Skip to main content

The exploration of continuous learning intention in STEAM education through attitude, motivation, and cognitive load



This study proposes a learning cycle and a comprehensive research framework that integrates Bloom’s taxonomy: the cognitive domain (cognitive load), affective domain (attitude and motivation) and psychomotor domain (implementation of science, technology, engineering, arts, and math [STEAM] activities) to explore the relationship between these learning domains and learning intention. The proposed innovative mediated-moderation model includes second-order factors derived from the technology acceptance model (TAM) (perceived usefulness, perceived ease of use, and perceived enjoyment), the attention, relevance, confidence and satisfaction (ARCS) model, and cognitive load (mental load and mental effort) to explain the continuous learning intention of STEAM education.


A teaching material was designed for the STEAM activity, and an empirical experiment was subsequently conducted. The empirical experiment of STEAM activities with our design teaching material (micro:bit with artificial intelligence-based concept) was conducted at a university and an elementary school; a total of 145 questionnaire survey data were collected after the activities. University student participants were 20–24 years old and the elementary school student participants were at the K5–K6 level. The results showed that perceived usability directly influenced learning intention and strengthened the relationship between learning attitudes and intention. The ARCS plays a critical moderating role that positively influenced perceived usability and strengthened its effects on learning attitudes. Regarding the mediating effects, cognitive load negatively influenced perceived usability.


The findings of this study revealed that critical factors affect students’ learning attitudes and intentions regarding STEAM education. The theoretical and educational implications of these findings were proposed to future instructors.


  • The STEAM learning cycle was proposed to describe four phases for successful STEAM learning.

  • Affect factors of learning intention in STEAM education were investigated.

  • Motivation factors positively influences perceived usability and learning attitude.

  • Cognitive load has a significant negative effects influences perceived usability.

  • Perceived usability directly influences learning intention and strengthens the relationship between attitudes and intention.


Recently, the concepts of STEAM (science, technology, engineering, arts, and math) have been examined to determine the benefits of enhancing the cognitive domain of learning of memory, reaction time, and innate intelligence (Pabalan et al., 2018); psychomotor domain of physical measure, coordination, and skill (Ariyanto et al., 2019); and affective domain of self-confidence, self-motivation, collaboration, personal grooming, and time administration (Ramma et al., 2018). In cognitive domain studies, several previous studies examined the effects of STEAM in enhancing creative thinking (Bassachs et al., 2020; Land, 2013), career decision-making (Abe & Chikoko, 2020), cognitive structures for engineering design thinking (Lin, Chai, et al., 2021; Lin, Wu, et al., 2021), and cognitive appraisals and boredom (Ekatushabe et al., 2021). The psychomotor domain refers to skills parallel with physical growth and development to implement a specific task (Gülen et al., 2019). In the psychomotor domain, the relationship between STEAM education and learning performances, such as excitement, manual control, skill, fitting situations, and creating improvement, has been examined in previous studies (Bassachs et al., 2020; Marín-Marín et al., 2021). In the affective domain, several previous studies adopted the technology acceptance model (TAM) to explain the attitude toward continuous learning intention (Haji et al., 2017; Huang & Liu, 2021; Wu & Chen, 2017), and motivations for STEAM education (Conradty & Bogner, 2020). Thus, considering technology usage in enhancing self-directed learning (Curran et al., 2019), learning motivation (Dunn et al., 2019), and skill improvement (Radhamani et al., 2021), more theoretical and empirical research is essential to discover the influence of the ARCS (attention, relevance, confidence, and satisfaction) motivation model (Keller, 2009). Technology usage in STEAM education also has the benefits of decreasing cognitive load (mental load and mental effort) and improving students’ learning intentions (Cheng, 2017; Costley & Lange, 2017).

In addition, previous educational studies have provided evidence of the benefits of STEAM education in student learning (Gao et al., 2020), design thinking (Lin, Chai, et al., 2021; Lin, Wu, et al., 2021), and professional development (Shernoff et al., 2017). However, artificial intelligence (AI) application and needs represent the increasing trend of education because it advances the learners’ capabilities of learning analytics, which may increase their workplace competitiveness (Zawacki-Richter et al., 2019). However, the integration of the above TAM concepts and AI application functions in the education process has not been well examined (Gülhan & Şahin, 2018).

Bloom (1956) proposed a taxonomy of learning domains that identified three domains of educational activities: the cognitive, affective, and psychomotor domains. Recently, Bloom's taxonomy concepts have been widely used in many education studies, such as those applied in science and technology courses, measuring learning outcomes (Zorluoglu et al., 2019), social studies curriculum (Koc et al., 2020), and estimating the objects in information technology and software course curricula (Ocak et al., 2020). In other words, Bloom’s taxonomy is a useful guide to help educators develop curricula. The three critical attributes are as follows: first, the cognitive domain focuses on mental skills that involve the knowledge and development of intellectual skills (Bloom, 1956). The affective domain concerns growth in feelings or emotional status, such as attitude or motivation (Krathwohl et al., 1973). The psychomotor domain is related to physical skills (Simpson, 1972). Accordingly, based on Bloom’s taxonomy (1956), the learner learns new knowledge (cognitive domain), attitudes (affective domain), and skills (psychomotor) in a learning activity.

This paper proposes a STEAM learning cycle that reveals four phases (cognitive, psychomotor, and affective domains, and continuous learning intention) for successful STEAM learning.

In the first phase, a learner starts to learn knowledge (cognitive domain), and the high/low level of cognitive load will affect the learner’s outcome of implementation of steam activity in the next psychomotor domain phase (Fig. 1). First, from the cognitive domain viewpoint, cognitive development can be seen as the resource allocation of the knowledge development process (Lu et al., 2020; Yafie et al., 2020). In the second phase (psychomotor domain), a heavy cognitive load may negatively affect learning performance because individuals cannot handle the complexity and the infinity of the abundant information at once as an input (Kozlovskiy et al., 2021). In the third phase (affective domain), the successful implementation experience of STEAM activity in the second phase would increase the positive attitude and enhance the motivation towards continuous learning intention in the fourth phase. Continuous learning intention refers to the learners’ willingness to engage or attend STEAM activity in the future. The definition of continuance intention is derived from a previous study (Dai et al., 2020) that denotes the intention of students to continue learning STEAM after the teaching activity. The same idea has been adopted in learning intention assessment of AI-enabled application (Fu et al., 2020) and MOOC (Dai et al., 2020). In the fourth phase, a high level of STEAM continuous learning intention motivates the learner to start the next phase in the STEAM learning cycle. The successful implementation of STEAM activity positively triggers the learners’ affective domains, such as attitude and motivation toward continuous learning intention in the future.

Fig. 1
figure 1

STEAM learning cycle

Previous STEAM studies focus on the relationship between Bloom’s cognitive domain (such as six layers of cognitive domain) and learning performance. Most STEAM studies have focused on developing engineering design thinking, teachers’ perceptions (Margot and Kettler 2019), and the cognitive domain applied in learning performance (Fletcher, 2018; Gao et al., 2020). However, few studies have integrated affective factors in the affective domain and examined how to apply them in STEAM education. Affective domain assessment for STEAM is one of the most important assessments for STEM education (Gao et al., 2020). A previous study investigated attitudes, beliefs, motivation, and interest towards the intention of disciplines in STEM. The well-designed STEM activity not only increased students’ knowledge bust, but also their affective intention for STEM (Apedoe et al., 2008).

The STEAM learning activity’s emphasis on learning-by-doing usually requires learners to finish a specific task that takes a longer time to assimilate new things learned and apply them to daily life practices (Gao et al., 2020; Wahono et al., 2020). Therefore, affective factors such as attitude or motivation for enhancing learners’ patience to successfully complete learning tasks need to be clarified. In the psychomotor domain, learning objectives focus on behavioral changes and skill development. Skills denote the ability to physically manipulate or instruments to complete a specific task. We believe that the assessment of the psychomotor domain, such as implementing STEAM activity, also plays a critical role in continuous learning intention in STEAM education. Thus, this study proposes a comprehensive research framework that integrates Bloom’s (1956) taxonomy: cognitive domain (cognitive load), affective domain (attitude and motivation), and psychomotor domain (implementation of STEAM activity) to explore the relationship between these learning domains and learning intention. We adopted cognitive load factors from cognitive load theory to the cognitive domain, attitude factors (perceived attitude and learning intention) from TAM and motivation factors from Keller’s (1983) ARCS theory to the affective domain, and implemented micro:bit with AI learning activity for skill training assessment in the psychomotor domain to develop a comprehensive assessment framework for STEAM learning intention. The research framework is illustrated in Fig. 2. Micro:bit is an embedded system based on the Advanced RISC Machine (ARM) architecture, designed by the British Broadcasting Corporation (BBC) for use in computer education in the UK (

Fig. 2
figure 2

Research framework

Previous studies have considered the factors of active online interaction and collaboration to improve students’ learning performance in STEM education (Barrett et al., 2020; Granić & Marangunić, 2019; Huang & Liu, 2021). However, several gaps in the existing literature need to be addressed. First, the critical factors influencing students’ learning intentions and attitudes toward STEAM education are unclear. Second, only a few education studies have examined the multi-perspectives that include cognitive, affective, and psychomotor domains affecting the attitude and learning intention toward STEAM education. With the increasing attention in STEAM education development and application, these unsolved issues have become critically important and need to be addressed further. Therefore, this study introduces a new theoretical model (Fig. 2) that integrates the above-mentioned multi-learning domains to explain how and why these factors affect STEAM learners’ learning intention in order to address these research gaps. Therefore, this study aims to explore the latent factors that can enhance STEAM continuous learning intention. These not only contribute to the research on learning motivation factors of learners, but also improve the application of TAM, cognitive load theory, and ARCS theory in STEAM education. Furthermore, the moderation effect analysis allows this study to provide new insights for STEAM educators in evaluating and improving teaching activities.

