3: Pattern/Attention

Question 1

What are the different types of pattern recognition?

Answer 1

A) Template Models
B) Feature Models
C) Object Recognition

Question 2

What are template models and its limitations?

Answer 2

– Match stimulus to template in memory (we match the pattern that has entered our senses to a template we have stored)
– Support for idea: computers / scantrons
– limits
> inefficient to store many templates
> people/objects move and are different (very irregular world)
> has to be a strict match

Question 3

What are feature models?

Answer 3

– We recognize patters that enter our senses by their features (everything in our world is made up of features)
» the letter T has a vertical line and a horizontal line
» we bring these features together into an object we recognize

Question 4

How did Neisser support the feature model?

Answer 4

• • When people are scanning a list of letters, they are scanning the features and comparing them to the features of the letter “z” (target letter)
• • harder (slower) if target (Z) shares features with distractors!

Question 5

How did Lettvin support the feature model?

Answer 5

– microelectrodes into cells of frog retina and recorded activity
» edge detectors (certain cells would only respond to edges)
» moving edge (other cells would respond to moving edges)
» Dimming (respond to dimming LS)
» convex edge (small, circular dot that move ex. a fly)
– shows existence of feature detectors

Question 6

How did Hubel & Wiesel support the feature model?

Answer 6

– Recorded cell activity in the striate cortex (visual cortex) of a cat
–showed stimuli to the cat and recorded single cells in the cortex
» Simple cells: simple patterns of light, location specific (pick up edges, slits, lines)
» Complex: same as simple, but NOT location specific
» Hypercomplex: moving lines
» w,x,y: information about speed

Question 7

What is top-down (conceptually driven) pattern recognition?

Answer 7

– Gestalt grouping principles: processing of “whole” object
– Pattern recognition is influenced by knowledge
» Picking out letters in a scrambled word (university) vs seeing the words all together form “university” and noticing much faster
– Reading; Role of top-down and not feature-by-feature since we can read so fast

Question 8

How did Avant & Lydall support the top-down processing?

Answer 8

– Masking of BOY vs. YOB
– Results: a shorter interval is
required to erase (mask) BOY than YOB because “BOY” is getting into our semantic memory (recognition) before the mask can do anything compared to random letters that don’t mean anything and stay in our sensory memory. Therefore, the mask has to come after the target (BOY) quicker to erase it before it gets into our semantic (short term) memory
– This shows us that pattern recognition is not strictly bottom up, or else the masking interval wouldn’t be shorter for BOY

Question 9

What is the Word Superiority Effect (Reicher & Wheeler)?

Answer 9

– Harder to mask words than random letters; if you mask WIND and then ask the participants if it was WING or WIND they will choose WIND even though there is a small difference. This is because there is top-down affect

Question 10

What is repetition blindness (Morris & Harris)?

Answer 10

– Use of RSVP paradigm (Rapid Serial Visual Presentation -rapidly presenting things visually)
– “When she spilled the ink there was ink all over.”
» People fail to see stimulus (“ink”) the second time
» Top-down influence: the cognitive system has identified the “ink” the first time it was shown so the system is going through a refractory period, it doesn’t expect to see “ink” so soon again because it has already processed it
– “I broke a wine class in my class yesterday”
» People read “wine glass” (top-down context biases)
» Also, expect repetition blindness to class–class, but there was no repetition blindness to the 2’nd “class”: top-down (“glass”) overrides effect (“glass – class”)

Question 11

What is object-recognition?

Answer 11

– Recognition By Components (RBC) Theory (Beiderman)
– Objects made up of combinations of fundamental shapes called “geons”
– Recognition involves:
o first parse objects into component geons
o note where geons join (find edges)
omatch geon combinations to representations in memory
overy bottom-up model
– Problems:
» Expertise with processing certain types of features and objects affects our early perception
» Often it is quicker to recognize the whole object rather than the bits and pieces so if our processing is just bottom up then our recognition of the whole object should not be faster than the bits and pieces, but it is

Question 12

What is agnosia?

Answer 12

– when pattern/object recognition goes wrong (a failure or deficit in recognizing objects)
» pattern of features cannot be synthesized into a whole
» or person cannot connect the whole pattern to a meaning
– Caused by specific brain damage

Question 13

What is Prosopagnosia?

Answer 13

– disruption of face recognition

but these patients typically are able to recognize other objects

Question 14

What is Apperceptive agnosia?

Answer 14

• • disruption in perceiving whole patterns
• • can process basic features (lines, colours)
• • cannot integrate into a whole object
• • located in right hemisphere parietal lobe

Question 15

What is Associative agnosia?

Answer 15

– can combine features into a whole, can copy and describe a drawing (can draw a cat but can’t tell you it’s a cat
– cannot associate with a meaning (cannot
identify)
– involves temporal lobes of both hemispheres

Question 16

What do agnosia studies tell you about object recognition?

Answer 16

Recognition of objects is not “immediate”, but has 3 steps:

1) Perception of features
2) Integration of features into a larger whole pattern (Gestalt)
3) Association of the pattern to meaning

Question 17

What is visual attention and the types?

Answer 17

• • Sensory memory has large capacity. However, to be efficient, we must be able to select the most relevant information from the environment.
    1) Space-Based Visual Attention
    2) Object-Based Visual Attention

Question 18

What is space-based attention?

