AUTHOR=Gu Jin , Liu Baolin , Yan Weiran , Miao Qiaomu , Wei Jianguo TITLE=Investigating the Impact of the Missing Significant Objects in Scene Recognition Using Multivariate Pattern Analysis JOURNAL=Frontiers in Neurorobotics VOLUME=14 YEAR=2020 URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2020.597471 DOI=10.3389/fnbot.2020.597471 ISSN=1662-5218 ABSTRACT=

Significant objects in a scene can make a great contribution to scene recognition. Besides the three scene-selective regions: parahippocampal place area (PPA), retrosplenial complex (RSC), and occipital place area (OPA), some neuroimaging studies have shown that the lateral occipital complex (LOC) is also engaged in scene recognition processing. In this study, the multivariate pattern analysis was adopted to explore the object-scene association in scene recognition when different amounts of significant objects were masked. The scene classification only succeeded in the intact scene in the ROIs. In addition, the average signal intensity in LOC [including the lateral occipital cortex (LO) and the posterior fusiform area (pF)] decreased when there were masked objects, but such a decrease was not observed in scene-selective regions. These results suggested that LOC was sensitive to the loss of significant objects and mainly involved in scene recognition by the object-scene semantic association. The performance of the scene-selective areas may be mainly due to the fact that they responded to the change of the scene's entire attribute, such as the spatial information, when they were employed in the scene recognition processing. These findings further enrich our knowledge of the significant objects' influence on the activation pattern during the process of scene recognition.