Date of Award
Doctor of Philosophy (PhD)
Neuroscientists have traditionally conceived the visual system as having a ventral stream of vision for perception and a dorsal one associated with vision for action. However functional differences between them have become relatively blurred in recent years, not the least by the systematic parallel mapping of functions allowed by functional magnetic resonance imaging (fMRI). Here, using fMRI to simultaneously monitor several brain regions, we first studied a hallmark ventral stream computation: the processing of faces. We did so by probing responses to motion, an attribute whose processing is typically associated with the dorsal stream. In humans, it is known that face-selective regions in the superior temporal sulcus (STS) show enhanced responses to facial motion that are absent in the rest of the face-processing system. In macaques, face areas also exist, but their functional specializations for facial motion are unknown. We showed static and moving face and non-face objects to macaques and humans in an fMRI experiment in order to isolate potential functional specializations in the ventral stream face-processing system and to motivate putative homologies across species. Our results revealed all macaque face areas showed enhanced responses to moving faces. There was a difference between more dorsal face areas in the fundus of the STS, which are embedded in motion responsive cortex and ventral ones, where enhanced responses to motion interacted with object category and could not be explained by their proximity to motion responsive cortex. In humans watching the same stimuli, only the STS face area showed an enhancement for motion. These results suggest specializations for motion exist in the macaque face-processing network but they do not lend themselves to a direct equalization between human and macaque face areas. We then proceeded to compare ventral and dorsal stream functions in terms of their code for spatial attention, whose control was typically associated with the dorsal stream and prefrontal areas. We took advantage of recent fMRI studies that provide a systematic map of cortical areas modulated by spatial attention and suggest PITd, a ventral stream area in the temporal lobe, can support endogenous attention control. Covert attention and stimulus selection by saccades are represented in the same maps of visual space in attention control areas. Difficulties interpreting this multiplicity of functions led to the proposal that they encode priority maps, where multiple sources are summed to form a single priority signal, agnostic as to its eventual use by downstream areas. Using a paradigm that dissociates covert attention and response selection, we test this hypothesis with fMRI-guided electrophysiology in two cortical areas: parietal area LIP, where the priority map was first proposed to apply, and temporal area PITd. Our results indicate LIP sums disparate signals, but as a consequence independent channels of spatial information exist for attention and response planning. PITd represents relevant locations and, rather than summing signals, contains a single map for covert attention. Our findings have the potential to resolve a longstanding controversy about the nature of spatial signals in LIP and establish PITd as a robust map for covert attention in the ventral stream. Together, our results suggest that while the distribution of labor between ventral stream and dorsal stream areas is less linear than what a what a rough depiction of them can suggest, it is illuminated by their proposed function as supporting vision for perception and vision for action respectively.
Polosecki, Pablo, "Specialized Signals for Spatial Attention in the Ventral and Dorsal Visual Streams" (2015). Student Theses and Dissertations. 409.