A Self-Organizing Neural Network That Learns to Detect and Represent Visual Depth from Occlusion Events

Visual occlusion events constitute a major source of depth information. We have developed a neural network model that learns to detect and represent depth relations, after a period of exposure to motion sequences containing occlusion and disocclusion events. The network's learning is governed by a new set of learning and activation rules. The network develops two parallel opponent channels or "chains" of lateral excitatory connections for every resolvable motion trajectory. One channel, the "On" chain or "visible" chain, is activated when a moving stimulus is visible. The other channel, the "Off" chain or "invisible" chain, is activated when a formerly visible stimulus becomes invisible due to occlusion. The On chain carries a predictive modal representation of the visible stimulus. The Off chain carries a persistent, amodal representation that predicts the motion of the invisible stimulus. The new learning rule uses disinhibitory signals emitted from the On chain to trigger learning in the Off chain. The Off chain neurons learn to interact reciprocally with other neurons that indicate the presence of occluders. The interactions let the network predict the disappearance and reappearance of stimuli moving behind occluders, and they let the unexpected disappearance or appearance of stimuli excite the representation of an inferred occluder at that location. Two results that have emerged from this research suggest how visual systems may learn to represent visual depth information. First, a visual system can learn a nonmetric representation of the depth relations arising from occlusion events. Second, parallel opponent On and Off channels that represent both modal and amodal stimuli can also be learned through the same process.

[1]  Jonathan A. Marshall,et al.  Challenges of vision theory: self-organization of neural mechanisms for stable steering of object-grouping data in visual motion perception , 1991, Optics & Photonics.

[2]  Stephen Grossberg,et al.  A solution of the figure-ground problem for biological vision , 1993, Neural Networks.

[3]  Jonathan A. Marshall.  Unsupervised learning of contextual constraints in neural networks for simultaneous visual processing of multiple objects , 1992, Optics & Photonics.

[4]  J. A. Marshall,et al.  Self-organizing neural network architectures for computing visual depth from motion parallax , 1989, International 1989 Joint Conference on Neural Networks.

[5]  Shinsuke Shimojo,et al.  An occlusion-related mechanism of depth perception based on motion and interocular sequence , 1988, Nature.

[6]  Jonathan A. Marshall,et al.  Adaptive perceptual pattern recognition by self-organizing neural networks: Context, uncertainty, multiplicity, and scale , 1995, Neural Networks.

[7]  K Nakayama,et al.  Experiencing and perceiving visual surfaces. , 1992, Science.

[8]  V. S. Ramachandran,et al.  Perception of illusory occlusion in apparent motion , 1986, Vision Research.

[9]  Jonathan A. Marshall,et al.  Self-organizing neural networks for perception of visual motion , 1990, Neural Networks.

[10]  K. Nakayama,et al.  Occlusion and the solution to the aperture problem for motion , 1989, Vision Research.

[11]  L. Chalupa,et al.  Stratification of ON and OFF ganglion cell dendrites depends on glutamate-mediated afferent activity in the developing retina , 1993, Nature.

[12]  George A. Kaplan,et al.  Kinetic disruption of optical texture: The perception of depth at an edge , 1969 .

[13]  James S. Albus,et al.  Outline for a theory of intelligence , 1991, IEEE Trans. Syst. Man Cybern..

[14]  Stuart Anstis,et al.  Entrained path deflection in apparent motion , 1986, Vision Research.

[15]  K Nakayama,et al.  Stereoscopic Depth: Its Relation to Image Segmentation, Grouping, and the Recognition of Occluded Objects , 1989, Perception.

[16]  W. B. Thompson,et al.  Relative motion: Kinetic information for the order of depth at an edge , 1987, Perception & psychophysics.