Explorative and Exploitative Learning Affected by Extraneous Cognitive Load and Gameplay Anxiety in a Gestalt Perception Game

Abstract

In line with the attention-to-affect model, this study employed a game that encompasses Gestalt perception to explore how extraneous cognitive load (ECL) and gameplay anxiety correlate with attitude towards exploitative learning and attitude towards explorative learning as students play the Visual-Saliency game (VSG) with images of artworks. The data of 56 sixth-grade students were collected and subjected to Smart PLS analysis to verify the pathways. The results of this study show that ECL is negatively related to attitude towards exploitative learning and attitude towards explorative learning. However, gameplay anxiety is negatively related to attitude towards exploitative learning, but is not significantly related to explorative learning when playing the VSG. The implication of this study is that Gestalt perception in game playing can give students more opportunities to practice reasoning, and they can gain more knowledge about the artworks themselves.

Keywords

exploitative and explorative learning extraneous cognitive load gameplay anxiety Gestalt perception visual-saliency game

The Gestalt mechanism model aims to use a formal object identification model to determine background connectivity and produce priority images (Yan et al., 2018). As Gestalt perception can portray the capabilities of visualization to yield whole forms of objects from a group of simple and even unrelated visual elements (Carlson et al., 2010), for example edges and regions, we employed the Gestalt model to design a game that explores participants’ affective and cognitive factors in relation to salient object detection. By locating a few parts to predict the whole image and designing interesting and challenging tasks (Wang et al., 2015), we designed a Gestalt perception game named the Visual-Saliency game (VSG). According to the theory of attention-to-affect (Satpute et al., 2013), all learning materials and designs will affect individual cognitive and affective states. However, Gestalt perception has not been implemented in a digital game, raising the question of how gestalt perception in the game affects players’ cognitive and affective states. This question has not been addressed in the prior literature. Thus, in this study, an attention-to-affect framework was developed to understand the capability of Gestalt perception and its learning effect on playing VSG.

This VSG proposes a new representation of the visual content of an image that allows participants to figure out what elements are part of an image and the holistic image that they form. Each image and its object annotations from a practice dataset (i.e., 500 paintings selected from the Palace Museum) were parsed to obtain the proposed visual representation. The VSG design allows participants to integrate object detection and image understanding in a single process. The holistic object detection is not agnostic to the overall content of the image, but is influenced by the image composition and the discovered parts (Vaca-Castano et al., 2019). These images and their parsed object representations are stored in long-term memory. That is, during play time, from some parts of an image, participants are able to recall the most prominent features of objects with a proper understanding of the image content through the output of the learning efforts (Inzlicht et al., 2018). Two types of learning attitude can be employed to elaborate the learning efforts: exploratory and exploitative learning. Exploratory learning involves finding various elements of the repertoire and combining them into problem-solving solutions (Li et al., 2012) that are related to applying or refining knowledge similar to the existing process (Zahra & George, 2002). Exploitative learning would lead players to practice epistemic action to achieve their goals (Barzilai, 2017). However, exploratory learning is a potentially productive site for exploring details of knowledge or problems (Gee, 2007). On the other hand, exploitative learning can be enhanced by the aim of achieving the performance goal (Chinn & Rinehart, 2016). How two types of attitude toward learning are affected by game playing has not been studied; thus, this study explored them in relation to participants playing VSG.

The VSG design involves players seeing parts of whole images, beginning with three pieces, but increasing up to 21 of a total of 25 pieces. It requires the players’ cognitive efforts related to exploratory and exploitative learning to practice their Gestalt perception. Moreover, the time meant for playing the VSG was limited to 30 seconds for each try. The players might have difficulty at the beginning, and this will evoke their psychological state differently, including the levels of cognitive and affective states, such as cognitive load and gameplay anxiety, which affected the learning attitudes of the participants in this study. There are many studies focused on cognitive load and gameplay anxiety (e.g., Hong et al., 2016a; Kozhevnikov et al., 2018), but little research has been done with a focus on the cognitive and affective effect on Gestalt perception games; thus, the present study examined how participants’ cognitive load and gameplay anxiety affect attitudes toward explorative and exploitative learning when playing the VSG.

Cognitive Load in the Context of Gameplay

Cognitive load theory explains that there is a certain amount of information that can be processed in the working memory at one time without overloading its processing capacity. Thus, when cognitive load is increased beyond our working memory capacity, learning is reduced (Paas et al., 2003). However, cognitive load consists of three types: extraneous, intrinsic, and germane (Sweller, 2010). Extraneous cognitive load (ECL) is generated by the manner in which information is presented to learners and is under the control of instructional design. Intrinsic cognitive load (ICL) is the inherent level of difficulty associated with a specific instructional topic, such that information that is hard for the learners to interpret increases cognitive load. Germane cognitive load (GCL) refers to load that is generated by instructional activities (Mayer, 2005). Regarding the interaction between cognitive ability and task difficulty, ECL has been considered as the main factor affecting visual perception (Polatsek et al., 2018). Moreover, while recent saliency models have been used to predict users’ visual attention in a task of visualization in order to explore objects, there is little information to determine how much influence ECL has on task-based visual analysis (Polatsek et al., 2018). Thus, this study took ECL as the cognitive factor that affects learning while playing VSG.

