Human eye movements as a model of visual prediction

在自我运动过程中，眼球运动具有稳定凝视的重要功能，并将目光转移到感兴趣的位置或物体上。由于它们与视觉感知密切相关，眼动提供了一个很好的模型系统来回答关于视觉和运动背后的大脑机制的基本问题。我们的大部分研究依赖于平滑追踪眼球运动，即眼睛对视觉对象运动的关键反应。在我们之前由NSERC资助的研究中，我们对驱动感知和追求的视觉信号进行了表征和比较，发现了感知和行动之间的相似之处和显著的分离。 最近，我们发现，在需要预测的任务中，追踪眼动提供了一个重要的性能优势。许多真实世界的运动，如拦截猎物或击球，都涉及到通过对物体运动的短暂一瞥来预测物体的轨迹。是否、何时、在哪里拦截的决定是基于当前的感官对象信息，如棒球的旋转或速度，以及过去的感觉经验，如曲线球的概率。整合现在和过去的感觉信号的机制的有效性和时间尺度还知之甚少。 我们将开发新的实验范式，并依靠包括人类视觉心理物理学、眼球跟踪、手跟踪和计算建模在内的技术方法来研究这些整合机制。我们将在不同的加工水平上操纵感觉体验，从对特定物体运动发生的低级预期到高级规则和启发式。 建立视觉预测过程中整合和积累信息的时间动力学将大大有助于我们对视觉和运动控制的理解。此外，我们的实验范式对许多自然任务(如球类运动)具有很高的生态有效性，并可转换为可移植的应用程序，用于技能测试和培训。多样化的技术组合以及我们多学科和协作的研究方法将为学员提供理想的研究环境。

Eye movements have an important function to stabilize gaze during self-motion, and to shift gaze to locations or objects of interest. Because of their close connection to visual perception, eye movements provide an excellent model system to answer fundamental questions about brain mechanisms underlying vision and movement. Most of our research relies on smooth pursuit eye movementsthe eyes' key response to visual-object motion. In our previous NSERC-funded research, we have characterized and compared the visual signals that drive perception and pursuit, and discovered similarities and striking dissociations between perception and action. Recently, we found that pursuit eye movements provide an important performance advantage during tasks that require prediction. Many real-world movements, such as intercepting prey or hitting a ball, involve prediction of an object's trajectory from a brief glance at its motion. The decision whether, when and where to intercept is based on the integration of current sensory object information, like a baseball's spin or speed, with past sensory experience, like the probability of a curveball. The efficacy and timescale of the mechanisms underlying the integration of current and past sensory signals are poorly understood. We will develop novel experimental paradigms and rely on a combination of technological approaches including human visual psychophysics, eye tracking, hand tracking, and computational modeling to investigate these integration mechanisms. We will manipulate sensory experience at different levels of processingfrom low-level expectations about the occurrence of a particular object motion to high-level rules and heuristics. Establishing the temporal dynamics of integrating and accumulating information during visual prediction will significantly contribute to our understanding of visual and motor control. Moreover, our experimental paradigms have high ecological validity for many natural tasks such as ball sports, and can be translated into portable applications for skill testing and training. The diverse combination of techniques and our multi-disciplinary and collaborative research approach will provide trainees with an ideal research environment.