This thesis aimed to probe the functional specializations present within several retinotopic divisions of human lateral occipital cortex (LO). The divisions of interest were LO1 and LO2, two neighbouring visual field maps that are found within object-selective LO; the posterior portion of a larger area referred to as the lateral occipital complex (LOC), and V5/MT, the well-known visual complex that is highly selective to visual motion. In order to seek out the causal roles played by these divisions in human visual perception, I used transcranial magnetic stimulation to temporarily disrupt neural processing within these areas, while observers performed visual tasks. The visual tasks I employed examined both spatial vision, through orientation and shape discriminations, and motion processing, through speed discrimination.
The data revealed a number of double dissociations. A double dissociation was present between LO1 and V5/MT in the perceptions of orientation and speed. A similar pattern of results was present during orientation and speed discrimination of the same moving stimuli, although this effect was markedly weaker. Additionally, a double dissociation was present between LO1 and LO2 in the perceptions of static orientation and shape, respectively. These double dissociations suggest that LO1, LO2 and V5/MT exhibit functional specializations for orientation, shape and speed, respectively and moreover, perform these specialized roles largely independently of one another.
It is unsurprising that I found evidence for parallel processing of motion and aspects of spatial processing because: (1) V5/MT has been shown to be a cluster of multiple visual field maps with a common foveal representation – a feature that has led to the idea that the maps within clusters perform related aspects of processing, but are independent of the processing undertaken in adjacent visual field map clusters like LO; (2) neuropsychological evidence, from studies of akinetopsia and visual form agnosia, points to a double dissociation in processing of motion and form and (3) there is evidence of parallel processing pathways from early visual areas and even subcortical structures to V5/MT.
The parallel processing of orientation and shape in LO1 and LO2 is a novel and more surprising finding for the following reasons: (1) These visual field maps are adjacent maps within a single cluster and therefore, might be expected to perform a series of related and dependent roles and (2) shape, as defined here by curvature, could be seen as a property that is dependent on orientation processing. These findings therefore, point to an architecture whereby the extrastriate visual maps in LO sample visual information from antecedent visual areas in parallel, to extract higher order spatial statistics. Mutual retinotopic information and parallel processing not only reduces replicated information across maps but also, provides a common mechanism for communication between maps which exhibit different specializations.
Importantly, the well-known category-selectivity of extrastriate regions, like LO, may simply emerge from patterns of unique and low-level visual computations, which encode category specific image statistics, performed by the individual visual field maps that subdivide these areas.
