Electronic Commerce Research and Applications

Electronic Commerce Research and Applications 44 (2020) 101011
Contents lists available at ScienceDirect
Electronic Commerce Research and Applications

journal homepage: www.elsevier.com/locate/elerap
Enhancing Computational Thinking Capability of Preschool Children by

Game-based Smart Toys
Szu-Yin Lin a, Shih-Yi Chien b, *, Chia-Lin Hsiao c, Chih-Hsien Hsia a, Kuo-Ming Chao d
a
Department of Computer Science and Information Engineering, National Ilan University, Yilan County, Taiwan
b
Department of Management Information Systems, National Chengchi University, Taipei, Taiwan
c
Department of Information Management, Chung Yuan Christian University, Taoyuan City, Taiwan
d
School of Computing, Electronics and Mathematics, Coventry University, Coventry, UK
A R T I C L E I N F O A B S T R A C T
Keywords: Computational thinking has become an important issue in the field of education. Because preschool and
Cognitive development of children kindergarten learners are capable of exercising their cognitive abilities to resolve basic computational logic, this
Computational thinking demographic has raised significant interest in studying their learning intentions and behaviors. However, prior
Game-based learning
research fails to examine the effects of teaching computational logic to kindergarten children. Therefore, this
Tangible user interface
Smart toys
study aims to investigate the influences of teaching approaches in guiding preschool children to learn compu
tational logic and programming concepts to enhance their problem-solving skills as well as computational
thinking abilities. A novel teaching framework is designed to develop the learner’s cognitive abilities, which
adopts the smart toy game-based learning approach along with a tangible user interface (TUI) to enhance
children’s learning performance and interests. The proposed teaching approach integrates the game-based
learning concepts into the TUI system, where the learning processes allow the learners to effectively practice
the conceptual knowledge and efficiently advance their problem-solving skills. The results suggest using the
developed game-based TUI system can increase preschool children’s learning behaviors as well as enhance their
learning interests and computational thinking abilities.
1. Introduction developing innovative knowledge in different domains (Furber, 2012).

For example, the STEM (science, technology, engineering, and mathe
The advanced achievements in computer science have altered the matics) field aims to enhance learners’ logical thinking abilities and
manner in how humans use and learn information technology. Nowa problem-solving skills through the practice of programming logic.
days, science and technology have become omnipresent. People not only Computational thinking has been suggested as a required skill that
browse science and technology knowledge on the World Wide Web, but every individual should acquire (Wing, 2006). Brennan and Resnick
also access the relevant information through a variety of mobile elec (2012) concluded learning programming languages can improve
tronics and smart toy applications. The next generation is also expected learners’ computational thinking abilities. Computational thinking
to advance from browsing users to inventors and creators who will processes require learners to approach problems by developing logical
intensify the competitiveness of global powers. There has been a solutions and practicing concise communication to explain their
worldwide initiative to teach programming logic to children for reasoning activities (Lye and Koh, 2014; Shafto, 1986). Lindberg et al.
enhancing their computational thinking abilities and problem solving (2019) also point out an increased global trend in learning computa
skills. The concept of computational thinking proposed by Wing (2006) tional thinking in education, which reveals the importance of studying
has been extensively discussed across various fields and has gradually programming lessons for preschool children. Three types of computa
become an important part of education. Computational thinking is a tional thinking are involved in developing a program: computational
process of comprehending and solving problems, which is not limited to concepts, computational practices, and computational perspectives
a single discipline or field. It has become indispensable in analyzing (Brennan and Resnick, 2012), where each component addresses a
numerous types of real-world computational problems, as well as different learning objective and can vary in degree of difficulty.
* Corresponding author.
E-mail address: [email protected] (S.-Y. Chien).
https://doi.org/10.1016/j.elerap.2020.101011
Received 8 March 2019; Received in revised form 21 August 2020; Accepted 23 September 2020
Available online 28 September 2020
1567-4223/© 2020 Elsevier B.V. All rights reserved.
S.-Y. Lin et al. Electronic Commerce Research and Applications 44 (2020) 101011
Additionally, the cognitive development theory from Piaget (1976) and contributions: first, to the best of our knowledge, this is one of the most
the scaffolding theory from Vygotsky which summarized by Berk and limited research that integrates game-based learning concepts into TUI
Winsler (1995) suggest that preschool children require more assistances systems to enhance preschoolers’ computational thinking abilities.
for converting conceptual ideas into tangible entities. To decrease Second, the proposed teaching framework has been empirically vali
learning barriers, prior studies have developed visual programming dated and shown promising results in improving students’ learning
language platforms to cater different age learners and skill levels, as well abilities, which can benefit the preschool educators and instructors in
as provide diverse teaching methods (Smith et al., 2019) and provide an delivering the conceptual course materials.
opportunity for developing computational thinking to preschool stu The article is organized as follows: the Section 2 reviews the factors
dents (Ching et al., 2018). With the graphical design and interface, vi affecting computational thinking; Section 3 introduces the theoretical
sual programming language (VPL), compared with traditional text- model used in the study; Section 4 includes the details of methods;
based programs, is substantially easier to learn programming skills Section 5 discusses the experimental results; Section 6 concludes the
and can enhance the students’ learning interests (Grover and Pea, 2013). study findings.
For example, the Lego EV3 system uses the VPL platform with pro
grammable bricks to provide a tangible user interface (TUI) to the 2. Literature reviews
learners in developing the (robotic) system applications. In addition,
through the use of a VPL system, learners can experience the rewarding This study aims to examine the effects of applying game-based
process of problem solving while encountering program bugs, and learning methods to design course materials, as well as improve pre
solving the task at hand. This method has been known as game-based school children’s computational thinking skills. To provide better un
learning, which combines entertainment and problem solving ele derstandings of the research questions, the following literature reviews
ments into the learning processes. Prior research conclude game-based include computational thinking, tangible user interface, smart toys,
learning can effectively enhance students’ learning interests and learning cycle teaching strategies, and digital game learning methods.
improve overall learning performance in high school education
(Papastergiou, 2009). 2.1. Computational thinking
Although great efforts have been devoted to developing hardware
and software tools for students to learn programming skills and improve Computational thinking is a process in which a person develops a
their computational thinking abilities, most current developments focus series of thinking strategies to approach a problem (Wing, 2006). The
primarily on junior and senior high school students rather than pre idea of educating children’s computational thinking first appeared in
school students (Lye and Koh, 2014). In other words, most resources are early 1960 (Rees et al., 2016). Due to the lack of evidence to support its
currently designed for experienced users and examine how the devel effectiveness, little attention was paid to construct the relevant educa
oped tools can further advance these users’ computational knowledge. tional tools to enhance learners’ computational thinking abilities and
As preschool children have limited cognitive abilities and may require problem solving skills (Lye and Koh, 2014). However, with the
more assistance to transform abstract concepts into real-world entities, advancement of technology, computational thinking education has
the game-based learning approach with TUI applications provide a become an important topic in recent years (Lindberg et al., 2019). For
better way to motivate preschoolers to actively participate in the example, the development of VPL (such as Scratch) allows students to
learning materials as well as increase their computational thinking learn the programming skills and required knowledge in an effective and
abilities. However, there lacks sufficient resources providing basic efficient manner. In other words, the reduced learning cost helps stu
concepts for beginner learners, such as the research gaps identified in dents reduce their cognitive loads during the learning processes and
Zhang and Nouri’s review study (2019). As the preschool children’s makes it easier to develop computational thinking abilities through the
learning capability is limited by their cognitive level (Koslowski, 1980), learning procedures (Shafto, 1986).
the conventional syntax-based programming languages can inject sig Wing (2006) concluded “computational thinking is for everyone and
nificant difficulties (Chien et al., 2018) and is therefore inappropriate for everyone must have the skills.” Computational thinking ability has been
these leaners. This implies that the traditional teaching tools may be suggested to significantly correlate with STEM (science, technology,
insufficient for preschool children. Due to children’s limited cognitive engineering, and mathematics) education (Khine, 2018). Since STEM is
ability, teaching program logic and determining a suitable curricular for highly related to the information technology fields, this suggests that
preschool students remain a challenge. computational thinking abilities should be considered as a critical sub
The objective of this study aims to create an appropriate teaching ject and take root in information science education. Various countries
framework for enhancing the preschoolers’ computational thinking have devoted considerable efforts to promote computational thinking
abilities, learning interests, and learning achievements. The game-based education in academia, industry and government units. For instance, in
learning method along with a TUI is created to guide the preschool the United Kingdom, the programming-related courses are regulated as
learners to study computational logic and programming concepts. This a compulsory course in the secondary school’s syllabus. Germany,
goal has the following sub-goals: Netherlands and Japan have also developed information technology
capability indicators to strengthen the course materials and ensure the
(1) applying a game-based learning method to strengthen a learner’s students have sufficient computational thinking abilities.
computational concepts, computational practices, and computa The computer science teachers association (CSTA) in the United
tional perspectives States established a core competency standard for computer science
(2) creating a TUI to engage a preschoolers’ learning interests as well education, which develops a framework for K-12 CS-related education
as encourage them to participate in the learning activities (Seehorn et al., 2011). The framework divides the CS courses into three
levels (Fig. 1). To better examine the preschool students’ computational
Two rounds of user studies were conducted to collect the empirical thinking abilities, our study focuses on the fundamental level (i.e., level-
data to examine the research questions. The results show that the 1). The first level is for students from kindergarten to sixth grade, where
developed teaching approach (game-based learning with the TUI sys the education guideline concentrates on the students’ understanding of
tem) can effectively increase preschool learners’ cognitive thinking the basic CS concepts. This level aims to develop students’ creativity,
performance, as well as improve their learning behaviors. In addition, active learning abilities, and explored capabilities to encourage students
the results also reveal that the participated preschoolers’ prior learning to apply computational thinking ability into basic or daily science and
experience can not only influence their study performance but also their technology matters.
learning behaviors and interests. Our study presents the following Computational thinking ability can be categorized into three
2
objects. One of the notable developments, Augmented Urban Planning