Theoretical background and hypothesis development

Technology acceptance model (TAM)

The TAM theory was proposed to explore the behavior of user acceptance of information service systems based on the social psychology perspective (Mu & Jong, 2018; Zhang et al., 2019). Davis’s TAM proposes that perceived usefulness effects significantly influence users’ attitudes and behavioral intentions (Davis, 1989). The TAM theory, one of the most popular and important theoretical models regarding technology acceptance, provides a fundamental background for understanding individuals’ perception of specific technology acceptance behavior of intention, such as massive open online courses’ (MOOCs) continuance intention (Wu & Chen, 2017). Considering the adoption of TAM in Web 2.0 and technologies in education, students’ and teachers’ perceived usefulness of the technologies had a positive impact on their attitudes toward the intention of using Web 2.0. However, contrary to the original TAM, in a later TAM, perceived ease of use had no significant influence on perceived usefulness (Gyamfi, 2017).

The TAM has been adopted in STEM education studies to explore the effects of perceived usefulness, ease of use, and fun as predictors of behavioral intention and perceived attitude (Mutambara & Bayaga, 2021; Shiau et al., 2018). The TAM combined with innovation diffusion theory (IDT) has been applied to examine the innovation diffusion of OpenStreetMap in STEM education. The results suggested that STEM students’ perception of the usefulness of technology and their perceived attitude toward it leads to their intention to continue using the technology (Shiau et al., 2018). In STEM education in rural areas, the relationship between perceived usefulness and attitude is a critical factor in determining the acceptance of action learning (Mutambara & Bayaga, 2021). Davis (1989) showed that perceived ease of use affects attitude because if users find the system difficult to use, it creates user barriers and affects users’ behavioral intentions and attitudes. Additionally, the relationship between the perceived usefulness of technology and intention to use was investigated by studying highly immersive virtual reality systems. However, the research results showed that the perceived usefulness of the system has a weak influence on the intention to use the system (Quaid et al., 2020). Perceived usefulness and attitude are critical for MOOCs’ continuance intention; however, it was found that perceived ease of use does not influence attitude (Wu & Chen, 2017). Therefore, we created a second-order construct, perceived usability, that combined the three major constructs of TAM—perceived usefulness, perceived ease of use, and perceived enjoyment—to explore the relationship between perceived usability, attitude, and intention. Thus, the following hypotheses were proposed:

Hypothesis 1 (H1)

Perceived usability has a positive and significant effect on attitude.

Hypothesis 2 (H2)

Attitude has a positive and significant effect on intention.

Hypothesis 3 (H3)

Perceived usability has a positive and significant effect on intention.

Cognitive load

Cognitive load theory was first proposed by a cognitive psychologist based on empirical research on short-term memory ability (Sweller et al., 1998b). Cognitive load has been widely applied in education research since Sweller and colleagues (Sweller et al., 1998b) applied cognitive load to the field of teaching (Chen & Huang, 2020). Cognitive load consists of two constructs: mental load and mental effort. The higher the perception of learners’ mental load, mental effort, or task difficulty, the higher the perceived cognitive load for the learning process (Paas, 1992).

The relationship between cognitive load and learning performance has been examined in several previous studies (Chen & Huang, 2020; Liu et al., 2021). Regarding the effect of STEAM-based mobile learning on learning achievement and cognitive load, the research results showed that the learning effectiveness of the experimental group (game-based learning system) was superior to that of the control group, and that the experimental group (game-based system) generated a lower cognitive load than that of the control group (transportation vehicles) (Chen & Huang, 2020). Regarding the effect of cognitive load on learning with augmented reality (AR), learning with 3D technology, and traditional learning, the research results found that the AR group performed better than the 3D and traditional groups in their knowledge improvement, and the AR group students had the lowest cognitive load among the three groups. These results suggest that integrating AR in experiments that help students construct knowledge by providing a virtual–real fusion environment significantly reduces the cognitive capacity that students need to allocate to deal with learning tasks (Liu et al., 2021). Thus, most of the studies that focus on cognitive load theory examine the difference between the experimental and control groups through t-tests, and only a few studies directly explore the connection between cognitive load and STEAM learning intentions.

Additionally, a previous study aimed to identify the factors affecting higher education students’ behavioral intention toward learning management systems via the TAM, including perceived usefulness and perceived ease of use, and external factors, including self-efficacy, enjoyment, subjective norms, satisfaction, interactivity, and control. The research results confirmed that the relationships between the influencing factors provided insight into students’ behavioral intentions toward the use of learning systems (Findik-Coşkunçay et al., 2018). Therefore, based on the above-mentioned studies, we believe that cognitive load is negatively correlated with learning performance and intention. Additionally, learning intention is positively affected by perceived usefulness, perceived ease of use, and perceived enjoyment. Cognitive load plays a critical role in augmented reality (AR) continuous learning intention. A previous study showed that less cognitive load, stronger motivation, and more positive attitudes towards learning intention in AR (Cheng, 2017). High-quality instructional design enhances the level of germane cognitive load and continuous learning intention (Costley & Lange, 2017). A previous study examined the relationship between perceived cognitive load, motivation, attitudes, perceived usefulness, and learning intention in AR learning. Motivation factors mediated the relationship between learners’ cognitive load and learning intention (Cheng, 2017). Thus, the following hypothesis was proposed to examine the connection between cognitive load and STEAM learning intention:

Hypothesis 4 (H4)

Cognitive load has a negative and significant effect on perceived usability.

Attention, relevance, confidence, and satisfaction (ARCS) theory

The ARCS model was developed by John Keller to provide an instructional model to explain how to motivate learners and ensure the continuity of motivation during the teaching activity (Keller, 1983). This model has been widely applied in evaluating the motivation of e-learning and digital teaching material design areas in recent decades (Karakış et al., 2016). The ARCS model includes four components: attention, relevance, confidence, and satisfaction (Keller, 1983). Students’ attention can be attracted in two ways: (1) perceptual arousal: using amazing or remarkable points of interest to attract students’ interest, and (2) inquiry arousal: stimulating students’ curiosity by posing challenging questions or problems to be solved (Keller, 2009). Relevance aims to establish relevance regarding the target knowledge in order to increase learners’ motivation. Confidence describes the relationship between learners’ expectations of success and confidence levels during the learning process. Confidence helps learners understand their likelihood of successful learning and helps them avoid feeling that they cannot finish the learning objectives. Satisfaction aims at learning to be rewarding or satisfying, as a form of achievement. Learners should be satisfied with what they achieve during their learning activities (Keller, 2009). The ARCS theory argues that learners’ motivation can be increased if the teaching material satisfies the above-mentioned four components.

The ARCS theory has proven to be an effective tool for enhancing learner motivation and performance (Jason Bond Huett, 2006). The ARCS motivation model was used to investigate the effects of computer-assisted instructional materials designed for the ASSURE model (an instructional system) on students’ performance and attitudes in mathematics classes. The research results showed that computer-assisted instructional materials have a positive impact on students’ attitudes toward computer-assisted instruction and enhance their academic achievement (Karakış et al., 2016). Keller’s (1983) ARCS model was used to examine the effects of the four components on students’ attitudes toward the use of gamification for competency development in higher education in Spain. The study revealed that perceived attention, perceived relevance, and perceived confidence directly and positively influence students’ attitudes toward using online educational video games to develop competencies (Galbis-Córdoba et al., 2017). A previous study adopted the ARCS model to examine the influence of the gamified learning approach on science learning, achievement, and motivation by using a context-aware mobile learning environment. The ARCS model explains and verifies the effects of the four components on students’ motivation and learning achievement (Su & Cheng, 2015). For the problem-based learning (PBL) method, the ARCS model has been used to explore the relationship between learning motivation and entrepreneurial attitudes by focusing on the impact of the PBL method on the learning motivation of entrepreneurial attitudes (Munawaroh, 2020). Thus, the following hypotheses were proposed:

Hypothesis 5 (H5)

ARCS has a positive and significant effect on attitude.

