Keywords

1 Introduction

According to the Next Generation Science Standards [1], science education is more than memorizing scientific facts and concepts. Science education should engage students in science and engineering practices, e.g., planning and carrying out investigations, hypothesis testing, analyzing and interpreting data, and arguing based on evidence [1]. Game-based learning, i.e., learning through playing games designed for educational purposes [2], is a promising method to help students gain conceptual understanding by applying science knowledge in solving game problems [3, 4].

Learners' reflection on their gameplay is necessary for a meaningful game-based learning experience [5,6,7,8]. Derived from Dewey's pragmatism theory, reflection refers to a metacognitive process that allows one to be cognitively aware of one's actions [6, 8, 9]. Moreno and Mayer [7] have argued that meaningful learning in multimedia learning environments (e.g., game-based learning) occurs when students select relevant information and organize and integrate it with prior knowledge. Students’ reflection contributes to the organizing and integrating process [7], thus making it essential to the game-based learning outcome. Kiili [5] has further asserted that reflection enhances students’ conceptual understanding by connecting gameplay with learning. According to the experiential gaming model [5], both direct experience and reflective observation are critical to game-based learning. By reflecting on their previous direct experience and corresponding feedback, learners conceptualize their previous gameplay experience to propose a new hypothesis concerning game problems [5]. Moreover, Kori et al. [6] have noted that reflection is especially important for exploratory science learning environments (e.g., game-based science learning) because students tended to hold misconceptions and misunderstandings in their initial explaining of natural phenomena.

Despite the necessity of reflection, few students proactively conducted reflection in game-based learning without extra guidance [4, 10, 11]. Students may solve game problems by guessing or trial-and-error without being metacognitively aware of underlying content knowledge [4, 11], thus impairing their game-based learning outcome [2, 5]. Moreover, reflection is a high-level metacognitive process where students rarely practice in traditional teaching settings [6, 8, 9, 12]. Therefore, it is valuable to investigate how to support students’ reflection in game-based science learning to enhance their science understanding. Such research can provide educators and game designers with insights on the design of reflective game-based learning experience to facilitate students’ science learning outcome.

A few research papers reviewed the literature on reflection support in game-based learning, but none of them focused specifically on supporting students’ reflection in science education. For example, Wouter and Van [9] conducted a meta-analysis to assess the effects of instructional support (e.g., reflection support) on game-based learning outcome based on relevant empirical studies from 1990 to 2012. Results showed that students significantly learn better by playing educational games with reflection support than those without such support (d = .29, p < .01). Additionally, Taub et al. [8] reviewed studies concerning embedded reflection prompts in game-based learning. They analyzed when reflection was prompted (e.g., time-based, activity-based), how students were prompted (e.g., written, spoken, menu-based), and how students responded to the prompts (e.g., spoken, written). Taub and colleagues found that embedded reflection prompts were effective in improving students’ game-based learning outcome. But the researchers did not include out-of-game reflection support in game-based learning, such as teacher guidance.

This literature review is inspired by Wouter and Van [9] and Taub et al. [8]. In the current study, we review the literature regarding the design and effects of students’ reflection support in game-based science learning to shed light on the effective game-based science learning design. But unlike Taub et al. [8], we will focus on both in-game (e.g., embedded prompts) and out-of-game reflection support (e.g., instructor guidance). The findings will help educators and game designers gain a better understanding of how to support students’ reflection in game-based learning to facilitate their science learning outcome. We will answer the following research questions:

  1. 1.

    How is reflection support designed in game-based science learning?

  2. 2.

    What is the effect of the reflection support on students' game-based learning outcome?

2 Methods

2.1 Database and Search Terms

We used four databases through ProQuest: ERIC, APA PsycINFO, Sociological Abstract, and Education Database. ERIC is sponsored by the U.S. Department of Education to provide extensive access to educational-related literature. PsycINFO is a ProQuest online version of Psychological Abstracts, covering journal articles, book chapters, books, technical reports, and dissertations in psychology and psychological aspects of related disciplines from 1887 to the present. Sociological Abstract covers sociology-related topics including education, social psychology, urban studies, and so on. Education Database supports the study and application of education across all education levels, including early childhood education, primary and secondary education, and higher education.

