DissLiteratur/storage/SD9DM67C/.zotero-ft-cache

Using Eye Tracking to Identify Cognitive Differences: A Brief Literature Review

George E. Raptis
HCI Group Department of Electrical and
Computer Engineering University of Patras Patras, Greece
raptisg@upnet.gr

Christos A. Fidas
Department of Cultural Heritage Management and
New Technologies University of Patras
Patras, Greece
ﬁdas@upatras.gr

Nikolaos M. Avouris
HCI Group Department of Electrical and
Computer Engineering University of Patras Patras, Greece
avouris@upatras.gr

ABSTRACT
Being the windows to the soul, eyes reveal information about individuals’ feelings, emotions and behaviour, aﬀecting various cognitive tasks, such as focus of attention, spatial cognition and navigation, cognitive load, etc. With the increased use of computer systems, complex information is visualized and communicated through visual interfaces as a mean of information presentation to and processing by the users. However, people diﬀer regarding the way they seek, retrieve, process, comprehend, organize and recall information, based on their individual perceptual characteristics, cognitive skills, abilities and styles. Therefore, the point and the motion of the eye gaze could reveal behavioural patterns related to individual cognitive diﬀerences; patterns that are extracted using eye tracking tools which quantify and provide compelling data regarding eye gaze movement. In this paper we review the current literature regarding the eﬀect between the FD-I cognitive style of users in visual exploration and search activities and to correlate these with objective measures gathered through eye-tracking.
CCS Concepts
•Human-centered computing → HCI theory, concepts and models; •Computing methodologies → Cognitive science;
Keywords
Cognitive styles; Eye tracking; Visual perception; Cognitive diﬀerences; Literature review
1. INTRODUCTION
Several studies [11] have indicated that people with different iris features, gazing patterns, and eye characteristics tend to develop along diﬀerent personality and cognition lines, reﬂecting on their individual diﬀerences and their cog-
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for proﬁt or commercial advantage and that copies bear this notice and the full citation on the ﬁrst page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior speciﬁc permission and/or a fee. Request permissions from permissions@acm.org. PCI ’16, November 10-12, 2016, Patras, Greece c 2016 ACM. ISBN 978-1-4503-4789-1/16/11. . . $15.00 DOI: http://dx.doi.org/10.1145/3003733.3003762

nitive styles. The evolutionary technological era has enabled the development of highly accurate eye tracking tools which collect and analyse data regarding eye movement reﬂecting on visual perception abilities. Elaborating such tools allows for identifying individual cognitive diﬀerences and enables researchers to further understand user visual behaviour.
1.1 Cognitive Styles
Cognitive styles refer to “people’s characteristics and typically preferred modes of processing information”[26]. There is a number of related terms that describe characteristics similar to cognitive styles, including learning, thinking and information processing styles. Lying at the junction between cognition and personality, cognitive styles describe individual diﬀerences in how people use their cognitive skills and abilities to solve problems and process information in their preferred way.
A number of cognitive styles have been proposed [1, 10, 23, 32], modelling the individuals’ preferred way of information processing, based on their cognitive diﬀerences. One of the most well established, credible and validated [2, 5] cognitive styles is the Field Dependence-Independence (FD-I) style [32]. It is a single dimension model having the ﬁeld dependence on the one side and the ﬁeld independence on the other. According to this model, the individuals are classiﬁed either as ﬁeld dependent (FD) or as ﬁeld independent (FI). However, several studies on FD-I style have introduced a third cognitive group: the ﬁeld neutral (FN) or ﬁeld mixed (FM), which reﬂects on individuals who do not have a clear orientation towards FD or FI [12, 25]. FD individuals tend to prefer a more holistic way when processing information, have diﬃculties in identifying details from information in complex schemes and perform better on inductive tasks [32]. On the other hand, FI individuals tend to prefer impersonal orientation, follow a more analytical way when processing information, pay attention to details and easily separate simple elements and structures from the surrounding composite context [32]. Several cognitive style elicitation tools have been developed to classify individuals as FI or FD. The most credible, validated and widely used tools are based on traditional techniques, like ”paper-and-pencil” and they include the tools:
– Hidden Figures Test (HFT) [7];
– Group Embedded Figures Test (GEFT) [19];
– Children Embedded Figures Test (CEFT) [31];
– Embedded Figures Test (EFT) [31].

