Lecture 3 Sept. 23rd Cognitive Psychology 2700
Lecture 3 Sept. 23rd Cognitive Psychology 2700 •
Pattern Recognition: We recognize patterns by only selecting certain information for further processing. I.
Template Models i.
Match stimulus to a mental template stored in memory. Support/examples for this idea are barcodes (Every time the code is read the same product comes up).
ii.
Limits: a) Inefficient: generally, it is not possible to have enough templates for everything. b) Irregular world: A lot of the objects we process are irregular. c) Strict match: they need a very strict match in order to work.
II.
Feature Models i.
Feature more regular than patterns: Features are easier to recognize than complex objects. We can recognize more complex things by first recognizing their features.
ii.
Complex objects composed of simple features: even people are composed of simple objects.
iii.
Gibson 1969: Studied Features of letters. Said that we actually perceive letters as a breakdown of lines and curves.
iv.
Neisser 1964: Had people look for letters amongst a whole bunch of other letters. He found that it was harder to pick out the letter if it had similar lines to the others. Ex. Easier to find the letter z in the following: OUQZG than in these: KWYZX.
v.
Hubel & Wiesel: there are different cells in the eye. He found this by studying cats. 1. Simple Cells: Location specific a) Simple patterns of light b) Location specific
c) Edge, slit, line 2. Complex: Same as simple but NOT location specific i.
Hypercomplex: Moving lines
ii.
W,x,y a) Speed of transmition b) Y (movement speed)
vi.
vii.
Word Superiority Effect: i.
Subject is shown the letter D, then a mask, then asked if they saw a D or a G.
ii.
Subject is shown the letter WIND, then a mask, then asked if they saw the word WIND or WING.
iii.
Subjects are better remembering the word WIND than the letter D. This is due to top down processing.
Repetition Blindness: we fail to see something when it happens a second time or is repeated. i.
vii.
RSVP: Rapid Serial Visual Presentation. Says that people fail to respond to/see stimulus the second time.
viii.
Avant & Lydall: Masking of BOY vs YOB. Found a shorter interval is required to erase BOY than YOB. In other words, the interval between the letters and the mask needed to be shorter for a real world. This is because real words are processed much more quickly.
ix.
Top Down: knowledge we already have is used to process information more quickly.
Object Recognition i.
RBC Theory: All objects are made up of features, and all features are common forms called Geons (fundamental geometric forms). Our system breaks objects down onto these geons, and we note where the geons connect. These geometric combos are remembered. Bottom up model.
ii.
Agnosia: Failure/difficulty in recognizing objects. Causes by specific brain damage. a) Apperceptive Agnosia: disruption in perceiving patterns. Can see lines, colour, etc. but cannot process into a whole. Located in the right parietal lobe.
b) Associative Agnosia: Can form gestalt, can copy a drawing, but cannot associate with meaning. Temporal lobes of both hemispheres. iii. •
Prosopagnosia: disruption of face recognition. Can be visual or auditory
Selective Attention I.
Filter Theory (Broadbent): i.
Sensory Register: Information (input) is absorbed into the sensory register
ii.
Selective filter: We selectively filter out info from our sensory memories. a) Limited Capacity Decision Channel: This carries the selected info to the detection device.
iii.
Detection Device
iv.
Short term memory
v.
Ex. Gave people a bunch of numbers at the same time to both ears (LE 1, RE7, etc.) When asked to recall the numbers based on what ear heard them 65% of numbers could be remembered. If asked to recall the numbers in order, they only got about 25%. Suggests select info based on channel.
vi.
Cocktail Part Theory: Disproves the above to an extent. Important info on the channel that isn’t being used “breaks through”.
vii.
Grey & Wedderburn: Showed that info is organized by word/digit not just channel.
viii.
Corteen and Wood: Two Phase study. a) Classical conditioning where they paired names of cities (CS) with a shock (UCS). After a while, the name of the city will illicit the same response as the shock. b) Dichotic listening task. Old city names are presented to one ear, new cities to the other. Every now and then, they are given a city name they had previously been conditioned to associate with shock; sometimes these cities are given to the attended ear, sometimes to the unattended. c) They found that the UCR occurred when the cities associated with shock came up in either ear. It was also found that the UCR also occurred for new city names (other names in the same category) which had not been conditioned to illicit the UCR.
II.
III.
Attenuation Theory: Treisman. Attenuation is like having taps you can turn up or turn down. i.
Sensory Register
ii.
Attenuation Control: there is limited capacity between this and the detection device
iii.
Detection Device: dictionary units.
iv.
Short term memory & Response
v.
Filter: Is not all or none, just attenuates. Uses physical attributes
vi.
Dictionary units: deals with words. We have a dictionary in our mind with all the words we know. This dictionary forms our mental “lexicon”. Logogens are our source of word knowledge and we have one for every word we know. This affects detection
vii.
Thresholds: Logogens are devices we use to recognize words, and they have a threshold. They are like a little cup and when we get enough info to fill the cup, we recognize the word. In other words, there is a build-up of activation until we reach threshold, then we recognize the word. Words that are very common have a lower threshold than words that we do not see all the time. Words that are significant to us (like our name) also have a much lower threshold than insignificant words. On the unattended channel, words need to reach an even higher threshold than normal to be noticed.
Multimode Model of Attention: selective attention at many levels i.
Cognitive economy: It is economical to selectively choose what we pay attention to.
ii.
Appropriate selection: Select the right things to pay attention to.
iii.
Johnson & Heintz: Had people do a shadowing task which was either easy or hard. The difficulty varied physically and/or semantically. People were also given a secondary task of needing to push a button as soon as a light changes (easy task). If the task the participant was doing was really hard, they were slower at the easy secondary change.
iv.
