Receptive Fields Summary
Rods and Cones => Small Spots
Retinal Ganglion Cells, LGN => Larger Spots
Primary Visual Cortex => Bars/Edges etc…
Beyond V1 = ????
Receptive Fields:
Rods & Cones
Small Spots
Receptive Fields:
Retinal Ganglion Cells and LGN
Larger Spots
Receptive Fields:
Primary Visual Cortex
Bars/Edges etc…
WHAT do you see in this image?
At the level of the retina, you “see” an array of point-lights bouncing off the page and exciting your rods and cones.
In early visual brain areas, you “see” a collection of oriented lines and a collage of red, green, yellow, and blue color patches.
But your response to this question was almost certainly not “light” or “lines” or “colors”; what we all perceive in this scene are “toys.”
The ability to organize visual sensations into coherent objects and then assign meaningful category labels to these objects is in many ways the ultimate accomplishment of vision.
Ambiguous figure
A visual stimulus that gives rise to two or more interpretations of its identity or structure.
In a way, all images are inherently ambiguous. Any 2D image can have an infinite number of 3D interpretations.
An “ambiguous figure” is usually obviously ambiguous.
*Black and white spotted ambiguous picture of dalmatian.
Adaptation that Is Specific to Spatial Frequency
This previously shown adaptation demo fatigued small groups of neurons thus revealing their specific responsibilities (i.e. orientation & spatial frequency
Can we adapt to more complex stimuli?
Dalmatian Adaptation?
No.
“THIS IS A JOKE. YOU CAN NOT HAVE ADAPTATION TO DALMATIONS!”
Middle vision
A loosely defined stage of visual processing that comes AFTER basic features have been extracted from the image and BEFORE object recognition and scene understanding.
- Involves the perception of edges and surfaces.
- Determines which regions of an image should be grouped together into objects.
High level vision
Loosely defined stage that involves complex image analysis including 3D vision, object recognition, scene understanding and more.
Extrastriate cortex
Region of cortex bordering the primary visual cortex (V1) and containing multiple areas involved in visual processing
-V2, V3, V4, etc.
After extrastriate cortex, processing of object information is split into “WHAT” and “WHERE” pathways of High Level Vision.
“Where” pathway
concerned with locations of objects, but not their names or functions
Dorsal pathway = “Where/how” system
Object Localization/Manipulation
Parietal Cortex
“What” pathway
concerned with names & functions of objects, regardless of location
Ventral pathway
Object Identification
Inferior Temporal (IT) Cortex
Middle Vision Receptive Fields
The same visual input occurs in both (a) & (b) and a V1 neuron would respond equally to both.
A V2 neuron might respond more to (a) than (b) because the black edge is owned by the square in (a), but not in (b)
Extrastriate Receptive Fields
The receptive fields for cells in extrastriate areas are more sophisticated than those in striate cortex.
They respond to visual properties important for perceiving objects.
For instance, “boundary ownership.”
For a given boundary, which side is part of the object, and which side is part of the background?
Lower Vision
The functionally defined primary visual cortex (V1) is approximately equivalent to the anatomically defined striate cortex.
Structuralism
A school of thought believing that complex objects or perceptions could be understood by the analysis of the components.
Developed by Wilhelm Wundt and his mentee Edward Titchener.
Structuralism:
Lines
Edges are important, they give us: LINES,
How do we know which lines go together? Which belong to the same object and which do not?
We see edges that computer algorithms can’t.
Picture of gradient colored arrow.
Picture of image filtered with Photoshop’s “Find Edges” filter.
But you SEE the edges as continuous…. Ha!!
A better illustration of “dumb” edge detecting algorithms…is next.
Kanisza triangle
These are “imaginary” edges.
Nothing is there…. But…
- Sometimes people actually report seeing different shades of white inside and outside of the house.
- Brain responds differently to both types of white surface!
- This image illustrates the point that the visual system knows about physics!! Objects tend to be opaque and obstruct the view of other objects behind them.
- Monkeys also see this as a house…
Finding Edges
We see objects that computer algorithms can’t.
