Psych 316 Exam 2
Context effects
"the cat" example -information is ambiguous at the feature and letter levels -information is disambiguated at the word level Bottom-up: objects are recognized by their parts and combined to create the whole Top-down: When the interpretation is ambiguous, or when it is possible to parse the features in multiple ways, we use context to guide the parsing mechanism.
some basic characteristics of perception
-perceptions can change based on added info -perceptions can involve a process similar to reasoning or problem-solving -perceptions can be based on a perceptual rule (when objects overlap, the one underneath usually continues behind the one on top) which may be based on the persons past experiences -perception occurs in conjunction with action -perception extends beyond identifying objects or helping us take action within our environment -perception is the gateway to all other cognitions
3 examples of top-down processing involved in perception
1. perceiving objects: 2. hearing words in a sentence 3. experiencing pain -pain can be influenced by what a person expects, how the person directs their attention, and the type of distracting stimuli that are present perception is created by a combination of bottom-up and top-down processing
Edge Grouping by Collinearity
An early stage of processing where edges of objects are derived -collinearity means having a common line
Other shortcomings of RBC
Geon recognition alone is not enough to recognize all objects -what about different objects with the same geons? (polar bear and a panda bear) -what about individual objects within the same category (your cat vs. other cats?) Some objects cannot be recognized based on geons (e.g., mass nouns objects such as water, sand, for which texture and color are important for recognition). Another system is needed
Which of the following is an assumption of Biederman's Recognition-By-Components (RBC) theory?
Geons can be recognized from any angle, except accidental angles
recognition by components (RBC)
Geons: "geometrical ions" -36 geons make up the visual object alphabet -objects are made by combining geons at their edges (like LEGO)
Beyond the given info
Gestalt Psychologists argued that our perception of the visual world is organized in ways that the stimulus input is not. -the perceptual "whole" is often different from the sum of its parts According to Jerome Bruner, our perception of a stimulus differs from, and goes beyond, the stimulus itself
Figure-ground separation
Gestaltists claimed that the figure is perceived as having a distinct shape or form and the ground lacks form -as your perception changes on what is the figure and what is the background, the quality of the stimulus changes --ex: face-vase figure/pic -This suggests that figure perception and ground perception may be governed by different processes separating the object from the background
Evidence for feature binding
Interactive agnosia: patients with this condition have trouble combining or integrating features of an object -ex: patient cannot find an inverted T among upright T's although he could identify an inverted T when it was presented alone -this is because he found it hard to group the distractors together (Humphreys et al 1985) -can recognize objects if standing alone, but struggle when objects are placed together
Testing viewpoint independence
Isabel Gauthier and Michael Tarr -P's learned Greebles over a 2-week period (7 hour) -P's decide whether 2 greebles, presented in succession, are the same. The greebles are either at the same angle or are rotated at up to 75 degrees found: object recognition performance is not viewpoint independent
Feature vs. Configural Processing
Martha Farah proposed 2 pattern-recognition systems: 1. Feature Processing: recognition of simple parts and the assembly of those parts into larger wholes -cannot recognize someone (a face) from this (recognition of 2 eyes, a nose, a mouth) 2. Configural Processing: recognition of larger configurations. responsible for analyzing the spatial relations among features in a face. -this system is damaged in ppl with prosopagnosia
Are faces special? (contradiction)
No examples: -a prosopagnosic farmer could not recognize his individual cow -a prosopagnosic bird watcher lost his ability to discriminate warblers -another one lost her ability to tell cars apart -dog show judges show the inversion effect for dogs as they do for humans
What happens if you lose the ability of pattern recognition?