Workbench (Ishii et al., 2002), simulates real world phenomena (e.g.,
airflow, shadows, and reflections), constructs landscapes with real world
materials (e.g., clay or sand), and integrates these properties into a
three-dimensional space. Additionally, a variety of smart toys can be
seen as an extension of TUI computational thinking education. For
example, BRICKO, an educational computing thinking robot, supports
tangible interface, social and entertainment interactions while
educating children (Pedersen et al., 2018).
Because most tangible computing environments are too complex for
young children, prior research proposed Tangicons (the non-electronic
physical programming cubes for beginner learners) to incorporate
tangible bricks into physics games, allowing children to learn pro
Fig. 1. K-12 computer science core indicators (Seehorn et al., 2011).
gramming in a fun way (Scharf et al., 2008). Researchers who have
compared the ease of understanding between graphics and tangible user
interfaces, and found that tangible and graphical systems are equally
dimensions (Brennan and Resnick, 2012; Lye and Koh, 2014), including
easy to understand. However, through a tangible interface, users are
computational concepts, computational practices, and computational
more likely to try and actively participate (Horn et al., 2009). Examples
perspectives. This study adopted these three dimensions as the basis to
of TUI that are actually available for children to work with include
design the course materials and learning outcomes, which allowed the
puzzle pieces; their inherent physical syntax of connectable elements
learners to practice logical thinking as well as comprehend the compu
provide a powerful and expressive metaphor for building tangible sys
tational knowledges. Therefore, through the learning processes, a
tems (Oh et al., 2013). Another example is TanProRobot 2.0, which
learner first studied the fundamental computational concepts and then
consists of three parts: a tangible programming block, a robotic car, and
resolved the advanced programming questions in the computational
several manipulators. The child can program the robot car by perform
practices phase. Once accomplished, the user should be able to fully
ing programming blocks to perform certain operations and interact with
understand the relationships in a given programming context. Detailed
the car through manipulation. It can help children learn programming
descriptions of the three dimensions are summarized in Table 1.
concepts and get a glimpse of event handling concepts (Wang et al.,
A considerable amount of effort has been devoted to examine
2016). Other examples of the latest tangible systems include Maker
learners’ computational thinking abilities. For instance, Yadav et al.’s
Wear, a tangible interface for wearable design that allows children to
(2017) and Denning’s (2017) studies developed the teaching guidelines
apply computational thinking (Kazemitabaar et al., 2017). MakerWear
to introduce the computational thinking related materials to the
is a great example of how new technology settings can fundamentally
learners. In addition, Shute et al. (2017) proposed a model to assess
change the interaction of electronic media and behavior, as well as
students’ computational learning outcomes; whereas Sullivan and Bers
perception, physical manipulation and overall social activities (Bergs
(2018) and Pérez-Marín et al. (2020) focused mainly on the preschool
mark and Fernaeus, 2016).
children’s computational thinking learning performance. These studies
This study will let children learn the concept of computational
suggested that computational thinking has become the fundamental and
thinking ability through TUI interface and smart toys, such as color
critical skills in this digital age.
answer cards and Arduino toy car robots. The review of smart toys will
be introduced in the next section.
2.2. Tangible user interface
2.3. Smart toys
Tangible user interface (TUI) is a user interface in which people can
interact with digital information through the physical components. The A smart toy is a toy that can respond to user feedback and change its
purpose of TUI is to enhance collaboration, learning and design ability behavior according to environmental stimuli. It can act according to the
by providing a physical form for digital information, thereby enhancing design model, and usually adapts to the player’s abilities. Smart toys
human ability to learn and manipulate physical objects and digital in typically have electronics consisting of microprocessors or micro
formation (Ishii, 2008). In addition, Ishii’s group, one of the pioneers of controllers, memory storage devices, and numerous forms of
the TUI, develops a TUI application called Tangible bit (Ishii, 2008), input–output devices (Boss et al., 2001). While computers can represent
which provides a physical form of digital information that allows people children in the medium of social and intellectual development, some
to operate digital signals and connect the physical objects with the researchers believe that using computers before the age of 7 reduces
digital data. A simple example of a TUI is a computer mouse, an interface important developmental tasks and other types of learning (Healy,
that allows people to interact with digital information through physical 2000). Therefore, some scholars have proposed an interactive interface
with smart toy that children can use alone or in combination with a
Table 1 computer, combining current popular mobile device learning, tangible
Dimensions of computational thinking ability (Lye and Koh, 2014). interfaces, and a variety of home technologies (Plowman and Luckin,
Dimension Description Examples 2004). If a toy contains only a unilateral action or a single message-
transmitting display but has little ability to adapt user’s intentions, it
Computational The concepts used in Variables
concepts programming processes The loops
should not be classified as a smart toy. In other words, the distinguishing
Computational The methods of solving Being incremental and factor a smart toy has is the ability to integrate machine/system appli
practices programming problems iterative cations into gaming experience to create human-like intelligence.
Testing and debugging Additionally, most of today’s smart toys have a networked mecha
Reusing and remixing
nism, requiring a demand for people to pay higher attention to privacy
Abstracting and
modularizing rights. Because the definition of privacy may not be fully understood by
Computational The student’s understanding of Expressing and children in the early childhood stage, children are more likely to un
perspectives the relationships between oneself questioning the use of knowingly reveal private information (Rafferty et al., 2017). Therefore,
and others in a technological technology related studies should refer to and comply with the privacy protection
context
requirements of Hung’s study (2016), including the concept that
3
children may not understand privacy and children will disclose as much Table 3
information to smart toys as they can trust. For example, Delprino et al. 5E Learning cycle.
(2018) develop ABBOT that combines intelligent tangible objects with Stage Phase Introduction
outdoor sports. ABBOT’s tangible objects help children capture images
1 Engagement The teaching model of this period completely imitate the
of the elements they find interesting in the physical environment, learning task. Activities should be able to connect past and
inspiring a greater interest to explore the outdoor environment. present learning experiences and focus on the thinking
Evidently, several smart toy robots have recently been invented for the process of students’ learning outcomes at present activities.
purpose of educating and engaging. Additionally, Relkin and Umaschi This phase should engage students to explore the concepts,
processes, and techniques of the mind.
Bers (2019) designed TACTIC-KIBO, a tangible learning tool, to 2 Exploration This phase of the teaching model provides a common
encourage preschool children in activity learning programming con empirical basis for students to identify and develop current
cepts and skills. Sullivan et al. (2017) concluded the use of KIBO robot concepts, processes and techniques. During this period,
kit can effectively advance learners’ computational thinking abilities as students actively explored their environment and
manipulated teaching materials and teaching aids.
well as increasing their learning interests. Evidently, several smart toy
3 Explanation In this period, students learn to communicate their
robots have recently been invented for the purpose of educating and observations. Students also learn to interpret their
engaging. Table 2 summarizes the smart toys with its educational observations and draw a meaningful conclusion. Teachers
functions. can simultaneously introduce formal definitions explaining
In this study, the mBot Arduino robot is selected as an intermediary concepts, processes or behaviors.
4 Elaboration Elaboration teaching mode challenges students to gain a
TUI interface that can be used in conjunction with teaching. Addition deeper understanding of the material through application.
ally, this study also required self-designed answer color cards, a series of Through additional investigation or creating presentations,
teaching courses, and our own image recognition program written on students ensure they have a firm understanding on the
the mBot robot. content.
5 Evaluation This period allows students to self-assess and reflect,
ensuring they fully comprehend the material. This also
2.4. Learning cycle teaching strategies provides teachers with an opportunity to assess the progress
of students in achieving educational goals.
The learning cycle is a concept in which people can learn different
knowledge or skills through experiences. Usually learning cycles have
multiple phases, and the last phase can follow the first phase to be a 2.5. Digital Game-based learning
complete learning cycle. The 5E learning cycle was developed by Bio
logical Sciences Curriculum Study, and specifically for purpose of Digital game-based learning is learning through a technological
teaching science. This model describes a teaching sequence that can be gaming platform. In the game, learners achieve a sense of accomplish
used for entire programs, specific units, and individual lessons (Duran ment through solving and overcoming challenges. Digital game-based
and Duran, 2004; Bybee et al., 2006). Table 3 shows each phase of the 5E learning should take into account both gameplay and education, and
learning cycle. achieve the goal of entertaining and learning. Prensky (2003) pointed
In this study, we apply the 5E learning cycle as the basis for the out that the features of digital game-based learning include the
teaching methodology and experiments. In addition, we also combine following 12 features. The various features mentioned below will be
the methods used by Garris et al.’s (2002) Input-Process-Outcome Game applied in this study.
Model to design the teaching concept. The detailed game-based TUI for
computational thinking methodology will be introduced in Section 3. (1) Entertaining: the game is fun and engaging for the learner.
(2) Gameplay: provides a form of play. Motivates learners in a fun
and appealing way.
(3) Regularity: make the content of the game structured. It will make
it easier for learners to organize gameplay, and interact in the
Table 2 game.
Smart toy robots with educational function. (4) Goals: the specific tasks in the game can clearly guide the users to
Name Educational Functions learning through play.
KIBO KIBO is the screen-free robot kit for kids that lets 4–7 year-olds create, (5) Interaction: the game interface is user-friendly and intuitive.
design, decorate and bring their own robot to life. KIBO is an easy and (6) Adaptability: the game design can vary in degree of difficulty
fun way to bring robotics and coding to young learners and spark according to the learner’s level.
their interest in STEAM. (KinderLab Robotics, 2014) (7) Outcomes and feedback: provide opportunities for users to learn.
Dash & Dot A robot that can walk and identify obstacles. It can also respond, sing
or dance. This robot allows children aged 5–12 to learn and practice
(8) Sense of victory: learners achieve a sense of accomplishment
programming skills. (Wonder Workshop, 2015) through overcoming barriers in the game.
mBot mBot is an educational robot for beginners that makes teaching and (9) Conflict: competing and challenging: Challenges users with bar
learning robot programming simple and fun. mBot also aids in the riers and tasks for them to face and overcome.
development of logical thinking and design skills. (Makeblock, 2015)
(10) Problem Solving: design questions in the context of the game to
Cubetto The Cubetto Playset is a Montessori inspired coding toy that allows
children aged 3–6 to program a friendly wooden robot. The toy is inspire learner creativity.
powered by tangible programming language made of colorful (11) Social interaction: learners build a relationship with other game
wooden blocks (Primo Toys, 2016) players, creating a sense of community.
Codey A robot that includes a combination of software and hardware that (12) Representation and story: the learner is interested in the storyline
Rocky allows children to learn programming concepts through play and
creation. This robot uses mBlock as its programming language. (
and game tasks, and is emotionally invested in the game.
Makeblock, 2017)
KUBO KUBO aims to teach younger children the basics of programming and 3. Theoretical guidelines and research hypotheses
computational thinking in a simple and intuitive way. This game also
guides children through a series of challenges using small robots,
This study aims to improve how program logic is taught to preschool
square cards and maps. (KUBO Education, 2017)
ROBOPAL A programmable learning robot that uses magnetic coding blocks as children, bettering the students’ computational thinking capabilities.
its programming language. It also encourages children to have fun Game-based learning is used as the strategy, teaching through hands-on
while learning computational thinking. (Kickstarter, 2017) practice and tangible interaction. The Input–Process–Outcome game
4
model is also adopted to assist preschool children in acquiring compu Table 4

