Our research goal is to understand how internal representations of incoming stimuli are updated as we acquire new information: How do individual neurons and networks in the visual cortex encode image features that vary dynamically between eye movements, and how does this encoding relate to visual behavior?
To achieve this goal, we employ techniques that allow us to record simultaneously the activity of multiple neurons in the visual cortex of alert monkeys during specific behavioral tasks, in combination with human psychophysics and computational models of neural network function.
Specific questions of interest in our laboratory are:
How do rapid changes in the structure of images caused by eye movements influence how images are coded in the visual cortex?
What is the relationship between feature encoding in early cortical areas (e.g., primary visual cortex), object encoding in higher cortical areas (e.g., inferotemporal cortex), and object perception?
How do changes in the internal state of cortical networks (e.g., induced by behavioral or perceptual context) influence neural coding in visual cortex and the corresponding behavioral performance?
We believe that our research on the neural coding of dynamic image representations has the potential to advance our understanding of the neuronal mechanisms underlying visual perception and learning, and, at the same time, help develop chronically-implantable human cortical prostheses to assist visually impaired people.
Example of an eye scan path (white line) and fixation location centers (circles) during one free viewing episode in one monkey. During a typical free viewing session animals viewed a sequence of 10 natural images, each presented for 10 s, separated by intervals of 20s. Eye position was continuously monitored using an infrared eye tracking system.
Psychophysical studies in monkeys and humans are used to test the relationship between perceptual performance and the cortical processing of visual information.
Actions potentials recorded during three consecutive multiple-electrode recording sessions from the primary visual cortex (V1) of a fixating monkey. Stimuli consisted of movie strips in which each frame was a grating of random orientation and spatial phased, flashed at 60 Hz in the center of the receptive field.
Orientation preference map obtained by intrinsic signal imaging in V1 of anesthetized cat (top). The angle of preferred orientation of each pixel is shown according to the color key. Circles indicate the electrode penetration locations (top). Vascular pattern of the cortical surface from the region shown above. Blood vessels are used to guide electrode penetrations aimed at specific cortical locations (bottom).
