Term
| small scale modules: hypercolumns & columns |
|
Definition
| Hypercolumns are made out of many different little columns. A column is a bunch of cells that respond to bars of a single orientation. |
|
|
Term
| hypercolumn structure and function |
|
Definition
| Hypercolumns are made up of all the simple cells, complex cells, and hypercomplex cells for ONE patch of visual space. Code for images right eye vs left eye. If you travel down one direction, you get cells that respond to different orientations. (vertical, horizontal, diagonal). A hypercolumn is any little patch of the V1 that includes both a left and right ocular dominance slab |
|
|
Term
|
Definition
| Primary visual cortex (v1) first cortical visual area, where most visual information is processed. The cells are like stripes. |
|
|
Term
|
Definition
| A hypercolumn is any little patch of V1 that includes both a left and right ocular dominance slab. |
|
|
Term
|
Definition
| A bunch of cells that respond to bars of a single orientation, just like how simple cells are bar/edge detectors. |
|
|
Term
|
Definition
| Almost everywhere you look, what you "see" is the surface. We don't represent things behind the first opaque surface. Surfaces are a level of description between low level receptive fields and the objects. |
|
|
Term
| Gestalt grouping laws (all) |
|
Definition
| Laws that the visual system uses to decode images. Proximity: the discs appear to group in pairs due to proximity. Bad form: certain cues indicate that two bits do not belong together (T junctions). Amodal completion: surfaces complete behind an occulating surface. Modal: surfaces complete in front of each other even though the border is not there - the completd surface is seen (illusory surface). |
|
|
Term
|
Definition
| Modal: two surfaces have the same color and brightness. There is no contour in the image. But we see a subjective contour. Amodal: surfaces complete behind an occulating object; the surface is not seen but it is registerd. Evidence that the two black regions behind a gray rectangle is registered as "joined". You can only get modal completion if your visual system assumes the AMODAL. |
|
|
Term
| physiology of subjective contours |
|
Definition
| Recordings from a single cell in area V2 of monkey - V2 cells can't tell the difference between a real ine and an illusory one. |
|
|
Term
|
Definition
|
|
Term
|
Definition
| Border is now owned by the surface in the back. |
|
|
Term
|
Definition
| Figure is in the front. Modal completion in front for camouflaged contours. Figure in front owns its borders. |
|
|
Term
|
Definition
| top down help from object knowledge |
|
|
Term
| line drawings (cartoons, outlines) |
|
Definition
| Cartoons go straight from the outline to the ojbect. Our vision is more than just a cartoon line drawing. |
|
|
Term
|
Definition
| Our eye gets the light that is reflected from a surface - this is the product of the light's intensity I and the surface reflectance R. We see reflectance and light but only get the product L = I x R |
|
|
Term
|
Definition
| complex shapes in scene determined by direction o the light; shape of object casting the shadow and surface on which it falls. Darker but same color; similar texture inside and outside. X JUNCTIONS of shadow and surface contours. An area where light is blocked by an intervening object. Not common in art until modern times. |
|
|
Term
|
Definition
| Does not belong to the surface on which it is seen. Changes as the observer moves, scene changes, object moves. Reflected pattern does not have to match the pattern accurately or the environment. REFLECTIONS DO NOT MOVE WITH THE OBJECT. Add to the realism of materials. No shadows on reflective surfaces. Can be the wrong shape, color, scene, angle, but are always lighter than surface they are seen on, vertically aligned with reflected objects if reflection is on a horizontal surface (water), and natural. |
|
|
Term
|
Definition
| Checkerboard illusion - when we correct and discount for the illuminant we make two areas that send equal amount of light look very different. What you perceive is not what hits your eye, it's what hits your eye times a discounted illumination. |
|
|
Term
|
Definition
| Shading is the variation in reflected light due to the change in the orientation of the surface. Shading helps us determine the shape of the object. |
|
|
Term
| feature detector, line detector |
|
Definition
| Ex: the frog's eye tells the brain because it has ganglion cells that respond to small moving spots (trigger motor response "bug detectors") and seagulls respond to open their mouths to gape for food. Hubel & Wiesel suggest that a shape is a piecewise set of adjacent bars and corners. Is a single line encoded as a set of adjacent line segments? We know now that this is NOT THE CASE. |
