AllLectures PDF

Introduction to
Cognitive
Psychology
Greg Francis
Spring 2020
MWF 9:30 - 10:20 am

Matthews Hall, Room 210
http://www.psych.purdue.edu/~gfrancis/Classes/PSY200/indexF20.html
Prof. Greg Francis
Four great mysteries

Introduction
● Humans face four great mysteries about the
universe
PSY 200 ● 1) Why is there something instead of nothing?
w This is the domain of physics
Greg Francis w Most of us are not going to understand the ideas
Lecture 01
Four great mysteries.
Purdue University Purdue University
Four great mysteries Four great mysteries

● Humans face four great mysteries about the ● Humans face four great
mysteries about the
universe
universe
● 2) How did life form? ● 3) Why is there so much
w This question is addressed at the boundary between diversity of life?
chemistry and biology w This is the domain of
biology
w Evolution and natural
selection answer this
question

● Humans face four great mysteries about the universe
Topics
● 4) What is the basis of human intelligence and
consciousness? ● Discuss a sample of issues in cognitive
w Cognitive psychology and neuroscience psychology / cognitive neuroscience
w Far from a complete answer
w Lots of issues to discuss
● Try to relate cognitive psychology to
stories you may have heard in the popular
press
● Identify how the topics can help you to be
a better person
PSY 200: Intro. to Cognitive Psychology

1
Prof. Greg Francis
Topics Textbook
● There is no textbook
● For example
● Lecture notes are used instead
w What s the deal with left and right brains?
w Why does everyone love Prozac? ● If you want a book, borrow from a past
class
w Why telephone operators seem rude.
w Why there is a gate at the first floor stairway ● There are optional readings in the
in the Psychology building. syllabus
w What to do if you are drunk while studying for w Not for every subject
an exam.
w What is the plural of walkman?
Lecture notes Lecture vodcast

● Vodcast of the lectures will be provided
Downloadable from the
Prof. Greg Francis
●
● To me, these are a poor substitute for attending
class web page
Introduction
lecture
●  Humans face four great mysteries about the
universe
PSY 200 ●  1) Why is there something instead of nothing?
!  This is the domain of physics
Greg Francis !  Most of us are not going to understand the ideas
Lecture 01
w Adobe Acrobat (pdf) Four great mysteries.

● Vodcasts will be posted on Blackboard as they
format Four great mysteries Four great mysteries
become available
●  Humans face four great mysteries about the ●  Humans face four great
w Sometimes takes a few days

mysteries about the
universe
universe
2) How did life form?
w Reduced form (6 to a
● 
●  3) Why is there so much
!  This question is addressed at the boundary between diversity of life?
chemistry and biology !  This is the domain of
biology
!  Evolution and natural
selection answer this
w Sound may not be very good

question
page) Purdue University Purdue University

Topics
●  Humans face four great mysteries about the universe
●  4) What is the basis of human intelligence and
consciousness? ●  Discuss a sample of issues in cognitive
!  Cognitive psychology and neuroscience psychology / cognitive neuroscience
!  Far from a complete answer
!  Lots of issues to discuss ●  Try to relate cognitive psychology to
stories you may have heard in the popular
press
●  Identify how the topics can help you to be
a better person

1
Attendance Course web page

● When you enter the room each day, be sure to
get a scantron
● Syllabus on the web
w fill it out with your Purdue information
● http://www.psych.purdue.edu/~gfrancis/Classes/PSY200/indexS20.html
w Sign it
w Answer the first question with anything
w updates to the syllabus
● Hand in the scantron at the end of class w Links to labs
w It will function as a check on your attendance w Links to writing assignments
● You can miss up to 6 lectures without any w Study guides for the exams
penalty
w Beyond that you lose a proportion of points
w (old) practice exams
w 5% of your class grade w Links to optional readings
● Scantrons are only available for the first 15 w Grades will be posted after the first exam
minutes of class
● The course sparingly uses Blackboard

2
Prof. Greg Francis
Course outline Exams

● Neuroscience -- EXAM 1 (10%) ● Multiple choice (scantron)
● Perception, Attention & Memory – EXAM 2 (10%)
● Detailed study guides are already on the
● Memory & Mental representation -- Exam 3 (15%) class web site
● Language -- Exam 4 (15%)
● Beware the scheduling of the final exam!
● Reasoning
w The exam is during the final exam week
● Cumulative Final (15%)
w There are few excuses for changing the date
CogLab
CogLab
● Labs are listed on the syllabus
● Homework
● They must be completed by 8:00 am at the date
● You participate in classic experiments indicated in the syllabus
● Total lab grade contributes to 15% of your class w else you get no credit
grade. w Better to do it the night before
● Grade is based solely on completing the ● Since I wrote CogLab, you get access to the
experiment, not on the quality of the data experiments for free
w (a $50 value!)
● See handout for instructions on getting started
● Registration code is on a label on the instructions
● First lab is due at 8:00 am on Wednesday!
Writing assignments ● Straight scale Grading

● You need practice writing! w 98% -100% A+
w 93% - 97% A
● Four assignments, 3-5 pages of single spaced
w 90% - 92% A-
text.
w 88% - 89% B+
w If you struggle to fill 3 pages of text, you probably do w 83% - 87% B
not understand the assignment
w 80% - 82% B-
● Assignments are due in Blackboard w 78% - 79% C+
w 73% - 77% C
● First assignment is January 29
w 70% - 72% C-
w By the start of class (not one second later!) No extra credit
w 68% - 69% D+
● 15% of your class grade w 63% - 67% D
w 60% - 62% D-
w 0% - 59% F
● No rounding up: 82.99 is a B-

3
Prof. Greg Francis
Grading Grading
● Last semester’s grades (Spring 2019) ● Last semester’s grades (Spring 2019)
Frequency
A 10
1 00 0
.0 B 67
C 59
Final class grade

9 0.00
E1 E2 E3 E4 Final CogLab Writing Attendance Total 8 0.00

D 29
7 0.00
Average 72 69 73 81 71 84 69 91 75 6 0.00 F 22
Max 96 96 98 99 99 5 0.00
4 0.00
3 0.00
2 0.00
1 0.00
00
.0
00
.0 2 0.00 4 0.00 6 0.00 8 0.00 1 00 0
.0 1 20 0
.0
CogLab score
Grading Instructor office hours

● Monday, Wednesday, Friday, 2:00 – 3:00 pm
● Last semester’s grades (Spring 2019)
w Or by appointment
Frequency
A 10 w Psych 3186
1 00 0
.0
9 0.00
B 67 w Email: gfrancis@purdue.edu
C 59
Final class grade
8 0.00
7 0.00 D 29
6 0.00
F 22
5 0.00
4 0.00
3 0.00
2 0.00
1 0.00
00
.0
00
. 2 0.0 4 0.0 6 0.0 8 0.0 1 00 0
. 1 20 0
.
Attendance score
Attitude/Advice
Teaching assistants ● During lectures: turn off cell phones, don t read newspapers, don t play
games
● Questions are always welcome. I can adjust my lecturing pace accordingly
● Maria Kon and Dani Larranaga ● Print out the lectures and bring them to class. Take notes during class. Not
everything is on the slides.
● Grade writing assignments ● Everything we talk about in class is important
● Keep track of grades and attendance ● Work on the study guide every week, so the ideas/answers are fresh in your
mind.
● Have office hours ● This class is an introductory class, but that does not mean it is easy
w It’s like Introduction to Physics or Introduction to Chemistry
● May provide out-of-class study sessions w Almost every other subtopic in psychology depends on the ideas in cognitive
psychology
for exams w Everything is at least 10,000 times more complicated than what we discuss
● If you don t find a topic interesting, just wait a week

4
Prof. Greg Francis
Next time
● Cognitive neuroscience
● The brain
● The modularity hypothesis
● CogLab on Brain asymmetry due!
● What s the deal with left and right brains?
Purdue University

5
Prof. Greg Francis 1/6/20
Brain parts The brain
PSY 200 ● The source of cognition

(consider transplant!)
Greg Francis ● Weighs about 3 pounds

● Damage to some parts result
in immediate death or
Lecture 02 disability
● Damage to other parts
seems to have no effect!
● What brain parts are
What s the deal with left and important to cognition?
How do we discover the role
right brains? ●
of each brain part?
The brain The brain

Hypothalamus
● Fore-brain (appetite, thirst,
● Hind-brain temperature,
w cortex hormones)
w (brain
● Hind-brain stem)
w brain stem)
Thalamus
Limbic Lobe (sensory
(sexual behavior, gateway,
emotional behavior, except
memory) smell)
Cerebellum
Purdue University (muscle control, learning)
Purdue University
Fore-brain Fore-brain
● Cortex ● Cortex
● Similar to a thick, ● Similar to a thick,
crumpled crumpled
newspaper page newspaper page
● Grooves (fissures ● Grooves (fissures
or sulci) separate or sulci) separate
regions regions
PSY 200: Intro. to Cognitive Psychology 1

Fore-brain Fore-brain
● Cortex ● Cortex
● Similar to a thick, ● Similar to a thick,
crumpled newspaper crumpled
page
newspaper page
● Grooves (fissures or
● Grooves (fissures
sulci) separate regions
or sulci) separate
regions
Contralateral processing Contralateral processing

● Processing in the brain is done on the
● Neural fibers
opposite side of your organs from the eye
● Control of your right arm is from the left cross on way
to cortex
side of your brain
● Information from your left field of view
goes to the right side of your brain
Contralateral processing Cut corpus callosum

● Brain
hemispheres
● Behavior
are
changes very
connected by
a mass of little
neural fibers ● Subtle effects
called the
corpus
callosum
http://www.youtube.com/watch?v=ZMLzP1VCANo

Brain sides Brain sides

● If nut flashes on left ● Results led to further study and common belief that
screen w Left side: language, analytical, classification, Western rationalization
w subject cannot name it w Right side: art, music, recognition of faces and shapes, Eastern
mysticism
w subject can pick up nut
with left hand ● Vast oversimplification
w in a normal brain, both
● If nut flashes on right
sides are involved in
side many tasks
w subject can name it ● Results do support the idea
w subject cannot pick up that different parts of the
nut with left hand until brain are involved in
he says nut out loud different cognitive tasks
(modularity hypothesis)
1981 Nobel Prize for Roger Sperry!
CogLab CogLab
● In the CogLab + ● Your task was to +
experiment Brain judge whether the
asymmetry presented word was
● You stared at a “old” (seen on an
central fixation point + wood earlier trial) or “new” + wood
(not previously seen
● A word was
in this experiment)
presented to either
the left or right side
of fixation + Time + Time
CogLab CogLab
● Federmeier & Benjamin (2005) found better memory ● Other explanations than hemispheric
performance for words presented in the right visual field
specialization
● Words in the right visual field go to the left hemisphere
w Which is known to be specialized for language w Reading goes from left to right, from fixation to
right visual field
w Perceptual advantage to right visual field?
w Attentional advantage to right visual field?
● It is difficult to come up with an experiment
that isolates hemispheric specialization

CogLab Data CogLab Data

● Class (149 right-handed participants; 19 left-handed ● Global (20,233 right-handed participants; 2106 left-handed
participants) participants)
w Very tiny effect (in wrong direction) for right handers (probably ● Very tiny effect (but bigger for left-handers than right
just noise) handers – weird; probably noise)
Anatomy Occipital lobe

● The cortex contains large fissures that
separate five major areas ● Receives
w Limbic (already discussed) information from
w Occipital the eye
w Parietal w Most investigated
w Temporal area of the brain
w Frontal
● Each has distinct properties
Parietal lobe Parietal lobe
● Sensations of ● Primary sensory

pain, area
temperature, w sensitivity
touch, pressure involves
disproportionate
areas of the brain,
Primary relative to size of
sensory body part
area

Temporal lobe Frontal lobe

Hearing
Largest part of
●
●
w speech (left) cortex

w music (right)
w planning
● Memory and w prediction
attention w motor area
w visual recognition w speech area
Broadman areas Area function

● Divide lobes into areas ● Can partly identify
● e.g. Broadman function by looking at
w Area 1, Area 12, … Area 200
nerves coming in and
w Some special names: V1, V2, V3,…
out of area
● Pathways through Color
areas seem to be
involved in different
kinds of cognitive tasks
Brain layers
Conclusions
● There is order and
function even within ● Lots of research in this area
an area ● New brain regions are being mapped out
● The cortex is a daily with ever increasing resolution
sheet of neurons
● Cognitive neuroscience relies strongly on
● In its thickness are
the modularity hypothesis
6 layers of neurons
w numbered 1-6 ● Putting everything together is very difficult
w sometimes include
subdivisions (4a, 4b,
4ca, 4cb, …)

Next time
● Brain scans
● EEG recordings
● MRI scans
● Functional MRI
● How to study the brain without killing someone.
Purdue University

Brain scans Scanning

● Technology provides insight into brain processes
PSY 200 w EEG recordings
w MRI
Greg Francis w Functional MRI
● Non-invasive
Lecture 03 ● Maps of brain activity
● The goal is to relate brain events to cognitive
events
How to study the brain without
killing someone.
Resolution
Electroencephalogram
● For almost every technique we have to worry about its
ability to discriminate differences in
w Space: which ● EEG
place is active?
w Time: when ● The brain
does something
happen? produces
● Finer resolution electrical activity
is usually better
w But can be ● Put electrodes
difficult to deal
with so much
on the head
data
Brain maps
EEG ● You can analyze the EEG signals in many different ways
● Watch the electrical current change through time while ● Compare the signal strength for different situations
reading sentences (averaged across many trials) ● Ayahuasca is a Brazilian psychoactive tea
w Good temporal resolution Semantic
w Kutas & Hillyard (1980) anomaly

Magnetic Resonance Imaging

Spatial resolution ● Magnetic field
forces protons in
● Poor spatial your body to line
up
resolution
w pulses of radio
● You never really into field
bounces
know which part of protons around
the brain is making w as they return
to normal
the current
position, they
w Lots of work to emit a signal
improve that can be
decoded into a
map
Magnetic Resonance Imaging Magnetic Resonance Imaging

● MRI Scans: Like an x-ray machine, but ● MRI Scans: Like an x-ray machine, but can look at soft
tissue (like ankles, my brain,..)
can look at soft tissue (like lungs, heart,..) w Very good spatial resolution
w Very good spatial resolution w millimeters
MRI Scans MRI Scans

● MRI provides ● Can identify
a slice at a anatomical
time differences between
● Take multiple brains
slices to build ● Alcoholic has larger
up full image ventricles and thinner
corpus callosum
● Note, comparing
across brains is a bit
Nobel prize
tricky!
winning work!
w Everyone s brain is a
bit different

MRI Scans MRI Scans

● Non-invasive, no side effects
● 3-D maps ● Allows early detection of brain disease,
tumors,…
● Normal ● Fantastic spatial resolution
● But…
● Ataxia: w it only shows structure
w loss of motor w no way to know what a brain area does

control
Functional MRI Functional MRI

● Just like MRI, but with a new analysis
● Color maps show
w MRI differentiates between different types of tissue (cell
strongest
types)
responses
w Functional MRI differentiates between active and
inactive neurons: concentration of oxygen ● E.g., fMRI scan of a
w The measurement is called the blood oxygen level woman after a stroke
dependent (BOLD)
w Blue/green:
» It roughly tracks the flow of blood in the brain
normal blood flow
» More active neurons recruit more blood
w Red/black:
abnormal blood
flow
Functional MRI Scanning

● Consider this fMRI scan
● Very good spatial ● It shows regional cerebral

blood volume (rCBV)
resolution
● You cannot tell how/if
w millimeters
different regions are involved
● Pretty good in different activities
temporal resolution w Breathing
w Seconds w Digestion
w Thinking about exams
w (Silva, 2002)
w …..

