| Summary: | The visual system is sensitive to differences in features, like orientation or luminance, over space. Local differences in features allow us to detect boundaries in the environment â a critical step for segmenting and detecting meaningful figures or objects. However, additional feature-tuned mechanisms can yield enhancement of the entire figure region or of salient regions in the visual field, which appear subjectively to âgrabâ our attention. Here, we explored these feature-tuned contextual effects in the human early visual system using high-resolution functional magnetic resonance imaging (fMRI) at 7Tesla. We asked how visual salience and figure-ground processing are instantiated in visual cortex and the lateral geniculate nucleus (LGN) of the thalamus; we also evaluated the role of directed attention in these perceptual mechanisms. We found that orientation-defined figures yield enhanced responses in both early visual cortex and in the LGN, and that this enhancement does not depend on the locus of directed attention. We further demonstrated that in the LGN, figure-ground modulation is likely driven by feedback from cortical regions. In V1, we used a novel voxel-wise analysis to show separable effects of local boundary detection and widespread figure enhancement in the absence of attention, arguing further for the automaticity of these effects in the early visual system. We also asked how differences in features across a wider portion of the visual field yield a computation of salience. We found that early cortical areas are sensitive to feature-defined salience while the LGN is not. While attention modulated responses in all visual areas, its effects did not interact with representations of salience, suggesting that these two mechanisms provide independent sources of information in the early stages of visual processing. In all, these results inform our understanding of the role of the LGN in vision, demonstrating its sensitivity to complex orientation-tuned perceptual computations, which likely arise via corticothalamic feedback. More broadly, this work clarifies and characterizes mechanisms of automatic, stimulus-driven contextual effects in the human early visual system, which segment and enhance meaningful information in the visual environment.
|
|---|
