A Biologically Interpretable Cognitive Architecture for Online Structuring of Episodic Memories into Cognitive Maps
This work addresses the need for biologically interpretable models in AI and computational neuroscience, offering a more plausible approach for adaptive agents, though it is incremental in combining existing frameworks like Successor Features with episodic memories.
The paper tackled the problem of creating biologically plausible cognitive maps for artificial agents by proposing a novel architecture that uses local, Hebbian-like learning rules instead of global optimization, and demonstrated its efficacy in a partially observable grid-world by enabling autonomous organization of episodic memories into structured representations.
Cognitive maps provide a powerful framework for understanding spatial and abstract reasoning in biological and artificial agents. While recent computational models link cognitive maps to hippocampal-entorhinal mechanisms, they often rely on global optimization rules (e.g., backpropagation) that lack biological plausibility. In this work, we propose a novel cognitive architecture for structuring episodic memories into cognitive maps using local, Hebbian-like learning rules, compatible with neural substrate constraints. Our model integrates the Successor Features framework with episodic memories, enabling incremental, online learning through agent-environment interaction. We demonstrate its efficacy in a partially observable grid-world, where the architecture autonomously organizes memories into structured representations without centralized optimization. This work bridges computational neuroscience and AI, offering a biologically grounded approach to cognitive map formation in artificial adaptive agents.