Spatial Frequency Tolerant Visual Object Representations in the Human Ventral and Dorsal Visual Processing Pathways

Abstract

Primate ventral and dorsal visual pathways both contain visual object representations. Dorsal regions receive direct input from magnocellular layers of LGN while ventral regions receive directly input from both magnocellular and parvocellular layers of LGN. Because magnocellular layers prefer low SF and parvocellular layers prefer mid- to high-SF components of the image, object representations in ventral and dorsal regions may differ in how they represent visual input from different spatial scales. To test this prediction, we asked observers to view blocks of images from six object categories shown in full spectrum, high SF, or low SF. We found robust object category decoding in all SF conditions as well as SF decoding in nearly all the early visual, ventral, and dorsal regions examined. Cross-SF decoding further revealed that object category representations in all regions exhibited substantial tolerance across the SF components. No difference between ventral and dorsal regions was found in their preference for the different SF components. Further comparisons revealed that, whereas differences in the SF component separated object category representations in early visual areas, such a separation was much smaller in downstream ventral and dorsal regions. In those regions, variations among the object categories played a more significant role in shaping the visual representational structures. Our findings show that ventral and dorsal regions are similar in how they represent visual input from different spatial scales and argue against a dissociation of these regions based on differential sensitivity to different SFs.

Roles of Category, Shape, and Spatial Frequency in Shaping Animal and Tool Selectivity in the Occipitotemporal Cortex

“…While the univariate effects for Shape and Spatial Frequency were not as reliable and consistent as the category effects in the ROIs, classification analyses (SVM) revealed that the neural response patterns were distinguishable not only between categories, but also between shapes and between SFs in most ROIs. These results are consistent with previous findings that both visual and category differences contribute to the neural responses in the occipitotemporal cortex (shape: Op de Beeck et al, 2008 ; spatial frequency: Vaziri-Pashkam et al, 2019 ; category: Haxby et al, 2001 ; Cox and Savoy, 2003 ; Pietrini et al, 2004 ; Kriegeskorte et al, 2008 ), and that the contributions of visual and category information may be independent ( Bracci and Op de Beeck, 2016 ; Proklova et al, 2016 ). Extending from these findings, the current study directly compared the relative contributions of these factors to the representations in the animal- and tool-selective ROIs.…”

Express detection and discrimination of visual objects by primate superior colliculus neurons

“…Because spatial frequency is relevant for visual object recognition [33][34][35][36][37], and because primate SC neurons exhibit spatial frequency tuning [31,32], we next asked how object detection performance as in Figs. 1, 2A, 2B, 2D, 2E depended on spatial frequency.…”

An integrative computational architecture for object-driven cortex

“…In the recognition models discussed here, physical properties of objects are fed to the physics engine for integration of information across time (e.g., updating beliefs about an object's mass), future prediction, and reasoning (e.g., computing the force to apply to keep a tower stable). In line with this computational pipeline, recent brain imaging work suggests that abstract object information such as shape is "uploaded" from ventral pathway to regions in the parietal cortex [8,9,10] where it may adaptively support aspects of cognition and action [55]. If, in addition to shape, visually computed representations of objects' dynamic physical properties such as their mass are uploaded from ventral stream to an intuitive physics engine in parietal and premotor cortex, then we should expect to see representations of these properties in those regions.…”

“…1D, E). These regions include the traditional objectselective occipitotemporal regions (e.g., [7]), such as the lateral occipital cortex (LOC) and posterior fusiform (pFus), as well as regions in the intraparietal sulcus [4,8,9,10] and frontal cortex that show large overlaps with networks implicated in tool use and action planning [4]. Presumably, these different regions process dynamic objects in different ways and for different functional purposes [11].…”