Differences A simple experiment

● We have to look for differences in activity ● Suppose you run an fMRI experiment where a person alternates
between seeing a blank screen and a face
● Alzheimer s patients have reduced brain activity
● You take multiple fMRI scans with half recording brain activity during
the blank and half recording brain activity during the face
● Add them up pixel by pixel for each condition
Viewing blank Viewing face
Subtraction method Reporting

● What is usually
● Subtract the fMRI signals produced by one condition from the fMRI
reported is just the
signals produced by another condition
difference map
● The difference map indicates those brain regions that are involved in
● Colors mark places
the different cognitive tasks
in the brain that are
● It requires a sophisticated statistical analysis to avoid false positives!
statistically different
Viewing blank Viewing face Difference map between conditions
● Czisch et al. (2009)
for rare tones vs.
frequent tones
● The map would be
different if it
compared rare tones
versus speech
Functional MRI
Functional MRI
● Color maps show ● When moving a

strongest
pointer to a target
responses
box compared to no
● e.g., during a task
that requires covert movement
spatial attention w activity in areas
compared to one that involved in vision,
does not require planning, and motor
attention control

Connectome Connectome
● You can use similar technology (diffusion ● Gives an
spectral imaging) to focus on particular types of anatomical
cellular material map of how
w E.g., identify axons (discussed later) that connect information
brain cells can travel
● Gives an
anatomical
map of how
information
can travel
Limitations Common misconception

● Brain scans do not really tell us how the brain ● Brain scans demonstrate a physiological basis to
works things that were thought to be emotionally or
cognitively based
w the scans just tell us approximately where in the brain
something occurs w e.g., MRI scans
w sometimes it can tell approximately when of stutters
w in fact, all
● Even trying to find the place may be problematic
behavioral traits
w Lots of cognitive abilities involve many different areas are physiologically
of the brain based
● Most theories of cognition are derived from
experimental psychology
w Brain studies explore how to implement the theories
Conclusions Next time

● Lots of research in this area ● How do we use brain scans to study
● Technology is improving in many ways cognition?
● There are many other types of scanning ● How good are the scans?
technologies
w Computerized Axial Tomography (CAT) ● What is really being measured?
w Diffusion tensor imaging (DTI)
w Single Photon Emission Computed Tomography
(SPECT) ● How to read someone s mind.
w Near Infrared Spectroscopic Imaging (NIRSI)
w Magnetoencephalography (MEG)
w Positron Emission Tomography (PET)

Brain scans Scanning

● Brain scanning techniques like fMRI
PSY 200 provide spatial and temporal patterns of
activity across the brain
Greg Francis
● We want to analyze those patterns to
Lecture 04 discover how the brain works
How to read someone s mind.
fMRI fMRI
● Passive listening vs. active listening Passive
● The colors show the
w Vannest et al. (2009) listening
difference maps
● Twenty children (ages 11-13) complete three relative to listening to
the tones
tasks
● Common activity
w Passive listening: hear a female speaker tell a 30-
(breathing, digestion,
second story
hearing machine
w Active response: hear the same speaker tell a story in
noise,…) is
5 second segments of two sentences. Scanning Active
subtracted out
occurred after the sentences (silence). Answer response
questions ● The colors are not
brain activity!
w Random tones: no task, just listen
fMRI fMRI
● More signals and Passive ● Does more signal for the Passive
listening listening
different patterns for active response mean active
active listening response listening is better
compared to passive than passive listening?
listening ● Tested children on
● (Could it be comprehension of stories
otherwise?) w PL: 75.1% correct, SD=12.7
w AR: 79.1% correct, SD=9.1
Active Active
response ● No real difference in response
comprehension

Sensory substitution Sensory substitution

● Some scientists look for replacements to lost perception ● People can use a TDU to discriminate shapes
● For example, there is a tongue display unit that attempts to present spatial w Kaczmarek, Bach-y-Rita & Tyler (1998)
information for blind people w Link to video on class web page
Sensory substitution Sensory substitution

● So what happens in the brain? ● fMRI differences
● Are there responses from areas typically involved in shape suggest that using the
perception? Or in areas related to touch on the tongue? (or both or TDU involves areas of
neither?) motor cortex
● Not areas that are
traditionally for visual
perception
● This is the kind of

question that can best
be answered with brain
scan technology
Walking Walking
● There is no portable
MRI machine
● But scientists are
creative about how to
use it to study a wide
variety of activities
● Volta et al. (2015)
studied walking by ● Can compare walking “indoors” versus “outdoors”
having participants ● Execution (actually “walk”) versus observation (not “walk”)
“walk” on a cylinder
● Complicated controls
outside the MRI
w Press feet against cylinder
machine
w Still (non-moving) image
w Gray image (no picture)

Walking Reading minds

● Long-term, the goal of fMRI research is to be able to analyze a brain scan
and identify what a person is thinking
● There are several attempts to do this (Haynes et al., 2006)
w Adding or subtracting numbers
● Top: Executing or observing walking versus gray image

w Interesting that they involve some similar brain regions
● Bottom: Hallway walking versus open field walking
Reading minds Mind reading

● Based on the choice at the end, we can deduce whether the subject chose ● It is possible to build a recognition system that distinguishes (with
addition or subtraction for that trial 71% accuracy) the brain patterns for addition and subtraction
● Make an fMRI scan during the selection process w Depends on the place in the brain
w Whether to add or subtract numbers w Different places for intention and execution
● You can read the mind of these subjects!
Scan
now
Or scan
now
Thought reconstruction Thought reconstruction

● Another research group analyzed fMRI responses to ● Performance depends on where the signals come from
reproduce a shown image ● Fewer errors for lower brain areas
w Where do you stop?, the retina?, the lens of the eye?

Thought reconstruction Mind reading limits

● fMRI: If subjects decide to multiply numbers, a system trained to
● These kinds of studies are mostly a demonstration of distinguish between subtraction and addition is clueless
technology
● Thought reconstruction: As the number of possible images to be
w we already know the brain represents visual information! shown increases, it becomes harder to reconstruct the shown
● Before the study was run, we knew that there were image
differences in the brain when we see different images ● In general, brain scans provide a very limited form of mind reading
w The percept is the brain’s behavior, so there must be w People do better than this every day by watching people behave
(posture, eyes, skin tone)
differences!
● These kinds of studies tell us that the
neurophysiological differences between cognitive
events can be measured by these brain scanning
technologies
w Failure would only indicate limits of the technology
Problems / limitations with

Statistics
scanning
● So much data that it is difficult to know what to do with everything ● It is easy to do the statistics incorrectly (it has taken a while for the
w Statistical analysis is complicated field to sort this out)
w In a small brain scan, you may have 64 x 64 voxels x 10 slices ● Bennett et al. (2010) ran a study where the subject was shown a
» =40,960 voxels overall series of photographs depicting people in social situations with a
specified emotional valence, either socially inclusive or socially
w Some of those voxels will give different responses just by chance
exclusive. The subject was asked to determine which emotion the
● Difficult to compare across subjects
individual in the photo must have been experiencing.
w Slightly different anatomy
● fMRI contrasts were computed between the scans for the two
● Blurring of images is difficult to deal with (subjects move in the
types of emotional valence
scanner)
w Sometimes blur together brain areas, across a fissure, that are
actually far apart on surface of cortex
● Some cognitive events are faster than the technology can track
● Can only measure the brain, cannot manipulate it
Statistics Statistics
● There is a significant difference in fMRI activity for some regions of ● The subject in this study was a mature Atlantic
the brain
salmon (sex unknown)
w Medial brain cavity and upper spinal column
● The active regions identified by the fMRI are due to
chance
● Even with purely random noise, there will be some
statistically significant findings
w The brain has lots of random noise
● These problems can be reduced but never entirely

eliminated
w They are common to many areas of psychology, not just brain
scans

Neurons A neuron
● The brain cells that are responsible for cognition ● Dendrite ● Axon
w input w output
are neurons Soma
Myelin sheath
●
●
w integrate
w insulate
Myelin A neuron
● Diffusion Spectral Imaging detects properties of ● There are many different types of neurons
the myelin sheath (“white matter”) ● We will describe only the most common characteristics
● Allows imaging of human brain connectome
How many neurons? Conclusions

● Estimates of 10^11 neurons in the human brain ● Brain scans usually look at differences in brain
activity
w 100,000,000,000; one hundred billion
● Lots of technical (and ethical) issues
w estimates of 100,000 per cubic millimeter
w (about the resolution of functional MRI) ● The goal is to be able to look at a map (or movie)
and be able to read someone s mind
● Millions are active at any given time
● That is many years off
w questionable if it can even be done with these methods
No. 2 1 mm
alone
https://www.youtube.com/watch?v=2qTuZlMvFgY
(time 2:55) Purdue University Purdue University

Next time
● What is the neural activity that produces
brain scans?
● How do neurons transmit information to
other neurons?
● Why does (nearly) everyone like Prozac?
Purdue University

Neurons and neurotransmitters Neurons

● The brain cells that are responsible for cognition
PSY 200 are neurons
Greg Francis
Lecture 05
Why does (nearly) everyone

love Prozac?
Input at dendrites
Input / output
Changes the cell membrane
-70 millivolts ●
● Electrical signal potential

electrode resting level
w Established by the w which causes further
relative amount of changes in the cell s
charged ions inside NA+ chemistry
versus outside the cell w which causes further
membrane changes in the membrane
NA+ potential
● Inputs change the
Strong enough input crosses
resting potential of the
cell
K+ ●
a threshold and the cell fires

● Output identifies when K+ w action potential
the cell potential has

increased a lot
http://www.mind.ilstu.edu/curriculum/neurons_intro/action_potential_1.0.swf
A neuron A neuron
● An action potential ● The action potential then affects the membranes of
generated at the other cells dendrites
soma travels
down the axon to
the terminals

Output Output
● Myelin is like insulation for the ● The output of a neuron is either excitatory or inhibitory on the
cell s axon
other neuron it reaches
w it insures that the signal
generated by the action ● Excitatory: when our neuron sends an output, the receiving
potential is strong neuron is more likely to produce an action potential
w Jumps electrically rather than ● Inhibitory: when our neuron sends an output, the receiving
the normal chemical exchanges
neuron is less likely to produce an action potential
● In multiple sclerosis the body s
immune system attacks myelin
w physical problems (paralysis)
w cognitive problems (memory,
reasoning, judgement)
w cause unknown (300,000
people)
Networks Epilepsy
● Cognitive behavior is related to ● Disease of central nervous system
groups of neurons working w causes mostly unknown
together
● Seizures
● Include excitation and inhibition
w bursts of electrical activity travelling through networks in
w more later
the brain
w brain activity is out of control
w epileptic fits
● Isolated seizures also occur due to high fever,

lack of oxygen, or head injury
Epilepsy Epilepsy
● One theory (but not yet proven) is that epilepsy
● EEG recordings patients inhibitory cells are not working properly
are often used to ● Excitatory cells activate everything until they
diagnose epilepsy exhaust themselves
● Many different
types of epilepsy,
with different EEG
patterns

Epilepsy A balanced brain

● The brain is a dynamic system
● Treatment generally involves w at multiple levels
w drugs, diet, avoiding stress, ● Neurons

keeping regular schedule
w balance between forces inside and outside of cell membrane
● In extreme cases surgery allows for action potentials
prevents seizures from
● Networks
spreading throughout the
w balance between excitation and inhibition
brain
● Without these balances you do not think
● Contrast with ideas about using “more” of your brain
Neural connections
Molecular structure
● Axon --> dendrites ● Molecules have a particular three-

dimensional shape
water benzene
http://www.mind.ilstu.edu/curriculum/neurons_intro/neurotransmission_classic_3.0.swf
Molecular structure Molecular structure

● Neurotransmitters are just molecules
● Different molecules have different
● At least 50 different neurotransmitters
shapes
aspirin w dopamine, norepinephrine, serotonin,
acetylcholine, glutamate, gamma-
TNT aminobutyric acid (GABA)
● All with different shapes!

Receptor Receptor
● Very large
molecules called ● When it accepts a neurotransmitter, it
proteins
starts a chain reaction of events
● Similar to a filter
w accepts some
w physical, chemical, electrical
neuro- w locally changes the cell membrane
transmitters
» depolarization (excitation)
w rejects others
» hyperpolarization (inhibition)
Neurotransmitters Tourette s syndrome

● Inherited (~200,000 in US)
● Different neurotransmitters are associated ● Behavior
w Swearing
with different properties
w Tics
w actually neurotransmitter and receptor pairs » Simple: eye blinking, facial grimacing, sniffing
» Complex: coordinated patterns, sniffing objects,
● neural jumping, twisting
● Too much dopamine
● cognitive
● Treated with Haldol (among others)
● behavioral w blocks dopamine
https://www.youtube.com/watch?v=NlFpkruxrCI
(6:15 in)
Parkinson s Drugs
● Lack of dopamine ● Interact with neurotransmitters in lots of ways, for
w Many different causes example
w Replace: accepted by receptor and with similar effect
w In extreme cases, patients are frozen
w Production: increase or decrease
● Give patients large doses of L-DOPA w Reuptake: knock out enzymes that remove
neurotransmitter from receptor, neurotransmitter has a
w a precursor of dopamine
bigger effect
w sometimes solves the problem w Blocking: enter receptor but does not trigger reaction,
partly closes receptor protein so neurotransmitter
w lots of side effects
cannot enter
● Awakenings, by Oliver Sacks
https://www.youtube.com/watch?v=koL0PWCJ4lo

Prozac Other drugs

● Some forms of depression ● Amphetamines: release of norepinephrin or dopamine
seem to be related to limits in ● LSD: resembles serotonin
the use of the
neurotransmitter serotonin ● Phenothiazine drugs: block dopamine
● Prozac is a selective ● Curare: blocks acetylcholine

serotonin reuptake inhibitor
● Cocaine: prolongs effects of dopamine
(SSRI)
w it keeps serotonin bound to a ● Morphine: resembles a small set of neurotransmitters called
receptor for longer than endorphin peptides (modulate pain perception)
usual, thereby increasing its
effect ● Tetrahydrocannabinol (active ingredient in marijuana): binds to
● Prozac is one of the most some neuroreceptors, but it s not clear what it does
widely prescribed drugs in the
world!

● Neural action potentials
Shape of proteins
Neural sensitivity
●
●
● Specific use of neurotransmitters for certain behaviors
● Current work on identification of role of neurotransmitters ● Neural codes
Lots of money to be made
Receptive fields
●
●
● Lots more complicated than what we ve seen here
● CogLab on Blind spot due!
● How do you recognize your grandmother?

