Investigating topographic signatures of global object perception and multifocal attention in human visual cortex
To organize and filter visual information across visual space, the visual system draws upon receptive fields, spatial attention, and perceptual grouping.
Receptive fields refer to the portion of visual space altering a neuron’s response when visually stimulated. They represent the building blocks of topographic maps in visual cortex, where receptive fields of neighboring neurons
correspond to neighboring portions in the retina and visual space. This retinotopy preserves the spatial arrangement of a visual scene. However, receptive fields are not hard-wired spatial filters. Specifically, research on
perceptual grouping and spatial attention showed that receptive field properties depend on the observer’s perceptual-cognitive state. Spatial attention
allows us to prioritize information processing at particular locations in visual
space, and evidence suggests receptive fields shift towards attended locations.
Perceptual grouping allows us to bind together image elements into global
objects and segregate them from the background, and evidence suggests
neurons increase their response when their receptive fields contain elements
perceived as a figure compared to a background. Across several projects, this
thesis investigated this state-dependency at the level of topographic maps in
human visual cortex. To this end, an approach to back-project brain measures into visual space was developed, allowing for fine-grained read-outs
of topographic signatures. The first project focused on topographic signatures of global object perception and points to a non-generic involvement
of higher object-sensitive cortex in perceptual grouping and a suppressionenhancement mechanism, possibly mediating figure-ground perception. The
second project investigated topographic signatures of multifocal attention
and shows that changes in topographic signatures under uni- or multifocal
attention conditions cannot be distinguished from changes for test-retest data. The third project investigated flaws in quantifying topographic signatures and highlights that prior summaries of topographic signatures might be
contaminated by regression artifacts. Collectively, these results underscore
the potentials and challenges of investigating state-dependent topographic
signatures.
https://discovery.ucl.ac.uk/id/eprint/10148275/1/Stoll_Thesis.pdf