Continuously Decoding Visual Perception and Working Memory

描述（由申请人提供）：早期视觉区域通过基于特征的编码对详细的视觉输入进行编码，从而为表示视觉刺激之间的细粒度差异提供有效的编码。例如，V1中的神经元选择性地对特定的视网膜位置、方向、空间频率、颜色、运动方向和双眼视差做出反应。我们已经从神经生理学研究中学到了很多关于这些特征调节特性的知识。最近，利用模式分类技术的功能磁共振成像研究已经允许检查人类视觉系统中的特征调谐特性。在这些研究中，研究人员训练一个fMRI活动分类器来区分与一小组刺激条件相关的活动模式，识别准确性提供了这些区域传达多少刺激条件信息的衡量标准。例如，通过功能磁共振成像（fMRI）的V1活动模式，可以分辨出参与者正在观看的是8个方向中的哪一个，类似的方法已经成功地识别了其他基于特征的反应模式。然而，感兴趣的刺激维度，如方向、运动或颜色是典型的连续变量，可以在一定范围内取任意值。此外，与这些刺激变量相关的皮层反应模式可能在一个连续体上变化。本提案的目标是开发一种方法来解释fMRI激活模式与在某些维度上经历的特征的实际值之间的关系，而不是区分一小组值。这种连续解码方法的存在将提供比目前可能的更丰富的实验范式，允许对视觉皮层中特征的表征进行新颖的研究。Aim 1的目标是使用连续解码从人类视觉皮层提取有关定向选择反应的新粒度信息。目的2是通过在视觉错觉体验中应用连续解码来研究这些定向选择反应与定向知觉之间的关系。目的3是研究在没有视觉刺激的情况下，工作记忆任务的行为表现与记忆特征的皮层表征之间的关系。该奖学金的总体目标是培训申请人使用功能磁共振成像，重点是解码方法。此外，申请人将通过指导阅读、每周研讨会和范德比尔特视觉研究中心教师讲授的课程，接受视觉科学领域的培训。该培训将使申请人成为一名独立研究人员的长期目标成为可能，开发新的功能成像数据分析技术，并利用它们进一步了解视觉感知和高阶认知的神经基础。

DESCRIPTION (provided by applicant): Early visual areas encode detailed visual input through feature-based coding, thereby providing an ecient code for representing ne-grained dierences between visual stimuli. For example, neurons in V1 respond selectively to particular retinotopic locations, orientations, spatial frequencies, colors, motion directions, and binocular dis- parities. We have learned a great deal regarding these feature-tuning properties from neurophysiology studies. More recently, fMRI studies utilizing pattern classication techniques have allowed for the examination of feature- tuning properties in the human visual system. In these studies, researchers train a classier on fMRI activity to discriminate between activity patterns associated with a small set of stimulus conditions, and discrimination accuracy provides a measure of how much information those areas convey about the stimulus conditions. For example, it is possible to discern which of eight orientations a participant is viewing from fMRI activity patterns in V1, and similar approaches have been successful in identifying other feature-based response patterns. However, the stimulus dimensions of interest, such as orientation, motion or color are typically continuous variables that can take on any value within some range. Moreover, the cortical response patterns associated with these stimulus variables likely vary on a continuum. The goal of this proposal is to develop methods that account for relationships between fMRI activation patterns and the actual value of a feature experienced on some dimension, rather than discriminating between a small set of values. The existence of such continuous decoding methods will provide for a far richer set of experimental paradigms than are currently possible, allowing for novel investigations into how features are represented in visual cortex. The goal of Aim 1 is to use continuous decoding to extract ne-grained information about orientation-selective responses from the human visual cortex. The goal of Aim 2 is to inves- tigate the relationship between these orientation-selective responses and perceptions of orientation by applying continuous decoding during experiences of visual illusions. The goal of Aim 3 is to study relationships between behavioral performance on a working memory task and cortical representations of the remembered features in the absence of visual stimulation. The overall objective of this fellowship is to train the applicant in the use o fMRI, with an emphasis on decoding methods. In addition, the applicant will be trained in the eld of vision science through directed reading, weekly seminars, and course work taught by faculty in the Vanderbilt Vision Research Center. This training will make possible the applicant's long-term goal of becoming an independent researcher, developing novel analytic techniques for functional imaging data and using them to further our understanding of the neural bases of visual perception and higher-order cognition.