Moderation effect

Affective domain assessment for STEAM is one of the most popular assessments for STEM education (Gao et al., 2020). A previous study investigated attitudes, beliefs, motivation, and interest towards the intention of disciplines in STEM. The well-designed STEM activity not only increased students’ knowledge, but also increased their affective intention in STEM (Apedoe et al., 2008). The STEM program activities can help students increase their intention and factor in the ARCS model, such as confidence, to improve STEM knowledge and skills. The students’ confidence in success in STEM strengthened the effects of perceived usability and attitude (Musavi et al., 2018). The ARCS motivation model reveals the key motivation factors adopted to explain the relationship between learners’ perceived usability and improvement of learning attitudes (Chang et al., 2019). Thus, the following hypotheses of moderation effects are proposed:

Hypothesis 6 (H6)

Perceived usability positively moderates the relationship between attitude and intention.

Hypothesis 7 (H7)

ARCS positively moderates the relationship between perceived usability and attitude.

This study proposed a conceptual research framework that includes three second-order factors derived from TAM (perceived usefulness, perceived ease of use, and perceived enjoyment), ARCS (attention, relevance, confidence, satisfaction), and cognitive load (mental load and mental effort) to explain the continuous learning intention of STEAM education. The proposed conceptual research framework is illustrated in Fig. 3 to address the research objectives.

Fig. 3
figure 3

Conceptual research framework


Population and sample

In this study, elementary and university students were recruited in the classroom after their learning activities. The participants voluntarily participated in the questionnaire survey. Similar participant selection and statistical methods were adopted in a previous study (Mutambara & Bayaga, 2021). A non-probability voluntary response self-selection sampling method was adopted to recruit students following the sampling procedure in a previous empirical study of undergraduate students (Barrett et al., 2021). The present study used two STEAM teaching activities to collect survey data. The data collection period was from February to May 2021, and 145 questionnaire responses were collected, yielding 141 valid responses. Participation in the questionnaire survey was voluntary (84 undergraduate students and 57 elementary students), and the participants completed the questionnaire after STEAM teaching activities.

We developed STEAM learning materials for elementary and college students that suited their level, although the experimental equipment used was the same. Since the goal of this study was to investigate participants’ continuous learning intention and not the effectiveness of their learning, to collect more samples and encourage more students to participate, we invited these two groups of students to verify our research model. The learning materials used in the study were micro bit and AI neural network programming, which has never been taught to elementary school or college students. The learning activity was designed to be more difficult for college students than for elementary school students. Primarily, the learning materials and activities were designed keeping in mind the difficulty that college students and elementary school students could accept.

We conducted two similar STEAM activities and used the same questionnaire to collect survey data after the STEAM activities in the university and elementary school. The two versions of STEAM activities have the same learning topic with different learning materials: a difficult one for college students and an easier one for elementary students. The participants voluntarily participated in the questionnaire survey. University student participants were 20–24 years old and the elementary school student participants were at the K5–K6 level. Participants from the K5–K6 level were selected because their knowledge level fits the teaching of AI-based knowledge for the syllabus in elementary school. We conducted STEAM learning activities on undergraduate and elementary students because we hoped our research results could have better interpretive results that would be suitable for university and elementary students. The instructions were the same for both the elementary and university students.


This study integrates three major theories—TAM, cognitive load theory, and ARCS theory—into the proposed STEAM research framework. The research framework consists of 11 constructs that include the learning intention of STEAM as the main construct and a second-order construct. The first part of the main construct was derived from TAM, which included perceived usefulness, perceived ease of use, perceived enjoyment, attitude, and intention. This study created a second-order construct, namely perceived usability, which includes three constructs (perceived usefulness, perceived ease of use, and perceived enjoyment) derived from TAM (Davis, 1989; Davis et al., 1989; Kanchanatanee et al., 2014; Teo, 2009; Weng et al., 2018). The cognitive load theory construct is divided into two sub-constructs: mental effort and mental load (Chen & Huang, 2020). The ARCS theory construct consists of four sub-constructs: attention, relevance, confidence, and satisfaction (SAT) (Karakış et al., 2016; Keller, 1983, 2009; Lin, Chai, et al., 2021). The model constructs, definitions, and references are listed in Table 1.

Table 1 Model constructs, definitions, and references

Design of AI-based STEAM education game and experiment procedure

In this study, STEAM teaching materials contained instructions regarding both STEAM and AI concepts. The STEAM teaching materials used in this study were designed based on the STEAM 6E framework. This six-step framework included engagement, exploration, explanation, engineering, enrichment, and evaluation to complete the STEAM learning task. Using these concepts, students designed an AI-based STEAM game. The game uses computer vision recognition techniques to automatically recognize a user’s hand gestures and control a robot to play a game of rock–paper–scissors with users. The details of the game are discussed below.

STEAM concepts

  1. 1.

    Science: The students were required to understand computer vision recognition theory and neural network theory to design and implement a neural network in order to develop an AI-based STEAM rock–paper–scissors game.

  2. 2.

    Technology: The students used Kittenblock, a block-based visual programming language development tool, to code the program for the AI-based STEAM game.

  3. 3.

    Engineering: In the process of assembling the game, in addition to understanding the structures of various parts of the micro:bit and motor, students had to use engineering concepts.

  4. 4.

    Arts: The Arts part of our STEAM project required students to design unique graphs on the micro:bit of the AI-based STEAM rock–paper–scissors game. In this study, the lecturer asked students to use their creativity to design the images in the motor, and the graphs denoted the rock, paper, and scissors on the micro:bit LED screen to increase the appeal of the game, as shown in Fig. 5c.

  5. 5.

    Mathematics: Students had to use mathematics to evaluate the correct rate of computer vision recognition tasks for gesture recognition and tune the optimal parameters in the neural network of gesture recognition.

In this study, the learning activity was designed to allow learners to understand image recognition and implement a hardware-controlled (micro:bit + motor) rock–paper–scissors game in 2 weeks. The features of the project are as follows:

Step 1: Obtaining images of scissors, rocks, and paper through the webcam image acquisition program and storing them in the computer.

Step 2: Training an AI deep learning neural network to learn the image features of scissors, rock, and paper from a user’s gesture.

Step 3: Coding a program in the Scratch language in the Kittenblock environment that automatically recognizes the user's hand gestures (rock, paper, and scissors) via a webcam.

Step 4: Testing the performance of the gesture recognition system.

Step 5: Tuning of the parameters of the neural network or retraining the neural network model, ensuring the game’s best performance.

Step 6: Assembling the micro:bit and motor to complete the game design.

The students were taught to employ the concept of AI to design an intelligent system that could recognize scissors, rocks, and paper expressed in human gestures via a webcam in real time; subsequently, they were taught STEM subjects to create an intelligent system that can automatically recognize the gestures. The program designed for this game determines the player’s gestures as either scissors, rock, or paper, and subsequently uses the micro:bit to control the motor to raise the corresponding Lego arm. The learning activities of STEAM + AI are illustrated in Figs. 4, 5.

Fig. 4
figure 4

An illustration of STEAM teaching materials

Fig. 5
figure 5

a Coding the program in Kittenblock. b AI program can recognize user’s gestures (blank, scissors, rock, paper) in real time through webcam image. c Designed graphs by students in the STEAM activity

Data collection procedure

The experimental procedure was conducted over two weeks and organized into the following two procedural stages:

(1) Orientation and teaching

The development of the teaching activities in this study followed the 6E model, wherein the teaching objectives, STEAM knowledge content, and AI concepts were considered. In this model, the teacher plays the role of a guide, while the students are guided by the teacher or the teaching students on how to complete the tasks and produce the learning outcomes at each stage. The teacher taught students how to finish an AI-based STEAM game. The students must have the ability to design their own AI-based STEAM game by retraining the neural network with their own photos.

(2) Data collection instrument

Participants were asked to complete an online questionnaire designed by the authors of this study. Researchers were available on site to resolve any query of the participants regarding the questionnaire items.

Measurement tools

The questionnaire used in this study included items from the following scales: TAM scale, cognitive load, and ARCS scale. The TAM scale elements used in this study were derived from the scale developed by (Mutambara & Bayaga, 2021), including three major concepts: perceived usefulness, perceived ease of use, and perceived enjoyment. The cognitive load scale, which includes mental load and mental effort, was developed by (Chen & Huang, 2020; Hwang et al., 2013) based on the concepts of cognitive load proposed by Sweller and colleagues (Sweller, 1988; Sweller et al., 1998a). The ARCS scale, which includes items measuring attention, relevance, confidence, and satisfaction of respondents, was developed by (Li et al., 2018) based on Keller’s proposed ARCS theory (Keller, 1983, 2009). Thus, an instrument consisting of 24 items (TAM scale), 8 items (cognitive load scale), and 20 items (ARCS scale) rated on a 7-point Likert scale ranging from 1 (“strongly disagree”) to 7 (“strongly agree”) was proposed in this study. Detailed information on each item is presented in Appendix A. The data were imported into the PLS-SEM software to calculate reliability, validity, and structural model estimation. PLS-SEM has been used in education assessment (Olmedo Moreno et al., 2014) and STEM education (Badri et al., 2016; Shiau et al., 2018).


