CALM: Continual Associative Learning Model via Sparse Distributed Memory

Sparse Distributed Memory (SDM) provides a biologically inspired mechanism for associative and online learning. Transformer architectures, despite exceptional inference performance, remain static and vulnerable to catastrophic forgetting. This work introduces Continual Associative Learning Model (CALM), a hybrid framework that integrates SDM with lightweight dual-transformer modules to enable continual, lifelong adaptation without retraining. The architecture maintains an always-online associative memory for episodic storage (System 1) while pair of asynchronous transformer consolidate experience in the background for uninterrupted reasoning and gradual model evolution (System 2). The framework remains compatible with standard transformer benchmarks, establishing a shared evaluation basis for both reasoning accuracy and continual learning stability. Preliminary experiments using the SDMPreMark benchmark evaluate algorithmic behavior across multiple configuration sets, revealing a critical radius-threshold phenomenon in SDM recall and demonstrating near-linear computational scaling with substantial memory efficiency over dense-vector systems. These results represent deterministic characterization of SDM dynamics, establishing empirical baselines for future integration with transformer-based semantic tasks. Planned benchmarks on continual-learning and reasoning datasets (e.g., GLUE, SQuAD, episodic recall) will quantify retention stability, adaptation speed, and long-term reasoning performance. The CALM framework thereby provides a reproducible foundation for studying continual memory and associative learning in hybrid transformer architectures.