...
首页> 外文期刊>Journal of vision >Contextual location information relevant to visual search in natural scenes is encoded in extrastriate visual cortex and the anterior intraparietal sulcus
【24h】

Contextual location information relevant to visual search in natural scenes is encoded in extrastriate visual cortex and the anterior intraparietal sulcus

机译:与自然场景中的视觉搜索相关的上下文位置信息被编码在外视皮层和顶壁前沟中

获取原文

摘要

Humans are highly adept at detecting objects within natural scenes taking advantage of contextual cues to increase detection and recognition performance (Oliva & Torralba, 2007). We show that both retinopically defined dorsal areas (V3d, V3A, V3B) and ventral functionally defined areas (LO, PPA) contain information able to accurately discriminate the location contextually relevant to a cued target object. Eleven observers viewed a series of natural images and determined whether a cued object was present whilst whole brain fMRI data was recorded. Trial durations were 4.86 seconds with an initial 400 ms presentation of the cue word followed by a 250 ms ISI and stimulus presentation of 250 ms. Stimuli consisted of natural scenes (640 images, 17.5?° square) with targets present in 50% of the images. Critically, target absent images were selected so that a single expected location (left/right lateralized) was contextually consistent with the cued target word. Experimental runs consisted of 16 presentations of each condition with observers completing 10 runs. We used multivariate pattern analysis to assess the accuracy of discriminating the location contextually relevant to the cue word. Only object absent scenes were used to isolate effects of context from that of the target objects. Accuracy for all regions-of-interest was greater than chance for identifying the location (left vs. right) contextually relevant to the cue. Accuracies were highest in dorsal areas V3d, V3B and the anterior intraparietal sulcus. In contrast, discrimination accuracy for the location (left vs. right) which contained the most salient object within the scene (determined independently [n = 100]) showed that only higher visual areas displayed significantly greater accuracy for context relative to saliency. Our results suggest that perceptual decisions utilizing contextual information for target location involve interactions between dorsal and ventrolateral visual areas containing task-relevant information and mediating parietal cortical areas governing attention.
机译:人类非常善于利用上下文线索来检测自然场景中的物体,以提高检测和识别性能(Oliva&Torralba,2007)。我们显示,既是视网膜定义的背侧区域(V3d,V3A,V3B)又是腹侧功能定义的区域(LO,PPA)均包含能够准确地区分与提示目标对象相关的位置的信息。 11位观察者查看了一系列自然图像,并确定了在记录全脑fMRI数据的同时是否存在提示的物体。试验持续时间为4.86秒,其中提示词的初始显示时间为400 ms,随后是250 ms的ISI和刺激时间,显示为250 ms。刺激由自然场景(640张图像,17.5?°正方形)组成,目标出现在50%的图像中。至关重要的是,选择了目标缺失图像,以使单个预期位置(左/右横向化)与提示的目标单词在上下文中保持一致。实验运行包括每种情况的16次演示,观察者完成10次运行。我们使用多元模式分析来评估区分与提示词相关的位置的准确性。仅使用没有对象的场景将上下文效果与目标对象的效果隔离开。所有感兴趣区域的准确性都比在上下文中确定与提示相关的位置(左与右)的机会要大。在背侧区域V3d,V3B和前壁顶沟中的准确性最高。相反,包含场景中最显着物体的位置(左与右)的辨别精度(独立确定[n = 100])显示,相对于显着性,只有较高的视觉区域显示出更高的上下文准确度。我们的结果表明,利用上下文信息进行目标位置的知觉决策涉及背侧和腹侧视觉区域之间的相互作用,该区域包含与任务相关的信息,并调节顶叶皮层区域来控制注意力。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号