多感觉模态信息处理的认知神经基础

The Neurocognitive Basis of Multisensory Processing

  • 摘要: 多感觉模态信息处理是人类基本的心智功能,是漫长进化塑造而成的稳健、灵活、高效的信息加工方式,是贯穿感知、注意、记忆、语言、学习的重要特性。解析人脑多模态信息处理的认知神经和计算机制,对理解大脑功能组织原则有重要理论意义,对启发多模态人工智能发展有重要实践价值。然而,人脑的多模态信息处理机制异常复杂,来自神经科学、认知科学等领域的研究者从多种问题视角开展研究,分别提出了不同的多模态信息处理原则、理论、计算模型。系统回顾、梳理人脑多模态信息处理机制,对于准确把握该领域的前沿进展、启发后续研究非常关键。以“视听感知”这一重要多模态信息处理功能为切入点,在神经元、皮层、大脑通路等层面综述了人脑多模态信息处理的认知神经机制。首先,介绍了人脑多感觉神经元的功能特性和响应原则;其次,讨论了初级皮层间跨模态信息调制的关键机制;再次,分析了以多感觉脑区为核心的视听加工大脑网络的主要功能和计算机制;最后,讨论了人脑多模态信息处理机制对多模态人工智能系统开发的启示。

     

    Abstract: Multisensory processing is a fundamental cognitive function that underlies perception, attention, memory, language, and learning. Understanding the neurocognitive mechanisms of this function is of great theoretical and practical significance for understanding the principles of the human mind and for guiding the development of multisensory artificial intelligence. However, the mechanisms of multisensory information processing are extremely complex, and systematic review is essential for accurately understanding the cutting-edge progress in this field and inspiring future research. In this review paper, we first introduced the functional characteristics and response principles of multisensory neurons in the brain. Then, we discussed the mechanisms of cross-modal modulation between primary cortical areas. After that, we focused on the functions and computational mechanisms of multisensory areas and hierarchical multisensory brain networks. Lately, we discussed how studying human multisensory processing can inform the development of multisensory artificial intelligence.

     

/

返回文章
返回