Learning by the Intelligent Pedagogical Agents: Can Eye Tracking Improve Attention Performance?

Attention-deficit/hyperactivity disorder (ADHD) is a persistent condition associated with impairment in educational functioning, professional position, and social relationships (Fabio and Caprì, 2015). The disorder is determined by three basic symptoms: inattention, hyperactivity, impulsivity (APA, 2013). The disorder is closely linked to a series of neuropsychological deficits such as executive functions, working memory, and cognitive processes.
 Problem-solving is one of the areas in which ADHD students experience problems as a result of deficits in attention and working memory. Although normal developing children may easily learn problem-solving skills; Children with ADHD need help to learn these skills, and instruction must be provided in a controlled manner for this purpose.
However, flexible learning environments in terms of content and presentation time can be helpfull to reduce the problems of these children. For example, the Attention of ADHD students can be improved when doing teacher-led homework, instead of independent tasks. However, due to the short range of attention, assignments should be presented in small pieces with just-in-time feedback on how to do it (Brock, Grove & Searls, 2010).
Although the presence of a teacher and the provision of guidance can be a positive aspect of face-to-face education, the need to spend special guidance, and personalized instruction according to the needs of these children, requires another solution.
The purpose of the present study was to investigate the role of the intelligent pedagogical agent on complex problem-solving ability and duration of task engagement in students with attention deficit disorder in the e-learning environments.
Pedagogical agents are virtual characters used in online learning environments to serve various instructional goals (Veletsianos and Russell, 2014). They have human-like gestures, speech, gaze, and behaviors to address some roles, such as tutor, coach, actor (Kim and Baylor, 2007)
The manifestation of the pedagogical agent is supported by theories such as Computers as Social Actors-CASA (Nass and Brave, 2005), Cognitive Load Theory (Sweller, 2011) that emphasize on the social aspect in the learning process. As Mayer (2014) mentions in "Presence principle" the students learn more deeply when on-screen agents display human-like gesturing, movement, eye contact, and facial expression. In general, what is discussed in the social effects of the pedagogical agent is a concept called "persona effect"(Veletsianos and Russell, 2014); This concept means that the presence of a human-like agent in interactive learning environments have a positive effect on the learner's perception of their learning experiences. the presence of an educational agent causes the learner to interpret their learning experiences as a result of interaction with the computer in the role of a social actor (Choi and Clark, 2006).
Thanks to the support of the theories proposed in the pedagogical agent literature, researchers have examined their roles on learning, performance, and motivation of learners. The results of the meta-analysis of Martha, and Santoso (2019) indicate the positive and significant role of pedagogical agents on learning and learners' performance.
Kim and Baylor (2016) in a meta-analysis found that in agent-based situations 1) learners expect agents to show acceptable teaching ability and motivation 2) and they learn the meaning of the material better and have a higher motivation in learning
Lin et al. (2020) investigated the effect of conversational style instruction with the presence of pedagogical agents on learning outcomes, cognitive load, and intrinsic motivation. Their study also showed that learning with a pedagogical agent is more attractive than learning without pedagogical agents.
Daradoumis and Arguedas (2020) in a study entitled "To foster reflective learning during metacognitive activities through the pedagogical agent" concluded that the presence of the pedagogical agent could lead to better scores in the process of reflection in the experimental group.
One of the areas that has been less addressed in research is the effect of pedagogical agents on attention, visual attention, and their guidance to related issues. If we consider the role of agents as a teacher in face-to-face space, then we can expect similar performance from them in online environments. In line with this assumption, the present study aims to investigate the role of the intelligent pedagogical agent on complex problem-solving ability and duration of task engagement in students with attention deficit disorder.
The target population was selected from 7 to 13 years old students (M 10.5) who were attending elementary and secondary school in public schools in Sicily, Italy. The sample group (N: 45) who selected by the Italian version of the ADHD Rating Scale for Teachers was randomly assigned to three groups: 1) without the agent 2) the agent only provided instructions during the problem solving 3) the agent provided instructions during the problem solving and provided feedback on learner's attention. The tools used in this study were the web-based complex problem-solving test and problem-solving time recording.
The test and situation were the same for three groups except for agent presence in groups 2 and 3; and in the group 3 by the interactive intelligent agent version, instructions and recommendations were designed based on the user's choices, and in addition, through the process of face recognition and image processing, the direction and position of the face and tracking of the eye movements.
Finally, in order to analyze the data, descriptive statistics, one-way analysis of variance, and Scheffe post hoc test was used.
The results showed that the presence of the intelligent pedagogical agent improves the performance of students with attention deficit disorder (F: 7/539, sig: 0.002) but the effect of agent on the time of learner’s engagement with the task was not significant. In general, based on the results, the guidance and feedback provided by the agent improves the attention process and consequently improves their performance in problem-solving.
According to the findings of the present study, although the effect of intelligent pedagogical agents has been proven in many studies, for specific groups same as students with attention deficit disorder it requires special instructional design. Accordingly, it is suggested to use technology such as image processing and eye-tracking as a visual element of attention in order to design intelligent environments.