Current Research Projects

Research in my laboratory seeks to elucidate the computational principles and neural mechanisms underlying visual perception. We use computational, mathematical and neurophysiological experimental techniques to address the following fundamental issues in the fields of computational and biological vision.

Statistcal and ecological approaches to higher order neural codes.

The study of neural representation should start with a rigorous study of the statistical regularities in the visual environment. I have adopted the Gibsonian ecological approach in our quest to understand how the brain represents and computes 2D curves and 3D surfaces. We developed databases of 2D and 3D natural scenes and applied statistical and machine learning techniques to discover the statistical structures in the data. This has allowed us to predict plausible neural representations based on the principle of efficient coding. We are currently carrying out neurophysiological experiments to evaluate the various possible candidates for the neural representations of mid-level vision (2D shapes and 3D surfaces). We have developed methods to discover how neurons encode information and to decode visual stimuli based on neural responses. We are experimenting with the implantation of microelectrode arrays that can record from hundreds of neurons simultaneously. Our long term goal is to decode and reconstruct computationally the mental images that are represented in our visual cortex.

Learning, adaptation and development in neural systems.

Learning and adaptation are what make biological systems so much more robust and powerful than current man-made vision systems. My current research explores the principles underlying adaptation and learning in the visual system at different time scales and at the level of neurons and of neural systems. We have studied theoretically how neurons adapt dynamically to the statistical context of the visual stimuli. We have determined biophysical features in spiking neurons that make them adapt, and have isolated the statistical features in natural stimuli that drive neuronal adaptation. In our neurophysiological studies, we have demonstrated that the neural machinery of perceptual processing is very flexible and subject to modification by behavioral experience. We are currently undertaking computational and neurophysiological studies on the time course and dynamical processes of neural plasticity and visual development. We hope these studies will provide insights to the design principles underlying the adaptation of adult visual systems and the development of infant vision.

Principles and algorithms of hierarchical perceptual inference.

Perceptual inference is an active and creative process that constructs an internal interpretation of the outside world in our mind. Bottom-up information from the retina is only a clue that starts the inference process, which is affected by various global scene contextual factors and perceptual experience. We have carried out a series of neurophysiological experiments to demonstrate the influence of a variety of contextual factors in shaping visual processing in the early visual areas. We have demonstrated experimentally that visual processing likely involves the entire hierarchical circuit interactively. We are developing a computational framework for perceptual inference based on hierarchical Bayesian inference to elucidate the rules of recurrent feedback in cortical circuits, and to understand the computational algorithms underlying the inference of shapes and surfaces of visual objects in our mind.

Our research is currently supported by two grants from the National Science Foundation, a center grant from NIMH and a center grant from NIH, two NSF graduate student fellowships and a NRSA postdoctoral fellowsip from NEI. For further details of research in my laboratory, please visit the Active Perception Laboratory web page.