We used search terms related to game-based learning, science education, and reflection. We identified similar terms using the thesaurus function of the above four databases. The final search terms we used were: (“Game-based learning” OR “educational games” OR “learning games” OR “serious games” OR “video games” OR “computer games” OR “digital games”) AND (“science”) AND (“reflect*”). The asterisk in the ProQuest platform means searching for words starting with the letters before the asterisk. We set the searching range as abstract and limited the searching results to English peer-reviewed scholarly journals.

2.2 Inclusion and Exclusion Criteria

We identified 131 papers after the database searching process. Then we screened their titles and abstracts to determine the papers included in the current literature review based on the inclusion and exclusion criteria. Our inclusion criteria were:

  • Games discussed should be digital or computer games that were used for educational purposes. We focused on digital or computer games instead of board games because the former could have multimedia in-game reflection support.

  • Subjects should be natural science, e.g., physical sciences, life sciences, and earth and space sciences.

  • Participants should be secondary school students to college students.

  • The paper should discuss students' reflection on game-based science learning.

Theoretical or review articles that did not meet items 2 and 3 were also included to enhance the current research topic’s theoretical foundation.

Our exclusion criteria were:

  • Games were board games.

  • The paper discussed researchers' or teachers' reflection instead of students’ reflection.

After the screening process, a total of 14 studies were included in the literature review. Among them, ten were empirical studies and four were theoretical or review articles. We conducted systematic analysis mainly based on the empirical studies, while the theoretical or review articles were used to discuss the results.

3 Results

In this section, we first synthesize researchers’ definitions of students’ reflection in game-based learning. Then, we identify two types of reflection support in the literature: in-game prompts and instructor guidance. We discuss the design and effects of these two types of reflection support on students’ game-based science learning.

3.1 Definition of Reflection in Game-Based Learning

Overall, researchers shared the basic understanding of students’ reflection in game-based learning: students looking back to their prior experience for new understanding [6,7,8, 13]. However, researchers focused differently on what “prior experience” students reflected on and what “new understanding” they were supposed to gain from the reflection. Most researchers investigated reflection on students' game strategies or solutions to game problems [5, 7, 9, 10, 13,14,15,16]. For example, Wounter and Van [9] and Moreno and Mayer [7] defined reflection as students thinking about and explained their answers or solutions. And the purposes of such reflection were to help students be cognitively aware of their game actions, connecting game problem solving with underlying content knowledge, and detecting potential misconceptions [7, 9, 10, 13, 14].

Other researchers (e.g., [8, 17,18,19]) held a broad understanding of students’ reflection. They focused on reflection on students’ whole game-based learning experience, including game information, game environments, students’ perceived learning gains and feelings. For example, Nelson [18] investigated students’ reflection on evidence or clues they gathered in the game. Nilsson and Jakobsson [19] asked students to reflect on what they have learned from the intervention. In addition to facilitating learning, such reflection was designed to contribute new hypotheses or solutions to game problems and students' self-reported assessment of their learning experience [17, 18]. The next section describes the design and effects of two types of reflection support in game-based science learning.

3.2 Reflection Support Design and Effects

For the empirical studies, we coded their research method (e.g., quantitative and qualitative), participants, reflection support design (e.g., in-game prompts and instructor guidance), reflection support time (i.e., when was the reflection support triggered), students’ response type (i.e., how did students respond to the reflection support), and effects of such supports on students game-based learning outcome (e.g., not discussed, positive effects, non-significant effects). Results are shown in Table 1.

Table 1. Description of the empirical articles analyzed in this study
Full size table

Results indicated that most (70%) empirical studies adopted the quantitative research method and two were qualitative research (design-based research, to be specific). The sample size ranged from 22 to 272. Most participants were in the U.S., while three studies were conducted in Europe and one was in Asia.

