Lecture 9: Representation of Knowledge 

When did mental imagery become a popular topic?

Types of Knowledge

Procedural

-Knowledge that we have acquired through learning about how to do things  -Acquired through repetition

-It’s implicit, you can’t really reflect on it

-Hard to change once you have it  

Declarative

-Knowledge of facts and events we’ve experienced  

-We acquire it through organization

-It is easily forgotten and changed  

-Episodic???? autobiographical, your knowledge based on your experiences  -Semantic???? facts, words, concepts, don’t reference our personal experiences; we  acquire it through school, learning etc.…

Explicit Episodic learning vs. Implicit Procedural learning:

What is the scope of hemisphere specialization?

-compared normal people to amnesiacs  

-had the participants solve the tower of Hanoi puzzle  Don't forget about the age old question of What are the different schemes observed in saltworks at chaux?

-they found that normal people can learn the procedure, transfer it to other puzzles  and they remember doing the task

???? the amnesiacs can learn the task even though they don’t remember doing it, they  can also transfer it to other puzzles, but they have no episodic memory  -results show there is a dissociation between episodic and procedural memory

Hm:

-functional STM

-could not transfer new declarative memory into LTM Don't forget about the age old question of What are examples of endothermic matter?
We also discuss several other topics like Which scientific attitudes should be exhibited in critical analysis?

-he performed a tracing task with a mirror, he improved over time even though he  never remembered doing it  

???? he learned, showed that implicit memory is for learning and memory

What are the rating scales like in dual-coding?

Semantic Memory

-stable memory

-uses different processes than episodic

-episodic involves hippocampus and anterior pre frontal cortex -semantic memory involves lateral temporal lobes and posterior pre frontal cortex  -highly organized

Tulving:

-subjective organization

-gave people a list of 30 words, look at how well they could recall the words -after enough trials, your recall would be higher  We also discuss several other topics like How does french romanticism differ from german romanticism?

-every trial presented the same words in a different order

-found that people imposed their own order on the words they were recalling ???? subjective organization, we have a natural tendency to organize our knowledge

Models of Semantic Memory

-descriptions where we formalize the structure of semantic memory and decision  processes  

Feature models: we compare features

1. Feature Comparison Models

Structure:

-knowledge consists of features

-features are weighted of importance  

???? defining vs. characteristic

Decision process: If you want to learn more check out Why do producers give some of the selling job to channel partners?
If you want to learn more check out How does implied consent work?

-all features are overlapped

-high overlap between common features and target species results in fast  response (first stage)

-low overlap between common features and target species results in slower  response  (second stage)

Evidence:

-Typicality rating

-consistent among most people

-less typical items how slower RT

-more typical items have fast RT  

2. Hierarchical Network Model

Why are true/false responses faster to “is a robin a bird” than “is a robin an  animal”

Structure:

-we have associations/links that have properties (nodes)

???? directional link: a robin is a bird, a bird is not always a robin -attributable links

-the higher up in the hierarchy, the more general  

???? common info is stored at one level

Processing:

-we retrieve info and search pathways between nodes  

-higher levels are more inclusive  

-going from one level to the next takes time

-info is stored at the level in which it is the most inclusive

-not quite an accurate model

3. Spreading activation model

Structure:

-no hierarchy

-things are interconnected  

???? the shorter the line, the stronger the relationship

Process:

-There is a spread of activation, when you activate one node, it activates  other relevant nodes  

Evidence:

semantic priming

-presented with a lexical decision task

-decide whether or not string if letters makes a word

-presented with 2 strings of letters at once  

-some words are related, some aren’t  

-if items are related, recognition is faster  

???? when you activate the first node for the first word, it spreads and activates  nodes that are related to the first word  

Ex: doctor, nurse

Lecture 10 

Mental Imagery

**not a popular topic until the 1950s because behaviourism was the main school I. Picture Memory

-memory for non verbal info is pretty good

Shepard: gave participants 200 pictures, for 10 seconds each. Had a retention  break and then conducted a recognition task. There was a 50/50 chance the  picture would be one you had seen. They got 98% correct

Nickerson: gave 600 pictures, for 5-10 seconds. Recognition test was done  one day later, accuracy was 92%. A year later, the accuracy was 63%.

