Form Perception - Module notes + Lecture notes PDF

Preview

Summary

Module notes + Lecture notes ...

Description

Form Perception Sunday, March 4, 2018

8:32 PM

Introduction to Form Perception - Distinguishing figure from background is a complex visual function Gestalt Principles Gestalt philosophy - Gestalt psychologists --> German, 1920s and 1930s - Believed "the whole is diff than the sum of its parts" ○ People tended to perceive the whole stimulus rather than just putting together a collection of the stimulus' discrete parts - Gestalt movement --> reaction to structuralist approach - Ex. movies made from quickly flashing static pictures ○ Motion is an emergent property of a sequence of pictures - Gestalt principles: laws describing how we organize visual input ○ Believe people either born with the laws or acquire them very rapidly Six gestalt principles 1. Figure ground - Ability to distinguish an object from its background in a visual scene - Tend to have distinct borders that give it form over the background (perceived as in front of the background) - People constantly figuring out what's figure and what's background - Difficult process if cues that make figure-ground decisions are n't clear

2. Proximity - Tendency to group elements that are close together in space - We naturally see regions of high density as one group due to proximity - More likely to group objects that are closer together than far apart

3. Closure - Tendency to fill in gaps in a contour to perceive a whole object - We automatically fill in the parts we can't see to perceive a single object - Ex. perceiving a rectangle, despite obvious gaps

4. Similarity - We tend to group together objects of the same type - Tend to see columns of the same elements as belonging together

RE

5. Continuity - The ability to perceive a simple, continuous form rather than a combination of awkward forms - Ex. we perceive an X as 2 continuous lines rather than 2 "V's"

6. Common fate - The tendency to group together elements that change in the same way - Objects moving together in the same direction tend to be grouped together - Explains why we can suddenly see a camouflaged animal ▪ Allows contour to be perceived Pattern/object recognition Expectations - Shapes what we see

Process of object recognition - Establishing figure from background --> first step to object recognition - Then --> parts of figure identified and grouped together into a single object - Recognizing an object involves bottom-up and top-down processing Bottom-up processing - Features that are present in stimulus itself guides object recognition - Ex. people recognize a cow cuz its characteristics

- People recognize what they see by analyzing the indv's features to things with similar features in the person's memory Top-down processing - A person's beliefs or expectations guide their object recognition - Ex. "THE CAT" --> H and A are physically identical but still read as "THE CAT" cuz of influence of context

Priming experiments - Experimenter measures how fast a participant can read a word that's flashed on a screen - If participant is told that the next word is an animal --> priming effect will be seen - Words like dog or duck will have a faster reading time than log or puck

- Process of a word is more efficient if participant is primed to expect a word from a certain category - Top-down processing needs some input from the stimulus to work - Both top-down and bottom-up processing guide object recognition - Called bi-directional activation cuz processing occurs in both directions - Features of the object combined with our expectations guide object recognition Theories of object recognition - Object recognition can rely on varying degrees of processing Biederman's geon theory - Suggests that we have 36 different geons (simple geometrical forms) stored in memory - Possible to recognize over 150 million diff objects - Difficult to imagine the geons used in more complicated objects - Brain injury can affect recognition for only certain types of objects Template theory - Many diff templates are stored in memory --> when object is come across it is compared to the template in memory - If its matched, then it’s a familiar object and it can be named by activating connections to other language areas in brain - No match --> unfamiliar object and new template is stored in memory - Too many different stimuli exist to feasibly store in memory Prototype theory - The most typical or ideal example of an object is stored in memory - More flexible cuz exact match isn't needed between observed object and what's stored in memory The importance of parallel processing - Multiple brain systems process diff components of visual signal simultaneously (in parallel) Perceptual constancies Perceptual constancy - The ability to perceive an object as unchanging even though the visual image produced by the object is constantly changing Five perceptual constancies 1. Shape - An object is perceived to have a constant shape despite the shape of its retinal image changing with in point of view or change in object position - Ex. a door is perceived as rectangle despite its visual change when opening

2. Location - An object is perceived to be stationary despite changing location on our retina due to body movements - Ex. while driving, we don't perceive the objects outside to be moving 3. Size - An object is perceived to be the same size despite the size of its retinal image varying with distance - Ex. someone walking away is not perceived as shrinking in size

4. Brightness - An object is perceived to be the same brightness despite reflecting more or less light onto our retina - Objects have the same brightness whether in high or low illumination

5. Colour - An object is perceived to have a constant colour despite diff illumination conditions - Ex. a white dog is still recognized even under a reddish light Existing knowledge - Existing knowledge provides top-down influence on perception - Existing knowledge makes sense of changing visual stimuli Cues in scene - Cues in the scene indicate perceptual constancies - Depth cues indicate the size of an object relative to its distance - Colour cues indicate the influence of a light on an object's colour - The brain integrates the motion of of all elements in a scene - Perceptual constancies exist from prior knowledge and cues in our scene that the perceptual system automatically factors in Visual illusions - When brain is given ambiguous or partial info --> visual illusion - Occur cuz perceptual strategies which work most of the time, are used in particular situations where they don’t belong - Perceptual constancies can be overcome by removing the relevant contextual info

