1991. Color in visual search. Vision Research, 31, pp.951–966.
Elmer, M., E. 2013. Symbol Considerations for Bivariate Thematic Maps.
Proceedings of ICC Conference, [pdf] Dresden, 25-30 August 2013. Available
at: <
http://icaci.org/files/documents/ICC_proceedings/ICC2013/_extendedAbstract/
278_proceeding.pdf>[Accessed 10 April 2016].
Gershon, N. 1998. Visualization of an imperfect world. Computer Graphics and
Applications, 18, pp.43–45.
Graff, M. 2003. Learning from web-based instructional systems and cognitive
style. British Journal of Educational Technology, 34(4), pp.407–418.
Hammond, K. R. 1996. Human judgment and social policy: Irreducible
uncertainty, inevitable error, unavoidable injustice. New York: Oxford University
Press.
Hofmann, A. Hošková-Mayerová, Š. Talhofer, V. and Kovařík, V. 2015. Creation
of models for calculation of coefficients of terrain passability. Quality and
Quantity, 49(4), pp.1679-1691.
Horváth, J. 2012. Funkce kognitivního stylu při čtení mapy v kontextu krizového
řízení. (Role of cognitive style with respect to map reading in context of crisis
management). Diploma Thesis. Brno: Masaryk University.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 39
Hus, V. and Hojnik, T. 2013. Comparative Analysis of Cartographic Literacy in
the Selected Curricula at the Primary Level. Creative Education, 4(12), pp.757-
761.
Itti, L. and Koch, C. 2000. A saliency-based search mechanism for overt and
covert shifts of visual attention. Vision Research, 40, pp.1489–1506.
Jones, Ch. 1997. Geographical Information Systems and Computer
Cartography. London and New York: Routlege.
Kirton, M. J. 1989. Adaptors and innovators: styles of creativity and problemsolving.
London: Routledge.
Koláčný, A. 1977. Cartographic Information – A Fundamental Concept and Term
in Modern Cartography. Cartographica: The International Journal for
Geographic Information and Geovisualization, 1(14), pp.39-45.
Konečný, M. et al. 2011. Dynamická geovizualizace v krizovém managementu.
Brno: Masarykova univerzita.
Kozhevnikov, M. 2007. Cognitive styles in the context of modern psychology:
Toward an integrated framework. Psychological Bulletin, 133(3), pp.464-481.
Kubíček, P. Šašinka, Č. and Stachoň, Z. 2014. Vybrané kognitivní aspekty
vizualizace polohové nejistoty v geografických datech. Geografie - Sborník
České geografické společnosti, Česká geografická společnost, 119(1), pp.67-90.
Kubíček, P. et al. 2016. Cartographic Design and Usability of Visual Variables
for Linear Features. The Cartographic Journal, pp.1-11.
Kunz, M. and Hurni, L. 2011. How to enhance cartographic visualizations of
natural hazards assessment results. The Cartographic Journa, 48(1), pp.60-71.
Kunz, M. 2011. Interactive visualizations of natural hazards data and associated
uncertainties. PhD. ETH Curych.
Larkin, J. H. and Simon, A. H. 1987. Why a Diagram is (Sometimes) Worth Ten
Thousand Words. Cognitive science, 11, pp.65-99.
Leitner, L. and Buttenfield, B. 2000. Guidelines for the display of attribute
certainty. Cartography and Geographic Information Science, 27, pp.3–14.
Leech, N. L. and Onwuegbuzie, A. J. 2009. A typology of mixed methods
research designs. Quality & Quantity, 43(2), pp.265-275.
Longo, L. and Barrett, S. 2010. A Computational Analysis of Cognitive Effort.
In: N. T. Nguyen, M. T. Le and J. Świątek, eds. Intelligent Information and
Database Systems. Volume 5991 of the series Lecture Notes in Computer
Science. New York: Springer, pp 65-74.
Maceachren, A. 1992. Visualizing uncertain information. Cartographic
Perspectives, 13, pp.10–19.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 40
MacEachren, A. M. and Taylor, D. R. F. 1994. Visualization in Modern
Cartography. Oxford: Pergamon.
