Closure of schools, colleges and other educational institutions have migrated billions of students across the globe to the web for attending online classes. World Economic Forum estimates that the COVID-19 pandemic has moved 1.2 billion students out of the classroom in 186 countries. The education landscape has transformed dramatically with the distinctive growth of online learning, as remote teaching on digital platforms in becoming the new order. Besides the COVID-19 pandemic crisis, online education business powered by digital technologies had reached USD 18.66 billion in 2019 and is projected to reach USD 350 billion by 2025, according to Research and Markets forecasts. Democratization of education through online massive online open courseware (MOOC) platforms has also enabled the top-tier global universities to teach students through extended virtual classrooms.
Enterprises are offering innovative learning technologies to fuel an exponential growth in the online learning landscape. Technology solutions for online learning provide capabilities in unlimited video conferencing time, smart calendar scheduling, real-time co-editing of project assignments and auto-translation capabilities. Online learning methods have shown retention rates of 25-60%, which are higher than the retention rate of 8-10% in offline classroom training. Employee training activities have also shifted online, which has resulted in participants learning approximately five times more material without increasing time spent in training, while taking 40 to 60 percent less time to learn. The changing education imperative globally has asserted the loss of relevance of the traditional education systems. Education sector is shifting towards online teaching methodologies in the urban areas of different countries, while the rural and the marginalized sections still rely largely on the traditional teaching methods.
The effectiveness of online learning methods varies across age groups. For young children, a structured environment is crucial to enable remote learning, as they are prone to distractions. It becomes imperative to replicate the physical classroom environment through video capabilities and collaboration tools that enable inclusion, intelligence and personalization. Attention lapses risk failing the purpose of online education programs as students miss certain sections of information on fundamentals, which are instrumental in building understanding of advanced concepts later. Digital monitoring technologies are, therefore, a necessity to ensure detection and prevention of voluntary distractions. Maintaining a track of attention lapses can also be instrumental in behavioral analytics to understand the causes of distraction and the necessary interventions that can be made by the teachers to ensure an attentive remote classroom.
Various technological interventions available in the online education space can be leveraged to monitor the behavior of students in the virtual space. Voluntary deviant behavior of students can disturb the decorum of a virtual classroom and can derail the process of effective remote education. Tracking the eye movements of the students during a session can be helpful to monitor the attention span of students and to alert the teacher of any loss of decorum in the online classroom. Digital solutions enabled with computer vision and machine earning models can provide deeper visibility into the attention lapses of virtually connected students.
Attention-related challenges in online education
Students participating in virtual education are prone to attention lapses due to lack of active monitoring during the online sessions. Multitasking behavior among students has been reported to be higher in the case of online courses, which is linked to poor academic performance. Kent State University research on 300 university students attending online courses discovered that 25 percent of the students were more likely to chat on social networks, share text messages listen to music and browse Internet. The students who attend online sessions exhibit greater tendency to multitask and score lower in academic evaluations compared to their academic performance in offline courses delivered face-to-face. The academically disadvantageous behavior is difficult for the instructors to detect from behind the camera.
The low completion rate of massive online open courseware (MOOCs) points in the same direction. MIT researchers examined the data of 5.63 million students on edX learning platform and found only 3.13 percent of those online students had completed their coursework during 2017-18. The course completion rate was 4 percent in 2016 and 2015, and was 6 percent in 2014. Distractions in the form of the web notifications and applications running in parallel undermine the effectiveness of virtual education programs.
The Kent University study also found that students showing higher tendency of multitasking feel encouraged by the absence of physical monitoring techniques. Students with higher scores in Polychronic-Monochronic Tendency Scale display higher multitasking tendencies during the online classes compared to sessions in physical classrooms. Students and teachers collectively establish the norms of the classroom that control the academically disadvantageous behavior of the multitasking students. Research by University of Nebraska at Lincoln found in 2016 that students spent approximately 20% of their time on digital platforms and devices for “non-class purposes”, which equaled 11.43 times a day on average. Lack of monitoring techniques allow students to stream online videos on devices in parallel with ongoing lecture sessions, while sending text messages to their peers.
