Skip to main content
. 2016 Oct 21;6:35723. doi: 10.1038/srep35723

Figure 2.

Figure 2

(a) Amplitude spectrum of target images: For comparability with previous studies, we used four greyscale natural images in Experiment 1. To select an appropriate masker, these images were first analysed with a two-dimensional FFT. The results reveal a 1/f spatial profile for each of the target images, thus maskers were given a 1/f amplitude spectrum. (b) Spatiotemporal filtering: A three-dimensional FFT was first computed for a stack of 205 randomly generated noise images. Filtering the z-axis with a log-Gaussian filter controlled masker temporal frequency, whereas individual noise images (x and y axes) were convolved with a circularly symmetric inverse frequency filter for spatial frequency. The resultant stimulus was temporally narrowband, continuously modulating pink noise. (c) Ramped onsets: To avoid abrupt transients, both masker and targets were ramped up to their maximum contrast over a period of 1000 ms and the masker preceded the target by 50 ms to allow cumulative suppressive effects.