Hemisphere specialization???? processing of words vs. processing of pictures Leehey: Stimuli is presented to either left or right hemisphere. Visual fields  project to opposite hemispheres. They found that words presented to the left  hemisphere were processed better. Pictures presented to the right  hemisphere were processed well.  

***Verbal and nonverbal stimuli are processed using different codes and  different brain structures (LH vs. RH)

II. Dual-Coding Theory

-the idea that there are 2 ways to represent or code information  Paivio: type of code depends on the nature of the info we are dealing with,  the task and the individual differences in people  

-concrete things are visable words, we can readily generate images for them -abstract things are words we can not readily imagine  

???? not specific to an object

???? rated the imagability of words

???? matched words based on meaning

???? paired association task

-if you have concrete items, the first image can provide cues to reinstate the  compound image

-results showed that the best remembered items were images that were  highly imagable and highly matched on meaning  

III. Dual-Coding and Individual Differences  

-subjective imagery ability can be tested using VVIQ, QMI and TVIC -the scale ratings tends to correlate  

-mental practice shows an impact on the subjectivity of mental  

representations

IV. Functional Equivalence

-the idea that there is an equivalence between images and what we can  actually do with them

-images are an abstract representation of something we image

-doesn’t have to be tied to visual input

-not a perfect representation of real world  

-can simulate relationships and changes  

Evidence for imagery code:

Mental rotation???? shepard and metzler task. The amount of response time  depends on how far the object needs to be rotated (physically or mentally) Mental object size???? our clarity depends on the size of the object in the visual  field. Visual acuity is improved when things are bigger in the visual field.  

Kossalyn:

-controlled size of object

-large objects were processed better

-showed that our internal images function in  similar way to perception of  visual objects

Lecture 11 

1. Visual Processing

-translating written squiggles into meaning

-our eye movements include saccades and fixation

-we don’t fixate on every word, we fixate on content words, words that  provide meaning  

-saccades jump 7-9 letters in distance (there is no info processing) -fixation occurs so we can foveate and pick up detail  

Eye Tracker: studies fixation and saccades  

Two Assumptions:

-The reader interprets the word as they read it, each word is processed as it  is read

-the eye fixates as long as the word as being processed  

Evidence

• Gaze contingent paradigm

-you have to foveate to process words

-they masked the letters in which you foveated on  

• Skipping words

-when you skip words, they usually are function rather than content  words

-you can skip words more often when text is easier to comprehend  • Normal Reading

-do we pre process the next word while fixating on a word?

-change the next word when the eye is moving  

???? people don’t notice

???? the next word is not pre processed  

• Fixations and Reading skill

-low skill leaders fixate more often and back track

2. Visual Word Recognition

-word recognition includes phonology, semantics and morphology -eye tracking, lexical decision tasks and naming are used to examine this

Word Recognition Model:

Marker Effect

Logogen Model

-an accumulator of info

-includes phonological and orthographic info of each word

-1 logogen for each known word

-each logogen has a resting level for activation

???? it doesn’t pass on info unless activation threshold is exceeded  -uncommon words needs to accumulate info longer in the logogen, which is  why we take longer to process (higher threshold)

-threshold is lower for common words  

*as we experience words more and more, the thresholds for the logogens for  those words gets lower and lower  

Context

-priming occurs when a word is preceeded by a word that is semantically  similar (Ex: doctor, nurse)

-Logogen model doesn’t account for meaning

-context system provides meaning in a separate network  

???? Semantics are organized in a separate system

-your thresholds are lowered when they are primed by other words and their  logogens  

Repetition Effects

-recognition will be faster the second time you see a word

-you see the word, activate the threshold, it slowly goes back down to  baseline  

-if the word is presented a second time in a short period of time, before the  threshold reaches baseline, it has less distance to go to activate threshold -auditory???? visual repletion does not show repetition effects

Stimulus Quality

-degrading stimulus quality results in longer reaction time in lexical decision  asks and naming tasks

-stimulus quality being degraded just adds time to the word processing,  regardless if its frequent or not

-encoding is affected by stimulus quality

???? low quality input is cleaned up by encoding stage  

Role for Phonological Coding in Reading

-sound plays a role in reading

• Edfelt

-used EMG during silent reading, showed evidence for speech being  sounded out at a sub vocal level