1.2 Visual Perception
FD-I cognitive style is closely related to visual perception, i.e. the ability to organise, identify and interpret the surrounding environment by processing visually displayed information [32]. One of the fundamental approaches regarding the visual perception is the Gestalt theory and its derived principles on how people acquire and maintain meaningful perceptions in a chaotic world. The cognitive elicitation tools discussed in the previous section are greatly inﬂuenced by the Gestalt framework. The ﬁrst stage of the visual perception process comprehends the phases of exploration and search. These stages diﬀer in great extent, and they inﬂuence people’s behaviour when they are looking for information in chaotic environments [20]. During the visual exploration, people get an overview of the data and they interactively browse through diﬀerent portions of the data, as they have only vague hypotheses about the nature of the data. In this sense, visual exploration can be understood as an undirected search for relevant information within the data [28]. To support users in the search process, a high degree of interactivity must be a key feature of visual exploration techniques. On the other hand, visual search requires high degree of attention, which typically involves an active scan of the visual environment, seeking for a particular object, i.e. target, within a cluttered visual set of other objects, i.e. distractors [29].
1.3 Eye Tracking
Measures of eye movements and visual perception are highly correlated, especially in visually complex scenes, where many important details cannot be easily resolved [33]. Furthermore, since eye movements can be observed directly, unlike shifts of covert attention, they provide a rich dataset to improve our understanding of visual search and exploration [33]. A very popular technique on analysing visual perception is the eye-tracking, which is used to investigate the approach an individual follows to solve a problem visually, identify where attention is drawn and understand the required eﬀort and the cognitive workload of the users when solving the problem [22]. The recent technological advances had a major impact on the eye tracking industry, which provides accurate solutions of high standards to researchers, helping them understand the users’ behaviour while interacting with computer systems.
An eye tracker mechanism captures gaze information in terms of ﬁxations and saccades. Fixations occur when a person focuses their vision on a particular point over a period of time, while quick changes in eye position are referred as saccades. These two parameters along with ﬁxation duration, i.e. the amount of time the eye ﬁxate within an area of interest (AOI), are the fundamental features of the analyses used in the literature in order to derive viewers’ attention patterns. A number of more complex high-level eye tracking metrics have been built on the aforementioned fundamental features, which are widely used in eye tracking studies [27]. Such metrics include:
– Fixation count, rate, and duration;
– Relative and absolute saccade angle;
– Heat maps and scan paths;
– Fixation proportionate in AOIs;
– Fixation transitions in AOIs.

2. EFFECTS OF FD-I IN VISUAL SEARCH AND VISUAL EXPLORATION
2.1 Effect on Visual Exploration
Visual exploration occurs when people do not search for speciﬁc information within a complex environment, but they visually browse the given data [20, 28]. Being the initial stage of the visual perception, many researchers have studied the eﬀect of FD-I style on visual exploratory tasks. In 1978, Shinar [24] studied the driving behaviour diﬀerences between FD and FI individuals with the use of eye tracking mechanisms. The results of his study indicated that FD individuals spent more on the focus of expansion; an overall good approach when driving on straight road, but ineﬃcient while negotiating curves. Moreover, FD drivers were more concentrated on their ﬁxations, and thus they did not easily adapt themselves to changing environments, thus they had slower visual search process than FI drivers. Therefore, FD drivers tend to be less adaptive and eﬃcient in curve negotiating environments where the perceptual load is drastically increased and the target area, i.e. the road, changes iteratively within their visual ﬁeld.
Yekan and Cagiltay [34] investigated the eﬀects of FDI style on students’ interaction with a computer-based instructional interface through eye-tracking measures. They examined whether there are diﬀerences between FD and FI individuals in terms of ﬁxation count and duration when exploring an instructional website. The website was built by the researchers and was divided into ten AOIs. Their ﬁndings revealed no statistically signiﬁcant diﬀerences, as none of the AOIs was preferred by one or the other group; however, the FD individuals tended to ﬁxate in the AOIs for longer time periods than FI students. This could imply that FD individuals have lower performance in hypermediabased activities; no statistically signiﬁcant diﬀerences were identiﬁed though. The fact that the instructional website was used for the very ﬁrst time by the study participants could have an impact on the study ﬁndings, as the learning curve was not taken into account by the researchers.
Mawad et al. [14] investigated the eﬀect of FD-I style on information processing and selection of food product labels. The study participants had to choose a yoghurt to consume, after looking the labels of two diﬀerent products. The analysis revealed that FD and FI individuals had diﬀerent visual approach on information seeking in order to decide which product they would choose. FI consumers had longer ﬁxation durations and spent more time evaluating the two product labels and they tended to seek information more exhaustively, having greater ﬁxation ratios in speciﬁc areas of the yoghurt labels such as nutrition facts. This ﬁnding suggests that FI consumers tended to sustain their attention on the product labels and performed a deeper evaluation of the presented information before choosing a product.
Raptis et al. [21] investigated how visual attention on interactive game elements is distributed among players with diﬀerent cognitive styles. The study participants were asked to play a video game, in which they had to explore an ancient civilisation, and their eye movements were analysed based on ﬁxation count and duration. The eye tracking analysis revealed signiﬁcant diﬀerences between the game-playing approaches of FD and FI individuals. FI gamers had greater ﬁxation count and ﬁxation duration than FD gamers; a ﬁnding that could be explained by the analytical nature of FIs.