tational thinking. TUI Perception and behavior mapping for computational thinking.
TUI mapping concepts Corresponding Corresponding skills
3.1. Design concepts abilities
Perception mapping Visual appearance/position/

Based on constructivist learning, this study plans to let preschoolers location/shape/size/
learn through more specific means such as interactive interface and distance/color
Tactile click/slide/touch
hands-on operation of real objects. Programming is taught through
Behavior mapping Gross motor movement jump/step/turn/flick
games to improve computational thinking. A review of the literature on Fine motor movement grab/grip/twist/slide
digital game-based learning and children game-based teaching reveals
Computational thinking TUI perception TUI behavior mapping
that games could improve the study interest and outcome of preschool fundamental concepts mapping
children (Hogle, 1996). The main design concept includes three aspects:
Sequences vision: color gross motor movement: jump/
(1) curricula designed on the three dimensions of computational
tactile sense: click/ step/turn/flick
thinking proposed by Brennan and Resnick (2012), i.e., computational slide fine motor movement: twist
concepts, computational practices, and computational perspectives; (2) Parallelism vision: color/location
framework constructed on Input–Process–Outcome game model; (3) tactile sense: click
appropriate tangible user interface (TUI) adopted or designed for game- Events vision: color/
location/appearance
based teaching to aid preschool children in their acquisition of compu tactile sense: click
tational thinking and to function as their instructional scaffold. Loops vision: color/shape gross motor movement: jump/
tactile sense: click/ step/turn/flick
3.2. System planning slide fine motor movement: grab/
grip
Conditionals vision: color/ gross motor movement: jump/
This section discusses the curricular design and game design. The location/appearance step/turn/flick
curricula primarily follow the three dimensions of computational tactile sense: click fine motor movement: twist/
thinking by Brennan and Resnick (2012): computational concepts, Operators vision: location slide
tactile sense: click
computational practices, and computational perspectives, to enable
Data vision: color/position
preschool children to grasp the logic of programming. Additionally, tactile sense: click
contents for different study-levels are designed based on the 5E learning
cycle. The concept of a learning cycle came from Piaget’s theory of
cognitive development, which are teaching strategies (Llewellyn, 2005). theory for children; moreover, the content divides the curricula into
The proposed curricular design scheme is as shown in Fig. 2: several steps: engage, explore, explain, elaborate/extend, and evaluate.
The curricula are framed around computational concepts, where Computational practices and computational perspectives refer to the
several fundamental concepts are applied to the TUI designs to match problems encountered and solved by the learners during game-based
the participants’ knowledge levels, including sequences, loops, paral learning. Their elements could be used for game design, such as
lelism, events, conditionals, operators, and data concepts. As the par increasing the difficulty of the game, adding passing criteria and re
ticipant’s abilities of cognitive perceptions and motor movements strictions, or breaking a problem down into several smaller ones as game
greatly contribute to the computational thinking performance in this checkpoints, i.e., creating individual checkpoints or a series of related
study, we therefore used the perception and behavior mapping ap checkpoints. The process includes tasks to be completed, and preschool
proaches in the TUI designs (Table 4). The perception mapping includes children are guided to identify the problems and identify solutions.
visual perception and tactile perception. The visual perception is based Teachers, students, and teams should communicate and ask/answer
on the properties of appearance, position, location, shape, size, distance, questions to accomplish the goal of the game, encouraging each person
and color; on the other hand, the tactile perception includes sliding, to speak out his/her ideas. The game is designed with the Input–Pro
clicking, and touching senses. The behavior mapping contains gross- cess–Outcome game model and according to the game features put
motor (e.g., jumping, stepping, turning, and flicking) and fine-motor forward by Prensky (2003). The game design scheme is shown in Fig. 3.
movements (e.g., pinching, grabbing, gripping, and twisting). For The three stages of input, process, and output are discussed as
example, to complete a sequence concept, the participant has to first use follows:
color and click/slide as the visual and tactile senses in TUI perception,
and then integrate these perception concepts to jump/step/turn/flick in 1. Input:
the gross-motor movements and twist in the fine-motor movements. This stage includes the curricula and game features. The curricula
The TUI is embedded into game-based learning by designing games are developed around the computational thinking framework in
that suit the tangible interface. The actual teaching content is developed curricular design according to the steps of the 5E learning cycle and
based on the 5E learning cycle according to the cognitive development
Fig. 2. Schematic illustration of the curricular design.
5
Fig. 3. Schematic illustration of the game design.
with tangible user interface as the instructional scaffold for preschool output is related to computational concepts. Affective learning
children, as shown in Fig. 4. output relates to computational perspectives, which is the expression
2. Game cycle: of thoughts by students on their work. Fundamental concepts of
In this step, a game cycle is formed by user judgment, user action, programming, which include declarative, procedural, and strategic,
and system feedback. The user’s judgment, as indicated in the are acquired through game-based learning. The first refers to the
Input–Process–Outcome game model of Garris et al. (2002), consti information and facts required to perform tasks. The second is the
tutes interest, enjoyment, task involvement, and confidence. The programming knowledge on the procedure to perform the tasks. The
preferences and tastes of preschool children are demonstrated third is the application of programming concepts learnt, to different
through games; therefore, the game-based learning of this study circumstances in order to arrive at new programming rules for gen
takes into account the learning characteristics of these children and eral or new scenarios. The correspondence between the psychomo
adopts the experiential learning by John Dewey to develop an tor, cognitive, and affective learning results acquired in the game
instructional scaffold, used with an appropriately interactive cycle and the three dimensions of computational thinking is sum
tangible user-interface, such that the learning interest of preschool marized in Table 5.
children is kindled during the course of the game. As mentioned by
Garris et al. (2002), user action is realized in the design of course
content and game features, which include fantasy, rules, targets, Table 5
sensory stimulation, challenge, mystery, and controllability. The The three dimensions of computational thinking.
system feedback affects the actions of the users through “learning by Computational Ability Learning Outcome
doing.” The feedback received from students is used to influence user thinking outcome performance
judgments in the subsequent rounds. As a result, students exhibit Computational Fundamental Cognitive Declarative
higher motivation to learn, and the memory and training of pre thinking programming learning knowledge,
school children are reinforced in this iterative process to achieve the concepts (CT-C) concepts such as outcome procedural
sequences and knowledge,
learning goal. looping. strategic
3. Outcome: knowledge
The expected results at the end of learning program-logic can be Computational Problem solving Psychomotor Skill-based
observed in three aspects: psychomotor, cognitive, and affective. thinking skills such as testing learning learning results
practices (CT-P) and debugging. outcome
These results should be achieved through the teaching method pro
Computational Expressing/ Affective Confidence, self-
posed in this study. Psychomotor learning output pertains to thinking connecting/ learning efficacy, attitude,
computational practices, which is the application of skills learnt perspectives questioning outcome and preferences
during programming to the solving of problems. Cognitive learning (CT-V)
Fig. 4. Incorporating computational thinking into the input stage through 5E learning cycle.
6
To improve the computational thinking capability of preschool 1. Physical objects