A 2019 study by University of Waterloo among 478 students and 36 instructors revealed that 49 percent of the undergraduate students found use of technology not related to learning, led to distractions. The study found that the use of social media in the classrooms had become normalized and off-task activity of peers further reinforced the indulgence of a student in academically disadvantageous behavior. The need to fight boredom in classroom, the urge to feel refreshed and the sentiment of having the right to use technology in classroom contribute to distraction-driven behavior. Technology impairs the ability of students to retain information and leads to wastage of classroom hours and instructors’ man hours. The University of Waterloo study also studied teachers’ concerns regarding students’ inclination towards off-task technology. Teachers found students’ distraction to be insulting and dismissed the idea of sharing responsibility of making the classroom sessions engaging enough to de-incentivize off-task technology use. Instructors held the belief that competing with social media to attract attention is a difficult task, and students must show due diligence to their classroom sessions.
III. Existing technologies for student monitoring
Online education programs need technological and pedagogical solutions to keep online students focused on their primary academic tasks in the absence of physical classrooms and instructors. Eye tracking is one of the techniques deployed to gather insights into human attention, online behavior and decision-making process. Eye movements can provide clues into the attention levels of students and their concentration at various regions of a digital device screen. Data generated from eye-tracking technology can provide deeper insights into cognitive processes that are difficult to monitor by a teacher during an online session. Eye-tracking technologies also provide a comprehensive analysis of students’ ability in reading and in focusing on the course content.
Eye tracking metrics
Eye movements are of two types, namely fixation and saccade. Fixations are periods of time with relatively stable eye movements, signifying processing of visual information. Saccades are rapid eye movements which signify lack of mental processing of visual stimuli. Saccade is the time taken by the eye to focus its attention from one visual stimulus to another, which makes it the time between two consecutive fixations. Eye-tracking technologies, however, cannot rely on fixations alone. Fixations are not indicative of attention in particular, as a student can be paying attention to other off-task computer applications. Moreover, a student can also be focused on a chat window to type a query for the teacher, which can lead to fixation on the edge of the screen. However, if the focused region does not cover the chat window and covers the lecture window alone, then the student can be penalized for wrong analysis. This requires eye-tracking technologies to consider blink velocity, saccadic velocity and eyelid’s degree of openness for deriving a meaningful approximation of a student’s attention.
Decreased saccadic velocity is indicative of fatigue, and increased saccadic velocity is suggestive of enhanced task difficulty. Reduced blink velocity and reduced degree of openness of eyelids symbolize increased fatigue. These parameters need to be factored in by the online learning platforms to analyze students’ attention patterns and optimize the teaching methodologies autonomously.
Gaze interaction systems collect eye gaze-related information and infer what the student is attending to on the digital device screen. If the eye dwell time on a certain non-academic related section of the screen crosses a pre-defined threshold, the eye-tracking system will detect possible distraction. Moreover, by checking the finer eye movements, pupil dilation and eye position changes due to head movements, inferences about the students’ area of concentration and highlight the current learning topic to bring back the students’ attention. Lower magnitude of pupil dilation and lack of focus of eye gaze on the academic section of the screen can also indicate loss of interest or fatigue. Eye-tracking mechanisms also need to monitor any incidence of fixated gaze on the screen, as it may suggest a fake image setup before the webcam to deceive the monitoring system.
Eye-tracking devices
Eye-tracking systems exist in two categories in the market, namely remote systems and head-mounted systems. The head-mounted systems are characterized by special devices that students require to wear as a headgear. The head-mounted devices are found to be suitable for studying the attention levels of a student as they closely monitor the eye movements and the head movements during an online education session. However, these do lead to discomfort for the students as they feel burdened by a helmet-like headgear atop their head at all times. The head-mounted devices direct infrared light towards the center of the eyes, where the pupil is usually located. The detectable reflections in the pupil and the cornea are detected by the systems and the vector between them is calculated. The corneal reflections enable optical tracking of eye movements. Since infrared light is not detectable by a human eye, the students don’t face distractions because of the ongoing eye-tracking.
The remote systems have a single or multiple cameras positioned in front of the students. These cameras can be integrated into the monitor of the laptops or other hand-held devices, and stay inconspicuous to the students’ eye. The remote eye-tracking systems enable cognitive load measurement by analyzing the pupillary response to visual stimuli on the digital device screen. These recording devices make the online learning sessions much similar to physical classrooms, where students are not required to wear any special gadget on their head while attending a session. However, since webcams are unable to emit targeted infrared light of the same quality as the head-mounted systems, the dependence on normal light sources result in lower contrast and subsequently, lower accuracy of eye-tracking. Specular reflections result from natural light, whereas infrared rays allow for a precise differentiation between the pupil and the iris.
