Shared Spatial Trajectories Facilitate Working Memory of Multiple Sequences Through Neural Replay

핵심 개념

Shared spatial trajectories across feature domains are spontaneously leveraged to facilitate working memory storage of multiple sequences through compressed encoding and neural replay.

초록

This study investigated how common structures shared across different feature domains, such as spatial locations and colors, can be spontaneously employed to facilitate memory of multiple sequences in working memory (WM).

The key findings are:

1. Behavioral performance: Sequences with consistent color-location trajectories (aligned condition) show enhanced memory precision and a significant correlation between reproduced color and location sequence trajectories, compared to the misaligned condition.

2. Neural encoding: During encoding, the brain represents the shared 1st-to-2nd and 2nd-to-3rd trajectory information for the aligned condition, but not for the misaligned condition.

3. Neural replay: When preparing to recall the location sequence, the brain spontaneously reactivates the color sequence in a temporally compressed, forward replay manner for the aligned condition, but not for the misaligned condition.

4. Structure-behavior relationship: The neural representation of the shared trajectory structure correlates with the behavioral trajectory correlation between color and location sequences.

These findings suggest that shared common structures are leveraged for storage of multiple sequences in WM through compressed encoding and neural replay, facilitating efficient information organization. The brain spontaneously extracts and utilizes the underlying relational structure across feature domains to overcome the limited capacity of WM.

통계

Participants showed better memory precision for color sequences in the aligned condition compared to the misaligned condition (paired t-test, t(32) = 2.446, p = 0.020, Cohen's d = 0.426).
The correlation between recalled color and location sequence trajectories was significant for the aligned condition (1st-to-2nd trajectory: one-sample t-test, t(32) = 5.022, p < 0.001; 2nd-to-3rd trajectory: one-sample t-test, t(32) = 3.113, p = 0.004), but not for the misaligned condition.

인용구

"Shared common structure is leveraged for storage of multiple sequences through compressed encoding and neural replay, together facilitating efficient information organization in WM."
"Instead of memorizing two 3-item sequences independently, subjects may just maintain their common trajectories."
"Neural replay serves to consolidate sequences that share a common structure."

핵심 통찰 요약

Shared structure facilitates working memory of multiple sequences

by Huang,Q., Lu... 게시일 www.biorxiv.org 07-19-2023

https://www.biorxiv.org/content/10.1101/2023.07.18.549616v3

더 깊은 질문

How might the findings from this study on working memory organization extend to long-term memory and learning?

The findings from this study on working memory organization can provide valuable insights into long-term memory and learning processes. The spontaneous extraction and utilization of relational structures across different feature domains, as observed in the study, can be crucial for long-term memory encoding and retrieval. When information is organized and stored efficiently in working memory through common structures, it is likely to be transferred to long-term memory in a more structured and coherent manner. This can enhance the encoding and consolidation of memories, leading to better retention and retrieval in the long term. Additionally, the neural replay mechanisms identified in the study, such as forward replay of sequences, can play a role in memory consolidation during sleep, a process known to strengthen long-term memories. Overall, the findings suggest that the organization of information in working memory based on shared structures can have a significant impact on long-term memory formation and learning.

What are the potential limitations of the current experimental design, and how could future studies address them?

One potential limitation of the current experimental design is the reliance on EEG data alone to infer neural processes underlying memory organization. While EEG provides valuable insights into brain activity, it has limitations in spatial resolution and may not capture deeper brain structures involved in memory processes. Future studies could complement EEG data with other neuroimaging techniques such as fMRI or MEG to provide a more comprehensive understanding of the neural mechanisms at play. Additionally, the study focused on a specific type of memory task involving color and spatial location sequences, which may limit the generalizability of the findings to other types of memory tasks. Future studies could explore memory organization across a wider range of tasks and stimuli to confirm the robustness of the observed effects. Moreover, the study primarily focused on the encoding and retrieval phases of memory, and future research could investigate the role of memory organization during the consolidation and retrieval stages to provide a more complete picture of memory processes.

How might the spontaneous extraction and utilization of relational structures across domains relate to broader theories of human intelligence and cognitive flexibility?

The spontaneous extraction and utilization of relational structures across different domains, as demonstrated in the study, aligns with broader theories of human intelligence and cognitive flexibility. The ability to identify and leverage common structures to organize information reflects cognitive flexibility, which is essential for adapting to new situations and solving complex problems. This process of integrating information from diverse domains into a coherent framework is a hallmark of higher-order cognitive processes associated with human intelligence. The findings suggest that individuals can flexibly apply common structures to facilitate memory storage and retrieval, highlighting the adaptive nature of cognitive processes. This ability to extract and utilize relational structures across domains may underlie the capacity for abstract thinking, problem-solving, and creative reasoning, all of which are key components of human intelligence. Overall, the study's results support the idea that cognitive flexibility and intelligence are closely linked to the efficient organization of information based on shared structures.