Receptive fields Action potential

With enough excitatory input, a cell produces an
PSY 200 ●
action potential that sends a signal down its axon

to other cells
Greg Francis
w But a single action potential has little effect
Lecture 06 ● If the input stays present, the cell produces

another, and another,…
w A rapid series of action potentials can influence other
cells
How do you recognize your ● The number of action potentials in a certain
grandmother? length of time determines the firing rate of the
cell
Firing rate Specificity

● Two key questions in cognitive neuroscience are
● 8 spikes (action potentials) during 300 ms
w What stimulus (or stimuli) makes a given cell fire at a strong
w Firing rate of 27 Hz (27 spikes per second)
rate?
» something red?
» a pen?
» your grandmother?
w What does it mean when a given cell fires strongly?
» You are thinking of something?
» Seeing something?
» Remembering something?
Receptive field Receptive field

● Receptive fields are very useful for studies of spatial perception
● The set of stimuli that reliably changes
● Touch involves sensitivity to pressure on skin
a cell s firing rate. ● The loops indicate the
● A stimulus could excite the cell regions where a single

neuron responds to
w above normal firing rate pressure
● Or inhibit the cell

w below normal firing rate

Receptive field Receptors

● Receptive fields are very useful for studies of visual perception
● The back of the eye
● Vision involves spatial patterns of light
contains tightly
Light
packed sensors
called rods and
cones that detect
light at a particular
location
Receptors Receptors
A receptor has a simple receptive field
● Light sensors (rods and cones) respond to light ●
w it responds to light of the right wavelength (color) and the right

at a particular location in the back of the eye
position
w produces a neural response
Light
Blind spot
Blind spot ● In CogLab you mapped your blind spot
● my
● Where nerves leave data
the back of the eye, looks
like
there are no light this
receptors
w light that hits this spot is
not visible

Blind spot Network

● Here s the average data for the whole class
● 108 participants ● Light receptors do not just pass information
to the brain
● Neurons are hooked together in an inhibitory
way
Light
Evidence of inhibition Evidence of inhibition
● Stimulation of a center region alone ● Adding light to a surrounding region

gives a strong response can reduce the response
40 40
Light 35
Light 35
Firing rate
Firing rate
30 30
25 25
20 20
15 15
10 10
5 5
0 0
Time Time
Receptive field Spatial responses

● The spatial
● The receptive field of this cell pattern of
includes any place on the retina excitation
(center) and
where light excites the cell and any
inhibition
place where light inhibits the cell
-+
(surround)
● On-center, off-surround means the cell
is sensitive to
the location of
a small spot of
light

Spatial responses Simple cells

● You have many ● On-center, off-surround cells send action potentials to
different such simple cells in parts of visual cortex, which have oriented
cells with receptive fields
receptive fields
that are
centered at
different
locations
● They respond
differently to an
edge
Simple cells Simple cells

● On-center, off-surround cells send action potentials to ● Simple cells come in a variety of types,
simple cells in parts of visual cortex, which have oriented but all are sensitive to bars or edges of a
receptive fields
preferred orientation at a particular
location
Information processing
Information processing
● For simple cells, an image like this
● For simple cells, an image like this
w is coded something like this
Strong
responses
at edges!

Complex cells Complex cells
● Many simple cells feed ● Many simple cells feed into a
into a complex cell, complex cell, which is

insensitive to direction of
which is insensitive to
contrast and responds to an
direction of contrast and oriented bar in many different
responds to an oriented places
bar in many different w Often these cells are also
sensitive to directions of
places
motion
Receptive field hierarchy Higher order cells

● Receptive fields inherit some properties from ● Complex cells feed into hypercomplex cells, which
lower-level cells are sensitive to some types of curves and visual
● But they also gain new selectivity by interacting forms
with each other (and across levels) ● Receptive fields seem to get ever more complex
● What does this mean?
● Is there a grandmother cell?
Receptive field Receptive field

● In the inferior temporal cortex of monkeys ● In the inferior temporal cortex of monkeys
w Desmione et al. (1984) w Desmione et al. (1984)
● some cells ● Other cells

appear to have appear to have
receptive fields receptive fields
that respond to that respond to
monkey faces, in hands
profile

Grandmother cells? Conclusions

● It could be that a single cell has a receptive field ● Receptive fields
selectively tuned to respond to the image of your
w any stimulus that affects a cell s firing rate
grandmother
w excitatory
w but it is unlikely
w not enough cells w inhibitory
w cell death ● Very useful for studies of the visual
● Receptive fields become less useful as we search nervous system
for neural representations of non-sensory concepts
● Lots of issues left unresolved
w What is the receptive field of a neuron that codes love
or trust ?
Next time
● Networks of neurons
● Connections between cells
● Feedback – resonance
● Seeing things that are not there.
Purdue University

Neural networks Receptive field

As we saw last time, a
PSY 200
●
cell s receptive field

depends to a large extent
Greg Francis on the receptive fields of
other cells
Lecture 07 w (e.g., complex cells depend
on simple cells)
● Today we look at some

Seeing something that is not issues involved in
networks of neurons
there.
https://www.youtube.com/watch?v=2qTuZlMvFgY
Feedback Resonance hypothesis

● Initially cell firing rates may vary a lot
● Cell 1 can affect cell 2, which ● In some networks cell firing rates stop
can affect cell 3, which can 2 changing much (unless outside input
affect cell 1 again, which… changes)
3 1
● What happens to cell firing ● The remaining active cells are those that
rates? support each others activities through
w high firing rate==> active excitation: resonance
w low firing rate ==> inactive ● Inactive cells are inhibited by the active cells
● Mental awareness ==> resonance
A simple model A simple model

● A cell s activation is on or off (one or zero) ● Neural connections
(synapses) are described as
● Cell connections (synapses or weights) are
reciprocal weights on the links w21
2
between cells
● Cells update activations one at a time 1
● Input to a cell is the summed 3
● Cell activations are calculated with the rule w41
multiplication of sending
w15
"$1 if ∑ wij a j > 0 activation and weight w34
ai = # ● Reciprocal weights have 4 5

$%0 if ∑ wij a j ≤ 0 wij = wij

Demonstration Feedback
● Feedback in networks can act to clean up
● Cell activities do settle down eventually
noisy sensory information to make it
● Final pattern of activities satisfies consistent with what our systems expect
constraints of the network connections ● In a very real way, what we see, hear, taste,
smell, touch, and think, is biased by our
● Error correction capabilities network s expectation
● Can tolerate the loss of some cells ● A network s expectation is established by its
connection weights
● Emergent properties of the network
w excitation -- inhibition
w no single cell has these properties
Seeing things that are not Seeing things that are not
there there
● Do you see a square in front of the ● Neurons in area V2 of your brain

pac men? create the illusory contours
Feedback Conclusions
● Similar situation with the illusory ● Networks of neurons have properties
circle seen here different from single cells
w emergent properties
w stable activities
w multiple constraints
w tolerance to errors and cell loss
● Structure of connections (synapses)
determines the final pattern of responses

Next time
● How networks learn
● Changing connections
● Learning rules
● Self-organization
● CogLab due for Implicit Learning
● A problem with virtual reality.

Purdue University

Neural learning Networks
PSY 200 ● As we saw last time, a

network of neurons can
Greg Francis have very complicated 2
behavior 1
3
Lecture 08 ● The behavior depends on
the connections between
cells
4 5
A problem with virtual reality. How do those connections
●
w45
get established?
Hebb s rule A simple model

● A cell s activation is on or off (one or zero)
● If two neurons are active
● Cell connections (weights) are reciprocal
simultaneously, then they
● Cells update activations one at a time
strengthen the connection
between them ● Cell activations are calculated with the rule
● Signals from the
environment change the
properties of the network
"$1 if ∑ wij a j > 0
ai = #
$%0 if ∑ wij a j ≤ 0
Simplified learning Self-organization

● A network of this type does not need an intelligence to set
● Initially, all connections are zero the connection weights
w wij = 0 ● The network self-organizes in response to stimulation
● Hebb s rule ● It can remember things it has previously experienced
w cells that are simultaneously active develop ● It can interpret new information on the basis of things it has
positive weights (excitation) previously learned
w an active cell develops negative weights with
inactive cells (inhibition)
● Demonstration

Deep learning
Learning
● Google used one version of a neural network to analyze
10 million YouTube stills ● This may not be the same type of learning you do when you
study for school
w 1000 computers (16,000 cpu’s) over 3 days
w but it is important just the same
● The network self-organized to identify common patterns ● Consider implicit learning
w Cats, faces, “tool-like objects oriented at 30 degrees” w A long sequence of trials,
where you press a key to
indicate the appearance
of a dot at a corresponding
location
Class data (~80 in

Implicit Learning each group) Learning
Global data
(~13,000 in each
● We are interested in how fast you respond to the dot group) ● Lots of learning happens that you do not
Two groups of subjects:
●
w Random: each sequence is random

notice
w Pattern: each sequence is the
same (but so long that people
● Consider the length of your arm
typically do not notice)
w to catch and throw objects your brain must know
● Subjects in the pattern
condition are faster
exactly your arm s length
(they have partly learned w but the length of your arm changes as you age!
the sequence, and generate
faster responses to
» And depends on unknown environmental
expected locations)
factors
● More generally, networks in your brain can learn information about your
environment without you being aware that something is being learned
Hand-eye coordination Coordination and learning
● We do not know the exact nature of the

vision arm network involved in this coordination
info info w but we know it continually modifies part of itself to
match up with the current situation
● This is actually a good design feature,

because the brain cannot know in advance
every detail of the eye-hand system
vision arm
info info

Virtual reality Cameras
● Using computer graphics to convince the ● Enhance visual perception

body it is someplace other than it really is ● Night vision for helicopter
● Useful for pilots
w architects, designers
w surgeons, pilots
w entertainment
Cameras Problem
● Enhance visual perception
● The network coordinating
w MRI overlaid on actual image
eye-hand systems, adjusts
of brain for surgeon
» highlight tumor itself
w Avoid other brain regions ● Extended use of the
» faster computer cameras makes
the user adapt so his eyes
are where the cameras are!
Problem Other adaptations
● After taking the cameras off, ● Inverted prisms

it takes some time to adapt
● Fortunately, the adaptations return to
back
normal pretty quickly
● Eye-hand coordination is off
● Kind of like the feeling you get after
● Could be a problem for
surgeons and pilots! roller-skating


● Learning in neural networks ● Review for Exam 1
w changing connections
w relatively simple rules
● Then
● Much of our perceptual and motor
● Neural networks for visual perception
behavior is based upon this type of
w brightness
continuous learning
w color
● It s not clear if more cognitive learning is
w form
similar
● Why we see color afterimages.

Visual perception Visual perception

● Light enters eye
PSY 200
● Signals sent to area V1 in cortex
Greg Francis ● Neural networks tuned to
w brightness
Lecture 09 w color
w form
w motion
w texture
Why you see color
w depth...
afterimages.
Seeing Measuring perception

● It’s kind of like a camera, but what we see is not just an image ● CogLab uses the method of constant stimuli
In particular, we do not see the world the way it “really” is
●
● Judge which of a pair of lines is longer
● When we notice this discrepancies, we call them illusions
● Vary the length
● Muller-Lyer illusion is one example
of the line
● How big is the illusion? How can you measure a subjective
without wings
experience?
● Repeat many
times
Psychometric function Neurophysiology

● Plot proportion of times line without wings is judged “longer” than the
line with wings ● How we see things is largely
● 50% is point of subjective equality (around 114 pixels) determined by the properties of
● A key property is that the perceptual experience is described in receptive fields
terms of physical units!
w on-center, off-surround
Class data w simple cells
(168 observers)
w complex cells
Global data ● And by network interactions among
(~31,000 observers)
cells

On-center, off-
On-center, off-surround
surround
● Characteristics of cell receptive ● Characteristics of cell receptive
fields force additional properties fields force additional properties
of the visual system of the visual system
w center-surround cells tend to not w center-surround cells tend to not
respond well to homogeneous respond well to homogeneous
light that covers both excitatory light that covers both excitatory
and inhibitory parts and inhibitory parts
w => percepts of the middle of an w => percepts of the middle of an
object is derived from the edges object is derived from the edges
Brightness contrast
Brightness contrast
● Two receptive fields inside the middle square
receive the same excitatory and inhibitory signals
● Edge responses are influenced by the surrounding
w Little response
light
w both center squares have the same light intensity
Brightness contrast Brightness contrast

● Receptive fields on the corner ● Thus, the visual system computes brightness as
something like local contrast
w Receive the same excitation at the center
w It s a property of the center-surround cells
w differ in the amount of inhibition in the surround w Our percept of brightness is determined by the responses
of cells at contrast edges
less w As a result, things that have equal physical intensities can

inhibition look dramatically different (next slide)
more
inhibition

Hermann grid Hermann grid
● Seems related to on-center,

off-surround cells
● Cells at intersections receive
more inhibition than cells at
single roads
● How do we explain the other

version?

Filling-in Filling-in
● We see color and brightness inside objects ● Brightness
w so edge information must fill-in to the interior information
● It sometimes gets things messed up spreads across
● Water color effect surfaces
w Craik-O Brien-
Cornsweet
effect
Filling-in Color system

● competition between
● Carefully fixate opposite colors
w red-green
the pink center
w blue-yellow
● If you keep your w black-white
eyes very still, it ● habituating gate
will disappear ● offset of one color leads to

rebound in other
● The yellow fills-in! ● Gated dipole circuit
Color system Color system

● competition between ● competition between
opposite colors opposite colors
w red-green w red-green
w blue-yellow w blue-yellow
w black-white w black-white
● Initial balance ● Extra input to green
w Neither color wins w Green wins competition
competition

Color system Color system

● competition between ● competition between
opposite colors opposite colors
w red-green w red-green
w blue-yellow w blue-yellow
w black-white w black-white
● Extra input to green ● Offset of green
w Fading of green signal w Rebound of red signal
Color system Color afterimage

opposite colors
w red-green
w blue-yellow
w black-white
● Recovery of green pathway
w Disappearance of rebound
w Return to initial state
Orientation competition Orientation competition

● competition between ● With additional input to
orthogonally tuned cells
horizontal pathway,
● habituating gate
horizontal channel wins
● offset of horizontal leads to
rebound in vertical
competition
● Same kind of gated dipole
circuit
w Principles of neural
computation!
● Baseline response
w Due to tonic input

Orientation competition Orientation competition

● With additional input to ● At offset of horizontal
horizontal pathway, input, the gated horizontal
horizontal channel wins signal is weaker than the
competition vertical signal
● But as horizontal gate ● A vertical rebound
habituates, horizontal appears
signal weakens
● It still wins the
competition, though
Oriented afterimages
Orientation competition
● As the horizontal gate ● Oriented
recovers, the system reset signals
returns to baseline and are also
the vertical after response implicated in
disappears an unusual
type of
afterimage

● Visual perception ● Visual dynamics
w brightness w Flicker
w Persistence
w color
w Motion perception
w form
● CogLab on Apparent motion due!
● Largely determined by the receptive fields
and network structure of visual circuits ● Why computer monitors work.
● Neurophysiology strongly determines
what we see!