Gameplay Anxiety in the Context of Playing VSG

Anxiety has been recognized as a cognitive response to a threat to one’s self-concept, by his or her subjective, consciously perceived feelings of tension (Spielberger, 1966). The state-trait anxiety theory was introduced by Spielberger et al. (1970), who described state anxiety as a “here and now” transitory feeling of tension that varies in intensity and fluctuates over time. In sporting competitions, individuals’ competitive trait anxiety directly affects their perception of threat, subsequently mediating the level of state anxiety experienced (Martens et al., 1990). Given that competing perspectives affect one’s state of anxiety (Mellalieu et al., 2004), in this study a dual competition mode was applied for participants to play against each other; this type of anxiety can be considered as competitive state anxiety. Additionally, if stimuli can create a local discontinuity in the visual environment, anxiety will be evoked (Sussman et al., 2016). The VSG particularly encompasses local discontinuity (i.e., it only provides three separate pieces of a painting for players to identify the painting at the beginning of the VSG) between pieces of paintings, indicating that gameplay anxiety as a psychological state might be raised while playing VSG. However, in game-based learning, the game activity must balance with players’ ability so that they can maintain the challenge experience. If the challenge is beyond that ability, the activity becomes so overwhelming that it generates gameplay anxiety (Hwang et al., 2013). Thus, this study explored gameplay anxiety as competitive state anxiety to understand participants’ feeling of tension when they played VSG.

Explorative and Exploitative Learning in the Context of Playing VSG

When playing the VSG, players are involved in a highly competitive setting. In attempting to win the game, they have to comprehend the nature or related knowledge of paintings, which would support them in finding the answers in the competition, i.e., the VSG does not ‘‘tell” players what they need to know. Instead, players need to engage in explorative learning to support their game playing (Li et al., 2012). In line with playing VSG, as exploratory learning refers to searching for or acquiring external knowledge that is entirely new to the existing process (Li et al., 2012), if students pay more attention to learning external information, they can process it more deeply and, thus, comprehend it better (Liu, 2014; Tsai et al., 2012). Previous studies seem to have confirmed this potential of digital games to support students' explorative learning. For example, most teachers who use games to promote explorative learning in class positively perceive students' cognitive outcomes (Huizenga et al., 2017).

For a number of years now, researchers have dedicated considerable effort to further the understanding of exploratory versus exploitative learning (Li et al., 2012). According to the dynamic capability view, mental agility represents the dynamic capability to integrate disconnected information into a pack of knowledge (Li & Huang, 2013). When playing the VSG, players will receive three out of 25 disconnected pieces of a painting and are required to identify the painting. In order to carry out this task while playing the game, a certain kind of mental agility is needed.

Moreover, a few studies in the exploratory and exploitative learning literature have examined attention-to-affect factors (Satpute et al., 2013) in a VSG involving the practice of Gestalt perception so as to understand students’ perceptions of meaningful learning. The two types of learning attitudes may need to differentiate the antecedent conditions of extraneous cognitive load and gameplay anxiety as the level of attention-to-affect. Thus, attitudes towards explorative and exploitative learning with Gestalt perception were explored in this study by having students play the VSG.

Hypotheses

The Visual-Saliency game requires players to have the ability to rapidly detect and respond to stimuli within the surrounding environment and to effectively reason visual information (Jian et al., 2018). For example, Steinkuehler and Duncan (2008) found that players often engage in practices such as scientific argumentation and model-based reasoning in order to develop and justify arguments about the game model and about reasoning strategies for playing. With reference to online games, attitude towards exploitative learning is defined as the extent to which playing online games is perceived to provide learning experiences (Griffiths et al., 2004). Individuals should be motivated to learn to think analytically and to develop cogent analysis of the relationship between whatever set of events they are interested in (Greene & Yu, 2014).

According to the model of attention and affect in learning (Moreno, 2006), cognitive effort can be defined as the level of perceived difficulty of the cognitive task and the subsequent cognitive effort. Some studies have explored the level of difficulty of a given task for the purpose of investigating the correlation between the level of cognitive effort in order to understand the effectiveness of efforts (Silk et al., 2010). In addition, when an individual experiences gameplay anxiety, the cognitive responses may lead to negative feelings and cognitions about competitive situations (Hong et al., 2017). Given the increasing recognition of the role of effort in emotion, cognitive control affects value-based decision-making, considering that this effort will provide clues about how to promote sustained cognitive effort across time (Inzlicht et al., 2018).

While playing a game, the complex representations and time-compression cause an increase in cognitive load in the working memory, thereby decreasing learning motivation and effect (Mayer et al., 2001). Pastore (2012) examined the effects of the redundancy principle within a time-compressed environment on multimedia learning and found that cognitive load resulted from time-compression and had an influence on learning. However, the effects of time-compression on participants’ learning emotion or attitude have not been studied in depth. Thus, the VSG design, which presents parts of images for players to identify the whole image within a limited amount of time, was expected to have an effect on the players’ gameplay anxiety and ECL.