-EMG of harder text is shows higher levels of speech sounds  • Meyer, Schvaneveldt & Ruddy  

-sound priming ???? rhyming words shows priming, showing that sound  was activated during processing the words  

-visual priming???? if you present words that look similar but don’t  sound similar, you don’t get priming  

-phonology reflects phonological representations in the logogens  • Rubenstein, Lewis  

-pseudohomopones are not real words but sound like real words  -we are faster to name pseudohomophones than plain old non words -we need sound during reading, to keep some memory traces alive

-some words have multiple meanings, we have to go back and  disambiguate the meaning, sometimes a few words later -sound persists longer than visual memory  

-we add a second route to the original model, the grapheme to  phoneme model (lexical route is the context system, sub lexical info is  the grapheme and phoneme route)

Lexical route:

-sensitive to frequency

-provides correct pronunciation  

-not sensitive to regularity

-fast

-regular low frequency words will be processed into the same code in  both the lexical and non lexical routes  

Non Lexical Route:

-not sensitive to frequency  

-regular mapping

-based on regularities  

-slow

Whole word vs. Phonics debate

Lecture 12 

Applied Cognitive Psychology

Change Blindness

-when a change in the visual world occurs, but is not detected

???? even though you may be looking for it  

-saccades create massive transience (changes)

-the flicker between images is what causes us to not see the change

Inattentional blindness

-when you’re attention is focused on one thing and you don’t notice other, very  obvious things

Ex: gorilla ex

-we can minimize this by appropriately allocating our attention

Head up Displays

-traditionally, pilots must look down into their cockpit

-HUD maximizes head up, eyes out time

???? allows them to see what’s going on outside the plane

-head up displays are basically clear plastic on the cockpit window that displays the  necessary info  

-they help with certain tasks, like landing because the info you need to land, is  located where you need to look in order to land

-BUT, they can cause issues….

-when using a HUD, pilots miss or are slow to respond to events in the external  scene  

???? runway incursions are less likely to be detected

-you need to be able to focus and see the external world

-the same is true for automobile HUDS

???? speed monitoring with HUD was good

???? Lane position monitoring with HUD was bad because people focus they’re  attention on the speed HUD but it compromised their ability to focus on the external  scene, which allows you to keep your car properly positioned within the lanes

Cognitive analysis of HUD:

-leads to inattentional blindness, the attention is focused on the HUD and not other  information

-Cognitive tunnelling: your attention is tunnel visioned onto the HUD -this is because our attention is object-based

-common fate is strong enough to create objectiveness that hold your attention ???? the HUD is static whereas the world is moving

 Lecture 9: Representation of Knowledge 

Types of

Knowledge

Procedural

-Knowledge that we have acquired through learning about how to do things

-Acquired through repetition

-It’s implicit, you can’t really reflect on it

-Hard to change once you have it

Declarative

-Knowledge of facts and events we’ve experienced

-We acquire it through organization

-It is easily forgotten and changed

-Episodicà autobiographical, your knowledge based on your experiences -Semanticà facts, words, concepts, don’t reference our personal experiences; we acquire it through school, learning etc.…

Explicit Episodic learning vs. Implicit Procedural learning: -compared normal people to amnesiacs

-had the participants solve the tower of Hanoi puzzle

-they found that normal people can learn the procedure, transfer it to other puzzles and they remember doing the task

à the amnesiacs can learn the task even though they don’t remember  doing it, they can also transfer it to other puzzles, but they have no episodic memory

-results show there is a dissociation between episodic and procedural memory

Hm:

-functional STM

-could not transfer new declarative memory into LTM

-he performed a tracing task with a mirror, he improved over time even though he never remembered doing it

à he learned, showed that implicit memory is for learning and memory

Semantic Memory

-stable memory

-uses different processes than episodic

-episodic involves hippocampus and anterior pre frontal cortex -semantic memory involves lateral temporal lobes and posterior pre frontal cortex

-highly organized

Tulving:

-subjective organization

-gave people a list of 30 words, look at how well they could recall the words

-after enough trials, your recall would be higher

-every trial presented the same words in a different order -found that people imposed their own order on the words they were recalling

à subjective organization, we have a natural tendency to organize our knowledge

Models of Semantic Memory

-descriptions where we formalize the structure of semantic memory and decision processes