Measurement model assessment

This study adopted confirmatory factor analysis (CFA) to assess the measurement model in terms of convergent validity and reliability (Findik-Coşkunçay et al., 2018). Table 2 summarizes the mean, standard deviation, standardized factor loadings, Cronbach’s alpha, composite reliability (CR), rho_A, average variance extracted (AVE), and variance inflation factor (VIF). All factor loadings were above 0.60 and at a significance level of p < 0.001, which revealed convergent validity (Liu, 2020). Each observed variable must have a factor loading > 0.7 to provide adequate convergent validity (Hair et al., 2006). The Cronbach’s alpha coefficients for each construct were as follows: perceived usability (0.96), perceived usefulness (0.95), perceived ease of use (0.91), perceived enjoyment (0.90), attitude (0.93), intention (0.89), cognitive load theory (0.94), mental effort (0.90), mental load (0.88), ARCS (0.97), attention (0.93), relevance (0.89), confidence (0.93), and satisfaction (0.90). They all satisfied the criterion of alpha > 0.7 (Dörnyei & Taguchi, 2009). The CR values of all constructs were greater than the suggested minimum of 0.7, all rho_A values were greater than 0.9, and all AVE values were greater than 0.5. All constructs in the measurement model satisfied the reliability criterion, and rho_A values of three constructs (intention, mental load, and relevance) were slightly lower than 0.9. This study used VIF to examine whether multicollinearity existed. The results demonstrated that all VIF values were acceptable (VIF < 5), indicating that the study data had no serious multicollinearity (Galeazzo et al., 2021). Because the factor loadings of item 5 of relevance, and items 2 and 3 of satisfaction did not have adequate VIF values (< 5) on the related latent variables, these items were extracted from the model. Therefore, all values exceeded the minimum threshold, indicating that the constructs were explained (Hair et al., 2006, 2016).

Table 2 Reliability

Discriminant validity is measured by the square root of the AVE of each latent variable, which indicates that each construct’s correlation is higher than the other constructs’ correlations (Peter, 1981). Table 3 shows the discriminant validity. The results support the discriminant validity of all the measured constructs in this study. Additionally, recent research (Henseler et al., 2015; Voorhees et al., 2016) suggests that the heterotrait–monotrait (HTMT) criterion could be better than the traditional Fornell and Larcker metric (Fornell & Larcker, 1981). The values of the HTMT criterion use the mean value of the item correlations across constructs relative to the geometric mean of the average correlations for the items measuring the same construct (Barrett et al., 2021). The results of HTMT inference at the 95% confidence level were obtained by conducting a bootstrapping procedure with a sample size of 5000. All HTMT values in Table 4 were less than 1, supporting and confirming discriminant validity (Hair et al., 2016).

Table 3 Fornell–Larcker discriminant validity results
Table 4 Heterotrait–monotrait ratio (HTMT)

Structural model assessment

This study employed partial least squares (PLS) using SmartPLS software to estimate the parameters of the proposed model. The multi-step procedure followed Baron and Kenny’s (1986) method to test the direct, indirect, and mediation effects of the variables. The estimation method utilized 5000 bootstrapping resampling estimations to examine the robustness of the research findings at the 95% confidence level. The structure of the proposed research model was examined by calculating the path coefficient values to assess the statistical significance of each hypothesis. Figure 6 shows the estimated path coefficients.

Fig. 6
figure 6

Structural model


Perceived usability has a positive and significant effect on attitude.


ARCS positively moderates the relationship between perceived usability and attitude.


Perceived usability has a positive and significant effect on intention.


Cognitive load has a negative and significant effect on perceived usability.


ARCS has a positive and significant effect on attitude.


Perceived usability positively moderates the relationship between attitude and intention.


ARCS positively moderates the relationship between perceived usability and attitude.

The direct path coefficients of the structural model were analyzed to examine the relationships between the constructs (see Fig. 6). According to the results of the structural model in Table 5, significant positive relationships were found between constructs H1, H2, H3, and H5 at the level of p < 0.05. Additionally, regarding the cognitive load theory construct (H4), a strong negative relationship was found between perceived usability and cognitive load. Analysis of the coefficient relationships within the original TAM constructs in STEAM education demonstrated that all TAM constructs proposed by Davis were supported (Davis, 1989); H1 (perceived usability → learning attitude) path showed a significant positive effect (B = 0.41, p < 0.005), H2 (learning attitude→ learning intention) path showed a strong significant effect (B = 0.34, p < 0.01), and H3 (perceived usability → learning intention) path showed a significant positive effect (B = 0.55, p < 0.01). The results revealed that high perceived usability (including perceived usefulness, perceived ease of use, and perceived enjoyment) enhanced learning attitude, and learning attitude positively influenced learning intention. Analysis of the effects of cognitive load (H4) (cognitive load → perceived usability) showed that the variable had significant negative direct effects on perceived usability (B = − 0.32, p < 0.01). The results revealed that the lower the cognitive load, the higher the perceived usability of STEAM education. Regarding the effects of Keller’s ARCS theory, H5 (ARCS → learning attitude) showed a positive and significant effect (B = 0.48, p < 0.01). The results indicate that strong ARCS motivation positively influences learning attitudes in STEAM education.

Table 5 Summary of structural model analysis

The coefficient of determination (R2) measures the in-sample predictive power (Hair et al., 2016). The f2 effect determines the change in the R2 value when a specified exogenous construct is omitted from the model in the range of 0.02, 0.15, and 0.35, indicating an exogenous construct’s small, medium, or large effect, respectively (Hair et al., 2016). The q2 value determines the impact of the model’s exogenous constructs on their reflective endogenous constructs. For PLS prediction, the Q2 prediction value for all indicators was above 0, indicating that the structural model has predictive power (Pangarso et al., 2020). Therefore, the R2, f2, and q2 values (Table 5) showed that our structural model had a medium predictive power.

The standardized root mean square residual (SRMR) is a goodness-of-fit measure for PLS-SEM that measures the mean absolute value of the covariance residuals by transforming both the sample covariance matrix and the predicted covariance matrix into correlation matrices (Hair et al., 2016). RMS_theta is the root mean squared residual covariance matrix of the outer model residuals (Lohmöller, 1989). The goodness-of-fit measures, SRMR and RMS_theta, should be less than 0.08 and close to 0, respectively (Hair et al., 2016). The SRMR value for the saturated model was 0.069, and the RMS_theta value was 0.16 close to 0, indicating a well-fitting model in this study.

Moderating effect analysis

Subsequently, we examined the moderating effects of these variables. The results in Table 6 show that the interaction term “perceived usability x attitude” is positive and significant for intention (B = 0.104, t-value = 3.13, p = 0.002), and the interaction term “ARCS x perceived usability” is positive and significant for attitude (B = 0.082, t-value = 2.03, p = 0.042). To reveal the moderating effects of intention and attitude more intuitively, this study plotted two interactive relationships.

Table 6 Moderating effect test results

According to the results of the moderating effect of perceived usability (see Fig. 7), the degree of positive influence (slope) of “perceived usability” on the relationship between learning attitude and learning intention at different levels (high and low) is significantly different.

Fig. 7
figure 7

Moderating effect of perceived usability (USAB) on the relationship between learning attitude and learning intention.

Perceived usability strengthens the positive relationship between attitude and intention. The relationship of “learning attitude → learning intention” is stronger when perceived usability is high; the relationship of “learning attitude → learning intention” is weaker when perceived usability is low. In other words, in the case of a high level of perceived usability, the intention of STEAM learning may be enhanced by a higher awareness of usability. Conversely, when learners’ awareness of usability is low, the effect of attitudinal enhancement of STEAM learning intention is lower. Therefore, teachers should consider how to design STEAM teaching materials to enhance students’ cognitive usability and attitudes, which in turn will enhance their learning intentions.

According to the results of the ARCS moderating effect (see Fig. 8), the degree of positive influence (slope) of “ARCS theory” on the relationship between perceived usability and attitude at different levels (high and low) is significantly different.

Fig. 8
figure 8

Moderating effect of the ARCS theory (ARCS) on the relationship between perceived usability (USAB) and learning attitude

Perceived ARCS strengthens the positive relationship between perceived usability and attitude. When learners’ perceived ARCS values were high, the “perceived usability → attitude” relationship was stronger; when the perceived ARCS values were low, the “perceived usability → attitude” relationship was weaker. In other words, when learners perceive a high level of ARCS regarding STEAM materials, the “perceived usability → attitude” relationship may be enhanced by the higher ARCS. Conversely, when learners perceive a low level of ARCS, the effect of perceived usability on learning attitudes is lower. Therefore, teachers should focus on designing instruction and/or instructional materials for ARCS to solve the particular motivational problem to enhance learners’ perceived usability (Chang et al., 2019) and improve their learning attitudes.

Multi-group analysis

The multi-group analysis allows us to test if our two types of student data groups have significant differences in their group-specific parameter estimates (e.g., outer weights, outer loadings and path coefficients) of our research structural model. The partial least squares multi-group analysis (PLS-MGA) method is a non-parametric significance test for the difference of group-specific results that builds on PLS-SEM bootstrapping results (Matthews, 2017). The PLS-GMA result showed that all hypotheses in our structural model have no significant difference between elementary student group and university group as shown in Table 7. Therefore, our research result can be interpreted as generalization for both elementary and university students.

