Affective Expressions in Conversational Agents for Learning Environments: Effects of curiosity, humour, and expressive auditory gestures
Loading...
Date
2021-12-13
Authors
Ceha, Jessy
Advisor
Law, Edith
Journal Title
Journal ISSN
Volume Title
Publisher
University of Waterloo
Abstract
Conversational agents -- systems that imitate natural language discourse -- are becoming an increasingly prevalent human-computer interface, being employed in various domains including healthcare, customer service, and education. In education, conversational agents, also known as pedagogical agents, can be used to encourage interaction; which is considered crucial for the learning process. Though pedagogical agents have been designed for learners of diverse age groups and subject matter, they retain the overarching goal of eliciting learning outcomes, which can be broken down into cognitive, skill-based, and affective outcomes. Motivation is a particularly important affective outcome, as it can influence what, when, and how we learn. Understanding, supporting, and designing for motivation is therefore of great importance for the advancement of learning technologies.
This thesis investigates how pedagogical agents can promote motivation in learners. Prior research has explored various features of the design of pedagogical agents and what effects they have on learning outcomes, and suggests that agents using social cues can adapt the learning environment to enhance both affective and cognitive outcomes. One social cue that is suggested to be of importance for enhancing learner motivation is the expression or simulation of affect in the agent. Informed by research and theory across multiple domains, three affective expressions are investigated: curiosity, humour, and expressive auditory gestures -- each aimed at enhancing motivation by adapting the learning environment in different ways, i.e., eliciting contagion effects, creating a positive learning experience, and strengthening the learner-agent relationship, respectively.
Three studies are presented in which each expression was implemented in a separate type of agent: physically-embodied, text-based, and voice-based; with all agents taking on the role of a companion or less knowledgeable peer to the learner. The overall focus is on how each expression can be displayed, what the effects are on perception of the agent, and how it influences behaviour and learning outcomes. The studies result in theoretical contributions that add to our understanding of conversational agent design for learning environments. The findings provide support for: the simulation of curiosity, the use of certain humour styles, and the addition of expressive auditory gestures, in enhancing motivation in learners interacting with conversational agents; as well as indicating a need for further exploration of these strategies in future work.
Description
Keywords
human-computer interaction, human-robot interaction, conversational agents, learning, education, pedagogical agents, social robots, motivation