Object agnosia: unable to recognize objects Prosopagnosia: unable to recognize faces Aphasia: unable to comprehend/produce language. pattern recognition can be quite domain specific impairment to the sensory organs is not the cause of recognition difficulties
A Double Dissociation (What and Where Pathways continue)
Object discrimination (What): -pick the correct shape (triangle or rectangle) for a reward --monkeys with temporal lesions cannot do this (but they can do landmark discrimination right) Landmark Discrimination (where): -pick the shape closer to the cylinder for a reward --monkeys with parietal lesions cannot do this (but they can do object discrimination right)
Evidence for Structural description
Object-decision task: p's see picture or drawings of real and pseudo-objects, and they must decide which are real -some p's perform poorly on this task even though they perform at normal levels on tasks designed to assess earlier stages of object recognition (ex: matching objects from different viewing angles) -some p's perform extremely poorly when they have to name objects presented visually, but can perform normally when naming objects presented verbally -struggle with what objects are supposed to look like bc they have trouble accessing it --they can't bridge the information correctly/ can decipher the image correctly but can't take that info and connect it --ex: looking at a flower but can't describe it
evidence for edge grouping (stage 1)
Patient DF had severely impaired object recognition ability. She recognized only a few real objects and could not recognize any objects in line drawings (D. Milner, 1991). She also had trouble recognizing line orientation, which is important for detecting -once patient feels like she's doing an action with the object, she's able to do it/recognize it
Associative Agnosia
Perceptual processes remain intact but object recognition is impaired because of difficulties in accessing relevant knowledge about objects from memory
The What and Where Pathways
Recognition requires these 2 things Ungerleider and Mishkin (1982): performed lesions in either the temporal and parietal lobes of monkeys. -Monkeys performed either an object discrimination or a landmark discrimination task -monkeys with parietal lesion could not perform the landmark discrimination task -monkeys with temporal lesion could not perform the object discrimination task
Multiple-view account
Tarr and Bulthoff (1995) -argued for a multiple-view account of object recognition Multiple-view account: suggests that we store a few prototypical views of an object in our mind and match what we see with those stored views. -When an object appears at a viewpoint that does not match what you have stored in memory, your object recognition performance should suffer. --ex: cat lounging/ sitting up right is a different view than we store of cats
what is visual agnosia?
The impairment of visual object recognition in people who possess sufficiently preserved visual fields, acuity and other elementary forms of visual ability to enable object recognition, and in whom the object recognition impairment cannot be attributed to... loss of knowledge about objects -The impairment is one of visual recognition rather than naming, and is therefore manifest on naming and non-verbal tasks alike.
Which of the following pathways from the occipital lobes enable one to recognize the IDENTITY (not location) of an object?
The ventral pathway to temporal lobe
Perception and Action
They're deeply intertwined -interacting with an object helps you recognize it visually We typically consider object recognition as involving perceptual and memory processes - you see an object and then identify it. This is the "sitting in the chair" approach. But object recognition often goes beyond visual recognition alone- we interact with objects -physical interactions with objects often facilitates object recognition
more evidence for importance of orientation in recognizing faces
Thompson 1980: The Thatcher Illusion (2 pictures of same face upside down, one distorted. Try to tell which one is distorted) Fat Face thin illusion: face looks thinner upside down What does this tell us?: -our perception of faces are specifically built to detect featural details and their spatial relations from an upright position. -when a face is inverted, that ability is lost. -our perceptions of faces upside down and right side up are very different
Evidence for normalization
Warrington and Taylor (1978): asked P's to recognize objects presented at either a normal or an unusual view -each depicting an object at a different angle and had p's judge whether they were the same object -these p's performed particularly poorly when objects were shown at an unusual angle -the same occurred when they showed 2 pictures simultaneously, with each depicting an object at a different angle and had p's judge whether they were the same object
Visual Perception
We receive far more perceptual input via vision than other modalities the cortex devotes much of it's surface to vision relative to hearing, etc one of the major achievements of perceptual processing is pattern/object recognition
Explaining Visual Agnosia
What we have talked about so far is a oversimplification
The What and Where Pathways (continued)
Where/how: parietal lobe -dorsal pathway What: temporal lobe -ventral pathway Occipital lobe: primary visual receiving area)
View Normalization (stage 3)
