Feb. 17, 11:45, Wei Ji Ma. Optimality and Probabilistic Computation in Visual Categorization

Naoki Kogo

University of Leuven

The side matters: Why line drawings only confuse us in understanding figure-ground perception

My talk will have two parts that are related each other. First, I will talk about some fundamental issues involved in investigating underlying mechanisms of a phenomenon called “completion”: filling-in of missing parts in an image. Computational models have applied contour-completion algorithms based on the detection of collinear boundary elements. However, there are examples that may indicate a fundamental problem of this approach. Importantly, for a signal to be qualified as a contour element, it has to be a part of a representation of a surface and hence the side of the boundary where the surface exists has to be specified. The same applies to interpreting neural activities at the location of completed contours. I hypothesize that completion is accompanied by a development of border-ownership signals. It means that the neural activities at the location of illusory contours not only indicate the existence of the boundary but also indicate that a surface on one side of the borderline is closer than the other surface. As the figure-ground perception is holistic, Gestalt, in nature, the global configuration of an image should be reflected in the computation of border-ownership signals. In the second part of my talk, I will talk about the computational model we developed recently (DISC model, Kogo (2010) Psy. Rev., 117, 406-439). I will show how border-ownership signals are computed in the model by global interaction between local elements in the image and I will explain an on-going research for further development of the model.

---home page (of our laboratory)

--- background readings

"Surface construction by a 2-D differentiation-integration process: A neurocomputational model for perceived border ownership, depth, and lightness in Kanizsa figures.", N. Kogo, C. Strecha, L. Van Gool, and J. Wagemans, Psychological Review, 2010, 117(2), 406-439.


"Linking depth to lightness and anchoring within the differentiation-integrationformalism.",  N. Kogo, L. van Gool, and J. Wagemans, Vision research, 2010, 50(15), 1486-1500


von der Heydt, R., Peterhans, E., & Baumgartner, G. (1984). Illusory contours and cortical neuron responses. Science, 224(4654), 1260-1262.


Zhou, H., Friedman, H. S., & von der Heydt, R. (2000). Coding of border ownership in monkey visual cortex. Journal of Neuroscience, 20(17), 6594-611.


"What is required for a signal to be qualified as a “grouping” tag?", N. Kogo, and J. Wagemans, British Journal of Psychology, in press.