Accumulated attentional landscape over a scene. The scene was viewed by 12 observers. (a) Position-based time course analysis, (b) positions weighted by fixation durations. To create these landscapes, a two-dimensional Gaussian distribution with a standard deviation of 1 degree of visual angle (equivalent to the fovea) was centered at each fixation point. For the duration-weighted landscapes, the maximum of each Gaussian was set to the duration of that fixation in milliseconds. The fixation-based Gaussians were summed up, incrementally for each time window (100 ms). For the lower panels, the current landscape was normalized (scaled between 0 and 1). The image was modulated such that it had full contrast within the attended regions (the fixation peaks) and much reduced contrast outside these regions. 

(inspired by Pomplun et al., 1996, and Henderson, 2003)