摘要
人类和其他动物的大脑时刻接收来自外界复杂环境中的多感官信息.为快速有效地感知并做出决策和反应,大脑需要整合共同来源的感官信号,并分离独立来源的感官信号.因此,生物体需要解决多感觉信息处理中的“因果推断”问题.本文综述多感觉信息处理领域中从早期强制整合模型到近期贝叶斯因果推断模型的发展历程,总结多感觉因果推断中常用的实验范式及基本计算原理.在神经层面,大脑动态编码因果推断信息,呈现层级递进加工,具体体现为初级感觉皮层快速表征单模态信息,随后在顶颞区强制整合,最终达到额顶区进行因果推断.最后,未来研究可以整合多种实验范式,探索特殊人群中因果推断的神经机制,克服贝叶斯因果推断模型的局限性,并关注神经网络在解释其生理机制方面的潜力.
Humans and other animals continuously receive multisensory information from complex environments.To efficiently perceive,make decisions,and respond,the brain must integrate sensory signals from common sources while segregating those from independent sources.This process requires organisms to solve the causal inference"problem in multisensory information processing.This review traces the evolution of multisensory models,from the early forced integration model to the recent Bayesian causal inference model,and summarizes commonly used experimental paradigms and foundational computational principles in multisensory causal inference.At the neural level,the brain encodes causal inference information dynamically and hierarchically.Specifically,this process begins with the rapid representation of unimodal information in the primary sensory cortex,progresses to forced integration in parietotemporal areas,and ultimately culminates in causal inference within frontoparietal regions.Future research should integrate multiple experimental paradigms to further explore the neural mechanisms underpinning the causal inference model,especially in individuals with cognitive disorders.Additionally,it should address the limitations of Bayesian causal inference models and examine the potential of network models to reveal the neural basis of causal inference.
作者
程羽慧
袁祥勇
蒋毅
曹一男
CHENG YuHui;YUAN XiangYong;JIANG Yi;CAO YiNan(School of Psychology,Nanjing Normal University,Nanjing 210024,China;State Key Laboratory of Brain and Cognitive Science,Institute of Psychology Chinese Academy of Sciences,Bejing 100101,China;Department of Psychology,University of Chinese Academy of Sciences,Bejing 100049,China;Department of Neurophysiology and PathophysiologyUniversiy Medical Center Hamburg-Eppendorf,Hamburg 20246,Germany;Department of Cognitive Studies,Ecole Normale Superieure-PSL,Paris 75005,France)
出处
《中国科学:生命科学》
北大核心
2025年第7期1395-1408,共14页
Scientia Sinica(Vitae)
基金
国家自然科学基金(批准号:32400864)
南京师范大学引进人才科研启动基金(批准号:184080H201A45)
欧盟玛丽·居里学者人才计划(MSCA)独立研究基金(批准号:101154160)资助。
关键词
多感觉整合
贝叶斯因果推断
层级加工
计算模型
神经网络
multisensory integration
Bayesian causal inference
hierarchical processing
computational model
neural network