children, curricula for the teaching of programming are developed on Instead of a virtual GUI, the learners are able to interact with the
the basis of the Input–Process–Outcome game model by Garris et al. physical objects via the TUI, including visual (e.g., colors), haptics (e.
(2002). Computational concepts, computational practices, and compu g., textures), audio (e.g., tones), and spatial properties (e.g.,
tational perspectives are taught in the course, with “learning by doing” locations).
as the learning strategy that fits into the cognitive features of preschool 2. Digital objects
children and a suitable TUI as the instructional scaffold to aid study. The Digital objects are also known as virtual entities, which include
comprehensive teaching and game designs are shown in the conceptu visual, audio and spatial effects. The hybrid TUI environment pro
alized game-based learning system exhibited in Fig. 5. vides multi-touch functions that allows the learners to directly
In the cognitive process during preschool children’s learning interact with the digital objects.
journey, teaching is dictated by cognitive level and requires the support 3. Actions
of instructional scaffold. Based on cognitive theory, this study uses The learner’s perceptual behaviors (such as manipulating a phys
tangible user interface (TUI) as the instructional scaffold for preschool ical object) can be transferred to a TUI system.
children to enhance computational thinking through the teaching of 4. Informational relations
program logic. According to Hiroshi Ishii and Ullmer (1997), TUI has the The informational relations include the mapping relation between
following characteristics: physical representations being coupled with physical objects and virtual objects, as well as the resulting behaviors
digital information, tangibles embodying operational information and between the real world components and simulated entities. For
feedback, physical representations embodying mechanisms for interac example, if a file folder is represented by a water bottle, twisting the
tive control, and physical representations perceptually coupled with bottle cap corresponds to opening the file folder.
actively mediated digital representations (images, sounds, etc.). As 5. Learning activities
discussed by Antle (2007) on children, the relations between TUI The assigned learning activities greatly influence the learners’
physical and digital representations are classified into perceptual map behaviors in interacting with the TUI.
ping, behavior mapping, and semantic mapping. In this experiment, we
design an interactive interface for the learning of program logic in These five components are developed by the cognitive and learning
preschool children based on the above principles and characteristics of theories, which can serve as the guidelines for the designs of learning
TUI. The course work targets computational thinking and teaches pre activities, as well as the architecture of a TUI. In addition, the guidelines
school children program-logic in a fun, game-based method. evaluate a TUI based on its relevance and usefulness. Table 6 summa
rizes the theoretical perspectives and provides 13 design guidelines in
3.3. Design of interactive tangible user interface TUI elements (Antle and Wise, 2013). For example, the guideline-1
(distributing information across modalities can increase effective
To examine the interactive behaviors between the TUI and learners working memory capacity) involves physical and digital objects, where
(Fig. 6), Antle and Wise (2013) suggest five interconnected components: using the physical and digital components allow the learners to decrease
Fig. 5. Conceptualization of game-based learning system.
7
Fig. 6. Interaction between TUI and learners (Antle and Wise, 2013).
in Fig. 7.
Table 6
Physical and digital objects are designed according to the first
Design guidelines in TUI for learning activities (1: physical objects, 2: digital
guideline of using multiple perceptions to enhance learning and mem
objects, 3: actions on objects, 4: informational relations, 5: learning activities).
ory. Visually perceived color is used in this study to define program
Guidelines, cited from (Antle and Wise, 2013) 1 2 3 4 5 instructions, and the children place tactile colored-cards and answer-
1. Distributing information across modalities can increase X X cards in response. Based on the second guideline of reducing excessive
effective working memory capacity. cognitive load, multiple inputs are summarized and simplified into four
2. Integrating spatial sources of information across and X X
types of robot motions, i.e., forward, backward, turn left, and turn right.
within modalities can minimize the extraneous cognitive
load imposed to synthesize inputs. These have their corresponding components in the colors and program
3. Using concrete representations can support interpretation X X X and can be applied with guideline 10, which permits the arbitrary
of symbolic representations of abstract concepts. matching of objects to support the creative thinking of preschool chil
4. Making mappings between the form and behavior of X dren. Guideline 3 is the use of specific representations for abstract ideas,
physical and/or digital objects and real-world entities
coherent can reduce extraneous cognitive load.
where Arduino robots are employed as specific representations to
5. Creating contextualized tasks or personal objects can X X showcase the results contributed by the children. The five elements for
support learners in forming individually meaningful goals the interactions with the tangible user interface depicted in Fig. 7 are
for interacting with the TUI. summarized as follows:
6. Using spatial, physical, temporal or relational properties X X X X X
can slow down interaction and trigger reflection.
7. Distributing parts of mental operations to actions on X X 1. Physical objects
physical and/or digital objects can simplify and support The design of the input interface adopts the visual attribute of
mental skills. color for preschool children, who are visually sensitive and whose
8. Leveraging image schemas in input actions can improve X hand muscles are under development. Colored cards of A4-size and
usability and system learnability.
9. Using conceptual metaphor(s) based on image schemas to X
10 × 10 answer cards are arranged and combined. The output
structure interaction mappings may bootstrap learning of interface uses the Arduino robots to display the program
abstract concepts instructions.
10. Designing objects that allow for spatial reconfiguration X X 2. Digital objects
can enable mutual adaptation of ideas.
The program instructions are split into forward, backward, left
11. Creating configurations in which participants can monitor X X
each other’s activity and gaze can support the turn, and right turn, based on the movement of Arduino robots.
development of shared understandings. Preschool children interact in their study through game-based
12. Distributing roles, information and controls across the X X learning and the game levels are set based on the Input–Pro
TUI learning environment can promote negotiation and cess–Outcome game model. Guideline 8 outlines the use of graphics
collaboration
to enhance the usability of inputs and their compatibility with
13. Creating constrained or co-dependent access points X
schemes can compel learners to negotiate with each other. learning. In this study, the graphics are simplified to a single color to
suit the study subjects. Guideline 13 lets the students discuss among
themselves in a restricted or dependent manner. To promote inter
the cognitive load and extend the working memory capacity. action between preschool children, we design game levels so that
Studying program logic through TUI without increasing the cogni they can discuss and encourage one another.
tive load for learning and also aids preschoolers in acquiring computa 3. Actions
tional thinking (Lin, 2015). This study uses modified Arduino robots to The students arrange and place colored cards into different com
design interactions with TUI suitable for preschool children based on the binations to accomplish the tasks of the game, and observe the out
thirteen TUI design guidelines (mentioned in the teaching methods) and puts of Arduino robots to amend the execution results. Guideline 4
the cognitive limitation of children. The overall interactions are shown and 7 show the information mapping between physical and digital
8
Fig. 7. Interactions with tangible user-interface.
objects. The program instructions forward, backward, left turn and

Table 7
right turn of Arduino robots are mapped to the colors red, yellow,
Design of TUI interactions.
blue, and green, respectively, and the symbols are limited to a single
color to reduce the cognitive load. From guidelines 3 and 9, the Physical Digital objects Action Informational
objects relations
mapping is classified into perception and behaviors. Because pre
school children have limited vocabulary, the TUI here is constructed Input red cards programming- Select and Perceptual mappings:
based primarily on perceptual and behavior mappings, tailored to sequences – place red – forward
forward
the cognitive characteristics of the students. yellow programming- Select and Perceptual mappings:
4. Informational relations cards sequences – place yellow – backward
To accommodate the cognitive level and locomotion capability of backward
preschool children in perceptual mappings, colored cards are adop blue cards programming- Select and Perceptual mappings:
sequences – left place blue – left
ted to satisfy the visual requirements. The red, yellow, green, and
green programming- Select and Perceptual mappings:
blue cards are assigned as forward, backward, left turn, and right cards sequences – right place green – right
turn movements, respectively. In behavior mappings, preschool Output Arduino programming- Execution Behavior mappings:
children arrange the colored cards to exhibit their thought process on robots execution order of colored cards
the causal relationship of an event. The combination of these cards – robots
represents the mechanism by which preschool children solve the