Given that the two types of learning involve complex information-processing tasks, we highlight two attention-to-affect factors, ECL and gameplay anxiety, which affect the information-processing capacity of playing the VSG. Therefore, the hypotheses of this study were proposed as follows:
H1: ECL is negatively related to attitude towards exploitative learning

H2: ECL is negatively related to attitude towards explorative learning

H3: Gameplay anxiety is negatively related to attitude towards exploitative learning

H4: Gameplay anxiety is negatively related to attitude towards explorative learning

Previous researchers have found that the factors that interact with the learning outcomes of gameplay are cognition, affection, and behavior (Kiili, 2005). This is of critical importance in the study of digital game-based learning because of the linkage between attention-to-affect and learning orientation (Slavin, 2011). As cognitive state will influence affective state during game playing (Kozhevnikov et al., 2018), cognitive load is positively related to competitive anxiety (Hwang et al., 2013); by excluding these relationships, the present study aims to explore how ECL and gameplay anxiety correlate with attitudes towards exploitative and explorative learning. The research model is therefore proposed as follows (see Figure 1).

Figure 1.
Research Model.

Game Design

Visual attention and perception play a critical role in learning (Polatsek et al., 2018). Accordingly, the VSG requires players to recall their prior knowledge, pay visual attention, and use Gestalt perception to respond to the questions that appear on the screen.

Game Features

In the VSG design, visual searching is an important component of the process of interpreting visualizations. It is the process of finding a specific target image in a scene from a few segments of that image. Considering the features of the VSG, players are required to identify a whole image by looking at three out of 25 pieces of the image (i.e., the three pieces that randomly appear in separate locations on the screen and cannot be controlled by players) at the beginning of the VSG. If they are unable to predict the whole image on their first trial, they will then be shown five, eight, 13, and 21 of the 25 pieces of the whole image respectively, but their score will decrease accordingly. The website of the National Palace Museum, which contains more than 10,000 paintings, allowed us to use their paintings for free. Therefore, to ensure effective practice with the attitude toward exploitative and explorative learning by players, they were asked to learn about the details of 500 paintings, i.e., they had to go to the website of the National Palace Museum to download the paintings before they joined the competition so as to practice exploitative and explorative learning to familiarize themselves with the traits of the paintings.

How to Play

Each of the artworks is divided into 25 segments. At the beginning of the VSG, three of the 25 segments are randomly selected and displayed. The player uses Gestalt perception to guess the whole image and choose the correct option from the six choices, the names of paintings, shown at the bottom of the screen (Figure 2). The functional bag in the bottom-left corner can be used to eliminate the displayed answers, reducing them from six to four and further from four to two by paying coins. Figure 2 shows that a player has used the functional bag once, so the answers were reduced from six to four. If the player cannot answer the question, he/she can use the magnifier which is located in the bottom-right corner to show more parts of the segments (i.e., the function of the magnifier is to increase the number of segments displayed to 5, 8, 13 and 21). A screenshot of the game is shown in Figure 3, and the answer is shown in Figure 4.

Figure 2.
A Sample Question From the Game (at the Beginning of the Game, Only 3 of the 25 segments Are Shown for Players to Select an Answer [the Name of the Painting or the Style of Painting] From the Options at the Bottom of the Screen).

Figure 3.
An Example of Using the Magnifier to Display Five Segments Which Will Randomly Appear on the Screen.

Figure 4.
The Whole Image Is Presented.

Considering the features of the VSG, players are required to identify whole images by looking at three out of 25 pieces of the image (i.e., those three that randomly appear on the screen) at the beginning of gaming. It is possible for two of the three pieces to be merged at random, but the same screen will appear at each side of the players’ smartphone. Dual-player mode is to control the equality of the gameplay. In gaming terms, when competition is introduced into a game, some individuals may try to “win”; specifically, if competition spurs individuals to want to defeat others, it may affect their individual cognition-affection and performance (Landers et al., 2019); thus, in this study, dual competition mode was applied for participants to play against each other. Moreover, according to the difficulty level, the game gives the player a maximum of 30 seconds to answer each question. If the player gets the answer right, he/she is awarded coins. Totally, 20 paintings are subjected to play (i.e., identifying the painting name or style of that piece) in each tournament round, and half of the teams are knocked out in each tournament. Three tournament rounds were completed to determine the final winning team in a day.

Method

Procedure

The VSG has two modes, the single-player mode and the dual-player mode. The dual-player mode can involve at most six players divided into two teams. The winner is the team with the highest score. An elimination tournament was implemented in this study. Four sixth-grade elementary schools were purposively sampled to compete on the same day (with a total of 16 students from each of the four schools, who were selected to represent their schools). One week after the competition, we administered 64 questionnaires to the participants. In the preface of the questionnaire, participants were informed that they were taking part in an evaluation study, that the data they provided was anonymous and that the study might be published. After the data were returned, eight incomplete questionnaires were deleted and 56 were counted as useful returns.

Participants

The demographic information of the participants was as follows: 31(55.4%) were male and 25 (44.6%) female; the average age was 11.5 years (sixth-grade students), and regarding their learning domains of interest, 40 (71.4%) chose humanities (i.e., arts, Chinese and history), and the other 16 (28.6%) chose non-humanities (i.e., science and math).

Research Instrument

The questionnaire items were based on those of previous studies and included four constructs: ECL, gameplay anxiety, attitude towards exploitative learning, and attitude towards explorative learning. The original English items were translated into Mandarin using the forward-backward translation method. This meant that the accuracy and clarity could be verified to ensure the face validity of the translated questionnaire. The questionnaire adopted a 5-point Likert scale anchored at 1 (Disagree strongly) and 5 (Agree strongly) as shown in Table 1. This study used the confirmatory analysis method through PLS to examine the reliability and validity of the collected data for both questionnaire items and constructs. The final version of the questionnaire can be seen below in Table 1.

Table 1.
Reliability and Validity Analysis.

