A Vector-Symbolic Decision Making Model for Disaster Recovery
Research Poster Engineering 2025 Graduate ExhibitionPresentation by Meera Ray
Exhibition Number 207
Abstract
The placement of resource centers during disaster response and recovery has systematically deprioritized communities already marginalized before the disaster. We are developing a cognitive AI agent to show emergency managers how individuals make decisions in disaster scenarios depending on factors including their race, class, and location. This agent uses the computational cognitive architecture called ACT-R, but extended with a more scalable memory model called holographic declarative memory (HDM). We’ve adapted HDM to work with the most comprehensive and widely-used implementation of ACT-R, set up a text processing pipeline to add the contents of large documents to the agent’s memory, and most significantly created a useful and novel mechanism to retrieve an entire chunk of memory based on a request using only vector representations of tokens. Preliminary results indicate that we can maintain vector-symbolic advantages of HDM (e.g., chunk recall without storing the actual chunk and other advantages with scaling) while also extending it so that previous ACT-R models may work with the system with little (or potentially no) modifications within the actual procedural and declarative memory portions of a model. As a part of iterative improvement of this newly translated holographic declarative memory module we will continue to explore better time-context representations for vectors to improve the module's ability to reconstruct chunks during recall. To more fully test this translated HDM module, we also plan to develop decision-making models that use instance-based learning (IBL) theory, which is a useful application of HDM given the advantages of the system.
Importance
Our work spans disciplines: It contributes to cognitive science, AI, emergency management and critical disaster studies. To cognitive science, we contribute to the prediction and explanation of how society influences a human’s decision making through memory. To AI, we better ground models trained on large amounts of text in cognitive science in order to make the model’s decisions more explainable and better grounded in how our brains learn. For emergency management, our work aims to help government officials allocate strained public resources as natural disasters increase due to the climate crisis. For critical disaster studies, our work could help explain the persistently documented inequality in who receives public assistance after natural disasters.