allows a viewpoint-invariant representation to be derived. - controversial idea, bc evidence generally suggests that successful object recognition does not require viewpoint-invariant representations -(only stage that's not actually required for visual recognition bc we dont need to form a 3D-view to recognize the object)
top-down processing
ambiguous figures provide a good example of what we mean by "beyond the information given" -although the stimulus remains the same, our perception can change drastically depending on how we interpret the stimulus -ex: old/young lady illusion or bunny/duck illusion book: processing that originates in the brain, at the "top" of the perceptual system -the basis of top-down processing is other info such as: person's knowledge of the environment, the expectations people bring to the perceptual situation, and their attention to specific stimuli
2 types of Visual Agnosia
apperceptive and associative
Viewpoint independence
as long as geons can be clearly identified, object recognition performance should be the same regardless of viewing angle
Cooper and Biederman (1993) -evidence In support of RBC
asked P's whether 2 objects presented simultaneously had the same name -2 conditions for objects with the same name: 1. one geon was changed 2. all geons remained the same but the metric properties of the geon (ex: length) changed found that altering a geon impaired object recognition performance
The RBC model (Irving Biederman)
assumes a hierarchy of detectors. -the first level detects curves, lines, corners, edges, etc. These feed into the next level -next level detects geons, which then feed into the next level -the next level detects geon assemblies, which allows specification of relations among geons
Semantic system (stage 5)
at this stage, people gain access to stored knowledge of semantic (non visual) information relevant to the object ex: not what a horse looks like, but what is sounds like, can it run, etc. ex: what something is for, how you hold something, etc.
information for human perception
bc one purpose of perception is to inform us about what is "out there" in the environment, it makes sense that perception is built on a foundation of info. from the environment
Bottom-up processing: feature theory
complex patterns are identified by first recognizing the simpler components or elements of the pattern, and then combining the elements into the complex structure Distinctive features matter: Features that distinguish one stimulus from another should be particularly important.
Configural processing:
computes the spatial relations among objects in a face, such as the distance between the eyes and the nose
Contrast reversal
contrast reversed faces are also much more difficult to recognize than contrast reversed objects -"negative images" or black turned white and white turned black
Face inversion effect
dependence on orientation. -face recognition is heavily dependent on being upright, but you could recognize an object upside down Yin (1969): showed p's pictures of houses or faces during a study phase -test phase: p's were shown upright or inverted houses and faces (some of these were presented during the study phase and some were not) -p's were asked to judge whether each picture was seen in study phase or not found there was a bigger difference in identifying faces correctly
structural description (stage 3)
during this stage, individuals gain access to stored knowledge about the structure (ex: visual appearance) of objects -if you have deficits in this, then you'll have trouble recognizing some objects
Top-down processing: embedded figures
embedded figures are perceived if the overall form is parsed in a particular war. -if we know what we are looking for we can impose that order on the stimulus
perception:
experiences resulting from stimulation of the senses
experience-dependent plasticity
faces are typically recognized at the individual level but other objects are not (within-vs between- class discrimination) -we have more expertise with faces -if we can develop expertise at identifying objects in a particular category, recognition of these objects activates the fusiform gyri, which are allegedly used only for face recognition -ex: telling your cat apart from cats of the same breed -gauthier et. al study (1999)
Wong and Weisstein (1982)
flashed a vertical line that is slightly tilted to the left or right and P's must discriminate its direction of tilt The line is flashed either alone on a plain background or on the face-goblet picture. -when the line appears in the face-goblet picture, P's were asked to perceive either the goblet or the face, so the line is shown on the ground or the figure subjects must fixate their gaze at a crosshair and not move their eyes regardless of where the line is presented "perception of figure and ground has been suggested to involve two systems with different information processing characteristics. Figure perception is characterized by detail analysis and high resolution, while ground perception is characterized by low resolution and insensitivity to phase information"
2 assumptions of RBC
geons can be recognized from any angle (except accidental angles) geons are the alphabet of objects. Color and texture are not important for object recognition
Prosopagnosia
inability to recognize faces
Object agnosia
inability to recognize objects Most patients with object agnosia can identify objects via touch -Visual agnosia: patients who have much more visually specific agnosia -tactile agnosia: some patients can recognize objects visually but not by touch
Why is it difficult to design a perceiving machine?