Maceachren, A. et al. 2005. Visualising geospatial information uncertainty.
What we know and what we need to know. Cartography and Geographic
Information Science, 32, pp.139–160.
MacEachren A. M. et al. 2012. Visual semiotics & uncertainty visualization: An
empirical study. Transactions on Visualization & Computer Graphics. 18(2)
pp.2496-2505.
Milne, E. and Szczerbinski, M. 2009. Global and local perceptual style, fieldindependence,
and central coherence: An attempt at concept
validation. Advances in Cognitive Psychology, 5, pp.1-26.
Montello, D. R. 2009. Cognitive Research in GIScience: Recent Achievements
and Future Prospects. Geography Compass, 3(5), pp.1824–1840.
Navon, D. 1977. Forest Before Trees: The Precedence of Global Features in
Visual Perception. Cognitive Psychology, 9(3), pp.353–383.
Nothdurft, H. C. 2000. Salience from feature contrast: Additivity across
dimensions. Vision Research, 40, pp.1183–1201.
Olson, J. M. 1979. Cognitive cartographic experimentation. Cartographica: The
International Journal for Geographic Information and Geovisualization, 16(1),
pp.34-44.
Ooms, K. et al. 2012. Investigating the effectiveness of an Efficient Label
Placement Method Using Eye Movement Data. The Cartographic Journal, 49 (3),
pp.234-246.
Pask. G. 1976. Styles and strategies of learning. British journal of educational
psychology, 46, pp.128-148.
Poirel, N. Pineau, A. and Mellet, E. 2008. What does the nature of the stimuli
tell us about the Global Precedence Effect?. Acta Psychologica, 127, pp.1–11.
Popelka, S. and Brychtová, A. 2013. Eye-tracking study on different perception
of 2D and 3D terrain visualisation. Cartographic Journal, 50(3), pp.240-246.
Popelka, S. and Dědková, P. 2014. Extinct Village 3D visualization and its
Evaluation with Eye-Movement Recording. In: B. Murgante et al., eds.
Computational Science and Its Applications - ICCSA 2014, Part 1. New York:
Springer, pp. 786-795.
Popelka, S. Stachoň, Z. Šašinka, Č. and Doležalová, J. 2016. Eyetribe Tracker
Data Accuracy Evaluation and Its Interconnection with Hypothesis Software for
Cartographic Purposes. Computational Intelligence and Neuroscience, [ejournal]
Available at: <doi:10.1155/2016/9172506>[Accessed 30 August
2016].
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 41
Rayner, S. G. 2000. Reconstructing style differences in thinking and learning:
Profiling learning performance. In: R. J. Riding and S. G. Rayner, eds.
International Perspectives on Individual Differences. Volume 1 Cognitive Styles,
Stamford: Ablex, pp.115-177.
Rezaei, A. R. and Katz, L. 2004. Evaluation of the Reliability and Validity of the
Cognitive Styles Analysis Inventory and Recommendations for Improvement.
Personality and Individual Differences, 36(6), pp.1317-1327.
Roth, R. E. et al. 2011. Card sorting for cartographic research and practice.
Cartography and Geographic Information Science, 38(2), pp.89-99.
Smith Mason, J. et al. 2016. Special issue introduction: Approaching spatial
uncertainty visualization to support reasoning and decision making. Spatial
Cognition & Computation, 16(2), pp.97-105.
Staněk, K. Friedmannová, L. Kubíček, P. and Konečný, M. 2010. Selected issues
of cartographic communication optimization for emergency centers.
International Journal of Digital Earth, 3(4), pp.316-339.
Spirduso, W. W. 1980. Physical Fitness, Aging, and Psychomotor Speed: A
Review. Journal of Gerontology, 35(6), pp.850-865.
Štěrba, Z. et al. 2014. Mixed Research Design in Cartography: A Combination
of Qualitative and Quantitative Approaches. Kartographische Nachrichten,
64(5), pp.262-269.
Štěrba, Z. et al. 2015. Selected Issues of Experimental Testing in Cartography.