Visual dynamics Flicker

A flashing light looks constant if it is
PSY 200
●
presented rapidly enough

Greg Francis ● The frequency of flashing at which
subjects do not detect flicker is called the
Lecture 10
Critical Flicker Frequency (CFF)
w about 50 Hertz (50 on-off cycles in a second)
Why computer monitors work. w 20 millisecond durations
CFF Phosphor
● Establishes minimum characteristics of ● The phosphor on a computer screen
electronic devices typically glows less than 10
milliseconds
● Lights flicker at 120 Hz
w ten thousandths of a second
w we spend a lot of time in darkness
● The gun reactivates the phosphor
● Computer (Cathode Ray Tube, CRT)
every 17 milliseconds
monitors and TV s flicker at around 60
● Thus, at any given time 1/3 of the
Hz
screen is dark
w better monitors go faster
w the percept persists in your head!
w Liquid Crystal Display (LCD) monitors work
differently
» Although some still flicker
Persistence Experiment
● What is the source of the persisting percept? ● Bowen, Pola & Matin (1973)
w Receptors in the eye? w subjects adjust duration of a blank
w Receptive fields? stimulus so onset of probe matched
perceived offset of the target
w Network interactions?
w Cognitive (memory)?
● Studies support network interactions +

Strange property Networks

● As the target s
● Feedback is
duration or
important
luminance
increases ● Produces a persisting
w its persistence
response
decreases ● Demonstration
Input from eyes
Explanation Explanation
● Francis, Grossberg & Mingolla (1994) ● Offset of input from
the eyes produces an
● Something has to reset the network
after response
w else it would keep persisting forever
w e.g., due to
● Two mechanisms competition from
w (1) new inputs inhibit old responses orthogonally tuned
cells
w (2) afterimages act as new inputs
Offset response
Note: afterimages get stronger as duration
●
●
inhibits persisting
and luminance increase!
response Input from eyes
● As the target s Explanation Wait a minute

duration or
luminance ● If visual percepts persist for over 100
increases milliseconds, why doesn t the world seem
w the afterimage blurry?
produced at target
offset increases in w There should be smears of objects as they
strength move or as we move
w so there is stronger
inhibition to break
● There must be something else preventing
the feedback such blurring
w so the persistence w masking
of the original
percept decreases

No mask No mask
● Write down all the letters you see ● Write down all the letters you see
R F
L P
M Q
No mask Masking demonstration
Masking demonstration Masking demonstration
Y S X X X X
D H X X X X
W F X X X X

Metacontrast
Significance
● Masks do not have to “write over” the target to have an
● The mask appeared after the target turned off
effect
● The target was presented all by itself for a
● In metacontrast masking the mask and target do not
brief period of time
overlap in space and (often) in time (CogLab)
● However, our visual system is unable to
develop a complete percept of a scene in a Target Mask
such a period of time
w Thus, the XXX mask interferes with processing of
the letters by shortening their persisting responses
w And prevents perceived blurring of changing scenes
Metacontrast Motion
● Correct identification of the narrow target is affected by ● In simple animals (like flies and frogs), we know how
the Stimulus Onset Asynchrony (SOA) of the target and motion is detected
mask w Demo on web page
● Reichardt detector
w Worse between 60-90 milliseconds
Detecting Motion Apparent Motion

● Humans have something like Reichardt motion detectors ● When objects move, there is a continuous path of
w at lots of different positions in the visual field motion
w sensitive to lots of different motion directions
● Reichardt motion detectors do not require
w sensitive to lots of different motion speeds
continuous motion
● Think of them as receptive fields that vary in both space and time
w and, continuous paths are not necessary for motion to be
● Many aspects of how we perceive motion follow from the properties seen
of Reichardt motion detectors
Time1
Time2

Apparent Motion Apparent Motion

● For just two stimuli, it does not depend on ● The percept of motion does depends on
w color
w stimulus duration
w shape
w interstimulus interval (50-200 msec)
w attention
w cognitive priming w distance
Apparent Motion Timing

● Weird grouping effects (not fully understood) ● For a Reichardt detector to
indicate motion, the signal from
the second area must follow the
signal from the first by just the ISI too short
right length of time
● Vary the Interstimulus Interval
(ISI) between the stimuli
ISI about right
w The time between offset of the
first stimulus and onset of the
second stimulus
ISI too long
Korte s laws Motion representation

● Apparent motion was highly studied at the beginning of the 20th ● One conclusion of studies of apparent motion
century
w Korte (1915) noted that to get good motion, you needed to increase is that motion is a fundamental percept
the ISI between the stimuli as the distance between them increased w It has an explicit representation in the visual system
● CogLab data ● You could imagine otherwise, we can be aware of
w (176 participants- from class, 8383 from global)
something moving without
actually seeing the movement
● Apparent motion is the source
of motion for all movies and
animation

Motion aftereffect After effect

● Just like for color and ● Motion can be adopted by non-moving stimuli
orientation, we might w http://www.michaelbach.de/ot/mot_adapt/index.html
expect an aftereffect of
motion
opposite directions of
motion
w Left-right
w Up-down
● habituating gate
● offset of one direction leads to
rebound in other

● Dynamic vision
● Attention
w flicker
w persistence ● What is attention?
w network dynamics
● What does it do?
w Masking
w Reichardt detectors
w Apparent motion
w Motion aftereffect
● How could you not see it?
● Also used to investigate other areas of cognition
and types of mental problems

Attention Attention
PSY 200 ● The world contains more information than

we can fully interpret or process all at once
Greg Francis
● The ability to deal with some stimuli and
Lecture 11 not others is attention
w not clear if there is an attentive system
w or if attention derives from other systems
How could you not see it?
Attention
Information processing ● Part of attention seems to be due to mental effort on
● Modern theories see cognition as information processing
your part
w much like a computer w attending a lecture
● Different systems have different capabilities, capacities, and speeds w ignoring whispering around you
● Necessarily, some information is ignored because it is not ● Part of attention seems a natural side effect of mental
processed effort
w ignoring the uhs and ums from a speaker
w ignoring the feel of clothes on your body
● Part of attention seems effortless

w a loud noise
Magic trick Magic trick

● Pick any card ● Now the computer will shuffle the cards and present
them again
● Memorize it
w really study the card carefully to be certain you have it
memorized

Magic trick Drawing attention

● Hey, the card you selected is missing ● Attention can be focused by meaningful stimuli
● How did I know which card you would select? ● Attention can be focused by environmental
w Try it on the web at characteristics
http://members.tripod.com/~andybauch/magic.html
Automatic attention Automatic attention

● Simon effect (Simon & Wolf, 1963) ● Location of the square is irrelevant, but it sometimes is
congruent with the response location (left-green; right-red)
● An irrelevant cue can affect response time to a while sometimes it is incongruent (left-red; right-green)
stimulus
● Task: respond as quickly as possible to identify
the color of the square
● The square is sometimes on the left and
sometimes on the right side of the screen
(irrelevant)
● You respond with a keypress on the left (green)
or on the right (red)
Automatic attention Automatic attention

● People are faster identifying color for congruent compared ● The Simon effect is, in some sense, a failure of attention
to incongruent conditions ● You want to ignore the location of the target square and
w 53 millisecond difference only attend the color
w But you cannot ignore the target location
CogLab global data ● CogLab has several labs that play on similar ideas
(~13,250 observers)
w Stroop effect (more next time)
w Spatial cueing
w Several labs related to memory and decision making have
similar properties

Automatic attention Human Factors

● These kinds of effects are ● Applied cognitive psychology
small (~50 ms), but they ● Among other things, design interfaces so that stimuli and responses
matter a lot
are compatible
● Consider the remote control
● Products “feel” better, are used as intended, and users make fewer
and on-screen channel guide errors
provided by my cable
● Really important in high stress situations
provider
w Aircraft cockpits, nuclear power plant control stations
● To move the “cursor” up on
● Really important in everyday (low stress) situations that are used a
the screen, I press the “+”
lot
button the right side
w Your phone
● The effect on the screen is to w Doors
go “up” one line, but that is to
a channel with a lower
number! Purdue University Purdue University
Drawing attention Drawing attention

● Raise your hand when you spot what
● In some situations, attention can be focused by changes in the two images
certain stimulus characteristics, especially
changes
w Flashes of light
w Movement
w Color
w Think of advertising signs
● We depend on these characteristics a lot

w Removing these cues can make simple tasks rather
difficult

● Raise your hand when you spot what ● Raise your hand when you spot what
changes in the two images changes in the two images


changes in the two images
● Suppose these cues were masked by other
changing stimuli
● You might not notice the change at all

● Raise your hand when you spot what ● Raise your hand when you spot what
changes in the two images changes in the two images
Attention
Drawing attention
● Masking the changes makes it difficult to identify
the changed parts of the image
changes in the two images
w Suggests that you do not actually “see” the entire
image with each presentation
● Attention seems to be necessary to detect
stimulus changes
● Explains how people can look but not see
w walking into doors
w driving into trains
w detecting changes on a radar screen
Purdue University w why magicians use flashes of light!

Purdue University

What does attention do?

● It is not clear, and it is probably different things for different situations
Attention illusion
● To many people, attended information feels “stronger”, so they think ● Sometimes attention
can change perceptual
neural representations must be stronger in the brain.
properties
● But if attention strengthens perceptual representations, we should lose
perceptual veridicality ● But then we have an
incorrect perception of
w We might expect what is schematized below
the properties of the
w But we normally do not experience this
visual scene
● Attention generally seems to strengthen information about a stimulus
● So it is difficult to
that is not perceptual
understand how
Stimulus Attend Attend green attention is helping
here
red
Tse, 2005
More demos Conclusions

● If time permits, here s some more demos
w http://viscog.beckman.uiuc.edu/djs_lab/demos.html ● Attention can have very powerful effects
● Field w help processing of focused on things
● Living room w can cause unawareness of unattended things
● Phone call ● Not precisely defined
● Lunch conversation (9 changes)
w characteristic of processing?
● Paris scene
w An extra system?
Next time
● Methods of studying attention
● What things influence attention
w Timing, features
● CogLabs on Attentional blink and Visual
search due!
● Should you pay $59.95 for Mega-speed
reading?
Purdue University

Attention Attention
We saw last time that attention can have
PSY 200 ●
very powerful effects

Greg Francis w when it is focused on one thing, you ignore
other things
Lecture 12
● Today we want to consider some more
specific properties of attention
Should you pay $59.95 for w and look at experimental methods that are
used to study attention
Mega-speed reading?
Characteristics of attention Attentional blink

● By identifying the properties and ● Suppose you have to identify rapidly presented (100
characteristics of attention we can deduce ms) letters
w e.g., detect J and/or K in a stream of letters
properties of the underlying systems that
are involved in cognition
M
w whether attention is thought of as a system P
K
w or as a by-product of other systems R
W
S
● Look at
w temporal
w featural
Attentional blink
Attentional blink
● Measure frequency of detection
● Turns out that detection of first letter tends to make w class data (182 observers)
detection of the second letter very difficult
w if it immediately follows the first
w Attentional blink
Implies that
M detecting the
P first letter
K causes
R you to miss the
J second letter!
S

Attentional blink Speed reading

● Suggests that processing the first target
● Ever seen the ads for speed reading?
letter, J
w uses up resources that would otherwise be used
● One is for a program called Mega Speed
to process second target letter, K Reading
w attentional focus and refocus takes time and for w claims to teach you to read 25,000 words per minute
this task takes approximately 400 ms
w ==> 2.4 milliseconds for each word
J
S w impossible, if only by attentional blink!
L
Once processing is done (a
N ● The seller is skimming at best,
few hundred milliseconds),
B
J
there is no difficulty and lying at worst
detecting second letter.
R
K
H
...
Visual search Visual search

● Proofreading ● Time and type of processing can also be measured by
having observers respond as quickly as possible when
● Much of our time is spent looking for various things they detect a target
w Staplers, Road signs w hypothesize that tasks that involve attention will be
slower than perceptually based tasks
w expect attention to depend on the number of things that
must be searched
● Typically, we distinguish between a target and
distracters by one or more features
w we vary the number of distracters
w and measure reaction time
Set Size Set Size

● In some situations, more distracters make search take ● In some situations, more distracters make search take
longer longer
w fast w slow

Feature Search Feature Search

● But if the target differs from the distracters in the right way, ● But if the target differs from the distracters in the right way,
search can be fast even with lots of distracters search can be fast even with lots of distracters
w pop out w pop out
● This often happens when the target has a unique feature ● This often happens when the target has a unique feature
relative to the distracters relative to the distracters
w shape w color
Conjunctive Search Visual search experiment

● But if the target has shared features with different ● Four types of responses
distracters, search is difficult
w 1) Feature - present (can respond as soon as see target)
w No pop out
w 2) Feature - absent (must examine all stimuli before sure
● This often happens when the target is defined by a
target is not present)
conjunction of features relative to the distracters
w 3) Conjunctive - present (can respond as soon as see
w Orange rectangle: color and shape
target)
w 4) Conjunctive - absent (must examine all stimuli before
sure target is not present)
CogLab feature search CogLab feature search

● Few distracters -easy ● Many distracters - still easy

Conjunctive search Conjunctive search

● Few distracters - fairly easy ● Many distracters - difficult
Visual search Interpretation

● Feature maps: color, shape
● CogLab data (179 observers) ● Feature search can identify target within either feature map
Color Shape
blue corners
green arcs
No searching
required!
Interpretation Interpretation
● Feature maps: color, shape ● Feature maps: color, shape
● Feature search can identify target within either feature map ● Conjunctive search cannot identify target within either feature map alone
Color Shape Color Shape
blue corners blue corners
Requires search
green arcs green arcs by comparison
across feature
maps.
Serial process
No searching that takes time
required!