inverse projection problem: the task of determining the object responsible for a particular image on the retina -involves starting with the retina image and extending rays out from the eye -perceptual system not concerned with determining an objects image on the retina, .. (above statement ) Objects can be hidden or blurred -people understand that part of an object this is hidden continues to exist Objects look different from different viewpoints: -Viewpoint invariance: the ability to recognize an object seen from different viewpoints
Hierarchical model of object recognition
meant to account for people with various deficits of visual agnosia -an information processing model, has 5 stages in serial manner Riddoch and Humphreys proposed a model to account for the various deficits in object recognition shown by brain damaged patients. 1. Edge grouping by collinearity 2. feature binding into shapes 3. view normalization 4. structural description 5. semantic system Apperceptive agnosia: -1. Edge grouping by collinearity 2. feature binding into shapes 3. view normalization -perceptual processing Associative Agnosia: -4. structural description 5. semantic system -
Feature Binding (stage 2)
object features that have been extracted during the edge grouping stage and the feature detection stage are combined to form shapes -this stage is akin to recognizing geons and/or geon assemblies in the RBC theory
Which assumption is held by the multiple views account of Object recognition?
object recognition is affected by the angle of presentation
Apperceptive Agnosia
object recognition is impaired because of deficits in perceptual processing
Evidence for Semantic System
patients with an impaired semantic system can show category-specific deficits in object recogntion Category-specific Deficit: -Imparied for natural objects but intact for artificial objects --ex: patient is impaired for natural things, like plants, but can recognize man-made objects, like a computer) or -impaired for artificial objects (though much less common)
Top-down, bottom-up How do we recognize objects?
pattern recognition is a stage between low-level sensory input and high-level cognition, and it is affected by both bottom-up and top-down processing
Pattern Recognition
patterns are complex and orderly configurations of sensory signals -sequence of sounds are recognized as words, music, or other noises. visual features are recognized as objects with color, depth, and motion pattern recognition is a process of identifying a stimulus by recognizing a correspondence between the stimulus and information in memory -it seems effortless, but has proven very challenging to recreate with artificial intelligence one remarkable feature of our ability to recognize patterns is that vastly different stimuli can be recognized as the same object
The importance of geons
people are much better at identifying objects when geons can be identified -if you can't recognize geons, you can't recognize objects
Double Dissociation: face vs Object (non-face) recognition
ppl can recognize objects but not faces, and others are opposite- can recognize faces but not objects
top-down processing:
processing that originates in the brain, at the "top" of the perceptual system -the basis of top-down processing is other info such as: person's knowledge of the environment, the expectations people bring to the perceptual situation, and their attention to specific stimuli
bottom-up processing
starts at the "bottom" or beginning of the system, when environmental energy stimulates the receptors -sequence of events from eye to brain
Aphasia
unable to comprehend/produce language
category-specific deficit
way more ppl struggle to recognize natural things/living things -natural things are more visually similar than man-made objects -natural objects are not made by humans, so they're not made to serve humans -man made objects are made to serve a purpose for humans ex: when you explain a polar bear, you'll probably explain what it looks like, but when you explain a screwdriver, you'll probably explain what it does (functional vs. visual difference) Natural objects are defined by visual features, artificial objects are defined by functions
Are faces special?
yes -individuals with prosopagnosia : lose their ability to recognize faces, though they can recognize other objects --fMRI evidence suggests that the fusiform face area specializes in face recognition Faces are different from other objects, a completely different region of the brain controls recognition of faces than object recognition