Foveal and peripheral influences on the control of gaze and attention during scene perception: Experiments and mathematical modeling

高分辨率视觉仅限于我们视觉系统的中央凹，因为外围的视觉信息由于有限的敏锐度而模糊。中央凹对中高空间频率特别敏感，而外围则对低空间频率最敏感。因此，中央凹信息用于识别和详细分析注视刺激，周边信息对于空间定位和选择即将到来的扫视目标非常重要。该项目的科学目标是基于过程模型表征中央凹和周围信息处理对自然场景中眼球运动和注意力的时间控制的影响。在第一个资助阶段，我们在中央和外围视野中进行了9个使用注视随动高通和低通滤波的实验。此外，我们还建立了一个控制注视时间的数学模型。在该模型中，中心假设中央凹和外周加工对注视时间的控制有相反的动态影响。基于注视条件滤波的实验数据可以识别模型参数。第二阶段的主要研究目标是：(1)研究任务变化对眼球运动时间控制的影响；(2)对空间频率和颜色的范围进行经验描述，并随着时间的推移在视野的不同区域优先使用；(3)建立一个融合注视时间和空间控制的视觉场景中眼跳产生的数学模型。(4)最后，我们计划使用自然场景的注视随动空间频率滤波来创建一个带注释的眼动数据参考语料库。该语料库将为视觉注意和眼动控制模型的验证提供依据。

High-resolution vision is limited to the fovea of our visual system, since visual information in the periphery is blurred because of limited acuity. The fovea is particularly sensitive for medium and high spatial frequencies, while the periphery is optimized for low spatial frequencies. Therefore, foveal information is used for the identification and detailed analysis of fixated stimuli and peripheral information is important for spatial orienting and selection of the upcoming saccade target. The scientific goal of the project is the characterization of the influence of foveal and peripheral information processing on the temporal control of eye movements and attention in natural scenes based on a process model. During the first funding phase, we carried out 9 experiments using gaze-contingent high- and low-pass filtering in the central and peripheral visual field. Furthermore, we developed a mathematical model for the control of fixation durations. In the model, the central assumption was that there are opposing dynamical influences of foveal and peripheral processing on the control of fixation duration. Model parameters could be identified based on experimental data from gaze-contingent filtering. The main research goals for the second funding phase are as follows: (1) Investigation of foveal and peripheral influences on the temporal control of eye movements across task variations, (2) empirical description of the range of spatial frequencies and color, with preferential use in different regions of the visual field over time, (3) development of a mathematical model of saccade generation during scene viewing, which integrates fixation durations and spatial control. (4) Finally, we plan to create an annotated reference corpus of eye-movement data using gaze-contingent spatial frequency filtering of natural scenes. This corpus will be provided for the validation of models of visual attention and eye-movement control.