game tasks. After the cognitive construction of preschool children, scaffold is
adopted to aid their learning, with teaching assistants and teachers
The above physical and digital objects, actions, and informational supporting the scaffold as per design guideline 2 and 4. When helping
relations are summarized in Table 7 below. with the input interface, the teaching assistants use cell phones or tablets
To comply with guideline 5, i.e., supporting learners to achieve the to operate image recognition APP, and put the instructions by students
learning goal with scenarios, students learn in this study through game- as arranged and combinations of colored cards into Arduino robot.
based learning, with game levels designed on the Input–Pro
cess–Outcome game model. According to guidelines 6 and 12, the per
formance of tasks in this study enables collaboration among learners. 3.4. Learning Activities-Game levels
5. Learning activities The learning activities are implemented with the 5E learning cycle
The design of learning activities affects the students in their use of into game-based learning. The programming course teaches program
the tangible user interface, i.e., human–computer interaction. In this logic to preschool children through games. Three appropriate game
study, game levels are set according to the Input–Process–Outcome levels are planned to satisfy the attention and cognitive levels of pre
game model. Details of the game level design are described in the school children. The first explains the way robots move. Preschool
next section. children bring their life experience into the game through role-play,
where they understand the operation of robots and are engaged more
9
attentively to learn the basic logic and concepts of programming. The Table 9
second is challenge game, where tasks, such as finding the shortest different game levels in 5E learning cycle.
distance or being the first to reach the destination, are assigned. Addi Game level Game outline 5E strategy Activity
tionally, the accuracy of program instructions are gradually improved in
Level 1 – Robot Simulating the robot 1E engage Arousing interests
the game cycle, such as moving a few steps forward and walking to the simulation Initiating the robot through games
left. The third level focuses on debugging and analyzing problems to game Explaining program
identify the errors in logic instructions. The game levels are shown in logic and robot
Table 8 below. construction with
picture cards
The course content is embedded in the game-based learning model Level 2 – Reaching the 2E explore Motion of robots
and illustrated from the aspects of teaching and learning. Learning is Challenge destination on time 3E explain Discussing the ways of
initiated in preschool children to study program logic through game- game via the shortest path robots’ moves
based methods. The three game levels are explained with 5E. In the Level 3 – Task Identifying and 4E Task execution
game correcting the elaborate/ Reviewing and
first level, preschool children study program logic, using picture cards to
wrong path extend supervising the
initiate robot movements, and understand the mechanisms of robot 5E evaluate children’s problem
movements by simulating robots. Logic of sequences is also taught. In solving processes
the second level, preschool children execute tasks, such as directing
robots to reach the destination and move accurately. The third level is to
correct the mistakes and identify the reasons behind them. The above Table 10
are summarized in Table 9. 5E learning cycle in teaching activity.
At the engagement stage of the 5E learning cycle, picture cards are
5E strategy Game function/ Teaching activity
used with questions to guide the preschool children in their under interface
standing of robot movement mechanisms and their attempt to provide
1E – Engage Level 1: colored Guiding problem-solving
answers. In the exploration stage, they are permitted to investigate cards + Arduino Warm-up activities
freely. In the explanation stage, preschool children state the challenges robots
they encounter and discuss together. However, standard answers are not 2E – Explore Level 2: colored Explaining program sequences with robot
offered, such that they could be guided to arrive at their own solutions. cards + Arduino movements
robots Let preschool children play freely
In the elaboration/extension stage, they are encouraged to apply the
3E – Explain Level 2: colored When a robot executes a task, let
program logic learned. In the evaluation stage, open questions are asked cards + Arduino preschool children state the difficulties
to let the children express their feelings, thoughts about what they have robots encountered
learnt, etc. The roles that teachers play at each stage and responsibilities 4E – Level 3: Arduino When a robot executes a task, let
are shown in Table 10 below. Elaborate/ robots preschool children make flexible use of
extend the fundamental skills learned
The three game levels and the 5E learning cycle ensure participation, 5E – Evaluate Level 3: Arduino Asking open questions to let preschool
exploration, interpretation, elaboration, and evaluation at the respective robots children express themselves
stages of learning. Two experimental studies are conducted, which
contribute to different learning objectives, leading to different game
levels, class content, and computational thinking abilities (Table 11). completion results. In this method, a suitable user interface can be
In this instructional model, the operation of Arduino robots tangibly selected as required, and the game levels can be designed according to
displays the input and output information. The teaching focuses on the the course content and interface.
procedures and methods employed by preschool children during the
learning process to achieve the game goals and to overcome the chal 3.5. Hypotheses
lenges encountered. Their grasp of program logic is evaluated by task-
The designs of the proposed research as well as the experimental
Table 8
studies were based on the aforementioned computational thinking ele
Game level setups. ments (CT-C, CT-P and CT-V) and 5E learning cycle, in which these
conceptual frameworks have been suggested to enhance learners’
Game level Design goal Design function Literature
background
computational thinking ability as well as their learning attitudes and
behaviors. Therefore, based on these theoretical guidelines, we form the
Level 1 Robot Scenario simulation; Introducing the Game feature-
following hypotheses:
simulation role projection; game Fantasy
game familiarization with Input interface- Children cognitive H1: the developed game-based learning approach along with the TUI
interface colored cards feature-mimicking applications will enhance the preschool learners’ computational
TUI-perceptual thinking ability.
mappings-vision
H2: the developed game-based learning approach along with the TUI
Level 2 Strengthen logic; Practicing Game feature-
Challenge mode of action; Time limit/getting Control
applications will encourage the preschoolers to actively participate in
game team work around barrier/ Game feature- the course activities, as well as increase their learning interests.
shortest distance/ Challenge H3: the effects of learning experiences will not only influence the
minimum Children’s learners’ task performance, but also change their learning intentions and
instructions cognitive feature-
behaviors.
Output interface- perceptual
colored cards dominance/ The influences of the above hypotheses will be examined via the
limited attention empirical user studies, where the details can be found in the next
TUI-perceptual section.
mappings- vision
Level 3 Task Analyzing problems; Locating mistakes Game feature-
game finding mistakes Output interface- Goals 4. Methods
colored cards Game feature-
Mystery This study develops a learning framework (game-based learning
TUI-perceptual along with tangible user interface) to enhance preschool learner’s
mappings- vision
computational thinking abilities as well as improve their learning
10
Table 11
Game levels and teaching strategy.
5E learning cycle Game level Experimental schedule Teaching content Computational thinking (CT) ability
1E – Engage Level 1 1st study Introduction and explanation Fundamental capability

Simulating/exploring
2E – Explore Level 2 1st study Sequences CT-C

3E – Explain Level 2 1st study Expressing/connecting CT-V
Level 2 2nd study Initiation on condition CT-C
4E – Elaborate/Extend Level 3 2nd study Scoring/gaining experience CT-P

5E – Evaluate Level 3 2nd study Testing/debugging CT-P
Level 3 2nd study Expressing/connecting CT-V
achievements. The teaching methods integrate the course materials

