Deck 2: Pattern Recognition

Question 1

In an experiment by Biederman, the recognition of objects was more difficult when lines were deleted at

A)midsegments, supporting a template theory.
B)midsegments, supporting a structural theory.
C)vertices, supporting a template theory.
D)vertices, supporting a structural theory.

Answer 1

D

Question 2

Emphasizing distinctive features when teaching young children to recognize letters

A)helps them to distinguish between letters afterwards and minimizes frustration due to errors.
B)helps them to create a holistic template for letters.
C)helps them to understand the underlying principles of phonics and learn to read more quickly.
D)is of no demonstrable benefit.

Answer 2

A

Question 3

In an experiment by Phillips, subjects had to decide whether two checkerboard patterns were the same or different. They could not make a template match

A)when the two patterns were in different locations on the screen.
B)after the visual information store decayed.
C)if the two patterns were separated by more than .5 seconds.
D)All of these

Answer 3

D

Question 4

If I were to describe my friend Bob by saying he has dark hair, blue eyes, and he's very tall, which kind of theory would I be using?

A)template theory
C)structural theory
B)feature theory
D)geon theory

Question 5

Egeland taught kindergarten children to distinguish effectively between confusable letter pairs by emphasizing

A)all the features of the letters.
C)the shared features.
B)the distinctive features.
D)none of the features.

Question 6

The importance of distinctive features in pattern recognition is demonstrated by the finding that people were

A)faster in identifying caricatures of faces than accurate line drawings of faces.
B)faster in identifying accurate line drawings than caricatures.
C)more often correct in identifying caricatures than accurate line drawings.
D)more often correct in identifying accurate line drawings than caricatures.

Question 7

The skeleton structure of animals can best be described by a

A)template theory.
C)structural theory.
B)feature theory.
D)None of these

Question 8

The results of the Phillips (1974) study discussed in your text indicates that

A)the template model may describe events within the sensory store.
B)the feature model may describe events within the sensory store.
C)the template model may describe events within long term memory.
D)the feature model may describe events within long term memory.

Question 9

Which of the following models does not belong?

A)filter
C)structural
B)feature
D)template

Question 10

A major difference between a feature theory and a template theory is

A)a feature theory specifies the relations between the features.
B)a template specifies the relations between the features.
C)a feature theory processes the input pattern as a single unit.
D)a template processes the input pattern as a single unit.

Question 11

The duration of the visual sensory store is approximately

A)2.5 msec.
C)250 msec.
B)25 msec.
D)2500 msec.

Question 12

A method for distinguishing between real people and intelligent computer programs is to require both to recognize

A)a face
C)a mangled word.
B)typed digits on a check.
D)a secret code.

Question 13

A set of proposed features is usually evaluated by

A)asking people whether it looks reasonable.
B)determining whether it can account for perceptual confusions.
C)observing how people draw patterns.
D)All of these

Question 14

According to Gibson's feature theory, features should remain unchanged despite changes in

A)brightness.
C)perspective.
B)size.
D)All of these

Question 15

Template theories

A)are designed to explain our ability to read words faster than letters.
B)take an unanalyzed pattern and match it against stored alternative patterns.
C)analyze the specific features of a pattern.
D)specify how the features of a pattern are joined to each other.

Question 16

Which theory seems to best describe the contents of the sensory store?

A)template theory
C)feature theory
B)structural theory
D)geon theory

Question 17

Biederman's component model is an example of

A)a feature model in which all the features are characterized.
B)a template model in which there are only a limited number of templates.
C)a structural model in which a limited number of components can be used to build many different objects.
D)None of these

Question 18

The theory that describes patterns by listing their parts is

A)template theory.
C)structural theory.
B)feature theory.
D)prototype theory.

Question 19

Which theory states that we compare patterns with each other and measure how much they overlap?

A)feature theory
C)template theory
B)Sperling's theory
D)Rumelhart's theory

Question 20

Structural theories of pattern recognition

A)deny the existence of features.
B)are extensions of feature theories.
C)assume that a pattern can be described by listing its features.
D)ignore the relationship among features.

Question 21

An important effect of the interactive activation model proposed by Rumelhart and McClelland is that it stimulated interest in

A)feature models.
C)template models.
B)neural network models.
D)structural models.

Question 22

Imagine that you view a brief flash of letters. In which series will it be easiest to judge whether the letter C or M was the third letter?

A)AFCE
C)--C-
B)FACE
D)All are equal

Question 23

Serial processing is _____ while parallel processing is _____.

