A novel texture backward masking method to locate critical recurrent processes in human vision
This paper introduces a novel texture backward masking paradigm using deep network-synthesized masks to demonstrate that recurrent processes critical for challenging object recognition are already recruited in the early stages of the human visual system.