2017年第二届曲阜视觉科学会议知识库

PURPOSE: A core function for any visual system (biological and artificial) is recognising objects by grouping and filling-in image elements as a figure and segregating them from the background. Neural mechanisms of Figure–Ground (F–G) segmentation have been linked to processes operating across several cortical areas including primary visual cortex (V1), and depend heavily on re-entrant feedback connections. F–G is thought to be a cortical phenomenon, however we recently observed strong F–G modulation in the earliest visual area in the brain, the dorsal lateral geniculate nucleus of the thalamus (LGN). Our previous study assumed that F–G segmentation observed in the LGN was mostly stimulus, and not task driven. By adding a second figure target, we wished to test whether this was the case.
METHODS: We performed tetrode recordings of spiking and local field potentials (LFP) in behaving macaque dLGN. We used speeded reaction–time where the subject had to find, and then saccade to either of two moving-dot defined visual figural objects against a moving background. The choice depended on a simple distance metric.
RESULTS: When we divide attentional demand across two figural objects, we find that LGN cells are differentially driven by both the visual component and the saccaded to location of the task. LFP signal analysis suggests this task activity is driven predominantly within the α–band.
CONCLUSIONS: We find that relay cells in the LGN respond not only to the visual cues alone, but also exhibit differential activity whether the figural object was going to be chosen for a subsequent saccade or not. The α–band dominated LFP suggests feedback as the most parsimonious route of this task driven modulation. Future experiments aim to orthogonally control spatial and feature attention while causally manipulating the feedback using optogenetic approaches.