However, the game had diﬀerent types of elements: items that their acquisition was compulsory in order for the players to proceed in the level, and optional items that would help players to enrich their knowledge about the ancient civilisation. No signiﬁcant diﬀerences was found regarding the ﬁxation count and ﬁxation duration towards compulsory objects; whereas a signiﬁcant eﬀect regarding the ﬁxation count and ﬁxation impact toward bonus items was revealed. Overall, FD players observed less times and for shorter time periods the game items, as they followed a more intrinsic approach and were less inclined in detecting details, whereas FI players were more analytical, and thus they observed bonus items more often and for longer time periods.
In a recent study Katsini et al. investigated the eﬀect of FD-I style on a graphical authentication mechanism. The study participants were asked to create an image based password by choosing ﬁve diﬀerent images of a set of 199 images. The primary ﬁnding of their study was that FD individuals ﬁxated on a smaller subset of the image grid than FIs. FDs visually explored about 30% less images, which reﬂected on the security level of their passwords, with FDs being more vulnerable. Moreover, FD and FI individuals tended to follow diﬀerent visual approaches, with FDs being more holistic and FIs more analytical, as it was shown through heatmaps.
Apart from adults, studies have performed on children as well. Baron [3] investigated whether FI individuals exhibit more proﬁcient scanning strategies than do FD individuals, when watching a television program. The study participants were 85 third-grade pupils who undertook CEFT [31] to be classiﬁed as FD or FI. The participants then watched the Children’s Television Workshop program ”The Electric Company”, and the researcher was freezing the moving segments at speciﬁc intervals, which were chosen for their individual stimulus qualities such as electronic bridges, animation, etc. The analysis of ﬁxation count and duration revealed that there were no major diﬀerences in eye movement patterns between FD and FI individuals. However, FIs showed more proﬁcient scanning strategies as indicated by signiﬁcant diﬀerences in percentage of ﬁxations on target and ﬁxation duration for speciﬁc segments. This ﬁnding raises questions about the reason why high-level cognitive diﬀerences have an impact on the way individuals scan, but it was not further examined in Baron’s paper.
2.2 Effect on Visual Search
In contrast to visual exploration, visual search requires high degree of attention, which typically involves an active scan of the visual environment, seeking for a particular object (target) within a cluttered visual set of other objects (distractors) [29]. In 1968, Conklin et al. [6] investigated the eye-movement patterns of high and low scorers on EFT elicitation tool. Thirty-two ﬁrst-year university students took part in the study, and a series of EFT problems were presented to them. When each subject felt that he had solved the task in a time frame of 20 seconds, he was asked to close his eyes. The dependent variables of their study were: the mean ﬁxation durations and the mean track lengths. Moreover the Informative Search Score (ISS) [13] was determined for each subject. Inspection of the track-length variable revealed that the observed diﬀerences in track length are associated with the degree of structure in the stimulus. The greatest diﬀerence in track length between FDs and FIs was observed on the very unstructured ﬁgures. The ﬁgures