There were overall two types of reflection support: in-game prompts and instructor guidance. The in-game reflection prompts varied in presentations: textual or pictorial. And instructor guidance varied in implementation forms: individually and collectively. All in-game reflection prompts were triggered within gameplay (e.g., after students completing a game level or task), while instructor-guided reflection was mostly conducted after the whole gameplay. Moreover, some in-game prompts did not require students to respond, neither spoken nor written, but all instructor-guided reflection required students’ response. Findings of six studies indicated that reflection support tended to positively affect students' learning outcomes, while one study yields non-significant results.

In-Game Prompts

In-game reflection prompts were predefined questions in a game to help students reflect on their previous game experience. Researchers designed textual or pictorial in-game prompts triggered when students completed a game level, finished a key game task, or collected an important clue in gameplay. For example, Geden et al. [14] designed a 3D video game to help middle school students learn microbiology knowledge. Students role-played as microbiologists to explore the transmission source and treatment of a pandemic in a virtual island. In-game reflection prompts were presented as the message in the students' in-game cellphone asking them to reflect on their game progress and upcoming plan (e.g., “In your own words, please describe the most important things that you’ve learned so far, and what is your plan moving forward?”, other examples of predefined reflection prompts are showed in Table 2). Such prompts were triggered at five milestones in the game (e.g., after finishing a critical task or collecting a piece of important evidence). Students were asked to type in and send their responses via the in-game cellphone. A total of 118 middle school students in the U.S. played the game for two to three class periods and spent approximately 6 min on writing reflection. Results showed that their responses to the reflection prompts varied significantly in length and quality.

Table 2. Examples of predefined reflection prompts
Full size table

Similarly, Moreno and Mayer [7] designed textual prompts delivered by a game agent. Moreno and Mayer [7] adopted a multimedia botany game called Design-A-Plant. The game took students to travel around five alien planets with different weather conditions. Students had to choose suitable plants for each planet based on their certain weather conditions to win the game. After making planting choices, a pedagogical agent asked students to justify their planting choices or the correct choice. However, unlike the written responses in Geden and college’s [14] study, students in Moreno and Mayer’s [7] research were asked to give an oral explanation to the prompts. Researchers recorded the audio and analyzed if students applied correct knowledge in their explanations. Barab et al. [3] also designed three textual in-game prompts to facilitate students’ reflection when they played a multiple-user virtual game called Taiga Virtual Park. However, researchers did not report which reflection questions they asked and how students were required to respond to the prompts.

Other researchers [16, 18] only presented the in-game reflection prompts but did not require students to respond. For example, Nelson [18] designed voluntary reflection prompts customized to students’ gameplay progress. In a biology game called River City, students interacted with predefined characters and digital agents in a virtual town to investigate why its residents were getting ill. After interacting with certain characters or digital subjects, one or three hint buttons were enabled on the screen. Students were voluntary to click the button(s) to get reflective guidance messages. Each hint contained one message. The messages asked students to reflect on clues they gathered about solving the game problems. For example, after visiting a hospital in the virtual town, a student received a message: “Is there anything about this hospital that is different than the ones you have seen?”.

All the above research focused on textual in-game prompts, but Huang [16] designed both textual and pictorial reflection prompts in a physics game to help students learn Newton’s Second Law of Motion. The in-game prompts asked students to reconsider the particular game level. For example, a game level required students to choose among three different sizes of blocks to create different forces to a lever. The corresponding textual reflection prompt showed up after students completed this level and asked, “When you chose the middle-size block, what did you expect to see?” For the pictorial version, researchers translated the same content into graphs and icons. In their study, 83 6th–8th grade students were randomly assigned into four conditions: gameplay without reflection support, gameplay with textual support only, gameplay with pictorial support only, and gameplay with both textual and pictorial support. To alleviate the interruption of the support to students' gameplay, the textual or pictorial prompts froze on the screen for specific seconds to trigger students’ reflection, but students were not required to answer these questions.

All five papers on reflection support as in-game prompts found positive relationships between using of such prompts and students’ learning outcome (e.g., scientific skills, scientific knowledge, retention), except Barab et al. [3], who did not discuss the effects of the prompts.

