要在复杂的三维环境中自由活动，我们必须准确地估算自己与周围物体之间的相对运动，这个过程称之为自身运动知觉（self motion perception），需要多种感觉信息的共同参与,包括视觉、前庭、触觉以及本体感觉等。到目前为止，其形成的具体机制尚不清楚。本项目以清醒灵长类为研究对象，运用虚拟现实的技术将视觉和前庭刺激精确匹配，对自身运动进行模拟，采用电生理、行为心理学、神经信息学方法，并结合微电流刺激和可逆药物失活技术，来研究：1）不同皮层区域神经元活动与自身运动知觉之间的功能性联系；2）不同感觉信息编码所采用的参考系；3）视觉和前庭信息整合的统计学概率。另外，本项目将首次采用垂直方向的多通道电极在有眼动/头动的朝向判别任务下，对皮层的多个部位进行同步记录，从神经网络的层次上对自身运动知觉的多感觉整合机制进行研究，并为运动病和空间定向障碍的发病机制及预防提供理论依据。

To navigate effectively through a complex three-dimensional (3D) environment, we must accurately estimate our own motion relative to objects around us. Perception of self-motioni is a multi-modal process involving integration of visual, vestibular, and proprioceptive cues. Until now, the neural basis of multisensory integration for self-motion remains poorly understood. A major reason for this is the lack of a good animal model in which both the psychophysics and neurophysiology of sensory integration can be examined simultaneously. We have developed a virtual-reality motion system that allows us to flexibly present visual and inertial motion stimuli that can accurately simulate self-motion through a 3D virtual environment. We have also developed a multimodal heading discrimination task, in which we can demonstate that rhesus monkeys perceptually integrate visual and vestibular cues in a near-optimal fashion consistent with the Bayesian framework. These developments leave us uniquely positioned to study the neural mechanisms of multisensory integration during a behavior in which cue integration is known to be taking place. In the current proposal, three major goals are concerned: 1) how the neural activities related with self-motion judgments; 2) what is the reference frame of heading selectivity based on visual and vestibular cues; 3) how visual and vestibular sensory signals are properly integrated. We propose a combination of computational modeling, psychophysics, electrophysiology, electrical micro-stimulation and inactivationi aimed at elucidating the neural basis of optimal multisensory integration.