Interpretation Visual search

● Feature maps: color, shape
● Conjunctive search for
● Conjunctive search cannot identify target within either feature map
target absent has a
Color Shape slope twice as steep as
for target present
blue corners w Because when the
target is present you
find it, on average,
Requires search after searching half the
by comparison items and then can
green arcs
across feature stop the search
maps. w For target absent
Serial process searches, you must
that takes time search all items to
verify each is not the
target
Automaticity Automaticity
● The process whereby a task goes from
● When a task is unfamiliar it
requiring a lot of attention to requiring little is
seems to require a lot of
attention to perform called automatization
● Later it requires less attention ● Many tasks are automatizable
w riding a bike
w color naming
w driving a car
w typing w word naming
w tying shoelaces
» http://www.fieggen.com/shoelace/knots.htm
● Can measure effects by pitting an
automatized task against a non-automatized
task
Stroop task
● Stroop (1935)
Stroop effect
● Identify the color of ink for words
● It takes longer when the words are color
names
● Demonstration
w measure reaction time

Stroop effect
Stroop effect ● Word name interferes with ink color
naming
w ink color does not generally interfere with
word naming
w lots of studies on Stroop effect
● Many effects that are similar to it

w Simon effect for pointing
● You can try them both on CogLab

w Not required, no credit
Explanation Conclusions
● Word reading is well practiced ● Methods of studying attention
w especially among college undergraduates w attentional blink
w so it occurs quickly and is automatic w visual search

w Stroop task
● Color naming is unpracticed, so it occurs
● Characteristics of attention
slowly and requires attention w timing
w role of perceptual features
● With two tasks, both trying to report on a color
● Automaticity
w the automatic one tends to mess up the
unpracticed one, it takes more mental effort (and
time) to do the unpracticed task
Next time
● Intersection of attention, perception, and memory
w Iconic memory
w echoic memory
● Serial position curves

● CogLab on Partial report due!
● Why telephone operators seem rude.
Purdue University

Sensory memory Memory

Humans demonstrate memory when they
PSY 200
●
behave in a way that could only be based

Greg Francis upon previous experience
w does not necessarily imply that there are
Lecture 13 memory systems
● Memory could be a by-product of other

systems (vision, audition, language,…)
Why telephone operators
seem rude.
Perception to memory Whole Report
● Suppose you want to know how much

information is available in a single visual ● Write down as many letters as you see
glance
● How would you measure it?
● It turns out it s a complicated task
because it involves perception, attention,
and memory
Whole Report Whole Report
● Write down as many letters as you see ● Write down as many letters as you see
B X P Q
M H E L
V N W A

Whole report results Partial report

● Same type of letter matrix
● Subjects report 4.5 letters on
● Indicate which row to report after the matrix
average (Sperling, 1960) disappears
● Subjects claim they saw more w choice of row is random
letters, but lost the percept while ● Suppose the subject reports 3 of 4 letters from
they reported any row
w they cannot report fast enough w =>3/4ths of each row was available
● How can we tell if percept is lost? w ==> entire field was available
● This is essentially how college tests are

designed!
Partial report Partial report
● Write down letters from the indicated row ● Write down letters from the indicated row
S T Y I
R F C Q
Z E V N
Partial report Sensory memory
● Write down letters from the indicated row ● Temporal characteristics

w delay

Sensory memory Sensory memory
● Temporal characteristics ● Temporal characteristics
w delay w delay
R Q V L
S D G F
P N B A
Sensory memory Results

● Vary delay to watch decay from memory
w compare to CogLab data
● Temporal characteristics
w delay
4
Number of letters correct
0
0 0.2 0.4 0.6 1
Delay of tone (seconds)

Results Results
● We can test on any row and get essentially the same
● CogLab data (148 participants) result
w 3 letters in each row w so, the number of letters that actually persist and is available is
found by multiplying by the number of rows (3)
4 .5
Number of letters available
Number of letters correct
4
Original published data CogLab data
3 .5
10 4 .5
3
2 .5 8 4
3 .5
6
2 3
1 .5 2 .5
1 4 2
1 .5
0 .5
2 1
0
0 0 .2 0 .4 0 .6 0 .8 1 1 .2 0 .5
0 0
0 0 .5 1 1 .5
Delay of tone (seconds) 0 0.2 0.4 0.6 1

Delay of tone (seconds)

Iconic/sensory memory Infant iconic memory

● Performance is better ● The partial-report
than the whole-report experiment can be
procedure because you modified in several
can focus attention on ways
just one row before the Blaser and Kaldy
Number of letters available 10 ●
percepts fade away (2010) modified it to

8
● Fading percepts are due test iconic memory
to visual persistence, 6 of infants
which we talked about 4 ● 60, 6 month old
earlier infants
2
● We call the use of this ● Infants show a
information iconic 0 preference to look at
the changed object for
memory 0 0.2 0.4 0.6 1
small enough set sizes
w Large capacity Delay of tone (seconds)
w Short duration Purdue University Purdue University
Infant iconic memory Masking

● Infant iconic memory is
actually quite similar to adults ● Masking effects can influence iconic memory
● Ask adults to report the w persistence-based memory is very brief, and is
location of the changed color
easily destroyed by a mask
item
● They do better than ● Iconic memory is
infants (who did not
understand the task ) w brief
● Look for sharp drop in w easily disturbed
performance as set size
increases
● Estimate items in memory
● Adults = 5.75
● Infants = 5.0 Purdue University Purdue University
Partial report with masking Partial report with masking
● Write down letters from the indicated row ● Write down letters from the indicated row
X V F R
W K D M
S N J Y

Partial report with masking Masking

● With the mask you do not have enough time to focus
● Write down letters from the indicated row attention on the indicated row
● Iconic memory is so brief (less than a second) that it
X X X X probably has little to do with normal memory

w Too brief to be useful for many situations (except maybe
knowing how to reach for something just after lights go out)
X X X X ● Other similar systems are more notable
X X X X
Echoic memory Echoic memory

● Properties
● Other senses have a similar type of
w Longer
persistence or sensory memory 5
Number of letters
duration
4
F (seconds)
3
H
w Smaller
available
V 2
A capacity 1
D
S E ● Significant for 0
W X 0 1 2 3 4
some memory
tasks Delay (seconds)
Immediate serial recall Serial position curve

● After given a list of items
w e.g., digits, letters, words,… ● Often, subjects recall first and last items
w subject must report them back best 1
Probability of recall
0.8
w 1) no delay (immediate)
0.6 Primacy Recency
w 2) in the correct order (serial)
0.4
w 3) no cues (recall, not recognition)
0.2
● Plot percentage correctly recalled against 0
position of item in list 1 2 3 4 5 6 7
Position in list

Modality effect Modality effect

● Recency depends on the modality of ● Explanation:
w In this task, recency depends on sensory memory
presentation
w It takes time to report all the items in the list, in order
1
w in the visual presentation, iconic memory of the last item
0.8 is gone before subject tries to report it (poor recall)

0.6 w in the auditory presentation, echoic memory of last item
Auditory
Visual
is still present when subject tries to report it (good recall)
0.4
● Thus, auditory presentation shows recency, but
0.2
visual does not
0
1 2 3 4 5 6 7
● We will explain the primacy effect later
Position in list
Suffix effect Suffix effect

● Recency when cue to report is a tone
● Auditory presentation only
● Loss of recency when cue to report is a word
● Cue to report is either a word or a tone
1
2 2, 4, 1, 6, 8, 7, 9 0.8
4 0.6 Tone
1
0.4 Word
6
8 0.2
7
0
9 1 2 3 4 5 6 7
Position in list
Suffix effect Phone operators

● Not affected by
w practice ● Call information for a number
w meaning of cue word
● Operators are very short
w common vs. rare word
w give the number
● Words are physically different from tones w do not say goodbye or have a nice day
w suffix word acts like a mask to wipe out last word in
● Avoiding the suffix effect!
list from echoic memory
w you would forget the last part of the phone
w the situation is similar to being unable to report the
number if they finished with pleasantries
letters in the partial report task with the X-masks


● Partial report experiment ● Lecture is a vodcast available on the class web site!
● Sensory memory ● Memory
w iconic memory (visual) ● Modal model
w echoic memory (auditory) w short term memory
● Relation to immediate serial recall (recency) w long term memory
w modality effect ● Experiments

w suffix effect
● CogLabs on Brown-Peterson and Serial position due!
w significance for phone operators
● Why it is difficult to win a pizza at Little Caesars.


Two store model Memory

● Simple view
PSY 200 w memory is a container of past impressions and
knowledge
Greg Francis w memories can leak-out, decay away
Lecture 14 ● Not very realistic

w need to explain why memories disappear
Why it is difficult to win a pizza at Little

Caesar s.
Simple view Ebbinghaus experiments

● First memory experiment (1885)
● The container theory of memory does not explain, for
example, ● Measure how long it takes to learn a list of nonsense
w why some memories are very long lasting (my childhood car syllables perfectly
trips to Utah) w NOF, QAP, HOS, LEQ, FIK, MEC, KIJ, HOM, NEM, MOJ
w why some memories are very brief (my wife asks me to take
● How long does the memory last?
out the trash)
● In what form does the memory last?
● We are not going to get a full theory of memory, but
we can start to get an outline ● How does it affect future behavior?
w and identify some misconceptions about memory ● Does it help relearn the list at a later time?
Ebbinghaus Forgetting curve

● Savings =1
● Relearn the list at later points in time w subjects do not need to relearn, perfect memory
w a different list each time ● Savings=0

w subjects show no evidence of earlier learning
● Measure how long it takes to relearn the list
1.2
● Calculate savings 1
0.8
Savings
0.6
Timeoriginal − Timerelearn 0.4

Still not 0!
Savings = 0.2
Timeoriginal 0
0 5 10 15 20 25 30
Time (days)

Significance Memory task

● Ebbinghaus results suggest that memories can
last a very long time, in some form ● See (or hear) a trigram of consonants
w Memories were believed to be stored in a memory
system and did not just fade away (otherwise, the ● Report it back in order
curve should not asymptote above zero)
● Ebbinghaus results suggest good
w Memory loss was believed to be due to interference of
other memories memory until other letters are also
● Other experiments challenge this view memorized
WRM
Retention
Retention
● Vary duration of counting backward
● Peterson & Peterson (1959) ● Numbers are different from letters, you
w Brown (1958) might not expect any interference
● Give subjects trigram w but they can have very strong interference
1
Proportion correct
w ask them to count backwards by 3 s and 0.8
then recall trigram 0.6
779, 776, 773,... Suggests
0.4
some memories
0.2
WRM 782 last only a
few seconds! 0
0 3 6 9 12 15 18
Retention interval (seconds)
Retention Retention
● The results of the Brown-Peterson study ● The results of the Brown-Peterson study also
suggest that some aspects of forgetting are
suggest that some aspects of forgetting
passive
are process driven w even if you are distracted, you can recall the trigram if
w keeping a memory active requires effort only a short time has passed
w if many seconds have passed, while you are
w if you are distracted by another task, you distracted, you cannot recall the trigram
cannot apply the effort to keep the memory w memory has decayed , or something like decay, while
w similar to our observations about attention you were doing the distracting task
and processing

Retention Another experiment

● CogLab data
w 85 subjects 9
● Memory span
w Not everyone shows interference on this task
5 7
w how many items can
you correctly recall 8 6 3
immediately after 4 9 2 7
exposure? 5 9 0 1 3
● The magic number 6 8 2 9 3 4
7+/-2:… 7 2 5 1 6 9 3
w Miller (1956) 9 1 0 3 5 7 3 2
3 9 0 5 7 4 2 16
Pizza
Interpretation ● The Little Caesar s in W. Lafayette used to
have a game where you could win a pizza
● There exist two types of memory systems
w must repeat a sequence of flashing lights
● Long Term Memory (LTM) (changes every time)
w high capacity (no limit) w The sequence gets longer until you make a
mistake
w long duration (forever)
w need a sequence length >7 to win much
w Ebbinghaus experiment » Counts number of correct button presses
» 56 (sequence of 11 buttons): win a soft drink
● Short Term Memory (STM)
» 110 (sequence of 15 buttons): win crazy bread
w small capacity (~7 items) » 210 (sequence of 20 buttons): win pizza
w short duration (seconds) » nearly impossible with STM properties
w http://www.freegames.ws/games/kidsgames/simon/simon.htm
w Memory span, Brown-Peterson
Modal Model of Memory Modal Model of Memory

● Atkinson & Shiffrin (1968) ● When something is memorized
● Multiple stages of memory w Items are first held in STM (temporary store)
● STM plays a dominant role in active memory w Items may transfer to LTM (permanent store)
● Requires transfer between STM (STS) and LTM (LTS) w Takes time to transfer

Free Recall Serial Position Curve Serial position curve

● The effect of position is robust across many types of lists
● Given almost any list of items w words
● Subjects remember the first and last few items w letters
w numbers
best (free recall, not immediate serial recall)
w pictures…
1
RECENCY ● Here s the

0.8
CogLab data
PRIMACY
0.6 w (91 subjects)
0.4 ● Demo
0.2
0
1 2 3 4 5 6 7
Position in list
Serial position curve

Serial position curve
● The effect of position is robust across many types of lists
w words
● In some situations the serial position curve can
w letters
be explained by different properties of STM and
w numbers
w pictures… LTM
1
RECENCY:
● Here s the
0.8 PRIMACY:
Use STM
CogLab Global
data 0.6 Use LTM
w (26,557 subjects)
0.4
● Demo
0.2
NEITHER LTM NOR STM
0
1 2 3 4 5 6 7
Position in list

● Expansion of STM into
● Short Term Memory (STM)
● Working memory
● Long Term Memory (LTM) w central executive
● STM / LTM distinction is one of the w phonological store

w visuo -spatial sketchpad
strongest conclusions of cognitive
psychology ● CogLab on Sternberg search due!
● Why there is a gate at the first floor stairway in
● Accounts for quite a bit of data
the Psych building.
● Many details are unresolved

Working memory Modal Model of Memory

● Atkinson & Shiffrin (1968)
PSY 200 ● Today we focus on the Short-term store (Short term
memory)
Greg Francis
Lecture 15
Why there is a gate at the first floor

stairway in the Psych building.
Search of memory Types of searches

● How is memory searched?
w Sternberg hypothesized three types of searches ● (1) parallel: target item is compared to all the items in
memory at the same time
● Explore by varying the number of items in memory set
w the answer (yes or no) is returned after all items have been
(similar to visual search experiments)
checked
w measure reaction time
w Sternberg (1969) NO NO
5329
5329 5329
5329 8
8888 8888
8 8
8
Types of searches Memory search

● (1) parallel: target item is compared to all the items in
memory at the same time
w the answer (yes or no) is returned after all items have been ● If parallel search
checked
w number of items does
not matter
Reaction time
YES Reaction time is w Yes and No responses

the same for a are both flat
5329 yes response
3333 3333
3 3
3
Set size

Types of searches Types of searches

● (2) serial terminating: target item is compared to each ● (2) serial terminating: target item is compared to each
item one after the other item one after the other
w the answer (yes or no) is returned after the target is found or w the answer (yes or no) is returned after the target is found or
all items are searched all items are searched
NO YES Reaction time is

faster for a yes
5329 5329 response
8 8 8 88 3 3 3
8 8 3 3
8 3
If self-terminating search Types of searches

● (3) serial exhaustive: target item is compared to each
● Go through items item one after the other
one-by-one until find w the answer (yes or no) is returned after all items are searched
(regardless of whether target is found or not)
target 700
600
Reaction time
● RT increases with set 500

400
NO
size YES
300 NO
5329
w YES RT s shorter 200
than NO RT s 100
8 8 8 88
0
● Lines have different 1 2 3 4 5 6 8 8
slopes 8
Set size
Types of searches If exhaustive search

● (3) serial exhaustive: target item is compared to each
item one after the other ● Go through every
w the answer (yes or no) is returned after all items are searched item and then report
(regardless of whether target is found or not)
answer 700
600
Reaction time
● RT s increases with 500

YES Reaction time is 400
the same for a set size YES
300 NO
5329 yes response as w YES RT increases the 200
for a no
3 3 33 3 response
same as NO RT s 100
0
3 3 ● Lines are parallel 1 2 3 4 5 6
3
Set size

Hypothetical searches Search of memory

● Sternberg s data support exhaustive
● So, we have three hypothetical ways of searching search
= Yes
STM = No
● Here s the
w They predict very different patterns of reaction time as a CogLab data
function of memory set size (153
● Sternberg runs the experiment to see how the data participants)
comes out
w You ran a version of the experiment in CogLab
Search of memory Search of memory

