Vanderbilt Researchers Receive NSF Grant to Explore How Conversation Shapes Memory, Informing Advances in AI and Education
The U.S. National Science Foundation (NSF) has awarded a $446,272 grant to Vanderbilt University’s Peabody College of Education and Human Development to study how human conversation influences memory, research that could improve the way artificial intelligence (AI) systems communicate and enhance learning strategies in education.
The three-year project is led by Sarah Brown-Schmidt, professor of psychology and human development, and Deon Benton, assistant professor of psychology and human development. Together, they will examine the cognitive mechanisms that shape what people remember from conversation, combining human behavioral experiments with computational modeling to explore how language, memory, and interaction work together.
“We’ve discovered that people tend to remember best what they say themselves in conversation and have a comparatively worse memory for what was said to them,” Brown-Schmidt explained. “The words we choose, for example, describing a dog’s ‘fluffy ears’, can shape what details are encoded and later recalled.”
Using behavioral data from more than 100 participants, Benton will create a computational model simulating how conversational memory operates. This model will allow researchers to test hypotheses about how language patterns influence memory encoding and retrieval, offering insights that bridge human cognition and AI design.
Advancing Conversational AI through Cognitive Modeling
By modeling the dynamics of memory during natural conversation, the research could inform the development of AI systems that more effectively communicate with humans. Insights from this work may guide designers of large language models (LLMs) and other conversational agents to better anticipate what users are likely to remember or forget, improving responsiveness, accuracy, and user trust.
“If the programmers designing LLMs know that human users are not going to have a very good memory for certain aspects of a conversation, they could improve how the AI interacts with humans to account for those weaknesses,” Brown-Schmidt noted.
The project integrates experimental psychology and computational modeling, providing interdisciplinary training for graduate students, a key goal for developing future researchers equipped to work at the intersection of cognitive science, machine learning, and human-computer interaction.
Educational and Societal Impact
The findings may also inform evidence-based teaching practices. Understanding that people remember what they articulate themselves could encourage instructors to use more interactive approaches, prompting students to restate or reframe concepts in their own words to improve long-term retention.
Looking Ahead
Over the next three years, the study will deepen understanding of how language use shapes memory, providing foundational insights for AI system design, educational innovation, and cognitive science. The project underscores Vanderbilt’s commitment to research that connects human learning and technological development, advancing both fundamental science and real-world applications in an increasingly AI-driven society.