Suppr超能文献

空间频率上的轮廓整合。

Contour integration across spatial frequency.

机构信息

Department of Psychology, Methods Section, Johannes Gutenberg Universität Mainz, Staudinger Weg 9, 55099 Mainz, Germany.

出版信息

J Exp Psychol Hum Percept Perform. 2009 Dec;35(6):1629-48. doi: 10.1037/a0016473.

Abstract

Association field models of contour integration suggest that local band-pass elements are spatially grouped to global contours within limited bands of spatial frequency (Field, Hayes, & Hess, 1993). While results for local orientation and spacing variation render support for AF models, effects of spatial frequency (SF) have rarely been addressed. To explore whether contour integration occurs across SF, we studied human contour detection in Gabor random fields with SF jitter along the contour, and in the embedding field. Results show no impairment of contour detection when the contour elements are 1.25 octaves apart. Even with a SF separation of 2.25 octaves there is only moderate impairment. Because SF tuning functions measured for contextual interactions of neighbored single band-pass elements indicate much smaller bandwidths (Polat & Sagi, 1993), the results imply that contour integration cannot rest solely on local locking among neighbored orientation and SF tuned mechanisms. Robustness across spatial frequency, and across color and depth, as found recently, indicates that local orientation based grouping integrates across other basic features. This suggests an origin in not too distal brain regions.

摘要

轮廓整合的关联域模型表明,局部带通元素在空间上被分组为具有有限空间频率带宽的全局轮廓(Field、Hayes 和 Hess,1993)。虽然局部方向和间隔变化的结果支持 AF 模型,但很少涉及空间频率(SF)的影响。为了探索轮廓整合是否跨越 SF,我们在具有 SF 抖动的 Gabor 随机域中研究了人类轮廓检测,以及在嵌入域中。结果表明,当轮廓元素相隔 1.25 个倍频程时,不会损害轮廓检测。即使 SF 分离为 2.25 个倍频程,也只有适度的损害。由于为相邻单个带通元素的上下文相互作用测量的 SF 调谐函数指示带宽小得多(Polat 和 Sagi,1993),结果表明轮廓整合不能仅仅依赖于相邻的方向和 SF 调谐机制之间的局部锁定。最近发现的跨空间频率以及跨颜色和深度的稳健性表明,基于局部方向的分组可以整合其他基本特征。这表明其起源于不太遥远的脑区。

文献检索

告别复杂PubMed语法,用中文像聊天一样搜索,搜遍4000万医学文献。AI智能推荐,让科研检索更轻松。

立即免费搜索

文件翻译

保留排版,准确专业,支持PDF/Word/PPT等文件格式,支持 12+语言互译。

免费翻译文档

深度研究

AI帮你快速写综述,25分钟生成高质量综述,智能提取关键信息,辅助科研写作。

立即免费体验