Lecture 4 Foundations and Cognitive Processes in Visual Perception From the Retina to the Visual Cortex 1.Vision Science 2.Visual Performance 3.The Human Visual System 4.The Retina 5.The Visual Field and the Cortex 6.Retinotopic Projections 1/47
Vision Science Bradford Books, MIT Press, Cambridge, MA 1999. Part of Cognitive Science How is the structure of optical images processed to extract useful information about the environment? Perceptual psychology, psychophysics, computer science, neurophysiology, neuropsychology Single most coherent, integrated and successful branch of cognitive science 2/47
Vision A Kind of Computation Eyes and brains Video cameras and digital computers The language of computation serves as a common language for unifying and communicating the findings from the different disciplines Great challenge Relating basic physical phenomena evoked on submicroscopic levels to high-level concepts that do not depend on visual appearance 3/47
Concepts: What defines a Chair? 4/47
Visual Perception Concerns the acquisition of knowledge Distinct from purely optical processes like photography Knowledge concerns objects and events in the environment It is not merely about subjective visual experiences 5/47
Perception as Constructive Act Vision as a window to reality Is vision veridical? Vision adapts to environment Increased / decreased sensitivity E.g. adaptation in darkness (movie theater / night sky) Afterimages (non-veridical!) Complimentary color and brightness Motion 6/47
A small experiment...! Instruction: - Focus on the center of the red circle. - After the red circle disappears, keep the focus.
Orientation After Effect (Nacheffekt) 8/47
Motion: The Spiral Illusion 9/47
The Moon Illusion What is the real moon size? Why does the moon appear much larger when it is close to the horizon than when it is high in the night sky? Source: wikipedia.org 10/47
Lecture 4 Foundations and Cognitive Processes in Visual Perception 1.Vision Science 2.Visual Performance 3.The Human Visual System 4.The Retina 5.The Visual Field and the Cortex 6.Retinotopic Projections 11/47
Visual Performance Visual Performance Computational Model Visual performance Object recognition, visual illusions, change blindness, Functional decomposition in visual perception Levels of visual perception Basic processes of perception Functional architecture of the brain / computer 12/47 Brain / Computer
Object Perception On the next slide you will see a picture Raise your hand if you recognize objects Please do not tell what you see Especially if you have seen the picture before!!! 13/47
Object Perception! R.C. James 14/47
Object Perception Bregman, 1981 15/47
Object Perception Doolittle, 1985 16/47
Ambiguous Figures (1) What causes viewers to see different figures in these images? Why does the interpretation switch? 17/47
Ambiguous Figures (2) What causes the periodical flipping in the interpretation of the Necker cube? 18/47
Ebbinghaus Illusion Is the left inner circle larger, equal, or smaller than on the right? Which processes cause a misjudgment wrt. the size of the inner circle? 19/47
Contour and Contrast Illusions What do you see? Hermann grid illusion: Which processes produce the images of grey spots at the intersections? Why is it impossible to suppress these images? 20/47
Lecture 4 Foundations and Cognitive Processes in Visual Perception From the Retina to the Visual Cortex 1.Vision Science 2.Visual Performance 3.The Human Visual System 4.The Retina 5.The Visual Field and the Cortex 6.Retinotopic Projections 21/47
The Human Visual System 22/47
Visual Pathway Hypothesis Different areas of the visual cortex are specialized in processing different types of visual information The flow of information follows pathways! LGN lateral geniculate nucleus (of the thalamus) V1, V2, MT (Medial Temporal cortex) = V5 23/47
Four Levels of Visual Perception Retinal image: 2-D projection of the environment Image-based processing: edge detection Surface-based processing: 2-D primitives: regions Object-based processing: 3-D primitives Category-based processing: Knowledge-based recognition 24/47