of medium level of structure complexity yielded a marginal diﬀerence on track length. This ﬁnding suggests that FI individuals tend to have longer and more random eye movements than FDs as the structure complexity increases. While no diﬀerences between the samples were found for ﬁxation duration, there was a tendency for study participants to ﬁxate longer on less structured ﬁgures. An interesting ﬁnding is the highly reliable diﬀerence found on the ISS variable. This result suggests that FI individuals are capable of attending to the more relevant information more precisely and can quickly solve such tasks as the EFT.
Wijnen and Groot developed the software system EMAS [30], to analyse eye movement data recorded during ﬁgural tasks. The study participants undertook the HFT [7] and the EFT [31] tools, and the collected eye movement data were examined through EMAS functions. It was revealed that FI individuals scanned systematically, for longer time periods at speciﬁc sectors, and had a greater overall ﬁxation count. In the contrast, FD individuals scanned unsystematically in an ”unarticulated” way with many short ﬁxations nearly all sectors during the two tests. Despite the fact that the scan paths followed by FD and FI individuals were diﬀerent, the researchers did not perform exhaustive analysis on them, to gain more insights on the way FDs and FIs scan an area when seeking for a particular visual clue. Nevalainen and Sajaniemi [15] investigated the short-term eﬀects of graphical versus textual visualisation of variables on program perception. The participants studied four simple Pascal programs using two diﬀerent tools, PlanAni programming animator and Turbo Pascal programming environment; a graphical and textual environment respectively. Then they had to write a brief description of each program they studied. During the eye–tracking study the screen was divided into three areas: the code area, the variables area and the rest area. The variables area was viewed more times in PlanAni than Turbo Pascal in general, and FD individuals paid more attention than FI individuals on it. Turbo Pascal is a textual visualisation environment, and it does not require the users to be able to separate meaningful items from structured perceptual ﬁeld like in PlanAni, which provides a rich colourful graphical interface. Therefore, the level of ﬁeld-independence had an eﬀect on the visual attention and perception which would had an impact on the constructed mental models of the programmers. However, we should mention that due to nature of the programming environments, the variables areas had diﬀerent size, with PlanAni having a larger variables area, which could contribute to the overall ﬁxation count of each area.
Nisiforou et al. [18] investigated the impact of cognitive abilities on visual search tasks. The study participants were classiﬁed as FD or FI and then each of them undertook a set of nine fact-ﬁnding tasks, with the information being sought placed on a particular location of the webpage. They examined both the completion time needed for each task and the visual search behaviour of the study participants. The visual search behaviour was analysed qualitatively through gaze plots, focus maps and heat maps. The analysis revealed that the behavioural patterns among each cognitive group were highly related to the complexity of the scanned webpage. For webpages of low complexity the visual search strategies were similar. However, the scan paths of FD individuals appeared to be more disoriented and scattered on webpages of medium and high complexity, in contrast to FI individuals,

Study Baron [3]
Conklin et al. [6]
Jia et al. [8]
Katsini et al. [9]
Mawad et al. [14]
Nevalainen and Sajaniemi [15] Nisiforou and Laghos [16]
Nisiforou and Laghos [17]
Nisiforou, Michailidou and Laghos [18]
Raptis et al. [21]
Shinar [24]
Wijnen and Groot [30]
Yekan and Cagiltay [34]

Elicitation tool CEFT [31]

Participants
– 85 third-grade pupils – 42 boys and 43 girls – CEFT: 3-24 (m=15.3) – CS: 36 FDs and 49 FIs

Apparatus
Polymetrics eye movement recorder (V-1164-1)

EFT [31]

– 32 UG students – Gender: 16 F, 16 M

Polymetrics eye movement recorder (V-1164-1)

EFT [31] – 30 UG students

n/a

– Gender: 24 F, 6 M

– CS: 15 FDs, 15 FIs

GEFT [19]

– 51 adults – Gender: 16 F, 35 M – Age: 18-40 (29.3±5.8) – GEFT: 3-18 – CS: 25 FDs and 26 FIs

Tobii Pro Glasses 2

GEFT [19]

– 133 adults; – Gender: 88 F, 45 M – Age: 18-46 (23.3±5.1) – GEFT: 2-18 (m=8.7) – CS: 77 FDs and 56 FIs

Tobii T60 Eye Tracker

GEFT [19]
HFT [7]

– 12 UG students – Gender: 5 F, 7 M – GEFT: 7-18 (14.75±3.22) – 16 PhD students – CS: 5 FDs 5 FIs 5FNs

n/a
SMI iViewX; BeGaze 3.1

HFT [7]

