Abstract
Object recognition unfolds through computations within the dorsal and ventral visual streams, which are differentially recruited depending on object semantic identity. If an object is typically manipulated by the hands (e.g., a mug) then there is a greater degree of dorsal stream processing than is recruited by an object which is not typically manipulated by the hands (e.g., a potted plant). We hypothesized that the differential recruitment of the two visual streams according to object semantic identity would have perceptual ramifications, predicting that non-manipulable objects would have a benefit in spatial resolution due to the higher proportion of parvocellular input to the ventral stream and that manipulable objects would have a benefit in temporal resolution due to the higher proportion of magnocellular input to the dorsal stream. Using a gap detection task, we found that non-manipulable objects are processed with higher spatial resolution than manipulable objects, while using a temporal discrimination task we found evidence for higher temporal resolution for manipulable objects. Additionally, it was observed that the spatial resolution difference between manipulable and non-manipulable objects was absent when the semantic identity of the objects was obscured through inverting the objects and we found that the spatial resolution difference was modulated by ambient red light, which is known to suppress magnocellular processing. These results speak to the nature of object representations, grounding them in neural circuits and closely associating them to the nature of the perceptual inputs.