|
|
Term
|
Definition
| evidence for wavelets: spatial frequency tuning of cortical neurons. Narrow tuning for specific frequencies. Selectivity to bar width vs grating spatial frequency. |
|
|
Term
| commonality between sound and vision: frequency (tone or pattern) |
|
Definition
| A simple shape is a superposition of components (like a sound). |
|
|
Term
|
Definition
| Can decompose sounds into frequency sound waves (spectrum is graph between amplitude and frequency). Like in a graphic equalizer. Vision can be decomposed into sine waves as well (component image sine waves). Variation in brightness from low luminance to high luminance (luminance over space). Complex patterns can be made of the sum of simple sine patterns of different frequency. |
|
|
Term
| creating images out of sine waves |
|
Definition
| vision can be deconposed into sine waves as well. The variation in brightness from low luminance to high luminance ranges. Luminance over space. Complex patterns can be made out of the sum of simple sine patterns of different frequency (decomposition or synthesis). |
|
|
Term
| receptive field - revised understanding, patches of sine waves |
|
Definition
| small patches of sine waves (wavelets/Gabors). The same summation property holds. Arbitrary spatial patterns can be created by summing small patches of sine waves of many orientations and signs. Low spatial frequency > high spatial frequency. |
|
|
Term
| CSF (not cerebrospinal fluid) |
|
Definition
| Contrast Sensitivity Function - taking all cells together and plotting their individual spatial frequency tuning curves, we get an "envelope" - outer edge of all the cells' responses (peak sensitivity of all the cells) of the limit of our visual ability. We can see everything within the envelope; don't have any cells outside of it and can't see it. This ENVELOPE OF PEAK SENSITIVITIES OF INDIVIDUAL NEURONS are the limits of our perception and make up the contrast sensitivity function. |
|
|
Term
|
Definition
| CSF of a single cell is the outer edge of all the cell's responses |
|
|
Term
| CSF for whole system (perception) |
|
Definition
| CSF of all the cells combined plotting their individual CSF tuning curves together gets the whole envelope of peak sensitivities of individual neurons (which are limits of our perception) and form the contrast sensitivity function :) |
|
|
Term
| CSF for different species, ages |
|
Definition
| As you get older, your CSF gets smaller (contrast sensitivity function declines) and you can see less high spatial frequency. |
|
|
Term
|
Definition
| Monet's paintings changed (loss of high spatial frequency) as he grew older |
|
|
Term
|
Definition
| Features: object parts. In the inferotemporal cortex there are cells that respond to "parts" and characteristics of spatial patterns that are part of familiar shapes. Meaningless on their own. |
|
|
Term
|
Definition
| Cells that respond to "parts". Columns in the inferotemporal cortex have different parts represented in different columns (may be transformed from one to another). Has a structure, different from other shapes. |
|
|
Term
| FFA, PPA, lateral occipital cortex |
|
Definition
| FFA: Fusiform Face Area. Activity here when performing facial recognition tasks. Damage leads to face recognition problems (like prosopagnosia). PPA: Place recognition. Activity here when viewing houses, outdoor scenes, buildings. Lateral Occipital Cortex has activity when responding to object recognition tasks. Damage leads to object recognition problems like visual agnosia. |
|
|
Term
|
Definition
| upright faces are special; inverted faces are like any other object; put parts together to perceive the whole. We are poor at recognizing upside down faces. Upright faces are special; perceive the whole face at once (not by parts) - automatic, easy. |
|
|
Term
|
Definition
| Faces are recognized by high level object recognition - holistic processing. Either distribution coding or grandmother coding. Distribution more likely. Population response and tuning. Faces are NOT just a simple combination of parts like other objects. |
|
|
Term
|
Definition
| we process the face all together at once, we don't put them together by parts independently. When we put two different face together, we perceive one new face. |
|
|
Term
|
Definition
| extrastriate cortex in occipital lobe: Extrastriate Body Area (EBA). fMRI shows activity here when viewing whole human bodies or parts from stick figures, silouettes, etc. Low activity to faces, objects, animal bodies. |
|
|
Term
|
Definition
| Fusiform Face Area (FFA) responds to faces. We don't put them together by parts - good at recognizing upright faces and bad at recognizing inverted faces. Perceive the whole face at once, not by parts. This is fast, automatic, and easy. |
|
|
Term
|
