The rapid advancement of technology and the COVID-19 pandemic have led to an increase in the use of online education. However, this has resulted in issues such as decreased attention and academic performance among learners (Chen, Jiao, & Hu, 2021; Munastiwi & Puryono, 2021; Purwanto, 2020). Despite these challenges, online education is now a necessity for personalized learning in the future. Therefore, the aim of this study is to measure and evaluate students' learning progress using objective biometric data, in order to develop an online education feedback system to support students' learning. Especially, the purpose of this study is to confirm whether there is a difference in cognitive load depending on the achievement of online education using eye blink frequency and duration.

Cognitive load refers to the sum of mental activities that affect working memory in processing information (Sweller, 1988). For something to be learned, information must be processed in working memory. The problem between the amount of information that working memory can process and the information is called cognitive load (Kim & Kim, 2004). It can be divided into three types: intrinsic, extraneous, and germane cognitive load. Cognitive load theory suggests that if the cognitive load exceeds an individual's processing capacity, it may lead to poor learning outcomes and decreased motivation. Therefore, it is important to consider cognitive load when designing instructional materials and activities, in order to help learners process and retain information more effectively (Lee, 2017).

Self-report questionnaires and interviews were mainly used to measure cognitive load. However, there is a way to objectively measure cognitive load based on physiological responses (Park, Kim, & Jo, 2019). In particular, Abdurrahman et al. (2021) emphasized the importance of physiological monitoring to measure mental workload during learning transfer. Coral (2016) confirmed the increasing relationship between eye-related measurement variables and cognitive load. It was explained that the higher the eye blink duration and blink frequency, the greater the cognitive load. The research question of this study is, can the difference in cognitive load according to the achievement group be known through eye blink frequency and duration in the online learning environment?

 We provided 73 university students with educational content on Korean grammar, and eye tracking was conducted. There was no difference in the pre-test results on Korean grammar between the control and experimental groups. Eye blink frequency and duration were measured using gaze points GP3 HD Eye Tracker (150 Hz) and self-reported survey. Based on the post-test average score, the gaze tracking data and self-reported data was analyzed by dividing the group into high-achievement and low-achievement groups. The post-test in the 10-point test had an average score of 5.521. Based on this, 33 students were placed in the high-achievement group and 40 students were placed in the low-achievement group.

As a result, the difference in cognitive load by the self-report questionnaire was not significant. But eye blink frequency was higher in the lower achievement group than in the high- achievement group, but it was not significant. Eye blink duration was longer in the lower- achievement group than in the high- achievement group and had a significant difference.

In this study, it was confirmed whether this measurement is meaningful to objectively measure learners' cognitive load based on eye blink frequency and duration. It was confirmed that the longer the blinking duration, the greater the cognitive load, and it was verified that the eye-tracking data could be meaningfully used to prepare a support plan for online learning. Through this, it was suggested that a system be developed so that the instructor can immediately recognize the learner's cognitive load status and give feedback even in real-time online education. 

This study targeted 73 university students (male: 28, female: 45) in Korea. For the stable collection of gaze data, those with a history of functional bodily impairment (e.g., eye disease) and those who underwent eye-related surgery (e.g., LASIK surgery) were excluded in advance. This study was approved by the H University Institutional Review Board (IRB) to protect research participants and comply with research ethics (H University 2022-01-006). Written consent to participate in the study was obtained from the research subjects, and permission was obtained for the study again after informing the purpose of the research and information about recording before proceeding with the experiment. A 10-minute and 42-second online content about the types of sentences in Korean grammar was produced. Although the research participants are fluent in Korean, Korean grammar is not familiar to most Koreans as it is a learning area that must be studied separately as an optional subject for the SAT. To check the Korean level, pre-and post-tests were conducted as homomorphic tests. Korean language teachers in high school reviewed the viability of the test. Based on the average score of the post-test, it was divided into a high-achievement group and a low-achievement group. In addition, a t-test was conducted to confirm the difference between the pre-test and post-test by the group. In addition, to check the cognitive load, gaze points GP3 HD Eye Tracker (150Hz) was used to measure the eye blink frequency and duration, and a self-report questionnaire was also conducted. The self-report questionnaire used the cognitive load tool measured by Ryu & Lim (2009) with 20 questions divided into self-evaluation, physical effort, mental effort, material design, and task difficulty. To confirm the significance of the difference in the number and time of blinking according to the group according the grade, a non-parametric test was conducted as it did not follow a normal distribution, and each data was analyzed through SPSS

This study examined whether there is a difference in cognitive load according to the higher and lower achievement in online education through eye blink frequency and duration. First, through the self-report questionnaire, which has been actively used to measure cognitive load, differences in sub-factors of cognitive load, such as self-evaluation, physical effort, mental effort, material design, and task difficulty, were confirmed by the achievement group. It was confirmed that the cognitive load of the group with higher achievement was lower than the low-achievement group, but the difference between them was not significant. Next, considering that the eye blink frequency and duration increases as the cognitive load increases in previous studies, the difference between the eye blink frequency and duration, which is biometric data, was confirmed. It was confirmed that the number of eye blink frequency was higher in the lower achievement group than the higher achievement group, but the difference between them was not significant. Instead, it was confirmed that the eye blink duration was longer in the lower achievement group than the higher achievement group, and the difference between them was significant. However, the cognitive load could not be confirmed through self-report questionnaires but verified through eye blink duration. Therefore, this study confirmed the meaning of eye tracking to check the learner's cognitive load in developing an online support system to enhance the learner's educational performance. Therefore, it was possible to confirm the implications of creating a system that can provide real-time feedback to reduce the cognitive load by immediately checking the learner's reaction.

Abdurrahman, U. A., Yeh, S. C., Wong, Y., & Wei, L. (2021). Effects of neuro-cognitive load on learning transfer using a virtual reality-based driving system. Big Data and Cognitive Computing, 5(4), 54. Chen, Z., Jiao, J., & Hu, K. (2021). Formative assessment as an online instruction intervention: Student engagement, outcomes, and perceptions. International Journal of Distance Education Technologies (IJDET), 19(1), 50-65. Coral, M. P. (2016). Analyzing cognitive workload through eye-related measurements: A meta-analysis. Doctoral dissertation, Wright State University. Kim, K., & Kim, D. (2004). The effects of modality of text and timing of information presentation on cognitive load, effectiveness and efficiency in web based learning. communication books. Journal of Educational Technology, 20(4), 111-145. Lee, E. (2017). Difference in cognitive load according to learner's prior knowledge level and learning section in video learning environment : Focused on the pupil dilation. Master dissertation, Ewha Womans University. Munastiwi, E., & Puryono, S. (2021). Unprepared management decreases education performance in kindergartens during Covid-19 pandemic. Heliyon, 7(5), e07138. Park, H., Kim, D., & Jo, I. (2019). Correlation between the change of cognitive load and learning performance in video-based learning. The Journal of Educational Information and Media, 25(4), 797-826. Purwanto, A. (2020). University students online learning system during Covid-19 pandemic: Advantages, constraints and solutions. Sys Rev Pharm, 11(7), 570-576. Ryu, J., & Lim, J. (2009). An exploratory validation for the constructs of cognitive load. Journal of Korean Association for Educational Information and Media, 15(2), 1-27. Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257-285.