MULTISENSORY INTEGRATION IN EXTRASTRIATE VISUAL CORTEX

描述（由申请人提供）：整合多种感官输入需要强健的感知和行为表现。最近的心理物理学研究表明，人类根据贝叶斯概率理论得出的统计最优加权方案来组合线索。当我们依靠两个不同的线索做出感知判断时，我们能够考虑到每个线索的可靠性，即使这种可靠性在不同的试验中随机变化。贝叶斯理论还预测，当存在两个感官线索时，与只有一个线索相比，行为表现会有所改善。一项涉及多感官整合的特别重要的任务是对自我运动或航向的估计。来自视觉（“光流”）和前庭线索的信息对头部感知都是有用的，但对线索整合的原理和神经基质知之甚少。利用复杂的虚拟现实系统，我们开发了一种新的行为范式来研究恒河猴的线索整合。在这里，我们建议测试关于背内侧颞上区（MSTd）和腹侧顶叶内区（VIP）在视觉/前庭线索整合中的作用的特定假设。在目标1中，我们将记录猴子在执行头部识别任务时的MSTd和VIP神经元，分别基于前庭信号单独、视觉信号单独和两种信号的组合呈现。我们假设具有“一致”视觉/前庭反应的MSTd/VIP神经元构成贝叶斯线索整合的神经基质，包括当视觉可靠性改变时线索重新加权。在目标2中，我们将通过使用微刺激和可逆化学失活技术操纵神经活动，探索MSTd/VIP神经元与头球知觉多感觉线索整合之间的因果关系。在目标3中，我们探讨了与视觉/前庭反应“一致”相比，MSTd/VIP神经元具有“相反”的功能作用。我们将测试特定的假设，其中最突出的是具有相反视觉/前庭偏好的神经元在消除由物体运动引起的视觉运动成分与由自我运动引起的视觉运动成分的歧异方面的潜在作用。这些目标的结合将为我们对多感觉皮层的理解提供一个根本性的突破，以及更普遍的大脑如何使用概率进行计算。我们发现的关于感觉统合的一般原则应该广泛应用于系统级神经科学的许多问题。了解多感觉统合和自我运动知觉的神经基础，也将促进治疗空间定向障碍的新策略，这是许多大脑功能障碍常见的，包括阿尔茨海默病。其中一个缺陷是通过光流判断方向的能力受损，而这种损害与患者在周围环境中导航的困难有关。在灵长类动物模型中更好地定位这些功能将有助于将新的阿尔茨海默氏症疗法靶向到适当的大脑区域。了解多感觉统合和自我运动知觉的神经基础也将促进治疗包括阿尔茨海默病在内的许多脑功能障碍常见的空间定向障碍的新策略。其中一个缺陷是通过光流判断方向的能力受损，而这种损害与患者在周围环境中导航的困难有关。在灵长类动物模型中更好地定位这些功能将有助于将新的阿尔茨海默氏症疗法靶向到适当的大脑区域。

DESCRIPTION (provided by applicant): Integration of multiple sensory inputs is required for robust perception and behavioral performance. Recent psychophysical studies indicate that humans combine cues according to a statistically optimal weighting scheme derived from Bayesian probability theory. When we make perceptual judgments that rely on two separate cues, we are able to take into account the reliability of each cue, even when this reliability varies randomly from trial to trial. Bayesian theory also predicts an improvement in behavioral performance when two sensory cues are present, as compared with only one cue. One particularly vital task that involves multisensory integration is the estimation of self-motion, or heading. Information from both visual (`optic flow') and vestibular cues can be useful for heading perception, yet little is known about the principles and neural substrates for cue integration. Using a sophisticated virtual reality system, we have developed a novel behavioral paradigm for studying cue integration in rhesus monkeys. Here we propose to test specific hypotheses regarding a role of the dorsal medial superior temporal area (MSTd) and ventral intraparietal area (VIP) in visual/vestibular cue integration. In aim 1, we will record from MSTd and VIP neurons while the monkey performs a heading discrimination task, based on vestibular cues alone, visual cues alone and combined presentation of both cues. We hypothesize that MSTd/VIP neurons with `congruent' visual/vestibular responses constitute a neural substrate for Bayesian cue integration, including cue re-weighting when visual reliability changes. In aim 2, we will probe for causal links between MSTd/VIP neurons and multi-sensory cue integration for heading perception by manipulating neural activity using microstimulation and reversible chemical inactivation techniques. In aim 3, we explore the functional roles of MSTd/VIP neurons having `opposite', as compared to `congruent', visual/vestibular responses. We will test specific hypotheses, the most prominent of which is a potential role of neurons with opposite visual/vestibular preferences on disambiguating the components of visual motion that are due to object motion from those due to self-motion. Combined these aims will provide a fundamental breakthrough in our understanding of multisensory cortex, as well as more generally how the brain computes using probabilities. The general principles that we uncover regarding sensory integration should have wide application to many issues in systems-level neuroscience. Understanding the neural basis of multisensory integration and self-motion perception would also promote new strategies for treating spatial disorientation deficits common to many brain dysfunctions, including Alzheimer's disease. One of these deficits is an impaired ability to judge heading from optic flow, and this impairment is correlated with patients' difficulty in navigating through their surroundings. Better localization of these functions in a primate animal model would help targeting new Alzheimer's therapies to the appropriate brain regions. PUBLIC HEALTH RELEVANCE Understanding the neural basis of multisensory integration and self-motion perception would also promote new strategies for treating spatial disorientation deficits common to many brain dysfunctions, including Alzheimer's disease. One of these deficits is an impaired ability to judge heading from optic flow, and this impairment is correlated with patients' difficulty in navigating through their surroundings. Better localization of these functions in a primate animal model would help targeting new Alzheimer's therapies to the appropriate brain regions.