Items M SD FL t value

Extraneous cognitive load M = 2.630, SD = 1.042, Cronbach’s α = .908, CR = .935, AVE = .784

1. With only 30 seconds for each trial in the game, I felt that it was very difficult to keep pace with it. 2.625 1.229 .924 86.306

2. With only 3 out of 25 pieces of the painting displayed at the beginning, it was difficult for me to play. 2.500 1.206 .860 11.189

3. It was difficult for me to associate those separated pieces that popped up with the answer in this game. 2.714 1.074 .903 33.732

4. I had to pay a lot attention to play this game. 2.679 1.193 .852 10.664

Gameplay anxietyM = 2.554, SD = .756, Cronbach’s α = .864, CR = .898, AVE = .689

1. During the competition, I was so nervous that I forgot the answer even if I knew it. 2.714 .889 .881 36.488

2. During the competition, I was easily nervous and found it hard to think of the answer. 2.518 .894 .774 8.223

3. During the competition, I worried that others would perform better than me, so it distracted my concentration while playing. 2.464 .852 .852 12.543

4. During the competition, if I encountered those paintings that were not so familiar to me, I would be at a loss. 2.518 .953 .808 7.502

Attitude towards exploitative learningM = 3.884, SD = .841, Cronbach’s α = .942, CR = .958, AVE = .851

1. For the competition, I would like to spend more time analyzing the artworks in detail. 3.875 .896 .948 72.783

2. For the competition, I would like to reason things from parts to the whole. 3.839 .910 .919 33.954

3. For the competition, I often practiced top-down saliency analysis (e.g., color, location). 3.929 .850 .897 30.447

4. For the competition, I often differentiated the details of different paintings. 3.893 .985 .926 42.118

Attitude towards exploratory learningM = 3.795, SD = .820, Cronbach’s α = .918, CR = .942, AVE = .802

1. For the competition, I would explore painting styles in art history. 3.804 .942 .885 37.931

2. For the competition, I would explore more painting skills than before. 3.821 .897 .892 27.390

3. For the competition, I explored why painters draw particular paintings. 3.786 .929 .911 28.219

4. For the competition, I would explore what skills were used in that painting. 3.768 .894 .894 22.437

Measurement

Extraneous cognitive load: Items were adapted from Paas et al. (2003) to measure ECL in terms of invested cognitive effort, which represents the aspects of perceived level of difficulty, in order to access the specific game design that requires cognitive demands on working memory. The ECL experienced was measured using eight items.

Attitude towards exploitative learning: Exploitative learning relates to reasoning or refining knowledge (Zahra & George, 2002). Epistemic action can form knowledge directly by executing sensing actions (Eppe & Bhatt, 2015) that involve learners’ relatively complicated mental activities to justify knowledge, such as analyzing, reflecting, annotating, elaborating, linking, and evaluating (Hong et al., 2016b). Accordingly, those items that are related to attitude towards exploitative learning were considered as the attitude towards exploitative learning when playing the VSG.

Attitude towards explorative learning: Explorative learning refers to searching for or acquiring external knowledge in a process (Li et al., 2012). Accordingly, in this study, we adopted McGill and Hobbs’ (2008) definition and highlighted that the attitude towards learning can be extended to the degree to which users perceive that there is learning, including perceived effectiveness and searching for helpful information. Thus, these components of learning attitude in relation to the VSG were considered in regard to attitude towards explorative learning.

Gameplay anxiety: Competitive state anxiety could be a multidimensional concept, possessing at least cognitive and somatic components (Martens et al., 1990). In order to understand how participants responded to their state anxiety after the competition, Hwang et al. (2013) took cognitive anxiety as competitive state anxiety to assess the intensity of anxiety (characterized by negative expectancies and self-doubts) excluding somatic anxiety. In other words, those items related to negative expectancies and self-doubts were taken as gameplay anxiety in this study.

Item Suitability Analysis

Originally, the questionnaires consisted of eight items for ECL, six for gameplay anxiety, six for attitude towards exploitative learning, and seven for attitude towards explorative learning. To examine the internal validity of the items, the first-order confirmatory factor analysis introduced by Kline (2010) was applied, and items with residual values greater than 0.5 were removed. The final questionnaire retained four items each for ECL, gameplay anxiety, and attitude towards exploitative learning. To examine the external validity of the items, the t-values were significantly higher than 3.0, indicating that all items were discriminative (see Table 1) (Green & Salkind, 2004).

Reliability and Validity Analysis

According to Hair et al. (2010), a Cronbach's α value above 0.6 indicates an acceptable level of internal reliability. Table 1 presents that the Cronbach's α values were higher than 0.864, indicating that internal reliability was achieved. All composite reliability (CR) values in Table 1 surpassed the suggested threshold value of 0.7 (Hair et al., 2010), indicating that CR was achieved.

Convergent validity was evaluated by checking the following conditions: (1) whether the average variance extracted (AVE) values were larger than 0.6, and (2) if the factor loadings of all items were significant and higher than 0.6 (Hair et al., 2010). Table 1 shows that all of the factor loadings (FL) were higher than 0.774 and all AVE values were above 0.768. All of the conditions were met, indicating acceptable convergent validity. To establish construct discriminant validity, the square root of every AVE value is applied (Hair et al., 2010). The rule states that the square root of the AVE of each construct should be significantly larger than the correlation between the specific construct and any of the other constructs. Table 2 shows that the square root of every AVE value was above the correlation coefficient with other constructs, which showed that discriminant validity existed in the research constructs.