Answer 18

• • The visual attention system leads the eyes as a “spotlight” that processes locations
• • there’s two ways this can happen
    1) endogenously (internal/intentional) –place attention according to expectancy (driving a car along the highway and you want to change lanes but you know someone is coming up behind you so your attention is on the rear mirror so you know how fast and where they are so you don’t cut in front of them, or being aware of tacklers when playing football)
    2) exogenously (external cue) – draw attention to potentially important events in space (a flash of light, someone raising their hand, etc)
• • spotlight facilitates processing of information at the attended location and information at unattended locations is inhibited
• • Directing of attention can influence our perceptions– a dot showing up quickly on the left side of screen and then a line starting from the left line very quickly will make it look like the line was drawn out from the left to the right because our attention was drawn to the dot and then the line started there and our attention was already there so it affected our perception (seeing a line get drawn out from left to right)

Question 19

What is object-based attention?

Answer 19

– Attention is placed on objects
» within-object processing is more efficient (faster, more accurate) than across-object processing
– major goal of vision is to identify objects

Question 20

How did Lavie & Driver demonstrate object-based attention?

Answer 20

– line (object) study
» Two different lines, the target (gap) is on one line but far apart: fast recognition of targets
» Two different lines, the target (gap) is on the two different lines far apart: slow recognition of targets
» Two different lines, the target (gap) is on the two different lines but close: slow recognition of targets (even though the targets/gaps were close to each other, responses were slow meaning it is not space-based attention but object-based attention)

Question 21

What are the two theories of selective attention?

Answer 21

Filter Theory (Broadbent (1958)

Attenuation Theory (Treisman, 1964)

Question 22

What is filter theory?

Answer 22

3 components:

• • selective filter (only the information that enters this filter gets processed but not the other stuff); Regardless of how many competing channels or messages are coming in, the filter can be tuned, or switched, to any one of them, based on physical characteristics such as loudness or pitch
• • sent through filter and into limited capacity decision channel (essentially short-term memory)
• • long term memory (detection)

Question 23

What was the study that provided evidence for the filter theory?

Answer 23

– list of numbers to one ear (4, 7, 3) and list of numbers to other ear (1, 9, 5)
– told recall by ear: 65% of items
» attend to one ear, group those items then switch to other
– told recall sequence (4, 1, 7, 9, 3, 5): only 20% of items
» requires rapid switching between left and right ear
– suggests select info based on channel

Question 24

What was the dichotic presentation study (Cherry) on filter theory?

Answer 24

– Dichotic presentation: one message to one ear and a simultaneous message to the other ear
– Shadowing: repeating just one of the messages back (focusing attention on the message from one ear)
– Varied content of unattended
» if you change the unattended message voice gender, it disrupts the shadowing because this feature is being used to shadow and it messes up your attentional selection
» if you change the language, it doesn’t disrupt people’s attention/shadowing because the information in the unattended voice is not being processed
– suggests that we attend / select using physical cues

Question 25

What are the problems with filter theory?

Answer 25

• • person’s name on unattended channel will affect the shadowing (they were disrupted)
• • Not everyone detects their name easily: people with less cognitive capacity (a.k.a. working memory) were more likely to detect their name (less able to focus on a task and more prone to distraction and are more likely to process and detect information that they are supposed to ignore)

Question 26

What was the Grey & Wedderburn study on filter theory?

Answer 26

-- L.E: 
car
9
ter
-- R.E.
3
pen
5
-- responses organized by word/digit (They say “carpenter 3 9 5” because they group it by semantics, not by channel)

Question 27

What was the Corteen & Wood study on filter theory?

Answer 27

Phase 1:
– Presented people with names of cities and a shock but if it was a non-city name they didn’t get a shock
– after a while of this classical conditioning, when people are presented a city name they show a galvanic skin response (GSR)
Phase 2:
– old city names that had been paired with a shock & new city names that had not been paired with a shock in attended vs. unattended ears
– According to broadband’s model, your filter (attention) should be on the attended channel so whenever you hear an old city name in that attended channel that had been paired with a shock, you should show a GSR response but if it comes up in the unattended channel you shouldn’t be processing the meaning of that word so you do not get a GSR
– What they found was that regardless of whether the city name came up in the attended or unattended channel, there was a GSR response
– also, the GSR happened with old and new city names that had not been paired with the shock
– This concludes that semantic info from unattended is processed
against Filter Theory

Question 28

What is attenuation theory?

Answer 28

• • filter is not just all-or-none, it attenuates and can be based on physical cues
• • Information from both channels gets in even if you are not attending to it
• • Each word has representation (logogen), the origin of a word is in our lexicon called the logogen for each word
• • thresholds: all incoming messages receive some amount of low-level analysis but when the unattended messages yield no useful or important information, they are attenuated (reduced not in their volume or physical characteristics but in their informational importance to ongoing processing)
• • This permits attention to be affected by the semantic aspects of the message, that is, a top- down effect
• • ex. hearing “house” in unattended channel is not as relevant so requires more activation to grab your attention but less activation if it is in the attended channel; hearing your name in the unattended channel is more relevant so requires less activation to stimulate the logogen