Inhibitory Cross-Talk Enables Functional Lateralization in Attention-Coupled Latent Memory

Researchers developed a memory-augmented transformer that uses attention for retrieval, consolidation, and write-back operations, with lateralized memory banks connected through inhibitory cross-talk. The inhibitory coupling mechanism enables functional specialization between memory banks, achieving superior performance on episodic recall tasks while maintaining rule-based prediction capabilities.

• →The model partitions memory into left and right banks with sign-controlled cross-talk, where inhibitory coupling prevents collapse and enables specialization.
• →Excitatory cross-talk causes one bank to monopolize inputs, while inhibitory cross-talk actively suppresses contralateral activation for better task separation.
• →On symbolic benchmarks, the inhibitory model reduced cipher-domain loss by 124× over baseline while maintaining arithmetic performance.
• →The research demonstrates that persistent lateralized memory is necessary for episodic recall but not for rule-based prediction.
• →The approach is biologically motivated by inhibitory callosal projections in human cortex.