Modeling causal inference in vestibular-visual heading perception

描述（由申请人提供）：头部方向由前庭信号和穿过视网膜的图像运动模式（光流）指示。前庭信号和光流相结合可提高定位精度。只有当两种线索是由同一原因产生时，即观察者在固定世界环境中的运动，这种多感觉整合才会有益。然而，光流具有固有的模糊性，这对头部感知提出了至关重要的挑战：光流表明了观察者和世界中物体的运动的结合。使用定量的人类心理物理学和数学模型，我在这里测试了一个假设，即大脑执行了一个决策过程，推断出与前庭和光流线索相关的原因。根据这种被称为“因果推理”的决策过程，如果推断出一个共同的原因产生了这两个线索，那么它们就被整合在一起。因此，因果推理预测，多感官头球知觉将取决于视觉和光流头球信息之间冲突的程度以及这些线索的可靠性。此外，我将在因果推理框架内研究运动物体对头球感知的影响，该框架预测运动物体对头球感知的影响将随着世界中物体运动速度的增加而减小。所有实验将使用一个带有大视场立体投影系统的6自由度运动平台和两种强迫选择方法。我将开发因果推理过程的数学模型，并将使用贝叶斯模型比较技术严格地将这些模型的预测与实验数据进行比较。本研究结果对理解基本的知觉前庭-视觉相互作用具有重要意义，为空间定向心理物理学的基础和临床研究开辟了新的方向。

DESCRIPTION (provided by applicant): Heading direction is indicated by vestibular signals and patterns of image motion across the retina (optic flow). Heading precision improves when vestibular cues and optic flow are integrated. Such multisensory integration will be beneficial only if the two cues are produced by the same cause, that is, movement of the observer in a world-fixed environment. However, optic flow possesses an inherent ambiguity which poses a crucial challenge to heading perception: optic flow indicates the combined movement of the observer as well as movement of objects in the world. Using quantitative human psychophysics and mathematical modeling, here I test the hypothesis that the brain implements a decision process that infers the causes associated with vestibular and optic flow cues. According to this decision process, termed "causal inference", the two cues are integrated if it is inferred that a common cause produced them. Causal inference thus predicts that the multisensory heading percept will depend on the magnitude of the conflict between visual and optic flow heading information as well as the reliability of these cues. In addition, I will investigate the influenceof moving objects on heading perception within the causal inference framework, which predicts that the influence of moving objects on heading perception will decrease as the velocity of object motion in the world increases. A 6 degrees-of- freedom motion platform with attached large field-of-view stereo projection system and two-alternative-forced- choice methodology will be used for all experiments. I will develop mathematical models of the causal inference process and will rigorously compare the predictions of these models with the experimental data using Bayesian model comparison techniques. Findings of this research are important for understanding basic perceptual vestibular-visual interactions and opening new directions in the fields of basic and clinical spatial orientation psychophysics.