– 54 UG/PG students – Age: 18-35 – CS: 24 FDs 14 FIs 16 FNs

SMI iViewX, BeGaze 3.1

HFT [7]

– 16 adults – Age: 18-28 – CS: 7 FDs 6 FIs 3 FNs

SMI iViewX; BeGaze 3.1

GEFT [19]
EFT [31]
EFT [31] HFT [7]

– 14 UG/PG students – Gender: 2 F , 12 M – Age: 18-23 (20.5±1.8) – GEFT: 11.7±2.9 – CS: 8 FDs and 6 FIs

Tobii T60 Eye Tracker

– 5 UG students; – Gender: 2 F, 3 M

Shinar’s et al. built eye-tracking mechanism

n/a

EMAS

software

system

GEFT [19]

– UG students – GEFT: 14.82±2.87

Tobii Eye-Tracker device

ET metrics Fixation count, duration and ratio
Fixation duration; track length; Informative Search Score (ISS)
Fixation count and duration
Fixation count and duration; heat maps
Fixation count, duration and ratio
Fixation count and duration
Fixation count; heat maps; scan paths
Fixation count; saccades count; heat maps; scan paths
Scan paths; heat maps; focus maps.
Fixation count and duration
Fixation count and duration
Fixation count and duration; scan paths
Fixation count and duration

Results
The analysis of ﬁxation count and ﬁxation duration revealed that there were no major diﬀerences in eye movement patterns between FD and FI individuals. However, FIs showed more proﬁcient scanning strategies as indicated by signiﬁcant diﬀerences in percentage of ﬁxations on target and ﬁxation duration for speciﬁc segments. The observed diﬀerences in track length were related with the degree of structure in the stimulus. The greatest difference in track length between FDs and FIs was observed on the very unstructured ﬁgures. FDs tended to have longer and more random eye movements than FIs as the structure complexity increased. The diﬀerence found on the ISS variable suggests that FIs are capable of attending to the more relevant information more precisely and can quickly solve such tasks as the EFT. FIs could focus their attention to cue-pointed objects relying on internal references, while FDs tended to scan both the targets and distractors without the cue using external references. The study results indicated that FIs could remove irrelevant information more eﬃciently than FDs. In order to create their graphical password, FDs ﬁxated on a smaller subset of the image grid than FIs. FDs visually explored about 30% less images, which reﬂected on the security level of their password, with FDs being more vulnerable. Moreover, FDs and FIs followed diﬀerent visual approaches, with FDs being more holistic and FIs more analytical, as it was revealed through heatmaps. FI consumers had more ﬁxations and spent more time evaluating the food products before choosing one. They sought information more exhaustively than FDs, having greater ﬁxation ratios in speciﬁc AOIs, such as the nutrition facts label. FI consumers sustained their attention on the product labels and performed a deeper evaluation of the presented information before choosing a product. The level of ﬁeld-independence had an eﬀect on the visual attention and perception which could had an impact on the constructed mental models of the programmers.
FDs and FIs followed diﬀerent visual search strategies in order to identify a given pattern. FD users scanned diﬀerent points of the given shape, as they could not recognise the hidden ﬁgure. On the other hand, FI individuals could easily recognise the hidden shape, and thus their quest was more organised and their ﬁxations more precise. FD individuals could not identify the correct shape as they were looking at diﬀerent areas than the shape of interest and they spent more time in ﬁxating into incorrect shapes. On the other hand, FIs could easily and quickly recognise the simple shape hidden in the complex pattern. The scan paths of the FI revealed a more oriented eye movement behaviour producing less number of ﬁxations and saccades than FDs. FDs generated double ﬁxations than FNs and quadruple ﬁxations than FIs. These ﬁndings suggest that FD-I cognitive style aﬀects the eye movement patterns, revealing a more ineﬃcient visual search behaviour as the ﬁeld dependence level increases. The visual search behaviour depends on the environment complexity. For webpages of low complexity the visual search strategies of FD and FI individuals were similar. However, the scan paths of FD individuals appeared to be more disoriented and scattered on webpages of medium and high complexity, in contrast to FI individuals, who displayed more oriented and organised scan paths. FD players observed less times and for shorter time periods the game items, as they followed a more intrinsic approach and were less inclined in detecting details. On the other hand FI players followed a more analytical approach, and they observed the game items more often and for longer time periods. FD drivers were more concentrated on their ﬁxations, being less adaptive eﬃcient in changing environment, such as curve negotiating, where the perceptual load is drastically increased and the target area (the road), changes iteratively within their visual ﬁeld. FD and FI individuals followed diﬀerent visual approaches (e.g. scan paths) to solve hidden ﬁgure problems. I individuals scanned systematically, for longer time periods at speciﬁc sectors, and had a greater overall ﬁxation count. In the contrast, FD individuals scanned unsystematically in an ”unarticulated” way with many short ﬁxations nearly all sectors during the two tests. FD individuals had longer ﬁxation duration on hypermedia areas of interest, but no statistically signiﬁcant diﬀerences were revealed.

