Research

A crucial question in vision science is how do we recognize objects? It seems trivial, easy, and self-evident, but taking a cursory look at today's artificial intelligence research (or at face detection software in cameras) is enough to suggest how incredibly complex the process is. I have been working on uncovering the mechanisms underlying object recognition, using visual crowding as a tool. Currently, I am working on determining the neural processes underlying attention and visual short term memory (VSTM) using a combination of psychophysics, EEG and TMS.

Object recognition

Object recognition takes place in two steps. First, the features (orientation, color, etc) of an object are detected independently. Then these features are put together to form the representation of that object. Crowding is a breakdown of the second step. When a target object is flanked closely by other objects, the features of the target and the flankers get mixed up leading to a jumbled percept. This is crowding. It offers a direct window in how feature integration occurs, and hence serves as a handy tool in investigating object recognition.

Crowding is illustrated below. Fixate (keep your eyes focused on) the black square in the center. If you try to read the central letter of the triplet on the right, you will find it very hard. The same letter at the same distance on the left is extremely easy to identify. It is the presence of the the flanking letters on the right that makes the target letter unidentifiable.

Attention

Visual attention directs the limited resources of the visual system to the currently relevant input. I study an important aspect of this allocation: how quickly does it move? We find that voluntary attention is much slower (250 ms) than involuntary attention (100 ms). Also, voluntary attention takes longer to go to a location that is much further away, but involuntary attention is largely independent of distance. Further, the attentional resources of the two cortical hemispheres can allocated, to some extent, in parallel.

Currently, I am investigating the neural events that enable this attentional deployment. We find that the state of the brain just before a stimulus is presented determines how quickly attention is deployed. Our data suggests that attention is deployed rhythmically - it samples the visual input in a periodic fashion, like a movie camera - rather than in a sustained manner.

Current research

Neural oscillations and attention and VSTM
Neural representation of object location
Subitization