Tracking of head movement
Eye-tracking alone is not enough for attention monitoring. Students can place a physical or digital image, or a short video clip of them paying attention at the center of the screen of the learning device. These methods can deceive a remote eye-tracking system. However, images are immovable constants, while video clips shall display a repeating pattern of the same head movement. Head movement tracking mechanisms are required for attention monitoring systems. Repeating patterns detected in head movement and body movement trigger an alert on a possible act of deceit by the student. Together with eye tracking, head movement monitoring add to the attention monitoring capabilities of the systems.
Active application window tracking
Possibilities of students running other applications in parallel to the online course exist. In such cases, the eye-tracking systems shall interpret the student to be actively engaged in the content, as the fixations and saccade metrics may appear to be in desirable range. However, students may be gazing at off-task applications, like chat applications, browser window or gaming portals. Therefore, a system to track the application active and currently in use is required to monitor students’ activities in real time. This also helps check the incidence of students using chat applications to share and copy answers from peers during a remote assessment. It can also detect possible off-task activities of students during an ongoing live education session.
Challenges to attention monitoring of students
Students’ attention monitoring technologies based on eye-tracking and head-movement monitoring can face a number of challenges. The primary among them is the lack of insight on the teaching methodologies being used by the instructor. Any detection of loss of attention that is recorded in the database is indicative of how focused a student had been during the online learning session. However, the tracking systems need to mature into comprehensive data collection and analysis tools that interpret the cause of attention lapses. An incidence of attention lapse is one student can be a case voluntary indulgence in off-task activities. However, simultaneous attention lapses in a number of students gives rise to primarily two possibilities. The first is the possibility of the instructor falling short of the responsibility of engaging the students, which results in boredom or disengagement of a number of students after a certain period of time. The second possibility is of peers engaging together in an academically-disadvantageous activity, like off-task chat using a private window, or social media activity in a group. Eye-tracking and head-tracking systems risk falling short of identifying the cause of distractions. This can render attention monitoring systems ineffective in data analysis, despite their efficacy in data collection.
Efficacy of the remote systems for eye-tracking is another concern. As described earlier in the text, infrared light has higher accuracy compared to natural light for tracking of the pupil, iris and the cornea. Remote systems integrated into the digital device screens may suffer from lack of precision during data collection, which may result in inaccurate reporting of attention level of a student. Moreover, prolonged exposure to infrared radiation can cause a gradual and irreversible opacity of the eye lens. Damage to the retina done by infrared rays can lead to scotoma, which results in loss of vision. Redness of eye, hemorrhaging and swelling of eyes are other detrimental effects of exposure to infrared radiation.
The eye-tracking systems also rely on the real-time monitoring of the currently active application window in the learning device. However, students can also place their non-learning devices and lean them on the screen. A student watching a movie on a personal smartphone that is placed on top of the active video conferencing window will display high levels of attention, without putting the virtual classroom session window in the background. In such cases, the attention monitoring systems will also need to read and interpret the facial patterns of the students to detect any emotion of informality.
The attention monitoring tools installed on the learning devices can also occupy RAM space and consume processing power, which can interfere with the seamless streaming of the online virtual classroom sessions. Course content developers need to ensure that the monitoring tools they advocate don’t intervene with the seamless operation of the virtual classrooms.
Implications
Virtual education has placed emphasis on the need of monitoring the academically-disadvantageous behavior of students. Attention monitoring systems need to be integrated into the online course infrastructure for effective monitoring of students’ off-task behavior. The pedagogical tools of the future will rely on tracking of eyes and head movements, which are suggestive of the emotions and attention level of the participants in an online classroom session. Monitoring the finer details of the eye movements also provide deeper insights into the learning patterns and attention span of various students.
Challenges in the form of lack of insights on the instructor, impact of eye health and technical issues arising out of real-time monitoring exist. The virtual course developers need to factor in these challenges and design their attention monitoring methodologies that balance the benefits against the challenges. With the growth in remote education, instructors will have to rely heavily on technology to maintain a physical classroom-like decorum in the online space.