They found that semantics had no effect on the secondary task. However, difficult physically different messages slowed down reaction time.
Pattern Recognition: We recognize patterns by only selecting certain information for further processing. I.
Template Models i.
Match stimulus to a mental template stored in memory. Support/examples for this idea are barcodes (Every time the code is read the same product comes up).
ii.
Limits: a) Inefficient: generally, it is not possible to have enough templates for everything. b) Irregular world: A lot of the objects we process are irregular. c) Strict match: they need a very strict match in order to work.
II.
Feature Models i.
Feature more regular than patterns: Features are easier to recognize than complex objects. We can recognize more complex things by first recognizing their features.
ii.
Complex objects composed of simple features: even people are composed of simple objects.
iii.
Gibson 1969: Studied Features of letters. Said that we actually perceive letters as a breakdown of lines and curves.
iv.
Neisser 1964: Had people look for letters amongst a whole bunch of other letters. He found that it was harder to pick out the letter if it had similar lines to the others. Ex. Easier to find the letter z in the following: OUQZG than in these: KWYZX.
v.
Hubel & Wiesel: there are different cells in the eye. He found this by studying cats. 1. Simple Cells: Location specific a) Simple patterns of light b) Location specific
c) Edge, slit, line 2. Complex: Same as simple but NOT location specific i.
Hypercomplex: Moving lines
ii.
W,x,y a) Speed of transmition b) Y (movement speed)
vi.
vii.
Word Superiority Effect: i.
Subject is shown the letter D, then a mask, then asked if they saw a D or a G.
ii.
Subject is shown the letter WIND, then a mask, then asked if they saw the word WIND or WING.
iii.
Subjects are better remembering the word WIND than the letter D. This is due to top down processing.
Repetition Blindness: we fail to see something when it happens a second time or is repeated. i.
vii.
RSVP: Rapid Serial Visual Presentation. Says that people fail to respond to/see stimulus the second time.
viii.
Avant & Lydall: Masking of BOY vs YOB. Found a shorter interval is required to erase BOY than YOB. In other words, the interval between the letters and the mask needed to be shorter for a real world. This is because real words are processed much more quickly.
ix.
Top Down: knowledge we already have is used to process information more quickly.
Object Recognition i.
RBC Theory: All objects are made up of features, and all features are common forms called Geons (fundamental geometric forms). Our system breaks objects down onto these geons, and we note where the geons connect. These geometric combos are remembered. Bottom up model.
ii.
Agnosia: Failure/difficulty in recognizing objects. Causes by specific brain damage. a) Apperceptive Agnosia: disruption in perceiving patterns. Can see lines, colour, etc. but cannot process into a whole. Located in the right parietal lobe.
b) Associative Agnosia: Can form gestalt, can copy a drawing, but cannot associate with meaning. Temporal lobes of both hemispheres. iii. •
Prosopagnosia: disruption of face recognition. Can be visual or auditory
Selective Attention I.
Filter Theory (Broadbent): i.
Sensory Register: Information (input) is absorbed into the sensory register
ii.
Selective filter: We selectively filter out info from our sensory memories. a) Limited Capacity Decision Channel: This carries the selected info to the detection device.
iii.
Detection Device
iv.
Short term memory
v.
Ex. Gave people a bunch of numbers at the same time to both ears (LE 1, RE7, etc.) When asked to recall the numbers based on what ear heard them 65% of numbers could be remembered. If asked to recall the numbers in order, they only got about 25%. Suggests select info based on channel.
vi.
Cocktail Part Theory: Disproves the above to an extent. Important info on the channel that isn’t being used “breaks through”.
vii.
Grey & Wedderburn: Showed that info is organized by word/digit not just channel.
viii.
Corteen and Wood: Two Phase study. a) Classical conditioning where they paired names of cities (CS) with a shock (UCS). After a while, the name of the city will illicit the same response as the shock. b) Dichotic listening task. Old city names are presented to one ear, new cities to the other. Every now and then, they are given a city name they had previously been conditioned to associate with shock; sometimes these cities are given to the attended ear, sometimes to the unattended. c) They found that the UCR occurred when the cities associated with shock came up in either ear. It was also found that the UCR also occurred for new city names (other names in the same category) which had not been conditioned to illicit the UCR.
II.
III.
Attenuation Theory: Treisman. Attenuation is like having taps you can turn up or turn down. i.
Sensory Register
ii.
Attenuation Control: there is limited capacity between this and the detection device
iii.
Detection Device: dictionary units.
iv.
Short term memory & Response
v.
Filter: Is not all or none, just attenuates. Uses physical attributes
vi.
Dictionary units: deals with words. We have a dictionary in our mind with all the words we know. This dictionary forms our mental “lexicon”. Logogens are our source of word knowledge and we have one for every word we know. This affects detection
vii.
Thresholds: Logogens are devices we use to recognize words, and they have a threshold. They are like a little cup and when we get enough info to fill the cup, we recognize the word. In other words, there is a build-up of activation until we reach threshold, then we recognize the word. Words that are very common have a lower threshold than words that we do not see all the time. Words that are significant to us (like our name) also have a much lower threshold than insignificant words. On the unattended channel, words need to reach an even higher threshold than normal to be noticed.
Multimode Model of Attention: selective attention at many levels i.
Cognitive economy: It is economical to selectively choose what we pay attention to.
ii.
Appropriate selection: Select the right things to pay attention to.
iii.
Johnson & Heintz: Had people do a shadowing task which was either easy or hard. The difficulty varied physically and/or semantically. People were also given a secondary task of needing to push a button as soon as a light changes (easy task). If the task the participant was doing was really hard, they were slower at the easy secondary change.
iv.
They found that semantics had no effect on the secondary task. However, difficult physically different messages slowed down reaction time.