Table 12
(computational thinking concepts, practices, and perspectives) with the
Independent and dependent variables.
TUI components to enhance participants’ cognitive, psychomotor, and
affective abilities. The overall experimental design follows the Input– Independent variable Description
Process–Outcome game model (Fig. 8), along with the 5E learning cycle TUI applications Applying the programming logic in the TUI system
and TUI interactions. The TUI system is used in the empirical user allows the participants to develop their cognitive
studies, which adopts Arduino robots as the learning smart toy to enable abilities and facilitate their learning activities, resulting
in different computational thinking abilities and
the participants to study the computational thinking knowledge and to learning interests.
practice the problem-solving skills via the physical device. The cogni Knowledge levels The 1st and 2nd studies are conducted one week apart,
tive, psychomotor, and affective abilities are developed from perceiving where the knowledge differences may contribute to
different course materials and the learning outcomes are examined in different learning strategies and/or behaviors.
various levels of questions. Dependent variable Description
And as the research focuses on examining how the proposed learning Computational thinking Three levels of logical questions (easy/normal/difficult)
scheme affects the learner’s cognitive knowledge and behaviors, the abilities are used to examine the participants’ cognitive abilities.
independent variables (IV) include the use of TUI application and the Learning interests Three types of learning behaviors (bystander/alone/
participants’ knowledge levels, whereas the dependent variables contain collaborate) are identified to indicate the participants’
learning interests.
learners’ computational thinking ability and learning interest
(Table 12).
components in order to correct or revise a robot’s path. Through this
4.1. Course materials process of method, it can support the importance of studying program
ming lessons for preschool children (Lindberg et al., 2019).
The course materials first introduce the basic computational thinking
concepts (CT-C) to engage (1E in Fig. 8) participants’ attention to learn 4.2. Apparatus- tangible user interface (TUI)
the fundamental logic of programming skills (e.g., sequence in a pro
gram) as well as to strengthen cognitive ability. The middle level focuses The TUI testbed system contains two major components, a cardboard
on computational thinking practices (CT-P) that applies explore (2E) with answer cards in the planning phase and a robot car with color cards
and explain (3E) strategies and involves in testing and debugging in the testing phase. While learning the computational thinking
practices to increase psychomotor ability. In this phase, the participants knowledge, the students have to apply the programming logic to guide a
have to understand the course materials and transfer the cognitive robot car to reach the assigned destination. In the initial planning stage
knowledge into physical behaviors to control the TUI components. The (Fig. 9a), the participants paste the answer cards (Fig. 10) in the grid to
advanced level emphasizes on computational perspectives (CT-V) and develop a sequence of routes, which allow them to effortlessly create or
exerts elaborate/extend (4E) and evaluate (5E) strategies to enhance revise the paths. Once finished, the paths are examined in the real
affective ability. The participants have to understand a variety of environment (Fig. 9b) to test whether the robot car can avoid all the
computational concepts and comprehend the relations among the TUI potential obstacles (i.e., bricks) and reach the destination (Fig. 9c).
Fig. 8. Input-Process-Outcome Game Model.
11
Fig. 9. TUI across all the experimental phases.
color and respond to the direction. For example, a red color card rep
resents going forward, whereas a yellow color card indicates the back
ward movement. This approach allows the participants to not only
practice their computational thinking concepts in the longitudinal
fashion but also develop the psychomotor skills by matching the
cognitive elements with physical objects.
Fig. 10. Color cards for directing a robot’s movements. 4.3. Assess learning performance
In the testing phase (Fig. 9b & c), the students need to place the color To evaluate the participants’ learning performance, 29 computa
cards (Fig. 10) on the map to direct the robot car’s movements (e.g., tional thinking related questions are retrieved from Bebras (n.d.). These
initiate or change paths). The robot car utilizes a sensor to measure the questions are adopted in our study to assess the students’ learning per
formance after perceiving the course materials and interacting with the
Fig. 11. Questions applied to evaluate students’ learning performance. “A” represents the starting point, “B” indicates the destination, and the red “X” signifies the
obstacles that blocks a robot’s path. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
12
TUI components (cardboard, answer cards, and the Arduino robot car). potentially experiencing fatigue from 40-minute experimental sessions,
To better match the participants’ cognitive capabilities, the questions the first and second studies are conducted one week apart. This also
are categorized into three levels, easy (Fig. 11a), normal (Fig. 11b), and allows us to examine the learning effects between the two experimental
difficult (Fig. 11c). Game level 1 (i.e., easy phase) concentrates on studies. It is also supported by the theory proposed by Lye and Koh
engaging (1E) the participant’s attention in solving the task of driving (2014) to teach learners explain their reasoning through concise
the robot car from point A to point B. These easy questions are used to communication to solve their problems.
ensure the participant realizes the basic control mechanisms underlie The students’ improvements of computational thinking capabilities
the TUI system. After familiarizing with the basic concepts, game level 2 are evaluated twice, before and after the experiments. Since conven
provides more challenging tasks to enhance learners’ computational tional assessments (e.g., final scores) provide little information about
thinking abilities, where the normal question requires the participants to learners’ underlying cognitive strategies and may fail to reflect their true
use explore (2E) and explain (3E) strategies and apply higher level of intents, the study is evaluated by multiple assessments including the
computational skills to resolve the task and driving the robot car to the correctness of answers and the observations of learning behaviors/in
destination without hitting any obstacles (indicated by the red X in the terests during the classes. In other words, the students’ learning activ
grid on Fig. 11b & c). Game level 3 (difficult question) focuses on ities while using the experimental testbed system are recorded to
improving the psychomotor and affective abilities, where an operator examine their learning awareness and information processing behav
has to elaborate/extend (4E) and evaluate (5E) the robot’s paths and iors. To assess the learning interest of students (i.e., second hypothesis)
avoid the potential flaws. In other words, the difficult question demands as well as examine the change of learners’ learning intentions and be
the students to identify the obstacles along a robot car’s path and haviors (i.e., third hypothesis), behavior observations are made at fixed
develop a path plan to guide the robot to the destination in a more time intervals to continuously record the game-related activities of the
complex condition. A sample answer of question 3 can be found in students. To evaluate the cognitive load, the event-based behavior
Fig. 11d. observation forms are used to record specific actions, such as idling,
giving up, refusing to learn, losing focus, talking about things unrelated
4.4. Participants to the class. The participants’ learning outcomes (including observed
learning behaviors and the correctness of answers) are the dependent
Seven kindergarten students participated in the experiments, with variables, and the TUI with game-based learning are the independent
ages ranging from 5 to 6 years old. The selection process of these seven variables.
students began with receiving permission from their respective parents.
Parents were informed of the research goals, experimental tasks, 5. Results and discussions
experimental procedures, types of data collection, methods, and mea
sures before providing consent. Parental consent letters were used to To examine the proposed teaching method, students’ learning be
document parents’ permission for their child to participate in this study. haviors/interests and learning performances are evaluated using quali
Parents were also informed of the potential risks associated with the tative and quantitative assessments. Seven students are recruited for this
experiment, including potential frustration, boredom or fatigue. Parents experiment, with two rounds of empirical studies conducted a week
were also ensured their child remained anonymous. In order to conceal apart from each other. The students’ computational thinking abilities
participants’ identities for security purposes, participants were are examined by three question assessments and their learning activities
randomly assigned user ID’s before conducting the experiment. There is are observed and recorded by two coders.
no record kept linking participants with their user ID, ensuring their
identity remains anonymous. Because students’ participation was 5.1. Computational thinking ability
voluntary, the students were allowed to refuse participation and the
students’ parents or kindergarten instructors were allowed to stop the The learning outcomes are measured by three levels of questions, in
study at any time. which the easy question includes simple questions to examine the
fundamental logic knowledge, the normal question assesses the psy
4.5. Experimental conditions chomotor learning outcomes by asking the users to match the concep
tual concepts with physical system components, and the difficult
The experiment follows a within-group design. Two studies are question requires the students to locate and correct potential mistakes to
conducted, each study includes a 20-minute teaching lesson and a 20- evaluate whether the participants can comprehend the relevant
minute question session. The participants first take question 1 and 2 in computational concepts. The first two questions (east and normal) are
the pretest to examine their computational thinking abilities before tested before the teaching sessions and the difficult question is examined
perceiving any teaching materials (Fig. 12). Question 3 aims to evaluate after perceiving the teaching materials and interacting with the TUI
the learning outcome and is therefore tested after taking the teaching system.
session. These questions aimed to validate first hypothesis (i.e., The results (Fig. 13) show while students first perceived the learning
computational thinking ability). In order to avoid participants from materials (1st study), only 1 out of 7 students completes the easy
Fig. 12. Experiment flow chart.
13
Fig. 13. Overall test results. Fig. 14. Probability of the observed learning behaviors.
question correctly and none of them respond to the normal and difficult can effectively encourage the students to collaborate with each other.
questions successfully. In the 2nd study, conducting one week after the Engaging a learner’s interest during the learning phases and direct
1st study, the results reveal most of the participants can answer the ing their attentions to the teaching materials are critical to facilitating
questions correctly across all different levels of questions. These results the overall learning performance. The bystander effect can be seen as a
suggest that the TUI system can effectively improve the learning per type of distracting behaviors, in which the students pay little interests to
formance as well as enhance students’ computational thinking ability. the course contents and show inappropriate activities instead, such as
idling, chatting with classmates, talking about irrelevant topics. The
observed results suggest the proposed TUI applications can effectively
5.2. Affective results – learning interest
attract learners’ attention and encourage them to devote their cognitive
resources to the appropriate affairs, leading to great improvements of
The participants’ learning interests can be observed through their
learning performances and low distracting behaviors. Additionally,
behaviors, which are identified into three major categories, bystander,
while experiencing the similar learning activities, the TUI approach can
alone and collaborate (Table 13). The bystander type indicates a student
successfully secure user attention to the suitable events and satisfy
who shows little interest to participate in the experiment or pay no
various learners’ behavioral preferences to match their needs. In fact,
attention to the course activities. The alone behaviors represent a user
based on our observations, most of the distracting behaviors are
who tends to work on the course materials by herself rather than discuss
occurred while the students are waiting in line for testing their answers
with other students. The collaborate type shows the participants prefer to
via the robot car on the map (Fig. 9b & c). In other words, these inactive
collaborate with each other and may share strategies to complete the
learning behaviors are not resulted from the TUI system itself.
course activities.
The participants learning activities are observed and encoded by two
6. Conclusions
coders every 3 min during the lecture. These variances can reveal the
participants’ strategies and behavioral differences while perceiving the
This study introduces game-based learning and tangible user inter
learning materials as well as interacting with the TUI system. To better
face (TUI) to enhance preschool children’s computational thinking
represent the distribution of the observed activities, the results are
abilities. The results show the proposed instructional methods are suit
computed and showed based on its probability (Fig. 14).
able for this purpose. The provided entertaining scenarios and user-
The results of the first study show 77% of the observed behaviors are
friendly interfaces are adopted to lower the learning difficulties and to
identified in the collaborate type and 23% belongs to the alone group.
increase the learning interests and behaviors of students such that their
When the students first attend to the lecture, due to the insufficient
computational thinking abilities could be improved. The learners can
computational thinking ability, they tend to work together and share
transfer the cognitive abstracts into physical behaviors through the in
their thoughts with each other. After the first lecture, in the 2nd study,
teractions with the TUI applications, thus enhancing their computa
the learning patterns reveal that more students prefer to work on the
tional thinking. Although smart toys can help the development for a TUI
course content by themselves, instead of discussing their ideas with
system, there are still many types of smart toys and various options for
others. The increased probability of alone type could be resulted from
designing learning materials to adapt learners’ age or knowledge dif
higher confidence in the course materials as the students have already
ferences. In this study, we used Arduino-based robot cars and color-
experienced the similar lecture a week ago in the 1st study. However,
labeled cards to develop the TUI applications. The developed frame
the result again shows more than half of the participants’ behaviors are
work (scenarios and systems) can benefit the research community in
falling into the collaborate type, which suggests the provided TUI system
improving preschoolers’ computational thinking without increasing
their learning difficulties.
Table 13
The learning outcomes are measured by three different levels of
Behavior classification.
questions, where the results suggested the game-based learning
Behavior Behavior type approach along with the TUI system can effectively improve the stu
code
dents’ learning performance as well as enhance their computational
Bystander Uninvolved in teaching activities: idle, look around, aimlessly walk thinking abilities. In other words, the participated preschool children’s
around, observe other students’ behaviors and provide no assistance
computational thinking was improved by learning program logic
Alone A student participates in the teaching events and work on the course
activities independently. The participate prefers to perceive the through the game-based learning materials and interactive interface.
teaching content by herself rather than discussing with others Our first hypothesis is therefore supported. Additionally, the students’
Collaborate The students tend to collaborate with each other to share their learning interests were evaluated by their engagement activities, where
thoughts and discuss the materials when taking the lecture. The both an individual’s behaviors and the group participations are exam
