PSYC*2650 Chapter 4: Recognizong Objects Flashcards
What is apperceptive agnosia?
Patients can see, but can’t organize the elements they see in order to perceive an entire object
What is assosiative agnosia?
Patients can see, but can’t link what they see to their basic visual knowledge
What is bottom-up processing?
Processing that is directly shaped by the stimulus
What is top-down processing?
Processing that is heavily shaped by the knowledge and expectations the person brings to the situation
How does recognition begin?
With the identification of visual features in the input pattern
What are two types of visual search tasks?
- A pop-out search/singleton search
- A conjunction search
What is the difference between a pop-out search and a conjunction search?
- Pop-out: All items except the target share the same features
- Conjunction: Target does not contain any unique features and must be identified based on a conjunction of two or more features
During a visual search task, does a participant’s speed in locating the target increase or decrease as the target is defined by more features?
Decreases
What is a tachistoscope?
A device that allows for the presentation of stimuli for precisely controlled amounts of time (modern research just uses computes)
What are brief presentations of stimuli called?
Tachistoscopic presentations
What are four factors that may influence the ability to recognize a briefly visible stimuli?
- Familiarity
- Recency of view
- Word-superiority
- Degree of well-formedness
What is the purpose of a post-stimulus mask?
To interrupt any continued processing for the initial stimulus
What is priming?
A process through which one input or cue prepares a person for an upcoming input or cue
What recency priming?
A pattern of priming where a stimulus that has been presented recently results in processing becoming more efficient
What is repetition/frequency priming?
A pattern of priming that occurs when a stimulus is presented on multiple occasions and processing becomes more efficient
What is the word-superiority effect?
People are more accurate and efficient in recognizing letters if they appear within a word or word-like string than when in isolation
What does degree of well-formedness measure?
The degree to which a string of letters conforms to the usual patterns of spelling
T or F: There is a strong tendency to misread less common letter sequences as if they were more common patterns.
True
What are feature nets?
Systems for recognizing patterns that involve a network of nodes, with nodes for features serving as the initial layer on each system, then moving up to more complex nodes like those for letters or words
What does activation level measure?
The current status for a node/detector
When does activation level increase?
If the node/detector receives the appropriate input from associated nodes/detectors
Will activation level be higher or lower if input has been received frequently or recently?
Higher
T or F: Only an input strong enough will increase the activation level of a node.
False. A series of weaker inputs can also increase activation level.
What is the response threshold?
The quantity of information or activation needed to trigger a response
T or F: The higher an activation level, the less stimulation it will need to reach a response threshold
True
What are bigram detectors?
Hypothetical units in a recognition system that respond/fire whenever a specific letter pair is in view
How is there bias built into feature networks?
Weak signals from ambiguous input will trigger a well-primed detector over one that is less primed
Does the bias built into feature networks make ambiguous/unclear input look more or less regular than they actually are?
More regular
What does it mean for information to be locally represented?
Information is encoded in a small number of identifiable nodes
What does it mean for information to be distributively represented?
Information is represented by a pattern of simultaneous activity across many nodes
Is “knowledge” about bigram frequencies contained within the network via local or distributed representation?
Distributed
What is improved, and what is sacrificed when the feature network is feed fast, yet ambiguous input, allowing the most common interpretation to occur?
Efficiency is improved, but minor sacrifices to accuracy are made
What are two ways in which the basic conceptualization of the feature net has been improved?
- The proposal of inhibitor connections among detectors
- The proposal that applies the network idea to the recognition of complex, 3D objects
What is different about the McClelland and Rumelhart feature network model compared to the basic model?
It includes both excitatory and inhibitory connections
According to the McClelland and Rumelhart feature network model, what are excitatory connections?
Connections that allow one detector to activate its neighbours
According to the McClelland and Rumelhart feature network model, what are inhibitory connections?
Connections from one node to another that decreases the activation level of the receiving node
What is the recognition by components model?
A model of object recognition where geons serve as the basic building blocks that construct all objects
What are geons?
3D shapes for which non-accidental properties are the same across most viewpoints
What are proposed as the building blocks of all complex 3D forms?
Geons
What are non-accidental properties?
Features of 3D images that are always present in any 2D viewpoint (excluding “accidental” views)
T or F: Geons can be identified from virtually any angle of view
True
T or F: Objects can be recognized from just a few geons.
True
T or F: cells in the inferotemporal cortex near the terminus of the “what pathway” seem to have specific targets and fire most strongly when that target is in view.
True
What brain area is proposed to provide the biological foundation for word and object detectors/nodes?
The inferotemporal cortex
What is viewpoint invariance?
The ability to recognize objects regardless of their viewpoint
What are the two theories that attempt to explain viewpoint invariance?
- Recognition by components theory
- Recognition by multiple views
What does the recognition by multiple views theory propose?
That 3D object recognition is based on the accumulation of encoded memories of multiple viewpoints
What are viewpoint-independent cells?
Cells that fire strongly to virtually any view of the target object regardless of orientation
What are viewpoint-dependent cells?
Cells that fire strongest to an object viewed from a particular angle
What is prosopagnosia?
A syndrome in which individuals lose their ability to recognize faces, despite their vision being intact
In terms of facial recognition, what are “super-recognizers”?
People who are incredibly accurate in the recognition of faces, but have no advantage in other perceptual or memory tasks
What is the inversion effect?
A pattern of recognition in which the recognition of faces is incredibly disrupted when viewed upside-down, compared to the recognition of ordinary objects viewed upside-down, which is only slightly disrupted
Which brain area only demonstrates increased activity when perceiving faces and face-like stimuli?
The fusiform face area
T or F: According to some researchers, the fusiform face area doesn’t just process faces, but any objects people have expertise with.
True
What is holistic recognition?
A process in which the ability to identify an object depends on viewing its entire configuration, rather than on an inventory of its parts
Is facial recognition based on the perception of features or holistic perception?
Holistic perception
What is the composite effect?
It’s harder to recognize half of a face when aligned with half of a different face
How does the recognition of familiar faces differ from the recognition of less familiar faces?
- More familiar: Recognition relies on the relationships among the inner facial features
- Less familiar: Recognition is more influenced by the outer facial features like hair and overall head shape
T or F: Most people are less accurate in the recognizing faces of people from their own racial background.
False. More accurate.
Can object recognition be viewed as a self-contained process?
No