● Implications: Search of STM ● These results were a bombshell in 1969
w finer analysis of cognition than anyone expected
w 1) is serial, one item at a time
was possible
» and checking each item takes approximately
the same length of time w used a thought experiment about different types of
» Approximately 40 milliseconds (CogLab data is searches to generate precise testable predictions
a bit slower, 49 milliseconds) about cognition
w 2) is exhaustive » subsequent research found that there were other
» search always goes through all items types of searches that complicate the conclusions
w counter-intuitive finding
» why should search be exhaustive?
» seems inefficient!
Interpretation Controller
● Exhaustive search makes sense if search of STM
● Controlling attentional system
is done by some process that is
w supervises
w very efficient (can search very quickly)
w coordinates
w dumb (doesn t bother to stop itself) w starts and stops relatively independent processes
w initiated by some other system (a controller)
● e.g.
w Search short term memory
Controller w Search long term memory
w walking down stairs
Search w gate in psychological sciences building
memory
w Doors
STM process

Other aspects of STM Separate systems

● At about the same time, another study
indicated important characteristics of
● A complicated experiment
phonological and visuo-spatial systems
F
● Part 1: spatial mental task
● Brooks (1968)
(diagrams)
w two types of tasks (visuo-spatial and phonological)
w visual imagery
w two types of responses (visuo-spatial and
w classify corners (top or bottom
phonological)
corner?)
● Identifies two types of systems that are w yes if top or bottom
relatively separate w no if not top or bottom
Two response types

Separate systems
no, yes, no,
no, yes, no,
● Part 2: verbal mental ● Either ...
task w verbally
w read sentence A bird in w spatially

the hand is
w categorize words (noun
not in the
or not?
bush. YES NO
YES NO
NO YES
YES NO
NO YES
Results
Results ● Results
w when you have to respond by pointing, it is easier to
● Measure time to finish mental task for each
work with sentence information than diagram information
response type
w when you have to respond verbally, it is easier to work
w diagrams -- pointing
with diagram information than sentence information
w sentence -- pointing
Mental task
w diagrams -- verbal
Diagrams Sentences
Response task
w sentence -- verbal
Pointing 28.2s 9.8s
Verbal 11.3s 13.8s

Significance Interference
● The results suggest that there are two
● These system have only limited resources and capabilities
relatively separate systems ● Asking a system to do two things at once (e.g., pointing and
w one deals with visuo-spatial information and mental diagram) slows down the system
must do the pointing response and mental ● Splitting responsibilities across the systems (e.g., spoken
diagram task response and mental diagram) can be done quickly
w one deals with verbal information and must do

the spoken response and the sentence task
Diagrams Visuo-spatial Verbal Sentences Diagrams Visuo-spatial Verbal Sentences
Pointing information information Verbal Pointing information information Verbal
All together now

● Sternberg s study suggests the existence of a controller
Working memory
that tells other systems what to do
● Current thought, awareness
Working memory
● Brook s study suggests separate systems that deal
w extension of short-term memory
specifically with viso-spatial and verbal information,
respectively w small capacity
● Baddley (1986) put these ideas together into a model of w rapid forgetting
working memory
Central ● Processor of information
executive w not a storage device
w hypothesizes mechanisms that lead to memory properties
Visuo- Phono-
spatial logical
sketchpad loop

● Sternberg s study ● Properties of phonological loop
w controller system ● Data
● Brook s study w phonological similarity effect
w articulatory suppression
w separate visual and verbal systems
w word length effect
● Baddley s working memory model w irrelevant speech effect
w Central executive ● CogLabs on Memory span and Phonological

w Visuo-spatial sketchpad similarity due!
w Phonological loop ● A problem with IQ tests.

Working memory Working memory
PSY 200 ● Current thought, awareness

w extension of short-term memory Central
Greg Francis
executive
w small capacity
w rapid forgetting
Lecture 16 Visuo-
spatial
Phonological
● Processor of information sketchpad

loop
w not a storage device

A problem with IQ tests. w hypothesizes mechanisms that lead to memory
properties
Phonological loop Loop capacity

● Two components ● How many items can be
w Articulatory control process
kept in the phonological
(ACP)
Phonological loop Phonological loop
» converts non-speech loop?
information into speech
code ACP ACP
» rehearsal / refresh
● Depends on two factors
w Phonological store (PS) w Duration before decay from
» similar to how we first PS
described STM (items
decay from memory) PS PS
w Speed of rehearsal
» Refresh restarts the decay
process
● Spinning coins!
Magic number? Effect of rehearsal rate

● We earlier noted that memory span was about 7 items (+/- 2)
● The phonological loop suggests that it is not the number of
items but their rehearsal duration ● Capacity of the
To recall a list of items you must rehearse them all before any phonological loop depends
●
Phonological loop
of them fade on the rate of rehearsal (r)
w The duration of decay in the PS
● A set of items that takes ACP
● Memory span should follow the equation
w Span = (Rehearsal Rate) X (PS decay time)
longer to rehearse should
● Measure memory span (s): around 7 items be harder to remember
● Measure verbal rehearsal rate (r): around 4 items per second w more likely that some items PS
for English speakers will drop out before you get
back to the first item
● Estimate duration of decay in PS (d)
w d=1.75 seconds

Effect of rehearsal rate Word length effect

● Explains differences across groups of people
● Age effects in children
● Memory span is
w Hitch, Halliday & Littler (1989) related to the length
● This implies that it is not the of words
loop size that changes with w Number of syllables
age, but the rate of
rehearsal ● Nicely matched by
changes in reading
speed
Span=1.68(Oral Reading Rate)+0.71
w Rate of rehearsal
CogLab data
Language effects
● The CogLab experiment on memory span shows
data in agreement with our expectations (169
● Some
subjects)
languages are
spoken more
quickly than
others
● Should allow
larger memory
span
w it does
Relation to IQ Articulatory suppression

● Ellis & Henley (1980) ● Subject sees (hears) a list of phonemes
w investigated complaints about WISC ● Also repeats a phrase over and over
intelligence scores w e.g., tippy-toe, tippy-toe, tippy-toe,...
w Welsch children tended to score lower ● Recall is worse

than English children w True for both auditory and visual presentation
w (Recall for visual may be better than auditory because there is
● Part of the exam checks memory span some information in the visuospatial sketchpad as well)
1
w and the slower rate of speech in Welsch partly 0.9
Proportion recalled
explains the difference 0.8

0.7
0.6
w bilingual Welsch students tested in English got better Silent
0.5
scores than when tested in Welsch 0.4 Suppression
0.3
0.2
0.1
0
Visual Auditory

Articulatory suppression Phonological similarity

Repeating phrase ties up the ACP
Memory of a list of items is worse
●
●
w Without rehearsal more forgetting occurs
when the items sound the same
Phonological loop
Visual Auditory
ACP B F H N Better
recall
tippy-toe
B G P T Worse
PS recall
Phonological similarity Phonological similarity

● All items are stored in phonological loop ● All items are stored in phonological loop
w similar sounding items interfere with each other in the w similar sounding items interfere with each other in the
phonological loop phonological loop
w two possibilities: w two possibilities:
» 1) harder to rehearse (effect in the ACP) » 1) harder to rehearse (effect in the ACP)
» 2) fade more quickly (effect in the PS) » 2) fade more quickly (effect in the PS)
Phonological loop Phonological loop Phonological loop Phonological loop
ACP ACP ACP ACP

N
G P
B
F F
PS PS PS N
PSB G
H B T H B P T
Locus of similarity effect CogLab data

● Studies find a phonological similarity effect for auditory stimuli ● The CogLab experiment on phonological similarity shows data
under articulatory suppression in (somewhat) agreement with our expectations (161 subjects)
w We suggested two possibilities: w Ideally want parallel lines
» 1) harder to rehearse (effect in the ACP)
» 2) fade more quickly (effect in the PS)
● Since the phonological similarity effect is there even when the
ACP is not involved, it must be possibility 2 (in the PS)
Proportion recalled

Irrelevant speech effect

Irrelevant speech effect
● The presence of phonemes in the
● Does irrelevant background sound affect background is critical to the effect
memory? w strong effect when background is spoken in
w E.g., studying with the TV on German, even for English speakers
● Three groups of subjects recall consonants ● Suggests that background phonemes

interfere in the PS
w 1) no background best
w 2) background = nonsense words worst ● Study with classical music if you need
w 3) background = noise bursts something!

● Data accounted for by phonological loop
w word length effect ● Review for Exam 2
w phonological similarity
w articulatory suppression
● After exam 2
w irrelevant speech effect
● Encoding specificity
● Don t listen to lyrical music while studying
● CogLab on Encoding specificity due
w Classical music is fine
● What to do if you are drunk while studying for an

exam.

Encoding specificity Context

● The context within which you learn and recall
PSY 200 can have a profound impact on your memory
w e.g., part-set cueing
Greg Francis w given part of a set that has been learned, subjects recall
fewer of the remaining items than with normal free recall
Lecture 17 w Interference of recall
» Have to keep checking if an item you recall is
already on the list
● Demonstration
What to do if you are drunk while
studying for an exam.
Context Context
● The effect of part-set cueing suggests that to measure memory ● But memory is not exactly the same as visual
you must consider the conditions at test
search
w Memory is more often about discrimination of memory traces and
not about the strength of memory traces ● Information must be encoded in memory as well as
w Similar to visual search experiments recalled
w Such encoding can alter what features are stored as part of the memory
Feature w Which changes the discrimination of subsequent recall
search
● It turns out, that to maximize recallability
w the effort and conditions at the time of learning must be
consistent with the properties and conditions of the test
Conjunctive ● Encoding specificity principle
search
Representative study Test

● Each group is then split into two subgroups that vary in the recall of
● Subjects in two groups information
w 1) Normal recognition task
w see the same words, but have different tasks
w 2) Shown a word and asked if any of the target words rhymed with this
w This changes the encoding of information in memory
word 9
Semantic judgement Rhyme judgement 8
Number of words
7
Study
recognized
CHEESE CHEESE 6
5 Semantic
4 Rhyme
3
The man 2
threw the ball SNEEZE 1
to the 0
Standard Rhyme
______.
Recognition test

Significance Encoding specificity

● It is not that one learning strategy is better than ● Memory is better when the cues available during
the other recall match the cues available during encoding
w or that one testing strategy is better than the other w Tulving & Osler (1968)
● Encoding (learning) of information and recall of

Cue No cue
information need to match
● This means it is very difficult to test for soar-EAGLE EAGLE
HEALTH-nurse HEALTH
absolute memory
WHEEL-auto WHEEL
w subject s performance depends on many factors paper-BOOK BOOK
... ...
Encoding specificity Encoding specificity
● Each group split into two groups for testing

● Results
recall w cue does not always help
16
w cue is always related to target word
14
12
Words recalled
Cue No cue
10
Study no cue
8
soar Study cue
6
nurse 4
auto 2
paper 0
... No cue Cue
Test display

● Class esults (170 participants)
● CogLab has a variation of this experiment
w Best recall when test with same cue as at study
● Instead of cue-no cue, we used strong or weak
cue (48 word pairs to study)
Weak cue Strong cue

“Lure” is a
word that was
fruit-FLOWER bloom-FLOWER not studied

Surrounding context Surrounding context

● Deep-sea divers learn words (Godden ● Test for recall on land (dry) or under water (wet)
& Baddeley, 1975), either
w on land (dry)
w under water (wet) Study
EAGLE
EAGLE
HEALTH
WHEEL HEALTH
BOOK WHEEL
... BOOK
...
Significance Forgetting
Decompression tables for divers
Forgetting is not always a characteristic of a
●
●
w want to remember when under water
memory system, or your brain
w generally study while on land
w although it could be in some cases, it is not
● Researchers working under water have difficulty
always
recalling their details on land
w E.g., counts of species ● Forgetting must be defined operationally
● How do you know if something is forgotten? w specify the task and context of retrieval
w changing context may allow subject to recall seemingly w You can never be certain that if you are placed
forgotten information
in a different context you will still show forgetting
w forgetting = retrieval problem?
Internal context Internal context

● Best recall if study and test states are the same
● Goodwin et al. (1969) w Similar effects for marijuana cigarettes (Eich et al., 1975)
● Subjects drink 10 oz. 80 proof vodka, mixed in
sugar-free lemon-lime drink
w or a similar tasting drink (bit of vodka on top of drink)
Study state
● Test memory 24 hours later
● Subjects are either:
w Sober at study, sober at test
w Drunk at study, sober at test
w Sober at study, drunk at test
w Drunk at study, drunk at test

Testing Mood
● Mood has a similar effect (Eich et al, 1994)
w Mood induced by music and directed thoughts
● So, if you are intoxicated while studying for an
exam
w and you didn t study before
Study mood
● You should be intoxicated while taking the
exam
But do not
expect to do GET HELP!
very well
Classrooms Classrooms
● Is memory better when you are tested in the same room as ● Subjects recall all words either in the context of Day 1 or Day 2
lectured? (different contexts for different subjects)
w significant for final exams! ● Recall was best for words that were studied in the test context
● Smith et al. (1978)
w Subject studied words in one of two contexts (on separate days)
w Varied classroom and dress of experimenter
Study context

● Context ● Discrimination in memory
● Proactive interference (PI)
● Release from PI
w memory best if study and test are similar
● Cues
● CogLab on False memory due!
● Environment
● State ● How to take a test.
● Mood
● Classrooms Purdue University Purdue University

Memory discrimination Discrimination

● Many cognitive tasks require you to
PSY 200 discriminate between events/stimuli
w Is this a real smile?
Greg Francis w Is this fruit ripe?

w Is there a stapler on the desk?
Lecture 18 ● The same kind of discrimination is required for memory
How to take a test.
Discrimination False memory

● An example list is
● Discrimination is difficult because memories
w smooth, bumpy, tough, road, sandpaper, jagged, ready,
can come from lots of different sources coarse, uneven, riders, rugged, sand, boards, ground,
gravel
● Consider so-called False memory studies
w the special target is rough, which is not shown to the
w as in CogLab
subject
w subject views a list of words
● After viewing the list, the subject must go through a
w the list of words have something in common
set of words and identify which ones were in the
» they are all related to a target word
just seen list
w some words were in the list
w some words were not seen
» including the special target
False memory False memory

● The main finding is that the special target is often
identified as part of the just seen list ● These types of findings suggest that our
w even though it was not memories are
● Sometimes people will even report that they w not necessarily accurate, we can remember things that
recall seeing the special target never occurred
w but this is impossible because it was never shown w able to be manipulated, to a certain extent, I can make
you have certain memories
● CogLab data (163 participants)
w Type of selected items Percentage of recalls ● Why does the false memory effect
w In original list 78.5 happen?
w Normal distractor (not in list) 7.9
w Special distractor (not in list) 78.5

False memory False memory

● With every to-be-remembered item you store some ● At the end of the trial, you have a lot of items in memory that
information, but not only information about the item are related to the list
w Other information is automatically generated as well w Some of them are items that were actually on the list and some of them are items that
were generated but not actually on the list
w smooth, bumpy, tough, road, sandpaper, jagged, ready, w Reporting all items from memory is not going to lead to good performance
coarse, uneven, riders, rugged, sand, boards, ground, gravel w Both types of memory items are real, but only one type matches the physical stimuli
● Good performance on this task requires discrimination between

Memory: Item 1 Memory: Item 2 Memory: Item 3
memories generated by physical stimuli and memories generated by
internal processes stones
bumpy bumpy w Source monitoring
jagged
bumpy
smooth smooth smooth smooth peaks Memory
rough rough tough rough tough rough
baby baby baby nails babysandpaper
road road nails
gravel road
uneven
Discrimination Interference
● Good memory recall usually requires not only ● Retroactive interference (RI)
recall of an item from memory w new information prevents recall of
previous information
● You also must identify the correct item relative
w e.g., Overwriting a computer file.
to the appropriate context or time frame
w The current trial ● Proactive interference (PI)
w The context of the experiment w prior learning prohibits new learning
w Relative to an earlier event w e.g., Learning new cultural customs.