Table 7 Multi-group analysis by PLS-MGA method


In this study, perceived usability (perceived usefulness, perceived ease of use, and perceived enjoyment) strongly and positively dominated learning attitudes and learning intentions. Additionally, positive learning attitudes influence STEAM learning intentions (Barrett et al., 2021; Mutambara & Bayaga, 2021; Shiau et al., 2018). The research results confirmed the relationship in Davis’s TAM constructs (Davis, 1989; Davis et al., 1989) that were applied to explain the behavioral intention of STEAM education (Mutambara & Bayaga, 2021; Su, 2019). This may be because the higher the learners’ perceived usability regarding STEAM education, the stronger their attitude toward their learning intention, in line with Bloom’s (1956) affective domain argument (Krathwohl et al., 1973).

In this study, ARCS motivation factors (attention, relevance, confidence, and satisfaction) strongly and positively influenced learning attitudes (B = 0.48, p < 0.001), confirming the relationship in previous studies (Galbis-Córdoba et al., 2017; Ngah et al., 2021). Most previous studies examined the relationship between ARCS and learning performance (Su & Cheng, 2015) or motivation (Wahyudi et al., 2017). This study examined the relationship between ARCS motivation factors and learning attitudes. As ARCS is a model used by the students to solve the problems based on the way they construct knowledge about the concept provided by the teacher, it can show the motivation of the students derived from external conditions. (Wahyudi et al., 2017). The results of this study confirmed that the affective domain in Bloom’s (1956) taxonomy concerns the growth in feelings and attitudes; in other words, motivation enhances learning performance (Krathwohl et al., 1973).

Conversely, learners’ cognition had strong and negative effects on perceived usability (B = − 0.319, p < 0.01), in line with previous studies (Findik-Coşkunçay et al., 2018). Cognitive load, including mental load and mental effort, denotes resource allocation for cognitive processing in learning tasks. Mental load represents the cognitive capacity required to process the complexity of a task, whereas mental effort reflects a learner’s cognitive capacity or resources that are allocated to complete the learning task (Liu et al., 2021). The high-level load of cognitive processes (such as knowledge recall, comprehension of understanding, application, and analysis) decreases perception, which is consistent with Bloom’s (1956) cognitive domain argument of knowledge development of intellectual skills. Therefore, STEAM educators should design teaching materials to decrease learners’ cognitive load in order to improve their perceived usability.

Surprisingly, this study found two significant moderating effects that were observed in only a few previous studies. Perceived usability moderated the relationship between learning attitudes and learning intentions. The moderation effect provides further information that a high level of perceived usability strongly enhances the attitude toward the intention of STEAM learning than that of a low level of perceived usability. ARCS motivation further strengthens the positive relationship between perceived usability and learning attitude. Additionally, the research results showed two mediation effects for STEAM learning attitude and learning intention.

A previous study stated that in a highly competitive education system, using SEM analysis is not only helpful in understanding both the consequences and sequences of learning performance, but also provides a boosted understanding of individual learning processes, which are simultaneously shown as a set of complex learning environments within individuals and a meaning of moderate or mediate effects (Horng et al., 2020; Sam Liu, 2017). The results show that cognitive load enables learners to easily understand and implement the STEAM task, which extends the findings of Weng et al. (2018), who found that cognitive load is driven by inherent complexity and deters valuable cognitive resources towards tasks irrelevant to learning.


The research results provide sufficient evidence to other educators in Asia, especially regarding the lack of student attention in the digital era, which disturbs students’ engagement with digital education. This study found that in ARCS and perceived usability, the influence of students’ learning attention was mediated by learning attitudes. A similar observation was made regarding ARCS; that is, ARCS positively significantly influenced perceived usability and strengthened its effects on learning attitudes. Our research results supported ARCS learning, which reflects several critical characteristics of STEAM, including cross-domain and hands-on learning, life application, problem solving, and sense learning, and applied such concepts to an AI-based task (Li et al., 2018). Thus, these findings should help educators recognize the importance of AI-based education and ‘playfulness’ in learning, and help them combine AI concepts with their teaching strategies to enhance students’ learning attitudes and explore learning opportunities. The findings of this study have various implications for helping students become familiar with micro:bit and AI-based learning that assists them in developing their abilities, skills, and beliefs to predict their future careers (Zhu et al., 2019). The potential theoretical and educational implications are explained in the following section.

Theoretical and educational implications

This study makes several valuable contributions to the literature. First, while previous research on AI-based course design has been conducted in Western countries because of more advanced technology and widely accepted AI concepts, this has not been the case in Asian regions because of the lack of development in technology and economic support, most higher education institutions still remain in the transitional education phase (Adukaite et al., 2017). In the current education scenario, implementing AI concepts for education and gamified learning could not only encourage students’ participation and engagement, but also add value to the literature on education (Tan & Cheah, 2021). This is especially the case for research design in higher education institutions, where current research on TAM of STEAM instruction is sparse (Maskeliūnas et al., 2020), specifically affecting the benefits of digital learning. Second, this study provides new perspectives for demonstrating how to use AI and digital learning in formal educational environments that have limited technological infrastructure, but which could be interesting and enjoyable despite this limitation by adjusting students’ attitudes toward learning intention. Third, this study incorporated a modified integrated model of the mediation–moderation mechanism. In the mediation mechanism, learning attitudes acted as mediating variables. This mediating effect was examined using the bootstrapping technique, which extends the existing education literature (Loughlin-Presnal & Bierman, 2017). The most critical aspect of this study is the use of second-order factors derived from ARCS and the use of perceived usability as a moderating variable, which has not been used in previous digital learning and education studies.

Further, the mediated-moderation model provides more meaningful and insightful information for the education literature (Horng et al., 2020). Therefore, the modified integrated model used in this study makes a critical contribution to the digital learning education literature and provides a foundational reference for future studies, while adding value to the existing research.

Moreover, several educational implications of the study results were identified for educators. First, samples were collected from undergraduates and elementary students in Taiwan. It provides sufficient evidence to other educators in Asia, especially regarding the lack of student attention in the digital era that disturbs students’ engagement with digital education (Huang et al., 2006). To harness the positive impact of AI and implement gamification learning design to develop students’ intention for learning, educators need to educate students on how to use micro and AI learning tools to enhance their learning interest (Mutambara & Bayaga, 2021). Second, this study found that in ARCS and perceived usability, the influence of students’ learning attention was mediated by learning attitudes. Micro:bit and AI-based lessons may not only improve students’ attitudes to increase the effectiveness of ARCS motivational strategies, but also help improve students’ ability to work in an AI-empowered society in the future (Li & Moore, 2018). Third, the findings also provide insights into cognitive theory and extend the education design of micro- and AI-based curricula. The findings of this study have various implications, from helping students become familiar with micro- and AI-based learning that assists them in developing their abilities, skills, and beliefs to predicting their future careers (Zhu et al., 2019). In this study, cognitive load was negatively correlated with learning performance and intention. This result supports instructors’ efforts to guide students in navigating unfamiliar STEAM learning processes, as this unfamiliarity might contribute to learning pressure, consequently decreasing learning performance. Thus, in order to enhance students’ learning and decrease their cognitive load, educators should design motivation incentive mechanisms to attract students’ interest along with less cognitive load to enhance their learning achievement.

Limitations and future research suggestions

Despite the contributions of this study, several limitations remain to be addressed. First, due to time limitations and lack of sufficient course support, the present study included a sample of students from only one region of Taiwan. A previous study suggested that collected data from different regions provide meaningful and accurate predictions of student learning behavior (Horng et al., 2020). Thus, future research may extend the results of the present study by including students from different regions in Taiwan (e.g., South, North, Middle, and offshore islands of Taiwan). Second, the cultural background of the study was another concern. In the “Western” education system, teachers dedicate a significant amount of time and shoulder responsibilities to guide students to include self-determination in their learning, which is different than the “Eastern” education system (Koul & Fisher, 2005). Thus, future studies may collect samples from different cultural backgrounds and provide a comparative study to extend the findings of this study. Third, the long-standing unsolved problem of AI-based education in STEM studies and employability needs to be considered with other possible variables to examine the hypotheses in future studies (Chu et al., 2019).

Availability of data and materials

Not applicable.


  • Abe, E. N., & Chikoko, V. (2020). Exploring the factors that influence the career decision of STEM students at a university in South Africa. International Journal of STEM Education, 7(1), 60.

    Article  Google Scholar 

  • Adukaite, A., van Zyl, I., Er, Ş, & Cantoni, L. (2017). Teacher perceptions on the use of digital gamified learning in tourism education: The case of South African secondary schools. Computers & Education, 111, 172–190.

    Article  Google Scholar 

  • Apedoe, X. S., Reynolds, B., Ellefson, M. R., & Schunn, C. D. (2008). Bringing Engineering Design into High School Science Classrooms: The Heating/Cooling Unit. Journal of Science Education and Technology, 17(5), 454–465.

    Article  Google Scholar 

  • Ariyanto, S., Munoto, M., & Muhaji, M. (2019). Development of Psychomotor Domain Assessment Instrument on Brake System Competence in SMKN 1 Jetis Mojokerto. International Journal for Educational and Vocational Studies.

