Friendly robot tutor teaching math in classroom

Can Robots Truly Understand and Boost Learning? The Rise of AI-Enhanced Education

Samir D’Costa in Mind & Education June 2025 • 5 min read.

"Explore how emotion and memory models in robots are revolutionizing mathematics education, offering personalized learning experiences and improved outcomes for students."

The integration of social robots in education is rapidly growing, particularly in how these robots can adapt to individual student needs. Recent studies highlight the importance of designing robots that understand and respond to user characteristics to enhance learning outcomes. Researchers are actively exploring machine learning methods to create adaptive behaviors in educational robots, with reinforcement learning algorithms gaining traction. These algorithms enable robots to learn and adjust their teaching strategies in real-time, significantly improving the effectiveness of tutoring compared to non-adaptive methods.

While many studies focus on implementing novel adaptation strategies in social robots, fewer explore their effectiveness in real-world settings within existing school curricula. Experts emphasize the necessity of conducting more in-the-field research to gain valuable insights into the future role of robots in education. Building on this need, recent work has developed emotion and memory models that enable robots to create memory accounts of children's emotional states. This advancement allows robots to tailor their behavior based on past interactions, enhancing long-term engagement and vocabulary learning.

This article delves into a pioneering study that applies an emotion and memory model to a social robot in a mathematics learning scenario. The primary aim is to assess the model's impact on children's engagement and learning in a real-life educational context. By implementing scenarios that closely resemble actual classroom activities, this research seeks to validate the effectiveness of adaptive robots in education beyond vocabulary learning. It also addresses a critical gap in understanding how robots can truly enhance learning in practical, curriculum-based settings.

Emotion and Memory Model for Personalized Learning

Friendly robot tutor teaching math in classroom

The emotion and memory model is designed to mimic how humans create and recall emotional memories. Humans store memories of both positive and negative events, categorizing them into conscious (explicit, declarative) and unconscious (implicit, procedural) memory systems. Emotional memory involves the implicit learning and storage of information about the emotional significance of events, while the memory of emotional situations refers to conscious memory. When emotional events occur, they are processed by the sensory system and transmitted to the temporal lobe or amygdala, forming either explicit or implicit memories. These memories are then retrieved when a related cue is encountered.

Based on this understanding, a computational model was developed for social robots to store conscious memories of users' emotional experiences. This model aims to make the robot more relatable and effective by incorporating these memories into future interactions. Both positive and negative events are defined within the model. Positive events are those where goals are achieved without immediate problems, while negative events are impediments that cause a loss in achieving a goal. Neutral events do not affect the outcome.

Key components of the emotion and memory model: Inputs: Game events, learning outcomes, and user emotional states. Emotional Event Calculation (EEC): Determines the type of emotional event based on inputs. Memory Mechanism Generation (MMG): Transmits information to the Memory Processing Unit and updates the database. Behavior Selection Unit (BSU): Selects appropriate behaviors or responses using a reinforcement learning framework.

In the BSU, an adaptive strategy is designed using a Multi-Armed Bandit (MAB) reinforcement learning framework. MAB is used to distribute resources to competing actions, with the goal of maximizing accumulated rewards. The robot selects from three classifications of actions, and the Exponential-Weight Algorithm for Exploration and Exploitation (Exp3) algorithm guides its decisions. This algorithm balances exploration and exploitation to optimize behavior, ensuring the robot adapts effectively to the child's learning needs and emotional state.

Future Directions

This study underscores the potential of integrating emotion and memory models into educational robots to enhance children's learning and social engagement. The findings suggest that robots equipped with these models can create more personalized and effective learning experiences. Future research should focus on evaluating these models in various real-life educational scenarios to further refine their impact on learning outcomes.

About this Article -

This article was crafted using a human-AI hybrid and collaborative approach. AI assisted our team with initial drafting, research insights, identifying key questions, and image generation. Our human editors guided topic selection, defined the angle, structured the content, ensured factual accuracy and relevance, refined the tone, and conducted thorough editing to deliver helpful, high-quality information.See our About page for more information.

This article is based on research published under:

DOI-LINK: 10.1145/3284432.3284451, Alternate LINK

Title: Emotion And Memory Model To Promote Mathematics Learning - An Exploratory Long-Term Study

Journal: Proceedings of the 6th International Conference on Human-Agent Interaction

Publisher: ACM

Authors: Muneeb Imtiaz Ahmad, Omar Mubin

Published: 2018-12-04

Everything You Need To Know

How do robots use emotion and memory to personalize the learning experience for students?

The core idea involves building robots that understand and respond to how students feel, creating 'memory accounts' of their emotional states. This allows the robot to adjust its teaching style based on past interactions, making learning more engaging and effective. This personalization is achieved through emotion and memory models which record the emotional impact of learning events.

Can you explain the key components of the Emotion and Memory Model used in educational robots and how they contribute to adaptive learning?

An Emotion and Memory Model mimics how humans store memories related to emotions, categorizing them into conscious and unconscious systems. The model takes in game events, learning results, and the student's emotional state. It includes an Emotional Event Calculation (EEC) to determine the type of emotional event, a Memory Mechanism Generation (MMG) to update the memory database, and a Behavior Selection Unit (BSU) that uses reinforcement learning to choose the best responses. If implemented accurately, the robot can modify its teaching behavior in real-time to optimize engagement and learning.

How does the Behavior Selection Unit (BSU) in social robots use reinforcement learning to adapt to a child's learning needs?

The Behavior Selection Unit (BSU) uses a Multi-Armed Bandit (MAB) reinforcement learning framework. Within this framework, the Exponential-Weight Algorithm for Exploration and Exploitation (Exp3) is employed. The Exp3 algorithm balances exploration (trying new strategies) and exploitation (using strategies that have worked well) to optimize the robot's behavior. This ensures that the robot adapts effectively to each child's unique learning needs and emotional state, improving engagement and learning outcomes.

Within the Emotion and Memory Model, what distinguishes positive, negative, and neutral events, and how do these classifications influence the robot's behavior?

Positive events are defined as situations where learning goals are achieved without problems. Negative events occur when something impedes the achievement of a goal. Neutral events are those that don't significantly affect the outcome. By categorizing events, the emotion and memory model can better understand the student's experience and adjust the robot's behavior accordingly. For example, if a student consistently experiences negative events when tackling a particular type of problem, the robot could adjust its approach to provide more support.

What are the next steps in researching and developing emotion and memory models for robots in education, and what aspects need further exploration?

While the initial results are promising, it is essential to evaluate these models in diverse, real-world classrooms and across different subjects beyond mathematics. Further research should focus on refining the emotion and memory models, and exploring new ways to personalize learning experiences. Additionally, future studies need to assess the long-term impact of these robots on student learning, social development, and emotional well-being.