The representation of the visual field in visual areas of the dorsolateral, lateral, and ventral cortices was studied by means of extracellular recordings and fluorescent tracer injections in anaesthetised marmoset monkeys. Two areas, forming mirror-symmetrical representations of the contralateral visual field, were found rostral to the second visual area (V2). These were termed the ventrolateral posterior (VLP) and the ventrolateral anterior (VLA) areas. In both areas, the representation of the lower quadrant is located dorsally, between the foveal representation of V2 and the middle temporal crescent (MTc), whereas the representation of the upper quadrant is located ventrally, in the supratentorial cortex. A representation of the vertical meridian forms the common border of areas VLP and VLA, whereas the horizontal meridian is represented both at the caudal border of area VLP (with V2) and at the rostral border of area VLA (with multiple extrastriate areas). The foveal representations of areas VLP and VLA are continuous with that of V2, being located at the lateral edge of the hemisphere. The topographic and laminar patterns of projections from dorsolateral and ventral cortices to the primary (V1) and dorsomedial (DM) visual areas both support the present definition of the borders of areas VLP and VLA. These results argue against a separation between dorsolateral and ventral extrastriate areas and provide clues for the likely homologies between extrastriate areas of different species.

On the basis of extracellular recordings in marmoset monkeys, we report on the organisation of the middle temporal area (MT) and the surrounding middle temporal crescent (MTc). Area MT is approximately 5-mm long and 2-mm wide, whereas the MTc forms a crescent-shaped band of cortex 1-mm wide. Neurones in area MT form a first-order representation of the contralateral hemifield, whereas those in the MTc form a second-order representation with a field discontinuity near the horizontal meridian. The representation of the vertical meridian forms the border between area MT and the MTc. In both areas, the fovea is represented ventrocaudally, and the visual field periphery is represented dorsorostrally. Analysis of single units revealed that 86% of cells in area MT show a strong selectivity for the direction of motion of visual stimuli. The proportion of direction-selective cells in the MTc (53%), whereas lower than that in area MT, is much higher than that observed in most other visual areas. Neurones in the cortex immediately rostral to area MT and the MTc are direction selective, with receptive fields predominantly located in the visual field periphery. In contrast, only a minority of the cells in the cortex ventral to the MTc are direction selective, and their receptive fields emphasise central vision. The results suggest that the MTc is functionally closely related to area MT, and distinct from the areas forming the dorsolateral complex. The MTc may have a role in combining information about motion in the visual field, processed by area MT, with information about stimulus shape, processed by the dorsolateral complex.

The organisation of the second visual area (V2) in marmoset monkeys was studied by means of extracellular recordings of responses to visual stimulation and examination of myelin- and cytochrome oxidase-stained sections. Area V2 forms a continuous cortical belt of variable width (1-2 mm adjacent to the foveal representation of V1, and 3-3.5 mm near the midline and on the tentorial surface) bordering V1 on the lateral, dorsal, medial, and tentorial surfaces of the occipital lobe. The total surface area of V2 is approximately 100 mm2, or about 50% of the surface area of V1 in the same individuals. In each hemisphere, the receptive fields of V2 neurones cover the entire contralateral visual hemifield, forming an ordered visuotopic representation. As in other simians, the dorsal and ventral halves of V2 represent the lower and upper contralateral quadrants, respectively, with little invasion of the ipsilateral hemifield. The representation of the vertical meridian forms the caudal border of V2, with V1, whereas a field discontinuity approximately coincident with the horizontal meridian forms the rostral border of V2, with other visually responsive areas. The bridge of cortex connecting dorsal and ventral V2 contains neurones with receptive fields centred within 1 degree of the centre of the fovea. The visuotopy, size, shape and location of V2 show little variation among individuals. Analysis of cortical magnification factor (CMF) revealed that the V2 map of the visual field is highly anisotropic: for any given eccentricity, the CMF is approximately twice as large in the dimension parallel to the V1/V2 border as it is perpendicular to this border. Moreover, comparison of V2 and V1 in the same individuals demonstrated that the representation of the central visual field is emphasised in V2, relative to V1. Approximately half of the surface area of V2 is dedicated to the representation of the central 5 degrees of the visual field. Calculations based on the CMF, receptive field scatter, and receptive field size revealed that the point-image size measured parallel to the V1/V2 border (2-3 mm) equals the width of a full cycle of cytochrome oxidase stripes in V2, suggesting a close correspondence between physiological and anatomical estimates of the dimensions of modular components in this area.

No abstract

The marmoset monkey has become an important primate model in Neuroscience. Here we characterize salient statistical properties of inter-areal connections of the marmoset cerebral cortex, using data from retrograde tracer injections. We found that the connectivity weights are highly heterogeneous, spanning five orders of magnitude, and are log-normally distributed. The corticocortical network is dense, heterogeneous and has high specificity. The reciprocal connections are the most prominent and the probability of connection between two areas decays with their functional dissimilarity. The laminar dependence of connections defines a hierarchical network correlated with microstructural properties of each area. The marmoset connectome reveals parallel streams associated with different sensory systems. Finally, the connectome is spatially embedded with a characteristic length that obeys a power law as a function of brain volume across species. These findings provide a connectomic basis for investigations of multiple interacting areas in a complex large-scale cortical system underlying cognitive processes.