Research

Preface

This document is for organizing Koike’s research project and explaining the immediate research topic with great visions. These visions aim to tackle the ultimate question — How can we develop systems to optimize human thinking and learning? To address this question, we explore several projects using different approaches. In this document, we introduce our research area and three projects that aim to tackle this profound and fascinating question.

Our Core Technologies, Theories, and Perspectives

Our research aims to optimize learning activities by treating learners’ cognition as an information processing process (representationalism, cognitive science) and manipulating and controlling the information sources processed by learners to alter the learning activities themselves (constructivism, operationalism, psychophysics). Therefore, we focus on understanding the characteristics of learning as information processing (learning sciences, educational psychology, pedagogy), designing information structures as sources of information processing (ontology, knowledge engineering, applied ontology/ontology engineering), and designing learning tasks and activities with these information structures (model-driven, learning engineering). Furthermore, we implement these learning tasks and activities in systems, incorporating knowledge processing and inference (computer science), along with interaction mechanisms to maintain and promote motivation (human-computer interaction, cognitive psychology, positive psychology), to realize intelligent tutoring systems (artificial intelligence, educational technology). By doing so, we aim to approach the essence of learning (analysis-by-synthesis). In short, our basic research process is to deeply understand how learners process, understand, and apply information and provide optimal support tailored to that process.

In this process, we strongly believe that the design of learning tasks and activities is particularly important. Learning tasks refer to specific problems or tasks that learners engage with, and learning activities encompass the entire process of engaging with these tasks. These are closely related; effective learning tasks naturally lead to high-quality learning activities, supporting learners’ cognitive activities as a whole.

Designing based on cognitive processes necessitates a detailed analysis of the learning tasks and the problem-solving process. By clarifying how learners process information and which steps they should take, learners can engage with tasks more naturally. This approach allows them to focus on essential learning without experiencing unnecessary cognitive load. Specifically, we aim to provide the necessary information at the appropriate timing and design steps that enable learners to deepen their understanding efficiently.

Moreover, implementing intelligent diagnosis and feedback is crucial. Learning tasks designed with this approach can analyze learners’ performance in real-time and provide optimal diagnosis and feedback tailored to each individual. For instance, by instantly identifying where learners are struggling and offering immediate support, we maximize the effectiveness of learning.

Our research aims not only to remain theoretical but also to be applied in real educational settings, demonstrating its effectiveness. Through specific learning tasks and the intelligent diagnosis and feedback based on them, we strive to optimize learners’ cognitive processes, fostering deeper understanding and sustained learning motivation.

Related Research Areas

Our research area can be succinctly described as “Intelligent Tutoring Systems.” These are systems that intelligently support human learning. Intelligent Tutoring Systems (ITS) are computer systems that provide personalized instruction or feedback to learners. They use artificial intelligence techniques to adapt to individual learners’ needs and provide tailored support. ITS can help learners acquire new knowledge and skills more effectively by providing targeted instruction, practice, and feedback. By analyzing learners’ performance and progress, ITS can identify areas where learners need additional support and provide them with appropriate resources. ITS can also help instructors track learners’ progress and identify areas for improvement in their teaching methods. Overall, ITS can enhance the learning experience by providing personalized support and guidance to learners.

The history of ITS is extensive, and details can be found in Wenger’s book [Wenger, 1987]. However, considering that ITS is a vast interdisciplinary field, we present the areas related to ITS as envisioned by me in Figure.

1. Science

   • Cognitive Psychology: Study of mental processes like attention, language, memory, perception, problem-solving, creativity, and thinking.
   • Cognitive Science: Interdisciplinary study of the mind and its processes, including how cognition works.
   • Learning Sciences: Field aiming to understand and improve learning scientifically.
   • Educational Psychology: Study of how people learn, focusing on student outcomes, instructional processes, and learning differences.
   • Pedagogy: Art and science of teaching, dealing with educational theory and practice.
   • Psychophysics: Study of relationships between physical stimuli and mental phenomena.
   • Positive Psychology: Study of what makes life most worth living.

