Suppr超能文献

期望在人脸识别过程中引导预测性眼动和信息采样。

Expectations guide predictive eye movements and information sampling during face recognition.

作者信息

Garlichs Annika, Lustig Mark, Gamer Matthias, Blank Helen

机构信息

Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.

Hamburg Brain School, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.

出版信息

iScience. 2024 Sep 10;27(10):110920. doi: 10.1016/j.isci.2024.110920. eCollection 2024 Oct 18.

Abstract

Context information has a crucial impact on our ability to recognize faces. Theoretical frameworks of predictive processing suggest that predictions derived from context guide sampling of sensory evidence at informative locations. However, it is unclear how expectations influence visual information sampling during face perception. To investigate the effects of expectations on eye movements during face anticipation and recognition, we conducted two eye-tracking experiments ( = 34, each) using cued face morphs containing expected and unexpected facial features, and clear expected and unexpected faces. Participants performed predictive saccades toward expected facial features and fixated expected more often and longer than unexpected features. In face morphs, expected features attracted early eye movements, followed by unexpected features, indicating that top-down as well as bottom-up information drives face sampling. Our results provide compelling evidence that expectations influence face processing by guiding predictive and early eye movements toward anticipated informative locations, supporting predictive processing.

摘要

情境信息对我们识别面孔的能力有着至关重要的影响。预测性处理的理论框架表明,从情境中得出的预测会引导在信息丰富的位置对感官证据进行采样。然而,尚不清楚期望如何在面孔感知过程中影响视觉信息采样。为了研究期望对面孔预期和识别过程中眼动的影响,我们进行了两项眼动追踪实验(每项实验有34名参与者),使用包含预期和意外面部特征的提示面孔变形图像,以及清晰的预期和意外面孔。参与者会向预期的面部特征进行预测性扫视,并且比起意外特征,更频繁、更长久地注视预期特征。在面孔变形图像中,预期特征吸引了早期眼动,随后是意外特征,这表明自上而下以及自下而上的信息都驱动着对面孔的采样。我们的结果提供了令人信服的证据,即期望通过将预测性和早期眼动引导向预期的信息丰富位置来影响面孔处理,从而支持预测性处理。

https://cdn.ncbi.nlm.nih.gov/pmc/blobs/2fd8/11439840/1983cbec28e6/fx1.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