Characterizing and decoding visual percepts of real objects in a blind individual using fMRI

The brain can generate vivid and concrete visual perceptions in the absence of retinal input (e.g. hallucinations and dreams) which are associated with visual cortex activation. However, these percepts are, by definition, disconnected from reality. Can the visual cortex support accurate visual representations of one's external environment using only non-visual sensory modalities? In other words, can the brain see without the eyes? We explore this question through a case study of patient NS, a woman who lost her sight to retinal degeneration and now “sees” the objects she infers are around her through touch, proprioception, and sound. Unlike imagery, these representations are determinate, involuntary, and persist as long as she infers the object to be within her line-of-sight. We triggered NS's non-optic visual perceptions while recording fMRI BOLD activity by having NS touch and place 3D objects on a plexiglass tray that held them suspended in her field-of-view. A GLM analysis found significant patterns of activation in her visual cortex that resembled the patterns of activation in a sighted control who viewed the same objects through typical retinal vision. This was in contrast to the very little significant activity found when both participants imagined seeing the same objects, substantiating NS’s reports that her non-optic sight is phenomenologically more similar to the retinal vision that she lost rather than mental imagery. Furthermore, activity patterns within the visual cortex could be used to accurately decode the objects that NS was “seeing” but no longer touching, and changes in object size resulted in a corresponding change in the topological extent of the increased brain activity. This demonstrates that the activity associated with non-optic sight is both specific to and representative of the individual stimuli. These findings suggest that the visual cortex can support concrete visual experiences that accurately interpret non-retinal sensory input.