Table 1: Reviewed eye tracking studies

who displayed a more oriented and organised scan paths. In another study, Nisiforou and Laghos [16] examined the relationship between HFT [7] scores and various eye tracking metrics. Sixteen participants undertook the HFT test, while their eye movements were recorded. The researchers studied the heat maps and scan paths of the two cognitive groups, which diﬀered greatly, and it was revealed that FD and FI individuals follow diﬀerent visual search strategies in order to identify a given pattern. FD users scanned diﬀerent points of a given shape, as they could not recognise the hidden ﬁgure. On the other hand, FI individuals could easily recognise the hidden shape, and thus their quest was more organised and their ﬁxations more precise. The researchers also found that there is a high correlation between the task performance, in terms of time, and ﬁxations on each given ﬁgure, with FD individuals being slower. They also designed an instrument based on the collected eye-tracking data, inspired by HFT, to measure the level of ﬁeld-independence. The comparison of the ﬁndings derived from their tool and HFT declared that the patterns developed using two tools were similar and that there is a correlation between the two tools. However, no further analysis was performed to measure the eﬀectiveness and the reliability of their tool.
In a later study, Nisiforou and Laghos [17] extended their previous study [16] by studying diﬀerences in visual search patterns between three cognitive groups: FD, FI and FM individuals. Fifty-four students were engaged in visual exploration tasks, proposed in researchers’ previous study [16]. Each time, a complex ﬁgure with embedded simple shapes was displayed to each participant, who had to identify which simple ﬁgure of a given set was correctly hidden into the complex one.Their analysis was based on the comparison of heat maps and scan paths (qualitative analysis) and the comparison of the ﬁxations and saccades (quantitative analysis). Regarding the heat maps and the scan paths, the eye movement patterns demonstrated that FDs could not identify the correct shape as they were looking at diﬀerent areas than the area containing the correct simple form. Additionally, they spent more time in ﬁxating into incorrect shapes. On the other hand, FIs could easily and quickly recognise the simple shape hidden in the complex pattern. The scan paths of the FI revealed a more oriented eye movement behaviour producing less number of ﬁxations and saccades than FDs. In particular FDs generated almost double ﬁxations than FNs and quadruple ﬁxations than FIs. These ﬁndings suggest that FD-I cognitive style aﬀects the eye movement patterns, revealing a more ineﬃcient visual search behaviour as the ﬁeld dependence level increases.
Jia et al. [8] investigated the eﬀect of FD-I style on the performance in a visual working memory task. The results showed that FD and FI individuals performed diﬀerently, especially in distraction condition. The correct response rates and contralateral delay activity (CDA) amplitudes in 2-item and 2-item-2-distractor conditions were comparable for FI participants. FD participants performed worse, and the CDA amplitude was enhanced when distractors appeared. FIs could focus their attention to cue-pointed objects even through the items had disappeared in the memory stage because they could rely on internal references, while FDs were more likely to scan both the targets and distractors without the cue because they mainly tended to use external references. The study results indicated that FIs could remove irrelevant information more eﬃciently than FDs.