w At a particular moment in time
Proactive interference Proactive interference

● May be due to a variety of effects ● Suppose you see this building as the test on the third trial
w You have a match in memory, but is from trial 1, not trial 3
w One is that memory involves discriminating new from old
w You may report it having been shown in trial 3
● Visual memory
● In general, previous trials make memory discrimination
w See a set of photos
more challenging
w Then see a test photo and decide if new or old
Memory: Trial 1 Memory: Trial 2 Memory: Trial 3 Memory: Trial 1 Memory: Trial 2 Memory: Trial 3


● One finds proactive interference for lots of memory tasks ● One finds proactive interference for lots of memory tasks
● CogLab serial position experiment ● False memory experiment
w I looked at recall of the first letter in each list, averaged across w I looked at recall identification of the normal words in the list
all students
1 6.2
Proportion of times first
0.9
Average number of words

6
0.8
letter is recalled
0.7 5.8
recognized
0.6 5.6
0.5
0.4 5.4
0.3
5.2
0.2
0.1 5
0
4.8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 2 3 4 5 6
Trial
Trial

● Inference does not happen for all experiments ● Inference does not happen for all experiments
w Just those related to memory w Just those related to memory
● Partial report experiment (first 12 trials had the cue before ● Attentional Blink experiment (detection of the first letter in
the letter matrix – to give you practice) the stream)
0.9 0.5
Proportion of letters reported
Proportion of letters reported
0.8 0.45
0.4
0.7
0.35
0.6
0.3
0.5 0.25
0.4 0.2
0.3 0.15
0.1
0.2
0.05
0.1 0
0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
1 2 3 4 5 6 7 8 9 10 11
Trial
Trial
Release from PI Release from PI

● Proactive interference weakens for different
stimulus types ● Trials 1-3 show build up of PI
● Run two Brown-Peterson type experiments ● Experimental group shows release of PI on
Control Trial 4 100
Experimental
Percentage correct
80
XJF Trial 1 XJF 60 Control
WRM Experiment
Trial 2 WRM 40
20
DBL Trial 3 DBL
0
NRX 1 2 3 4
Trial 4 942 Trial

Release from PI Memory system

● Every memory system must have at least two
● Works for many kinds of memory tasks components/processes
● Many kinds of stimuli w Storage
w retrieval
5, 7, 9, 1 NICE, SUNNY, ENJOY, PUPPY
vs ● We have described proactive interference as being
vs
5, 7, 9, HAND NICE, SUNNY, ENJOY, KILL due to difficulty discriminating new items from
previous items
D, D, D, D ● But there is an alternative explanation

News stories vs w Proactive interference might prevent items from being
D, D, D, F stored and thereby make them unrecallable
Working memory Testing models

● Test storage vs. recall of PI by changing
● For example, working memory has a storage instructions after the list is presented
interference hypothesis for the phonological loop ● Experiment
● Working memory suggests that interference can w stimuli are names of indoor and outdoor
Trial 1
games
FOOTBALL
occur Visual Phonological loop
Auditory
w subjects usually do not notice that word Trial 2 SOCCER
w by blocking ACP rehearsal (articulatory ACP
on the fourth trial is an indoor game and
suppression, Brown-Peterson task, 421-418-…
others are outdoor games Trial 3 BASEBALL
word length effect) PS ● Take two groups of subjects
w within the PS when items sound similar
w one has traditional PI type experiment
Trial 4 WALLYBALL
(phonological similarity effect)
w one is told of difference on fourth trial, at
w both of these interference types block the storage of items the time of test
(items fall out of the loop)
Interference at recall How to take a test

● If PI prevented the last item from being stored your telling a ● Avoid PI
subject that the fourth item was an indoor sport, should
● Answering successive questions on the
make no difference (other than guessing)
same topic hurts recall
w but it makes a big difference, they show release from PI
100
w after answering unrelated questions
Percentage correct
80
w go back to questions you cannot answer
w less proactive interference
60 Not told
Told w should recall more
40
20 T1, T1, T1, T1

0 ACT vs
1 2 3 4 T1, T1, T1, T2
Trial


● Discrimination ● Constructive memory
● Retroactive interference ● Flashbulb memories
● Proactive interference ● Memory misattribution
● Release from PI ● Misleading questions
● Strong effects
● Knowing about can help in everyday ● How good is eye-witness testimony?
tasks

Constructive memory Memory test

● Last lecture you watched me for 50 minutes
PSY 200 ● What was I wearing?
Greg Francis
Lecture 19
How good is eye-witness testimony?
Discrimination No forgetting?
● Brain surgeon (Penfield, 1959)
● The task is difficult because you have to do w Epilepsy patients
several things w stimulate brain regions before operating
w Recall information that might be related to the task from memory w want to know what is being removed
w Determine if the memory is actually for the correct event ● Conscious patients report vivid memories
w Determine if the memory is actually for the correct moment in w unable to recall normally
time » "she saw herself as she had been while giving birth to her
w Gauge your confidence in the memory s validity baby."
w stimulation of temporal lobes
● All of this suggests that performance on a memory task
involves discriminating information ● In the image, numbers indicate
places where stimulation
To address the discrimination problem, people engage
evoked different reported
●
in a constructive process to report memories

experiences
No forgetting? Penfield (1959)

● Suggests that memories are stored but normally ● Even worse…
unreachable (context things again)
w the memories are nearly impossible to verify
● Basis for ideas of memory repression (and a few self-help books) w the few attempts find that the memories are not true
w The results are usually misunderstood » people describe places they have never visited, impossible
events, fantasy,…
w Actually only occurred for 5% of patients w The patients have epilepsy
» Stimulation may have triggered something like an epileptic
seizure (which can have hallucinations)
● It is more likely that stimulation feels like a memory, even though
it is not
w your awareness of remembrance is a product of your brain
w it can be stimulated, even without a real memory
● What do we mean by a valid memory?

Flashbulb memories Flashbulb memories

● Highly emotional events tend to produce strong ● People vividly recall details
memories surrounding event
● e.g. w where they were when they
w JFK assassination heard
w what people said
w Challenger explosion
w clothing worn
w Oklahoma City bombing
w time of day,..
w Earthquakes
w September 11, 2001 ● People are confident about
w ... their reports
w however...
Flashbulb memories Flashbulb memories

● Talarico & Rubin (2003) ● The memories of the WTC
● On September 12, 2001 attack were more vivid
● Asked volunteers to answer questions ● Subjects believed those
about their memory of
memories were more likely
w The WTC attack
to be reliable
w An ordinary event (volunteer s choice)
● Three groups for follow up
● Ordinary memories
w 7 days later w Faded in vividness
w 42 days later w Belief decreased over time
w 224 days later
● Recalling of details was the same for
WTC attack and ordinary event
Flashbulb memories Memory misattribution

● Subjects confidence in their memory can be ● Donald Thomson was accused of rape and picked out of
misleading a lineup by the victim (Schacter, 1996)
w retelling of the story ( I remember vividly when Kennedy ● He was on live TV at the time of the rape
was shot. I was… ) probably reinforced the story w Ironically, he was discussing memory of faces for eyewitness
w Maybe not the true memory testimony
● Flashbulb memories are a real phenomenon ● The victim

about the experience of memory, w had the TV on at the time of rape
w misattributed the face on TV for the face of her attacker
w but probably not super-memory
w very accurate report of the crime, otherwise

Eyewitness testimony
● I will show you a series of slides and then ask you
some questions
Purdue University Dr. Kim Vu PSY 332: Cognition 14

Purdue University

Purdue University

Purdue University

Eyewitness testimony
Questions
● Later, show slides and ask subjects if they were part of the
● Did the bus, which came by, come from the left or the original set
right?
● Key test is for a pair of slides (between subjects)
● Did another car pass the Red Datsun while it was at w real slide contains YIELD sign
the intersection with the stop sign? w fake slide contains STOP sign
● Did you see a bicycle?
● Did you see the taxi cab?
● Did you see if the policeman wrote anything down?

Eyewitness testimony Eyewitness testimony

● Loftus, Miller &
Misinformation effect 80
●
Burns (1978)
70
w Loftus & Palmer (1974) The Delayed
Percentage correct
●
60
Compare accuracy according to pre-test questions misinformation Immediate
● 50
effect gets
w Subjects without a misleading question--90% accurate 40
stronger with a
w Subjects with a misleading question -- 20% accurate 30
week delay
before the 20
w In a follow-up, the experimenters asked those with misleading
memory test 10
questions if they thought they were misled
0
» 90% say no Consistent None Misleading
● Paying money for correctness also had no effect Types of questions
Memory implants Conclusions

● Loftus has a procedure that implants a ● Many techniques believed to provide accurate
memory of being lost in a mall memories, do not
w Basically just have subject read a plausible story (with w flashbulb memories
some details that could be true)
w brain stimulation
w Get family members to pretend the story is true
● Memory is constructive
w Later the subject remembers the story as something
that happened to him/her ● Memories can be easily influenced by questions,
● It is very easy for a therapist to implant false interpretation, and context
memories into patients
Next time
● Amnesia
● Anterograde amnesia
● Retrograde amnesia
● Unusual characteristics
● Repression
● CogLab on Forgot it all along due.
● What s wrong with myPurdue

wife?University

Amnesia Fundamental fact

● There is no method other than object
PSY 200 physical evidence to verify the accuracy of a
memory
Greg Francis
● Memory is a cognitive experience
Lecture 20 w Confidence in the memory is another cognitive
experience
w You can be very confident and still be wrong
What is wrong with my wife?
● Of course, we must be correct fairly often, or
our lives would be a total mess!
Amnesia Amnesia
● Scope and duration
● Loss of memory or memory abilities
● Retrograde amnesia for one patient
w retrograde: forgetting events prior to the injury
5 months 8 months 16 months
w anterograde: forgetting events after the injury
● In most cases amnesia is limited in scope and

duration
coma coma Time
w like when my brother Joe slipped while playing frisbee coma
of
1 year 2 weeks accident
2 years
4 years
to infancy to infancy to infancy
What s wrong with my wife? An unusual case

Side issues
Nothing!
●
●
w Sense of smell
w Mild anomia
w Odd aphasia (language deficit)
● She is able to learn and remember new information
● Remarkably unaffected by the loss of memories
w Personality
● But she cannot remember anything before w Parents
her senior year in high school w college

w makes study of retrograde amnesia difficult
w motor cycle accident
w complete retrograde amnesia

What is lost? What is lost?

● How can someone who loses their ● More generally,
childhood memories go to college the next w while patients with retrograde amnesia forget their
year? names, parents, addresses,…
w memories cannot be wiped clean w they do not generally forget how to walk, talk,
solve problems
w perhaps they are just not directly accessible
» Although they may have problems…
w forgetting = recall problem?
w Different types of memory systems
» controversial!
Memory systems Anterograde amnesia

● There are many different types of memory
● Some patients have amnesia that preserves past
● Amnesia seems to affect declarative more than nondeclarative memory
memories but prevents formation of new memories
w many are long-time alcoholics who did not eat properly
» which leads to a thiamine deficiency
» which leads to Korsakoff s syndrome
● Leonard in Memento
Anterograde amnesics
Patient HM ● Fairly normal STM digit span (~7 items)
● Surgery on hippocampus (to control epilepsy) ● But very difficult to extend digit span
● anterograde amnesia w how many trials to repeat back list correctly?
30
w unable to learn anything new
● Thought it was 1953 25
Mean trials to criterion
20
w shocked by age of face in his mirror
Controls
w Could not stand to read newspapers 15
Patients
w reintroduced himself to doctors, nurses,… 10
● Could carry on a conversation! 5
0
0 5 10 15 20 25
Number of digits to remember

Serial position curve Patient HM

● Normal recency
● Abnormal primacy ● Could learn
● Consistent with STM-LTM dichotomy some things!
100 w E.g.,
90
location of
80
Percent correct
70 hospital
60
Controls
cafeteria
50
Amnesics w E.g. mirror
40
30 drawing
20
task
10
0
0 5 10 15
Item number
Patient HM Amnesia-like memory

● Mirror drawing task (Milner, 1968)
● HM had no knowledge of doing the task before! ● Some aspects of memory seem very much like
amnesia
w infantile amnesia
w repressed memories
● Careful studies are difficult to come by because

the memories (and absence thereof) must be
verified
w remember the fundamental fact at the start of
Purdue University
today s lecture Purdue University
Infantile amnesia Infantile amnesia

● Most people report that they cannot remember ● Reason is unknown, but the best theory goes
anything that happened to them before age 4
like this…
years
20
w children younger than 4-years-old view the world
Number of memories
differently from adults

15
reported
w by encoding specificity, one needs to be in a

10
similar state as study to best recall something
5
w adults are very different from children, and this
0 prevents recall of early memories
0 2 4 6 8
Age at time of event


Repression Repression
● Psychotherapists (e.g. Freud) suggested that ● In a laboratory, showing evidence of repression
infantile amnesia occurred because much of requires
childhood is filled with painful events and memory
w being unable to remember something
of the pain is prevented by psychological defense
mechanisms (repression) w being able to recover the memory through therapy
w proving that the recovered memory is accurate
● This is very unlikely
w people do remember painful events well
w laboratory studies find no evidence of repressed memories
Discovered memories
Repression
● However, it is possible for information that was once known to be
forgotten and then (re)discovered
● In therapy, clinicians often claim evidence of
CogLab’s Forgot it all along experiment demonstrates this
repression with
●
property
w dream interpretation
● Phases I and II are like an encoding specificity experiment
w patterns in symptoms
Study with cue Test with same or different cue
w recovering a memory through hypnosis
cup-D_ _K
● None of these techniques demonstrate a verified cup-DESK
memory pan-D_ _K
● Among carefully controlled memory research, there is
no evidence of repression!
Discovered memories Discovered memories

● Results:
● Phase III: judge your memory for an item in phase II
● Phase II (cued recall): encoding specificity effect
● We only care about the items that you correctly recalled in Phase
II ● Phase III (memory judgment): remembering recall is also affected
by cue type
● Thus, it is possible to forget that you remembered, and a change
Did you recall the upper case word? of cue would allow you to “recover” a forgotten memory
(same or different cue)
cup-DESK
Class data
pan-DESK 172 participants

● Retrograde amnesia
● Levels of processing (CogLab due!)
● Anterograde amnesia
● Judgments of learning
● Learning in anterograde amnesics
● Practice testing
● Infantile amnesia
● Learning styles
● Repression
● How to improve your memory without spending
$20.

Memory
Improving memory
● We seem to be unable to control our
PSY 200 memories
w learn things we don t want to remember
Greg Francis w unable to learn things we want to remember
● Is there any reliable cue that something will be

Lecture 21 remembered?
w no
w but there are several tricks you can use
to improve memory in certain situations
How to improve your memory without
spending $20.