participants may act different roles or tackle different issues while
ined. The results again revealed the developed teaching approach can
working together
14
help the preschoolers to actively participate in the course activities as CRediT authorship contribution statement
well as increase their learning interests, which support the second hy
pothesis. Since the 1st and 2nd studies are conducted one week apart, Szu-Yin Lin: Conceptualization, Methodology, Data curation,
this allows us to examine the influences of participants’ learning expe Writing - original draft, Writing - review & editing. Shih-Yi Chien:
riences affecting their performance and behaviors. As expected, stu Methodology, Visualization, , Writing - review & editing. Chia-Lin
dents’ experience differences did affect their task outcomes, where Hsiao: Software, Investigation, Validation, Writing - original draft.
better performances were observed in the more experienced users. In Chih-Hsien Hsia: Writing - review & editing. Kuo-Ming Chao: Writing
addition, an experienced individual tends to work alone to solving the - review & editing.
tasks rather than collaborating with others. This may suggest once the
preschool learners believe they have sufficient knowledge regarding the Declaration of Competing Interest
task, they will have less incentive to share knowledge or cooperating
with others. These results also support our third hypothesis. In addition, The authors declare that they have no known competing financial
the transition between class activities and its effect on study-results interests or personal relationships that could have appeared to influence
could be further investigated. In terms of the cognitive load, as the the work reported in this paper.
game-based interactive interface is straightforward to operate, pre
school children familiarize themselves with it rapidly and are able to Acknowledgment
play while engaging in thinking; distraction usually occurs during the
wait time. Furthermore, other arrangements can be made in future to This research was supported by the Ministry of Science and Tech
permit better transition between activities. nology, Taiwan, under Grant MOST 109-2410-H-197-002-MY3 and
In terms of curricular planning, two experimental studies are con MOST 109-2410-H-004-067-MY2.
ducted in this study; here, two computational thinking elements are
implemented, i.e., sequences and events. More classes can be conducted
References
in future, and their duration could increase such that more aspects of
program logic can be covered. In terms of the instructional method, the Antle, A.N., Wise, A.F., 2013. Getting down to details: using theories of cognition and
teaching and testing are performed with the assistance of the game learning to inform tangible user interface design. Interact. Comput. 25, 1–20.
https://doi.org/10.1093/iwc/iws007.
cycle, in which advancement in game levels is achieved. In this way, the
Antle, A.N., 2007. Tangibles : Five Properties to Consider for Children. In: Workshop on
learners will have a sense of accomplishment. For the future, the game Tangibles at Conference on Human Factors in Computing Systems (CHI 2007). pp.
cycle can be used as a secondary assistance in passing the levels, such 1–10.
that students could have an opportunity to make amendments if they Bebras - International Challenge on Informatics and Computational Thinking [WWW
Document], n.d. URL <https://www.bebras.org/> (accessed 7.10.19).
commit mistakes. Cognitive features of preschool children in different Bergsmark, M., Fernaeus, Y., 2016. From patchwork to AppliquÉ: reflections from an
growth and development stages should also be analyzed while designing interaction design remake. In: Proceedings of the TEI ’16: Tenth International
teaching methods. The framework proposed here could be used to set up Conference on Tangible, Embedded, and Embodied Interaction, TEI ’16. ACM, New
York, NY, USA, pp. 236–244. https://doi.org/10.1145/2839462.2839473.
game levels to enable students to learn while playing on game-based Berk, L.E., Winsler, A., 1995. Scaffolding Children’s Learning: Vygotsky and Early
TUI. The present study adopts the input-process-outcome game model Childhood Education. NAEYC Research into Practice Series. Volume 7. National
to construct the research framework. The overall research model is Association for the Education of Young Children, 1509 16th Street, N.
Boss, S., Bruce, H., Case, C., Miller, K., 2001. Developing smart toys— from idea to
shown in Fig. 5. To use this proposed method, a researcher first needs to product. Intel Technol. J.
identify the course materials and expected outcomes, in which each Brennan, K., Resnick, M., 2012. In: New Frameworks for Studying and Assessing the
course content requires different TUI interaction input and output (as Development of Computational THINKING. American Educational Research
Association, Vancouver. Canada, pp. 1–25.
the perception and behavior mapping shown in Table 4) and can lead to
Bybee, W.R., Taylor, J., Gardner, A., Van Scotter, P., Carlson, J., Westbrook, A.,
different outcomes to enhance the learner’s ability. For example, the Landes, N., 2006. The BSCS 5E instructional model: origins, effectiveness, and
sequence concept is one of the CT-C contents, which uses colors/clicks as applications. BSCS 5, 88–98.
Chien, S.-Y., Hsiao, S., Kang, Y., 2018. In: Understanding Students’ Intentions in Learning
the input and flick/twist as the output to train a participant’s cognitive
Visual Robotic Programming: An Integration of IS Success Model, Utilitarian Value
ability. The proposed teaching approach has been proved that can and Hedonic Value. PACIS, p. 85.
effectively enhance students’ learning outcomes; however, there are Ching, Y.-H., Hsu, Y.-C., Baldwin, S., 2018. Developing computational thinking with
some limitations in this study. One of the major issues is the limited educational technologies for young learners. TechTrends 62, 563–573. https://doi.
org/10.1007/s11528-018-0292-7.
sample size. As the preschool students are the target users in this study, Delprino, F., Piva, C., Tommasi, G., Gelsomini, M., Izzo, N., Matera, M., 2018. ABBOT: a
this increases the difficulties in recruiting greater number of partici smart toy motivating children to become outdoor explorers. In: Proceedings of the
pants. In addition, since the preschool students can focus their attention 2018 International Conference on Advanced Visual Interfaces - AVI ’18. Presented at
the 2018 International Conference. ACM Press, Castiglione della Pescaia, Grosseto,
in a short amount of time, researchers may need to conduct several Italy, pp. 1–9. https://doi.org/10.1145/3206505.3206512.
separated studies to collect sufficient responses. Therefore, in this study, Denning, P.J., 2017. Remaining trouble spots with computational thinking. Commun.
two around of studies are organized, which inject extra challenges to ACM 60, 33–39. https://doi.org/10.1145/2998438.
Duran, L.B., Duran, E., 2004. The 5E instructional model: a learning cycle approach for
ensure the participants are available as well as willing to attend both inquiry-based science teaching. Sci. Educ. Rev. 3, 49–58.
experimental sessions. The limited experimental duration also restricts Furber, S., 2012. Shut Down or Restart? The Way Forward for Computing in UK Schools.
the number of measures applying to examine the learning activities and The Royal Society, London.
Garris, R., Ahlers, R., Driskell, J.E., 2002. Games, motivation, and learning: a research
outcomes.
and practice model. Simul. Gaming 33, 441–467. https://doi.org/10.1177/
As the results demonstrate the proposed method can encourage 1046878102238607.
learners to actively participate in the course activities and enhances Grover, S., Pea, R., 2013. Computational thinking in K–12: a review of the state of the
field. Educ. Res. 42, 38–43. https://doi.org/10.3102/0013189X12463051.
their learning performance, the developed teaching framework can
Healy, J.M., 2000. Failure to Connect: How Computers Affect our Children’s Minds–for
benefit the research community in enhancing preschool learners’ BETTER and Worse. Powells, Portland, Or.
computation thinking abilities. Future work will include further exam Hogle, J.G., 1996. Considering Games as Cognitive Tools: In Search of Effective
inations on various learning contents across different degree levels and “Edutainment”.
Horn, M.S., Solovey, E.T., Crouser, R.J., Jacob, R.J.K., 2009. In: Comparing the Use of
evaluate the improvements on learners’ cognitive thinking ability. To Tangible and Graphical Programming Languages for Informal Science Education.
verify how participants consume their cognitive resources, we plan to ACM Press, Boston, MA, USA, p. 975. https://doi.org/10.1145/1518701.1518851.
apply an EEG device to examine the brain activities and eye-tracker to Hung, P.C.K., Iqbal, F., Huang, S.-C., Melaisi, M., Pang, K., 2016. A glance of child’s play
privacy in smart toys. In: Sun, X., Liu, A., Chao, H.-.-C., Bertino, E. (Eds.), Cloud
detect the attention allocation. Computing and Security, Lecture Notes in Computer Science. Springer International
Publishing, pp. 217–231.
15
Ishii, H., 2008. Tangible bits: beyond pixels. In: Proceedings of the 2nd International Plowman, L., Luckin, R., 2004. Interactivity, interfaces, and smart toys. Computer 37,
Conference on Tangible and Embedded Interaction - TEI ’08. Presented at the 2nd 98–100. https://doi.org/10.1109/MC.2004.1266302.
international conference. ACM Press, Bonn, Germany. https://doi.org/10.1145/ Prensky, M., 2003. Digital game-based learning. Comput. Entertain. 1, 21. https://doi.
1347390.1347392. org/10.1145/950566.950596.
Ishii, H., Ullmer, B., 1997. Tangible bits: towards seamless interfaces between people, Primo Toys, 2016. Cubetto: A toy robot teaching kids code & computer programming
bits and atoms. In: Proceedings of the SIGCHI Conference on Human Factors in [WWW Document]. Primo Toys. URL <https://www.primotoys.com/> (accessed
Computing Systems - CHI ’97. Presented at the SIGCHI conference. ACM Press, 11.15.19).
Atlanta, Georgia, United States, pp. 234–241. https://doi.org/10.1145/ Rafferty, L., Hung, P.C.K., Fantinato, M., Peres, S.M., Iqbal, F., Kuo, S.-Y., Huang, S.-C.,
258549.258715. 2017. Towards a privacy rule conceptual model for smart toys. In: Tang, J.K.T.,
Kazemitabaar, M., McPeak, J., Jiao, A., He, L., Outing, T., Froehlich, J.E., 2017. Hung, P.C.K. (Eds.), Computing in Smart Toys, International Series on Computer
MakerWear: a tangible approach to interactive wearable creation for children. In: Entertainment and Media Technology. Springer International Publishing, Cham,
Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, pp. 85–102. https://doi.org/10.1007/978-3-319-62072-5_6.
CHI ’17. ACM, New York, NY, USA, pp. 133–145. https://doi.org/10.1145/ Rees, A., García-Peñalvo, F.J., Jormanainen, I., Tuul, M., Reimann, D., 2016. An
3025453.3025887. overview of the most relevant literature on coding and computational thinking with
Khine, M.S., 2018. Computational Thinking in the STEM Disciplines: Foundations and emphasis on the relevant issues for teachers.
Research Highlights. Springer. Relkin, E., Umaschi Bers, M., 2019. Designing an assessment of computational thinking
Kickstarter, 2017. ROBOPAL: Enrich your coding experience with stories [WWW abilities for young children. In: Cohen, L.E., Waite-Stupiansky, S. (Eds.), STEM in
Document]. Kickstarter. URL <https://www.kickstarter.com/projects/51942215 Early Childhood Education. Routledge, pp. 83–98. https://doi.org/10.4324/
1/robopal-build-stem-skills-build-robots-have-fun> (accessed 11.15.19). 9780429453755-5.
KinderLab Robotics, 2014. Robot Kits For Kids | KIBO. KinderLab Robotics. URL <https: Scharf, F., Winkler, T., Herczeg, M., 2008. In: Tangicons: Algorithmic Reasoning in a
//kinderlabrobotics.com/kibo/> (accessed 11.15.19). Collaborative Game for Children in Kindergarten and First Class. ACM Press,
Koslowski, B., 1980. Quantitative and qualitative changes in the development of Chicago, Illinois, p. 242. https://doi.org/10.1145/1463689.1463762.
seriation. Merrill-Palmer Q. Behav. Dev. 26, 391–405. Seehorn, D., Carey, S., Fuschetto, B., Lee, I., Moix, D., O’Grady-Cunniff, D., Owens, B.B.,
KUBO Education, 2017. KUBO Robotics - Discover how easy it can be to teach computer Stephenson, C., Verno, A., 2011. K–12 Computer Science Standards [WWW
science [WWW Document]. KUBO Education. URL <https://kubo.education/> Document]. Computer Science Teachers Association. URL <http://scratch.ttu.ee/fai
(accessed 11.15.19). lid/CSTA_K-12_CSS.pdf>.
V. Lin, V., 2015. Computational Thinking and Technology Toys. Honors Thesis Collection Shafto, S.A., 1986. Programming for learning in mathematics and science. In:
307. Proceedings of the Seventeenth SIGCSE Technical Symposium on Computer Science
Lindberg, R.S.N., Laine, T.H., Haaranen, L., 2019. Gamifying programming education in Education - SIGCSE ’86. Presented at the seventeenth SIGCSE technical symposium.
K-12: a review of programming curricula in seven countries and programming ACM Press, Cincinnati, Ohio, United States, pp. 296–302. https://doi.org/10.1145/
games. Br. J.Educ. Technol. 50, 1979–1995. https://doi.org/10.1111/bjet.12685. 5600.5635.
Llewellyn, D., 2005. Teaching high school science through inquiry: a case study Shute, V.J., Sun, C., Asbell-Clarke, J., 2017. Demystifying computational thinking. Educ.
approach. Corwin Press, Thousand Oaks (Arlington, Va.). Res. Rev. 22, 142–158. https://doi.org/10.1016/j.edurev.2017.09.003.
Lye, S.Y., Koh, J.H.L., 2014. Review on teaching and learning of computational thinking Smith, D.H., Hao, Q., Jagodzinski, F., Liu, Y., Gupta, V., 2019. Quantifying the effects of
through programming: what is next for K-12? Comput. Hum. Behav. 41, 51–61. prior knowledge in entry-level programming courses. In: Proceedings of the ACM
https://doi.org/10.1016/j.chb.2014.09.012. Conference on Global Computing Education - CompEd ’19. Presented at the ACM
Makeblock, 2015. Robot Kits for Kids : mBot | Makeblock - Global STEAM Education Conference. ACM Press, Chengdu,Sichuan, China, pp. 30–36. https://doi.org/
Solution Provider. Makeblock. URL <https://www.makeblock.com/steam-kit 10.1145/3300115.3309503.
s/mbot> (accessed 11.15.19). Sullivan, A., Bers, M.U., 2018. Computational thinking and young children:
Makeblock, 2017. Codey Rocky | Makeblock - Global STEAM Education Solution Understanding the potential of tangible and graphical interfaces, in: Teaching
Provider. Makeblock. URL <https://www.makeblock.com/steam-kits/codey-rocky> Computational Thinking in Primary Education. IGI Global, pp. 123–137.
(accessed 11.15.19). Sullivan, A.A., Bers, M.U., Mihm, C., 2017. In: Imagining, Playing, and Coding with
Oh, H., Deshmane, A., Li, F., Han, J.Y., Stewart, M., Tsai, M.Y., Xu, X., Oakley, I., 2013. KIBO: Using Robotics to Foster Computational Thinking in Young Children. Siu-
The digital dream lab: tabletop puzzle blocks for exploring programmatic concepts. cheung KONG The Education University of Hong Kong, Hong Kong, p. 110.
In: Proceedings of the 7th International Conference on Tangible, Embedded and Wang, D., Zhang, L., Xu, C., Hu, H., Qi, Y., 2016. A tangible embedded programming
Embodied Interaction. ACM, pp. 51–56. system to convey event-handling concept. In: Proceedings of the TEI ’16: Tenth
Papastergiou, M., 2009. Digital Game-Based Learning in high school Computer Science International Conference on Tangible, Embedded, and Embodied Interaction, TEI
education: impact on educational effectiveness and student motivation. Comput. ’16. ACM, New York, NY, USA, pp. 133–140. https://doi.org/10.1145/
Educ. 52, 1–12. https://doi.org/10.1016/j.compedu.2008.06.004. 2839462.2839491.
Pedersen, B.K.M.K., Andersen, K.E., Jorgensen, A., Koslich, S., Sherzai, F., Nielsen, J., Wing, J.M., 2006. Computational thinking. Commun. ACM 49, 33. https://doi.org/
2018. Towards playful learning and computational thinking — developing the 10.1145/1118178.1118215.
educational robot BRICKO. In: 2018 IEEE Integrated STEM Education Conference Wonder Workshop | Home of Dash, Cue, and Dot – award-winning robots that help kids
(ISEC). Presented at the 8th IEEE Integrated STEM Education Conference (ISEC) learn to code [WWW Document], 2015. Wonder Workshop - US. URL <https://www
2018. IEEE, Princeton, NJ, pp. 37–44. https://doi.org/10.1109/ .makewonder.com/robots/dash/> (accessed 11.15.19).
ISECon.2018.8340502. Yadav, A., Gretter, S., Good, J., McLean, T., 2017. Computational thinking in teacher
Pérez-Marín, D., Hijón-Neira, R., Bacelo, A., Pizarro, C., 2020. Can computational education. In: Rich, P.J., Hodges, C.B. (Eds.), Emerging Research, Practice, and
thinking be improved by using a methodology based on metaphors and scratch to Policy on Computational Thinking. Springer International Publishing, Cham,
teach computer programming to children? Comput. Hum. Behav. 105, 105849 pp. 205–220. https://doi.org/10.1007/978-3-319-52691-1_13.
https://doi.org/10.1016/j.chb.2018.12.027. Zhang, L., Nouri, J., 2019. A systematic review of learning computational thinking
Piaget, J., 1976. Piaget’s theory. In: Piaget and His School. Springer, pp. 11–23. through Scratch in K-9. Comput. Educ. 141, 103607 https://doi.org/10.1016/j.
compedu.2019.103607.
16