    Article  Google Scholar 

  • Badri, M., Alnuaimi, A., Mohaidat, J., Al Rashedi, A., Yang, G., & Al Mazroui, K. (2016). My science class and expected career choices—a structural equation model of determinants involving Abu Dhabi high school students. International Journal of STEM Education, 3(1), 12.

    Article  Google Scholar 

  • Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182.

    Article  Google Scholar 

  • Barrett, A., Pack, A., Guo, Y., & Wang, N. (2020). Technology acceptance model and multi-user virtual reality learning environments for Chinese language education. Interactive Learning Environments.

    Article  Google Scholar 

  • Barrett, A. J., Pack, A., & Quaid, E. D. (2021). Understanding learners’ acceptance of high-immersion virtual reality systems: Insights from confirmatory and exploratory PLS-SEM analyses. Computers & Education, 169, 104214.

    Article  Google Scholar 

  • Bassachs, M., Cañabate, D., Nogué, L., Serra, T., Bubnys, R., & Colomer, J. (2020). Fostering Critical Reflection in Primary Education through STEAM Approaches. Education Sciences, 10(12), 384.

  • Bloom, B. S. (1956). Taxonomy of educational objectives, handbook I: The cognitive domain. David McKay Co Inc.

    Google Scholar 

  • Chang, Y.-S., Kuo Jui, H., Chiang, C.-W., & Lugmayr, A. (2019). Applying Mobile Augmented Reality (AR) to Teach Interior Design Students in Layout Plans: Evaluation of Learning Effectiveness Based on the ARCS Model of Learning Motivation Theory. Sensors, 20, 105.

    Article  Google Scholar 

  • Cheng, K.-H. (2017). Reading an augmented reality book: An exploration of learners’ cognitive load, motivation, and attitudes. Australasian Journal of Educational Technology, 33(4).

    Article  Google Scholar 

  • Chen, C.-C., & Huang, P.-H. (2020). The effects of STEAM-based mobile learning on learning achievement and cognitive load. Interactive Learning Environments.

    Article  Google Scholar 

  • Chu, H.-E., Martin, S. N., & Park, J. (2019). A Theoretical Framework for Developing an Intercultural STEAM Program for Australian and Korean Students to Enhance Science Teaching and Learning. International Journal of Science and Mathematics Education, 17(7), 1251–1266.

    Article  Google Scholar 

  • Conradty, C., & Bogner, F. X. (2020). STEAM teaching professional development works: effects on students’ creativity and motivation. Smart Learning Environments, 7(1), 26.

    Article  Google Scholar 

  • Costley, J., & Lange, C. (2017). The Mediating Effects of Germane Cognitive Load on the Relationship Between Instructional Design and Students’ Future Behavioral Intention. The Electronic Journal of e-Learning, 15(2)

  • Curran, V., Gustafson, D. L., Simmons, K., Lannon, H., Wang, C., Garmsiri, M., et al. (2019). Adult learners’ perceptions of self-directed learning and digital technology usage in continuing professional education: An update for the digital age. Journal of Adult and Continuing Education, 25(1), 74–93.

    Article  Google Scholar 

  • Dai, H. M., Teo, T., Rappa, N. A., & Huang, F. (2020). Explaining Chinese university students’ continuance learning intention in the MOOC setting: A modified expectation confirmation model perspective. Computers & Education, 150,

    Article  Google Scholar 

  • Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340.

    Article  Google Scholar 

  • Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer technology: A comparison of two theoretical models. Management Science, 35(8), 982–1003.

    Article  Google Scholar 

  • Dörnyei, Z., & Taguchi, T. (2009). Questionnaires in Second Language Research: Construction, Administration, and Processing (2nd ed.). Routledge.

  • Dunn, T. J., & Kennedy, M. (2019). Technology Enhanced Learning in higher education; motivations, engagement and academic achievement. Computers & Education, 137, 104–113.

    Article  Google Scholar 

  • Ekatushabe, M., Kwarikunda, D., Muwonge, C. M., Ssenyonga, J., & Schiefele, U. (2021). Relations between perceived teacher’s autonomy support, cognitive appraisals and boredom in physics learning among lower secondary school students. International Journal of STEM Education, 8(1), 8.

    Article  Google Scholar 

  • Findik-Coşkunçay, D., Alkiş, N., & Özkan-Yildirim, S. (2018). A structural model for students’ adoption of learning management systems: An empirical investigation in the higher education context. Journal of Educational Technology & Society, 21(2), 13–27.

    Google Scholar 

  • Fletcher, J. D. (2018). Comments and reflections on ITS and STEM education and training. InternationalJournal of STEM Education, 5(1), 16.

    Article  Google Scholar 

  • Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39–50.

    Article  Google Scholar 

  • Fu, S., Gu, H., & Yang, B. (2020). The affordances of AI-enabled automatic scoring applications on learners’ continuous learning intention: An empirical study in China. British Journal of Educational Technology,.

    Article  Google Scholar 

  • Galbis-Córdoba, A., Martí-Parreño, J., & Currás-Pérez, R. (2017). Education Students’ Attitude towards the Use of Gamification for Competencies Development. Journal of E-Learning and Knowledge Society, 13, 1.

    Article  Google Scholar 

  • Galeazzo, A., Ortiz-de-Mandojana, N., & Delgado-Ceballos, J. (2021). Green procurement and financial performance in the tourism industry: the moderating role of tourists’ green purchasing behaviour. Current Issues in Tourism, 24(5), 700–716.

    Article  Google Scholar 

  • Gao, X., Li, P., Shen, J., & Sun, H. (2020). Reviewing assessment of student learning in interdisciplinary STEM education. International Journal of STEM Education, 7(1), 24.

    Article  Google Scholar 

  • Granić, A., & Marangunić, N. (2019). Technology acceptance model in educational context: A systematic literature review. British Journal of Educational Technology, 50(5), 2572–2593.

    Article  Google Scholar 

  • Gülen, S., & Yaman S. (2019). The Effect of Integration of STEM Disciplines into Toulmin's Argumentation Model on Students' Academic Achievement, Reflective Thinking, and Psychomotor Skills*.

  • Gülhan, F., & Şahin, F. (2018, 08/05). The effects of STEAM (STEM+ Art) activities 7th grade students’ academic achievement, STEAM attitude and scientific creativities<p>STEAM (STEM+Sanat) etkinliklerinin 7. sınıf öğrencilerinin akademik başarı, STEAM tutum ve bilimsel yaratıcılıklarına etkisi. Journal of Human Sciences, 15(3), 1675–1699.

  • Gyamfi, S. A. (2017). Informal Tools in Formal Context: Adoption of Web 2.0 Technologies among Geography Student Teachers in Ghana. International Journal of Education and Development Using Information and Communication Technology, 13(3), 24–40.

    Google Scholar 

  • Hair, J. F., Hult, G. T. M., Ringe, C. M., & Sarstedt, M. (2016). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). SAGE Publications.

  • Hair, J., Black, W., Babin, B., Anderson, R., & Tatham, R. (2006). Multivariate data analysis. Upper Saddle River, NJ: Pearson Prentice Hall.

    Google Scholar 

  • Haji, S. A., Moluayonge, G. E., & Park, I. (2017). Teachers’ use of information and communications technology in education: Cameroon secondary schools perspectives. Turkish Online Journal of Educational Technology-TOJET, 16(3), 147–153.

    Google Scholar 

  • Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. Journal of the Academy of Marketing Science, 43(1), 115–135.

    Article  Google Scholar 

  • Horng, J.-S., Liu, C.-H., Chou, S.-F., & Huang, Y.-C. (2020). The roles of university education in promoting students’ passion for learning, knowledge management and entrepreneurialism. Journal of Hospitality and Tourism Management, 44, 162–170.

    Article  Google Scholar 

  • Huang, W., Huang, W., Diefes-Dux, H., & Imbrie, P. K. (2006). A preliminary validation of Attention, Relevance, Confidence and Satisfaction model-based Instructional Material Motivational Survey in a computer-based tutorial setting. British Journal of Educational Technology, 37(2), 243–259.

    Article  Google Scholar 

  • Huang, Y., & Liu, X. (2021). The analysis and research of STEAM education based on the TAM algorithm model to improve the learning effectiveness of higher vocational engineering students. Evolutionary Intelligence.

    Article  Google Scholar 

  • Huett, J. B. (2006). The Effects of ARCS-based Confidence Strategies on Learner Confidence and Performance in Distance Education. Citations.

  • Hwang, G.-J., Yang, L.-H., & Wang, S.-Y. (2013). A concept map-embedded educational computer game for improving students’ learning performance in natural science courses. Computers & Education, 69, 121–130.

    Article  Google Scholar 

  • Kanchanatanee, K., Suwanno, N., & Jarernvongrayab, A. (2014). Effects of Attitude toward Using, Perceived Usefulness, Perceived Ease of Use and Perceived Compatibility on Intention to Use E-Marketing. Journal of Management Research, 6, 3.

    Article  Google Scholar 

  • Karakis, H., Karamete, A., & Okcu, A. (2016). The Effects of a Computer-Assisted Teaching Material, Designed According to the ASSURE Instructional Design and the ARCS Model of Motivation, on Students’ Achievement Levels in a Mathematics Lesson and Their Resulting Attitudes. European Journal of Contemporary Education, 15(1), 105–113.

