Superior colliculus neurons encode a visual saliency map during free viewing of natural dynamic video

• DOI: 10.1038/ncomms14263
• Publication Date: 2017-01-24T10:42:06Z
• Authors: Brian J. White; David J. Berg; Janis Y. Kan; Robert A. Marino; Laurent Itti; Douglas P. Munoz
• Keywords: Superior colliculus; Neocortex; ENCODE; Representation

Models of visual attention postulate the existence a saliency map whose function is to guide and gaze most conspicuous regions in scene. Although cortical representations have been reported, there mounting evidence for subcortical mechanism, which pre-dates evolution neocortex. Here, we conduct strong test hypothesis by comparing output well-established computational model with activation neurons primate superior colliculus (SC), midbrain structure associated gaze, while monkeys watched video natural scenes. We find that activity SC superficial visual-layer (SCs), specifically, well-predicted model. This representation unlikely be inherited from fronto-parietal cortices, do not project SCs, but may computed SCs relayed other areas via tectothalamic pathways.