Development of visual object recognition

视觉物体识别的发展

• 批准号: 10132340
• 负责人: Najib Judeh Majaj
• 金额: $ 38.44万
• 依托单位: NEW YORK UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-04-01 至 2024-02-29
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10132340
• 关键词: 3-Dimensional; Adolescent; Adult; Affect; Age; Applications Grants; Area; Behavior; Behavioral; Behavioral Assay; Behavioral Mechanisms; Blindness; Brain; Categories; Child; Chronic; Cognition; Computer Models; Computer software; Confusion; Crowding; Data; Databases; Development; Ensure; Goals; Human; Image; Implant; Individual; Infant; Inferior; Life; Link; Macaca; Masks; Measurement; Measures; Methods; Monkeys; Nervous system structure; Neurons; Pattern; Perception; Performance; Physiological; Positioning Attribute; Primates; Process; Property; Psychophysics; Publishing; Research; Role; Sensory; Shapes; Site; Stimulus; Stream; Techniques; Training; Variant; Vision; Vision Disorders; Visual; Visual Cortex; Visual system structure; Work; awake; base; behavior measurement; course development; density; developmental plasticity; disability; experience; experimental study; extrastriate visual cortex; improved; juvenile animal; neural patterning; neurodevelopment; neuromechanism; neuron development; neurophysiology; nonhuman primate; object recognition; preservation; receptive field; relating to nervous system; response; targeted imaging; three-dimensional modeling; vision development; visual neuroscience

Project Summary A central goal of visual neuroscience is to understand how the brain supports perception. A powerful approach to studying the relationship between brain function and perception is to study the processes in parallel during development. View-invariant visual object recognition is a computationally difficult but highly ecologically relevant task. Visual function is poor in infants and emerges over different developmental trajectories depending on the type of visual function. Understanding the developmental trajectory of visual object recognition behaviorally and neurophysiologically will help constrain computational models that quantitatively link neuronal responses to behavior. Recent progress in neurophysiological measurement techniques have demonstrated convincingly that the ventral visual stream, IT cortex in particular, contains a robust representation of objects. The experiments we propose in this grant application aim to explore the links between the behavioral development of invariant object recognition and the functional development of visual area IT. First, we will use psychophysical techniques to longitudinally track the development of object recognition using and easy to train object-oddity task. We will explore shape similarity and identity-preserving image variation, two computational challenges the brain has to overcome to support invariant object recognition. We will also characterize the robustness of object recognition to spatial clutter during development. We will use images of three-dimensional objects placed on natural backgrounds and vary position, pose, and size. Second, we will directly assess neurophysiologically the development of neural response properties and receptive field organization of area IT to the same images of objects. The awake recordings will be interleaved with behavioral assays in the same individuals to directly correlate behavioral and neural development. These experiments will advance our understanding of the role of IT in support of 1) object recognition performance that is robust to identity preserving image variation, and 2) the limitations that neural development pose on the development of perception and cognition. Our greater goal is to elucidate the mechanisms by which the ventral stream processes form to support invariant object recognition.

项目摘要 视觉神经科学的一个核心目标是了解大脑如何支持感知。一种有力的方法 研究脑功能与感知之间的关系是在同时研究这些过程 发展。视觉视觉对象识别在计算上是一个困难但在生态上高度高度的 相关任务。婴儿的视觉功能很差，并且在不同的发育轨迹上出现 取决于视觉功能的类型。了解视觉对象的发展轨迹 在行为和神经生理上识别将有助于限制定量的计算模型 将神经元反应与行为链接。 神经生理测量技术的最新进展令人信服地表明 腹侧视觉流（特别是皮层）包含对象的强大表示。实验我们 在此赠款申请中提出的旨在探索不变的行为发展之间的联系 对象识别和视觉区域的功能开发。 首先，我们将使用心理物理技术来纵向跟踪对象识别的开发 并易于训练对象效果任务。我们将探索形状相似性和具有身份的图像变化， 大脑必须克服的两个计算挑战，以支持不变对象识别。我们也会 表征对象识别在发育过程中空间混乱的鲁棒性。我们将使用 三维物体放置在自然背景和不同位置，姿势和大小。 其次，我们将直接从神经生理学上评估神经反应特性的发展和 接受区域的现场组织到相同的对象图像。清醒的录音将交错 在同一个人中的行为分析中，可以直接相关行为和神经发展。 这些实验将提高我们对IT在支持1）对象识别的作用的理解 对保留图像变化的身份的性能，以及2）神经发展的局限性 构成感知和认知的发展。我们更大的目标是通过 腹侧流过程形成的以支持不变对象识别。