Explaining the heterogeneity and topography in inferior temporal cortex with deep neural networks

Humans recognise objects almost immediately – a brief glance is sufficient to recognise the face of a friend. Our everyday actions and interactions depend on object recognition which depends on a complex neural architecture in the visual cortex. The inferior temporal (IT) cortex is a particularly important part of the brain for object recognition. IT contains regions that respond preferentially to faces, colour and places. Despite a recent period of extensive study, we still do not understand how the selectivity of these regions emerges and how they are organised.

I propose to use computational models called deep neural networks to understand why the IT cortex has the structure it does. I will test whether similar structures emerge in our models by comparing brain activity measured in monkeys and humans.

This research will reveal the basic principles of object recognition, which are essential for human cognition and everyday life.