3. DISCUSSION
In this paper we reviewed 12 studies which use eye tracking tools to investigate the eﬀect of Field Dependent - Independent cognitive style in various application domains, and we provided a summary of them in Table 1. The ﬁndings of the studies indicated that eye tracking provides valuable and credible information about the individual cognitive diﬀerences in visual perception tasks, including both exploration and search of information. Motivated by the need of designing better models of human performance and behaviour and the importance of understanding cognitive processes at the highest possible level of elaboration, we argue that eye tracking tools could be used as elicitation mechanisms, replacing the tools used today, which are based either on ”pen-andpaper” techniques or computerised tests, demanding high amount of resources, such as time, cost, test subjects, etc. The elaboration of eye-tracking techniques provides a more natural environment for the identiﬁcation of individual cognitive styles, and thus tools based on eye-tracking could be used in order to implicitly classify the individuals based on their cognitive styles, providing them with a personalised experience, both in design and run-time stages. However, only a few attempts have been made towards this direction, such as AdELE [4], EMAS [30] and ET tool [16, 17, 18], but no such framework has been proposed yet. Therefore, further research is required in order to provide an eye tracking implicit classiﬁcation mechanism of individuals based on their cognitive preferences.
4. REFERENCES
[1] C. W. Allinson and J. Hayes. The cognitive style index: A measure of intuition-analysis for organizational research. Journal of Management studies, 33(1):119–135, 1996. doi:10.1111/j.1467-6486.1996.tb00801.x.
[2] C. Angeli, N. Valanides, and P. Kirschner. Field dependence–independence and instructional–design eﬀects on learners’ performance with a computer–modeling tool. Computers in Human Behavior, 25(6):1355–1366, 2009. doi:10.1016/j.chb.2009.05.010.
[3] L. J. Baron. Interaction between television and child-related characteristics as demonstrated by eye movement research. ECTJ, 28(4):267–281, 1980. doi:10.1007/BF02766990.
[4] V. M. G. Barrios, C. Gu¨tl, A. M. Preis, K. Andrews, M. Pivec, F. M¨odritscher, and C. Trummer. Adele: A framework for adaptive e-learning through eye tracking. Proceedings of IKNOW, pages 609–616, 2004.
[5] D. M. Chapman and J. G. Calhoun. Validation of learning style measures: implications for medical education practice. Medical education, 40(6):576–583, 2006. doi:10.1111/j.1365-2929.2006.02476.x.
[6] R. Conklin, W. Muir, and F. Boersma. Field dependency-independency and eye-movement patterns. Perceptual and motor skills, 26(1):59, 1968. doi:10.2466/pms.1968.26.1.59.
[7] J. W. French, R. B. Ekstrom, and L. A. Price. Kit of reference tests for cognitive factors. Educational Testing Service, 1963.
[8] S. Jia, Q. Zhang, and S. Li. Field dependence-independence modulates the eﬃciency of

ﬁltering out irrelevant information in a visual working memory task. Neuroscience, 278:136 – 143, 2014. doi:10.1016/j.neuroscience.2014.07.075. [9] C. Katsini, C. Fidas, M. Belk, N. Avouris, and G. Samaras. The interplay among human cognition, interaction device and user authentication: An eye tracking study. 2016.
[10] M. J. Kirton and S. M. De Ciantis. Cognitive style and personality: The kirton adaption-innovation and cattell’s sixteen personality factor inventories. Personality and Individual Diﬀerences, 7(2):141–146, 1986. doi:10.1016/0191-8869(86)90048-6.
[11] M. Larsson, N. L. Pedersen, and H. Stattin. Associations between iris characteristics and personality in adulthood. Biological Psychology, 75(2):165 – 175, 2007. doi:10.1016/j.biopsycho.2007.01.007.
[12] M. Liu and W. M. Reed. The relationship between the learning strategies and learning styles in a hypermedia environment. Computers in Human Behavior, 10(4):419–434, 1995. doi:10.1016/0747-5632(94)90038-8.
[13] N. H. Mackworth and J. S. Bruner. Selecting visual information during recognition by adults and children. Harvard center for cognitive studies, 1966.
[14] F. Mawad, M. Tr´ıas, A. Gim´enez, A. Maiche, and G. Ares. Inﬂuence of cognitive style on information processing and selection of yogurt labels: Insights from an eye-tracking study. Food Research International, 74:1–9, 2015. doi:10.1016/j.foodres.2015.04.023.
[15] S. Nevalainen and J. Sajaniemi. Short-term eﬀects of graphical versus textual visualisation of variables on program perception. 17th Annual Psychology of Programming Interest Group Worskhop, pages 77–91, 2005.
[16] E. A. Nisiforou and A. Laghos. Do the eyes have it? using eye tracking to assess students cognitive dimensions. Educational Media International, 50(4):247–265, 2013. doi:10.1080/09523987.2013.862363.
[17] E. A. Nisiforou and A. Laghos. Field dependence–independence and eye movement patterns: Investigating usersaˆA˘ Z´ diﬀerences through an eye tracking study. Interacting with Computers, page iwv015, 2015. doi:10.1093/iwc/iwv015.
[18] E. A. Nisiforou, E. Michailidou, and A. Laghos. Using eye tracking to understand the impact of cognitive abilities on search tasks. In International Conference on Universal Access in Human-Computer Interaction, pages 46–57, Heraklion, Greece, 2014. Springer. doi:10.1007/978-3-319-07509-9_5.
[19] P. K. Oltman, E. Raskin, and H. A. Witkin. Group embedded ﬁgures test. Consulting Psychologists Press Palo Alto, CA, 1971.
[20] J. Pearce, S. Chang, B. Alzougool, G. Kennedy, M. Ainley, and S. Rodrigues. Search or explore: do you know what you’re looking for? In Proceedings of the 23rd Australian Computer-Human Interaction Conference, pages 253–256. ACM, 2011. doi:10.1145/2071536.2071576.
[21] G. Raptis, C. Fidas, and N. Avouris. Diﬀerences of ﬁeld dependent/independent gamers on cultural