Three types of visual illusions 1. Muller-lyer illusion - Misapplied size constancy and misinterprets depth - Two lines are the exact same size

- Cultures without exposure to right angle are less susceptible to this illusion

The ames room - Looks like a normal rectangular room --> actually trapezoidal in shape - One corner is actually much farther away from your point of view than the other - Brain is tricked by cues normally used for distance constancy when it applies computations that normally leaf to size constancy - Manipulates distance to trick size constancy

The ponzo illusion - Manipulates depth cues to trick size constancy - Illusion occurs cuz 2 vertical lines are converging giving a sense of depth - Depth is being used to gauge size - Retinal size is the same but perceived distance differs

Feature Detectors - The brain processes stimuli in one region before passing it on to the next

Magno and Parvo Cells (in retina) - Transduce light stimulus into a neural impulse - Magno cells ○ Fond mainly in the periphery of retina ○ Used for detecting changes in brightness, motion and depth - Parvo cells ○ Found throughout retina ○ Used for detecting colour, pattern and form - Have small receptive cells - First step to object recognition --> axons of cells exit eye via optic nerve --> travel to LGN --> go to primary visual cortex in occipital lobe - Feature cells: very particular about what will make them fire ○ Located in primary visual cortex

Hodgkin and Huxley - Recorded electrical activity in indv neuron in squid = created method for recording activity in neurons

Lettvin et al - Discovered neuron in optic nerve of a frog that responded only to moving black dots ○ Called it "bug detectors" Hubel and Wiesel - Worked on neural activity in visual cortex of cats and monkeys - Began exploring by trying to learn what type of stimuli indv cortical cells responded to ○ Put microelectrodes in cortex of cat --> recorded electrical activity of indv neurons and cat shown diff types of visual stimuli ○ Showing a slide with a crack prompted crazy neuron firing - Found out that the cells fire maximally to stimuli of a certain shape, size, position and movement which defines the receptive field of the cell Simple cells - Responds maximally to a bar of a certain orientation in a particular region of the retina - Responds most to horizontal bar - If orientation of bar changes, cell is inhibited - Receptive field organized in opponent fashion

Complex cells - Responds maximally to a bar of a certain orientation and direction of movement, regardless of where the bar is located within the receptive field - For most, where the bar is in the receptive field doesn’t matter, even if its moving within the field - For some it matters

Hypercomplex cells - Responds maximally to a bar of a particular orientation and direction of movement, ending at specific points within the receptive field - Fires most t horizontal bar of light moving upwards that appears anywhere in the "on" region of receptive field - Gives weak response if bar touches "off" region - Cells have an inhibitory region at end receptive field ○ Makes them more sensitive to length of bar

Topographical organization - (layout of a visual scene) preserved in visual cortex - Neighboring objects processed by neighboring areas of brain - Largest amount of cortex devoted to processing - Info from central part of visual field --> projects onto the fovea - Each region of cortex receives some input --> within each region cells analyze specific features of the scene

Ventral Stream Combining information in the Extrastriate Cortex - Visual info must be combined to make sense of a visual scene

Subregion in extrastriate - Combo begins in extrastriate cortex (visual association cortex) --> surrounds primary visual cortex - Subregion such as colours about a scene, movement, diff line orientations - Extrastriate cortex breaks into 2 streams ○ Dorsal stream ("where" stream) ▪ Processes where objects are located in visual scene ▪ Processes how objects are moving within scene ▪ Takes info from PVC --> processes spatial info ○ Ventral stream ("what" stream): ▪ Processes ingo about what object is (form, colour) ▪ Sends info from PVC to temporal cortex □ Where all bits of feature info come together

Columns in the temporal cortex - Arranged in vertical columns oriented perpendicularly to surface of cortex - Respond to very specific stimuli that are more complex than stimuli to which neurons in PVC respond ○ Ex. hands, faces, apples, chairs - Each column has 5 layers of neurons ○ Each layer responds to complex stimuli coming from same category ▪ Each respond to slightly diff features within that category

- Objects are represented by unique activity patterns across many cells in several diff brain areas rather than specific neurons Development of pattern/object/face recognition Infant pattern recognition - Don’t perceive, objects and faces in the same way adults do - Prefer to look at patterns more than plain stimuli - Prefer high contrast with sharp boundaries between light and dark regions - Look longest at most complex stimuli that they're able to perceive

- Infants prefer larger squares ○ If squares are too small, they're poor acuity will make it look grey - By 2 years old, they prefer the smaller squares cuz they're visual acuity has improved The preferential looking method - Resolves the patterns infants can perceive

