Shape similarity, better than semantic membership, accounts for the structure of visual object representations in a population of monkey inferotemporal neurons.
Shape similarity, better than semantic membership, accounts for the structure of visual object representations in a population of monkey inferotemporal neurons.
复制标题
DOI：
10.1371/journal.pcbi.1003167
复制
发表时间：
2013
影响因子：
4.3
通讯作者：
Zoccolan D
中科院分区：
文献类型：
--
作者：
Baldassi C;Alemi-Neissi A;Pagan M;Dicarlo JJ;Zecchina R;Zoccolan D
前颞下皮质(IT)是灵长类动物处理视觉对象的视觉区域层次结构中的最高阶段。虽然有几条证据表明信息技术主要代表视觉形状信息，但最近的一些研究认为，信息技术中的神经元集合编码了视觉对象的语义成员(即代表概念类，如有生命和无生命的对象)。在这项研究中，我们通过对IT对一组视觉对象的反应进行多变量分析，调查了语义信息在IT中的表现程度，而不是纯粹的视觉信息。通过依赖各种机器学习方法(包括最近在统计物理领域开发的尖端聚类算法)，我们发现，在大多数情况下，视觉对象的IT表示取决于它们在形状水平上的相似性，或者更令人惊讶的是，低水平视觉属性的相似性。只有在少数情况下，我们观察到语义类的IT表示无法通过其成员的视觉相似性来解释。总体而言，这些发现重申了信息技术作为显性视觉形状信息传送者的主要功能，并揭示了信息技术在更大程度上表现了低水平的视觉特性，这一点超出了之前的认识。此外，我们的工作展示了如何结合各种最先进的多变量方法，并仔细估计形状相似性对对象类别表示的贡献，可以显著提高我们对大脑皮质视觉对象的神经元编码的理解。为了建立对外部单词的有意义的表征，到达我们感官的感觉信息流由大脑持续处理和解释。最终，这样的处理允许大脑将感觉(例如，视觉)输入安排到类别(如有生命和无生命的物体)和子类别(如面部、动物、建筑物、工具等)的层次结构中。重要的是，虽然许多物体可以基于它们的视觉相似性(例如，橙子和苹果)被分配到同一类别，但大多数类别的形成还需要将具有相似功能/意义但形状不相似的物体(例如，香蕉和苹果)任意地关联起来。关于视觉对象在大脑的高级视觉中枢(如颞下皮质；IT)的表征是否纯粹反映了形状相似，或者也(可能主要是)与形状无关的范畴知识，存在着长期的争论。在这项研究中，我们通过应用各种计算方法解决了这个问题。我们的结果表明，与类别成员相比，形状相似性更好地解释了一组颞下神经元的反应模式。这重申了IT作为视觉区域的主要功能，并展示了最先进的计算方法如何提高我们对大脑中神经元编码的理解。
The anterior inferotemporal cortex (IT) is the highest stage along the hierarchy of visual areas that, in primates, processes visual objects. Although several lines of evidence suggest that IT primarily represents visual shape information, some recent studies have argued that neuronal ensembles in IT code the semantic membership of visual objects (i.e., represent conceptual classes such as animate and inanimate objects). In this study, we investigated to what extent semantic, rather than purely visual information, is represented in IT by performing a multivariate analysis of IT responses to a set of visual objects. By relying on a variety of machine-learning approaches (including a cutting-edge clustering algorithm that has been recently developed in the domain of statistical physics), we found that, in most instances, IT representation of visual objects is accounted for by their similarity at the level of shape or, more surprisingly, low-level visual properties. Only in a few cases we observed IT representations of semantic classes that were not explainable by the visual similarity of their members. Overall, these findings reassert the primary function of IT as a conveyor of explicit visual shape information, and reveal that low-level visual properties are represented in IT to a greater extent than previously appreciated. In addition, our work demonstrates how combining a variety of state-of-the-art multivariate approaches, and carefully estimating the contribution of shape similarity to the representation of object categories, can substantially advance our understanding of neuronal coding of visual objects in cortex. To build meaningful representations of the external word, the stream of sensory information that reaches our senses is continuously processed and interpreted by the brain. Ultimately, such a processing allows the brain to arrange sensory (e.g., visual) inputs into a hierarchy of categories (such as animate and inanimate objects) and sub-categories (such as faces, animals, buildings, tools, etc). Crucially, while many objects can be assigned to the same category based on their visual similarity (e.g., oranges and apples), formation of most categories also requires arbitrarily associating objects sharing similar functions/meaning, but not similar shape (e.g., bananas and apples). A long-standing debate exists about whether the representation of visual objects in the higher visual centers of the brain (such as the inferotemporal cortex; IT) purely reflects shape similarity or also (and, perhaps, mainly) shape-unrelated categorical knowledge. In this study, we have addressed this issue by applying a variety of computational approaches. Our results show that the response patterns of a population of inferotemporal neurons are better accounted for by shape similarity than categorical membership. This reasserts the primary function of IT as a visual area and demonstrates how state-of-the-art computational approaches can advance our understanding of neuronal coding in the brain.
登录
查看更多内容
影响因子：
8.6
作者：
Bayati, M.;Borgs, C.;Zecchina, R.
通讯作者：
Zecchina, R.
DOI：
10.1016/j.cub.2010.03.050
发表时间：
2010-05-11
期刊：
Current biology : CB
影响因子：
--
作者：
Agam Y;Liu H;Papanastassiou A;Buia C;Golby AJ;Madsen JR;Kreiman G
通讯作者：
Kreiman G
影响因子：
2.5
作者：
Bell, Andrew H.;Hadj-Bouziane, Fadila;Ungerleider, Leslie G.
通讯作者：
Ungerleider, Leslie G.
影响因子：
8.2
作者：
Bar, M;Biederman, I
通讯作者：
Biederman, I
影响因子：
3.7
作者：
Clarke, Alex;Taylor, Kirsten I.;Tyler, Lorraine K.
通讯作者：
Tyler, Lorraine K.