heritage playing: Preliminary ﬁndings of an eye-tracking study. In Proceedings of International Conference on Digital Heritage (Euromed2016), Nicosia, Cyprus, 2016. Springer. doi:10.1007/978-3-319-48974-2_22. [22] M. Raschke, T. Blascheck, and M. Burch. Visual analysis of eye tracking data. In Handbook of Human Centric Visualization, pages 391–409. Springer, 2014. doi:10.1007/978-1-4614-7485-2_15. [23] R. Riding and I. Cheema. Cognitive styles - an overview and integration. Educational psychology, 11(3-4):193–215, 1991. doi:10.1080/0144341910110301. [24] D. Shinar, E. D. McDowell, N. J. Rackoﬀ, and T. H. Rockwell. Field dependence and driver visual search behavior. Human Factors: The Journal of the Human Factors and Ergonomics Society, 20(5):553–559, 1978. doi:10.1177/001872087802000505. [25] S. Somyu¨rek, T. Gu¨yer, and B. Atasoy. The eﬀects of individual diﬀerences on learner’s navigation in a courseware. TOJET: The Turkish Online Journal of Educational Technology, 7(2), 2008.
[26] R. J. Sternberg and E. L. Grigorenko. Are cognitive styles still in style? American psychologist, 52(7):700–712, 1997. doi:10.1037/0003-066X.52.7.700.
[27] D. Toker, C. Conati, B. Steichen, and G. Carenini. Individual user characteristics and information visualization: connecting the dots through eye tracking. In proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pages 295–304, Paris, France, 2013. ACM. doi:10.1145/2470654.2470696.
[28] C. Tominski. Event based visualization for user centered visual analysis. PhD thesis, University of Rostock, 2006.
[29] A. M. Treisman and G. Gelade. A feature-integration theory of attention. Cognitive Psychology, 12(1):97 – 136, 1980. doi:10.1016/0010-0285(80)90005-5.
[30] J. L. C. Wijnen and C. J. Groot. An eye movement analysis system (emas) for the identiﬁcation of cognitive processes on ﬁgural tasks. Behavior Research Methods, Instruments, & Computers, 16(3):277–281, 1984. doi:10.3758/BF03202402.
[31] H. A. Witkin. A manual for the embedded ﬁgures tests. Consulting Psychologists Press, 1971.
[32] H. A. Witkin, C. A. Moore, D. R. Goodenough, and P. W. Cox. Field-dependent and ﬁeld-independent cognitive styles and their educational implications. ETS Research Bulletin Series, 1975(2):1–64, 1975. doi:10.1002/j.2333-8504.1975.tb01065.x.
[33] J. Wolfe and T. S. Horowitz. Visual search. Scholarpedia, 3(7):3325, 2008. doi:10.4249/scholarpedia.3325.
[34] E. Yecan and K. Cagiltay. Cognitive styles and students’ interaction with an instructional web-site: Tracing users through eye-gaze. In Sixth IEEE International Conference on Advanced Learning Technologies (ICALT’06), pages 340–342, Kerkrade, The Netherlands, 2006. IEEE. doi:10.1109/ICALT.2006.1652438.