This repo contains code for the NMA 2022 fMRI project by the " The Nomads" group from Injera pod.
Warning You can now open the notebook by opening it in Github (click above) and then use the "Open in Colab" link.
Animacy is a mid-level visual feature processed by the visual cortex and is primarily described as the difference between living and non-living objects. Despite the extensive research on animacy and other mid-level features, it is yet unclear how these features are encoded in the brain and which regions are likely to be selective to animacy-related visual stimuli. Recent studies suggest that animacy is most likely represented in the mid-level visual cortex, containing late V2, V3, and V4. To validate this hypothesis, we used the Kay dataset of natural images with their fMRI responses to measure the neural responses to animate and inanimate images in each of those regions. Animate and inanimate categories were extracted and manually relabeled from the original dataset. We then used a Support Vector Machine (SVM) to calculate the discrimination accuracy between animate and inanimate in each region. The regions with higher discrimination accuracy are expected to be more involved in the animacy recognition task since they provide better linear discrimination between animate and inanimate objects.
Kay natural images dataset
This group project was completed during Neuromatch 2022 (July 11-29). See project-summary.pdf for details!
Amira Fadl, Yazan Ziad, Yasmine Guenni, Ziad A. Gamil, Esraa R. Moawad