Table 2.
Construct Discriminative Validity Analysis.

Extraneous cognitive load Gameplay anxiety Attitude towards exploitative learning Attitude towards explorative learning

Extraneous cognitive load 0.641

Gameplay anxiety .602* 0.744

Attitude towards exploitative learning .64* .548* 0.725

attitude towards explorative learning .578* .421* .82* 0.643

*** p < .001.

Results

In structural equation modeling, Smart PLS can be used to verify a research model with small samples (Kline, 2010). By applying Smart PLS, the results of this study (see Figure 5) show the following: ECL was negatively related to attitude towards exploitative learning (β = –.46, t = 4.845); ECL was negatively related to attitude towards explorative learning (β = –.506, t = 4.236); gameplay anxiety was negatively related to attitude towards exploitative learning (β = –.321, t = −3.563); gameplay anxiety was not significantly related to attitude towards explorative learning (β = –.159, t = 1.497).

Figure 5.
Verification of the Research Model.

From the above bootstrapping results, Figure 5 shows the results of the path relationships among the hypotheses, and indicates that H1 through H3 were supported, but H4 was not supported. In addition, the explanation power for ECL and gameplay anxiety on attitude towards exploitative learning was 52.2%, and the explanation power for ECL and gameplay anxiety on attitude towards explorative learning was 39.4%. The R ² values were more than the recommended threshold value of 10% proposed by Hair et al. (2010). Therefore, all variables of this research had acceptable predictive power (Hair et al., 2010).

Discussion

How the players’ ECL and gameplay anxiety were raised while they were competing against each other was explored in this study. The mean of ECL was below average (M = 2.63, SD = 1.042) on a 5-point Likert scale, indicating that the participants did not use too much cognitive effort while playing the VSG. This shows that the VSG was well designed for the target participants. The mean for gameplay anxiety was also below average (M = 2.554, SD = 0.756) on a 5-point Likert scale, revealing that the VSG did not cause the players too much anxiety. The mean of attitude towards exploitative learning was higher than average (M = 3.884, SD = .841), showing that the players perceived themselves as using deep cognitive processes to master the details of the paintings. Finally, the mean of attitude towards explorative learning was also higher than average (M = 3.795, SD= .820), indicating that the respondents recognized that they would learn more concepts related to paintings by preparing to join the competition. As Slavin (2011) argued, gameplay learning attitude is of critical importance due to the linkage between attention-to-affect and learning orientation. Our results show that the VSG was effective in terms of enhancing the participants’ two types of learning attitude.

With regards to H1 and H2, the results of this study revealed that ECL was negatively related to attitude towards exploitative and explorative learning, indicating that the lower the level of ECL, the higher the level of the two types of learning attitude. As studies have explored the correlation between the level of difficulty in a given task and cognitive effort as a necessity for understanding the effectiveness of learning (Silk et al., 2010), the results of this study indicate that while playing the VSG, the players had to connect pieces of paintings, which may be considered as needing high ambidexterity; thus, the participants’ cognitive effort in employing working memory and long-term memory from the perspective of the information processing theory was affected (Brunken et al., 2003). As such, attitude toward exploitative and explorative learning would be initiated (Li & Huang, 2013). Thus, decreasing ECL will increase the attitude towards exploitative and explorative learning while players involve Gestalt perception in playing the VSG. Thus, individuals with a high level of ECL should be motivated to engage in the two types of learning, namely thinking analytically and developing cogent analysis.

With respect to H3 and H4, the results of this study revealed that gameplay anxiety was negatively related to attitude towards exploitative learning, but was not significantly related to explorative learning, indicating that only a lower level of gameplay anxiety can predict a high level of attitude towards exploitative learning. In line with the theory of attention-to-affect (Satpute et al., 2013), if students pay more attention to learning information, they can process it more deeply, and thus comprehend it better (Tsai et al., 2012). Given the increasing recognition of the role of effort in emotion, gameplay anxiety affects the value-based decision-making to realize more content-specific knowledge, considering that gameplay anxiety influences how to learn more to enhance Gestalt perception across game playing duration (Inzlicht et al., 2018). As such, decreasing gameplay anxiety will increase the attitude towards exploitative learning when players involve Gestalt perception in playing the VSG. If players put more cognitive effort into learning to gain more knowledge about the content, their VSG performance will also improve (Liu, 2014; Tsai et al., 2012). However, this statement is supported by the influence of gameplay anxiety on exploitative learning, but not on explorative learning. This may be due to the fact that when an individual experiences gameplay anxiety, the cognitive responses may lead to a focus on competitive situations (Hong et al., 2015, 2017).

Conclusion

This study applied two attitudes toward learning in a micro game-playing world; the VSG was designed to incorporate Gestalt perception. Dual attitudes toward learning were explored in this study, and the results showed that ECL can be considered as the level of perceived difficulty of the task and the subsequent gameplay anxiety that affects the extent to which the individual achieves the cognitive aim. The results showed that participants’ two attitudes toward learning can be promoted, and provide an understanding of game design and application: embedding some kind of mechanism similar to the VSG may benefit students by giving them materials that fit learners' cognitive development and clarify the learning topics of VSG.