Gestalt Psychology
“The whole is greater than the sum of its parts”
Gestalt: In German, “form” or “whole”
Wertheimer, Köhler, Koffka (1920s–1950s)
Gestalt psychology opposed to other schools of thought, such as structuralism.
Gestalt grouping rules:
A set of rules that describe when elements in an image will appear to group together.
Patterns are spontaneously organized by the brain into the simplest possible configurations.
Gestalt grouping rules:
A set of rules that describe when elements in an image will appear to group together.
Patterns are spontaneously organized by the brain into the simplest possible configurations.
Good Continuation
A Gestalt grouping rule stating that two elements will tend to group together if they lie on the same contour.
Some contours in an image will group because of good continuation.
Can you find the shape embedded within the field of lines at right?
When should we complete edges behind occluders?
When the edges are relatable by an “elbow curve”
S curves = NOT LIKELY THE SAME EDGE
Elbow curve = likely the same edge.
Also, the steeper the curve, the less likely it is the same edge. (PARAMETRIC RULE)
HEURISTIC = mental shortcut (not always right!).
Texture segmentation
Carving an image into regions of common texture properties
Similarity & Proximity
Gestalt grouping rule
*Absolut Boston beer advertising
Parallelism
Gestalt grouping rule
Parallel contours are likely to belong to the same figure
Symmetry
Gestalt grouping rule
Symmetrical regions are more likely to be seen as figure
Symmetry & Parallelism help group objects.
More symmetry/parallelism is generally considered “better” (but not always).
Common region
Gestalt grouping rule
Two features will group if they appear to be part of the same larger region
Connectedness
Gestalt grouping rule
Two items will tend to group if they are connected
Dynamic grouping principles
Common fate: Elements that move in the same direction tend to group together
Synchrony: Elements that change at the same time tend to group together
Common fate
group elements moving in the same direction together.
They are going to get to the same place at the same time.
Synchrony
group elements changing at the same time together
flashing “open” sign, the lights flash at the same time
Camouflage
An organism’s attempt at breaking Gestalt rules so that its features are not perceived as an object on their own, but as parts of a larger object.
Figure–ground assignment
The process of determining that some regions of an image belong to a foreground object (figure) and other regions are part of the background (ground)
Gestalt figure–ground assignment principles: surroundedness, size, symmetry, parallelism
Gestalt figure–ground assignment principles:
- Surroundedness: The surrounding region is likely to be ground.
- Size: The smaller region is likely to be figure.
- Symmetry: A symmetrical region tends to be seen as figure.
- Parallelism: Regions with parallel contours tend to be seen as figure.
Ambiguous figures are created when these principles compete
Surroundedness
The surrounding region is likely to be ground
Size
The smaller region is likely to be figure
Symmetry
A symmetrical region tends to be seen as figure
Parallelism
Regions with parallel contours tend to be seen as figure
The Rubin Vase
Edgar Rubin: Danish Psychologist.
Isn’t that powerful (that OR starts before Figure-Ground segregation)??
Flips in ambiguous images: we think perception is as stable as the world! But that is not always the case.
Other Example: M. C. Escher, Encounter and Sky & Water I
Figure-Ground in Advertising
Absolut Malta beer advertising images
Which circles are “figures” and which are holes?
Context is important.
Both circles are small, surrounded, symmetrical, but one is seen as an object (penny), the other as a hole, based on their Context.
+ Figure/Ground segregation is not a simple process, even if the rules themselves seem easy…
This is an ambiguous image. Either a strip and a dot, or a T behind, or a strip and two dots (Accidental view).
- Global superiority effect
Parts and wholes
(Navon, 1977):
“Forest before the trees”
The properties of the whole object take precedence over the properties of parts of the object
- Why do gestalt principles work?
The Gestalt grouping processes capitalize on certain regularities that characterize the physical world.
In other words, the grouping principles have ecological validity.
Awkwardly worded summary provided by book:
- Bring together what should be brought together.
- Split asunder what should be split asunder.
- Use what you know.
- Avoid accidents.
- Seek consensus and avoid ambiguity.
Prof’s summary:
- GROUPING MECHANISMS.
- FIGURE GROUND SEGMENTATION.
- Use heuristics for findings corners, etc… and divide objects into parts.
- Do not consider as likely the accidental view point representations