Pattern Recognition

Question 1

What are template theories?

Answer 1

Template descriptions are when we measure the overlap between two patterns

Question 2

What are the problems with templates?

Answer 2

–Great variability across templates
–A precise comparison requires the template to be in the same position and orientation

Question 3

What are feature theories?

Answer 3

Focuses on the parts of the pattern being recognized

Question 4

What is Gibson on Letters?

Answer 4

–A feature theory
–Features should be critical to provide a contrast
–The identity of important features should remain unchanged under changes in brightness, size, and perspective
–Should have a unique pattern
–The number of proposed features should be reasonably small

Question 5

What is perceptual confusion?

Answer 5

–P and R are likely to have more perceptual confusion as they differ on a single feature while P and W are unlikely to cause perceptional confusion because they are very different.

Question 6

What are distinctive features?

Answer 6

Features that help distinguish very similar patterns from one another

Question 7

What was Egeland (1975)?

Answer 7

–An effective method for emphasizing a distinctive feature is to initially make it a different color from the rest of the pattern and then gradually change it back to the original color
–Enables the children to learn distinctive features

Question 8

What are the types of distinctions?

Answer 8

–General distinction: this is a face
– Intermediate distinction: this is a face of someone with European heritage
–Specific distinction: this is my mother’s face

Question 9

What is Heugenberg et al (2007)?

Answer 9

–A study where participants were asked to perform a facial recognition task and they were either informed that it is more difficult to recognize differences in the faces of those from a different ethnic background from their own, or not.
–The participants who were informed of these difficulties did better because they realized they had to focus on specific distinctions rather than intermediate distinctions

Question 10

What are the problems with features?

Answer 10

–By themselves they are not an informative pattern
–A list of features ignores the relationship between those features

Question 10

What are Structural Theories?

Answer 10

–An expansion of feature theories, it focuses on how the features of a pattern are joined to other features of a pattern
–Emphasizes the relationship between features

Question 11

What is Biederman’s Component Model

Answer 11

–We only need 35 simple geons to describe the objects in the world
–Participants were better (70%) at recognizing objects if presented with observable relations
–Relationships between features are important for pattern recognition

Question 12

What is Sperling (1960)?

Answer 12

–Observers viewed a brief exposure of letters and were required to report all the letters in one of the rows
–There were two types of reports: partial- reporting a specific row and whole- reporting the entire array
– Results: Whole report- 4.5/12, Partial Report- 3.3/4
–The limitation of their ability to report all of the letters was not due to the pattern recognition stage of processing but limitations of the duration of the sensory store

Question 13

What is Sperling’s Model (1967)?

Answer 13

1. Visual information store is a sensory store that preserves information for a brief period lasting from a fraction of a second to a second
2. Scanning - for pattern recognition to occur the information must be scanned
3. Rehearsal (STM) - saying the information to oneself
4. Auditory Information Store (LTM) - remembering the information that was rehearsed

Question 14

What is Rumelhart’s Model (1970)?

Answer 14

–The experimenter presents a brief exposure of the letters F, R and Z
– IF the exposure is short you might only see some features of each letter
–IF you increase the display time you’ll have less errors
If you increase the number of letters you’ll have more errors
–This explains why the whole report is equal to the partial report since there are more items

Question 15

What was Reicher (1969)?

Answer 15

– Designed an experiment in which observers were shown a single letter, a four letter word, or a four-letter nonword
–The task was to identify a specific letter by selecting one of two alternatives
–The exposure was immediately followed by a visual masking field so that there was no lingering visual information in the sensory store
–The results showed that subjects were significantly more accurate in identifying the critical letter when it was part of a word than a non word.
–This is due to top-down processing since we need access to information from LTM to know that the patterns of letter provided is a word or a non word

Question 16

What was McClelland and Rumelhart (1981)?

Answer 16

–Created the interactive activations model to explain word superiority
–First assumption is parallel processing which is all four letters are processed simultaneously
–Second Assumption is that there are three separate levels of abstraction occurring simultaneously: Feature level, Letter level, Word level

Question 17

What are the components of neural networks?

Answer 17

–Nodes are represented by features, letters, and words in the interactive activation model. They can acquire different levels of activations
–Pattern of connection among node. Nodes are connected to one another by excitatory and inhibitory connections that differ in strength
–Activation rule specify how a node combines its excitatory and inhibitory inputs with its current state of activation