Understanding visual object representations in human occipito-temporal and posterior parietal cortices, convolutional networks

In recent studies we show that non-spatial visual information, including the content of visual working memory, may be directly represented in the human posterior parietal cortex (PPC). In this talk, I will compare visual object representations in the human PPC with those in the human occipito-temporal cortex and describe how together they provide us with both a stable representation of our visual environment and allow us to interact with the external world flexibly and efficiently. Using methods developed in neuroscience, I will also examine the nature of visual representations in convolutional neural networks trained to perform object recognition.