    Article  Google Scholar 

  • Keller, J. M. (1983). Motivational design of instruction. Instructional Design Theories and Models: An Overview of Their Current Status, 1(1983), 383–434.

    Google Scholar 

  • Keller, J. M. (2009). Motivational design for learning and performance: The ARCS model approach. New York: Springer.

    Google Scholar 

  • Koc, E. S., & Ontas, T. (2020). A comparative analysis of the 4th and 5th grade social studies curriculum according to revised bloom taxonomy. Cypriot Journal of Educational Sciences, 15(3), 540–553.

    Article  Google Scholar 

  • Koul, R. B., & Fisher, D. L. (2005). Cultural Background and Students’ Perceptions of Science Classroom Learning Environment and Teacher Interpersonal Behaviour in Jammu. India. Learning Environments Research, 8(2), 195–211.

    Article  Google Scholar 

  • Kozlovskiy, Y., Bilyk, V., Zakharyash, O., Voloshyn, M., & Babkina, М. (2021). The information and communication technologies in the learning process of students in the context of cognitive development. Laplage em Revista, 7(3C), 471–476.

    Article  Google Scholar 

  • Krathwohl, D. R., Bloom, B. S., & Masia, B. B. (1973). Taxonomy of Educational Objectives, the Classification of Educational Goals. Handbook II: Affective Domain. New York: David McKay Co., Inc.

  • Land, M. H. (2013). 2013/01/01/). Full STEAM Ahead: The Benefits of Integrating the Arts Into STEM. Procedia Computer Science, 20, 547–552.

    Article  Google Scholar 

  • Li, C.-L., Chen, Y.-H., & Li, H.-Y. (2018). Technical College Students’ ARCS Learning Motivation on Hospitality English Vocabulary. International Jouranl of Human Resource Studies, 8(1), 189–207.

    Article  Google Scholar 

  • Li, K., & Moore, D. R. (2018). Motivating Students in Massive Open Online Courses (MOOCs) Using the Attention, Relevance, Confidence, Satisfaction (ARCS) Model. Journal of Formative Design in Learning, 2(2), 102–113.

    Article  Google Scholar 

  • Lin, K.-Y., Wu, Y.-T., Hsu, Y.-T., & Williams, P. J. (2021). Effects of infusing the engineering design process into STEM project-based learning to develop preservice technology teachers’ engineering design thinking. International Journal of STEM Education, 8(1), 1.

    Article  Google Scholar 

  • Lin, P.-Y., Chai, C.-S., Jong, M.S.-Y., Dai, Y., Guo, Y., & Qin, J. (2021). Modeling the structural relationship among primary students’ motivation to learn artificial intelligence. Computers and Education: Artificial Intelligence.

    Article  Google Scholar 

  • Liu, C.-H. (2020). Local and international perspectives of the influence of creative experiences of Chinese traditional culture on revisit intentions. Current Issues in Tourism, 23(1), 17–35.

    Article  Google Scholar 

  • Liu, Q., Yu, S., Chen, W., Wang, Q., & Xu, S. (2021). The effects of an augmented reality based magnetic experimental tool on students’ knowledge improvement and cognitive load. Journal of Computer Assisted Learning, 37(3), 645–656.

    Article  Google Scholar 

  • Lohmöller, J.-B. (1989). Latent Variable Path Modeling with Partial Least Squares. Physica.

    Book  Google Scholar 

  • Loughlin-Presnal, J., & Bierman, K. L. (2017). How do parent expectations promote child academic achievement in early elementary school? A test of three mediators. Developmental Psychology, 53(9), 1694–1708.

    Article  Google Scholar 

  • Lu, Y., Yeung, W.-J. J., & Treiman, D. J. (2020). Parental migration and children’spsychological and cognitive development in China: differences and mediating mechanisms. Chinese Sociological Review, 52(4), 337–363.

    Article  Google Scholar 

  • Margot, K. C., & Kettler, T. (2019). Teachers’ perception of STEM integration and education: a systematic literature review. International Journal of STEM Education, 6(1), 2.

    Article  Google Scholar 

  • Marín-Marín, J.-A., Moreno-Guerrero, A.-J., Dúo-Terrón, P., & López-Belmonte, J. (2021). STEAM in education: a bibliometric analysis of performance and co-words in Web of Science. International Journal of STEM Education, 8(1), 41.

    Article  Google Scholar 

  • Maskeliūnas, R., Kulikajevas, A., Blažauskas, T., Damaševičius, R., & Swacha, J. (2020). An Interactive Serious Mobile Game for Supporting the Learning of Programming in JavaScript in the Context of Eco-Friendly City Management. Computers, 9(4), 102.

  • Matthews, L. (2017). Applying multigroup analysis in PLS-SEM: A step-by-step process. In (pp. 219-243).

  • Mu, R., & De Jong, M. (2018). The psychology of local officials: explaining strategic behavior in the Chinese Target Responsibility System. Journal of Chinese Governance, 3(2), 243–260.

    Article  Google Scholar 

  • Munawaroh, M. (2020). The Influence of Problem-Based Learning Model as Learning Method, and Learning Motivation on Entrepreneurial Attitude. International Journal of Instruction, 13(2), 431–444.

    Article  Google Scholar 

  • Musavi, M., Friess, W. A., James, C., & Isherwood, J. C. (2018). Changing the face of STEM with stormwater research. International Journal of STEM Education, 5(1), 2.

    Article  Google Scholar 

  • Mutambara, D., & Bayaga, A. (2021). Determinants of mobile learning acceptance for STEM education in rural areas. Computers & Education, 160, 104010.

    Article  Google Scholar 

  • Ngah, A. H., Rashid, R. A., Ariffin, N. A., Ibrahim, F., Osman, N. A. A., Kamalrulzaman, N. I., Mohamad, M. F. H., & Harun, N. O. (2021). Fostering Students’ Attitude Towards Online Learning: the Mediation Effect of Satisfaction and Perceived Performance.

  • Ocak, G., & Tepe, M., & Olur, B. (2020). Investigation of the objectives of the information technologh and sofeware course to the revised Bloom taxonomy. European Journal of Education Studies, 6(12), 56–68.

    Article  Google Scholar 

  • Olmedo Moreno, E. M., de Luna, E. B., Olmos Gómez, M. D. C., & López, J. E. (2014). Structural Equations Model (SEM) of a questionnaire on the evaluation of intercultural secondary education classrooms. Suma Psicológica, 21(2), 107–115.

    Article  Google Scholar 

  • Paas, F. (1992). 12/01). Training Strategies for Attaining Transfer of Problem-Solving Skill in Statistics: A Cognitive-Load Approach. Journal of Educational Psychology, 84, 429–434.

    Article  Google Scholar 

  • Paas, F., Tuovinen, J. E., Tabbers, H., & Van Gerven, P. W. (2003). Cognitive load measurement as a means to advance cognitive load theory. Educational Psychologist, 38(1), 63–71.

    Article  Google Scholar 

  • Pabalan, N., Singian, E., Tabangay, L., Jarjanazi, H., Boivin, M. J., & Ezeamama, A. E. (2018). Soil-transmitted helminth infection, loss of education and cognitive impairment in school-aged children: A systematic review and meta-analysis. PLoS Negl Trop Dis, 12(1), e0005523.

    Article  Google Scholar 

  • Pangarso, A., Astuti, E. S., Raharjo, K., & Afrianty, T. W. (2020). Data of innovation ambidexterity as a mediator in the absorptive capacity effect on sustainable competitive advantage. Data in Brief, 29, 105200.

    Article  Google Scholar 

  • Peter, J. P. (1981). Construct validity: A review of basic issues and marketing practices. Journal of Marketing Research, 18(2), 133–145.

    Article  Google Scholar 

  • Quaid, E., Pack, A., Barrett, A., & Zhou, L. (2020, 12/13). Students' intention to use high-immersion virtual reality systems for learning paragraph structure: a PLS-SEM exploratory study. 291–297.

  • Radhamani, R., Kumar, D., Nizar, N., Achuthan, K., Nair, B., & Diwakar, S. (2021). What virtual laboratory usage tells us about laboratory skill education pre- and post-COVID-19: Focus on usage, behavior, intention and adoption. Education and Information Technologies.

    Article  Google Scholar 

  • Ramma, Y., Bholoa, A., Watts, M., & Nadal, P. S. (2018). Teaching and learning physics using technology: Making a case for the affective domain. Education Inquiry, 9(2), 210–236.

    Article  Google Scholar 

  • Sam Liu, C.-H. (2017). Remodelling progress in tourism and hospitality students’ creativity through social capital and transformational leadership. Journal of Hospitality, Leisure, Sport & Tourism Education, 21, 69–82.

    Article  Google Scholar 

  • Shernoff, D. J., Sinha, S., Bressler, D. M., & Ginsburg, L. (2017). Assessing teacher education and professional development needs for the implementation of integrated approaches to STEM education. International Journal of STEM Education, 4(1), 1–16.