Infant object recognition - Their poor visual acuity may limit their ability to perceive whole forms - If a young infant is shown a triangle or a star and where they look is measured --> tend to stare at one corner and not at entire shape - Infants over 2 months of age begin to focus on entire shape - Infants naturally attracted to certain key features of a stimulus

3 month old infants and partial figures - Can perceive whole form when given only parts of the form - Infant can see invisible square cuz of previous habituation to a real square

Trouble with overlapping objects - Distinguishing between objects that are touching or overlapping involves requires ability to use cues like pattern and colours to tell what parts of an object belong together and what parts are separate and from another object ○ Young infants can't do this until they're 5 months old especially if they are still or moving together ○ If one moves, at 3 months they can distinguish Perceptual constancies in infancy - Sense of perceptual constancy is needed to recognize that when perceiving an entire form, it is the same under different viewing conditions - Infants develop some perceptual constancy by 4 months of age Ganrud's size constancy study - Tested infants of diff ages to see if they could understand that an object doesn't change in size when its viewed at diff distances - Shown teddy bear at specific distance, shown second teddy at farther distance - Infants between 4 and 5 months found the identical bears at 2 diff distances as familiar but stared longer at the larger bear that was viewed from a greater distance ○ Shows some understanding of size constancy and that object thats farther away should produce smaller retinal image Innate preference for faces

- We may be born to prefer face stimuli over other stimuli - Could serve to build necessary social bond with our caregivers - Some studies say infants don’t have a preference for faces exactly but rather for complex stimuli that have a lot of contrast between light and dark (eyes, mouth)and moving parts - Studies that tracked where an infant was looking when shown a picture of a face showed that infants under 2 months of age focused mainly on outer contours of the face (ex. hairline, chin) - Not until infant was over 2 months old that they looked at regions within face (ex. eyes and mouth)

Face preferences could facilitate bonding - Infants as young as 4 days old prefer looking at faces over other stimuli - By 2 months, prefer looking at attractive faces rather than unattractive faces ○ Look longer at mothers face than others - By 5 months, detect diff emotional expressions

Early experiences and learning - Early experiences with faces develops our preference for them Normal and abnormal visual development Interaction between innate readiness and environment - Innate readiness and external stimulation interact in normal development Cat studies using cylinder environments - Kittens with early visual deprivation raised in cylinder with vertical stripes on walls only - Failed to develop proper feature detectors for horizontal stripes ○ Unable to see horizontal edges and objects in environment Dr. Deda gillespie's work on dark deprivation in cats - 1 month old kittens kept in te dark for 3-4 days experience visual degeneration - 1 month old kittens kept in the dark for an entire week or longer suffer severe and permanent visual degeneration Humans born with cataracts - Sometimes babies are born with them - Lens is cloudy --> allows only diffuse light to reach retina = complete loss of ability to perceive any objects, patterns or details - Surgery can remove it and place artificial lens in it

Visual development - Humans require some form of early visual input for proper development Visual agnosia and prosopagnosia Primary visual cortex damage and "keyhole vision" - Suffering damage to PVC = loss of vision in some parts of visual field ○ Parts seen will seem normal ○ Able to perceive objects in intact areas of visual field like a keyhole ○ Everything in keyhole is seen perfectly - Damage to PVC doesn't disrupt object recognition Extrastriate damage - Entire visual scene might be intact - May be able to see all objects in scene - Much difficulty recognizing some or all objects - Agnosia: being able to see everything but not knowing what anything is Object agnosia - Inability to perceive objects - Unable to identify diff objects by sight - Sometimes the type of object that a person can't recognize is very specific - Many with object agnosia can still read ○ Shows recognizing words involves diff brain mechanisms that recognizing objects Prosopagnosia - Inability to recognize faces - Know that they're looking at a face and able to see eyes, a nose, and a mouth but can't put the indv features together to perceive whose face it is (even if they're looking in a mirror) - Rely on cues others cues to recognize other people (ex. voice, smell or way they walk) - Can have difficulty recognizing other specific stimuli but can recognize categories of objects Lecture Notes The importance of face perception - Big events are expected to have big causal explanations - Faces: a special form of pattern recognition - An innate preference for faces is influenced by experience - Pareidolia: finding a meaning in shapes and objects that's just random stimuli - People sometimes see religious faces Two strategies for face processing - Configural approach - Seeing the face holistically - Looking at the whole face - Top down processing - The composite face effect ▪ People asked if the two pictures of a face are the same face ▪ The picture was split into - Featural approach - Bottom up processing - The inversion effect ▪ Face would flash on the screen and then asked to choose which was the face shown

▪ When shown an inversed face that flashed on the screen participants were not as accurate and the reaction time was longer ▪ Thatcher illusion Are faces special? - The hollow illusion - Flashed face distortion effect - All images aligned at eye level...