Purchase this article with an account.
Sophia Robert, Leslie Ungerleider, Maryam Vaziri-Pashkam; Motion-defined object category responses in the human brain. Journal of Vision 2020;20(11):1247. doi: https://doi.org/10.1167/jov.20.11.1247.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Humans can label and categorize objects in a visual scene with high accuracy and speed—a capacity well-characterized with neuroimaging studies using static images. However, in a dynamically moving visual scene, motion is another cue that could be used by the visual system to classify objects. To determine how motion-defined object category information is processed in the brain, we created a novel stimulus set to isolate motion-defined signals from other sources of information. We used videos of 6 object categories, extracted their movement information, and applied the extracted motion to random dot patterns to create motion-defined stimuli. We first ran a behavioral categorization task with these stimuli and found that humans can accurately categorize the objects based on just their motion, without access to luminance-defined form. We then investigated whether fMRI responses elicited by motion cues could be decoded at the object category level in functionally defined regions of occipitotemporal and parietal cortex. In a block design, participants performed a one-back repetition detection task as they viewed three second presentations of either motion-defined stimuli or static images cropped from random frames of the original videos. Using linear classifiers, we were able to significantly decode object category in both stimulus formats in all functionally-defined regions, including lateral occipital cortex, extrastriate body area, posterior fusiform sulcus (pFS), and inferior intraparietal sulcus. Classification accuracies were similar across the two conditions in all regions except pFS, which had significantly higher classification accuracy in the image condition. Significantly above chance classification accuracies in all regions were also observed when training the SVM classifier on data from the motion-defined condition and testing it on the image condition and vice versa. These results demonstrate that motion-defined cues can elicit robust category responses on par with those elicited by luminance cues in regions of object-selective visual cortex.
This PDF is available to Subscribers Only