Definition
| Lateral occipital cortex. fMRI shows activity when object recognition tasks. Damage leads to visual agnosia. |
|
|
Term
|
Definition
| Inner mid portion of temporal lobe (PPA). fMRI shows activity when viewing houses, outdoor scenes, buildings. |
|
|
Term
|
Definition
| object recognition problems; patient DF cannot recognize object or even report them; yet her hands will orient correctly to openings or for grasping objects; supplementary vision in the dorsal "where" stream; association between recognition of what the thing is and the ability to act on it |
|
|
Term
|
Definition
| Ventral stream is the "what" stream, dorsal is the "where" stream. Significant because there is supplementary vision in the dorsal "where" stream" that can help visual agnosia patient orient correctly to openings for grasping of objects. Association between recognizing what an object is and being able to act on it. |
|
|
Term
|
Definition
| facial recognition problems; can see details but can't recognize familiar faces. Recognize friends from voice, gait, hair style. |
|
|
Term
|
Definition
| cues that support depth perception in flat, static images. Occlusion, linear perspective, known size, gradients, height in field, atmospheric perspective, shadows. |
|
|
Term
|
Definition
| perceived size on the retina x perceived distance. Size constancy = the image of an object on the retina gets smaller as the object gets farther away. Vision relies on many cues to judged distance and depth. Given the distance, an estimate on the size can be recovered. |
|
|
Term
|
Definition
| the apparent size of an image is directly proportional to the perceived distance of the surface on which you see it. |
|
|
Term
| special point of view effects |
|
Definition
| buildings with right angles and parallel lines. Geometry requires that parallel lines converge in the distance. Gives metric information about depth. Introduced into painting by romans. Given a viewing point, locate real world points on the picture plane. gives realistic capture of distortions in the world; linear perception cues in the world. generaly requires that objects be attached to surfaces. hard to use linear perspectives. |
|
|
Term
|
Definition
|
|
Term
|
Definition
| t-junctions that indicate which objects are in front of which others (but NOT by how much; relative, not metric) |
|
|
Term
|
Definition
| object knowledge - if an object is familiar and has a typical size, then reverse the relation and recover the distance to the object. Useful cue but often overruled ALSO METRIC. |
|
|
Term
|
Definition
| we assume that the things on the ground are uniform in size; the change in size must be due to the chance in distance. Doesn't require level ground plane but it helps. We have to assume that everything on the ground is the same size, or else the cue does not work. top down knowledge helps. |
|
|
Term
|
Definition
| assuming a flat, level ground plane. Height in field corresponds to distance. Horizon is at eye level - distance to horizon, square root (2rh), about 3 mi. metric cue. |
|
|
Term
|
Definition
| distant surfaces have less contrast because intervening atmosphere superimposes a haze. The farther the distance, the more the haze, the lower the contrast. This is important because in SF we think buildings are farther away when there is more fog between us and them. in the moon we midjudge distance a lot becaus there is no atmosphere |
|
|
Term
|
Definition
| The effects of light (shadow, shading, highlight) may be discounted, tell us about depth. Tell us about relative placement of objects (ordinal, not metric) and about the relief of the surface on which they fall. Place objects relative to the surface on which they cast their shadow. |
|
|
Term
|
Definition
| perspective cues to depth overrides the knowledge of size (in the exploratorium) - we assume that the room has 90 degree angles but it realy doesn't. |
|
|
Term
|
Definition
| when something is in smooth motion it is much easier to tell depth from them because objects (trees) closer to you move by faster than objects farther away from you. Ex: on a trip, looking out the window. |
|
|
Term
|
Definition
| difference between distant focus and close focus. Bringing a target into focus - lens is stretched or relaxed. The visual system senses the degree of strain on the lens and then translates this into a distance for the object. |
|
|
Term
|
Definition
| you converge your eyes on closer objects (more convergence means closer objects). Just move your eye from one object to another to know the depth (not useful for objects farther than 1m away) |
|
|
Term
|
Definition
| the difference in the separation of two images in two eyes. The dstance (difference in spacing) between the two fingers you hold out in front of you as seen in each eye. |
|
|