2. Engineering

   • Artificial Intelligence: Simulation of human intelligence processes by machines.
   • Computer Science: Study of algorithmic processes, computational machines, and computation.
   • Knowledge Engineering: Development of intelligent systems that can reason and learn.
   • Applied Ontology/Ontology Engineering: Practical applications of ontological analysis to structure knowledge.
   • Educational Technology: Facilitating learning and improving performance through technology.
   • Learning Engineering: Applying learning sciences to design effective learning experiences and environments.
   • Human-Computer Interaction: Design and evaluation of interactive systems for human use.

3. Philosophy

   • Ontology: Philosophical study of being, existence, and the categories of being.
   • Constructivism: Theory that knowledge and meaning are generated from experiences.
   • Representationalism: Position that the world is composed of mental representations.
   • Operationalism: Defining concepts by the operations used to measure them.
   • Model-Driven: Research approach using models to represent and reason about phenomena.
   • Analysis-by-Synthesis: Process of generating hypotheses and comparing them to sensory data for validation.

Our Projects

We have the following projects and themes to tackle the ultimate question:

CHUNK: Componentization of Human Understanding and Knowledge

Objective

To understand and decompose the fundamental components of human cognition and knowledge. This project aims to develop systems that support programming and computing education by breaking down complex concepts into manageable chunks.

Themes

• BROCs (Building method that Realizes Organizing Components) [Koike, 2017; Koike, 2019; Koike, 2021; Koike, 2023]
• Problem-Solving Process Model [Koike, 2019; Koike, 2020]
• Compogram [Koike, 2020]
• BEAR (Program Behavior Analyzer) [Koike, 2023]
• etc.

Keywords

• Learning by Chunking
• Indexing of Information
• Programming Education
• Computing Education
• Function-Behavior-Structure
• Working Memory
• Knowledge Components

Relation to Vision

By understanding and modeling the components of human knowledge, CHUNK contributes to the development of personalized and effective educational tools, thereby optimizing thinking and learning.

CLOVER: Computational Learning Optimization with Variform External Representations

Objective

To enhance learning through the use of diverse external representations and error-based simulations. This project addresses the psychological and cognitive barriers to learning, providing robust support systems that encourage learners to persist in their efforts and view mistakes as learning opportunities.

Themes

• EBS (Error-based Simulation) [Koike, 2021; Aikawa, 2023]
• TAME (Teachable Agent Module for Error-visualization) [Koike, 2022; Koike, 2023]
• ELMER (Explainable Model for Learning from Errors with Multiple External Representations) [Tomoto, 2024]
• etc.

Keywords

• Learning from Errors
• Learning from Failure
• Productive Failure
• Trial-and-Error
• Error-visualization
• Error-based Simulation
• Multiple External Representations

Relation to Vision

CLOVER supports cognitive and educational development by promoting trial-and-error learning, helping learners understand and optimize their learning processes through the use of multiple external representations and error-based simulations.

OCEAN: Optimizing Cognition by Engagement of Agent’s Navigation and Negotiation

Objective

To develop intelligent agents that assist learners in navigating complex information environments and making informed decisions. This project integrates the principles of cognitive apprenticeship to enhance self-regulated learning and decision-making abilities.

Themes

• Robot’s Eye Color to change Academic Emotions [Koike, 2019]
• WHALE (Wise Helper Agent for Learning Environment) [Koike, 2023]
• ARK (Action-Resource-Knowledge) Model [Koike, 2023]
• CORAL (Cognitively-Recalibrated Adaptive Learning)
• etc.

Keywords

• Present Bias
• Self-Regulation
• Self-Regulated Learning
• Motivation
• Behavioral Economics
• Decision Support
• Pedagogical Agents
• Teachable Agents
• Negotiation-Driven Learning

Relation to Vision

By optimizing information processing and decision-making, OCEAN helps learners thrive in information-rich environments, supporting effective decision-making and self-regulated learning.

Acknowledgement(s)

I would like to acknowledge many colleagues, students, and co-researchers who have discussed with me in the past and present. In particular, I sincerely thank Takahito Tomoto, Tsukasa Hirashima, Tomoya Horiguchi, Hiroaki Ogata, Izumi Horikoshi, Rwitajit Majumdar, H. Ulrich Hoppe, and Riichiro Mizoguchi. Their contributions have significantly shaped my current research agenda.