Pedagogic conversational agents are interactive systems that allow students to dialogue with them about a certain domain to learn. PCAs have been used in multiple domains from pre-primary education to university, in roles such as teacher, student, or companion. In this chapter, Alcody, a PCA to teach programming to children, is enhanced with a new proposal to manage emotions in the dialogue with students. The goal is that when children are learning to program, Alcody can help them with the emotions associated to the learning. Six emotions have been integrated into Alcody: happiness, anger, sadness, fear, surprise, and disgust. A description of how a PCA to teach programming can modify its face and verbal expressions according to the emotion detected in the student. This is given for any other researcher that would like to incorporate emotions in dialogues between PCAs and students.
TopIntroduction
A Pedagogic Conversational Agent (PCA) can be viewed as a computer system that interacts with a student in a natural language to learn about a certain domain (Johnson, Rickel, & Lester, 2000). Such agents can assume the role of a lecturer, instructor, tutor, or even of a peer student. One of the first sample PCAs in the role of teacher was Herman the Bug (Lester, Converse, Kahler, Barlow, Stone & Bhogal, 1997), in the role of student was Betty (Biswas, Roscoe, Jeong & Sulcer, 2009), and in the role of companion was Jake & Jane (Arroyo, Woolf, Royer, & Tai, 2009), which incorporated some basic affective interaction.
Figure 1. Snapshot of JARO (source: Pérez-Marín & Caballero, 2013)
The use of PCAs has proved to have benefits for education. One reported benefit is “the Persona Effect,” which is described as the presence of an interactive agent in an educational computer environment that has a positive influence in the students’ perception of their learning experience (Lester et al. 1997). Another reported benefit is “the Proteus Effect,” according to which students are motivated to achieve the features of the agents and become more like them (Bailenson, Yee, Blascovich, & Guadagno, 2008; Yee & Bailenson, 2007). Finally, another reported benefit is “the Protégé Effect,” in which students can make greater efforts to teach their conversation agents than to study on their own (Chase, Chin, Oppezzo & Schwartz, 2009).
Previous work with agents such as JARO (Pérez-Marín & Caballero, 2013) show the viability of designing agents that dialogue with the students to learn about a certain domain. JARO was friendly and good evaluated in a proof-of-concept experiment with adults. However, in the last decades, advances in natural language dialogue, affective computing, machine learning, virtual environments, and robotics are making possible to design enhanced versions of PCAs with a higher potential impact on education (Johnson & Lester, 2018).
As can be seen in Figura 1, JARO is a static figure. It does not have any possibility of affective interaction, answering the same to all students irrespectively of their emotional state. The need of integrating some kind of affective interaction between the students and the agent was reinforced in the experiments performed with Dr. Roland (Tamayo-Moreno, 2017). Dr. Roland was designed to be used with children from Pre-Primary Education to Primary Education. Children regard PCAs as their friends. Friends have emotions, and learning can be enhanced when emotions are associated (Arroyo et al. 2009).
Moreover, it has also been detected an increasing interest in teaching programming to children (Pérez-Marín, Hijón, Bacelo, Pizarro, 2018). Alcody (Morales-Urrutia, Ocaña, Pérez-Marín, Tamayo, 2017) is a PCA designed to be used by children to teach them programming. In a first version, it was as Dr. Roland or JARO and it did not integrate affective interaction. In this paper, it is presented, the next step that is to enhance the agent with animated facial expressions and gestures, and able to manage the students’ emotions in the dialogue with them. In particular, six emotions have been integrated into Alcody: happiness, anger, sadness, fear, surprise, and disgust. A description of how a PCA to teach programming can modify its face and verbal expressions according to the emotion detected in the student is given for any other researcher that would like to incorporate emotions in dialogues between PCAs and students.