Feature models: we compare features

1. Feature Comparison

Models Structure:

-knowledge consists of features

-features are weighted of importance

à defining vs. characteristic

Decision process:

-all features are overlapped

-high overlap between common features and target species results in fast response (first stage)

-low overlap between common features and target species results in slower response (second stage)

Evidence:

-Typicality rating

-consistent among most people

-less typical items how slower RT

-more typical items have fast RT

2. Hierarchical Network Model

Why are true/false responses faster to “is a robin a bird” than “is a robin an animal”

Structure:

-we have associations/links that have properties (nodes) à directional link: a robin is a bird, a bird is not always a robin -attributable links

-the higher up in the hierarchy, the more general

à common info is stored at one level

Processing:

-we retrieve info and search pathways between nodes

-higher levels are more inclusive

-going from one level to the next takes time

-info is stored at the level in which it is the most inclusive -not quite an accurate model

3. Spreading activation

model Structure:

-no hierarchy

-things are interconnected

à the shorter the line, the stronger the relationship

Process:

-There is a spread of activation, when you activate one node, it activates other relevant nodes

Evidence:

semantic

priming

-presented with a lexical decision task

-decide whether or not string if letters makes a word -presented with 2 strings of letters at once

-some words are related, some aren’t

-if items are related, recognition is faster

à when you activate the first node for the first word, it spreads and activates nodes that are related to the first word Ex: doctor, nurse

 Lecture 12 

Applied Cognitive

Psychology

Change Blindness

-when a change in the visual world occurs, but is not detected à even though you may be looking for it

-saccades create massive transience (changes)

-the flicker between images is what causes us to not see the change

Inattentional blindness

-when you’re attention is focused on one thing and you don’t notice other, very obvious things

Ex: gorilla ex

-we can minimize this by appropriately allocating our attention

Head up Displays

-traditionally, pilots must look down into their cockpit

-HUD maximizes head up, eyes out time

à allows them to see what’s going on outside the plane -head up displays are basically clear plastic on the cockpit window that displays the necessary info

-they help with certain tasks, like landing because the info you need to land, is located where you need to look in order to land -BUT, they can cause issues….

-when using a HUD, pilots miss or are slow to respond to events in the external scene

à runway incursions are less likely to be detected

-you need to be able to focus and see the external world -the same is true for automobile HUDS

à speed monitoring with HUD was good

à Lane position monitoring with HUD was bad because people focus they’re attention on the speed HUD but it compromised their ability to focus on the external scene, which allows you to keep your car properly positioned within the lanes

Cognitive analysis of HUD:

-leads to inattentional blindness, the attention is focused on the HUD and not other information

-Cognitive tunnelling: your attention is tunnel visioned onto the HUD -this is because our attention is object-based

-common fate is strong enough to create objectiveness that hold your attention

à the HUD is static whereas the world is moving

1. Visual

Processing

 Lecture 11 

-translating written squiggles into meaning

-our eye movements include saccades and fixation

-we don’t fixate on every word, we fixate on content words, words that provide meaning

-saccades jump 7-9 letters in distance (there is no info processing) -fixation occurs so we can foveate and pick up

detail Eye Tracker: studies fixation and

saccades

Two Assumptions:

-The reader interprets the word as they read it, each word is processed as it is read

-the eye fixates as long as the word as being processed

Evidence

∙ Gaze contingent paradigm

-you have to foveate to process words

-they masked the letters in which you foveated on

∙ Skipping words

-when you skip words, they usually are function rather than content words

-you can skip words more often when text is easier to comprehend

∙ Normal Reading

-do we pre process the next word while fixating on a word? -change the next word when the eye is moving

à people don’t notice

à the next word is not pre processed

∙ Fixations and Reading skill

-low skill leaders fixate more often and back track

2. Visual Word Recognition

-word recognition includes phonology, semantics and morphology -eye tracking, lexical decision tasks and naming are used to examine this

Word Recognition Model:

Marker Effect

Logogen Model

-an accumulator of info

-includes phonological and orthographic info of each word -1 logogen for each known word

-each logogen has a resting level for activation

à it doesn’t pass on info unless activation threshold is exceeded

-uncommon words needs to accumulate info longer in the logogen, which is why we take longer to process (higher threshold)