A)one at a time; multi-tasking
C)geons; features
B)Rumelhart; Sperling
D)All of the above

Question 24

Sperling proposed that people use the auditory information store to rehearse the names of patterns. The auditory information store is a part of

A)the sensory store.
C)short-term memory.
B)the scan component.
D)long-term memory.

Question 25

Neural network models consist of nodes that are

A)independent of each other.
B)connected to each other by excitatory connections.
C)connected to each other by inhibitory connections.
D)connected to each other by excitatory and inhibitory connections.

Question 26

How does the partial report technique differ from the whole-report technique?

A)The partial report technique presents information for only a very brief time.
B)The partial report technique presents each line individually.
C)The partial report technique requires the subject to respond with all recalled information.
D)The partial report technique requires the subject to respond with only certain items.

Question 27

The 'word superiority effect' refers to the finding that it is easier to recognize a letter in a word than

A)a letter by itself.
C)Both a and b
B)a letter in a non-word.
D)Neither a nor b

Question 28

Computers are superior to humans in their ability to recognize patterns.

Question 29

Learning in neural network models occurs by

A)creating new nodes.
C)changing weights of connections.
B)creating new connections.
D)eliminating excess nodes.

Question 30

The interactive activation model of word recognition proposes that information about the letters in a word comes from

A)the feature level.
C)Both a and b
B)the word level.
D)Neither a nor b

Question 31

Neural network models have been most widely used to model

A)pattern recognition.
C)problem solving.
B)STM.
D)text comprehension.

Question 32

According to Rumelhart's model, people do better in the partial report procedure than in the whole report procedure because

A)they can use the visual information store to read off letters in the cued row.
B)the clarity of the visual information store increases over time.
C)they can often correctly guess which row will be cued.
D)they have less to remember.

Question 33

A distinctive feature is one that is present among all exemplars of a given category and helps define that category.

Question 34

One of the interesting results of Sperling's partial-report technique results was that

A)there was never a difference between the partial-report and whole-report findings.
B)so long as subjects responded within 5 seconds, there was no difference between the partial-report and whole-report findings.
C)after a one second delay, the results were the same as those obtained by the whole-report technique.
D)None of these

Question 35

Neural network models include all except which of the following?

A)processing units called nodes
C)states of activation
B)connections among nodes
D)strictly serial processing

Question 36

The purpose of Sperling's partial report technique was to distinguish

A)auditory encoding from visual encoding.
B)sequential scanning from parallel scanning.
C)perceptual limitations from memory limitations.
D)short-term memory from long-term memory.

Question 37

Template matches can occur in the sensory store.

Question 38

Which of the following is true for Rumelhart's model of pattern recognition?

A)It is a mathematical model.
B)Recognition is influenced by the number of items in the display.
C)Recognition is influenced by the clarity of information.
D)All of these

Question 39

What was the major revision in Sperling's 1967 model for the visual report task?

A)the change from sequential scanning to parallel scanning
B)the change from parallel scanning to sequential scanning
C)the addition of the visual information store
D)the addition of the auditory information store

Question 40

The decay rate of the visual information store depends on all of the following except

A)occurrence of a second exposure.
C)intensity of the stimulus.
B)contrast of the stimulus.
D)rehearsal.

Question 41

Neural network models can have connections that are only on or off.

Question 42

Describe the partial-report technique. What were the major findings from these studies?

Question 43

There are three major perspectives on pattern recognition: template, feature, and structural models. Explain the strengths and weakness of each model.

Question 44

Compare and contrast Sperling's model of information processing with Rumelhart's.

Question 45

Rumelhart's recognition model is influenced by the number of items in a display.

Question 46

A geon is essentially a three-dimensional feature.

Question 47

Why do people have difficulty recognizing faces of other races?

Question 48

Detection paradigms are those in which one has to specify which of two possible target patterns is present in a display.

Question 49

What is the word superiority effect? Describe the model discussed in your text to explain this effect.

Question 50

Sperling modified his original information-processing model by changing a parallel scan to a serial scan.

Question 51

The term superiority effect indicates that accuracy in recognizing a letter is higher when the letter appears alone than when it is part of a word.

Question 52

What are neural network models? What are the components of a neural network?

Question 53

Describe the impact of Biederman's work on structural theories. How did it build upon the previous work of others?

Question 54

Why was Sperling's partial report technique an innovation in research methodology, and why was it so important?

Question 55

Of the major theoretical perspectives of pattern recognition which do you prefer, and why?

Question 56

Neural network models can have both inhibitory and excitatory connections.

Question 57

Explain how perpetual learning can be facilitated by the highlighting of distinctive features. What benefits result from this method?