Project summary: A family of deep convolutional neural networks (DCNNs) currently best explains primate ventral visual stream activity that supports object recognition. Such models are often evaluated with neurobehavioral datasets where the stimuli are presented in the subjects’ central field of view (FOV). However, the exact visual angle, type of images and level of contextual information often varies widely across studies. A unified model of the primate visual system cannot have these fluctuations.

As part of my work in the Kreiman lab, I am testing how the predictivity of macaque monkey inferior temporal (IT) neurons by DCNNs depends on the FOV and the image-context during passive fixation. I am also analysing the differences in context processing across humans, monkeys and ANNs during object recognition tasks. In parallel, I am working on improving this neural predictivity by developing more brain-aligned artificial neural networks that are trained on neural data.