In our laboratory, we study how we see. Specifically, we are interested in understanding visual perception as statistical inference.
The rationale behind studying vision as inference is this: The visual world is more complex, ambiguous and noisy than we generally realize. This means that the brain cannot afford to construct an internal replica of the external world we see. Instead, brain has to ‘make do’ with the available ambiguous information. Thus, what we think of seeing is really the brain’s best guess as to what is out there.
Recent studies show that in order to come up with this inference about the world, the brain must evaluate the ambiguous sensory information together with what it knows about the nature of the visual world and about the task at hand. The goal of my laboratory is to study how this decision-making process works.
Visual inference is not a special case scenario, but rather is part and parcel of visual perception under any circumstance. In other words, there is no visual perception without inference.
An obvious illustration of this is 3-D vision. The retinal image is a flat, 2-D image; it contains no unambiguous 3-D information whatsoever. This means that the third dimension, or depth, has to be
inferred from the cues to depth (such as stereoscopic disparity, motion, occlusion, etc) implicit in the image. Since each depth cue by itself is ambiguous, the visual system must combine ambiguous information from multiple depth cues to infer depth. Thus, there is no 3-D vision without inference. Given that there are many additional sources of ambiguity in the visual image, there is no vision, 3-D or otherwise, without inference.
Indeed, recent computational and psycho-physical studies indicate that studying vision as inference, especially as Bayesian inference, provides a far-reaching framework for understanding not only how we see but, in good measure, how the brain works.
OUR RESEARCH INTERESTS
Our research interests include, but are not limited to, the perception of 3-D visual objects and scenes. We are also interested in understanding the mechanisms of visual dysfunction and developing methods for treating visual impairments.
RESEARCH METHODOLOGIES WE USE
We use a variety of advanced research methods, including multi-electrode recording in awake, behaving monkeys, functional magnetic resonance imaging (fMRI) in humans and monkeys, psychophysics in humans and monkeys, and modeling. We are also engaged in translational research that seeks to devise new ways of rehabilitating patients with various neuro-ophthalmological impairments. Thus, our laboratory offers an excellent opportunity to learn a diverse set of advanced research methodologies while addressing many questions in the forefront of Systems Neuroscience.
Free Website Builder by TandemServer.org