Abstract:Substantial evidence suggests that working memory (WM) leverages relational representations to provide flexible support for cognitive functions, a capacity likely derived from the dynamic nature of neural codes in WM. However, how these dynamic codes represent and maintain relations remains unclear. Here, we examined the transformation of neural geometries in the dorsal prefrontal cortex of monkeys performing a visuospatial delayed-match/nonmatch task, where the monkeys were instructed to hold the spatial location of a white square in WM to match it with the spatial location of a subsequent square. We found that the sensory manifold during the square’s presence and the mnemonic manifold after the square’s offset both aligned with the stimulus manifold. However, significant differences emerged between the sensory and mnemonic manifolds, exhibiting little correlation in their neural geometries. Further analysis on the dynamic transformation from the sensory to mnemonic manifold revealed a process of expansion followed by flattening: the asymmetric sensory manifold first expanded into a symmetric neural geometry immediately after the square’s onset offset, which then gradually flattened along dimensions different from those initially expanded, culminating in an asymmetric mnemonic manifold. This dynamic process of reconstruction not only remained its faithfulness to the stimulus geometry but also gained the flexibility to meet task demands. In sum, this transformation from asymmetry to symmetry and back to asymmetry in neural geometry precisely illustrates the dynamics of memory reconstruction, shedding lights on the subjective nature of WM that generates both accurate and illusory representation of the world we lived in.

The superstring galaxy associative memory model with anticipant fault-tolerant field on m-value information space

A Fault-Tolerant Superconducting Associative Memory

The Geometric Invariance in Representing Multiple Objects in Visual Working Memory

Defect-Tolerance Analysis of Fundamental Quantum-Dot Cellular Automata Devices

How the Brain Formulates Memory: A Spatio-Temporal Model Research Frontier.

Spiking Neuron Based Cognitive Memory Model

Synaptic Conditions for Auto-Associative Memory Storage and Pattern Completion in Jensen Et Al.’s Model of Hippocampal Area CA3

A Computationally Efficient Associative Memory Model of Hippocampus CA3 by Spiking Neurons

Logic, Design & Organization of PTVD-SHAM; A Parallel Time Varying & Data Super-helical Access Memory

MemoryMamba: Memory-Augmented State Space Model for Defect Recognition

Multi-domain Compact Modeling for GeSbTe-based Memory and Selector Devices and Simulation for Large-Scale 3-D Cross-Point Memory Arrays

A General Auto-Associative Memory Model

Sparsely Connected Associative Memory Based on the Preferential Mechanism of Heuristic Annealed Topology

Non-Abelian Self-Correcting Quantum Memory

SUPER-ACTIVATING QUANTUM MEMORY WITH ENTANGLEMENT

Asymmetric Bidirectional Associative Memories.

Information is asymmetry: spatial relations were encoded by asymmetric mnemonic manifolds

Physical Mechanism Model of the Metal Magnetic Memory Testing Technology

A Chaotic Associative Memory

Sparsely Connected Associative Memory with Adaptive Topology Through Annealed Dilution

Cellular Neural Networks As a Model of Associative Memories