Microsaccades as an objective index to visual orienting and selection

点击翻译按钮获取中文摘要

Our prior work has shown that microsaccades (MS) reveal precise information both about the direction and timing of attention shifts and about the location of continuously sustained attention. Recent work with non-human primates suggests that microsaccades reflect the activity of so-called visual selection neurons in a functionally integrated network of anatomically distributed brain areas (superior colliculus, frontal eye fields, lateral intraparietal area). This 'selection map' network is thought to integrate several aspects of visual orienting and selection, including top-down attention, bottom-up saliency, and saccade preparation. We now propose a series of further human experiments to exploit the fact that microsaccades shadow these different aspects of orienting and selection. Specifically, we will combine our unmatched expertise in dual-task psychophysics with microsaccade observations and with computational modeling in order to characterize the interaction of top-down attention, bottom-up saliency, and saccade preparation in unprecedented detail.The particular objectives of the proposed project are as follows:(i) Quantify the modulation of MS statistics by top-down visual attention, bottom-up visual saliency, and saccade planning. (ii) Test the hypothesis that different orienting processes compete for a "limited resource" [Sperling & Dosher, 1986; Lavie, 2005]. the engagement of which is mirrored by MS statistics. (iii) Establish whether top-down and bottom-up selection modulate MS statistics in a comparable manner, as predicted by "saliency map" theories of visual attention [Koch & Ullman, 1985; Itti & Koch, 2000].(iv) Test the hypothesis that visual performance "outside the focus of attention" [Braun, 1998; Braun & Julesz, 1998; Lee et al., 1999; Braun et al., 2001] is associated with saliency-based modulations of MS statistics(v) Clarify the much disputed relation between attention and discrimination performance ('awareness') [Lamme et al., 2003; Koch & Tsuchiya, 2007; Dehaene et al., 2006].(vi) Model 'selection map' activity in a manner consistent with data from non-human primates (data from Dr. Hafed, Tübingen). Predict MS statistics from stochastic fluctuations in the model activity.