The Human Eye 25/47
Lecture 4 Foundations and Cognitive Processes in Visual Perception From the Retina to the Visual Cortex 1.Vision Science 2.Visual Performance 3.The Human Visual System 4.The Retina 5.The Visual Field and the Cortex 6.Retinotopic Projections 26/47
The Retina: Schematic Structure Optic nerve fibers! Ganglion cells Inner synaptic layer Amacrine cells Bipolar cells Horizontal cells Outer synaptic layer Receptor nuclei! Receptors Pigmented layer 27/47
The Retina The retina as part of the brain Thickness: 0.25 mm Function Light penetrates retina; light-sensitive receptor cells are positioned at the rear side of the retina Front layer of retina collects information for transmission through the optic nerve Types of neurons in the retina Rods and cones (~ 125 million / eye) Bipolar, amacrine horizontal cells Ganglion cells (~ 1 million / eye) 28/47
Receptor Cells Cones Color sensitive (3 types) Require bright light High density in the center; low density in the periphery Rods Brightness-sensitive Work in twilight High density in the periphery; low density in the center 29/47
Experiments: Blind Spot Perceive the blind spot! Blind spot processing 30/47
From Receptors to Ganglion Cells Ganglion cells collect and process sensory information Direct and indirect (via bi-polar cells) connections between receptor and sensor cells #receptors : #ganglions = 125 : 1 Ganglion cells integrate information from a large number of receptor cells 31/47
Receptive Fields of Ganglion Cells Receptive Field: Area of receptors that influences a neuron Two types of receptive fields: On-center cells Off-center-cells 32/47
On-Center / Off-Surround Ganglion Cell Response 33/47
Overlap of Receptive Fields /47
Lecture 4 Foundations and Cognitive Processes in Visual Perception From the Retina to the Visual Cortex 1.Vision Science 2.Visual Performance 3.The Human Visual System 4.The Retina 5.The Visual Field and the Cortex 6.Retinotopic Projections 35/47
The Visual Field and the Cortex /47
The Visual Cortex optic nerve optic chiasma lateral geniculate nucleus Source: www.airflag.com 37/47
Line- and Edge Detectors in the Visual Cortex /47
Neural Wiring of Simple Cells in V1 Retina LGN Cortical Area V1 Receptive Fields Center Surround Cells Simple Cell 39/47
Complex Cells in V1: Motion Detection 40/47
Detection of Moving Objects /47
Column Structure of Visual Cortex Interleaved ocular dominance columns Orthogonal layers for orientations of stimuli Principle of retinotopic projection 42/47
Lecture 4 Foundations and Cognitive Processes in Visual Perception From the Retina to the Visual Cortex 1.Vision Science 2.Visual Performance 3.The Human Visual System 4.The Retina 5.The Visual Field and the Cortex 6.Retinotopic Projections 43/47
Retinotopic Projections into the Visual Cortex 44/47
Retinotopic Projections into the Visual Cortex 36 45 /47
Computational Models of Edge Detection: Local Operators Vertical Operator Horizontal Operator - 1 + 1-1 + 1 10 10 02 02 10 10 0-8 0 +8 0 10 10 02 02 10 10 0-8 0 +8 0 02 02 02 02 02 02 0 0 0 0 0 02 02 02 02 02 02 0 0 0 0 0 10 10 02 02 10 10 0-8 0 +8 0 10 10 02 02 10 10 0-8 0 +8 0 46/47
Edge Detection with Neural Nets Image Intensity Retinal Image Excitatory (+1) and inhibitory (-1) connections implement a vertical edge detector 47/47
Second-Order Edge Operators +1-2 +1 +1-2 +1 +1 +1 +1 +1-8 +1 +1 +1 +1 Vertical Horizontal Omnidirectional 48/47
Explanation of Hermann Grid Illusion 49/47
Receptive Fields of Ganglion Cells Light alley! Dark borders light alley Light intersection Dark borders light intersection Homogeneously dark surface homogeneously homogeneously light surface dark borders dark borders dark surface 50/47
Hermann Grid 51/47
Link Collection Visual Illusions http://michaelbach.de/ot/index-de.html http://www.panoptikum.net/ optischetaeuschungen/ http://www.sehtestbilder.de/optischetaeuschungen-illusionen/ 52/47
Next Week From lines via regions to 3D perception Depth perception: Shadows Perceptual organization (Gestalt principles) Visual completion Object recognition & 3D models 53/47