Practical Implication

Previous studies have confirmed the potential of digital games to support students' learning, and most teachers who use games in class positively perceive students’ cognitive outcomes (Huizenga et al., 2017). This study divided students’ learning attitudes into attitudes towards exploitative and explorative learning in relation to saliency-based learning, and the result is supported by Huizenga et al.’s (2017) study. According to the result of this study, teachers may consider using similar mobile devices by adding or placing content that can be picturized in this type of VSG to enhance students’ ability in relation to Gestalt perception.

Understanding visual attention is essential for designing effective visualizations and appropriate visual practice. Based on the results of this study, one implication is that educational game application designers may take the VSG as an example to implement different types of content that contains or can be picturized as domain-knowledge learning other than Chinese paintings, such as landscapes, microorganisms, and so on.

Limitation and Future Study

This study used a self-rating questionnaire to obtain an understanding of the VSG players’ cognitive and affective states. However, it is difficult for some participants to retrieve their affective state when they were playing VSG. Future studies may include some biotechnology, e.g., the level of anxiety that has been evoked in participants may be detected by some devices, such as thermometers, or eye tracking may be used to understand the level of ECL that participants experience during gameplay. Besides, future studies could also analyze the differences in participants’ background, such as their domain interest.

The layout of multi-dimensional visualizations is a crucial concern for the Gestalt cues in effective visual analytic interfaces. If one considers the accuracy of measurement and gender as a between-subject distinction, males have been found to perform better than females in two Gestalt cues: proximity and connectedness (Polys et al., 2011). However, the gender factor was not considered in this study; future studies may analyze the difference between male and female students in playing saliency-based games.

Kozhevnikov et al. (2018) suggested that cognitive state will affect affective state during game playing, and Hwang, et al. (2013) found that cognitive load is positively related to competitive anxiety. However, their studies focused on a single player mode game, whereas no such study has considered the state interaction in competitive games; therefore, future studies may explore the correlation between cognitive load and competitive anxiety when playing a dual player mode game such as VSG.

Items	M	SD	FL	t value
Extraneous cognitive load M = 2.630, SD = 1.042, Cronbach’s α = .908, CR = .935, AVE = .784
1. With only 30 seconds for each trial in the game, I felt that it was very difficult to keep pace with it.	2.625	1.229	.924	86.306
2. With only 3 out of 25 pieces of the painting displayed at the beginning, it was difficult for me to play.	2.500	1.206	.860	11.189
3. It was difficult for me to associate those separated pieces that popped up with the answer in this game.	2.714	1.074	.903	33.732
4. I had to pay a lot attention to play this game.	2.679	1.193	.852	10.664
Gameplay anxietyM = 2.554, SD = .756, Cronbach’s α = .864, CR = .898, AVE = .689
1. During the competition, I was so nervous that I forgot the answer even if I knew it.	2.714	.889	.881	36.488
2. During the competition, I was easily nervous and found it hard to think of the answer.	2.518	.894	.774	8.223
3. During the competition, I worried that others would perform better than me, so it distracted my concentration while playing.	2.464	.852	.852	12.543
4. During the competition, if I encountered those paintings that were not so familiar to me, I would be at a loss.	2.518	.953	.808	7.502
Attitude towards exploitative learningM = 3.884, SD = .841, Cronbach’s α = .942, CR = .958, AVE = .851
1. For the competition, I would like to spend more time analyzing the artworks in detail.	3.875	.896	.948	72.783
2. For the competition, I would like to reason things from parts to the whole.	3.839	.910	.919	33.954
3. For the competition, I often practiced top-down saliency analysis (e.g., color, location).	3.929	.850	.897	30.447
4. For the competition, I often differentiated the details of different paintings.	3.893	.985	.926	42.118
Attitude towards exploratory learningM = 3.795, SD = .820, Cronbach’s α = .918, CR = .942, AVE = .802
1. For the competition, I would explore painting styles in art history.	3.804	.942	.885	37.931
2. For the competition, I would explore more painting skills than before.	3.821	.897	.892	27.390
3. For the competition, I explored why painters draw particular paintings.	3.786	.929	.911	28.219
4. For the competition, I would explore what skills were used in that painting.	3.768	.894	.894	22.437

	Extraneous cognitive load	Gameplay anxiety	Attitude towards exploitative learning	Attitude towards explorative learning
Extraneous cognitive load	0.641
Gameplay anxiety	.602***	0.744
Attitude towards exploitative learning	.64***	.548***	0.725
attitude towards explorative learning	.578***	.421***	.82***	0.643

Footnotes

Acknowledgment

This work was financially supported by the “Institute for Research Excellence in Learning Sciences” of National Taiwan Normal University (NTNU) from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Jon-Chao Hong is a professor in the Department of Industrial Education and a member of Institute for Research Excellence in Learning Sciences at National Taiwan Normal University. His research interest includes game-based learning.

Ming-Yueh Hwang is a professor in the Department of Adult and Continuing Education and a member of Institute for Research Excellence in Learning Sciences at National Taiwan Normal University.

Mei-Syuan Chen graduated from the Department of Industrial Education in National Taiwan Normal University with a Master’s degree.

Kai-Hsin Tai is a postdoctoral fellow in the Department of Industrial Education and a member of Institute for Research Excellence in Learning Sciences at National Taiwan Normal University. Her research interest is about using new technology enhance learning.

