Gießen - New Haven
German - US-American Collaboration in Computational Neuroscience
Neuronal theory of three-dimensional form perception
Under normal viewing conditions, whenever we look at an object, we immediately perceive its three dimensional (3D) shape. We can tell which points on the object’s surface are closest to us and which are furthest away, and can judge the orientations and curvatures of its various parts. Although it feels effortless for us to do this, in fact, reconstructing the 3D shape of an object from its 2D projection in the retinal image is an extremely difficult computational problem, which cannot be solved by any existing computer algorithm. And yet the brain can solve it. This project aims to advance our understanding of how the brain reconstructs the 3D shape of objects from images on the retina.
The brain contains cells that measure how the pattern of light intensities varies across the retinal image. Specifically, there are some cells in visual cortex that measure the 2D orientation of intensity gradients in the image. This project develops and tests the theory that these cells are the main input to the computations that allow us to perceive 3D shape. The project uses both computer models of neural circuitry to develop the theory, and psychophysical experiments using volunteers to test the predictions of the theory. The ultimate goal of the project is to build a model that explains how the brain pools measurements of image orientations into specific organizations so that it can infer 3D curvatures and orientations. This allows us to map out the process of 3D shape estimation from the first stages of the process — which deal with 2D image measurements — to the final stages — which represent 3D surface structure in the brain.
The following scientists take part in this collaboration:
- Prof. Roland Fleming, Dept. of Experimental Psychology, Justus Liebig University Gießen
- Prof. Steven Zucker, Dept. of Computer Science, Yale University, New Haven, USA
Justus Liebig University Gießen
Phone: +49 (0) 641 / 99-26140