Research
Cracking the Code
We recognize objects easily every day, but object recognition is in fact a very difficult problem. Even leading computer algorithms do not match human performance today. Object recognition is not easy for the brain either: a series of cortical areas, taking up ~40% of the brain, is dedicated to vision. But we know very little about the code in which the brain represents objects for perception, and about how the brain transforms what we see into what we perceive. How do we crack the code for objects? What are its features and what are its rules?
Approach
Our approach to this problem is best understood through an analogy to colour. We see millions of colors but it is well known that color perception is three-dimensional - any color we perceive can be represented using three numbers. Can we do likewise for the millions of shapes we see? Do shapes also reside in a low-dimensional space? We use a wide variety of experimental techniques in our lab to address fundamental questions about high-level vision. These techniques include:
Behavioral experiments in human participants. Here, we make participants to perform simple tasks such as visual search or categorization to understand systematic patterns in their performance.
Brain imaging (fMRI) in human participants. Here, we make participants to perform tasks inside an MRI scanner to characterize the underlying neural basis.
Extracellular recordings from single neurons in monkeys while they perform visual tasks. Here, we record brain activity from the visual cortex of monkeys while they perform complex tasks, to understand the underlying representations at the level of single neurons.
Comparing object representations in biological vision and machine vision algorithms. Here, we compare state-of-the-art computer vision algorithms with object representations in biological vision (in behavior and neurons). The goal here is to establish a two-way dialogue: to understand biological vision using computational experiments on machine vision, and conversely, to improve machine vision using insights from biological vision.
Do you want to participate in our experiments?
We regularly require human volunteers to perform behavioral experiments on object recognition. The typical experiment takes an hour and involves you looking at images on the screen and making responses using a key press. If you are based in/near IISc and are interested in being a subject in our experiments, please get in touch with us.
Do you have difficulty recognizing objects?
Disorders of object recognition are of interest to us because they help in understanding how normal object recognition occurs. An excellent example is prosopagnosia or face blindness - people with this disorder can recognize all other objects except faces. If your vision is otherwise normal but you have difficulty perceiving or recognizing objects - and you would like to advance scientific knowledge, please contact me and we can talk further. However please note that we are not doctors and therefore may not necessarily be able to help you.