-threshold is lower for common words

*as we experience words more and more, the thresholds for the logogens for those words gets lower and lower

Context

-priming occurs when a word is preceeded by a word that is semantically similar (Ex: doctor, nurse)

-Logogen model doesn’t account for meaning

-context system provides meaning in a separate network à Semantics are organized in a separate system -your thresholds are lowered when they are primed by other words and their logogens

Repetition Effects

-recognition will be faster the second time you see a word -you see the word, activate the threshold, it slowly goes back down to baseline

-if the word is presented a second time in a short period of time, before the threshold reaches baseline, it has less distance to go to activate threshold

-auditoryà visual repletion does not show repetition effects

Stimulus Quality

-degrading stimulus quality results in longer reaction time in lexical decision asks and naming tasks

-stimulus quality being degraded just adds time to the word processing, regardless if its frequent or not

-encoding is affected by stimulus quality

à low quality input is cleaned up by encoding stage

Role for Phonological Coding in Reading

-sound plays a role in reading

∙ Edfelt

-used EMG during silent reading, showed evidence for speech being sounded out at a sub vocal level

-EMG of harder text is shows higher levels of speech sounds ∙ Meyer, Schvaneveldt & Ruddy

-sound priming à rhyming words shows priming, showing that sound was activated during processing the words

-visual primingà if you present words that look similar but don’t sound similar, you don’t get priming

-phonology reflects phonological representations in the logogens

∙ Rubenstein, Lewis

-pseudohomopones are not real words but sound like real words

-we are faster to name pseudohomophones than plain old non words

-we need sound during reading, to keep some memory traces alive

-some words have multiple meanings, we have to go back and disambiguate the meaning, sometimes a few words later

-sound persists longer than visual memory

-we add a second route to the original model, the grapheme to phoneme model (lexical route is the context system, sub lexical info is the grapheme and phoneme route)

Lexical route:

-sensitive to frequency

-provides correct pronunciation

-not sensitive to regularity

-fast

-regular low frequency words will be processed into the same code in both the lexical and non lexical routes

Non Lexical Route:

-not sensitive to frequency

-regular mapping

-based on regularities

-slow

Whole word vs. Phonics debate

Mental Imagery

 Lecture 10 

**not a popular topic until the 1950s because behaviourism was the main school

I. Picture Memory

-memory for non verbal info is pretty good

Shepard: gave participants 200 pictures, for 10 seconds each. Had a retention break and then conducted a recognition task. There was a 50/50 chance the picture would be one you had seen. They got 98% correct

Nickerson: gave 600 pictures, for 5-10 seconds. Recognition test was done one day later, accuracy was 92%. A year later, the accuracy was 63%.

Hemisphere specializationà processing of words vs. processing of pictures Leehey: Stimuli is presented to either left or right hemisphere. Visual fields project to opposite hemispheres. They found that words presented to the left hemisphere were processed better. Pictures presented to the right hemisphere were processed well.

***Verbal and nonverbal stimuli are processed using different codes and different brain structures (LH vs. RH)

II. Dual-Coding Theory

-the idea that there are 2 ways to represent or code information Paivio: type of code depends on the nature of the info we are dealing with, the task and the individual differences in people -concrete things are visable words, we can readily generate images for them

-abstract things are words we can not readily imagine à not specific to an object

à rated the imagability of

words à matched words

based on meaning à paired

association task

-if you have concrete items, the first image can provide cues to reinstate the compound image

-results showed that the best remembered items were images that were highly imagable and highly matched on meaning

III. Dual-Coding and Individual Differences

-subjective imagery ability can be tested using VVIQ, QMI and TVIC

-the scale ratings tends to correlate

-mental practice shows an impact on the subjectivity of mental  representations

IV. Functional Equivalence

-the idea that there is an equivalence between images and what we can actually do with them

-images are an abstract representation of something we image -doesn’t have to be tied to visual input

-not a perfect representation of real world

-can simulate relationships and changes

Evidence for imagery code:

Mental rotationà shepard and metzler task. The amount of response time depends on how far the object needs to be rotated (physically or mentally) Mental object sizeà our clarity depends on the size of the object in the visual field. Visual acuity is improved when things are bigger in the visual field.

Kossalyn:

-controlled size of object

-large objects were processed better

-showed that our internal images function in similar way to perception of visual objects