ORCID iDs

Jon-Chao Hong

Kai-Hsin Tai

References

Barzilai

(2017). “ Half-reliable”: A qualitative analysis of epistemic thinking in and about a digital game. Contemporary Educational Psychology, 51, 51–66. https://doi.org/10.1016/j.cedpsych.2017.06.004

Brunken, R., Plass, J. L., & Leutner, D. (2010). Direct measurement of cognitive load in multimedia learning. Educational Psychologist, 38(1), 53–61. https://doi.org/10.1207/S15326985EP3801_7

Carlson

N. R.

Miller

Donahoe

J. W.

Martin

G. N.

(2010). Psychology: The science of behavior. Pearson Education.

Chinn

C. A.

Rinehart

R. W.

(2016). Epistemic cognition and philosophy: Developing a new framework for epistemic cognition. In Greene

J. A.

Sandoval

W. A.

Bråten

(Eds.), Handbook of epistemic cognition (pp. 460–478). Routledge.

Eppe

Bhatt

(2015). A history based approximate epistemic action theory for efficient postdictive reasoning. Journal of Applied Logic, 13(4), 720–769. https://doi.org/10.1016/j.jal.2015.08.001

Gee

J. P.

(2007). What video games have to teach us about learning and literacy. Palgrave Macmillan.

Green

S. B.

Salkind

(2004). Using SPSS for Windows and Macintosh: Analyzing and understanding data (4th ed.). Prentice Hall.

Greene

J. A.

S. B.

(2014). Modeling and measuring epistemic cognition: A qualitative re-investigation. Contemporary Educational Psychology, 39(1), 12–28. https://doi.org/10.1016/j.cedpsych.2013.10.002

Griffiths, M. D., Davies, M. N. O., & Chappell, D. (2004). Online computer gaming: A comparison of adolescent and adult gamers. Journal of Adolescence, 27(1), 87–96. https://doi.org/10.1016/j.adolescence.2003.10.007

10.

Hair

J. F.

Jr. Black

W. C.

Babin

B. J.

Anderson

R. E.

(2010). Multivariate data analysis: A global perspective (7th ed.). Prentice Hall.

11.

Hong

J. C.

Hwang

M. Y.

Tai

K. H.

Kuo

Y. C.

(2016b). Crystallized intelligence affects hedonic and epistemic values to continue playing a game with saliency-based design. Computers & Education, 95, 75–84.

12.

Hong

J. C.

Lin

M. P.

Hwang

M. Y.

Tai

K. H.

Kuo

Y. C.

(2015). Comparing animated and static modes in educational gameplay on user interest, performance and gameplay anxiety. Computers & Education, 88, 109–118.

13.

Hong

J. C.

Tai

K. H.

Hwang

M. Y.

Kuo

Y. C.

(2016a). Internet cognitive failure affects learning progress as mediated by cognitive anxiety and flow while playing a Chinese antonym synonym game with interacting verbal-analytical and motor-control. Computers & Education, 100, 32–44.

14.

Hong

J.-C.

Hwang

M.-Y.

Tai

K.-H.

Kuo

Y.-C.

Lin

P.-C.

(2017). Confusion affects gameplay. Learning and Individual Differences, 59, 119–126.

15.

Huizenga

J. C.

ten Dam

G. T. M.

Voogt

J. M.

Admiraal

W. F.

(2017). Teacher perceptions of the value of game-based learning in secondary education. Computers & Education, 110, 105–115. https://doi.org/10.1016/j.compedu.2017.03.008

16.

Hwang

M. Y.

Hong

J. C.

Cheng

H. Y.

Peng

Y. C.

N. C.

(2013). Gender differences in cognitive load and competition anxiety affect 6th grade students’ attitude toward playing and intention to play at a sequential or synchronous game. Computers & Education, 60(1), 254–263.

17.

Inzlicht

Shenhav

Olivola

C. Y.

(2018). The effort paradox: Effort is both costly and valued. Trends in Cognitive Sciences, 22(4), 337–349. https://doi.org/10.1016/j.tics.2018.01.007

18.

Jian

Jung

Jia

(2018). Visual saliency estimation using constraints. Neurocomputing, 290, 1–11. https://doi.org/10.1016/j.neucom.2018.02.004

19.

Kiili

(2005). Digital game-based learning: Towards an experiential gaming model. The Internet and Higher Education, 8(1), 13–24. https://doi.org/10.1016/j.iheduc.2004.12.001

20.

Kline

R. B.

(2010). Principles and practice of structural equation modeling (3rd ed.). New York: Guilford Press.

21.

Kozhevnikov

Wong

Obana

Amihai

(2018). Do enhanced states exist? Boosting cognitive capacities through an action video-game. Cognition, 173, 93–105. https://doi.org/10.1016/j.cognition.2018.01.006

22.

Landers

R. N.

Collmus

A. B.

Williams

(2019). The greatest battle is within ourselves: An experiment on the effects of competition alone on task performance. International Journal of Human-Computer Studies, 127, 51–61. https://doi.org/10.1016/j.ijhcs.2018.09.011

23.

C. R.

Chu

C. P.

Lin

C. J.

(2010). The contingent value of exploratory and exploitative learning for new product development performance. Industrial Marketing Management, 39(7), 1186–1197. https://doi.org/10.1016/j.indmarman.2010.02.002

24.

Y. H.

Huang

J. W.

(2013). Exploitative and exploratory learning in transactive memory systems and project performance. Information & Management, 50(6), 304–313. https://doi.org/10.1016/j.im.2013.05.003

25.