Brno: Masaryk University.
Trau, J. and Hurni, L. 2007. Possibilities of incorporating and visualizing
uncertainty in natural hazard prediction. In: Proceedings of the 23rd
International Cartographic Conference ICC. [pdf] Moscow, Russia, 4-10 August
2007, Available at: <
http://www.mountaincartography.org/publications/papers/ica_cmc_sessions/5_
Moscow_Session_Mountain_Carto/moscow_hurni_trau.pdf>[Accessed 17 May
2016].
Voßkühler, A. Nordmeier, V. Kuchinke, L. and Jacobs, A.M. 2008. OGAMA OpenGazeAndMouseAnalyzer:
Open source software designed to analyze eye
and mouse movements in slideshow study designs. Behavior Research
Methods, 40(4), pp.1150-1162.
Witkin, H. A. Goodenough, D. R. and Karp, S.A. 1967. Stability of cognitive
style from childhood to young adulthood. Journal of Personality and Social
Psychology, 7(3), pp.291-300.
Wehrend, S. and Lewis, C. 1990. A Problem-oriented Classification of
Visualization Techniques. In: A. Kaufman, ed. Proceedings of the 1st conference
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 42
on Visualization '90. San Francisco, 23-26 October1990, IEEE Computer Society
Press, Los Alamitos, pp.139-143.
Yovel, I. Revelle, W. and Mineka, S. 2005. Who Sees Trees Before Forest? The
Obsessive-Compulsive Style of Visual Attention. Psychological Science, 16(2),
pp.123-129.
Zhang, J. and Goodchild, M. 2002. Uncertainty in geographical information.
London: Taylor and Francis.
List of figures
Figure 1. Triarchic structural model of performance when solving
map-related task (p.
Figure 2. Snow avalanche hazard and hazard uncertainty map: left
– extrinsic visual representation; right – intrinsic visual
representation (adapted from Kunz, 2011)
Figure 3. Differences in the encoding of variables between the
extrinsic (first column) and intrinsic (second column) form of
visualization; third column shows the difference in salience
between a fully saturated yellow (1) and partially desaturated blue
(6) on white and black backgrounds
Figure 4. Structure of used combined extensive-intensive research
design
Figure 5. Structure of map-related tasks (students of cartography –
top; students of psychology – bottom)
Figure 6. Sample map-related tasks: Left (intrinsic visualization) –
Select the area with a moderate level of avalanche hazard. Right
(extrinsic visualization) – Select the area with a high level of
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 43
avalanche hazard accompanied by a low level of avalanche hazard
uncertainty.
Figure 7. A compound stimulus (left); an item from the Compound
Figures Test adapted into the MuTeP environment (right).
Figure 8. Reaction times (s) for both subtests of CFT
Figure 9. Process of investigating variables: 1) stimulus -> output;
2) stimulus -> cognitive processing and associate behavior ->
output
Figure 10. Structure of map-related tasks in the eye-tracking study
Figure 11. Trial duration (in ms; left) and number of transitions
(right) between the map and the legend
Figure 12. Number of transitions between the legend and the map
Figure 13. Dwell times related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in %)
Figure 14. Fixation durations related to the map and the legend in
extrinsic (left) and intrinsic (right) visualization tasks (in ms)
Figure 15. Fixation durations related to the map and the legend in
extrinsic (two boxplots on the left) and intrinsic (two boxplots on
the right) visualization tasks (in ms)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
The Impact of Global/Local Bias on Task-solving in Map-related
Tasks 44
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Čeněk Šašinka is a Post-doctoral researcher at the Department of Psychology at Masaryk
University in Brno. He is head of the Center for Experimental Psychology and Cognitive
Sciences. He is one of the founders of the HUME Lab (Experimental Humanities Laboratory)
and he is a member of executive board of the HUME Lab at the same time. His research
interests focus mainly on psychological assessment, cognitive visualization, and differences
among individual users in map perception and interpretation.
Biographical Note
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Biographical Picture Click here to download Biographical Picture picture_sasinka.jpg
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65