However, the effects of in-game prompts varied based on their content [7], number [18], and formats [16]. For example, in Moreno and Mayer's [7] study, a total of 78 fresh undergraduate students were randomly assigned into self-reflection group (i.e., students were prompted to justify their answers), program-reflection group (i.e., students were prompted to justify the correct answer), and no-reflection group. Results showed that students from self-reflection and program-reflection groups recalled significantly more focal scientific knowledge than those in the no-reflection group (effect size = 0.81). Further, students in the program-reflection group outperformed those in the self-reflection and no-reflection groups in the far-transfer test (effect size = 0.80). Students' spoken explanation was recorded and analyzed. Results showed that students reflecting on the correct answers generated a higher proportion of correct explanation than those reflecting on their own answers (effect size = 1.10).

Moreover, Nelson [18] suggested that the number of in-game prompts might impact the effects of such prompts on students' learning. In his study, 272 public middle school students were randomly assigned to three conditions: no guidance, extensive guidance (i.e., receiving three reflection messages after each interaction), and moderate guidance (i.e., receiving one reflection message after each interaction). All students played the game over three weeks and completed the pre-and post-content test to measure their science inquiry skills and content knowledge. Results showed no significant difference in the pretest scores across the three groups after accounting for their pretest score. However, results indicated a significant positive relationship between guidance message views and posttest score for students in the extensive guidance group, after accounting for their socio-economic levels, prior science grades, and computer game experience. But the benefit of guidance viewing on learning did not show in the moderate guidance group.

Furthermore, Huang [16] indicated that prompts in both textual and pictorial formats were more effective than prompts in a single representation. The researcher randomly assigned 83 middle school students into four conditions: gameplay without reflection support, gameplay with textual support only, gameplay with pictorial support only, and gameplay with textual and pictorial support. All students first took a demographic survey and physics knowledge pretest, then played the game individually for 15 min, followed by a posttest and engagement survey. Findings showed that students received both textual and pictorial reflection prompts scored highest in the scientific knowledge posttest among all participant, accounting for their pretest score.

Although students play educational games with in-game reflection prompts might outperform those who play the game without such prompts in knowledge tests, Huang [16] and Nelson [18] contended that some students tended to pay inadequate attention to these prompts, especially when they were not required to respond to the prompts spoken or written. For example, when in-game prompts were designed to freeze on the screen for specific seconds to trigger students’ reflection, Huang [16] found that some students thought the prompts distracting and tried to skip them. Similarly, Nelson [18] reported that a quarter of the students with access to the reflection prompts did not view any prompts at all. The average number of views of reflection messages was quite low. Student interviews and classroom observations revealed the reasons for low access to prompts: (1) Students did not know the purpose of the prompts, and (2) they exclusively focused on solving the game missions and purposely ignored the prompts [18].

Instructor Guidance

Instructor-guided reflection referred to the instructor providing post-gameplay collective or within-gameplay individual guidance with or without predefined prompts to help students reflect on their game experience. Unlike in-game reflection prompts, most instructor guidance was conducted after gameplay (see [10, 15, 17, 19]). Moreover, since instructor guidance involved human interaction, this type of reflection support tended to require students' responses (see [10, 13, 15, 17, 19]).

Three research [10, 15, 19] focused on instructor-guided reflection as post-gameplay collective debriefing. For example, Anderson and Barnett [10] conducted design-based research to investigate how to integrate a 3D action/racing game called Supercharged! in science lessons to help middle school students grasp the basic electromagnetic concepts. Class observation showed that some students were confused about how they were supposed to learn from gameplay, and few students critically reflected on their gameplay. Therefore, the teacher developed log sheets for students to record their game actions to detect patterns in their gameplay. And the teacher displayed some students’ solutions and corresponding feedback to the class and guide students to interpret the feedback and predict results if using another solution in the debriefing. Nilsson and Jakobsson [19] also conducted design-based research on collective instructor guidance, but the instructor in their study guided students' reflection by posting several predefined questions. In this empirical qualitative study, 42 students (ages 14–15) from four Swedish schools reflected on their game experience of an urban simulation computer game called SimCity 4. These subjects were randomly selected from all participants of the Future City competition, where participants applied sustainability-related science concepts (e.g., photosynthesis, the greenhouse effect, the carbon cycle) to create a sustainable city proposal via the game. Teachers guided students to reflect on their proposals in a small group for 20–30 min. During the reflection, students discussed with their peers freely regarding six predefined reflection questions (e.g., "How was the design work conducted? What determined your design choices? Did you learn anything from playing the game?").

