Abstract
In general, object recognition is invariant to changes in the size of the retinal projection of an object; we recognize an object as a dog whether viewed from three feet or thirty feet. However, the size of the retinal projection is sometimes behaviorally relevant. For example, an increasingly larger projection of an unfamiliar dog may indicate that you should run away. Here, we used fMRI to investigate whether the response to object size in visual cortex could be modulated by task demands. Six participants were scanned in a 3T GE scanner as they viewed line-drawings of animals and vehicles while fixating. Images were presented in a rapid event-related design in four different sizes:6, 12, 18, and 24 degrees of visual angle. Object trials were intermixed with fixation-only trials and scrambled-images that served as baseline conditions. In separate runs, subjects performed two tasks on the same stimuli. During the size-task, subjects indicated whether the physical size of the image on the screen was small, medium, or large. During the object-task, subjects categorized each image as an animal, vehicle, or not an object. Overall, sensitivity to image size decreased from V1 to LO (i.e., representations became more size-invariant). Importantly, size-sensitivity changed across visual cortex as a function of task: both V1 and LO were more sensitive to image size during the size-task compared to the object-task. In V1, sensitivity to image size was evident during both object and size-tasks. However, sensitivity to size increased in V1 during the size-task through suppression of responses to non-preferred sizes. In contrast, object representations in LO displayed sensitivity to image size only during the size-task, and were invariant to size during the object-task. These data provide new insights into the dynamic nature of visual representations involved in object perception.