    Article  Google Scholar 

  • Shiau, S., Huang, C.-Y., Yang, C.-L., & Juang, J.-N. (2018). A Derivation of Factors Influencing the Innovation Diffusion of the OpenStreetMap in STEM Education. Sustainability, 10, 10.

    Article  Google Scholar 

  • Simpson, E. J. (1972). The Classification of Educational Objectives in the Psychomotor Domain. Gryphon House.

    Google Scholar 

  • Su, C.-H. (2019). The Effect of Users’ Behavioral Intention on Gamification Augmented Reality in Stem (Gar-Stem) Education. Journal of Baltic Science Education, 18(3), 450–465.

    Article  Google Scholar 

  • Su, C. H., & Cheng, C. H. (2015). A mobile gamification learning system for improving the learning motivation and achievements. Journal of Computer Assisted Learning, 31(3), 268–286.

    Article  Google Scholar 

  • Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285.

    Article  Google Scholar 

  • Sweller, J., Van Merrienboer, J. J., & Paas, F. G. (1998a). Cognitive architecture and instructional design. Educational Psychology Review, 10(3), 251–296.

    Article  Google Scholar 

  • Sweller, J., van Merrienboer, J. J. G., & Paas, F. G. W. C. (1998b). Cognitive Architecture and Instructional Design. Educational Psychology Review, 10(3), 251–296.

    Article  Google Scholar 

  • Tan, D. Y., & Cheah, C. W. (2021). Developing a gamified AI-enabled online learning application to improve students’ perception of university physics. Computers and Education: Artificial Intelligence, 2, 100032.

    Article  Google Scholar 

  • Teo, T. (2009). Modelling technology acceptance in education: A study of pre-service teachers. Computers & Education, 52(2), 302–312.

    Article  Google Scholar 

  • Trujillo, L. T. (2019). Mental effort and information-processing costs are inversely related to global brain free energy during visual categorization. Frontiers in Neuroscience, 13, 1292.

    Article  Google Scholar 

  • Voorhees, C. M., Brady, M. K., Calantone, R., & Ramirez, E. (2016). Discriminant validity testing in marketing: an analysis, causes for concern, and proposed remedies. Journal of the Academy of Marketing Science, 44(1), 119–134.

    Article  Google Scholar 

  • Wahyudi, S., Joyoatmojo, S., & Sawiji, H. (2017). Learning Model of Attention, Relevance, Confidence, Satisfaction (ARCS) supported by video tutorial to improve the students’ learning motivation in vocational high school. Advances in Social Science Education and Humanities Research (ASSEHR), 158, 603–611.

    Article  Google Scholar 

  • Wahono, B., Lin, P.-L., & Chang, C.-Y. (2020). Evidence of STEM enactment effectiveness in Asian student learning outcomes. International Journal of STEM Education, 7(1), 36.

    Article  Google Scholar 

  • Weng, F., Yang, R.-J., Ho, H.-J., & Su, H.-M. (2018). A TAM-Based Study of the Attitude towards Use Intention of Multimedia among School Teachers. Applied System Innovation, 1, 3.

    Article  Google Scholar 

  • Wu, B., & Chen, X. (2017). Continuance intention to use MOOCs: Integrating the technology acceptance model (TAM) and task technology fit (TTF) model. Computers in Human Behavior, 67, 221–232.

    Article  Google Scholar 

  • Yafie, E., Nirmala, B., Kurniawaty, L., Bakri, T. S. M., Hani, A. B., & Setyaningsih, D. (2020). Supporting Cognitive Development through Multimedia Learning and Scientific Approach: An Experimental Study in Preschool. Universal Journal of Educational Research, 8(11C), 113–123.

    Article  Google Scholar 

  • Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on artificial intelligence applications in higher education – where are the educators? International Journal of Educational Technology in Higher Education, 16(1), 39.

    Article  Google Scholar 

  • Zhang, S.-N., Li, Y.-Q., Liu, C.-H., & Ruan, W.-Q. (2019). Critical factors in the identification of word-of-mouth enhanced with travel apps: the moderating roles of Confucian culture and the switching cost view. Asia Pacific Journal of Tourism Research, 24(5), 422–442.

    Article  Google Scholar 

  • Zhu, S., Yang, H. H., MacLeod, J., Yu, L., & Wu, D. (2019). Investigating Teenage Students’ Information Literacy in China: A Social Cognitive Theory Perspective. The Asia-Pacific Education Researcher, 28(3), 251–263.

    Article  Google Scholar 

  • Zorluoglu, S. L., Bagriyanik, K. E., & Sahintürk, A. (2019). Analyze of the Science and Technology Course TEOG Questions Based on the Revised Bloom Taxonomy and Their Relation between the Learning Outcomes of the Curriculum. International Journal of Progressive Education, 15(2), 104–117.

    Article  Google Scholar 

Download references


The authors would like to thank the Ministry of Science and Technology, Taiwan, for financial support (MOST 108-2511-H-142-007-MY2; MOST 110-2511-H-142 -008 -MY2; 110-2511-H-006-012-MY3).


Ministry of Science and Technology. Taiwan, R. O. C.

Author information




All authors read and approved the final manuscript.

Corresponding author

Correspondence to Yueh-Min Huang.

Ethics declarations

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix A

Appendix A

Item Question
Perceived usefulness (PU) (Mutambara & Bayaga, 2021)
PU1 Using micro:bit + AI learning in class will improve my willingness to participate
PU2 Using micro:bit + AI to learn STEM will improve my learning effect
PU3 Using micro:bit + AI would make it easier for me to learn STEM
PU4 I will find micro:bit + AI useful in learning STEM
PU5 Using micro:bit + AI to learn STEM will improve my learning interest
Perceived ease of use (PEOU) (Mutambara & Bayaga, 2021)
PEOU1 It will be easy to learn how to use micro:bit + AI learning to learn STEM
PEOU2 I will find it easy to use micro:bit + AI learning to learn STEM
PEOU3 I will find micro:bit + AI learning easy to use in STEM classes
PEOU4 I will find micro:bit + AI learning to be flexible to interact
PEOU5 It will be easy for me to become skillful using micro:bit + AI
Perceived enjoyment (PENJ) (Mutambara & Bayaga, 2021)
PENJ1 Learning this curriculum will be enjoyable
PENJ2 I will find learning STEM using micro:bit + AI learning fun
PENJ3 I will find using micro:bit + AI learning interesting
Attitude (ATT) (Mutambara & Bayaga, 2021)
ATT1 I believe it is beneficial to learn STEM
ATT2 I feel positive about learning STEM
ATT3 My experience of learning STEM will be good
ATT4 I like to learn STEM-related subjects
ATT5 Learning about STEM-related subjects will be a pleasant experience
Intention (INT) (Mutambara & Bayaga, 2021)
INT1 Assuming I have access to micro:bit + AI, I intend to use it to learn STEM
INT2 I am planning to use micro:bit + AI to learn STEM
INT3 I would like to use micro:bit to learn STEM in the future
ARCS (Li et al., 2018)
Attention (ATTEN)
ATTEN1 The curriculum can arouse my exploratory motivation
ATTEN2 I can concentrate on curriculum content
ATTEN3 The technique for AI in interactive image recognition attracts my attention
ATTEN4 I will find it fun to program game code for AI in the interactive image recognition
ATTEN5 I will find it fun to interact with micro:bit + AI flexibly
Relevance (REV)
REV1 STEM teaching resources will be helpful
REV2 I can understand this curriculum
REV3 I can apply the outcome of this practice to different fields
REV4 I can apply the outcome of this practice to various industries (e.g., education and marketing)
REV5 I can apply the outcome of practice to different products
Confidence (COF)
COF1 I am confident I will succeed in the curriculum content of STEM learning
COF2 I have confidence that I will finish the assignment in the curriculum
COF3 It is not difficult to complete this project
COF4 I am confident in finding other functions
COF5 If I work hard, I can perform well in other systems
Satisfaction (SAT)
SAT1 I like this type of STEM course. I will continue to study related courses
SAT2 I am very satisfied that I will be able to glean relevant knowledge from the course
SAT3 I am very happy to complete the game in the course material
SAT4 I think it is fun while learning
SAT5 I had a sense of accomplishment in micro:bit + AI and program coding
Cognitive Load (Chen & Huang, 2020)
Mental Load (ML)
ML1 The learning content in this activity was difficult for me
ML2 I had to put a lot of effort into answering the questions in this learning activity
ML3 It was troublesome for me to answer the questions in this learning activity
ML4 I felt frustrated answering the questions in this learning activity
Mental Effort (ME)
ME1 I did not have enough time to finish the project
ME2 During the learning process, the curriculum content caused a lot of stress
ME3 During the learning activity, the content or information required a lot of mental effort
ME4 The curriculum content was difficult to follow and understand

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wu, CH., Liu, CH. & Huang, YM. The exploration of continuous learning intention in STEAM education through attitude, motivation, and cognitive load. IJ STEM Ed 9, 35 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • STEAM education
  • Technology acceptance model
  • Cognitive load
  • ARCS motivation model
  • STEAM learning cycle
  • Artificial intelligence
  • Learning performance