In addition to collective instructor reflection guidance, researchers also focused on instructors providing individual guidance [13, 17]. Two types of individual instructor-guided reflection were discussed: in-class reflection prompts and reflection journal.

Koops and Hoevenaar [13] asked a teacher to provide in-class reflection prompts when students sought help when playing a 2D physics game called SPACE CHALLENGE. The researchers argued that students experienced two states during gameplay: game state and learning state. Students intuitively acted on the feedback from the game environments in the game state and generated spontaneous conceptual knowledge, which could only be used in the specific game situation. However, in the learning state, students rationally reflected on the gaming experience and generated formal conceptual knowledge, which could be transferred to other real-life situations. In their study, researchers investigated if suddenly increasing the complexity of the game between consecutive levels and providing teachers’ individual guidance would transit students from gaming state to learning state, thus helping students gain better conceptual understanding. They designed SPACE CHALLENGE, where players maneuvered a spaceship to collect diamonds. Three high school classes were randomly assigned into three groups, switch group—students played the game whose level difficultly increased substantially once a new physics concept was introduced, game group—students played the game whose levels were all consistent in difficulty, and control group—students received lecture-based instruction without gameplay. Both the switch group and the game group played the game in class for around 50 min with a teacher's facilitation. The teacher in the switch group would ask students reflection questions (e.g., “Did you notice what happened when you did not apply any force?” or “How do you effectively go through a corner?”) when students failed the game levels and asked for help. The teacher in Koops and Hoevenaar’s [13] study failed to implement the intervention as designed. It was hard for the teacher to provide individual guidance synchronously to the whole class. Students in the switch group frequently asked for help due to the complexity of game levels—most of their questions related to technical issues or game mechanisms. The teacher was unable to provide the predefined reflective prompts and offer assistance in time for everyone.

Compared to providing synchronous guidance, asking students to write self-reflective journals after their gameplay was more practical. Namgyel and Bharaphan [17] investigated if integrating simulation and game with teaching would improve students’ understanding of the photoelectric effect. The teacher asked 31 12th grade students in Bhutan to write self-reflective journals after the intervention to reflect on their learning experience. Rather than facilitating learning, such reflection served as a measurement of students' attitudes towards the game-based learning experience.

Among all five research on instructor-guided reflection, two reported positive effects on students' learning [15, 19], one reported non-significant result [13], and two did not discuss it [10, 17].

Nilsson and Jakobsson [19] and Herrero et al. [15] conducted qualitative research and draw their conclusion concerning instructor-guided reflection’s effects by thematically analyzing observation data. Researchers asserted that instructor-guided reflection helped students enhance their understanding of the underlying science knowledge by connecting the content knowledge with their game experience [15, 19]. Herrero et al. [15] provided an example of how the teacher heuristically guided students to think deeply and recall exact prior knowledge to explain their game actions. A student explained a certain game level using a Darwinian strategy. The teacher asked why the student said that. The student replied that natural selection happened. The teacher re-asked the why question. Finally, the student linked their game experience with exact content concepts and knowledge from Darwin’s Theory of Evolution. Furthermore, Nilsson and Jakobsson [19] argued that collective reflection could lead to generative discussion, that is, students inspired by each other and generate ideas based on their own game experience. And the reflection on constraints of gameplay (e.g., the gap between game environments and real-world) helped students develop critical thinking towards the application of targeted science concepts.

