Flashcards in Object and Face Recognition Deck (29):
Sensation refers to the effects of a stimulus on a sensory organ. It starts when a sensory receptor is stimulated and produces neural impulses that the brain then interprets as sound, visual images, taste, pain etc
Perception involves the elaboration and interpretation of a stimulus. The brain organizes information and translates it in a way that is meaningful and can be rationalized
In recognition, perceptions are matched with stored mental representations which then enable categorization and identification
Object recognition involves 4 broad stages:
1. Early Visual Processing (colours, edges, motion)
2. Grouping of visual elements
3. Matching grouped visual descriptions onto a representation of the object stored in the brain (structural descriptions)
4. Attaching meaning to the object (retrieved from semantic memory)
2. Grouping of Visual elements
This is implicated with the Gestalt principles which explain why certain visual elements are grouped together to form perceptual wholes
1. Law of similarity: visual elements are more likely to be grouped together if they share similar visual characteristics such as shape and colour
2. Law of proximity:
Visual elements are more likely to be grouped together if they are close together
3. Law of good continuity states that edges are grouped together to avoid interruptions or changes
4. Law of closure states that missing gaps are filled in
5. Law of common fate states that elements that are moving together are grouped together
Disorders in object recognition are referred to as visual agnosia. These disorders fall into 2 broad categories:
- Apperceptive agnosia: relates to impairments in perception
- Associative agnosia refers to impairments in ataching meaning
Is a type of apperceptive agnosia. Patients with this disorder are unable to follow Gestalt principles like other individuals and see the parts not the wholes.
They also cannot access stored objects from memory though can still see the basic visual elements
What is a patient with integrative agnosia unable to do?
- able to describe an object's parts even though he/she cannot recognize it
- unable to perform an object decision task in which a 'new' object is creating by recombining separate parts of an object
(this is because they are unable to see the object holistically and link the parts to a recognizable whole)
- able to copy drawings of objects which he cannot recognize, suggesting that he can still see them at some level
- able to draw objects from memory, suggesting that he can access structural descriptions from memory but not from vision
What can patients with associative agnosia do?
Are able to colour individual objects in complex drawings even though they cannot recognize what they are. Patients with apperceptive and integrative agnosia cannot do this.
Recognition of objects must occur in the face of obstacles
- changes in perspective
- changes in object orientation
- incomplete views
the ability to recognize an object as being the same from different viewpoints and different lighting conditions
The standard way of viewing the object.
Objects are recognized faster from this perspective
Routes to object constancy
- View-dependent theories
- View-invariant theories
According to view-dependent theories, object constancy requires the extraction of the principal axis of the object
According to view-invariant theories, object constancy requires the extraction of structural information about the object's components and the relationship between these components
View dependent theories support
Visual agnosic patients with damage to the right parietal lobe are able to recognize objects when shown in the standard (canonical) view but unable to recognize them when presented in an unusual view.
The parietal lobe is implicated in spatial processing and perception and may contain mechanisms which extract the principal axis of the object, rotating it to a standard view so to establish matching.
this means that these patients need to rely on a mechanism that is independent of the way the object is viewed.
View-invariant theories support
other patients with less severe damage in the parietal lobe are able to recognize objects in all views (canonical and unusual) including their names, but unable to discriminate the correct orientation of objects.
This is referred to as object orientation agnosia
this supports the view-invariant theories and also shows that the principal axis of an object is stored seperately from other aspects of objects
Perception for identification
- what is the goal of object recognition?
the goal of object recognition is to identify and object and determine its location
Ventral and dorsal pathways
A patient with visual agnosia
Ungerleider and Mishkin proposed that the dorsal (where) pathway is implicated in object perception and recognition and the ventral (what) pathway is implicated in spatial perception
A patient with visual agnosia, when asked to name the depicted object (combination lock) failed to come up with the correct answer and kept insisted that it was a telephone. However his hands mimed the action required to open a combination lock. This clearly shows that some information was being processed.
Perception for action
Goodale and Milner proposed that the dorsal pathway is more than just a "where" system but is also essential for guiding interactions with objects, thus it is also a "how" system
- Patients with visual agnosia can act on objects that they cant recognize
Dissociate the functions of the dorsal and ventral visual pathways
The dorsal pathway is implication in vision for action, working in real time with little memory and no access to consciousness
The ventral pathway is implicated in vision for perception, to create representations of the world around us and using conscious awareness
Patients with optic ataxia have their ventral pathway intact but their dorsal pathway impaired, they have no problem in recognizing objects but great difficulty when trying to reach out and grasp them
Thus they have poor visual guidance of reaching for objects due to damage in the parietal love and dorsal pathway
The problem with faces
Facial recognition is a within-category discrimination (all faces look the same) and object recognition requires a between-category discrimination (e.g. distinguishing between a cup and a pen)
The Bruce and Young model of face processing
This model assumes that the earliest level of processing involves computation of a view-dependent structural description
in fact, this model has a number of similarities with models of object recognition
Evidence for the The Bruce and Young model of face processing
- there is a double dissociation between recognizing familiar faces and matching unfamiliar faces across different viewing conditions (face constancy)
- in face naming, it is often possible to retrieve semantic facts (occupation) without retrieving the name, but the reverse pattern is not found (so name generation depends on semantic retrieval)
- there is a double dissociation between recognizing familiar faces and recognizing emotion, age and sex. this means that recognition of emotion, age and sex is independent of familiar face recognition
- there is a double dissociation between recognizing familiar faces and using lip-reading cues
Are faces special?
Faces are more distinct than any other visual object, and the process of facial recognition also requires identifying particular faces
several researchers have therefore claimed that faces require a different kind of processing.
The fact that face recognition can be selectively impaired also suggests that faces are different and might therefore be "special"
Patients with prosopagnosia are unable to recognize previously familar faces
□ A particular case of a patient with prosopagnosia was a man who could not recognize his wife but could still recognize her voice and other non facial information. He was also still able to recognize other objects and name them
- The fusiform face area is implicated in recognizing familiar faces in particular, and several researchers suggested this case as an example of domain-specificity, where neurons in this area are specialised to process only a particular type of information.
Some experiments show that faces may indeed be special.
1. Task difficulty
- Faces are very similar visual objects differentiated only by their features (nose, mouth etc), which could imply that their speciality is solely a cause of task difficulty.
An experiment was done consisting of an object recognition task which involved using different spectacle frames of equal difficulty to a face recognition task in controls.
The control group performed quite well on both tasks while the patient with prosopagnosia was only impaired on the face task but not the object recognition task
2. Holistic/ configural processing
□ Farah claimed that all object recognition lies on a continuum between recognition by parts and recognition by wholes, with faces lying on one end of the continuum (holistic-based) and words lying on the other end (parts-based). She also claims that face and word recognition always affect object recognition
□ Other researchers claim that the fact that faces, objects and words can be individually impaired shows that there are separate storage systems of stuctural information about these rather than a continuum between 2 types of underlying perceptual processes
- Evidence from human fMRI and monkey electrophysiological experiments shows that face-selective regions of the cortex respond to both whole faces and face parts
3. Visual Expertise
- Gauthier and colleagues note that faces require discrimination within a category as opposed to objects which require discriminating between categories. They claim that faces are special because through experience we become experts at within-category discrimination.
- This can also be applied to "Greebles" which involves becoming experts at discriminating between objects of the same category such as model car collectors. This also involves the fusiform face area.