Electronic Commerce Research and Applications

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Electronic Commerce Research and Applications

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Electronic Commerce Research and Applications

Uploaded by

Copyright:

Available Formats

Electronic Commerce Research and Applications 44 (2020) 101011

Contents lists available at ScienceDirect

Electronic Commerce Research and Applications

Enhancing Computational Thinking Capability of Preschool Children by

1. Introduction developing innovative knowledge in different domains (Furber, 2012).

objects. One of the notable developments, Augmented Urban Planning

model is also adopted to assist preschool children in acquiring compu­ Table 4

Perception mapping Visual appearance/position/

Fig. 2. Schematic illustration of the curricular design.

Fig. 3. Schematic illustration of the game design.

To improve the computational thinking capability of preschool 1. Physical objects

Fig. 5. Conceptualization of game-based learning system.

Fig. 7. Interactions with tangible user-interface.

objects. The program instructions forward, backward, left turn and

represents the mechanism by which preschool children solve the

1E – Engage Level 1 1st study Introduction and explanation Fundamental capability

2E – Explore Level 2 1st study Sequences CT-C

4E – Elaborate/Extend Level 3 2nd study Scoring/gaining experience CT-P

achievements. The teaching methods integrate the course materials

Fig. 8. Input-Process-Outcome Game Model.

Fig. 9. TUI across all the experimental phases.

Fig. 12. Experiment flow chart.

You might also like

model is also adopted to assist preschool children in acquiring compu Table 4