However, Koops and Hoevenaar [13] reported non-significant effects of instructor-guided reflection. The researchers randomly assigned three high school classes into three groups, switch group—students played a game in hard mode and received instructor’s reflection prompts when they sought help, game group—students played the same game in easy mode, and control group—students received lecture-based instruction instead of gameplay. The pre-and post-test results showed that students from the switch and game groups outperformed those in the control group in conceptual understanding. However, there was no significant difference between the switch group and the game group in the posttest score after controlling for their pretest scores. The non-significant results may be because the teacher failed to provide the reflection prompts for everyone as planned. The teacher was busy with answering students' questions related to technical issues or game mechanisms. Researchers recommended the teacher or game designers to embed answers to commonly asked technical questions in the games in advance to improve the implementation of instructor-guided reflection.

4 Discussion

Students’ reflection is important for them to make sense of their gameplay experience and connect gameplay with learning when playing educational games [5,6,7,8]. This study selected and reviewed 14 articles regarding the design and effects of students’ reflection support in game-based science learning. Results showed that both in-game prompts and instructor guidance designed to scaffold reflection could prompt students’ game-based science learning outcome. And such effects varied based on the design of reflection support.

In-game reflection prompts were predefined questions designed to help students reflect on their previous game experience (e.g., collected clues, game strategies, solutions) after completing milestone game events (e.g., completing a level). Findings revealed that students receiving in-game reflection prompts gained better game-based science learning outcome than those without such prompts. And such effects varied based on the content, number, and formats of the prompts. The findings align with the results of a systematic analysis conducted by Taub et al. [8], who asserted a positive effect of in-game reflection support on students’ learning.

In-game prompts were triggered during gameplay, so some researchers did not require students’ spoken or written responses to avoid distracting them from gameplay. However, it turned out that some students might ignore these prompts and did not exert reflection. Therefore, efforts are needed to direct students’ attention to such prompts and ensure students’ reflection.

Moreover, most research on in-game prompts employed the experimental method. Therefore, it is hard to generalize the findings to other populations and other science games of studies with participants from one school and one type of game. Further, some studies’ population was not clearly defined, and their selection methods were either not reported or the convenient sampling (e.g., [16, 18]).

Unlike in-game prompts, instructor-guided reflection was mainly conducted collectively after gameplay. Teachers guided students to discuss shared issues in their gameplay or reflect on exampled students’ game experience. Teacher-guided reflection tended to be longer (e.g., 20-min debriefing), deeper (e.g., asking follow-up questions), and more flexible (e.g., without predefined prompts) than reflection supported by in-game prompts.

Some research (e.g., [10, 17]) discussed the design of instructor-guided reflection but did not analyze its effects on students’ learning outcome. Results from other research revealed that instructor guidance on students’ reflection could help students enhance their understanding of the underlying science knowledge by connecting the content knowledge with their game experience. The findings were consistent with Kori et al. [6], who asserted that students significantly learn better by playing educational games with instructor-guided reflection than those without such supports. Moreover, results indicated that collective instructor guidance (i.e., class-wide debriefing) was more feasible than individual one (i.e., providing individual guidance to each student in class).

The effects of such support depend heavily on teachers’ guiding skills. For examples, teachers should know how to organize academically productive discourse in game-based science lessons [20]. Moreover, answers to commonly asked technical questions should be prepared in advance to avoid taking up the time for productive discourse [13]. Most research on instructor guidance adopted the qualitative method and their evidence for the support’s effect was researchers’ observation. Therefore, it is hard to draw casual conclusions on the relationships between instructor-guided reflection and students’ learning outcome.

5 Conclusion

This study analyzed the design and effects of reflection support in game-based science learning. Overall, both in-game reflection prompts and instructor guided-reflection are promising in facilitating students’ science learning. This study sheds light on the design of reflection support from the game designers’ and instructors’ perspectives.

Regarding research on in-game reflection prompts, qualitative studies are needed to analyze how the in-game prompt works on students, how students respond to these prompts, and how the reflection facilitates students’ learning. Compared to instructor guidance, in-game support is more likely to facilitate reflection tailored to students’ game experience and personal characteristics. Future researchers should examine the design of customized or adaptive in-game reflection support. Regarding research on instructor-guided reflection, rigorous experimental studies are needed to verify the effects of such support on students’ learning.