Liu

P. L.

(2014). Using eye tracking to understand learners' reading process through the concept-mapping learning strategy. Computers & Education, 78, 237–249. https://doi.org/10.1016/j.compedu.2014.05.011

26.

Martens

Vealey

R. S.

Burton

(1990). Competitive anxiety in sport. Human Kinetics.

27.

Mayer, R. E. (2005). Cognitive theory of multimedia learning. Cambridge, UK: Cambridge University Press.

28.

Mayer

R. E.

Heiser

Lonn

(2001). Cognitive constraints on multimedia learning: When presenting more material results in less understanding. Journal of Educational Psychology, 93(1), 187–198. https://doi.org/10.1037/0022-0663.93.1.187

29.

McGill

T. J.

Hobbs

V. J.

(2007). How students and instructors using a virtual learning environment perceive the fit between technology and task. Journal of Computer Assisted Learning, 24(3), 191–202. https://doi.org/10.1111/j.1365-2729.2007.00253.x

30.

Mellalieu

S. D.

Hanton

O'Brien

(2004). Intensity and direction of competitive anxiety as a function of sport type and experience. Scandinavian Journal of Medicine & Science in Sports, 14(5), 326–334. https://doi.org/10.1111/j.1600-0838.2004.00389.x

31.

Moreno

(2006). Does the modality principle hold for different media? A test of the method-affects-learning hypothesis. Journal of Computer Assisted Learning, 22(3), 149–158. https://doi.org/10.1111/j.1365-2729.2006.00170.x

32.

Paas

Renkl

Sweller

(2003). Cognitive load theory and instructional design: Recent developments. Educational Psychologist, 38(1), 1–4. https://doi.org/10.1207/S15326985EP3801_1

33.

Pastore

(2012). The effects of time-compressed instruction and redundancy on learning and learners’ perceptions of cognitive load. Computers & Education, 58(1), 641–651. https://doi.org/10.1016/j.compedu.2011.09.018

34.

Polatsek

Waldner

Viola

Kapec

Benesova

(2018). Exploring visual attention and saliency modeling for task-based visual analysis. Computers & Graphics, 72, 26–38. https://doi.org/10.1016/j.cag.2018.01.010

35.

Polys

N. F.

Bowman

D. A.

North

(2011). The role of depth and gestalt cues in information-rich virtual environments. International Journal of Human-Computer Studies, 69(1-2), 30–51. https://doi.org/10.1016/j.ijhcs.2010.05.007

36.

Satpute

A. B.

Shu

Weber

Roy

Ochsner

K. N.

(2013). The functional neural architecture of self-reports of affective experience. Biological Psychiatry, 73(7), 631–638. https://doi.org/10.1016/j.biopsych.2012.10.001

37.

Silk

T. J.

Bellgrove

M. A.

Wrafter

Mattingley

J. B.

Cunnington

(2010). Spatial working memory and spatial attention rely on common neural processes in the intraparietal sulcus. NeuroImage, 53(2), 718–724. https://doi.org/10.1016/j.neuroimage.2010.06.068

38.

Slavin

R. E.

(2011). Cooperative learning: Learning and cognition in education. Elsevier Academic Press.

39.

Spielberger

C. D.

(1966). Theory and research on anxiety. In Spielberger

C. D.

(Ed.), Anxiety and behavior (pp. 3–20). Academic Press.

40.

Spielberger

C. D.

Gorsuch

R. L.

Lushene

R. E.

(1970). Manual for the anxiety inventory. Consulting Psychologists Press.

41.

Steinkuehler

Duncan

(2008). Scientific habits of mind in virtual worlds. Journal of Science Education and Technology, 17(6), 530–543. http://doi.org/10.1007/s10956-008-9120-8

42.

Sussman

T. J.

Jin

Mohant

(2016). Top-down and bottom-up factors in threat-related perception and attention in anxiety. Biological Psychology, 121(Pt B), 160–172. https://doi.org/10.1016/j.biopsycho.2016.08.006

43.

Sweller

(2010). Element interactivity and intrinsic, extraneous, and germane cognitive load. Educational Psychology Review, 22(2), 123–138. https://doi.org/10.1007/s10648-010-9128-5

44.

Tsai

M. J.

Hou

H. T.

Lai

M. L.

Liu

W. Y.

Yang

F. Y.

(2012). Visual attention for solving multiple-choice science problem: An eye-tracking analysis. Computers & Education, 58(1), 375–385. https://doi.org/10.1016/j.compedu.2011.07.012

45.

Vaca-Castano

Lobo

N. D. V.

Shah

(2019). Holistic object detection and image understanding. Computer Vision and Image Understanding, 181, 1–13. https://doi.org/10.1016/j.cviu.2019.02.006

46.

Wang

Ruan

Yang

M. H.

(2015). Deep networks for saliency detection via local estimation and global search. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 3183–3192). IEEE.

47.

Yan

Ren

Sun

Zhao

Han

Marshall

Zhan

(2018). Unsupervised image saliency detection with gestalt-laws guided optimization and visual attention based refinement. Pattern Recognition, 79, 65–78. https://doi.org/10.1016/j.patcog.2018.02.004

48.

Zahra

S. A.

George

(2002). Absorptive capacity: A review, reconceptualization, and extension. Academy of Management Review, 27(2), 185–203. https://doi.org/10.5465/amr.2002.6587995