Intermediate Fellowship research summary

We recognize objects every day, but object recognition is in fact a very difficult problem. Even leading computer algorithms today do not match human performance on tasks as simple as recognizing distorted letters (websites require us to do this task to deny access to automated programs).

How does the brain accomplish object recognition? We think the brain solves this problem using a sophisticated way of representing objects. Understanding this representation will help in treating disorders of vision in humans and in improving computer vision algorithms.

The goal of my research is to address major gaps in our understanding of object recognition. To this end, we will study object recognition performance in humans (using behavioral tests) and also record the electrical activity from single neurons in the monkey visual cortex. We will focus in particular on how objects are represented in the inferotemporal cortex (IT), an area critical for object recognition. This approach allows us to understand object recognition at the level of single neurons.