Suppr超能文献

基于水平视差的场景表示的认知可穿透性。

Cognitive penetrability of scene representations based on horizontal image disparities.

机构信息

Department of Psychology, Stanford University, 450 Jane Stanford Way, Stanford, CA, USA.

Wu-Tsai Neuroscience Institute, Stanford University, 290 Jane Stanford Way, Stanford, CA, USA.

出版信息

Sci Rep. 2022 Oct 25;12(1):17902. doi: 10.1038/s41598-022-22670-7.

Abstract

The structure of natural scenes is signaled by many visual cues. Principal amongst them are the binocular disparities created by the laterally separated viewpoints of the two eyes. Disparity cues are believed to be processed hierarchically, first in terms of local measurements of absolute disparity and second in terms of more global measurements of relative disparity that allow extraction of the depth structure of a scene. Psychophysical and oculomotor studies have suggested that relative disparities are particularly relevant to perception, whilst absolute disparities are not. Here, we compare neural responses to stimuli that isolate the absolute disparity cue with stimuli that contain additional relative disparity cues, using the high temporal resolution of EEG to determine the temporal order of absolute and relative disparity processing. By varying the observers' task, we assess the extent to which each cue is cognitively penetrable. We find that absolute disparity is extracted before relative disparity, and that task effects arise only at or after the extraction of relative disparity. Our results indicate a hierarchy of disparity processing stages leading to the formation of a proto-object representation upon which higher cognitive processes can act.

摘要

自然场景的结构由许多视觉线索表示。其中主要的是由两只眼睛的横向分离视点产生的双目视差。视差线索被认为是分层处理的,首先是根据绝对视差的局部测量,其次是根据更全局的相对视差测量,从而提取场景的深度结构。心理物理学和眼动研究表明,相对视差与感知特别相关,而绝对视差则不然。在这里,我们使用 EEG 的高时间分辨率来比较仅隔离绝对视差线索的刺激与包含额外相对视差线索的刺激的神经反应,以确定绝对和相对视差处理的时间顺序。通过改变观察者的任务,我们评估了每个线索在认知上可穿透的程度。我们发现,绝对视差先于相对视差提取,并且只有在提取相对视差时或之后才会出现任务效应。我们的结果表明,视差处理的层次结构导致形成原始对象表示,更高的认知过程可以在此基础上进行。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/09d3/9596438/2e513afa61f0/41598_2022_22670_Fig1_HTML.jpg

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验