● We know that memory is best when study and test contexts are ● Test subjects in a neutral context
similar (after another 3 hour interval)
w For example, testing in the study classroom
● Look at proportion correct recall
● But variability in study promotes more general recall w Highest with variable study contexts
w Smith et al. (1978)
● Advice: if you want to remember
w Subjects studied words twice: either in same context or different contexts (3 hour
interval between contexts) something in lots of contexts, study in
lots of contexts
Session 1 Learning context
Session 2 Classroom 1 Classroom 2
Context
Classroom 1 0.41 0.69
Classroom 2 0.53 0.39
Study style
Level of processing
● Time spent studying is also “context” for memory
● Memory can be influenced by depth of processing at the time
retrieval
of study
● Generally, more study leads to better memory w Craik & Tulving (1975)
● Style of study matters too ● Subjects observe words with associated tasks
w distributed practice is better than massed practice

w avoid cramming! question YES NO
w true for many skills 1 hour In capital
BOOK book
letters?
1 hour Rhyme with
3 hours spring sprint
thing?
1 hour Synonym
bulky brown
for heavy?

Level of processing CogLab

● Recall is better as depth of processing increases ● Recall is better as depth of processing increases
w More distinctive memories are created, which helps subsequent recall w “Test” is what matters here, other data is just for completeness
w By varying depth of processing, you can construct memories that are w 175 participants
more likely to be recalled
25
Number of words recalled
20
15
10
0
Capital Rhyme Synonym
Study task
Levels of Processing
Intention
● Level of processing is more important than intent to
learn (Hyde & Jenkins, 1973) ● Recall (out of 24 words) varies a lot with task
● 11 groups of subjects ● Not much variation with intention to learn
w 1 control group: told they will be tested to recall the words
» not given any study task
w 10 experimental groups split to perform a study task
» Pleasant-unpleasant rating
» Estimate frequency of word usage
» E-G checking: does word contain an E or a G?
» Identify part of speech: noun, verb,…
» Sentence framing: which sentence does word best fit in?
w For all experimental groups, either
» (a) Intentional learning : told they will be tested to recall the
words
» (b) Incidental learning: not told they will be tested
Implications Judgments of Learning

● Advice: study interactively ● Nelson & Dunlosky (1991)
w read notes ● Subjects study a pair of words (e.g., OCEAN – TREE)
w rewrite notes
● Estimate how likely they are to be able to remember
w rephrase notes
one word if shown the other (JOL).
w teach someone else w Given OCEAN, how likely to remember the associated item
● Importantly, people are not usually good at estimating later?
whether something will be remembered w This is the subject s estimate of their ability to use LTM
● Make judgment either
w Immediately after studying the pair
w Delayed to later in the experimental trials
● Note: students studying for an exam often use the
immediate approach for a JOL to decide if they need
to continue studying

Judgments of Learning Practicing recall

● A common approach to studying is
● Immediate JOLs do to use flash cards (or something
not match memory similar)
performance (at the ● Two steps to studying
end of the experiment) w 1) Read material on both sides
w Especially for high (study)
JOLs w 2) Practice test the material (given
● Delaying the JOL one side, try to recall the information
on the other side)
leads to fairly accurate
JOLs ● What should you do when you
successfully recall the information
● Advice: study, wait, during the practice test?
estimate learning
w Continue to study?
w Continue to test?
w Set aside and focus on other cards?
Practicing recall Practicing recall

● Karpicke & Roediger (2008) ● Standard advice is that once you learn something, study something else
● Subjects study 40 Swahili - English word pairs ● This is not good advice Keep practice
w mashua – boat ● Performance is best when testing
w kaka -- brother every pair is tested, even if
w Test for English given Swahili: you have already
» mashua --??? demonstrated it is
memorized
● Four groups of subjects, that differ after an item is correctly recalled
w ST (study-test): subject studies and continually tested over every pair ● The amount of time spent
studying the words does
w SnT (study on non-recalled - test on all): when a subject recalls a pair, it is no
not matter so much Don t practice
longer studied, but it continues to be tested
test
w STn (study all, test only on non-recalled): when a subject recalls a pair, it ● Suggests that you learn how
continues to be studied, but it is not tested to recall the information
w SnTn (study on non-recalled, test on non-recalled): when a subject recalls a ● Advice: Test yourself!
pair, it is not studied or tested again
● A week later, everyone is tested
Learning styles Learning styles

● A common approach in education is to identify a student s learning style ● Unfortunately, there is absolutely no evidence that reported learning style
and then teach for that style preference has anything to do with learning
● Lots of tests to identify a student s learning style ● Pashler et al. (2009) observed that to demonstrate evidence that learning
● There do seem to be real differences in what style people indicate they style influenced learning, you have to show a particular kind of interaction
prefer of effect

Learning styles Learning styles

Pashler et al. (2009) then reviewed hundreds of studies purporting to show
●
● Why is the idea popular?
evidence for learning styles, but only ever found effects like these
● But these only indicate an advantage for a type of learning or a method ● It fits with the American ideal of everyone being
capable of learning if given the chance (no child
left behind)
● It allows parents (and students) to blame the
educational system for failure rather than lack of
motivation or ability
● It lends itself well to statistical quirks of finding
just the right method for a given student
● It s a generalization of the experience that a
given student benefits from a new explanation of
Purdue University material Purdue University

● Lots of ways to improve memory
w Encoding specificity ● Mental imagery
w level of processing ● Sleep
w Judgments of Learning ● Brain training
w Practice testing
● CogLab on Link Word due!
w Learning styles
● Get a good night’s sleep!

Memory
Improving memory 2
● We seem to be unable to control our
PSY 200 memories
w learn things we don t want to remember
Greg Francis w unable to learn things we want to remember
● Is there any reliable cue that something will be

Lecture 22 remembered?
w no
w but there are several tricks you can use
to improve memory in certain situations
Get a good night’s sleep.
Memory trick - grouping

SF: Digit span
● We often hear of people memorizing
Broke down and organized
pages of the phone book
●
each digit list

w how do they do it?
● Long-distance runner
w some cheat (frauds) w sequence like 3492 converted to
w others take advantage of organization and 3 minutes 49.2 seconds- near
memory tricks world record time
Eventually created a hierarchy

SF learned to increase his digit span to
●
●
of tricks (ages, dates)
79 digits (any random sequence)
● Technique did not transfer to
w 230 hours of practice (over 20 months)
other memory tasks (e.g.,
w Ericsson, Chase & Faloon (1980) letters)
Method of loci Method of loci

● Used by ancient Greeks to remember complicated speeches
● e.g., grocery list
● To remember a list of words or key ideas
w visualize walking around an area with distinctive landmarks
w link the items to be remembered with landmarks by using bizarre ITEMS LOCI Add vivid,
mental imagery
bizarre imagery
w to recall items in order, mentally walk through area
hot dogs driveway
w (any ordered sequence will work -- e.g., a children s rhyme)
Memory piggybacks on the easy recallability of the bizarre
●
cat food garage interior
imagery
tomatoes front door

Peg word system

Peg word system
● Associate items in list with a previously ● Hook to be remembered items to the list
memorized list w visual imagery helps again!
ITEMS Peg word
One is a bun. Six is a stick. recall by
Two is a shoe. Seven is a heaven. milk bun reciting poem
Three is a bee. Eight is a gate.
Four is a door. Nine is a line.
Five is a hive. bread shoe
Ten is a hen.
bananas tree
Link word method Link word method

● In a study of learning 120 Russian words (Atkinson
● Foreign language vocabulary
& Raugh, 1975)
w find an English key word that sounds like some part of the
foreign word ● Two groups
w form a mental image of the key word interacting with the w control: heard Russian words, saw English translation
English translation of the foreign word w experimental: heard Russian words, saw English
● E.G. translation, saw key words, and applied method
w pato -> Spanish for duck , sounds like pot-o ● Experimental group learned more words faster and
» imagine duck with pot on its head for longer
w zronok-> Russian for bell , sounds like zrahn-oak w 6 weeks later
» imagine an oak tree with bells as acorns » experimental (43% correct)
» control (28% correct)
Link word method Mnemonists

● CogLab Link word lab (154 participants) ● Some people seem to have extraordinary
● Study 50 French words (25 in each condition) memories
w Half with a provided link word to form an image
w professional - apply one of the techniques we ve
w Half without a provided link word (no image)
discussed
w spontaneous- seem to not consciously apply a
technique
● Photographic memory?
w Few documented cases
w Generally, not happy outcomes

S.: Luria S.: Luria

● Extreme synesthesia
● Luria: Russian psychologist
w sensory information from one modality evokes
w met S in 1920s sensation in another
● S w tone, 30 cps, 100 decibles --> saw a strip 12-14
w able to recall without error a list of 70 words cm wide the color of old, tarnished silver
» took 2-3 minutes » 50 cps--> brown strip, taste of sweet and sour
borscht
» able to report it again several months later
w voices gave rise to visual responses
w other unusual characteristics
w used the full sensation of events to help memory
Brain Training
S.: Luria ● Several companies market
activities to make you smarter
● Visual imagery w “Exercise” your brain with games that are
adapted from neuroscience
w used method of loci w Does that even make sense?
● Often aimed toward elderly

w such strong imagery it interfered with his ability to (Alzheimers) and young children
understand simple prose
» words kept evoking inappropriate images...
Brain Training Brain Training

● Much of the hype comes from a study that trained people for a ● Much of the hype comes from a study that trained people for a
few hours on an dual n-back task (Jaeggi et al., 2008) few hours on an dual n-back task (Jaeggi et al., 2008)
w Does the current stimulus match the one from n trials back? w n is also a measure of how well subjects do the task
w n is adjusted for each person so the task is always demanding ● Subjects do get better at the n-back task with training

Brain Training (WARNING!)

Brain Training
● Redick et al. (2013) cautions:
● Transfer effects for a measure of fluid intelligence (refers to
the ability to reason and to solve new problems independently ● The conclusions are based on 4 small studies that
of previously acquired knowledge) varied in many ways
w Training group does better than a control group
w It is probably a mistake to average scores across these studies
w Amount of training time is related to gain in intelligence
● Some selective reporting of measures of fluid
intelligence
w Measures that did not show an effect were not reported
● No comparison to an “active control”

w Where subjects complete a training task that should not improve
fluid intelligence
Brain Training (WARNING!)

● In October 2014, a group of memory researchers released a Sleep
statement with the following summary:
w We object to the claim that brain games offer consumers a scientifically
grounded avenue to reduce or reverse cognitive decline when there is no ● Many types of memory improve with sleep
compelling scientific evidence to date that they do. The promise of a magic bullet
detracts from the best evidence to date, which is that cognitive health in old age ● Some type of consolidation of memories
reflects the long-term effects of healthy, engaged lifestyles. In the judgment of
the signatories below, exaggerated and misleading claims exploit the anxieties of
older adults about impending cognitive decline. We encourage continued careful
● The effect is not just time
research and validation in this field.
w Although time also has an effect
● You should be similarly skeptical about claims for improving attention,
perception, and other mental capabilities ● We ll look at one representative study
w Playing video games does not seem to improve your attention or perception
● You can improve performance on specific tasks, but that does not typically w Ellenbogen et al. (2007)
transfer to other tasks
● You can make yourself smarter by learning new information
Sleep Sleep
● Subjects learn to identify ● There is a ordered arrangement to the stimuli
order relationships ● If you know this arrangement, deciding for any pair is easy
between random shapes w But subjects are never explicitly told about this arrangement
● Only shown one pair at a
time
● Subjects have to
learn/memorize the
appropriate answer to
each pair

Sleep Sleep
● Subjects are split in to three ● Subjects are split in to three
groups, according to when groups, according to when
they are tested they are tested
w 20 minutes later w 20 minutes later
w 12 hours later w 12 hours later
w 24 hours later w 24 hours later
● No differences when tested ● Big differences when tested

on the originally studied on new pairs that fit the
items ordered structure
w E.g., A>C, C>E, B>D
Sleep Conclusions
● Half of the 12 hour group
had sleep and half did ● Lots of ways to improve memory
not
● It makes a difference for
w Method of loci
pairs of items that are far w Imagery
apart in the ordered
structure w Mnemonics
● 1-degree: A>C, B>D,… w Brain training
2-degree: A>D, B>E,…
●
w Sleep
● Advice:
w study early!
w Get some sleep!
Next time
● Mental representation
● Prototypes
● Exemplars
● Propositions
● CogLab on Prototypes due!
● What is a shoe?
Purdue University

Professor Greg Francis 1/6/20
Concepts
Representation of knowledge
● What is the information in Long Term Memory?
PSY 200 w May be several different types
● We have knowledge about the world

Greg Francis w Due to personal experience
w Or due to language
Lecture 23
● Such information must be in some kind of format,
which we call concepts
But what are the concepts?
What is a shoe? ●
w what is the concept of dog, walking, or free-market

capitalism ?
Concepts Definitions
● We will look at three topics in concepts ● Plato (and Socrates) spent a lot of effort
w Definitions (don t really work) trying to define terms like virtue and
w Prototypes (closer to how humans think) knowledge
w Exemplars (more likely than prototypes) w they were largely unsuccessful
● And then combinations of concepts ● the 20th century philosopher Wittgenstein

w propositions wondered if definitions of even simple
concepts were possible
Definitions Definitions
● Consider the concept shoe, you might define it as ● Consider the concept shoe, you might define it as Webster s
Dictionary does
Webster s Dictionary does
w A covering for the human foot, usually made of leather, having a
w A covering for the human foot, usually made of leather, thick and somewhat stiff sole and a lighter top.
having a thick and somewhat stiff sole and a lighter top. w Anything resembling a shoe in form, position, or use.
w Anything resembling a shoe in form, position, or use. ● But now consider some situations and decide if they are really
shoes
● Lots of shoes fit this definition
w A shoe that is intended for display only

Definitions Definitions
● Consider the concept shoe, you might define it as Webster s
Dictionary does ● The difficulty is the same one that Plato and Socrates
w A covering for the human foot, usually made of leather, having a had trying to define virtue
thick and somewhat stiff sole and a lighter top. w for any definition you come up with, I can find examples that do
w Anything resembling a shoe in form, position, or use. not seem to fit the definition
● But now consider some situations and decide if they are really ● But we all know what a shoe is
shoes
w so our knowledge of this concept must not be based on some
w a shoe filled with cement, which cannot be worn
precise definition
w a covering worn on the hands
of a person without legs who ● Note, scientists can (sometimes) create precise
walks on his hands definitions (e.g., a dog is defined by a DNA pattern or
w And this? à by mating abilities)
w but the definition is somewhat arbitrary
Prototypes Prototypes
● Perhaps what defines a concept is similarity ● In prototype theory it is possible for an object to
among its members be more or less a certain concept
w there may be no absolutely necessary characteristics ● Consider the concept coffee cup
w there may be no absolutely sufficient characteristics
● Prototype theory supposes that similarity is

judged relative to a prototype example of the
concept
w e.g., an ideal, average, or most frequent version of the
concept
● In prototype theory it is possible for an object to ● In prototype theory it is possible for an object to
be more or less a certain concept be more or less a certain concept
● Consider the concept coffee cup ● Consider the concept coffee cup
w and variations (some are cup-ier than others) w and variations (some are cup-ier than others)

● In prototype theory it is possible for an object to ● In prototype theory it is possible for an object to
be more or less a certain concept be more or less a certain concept
● Consider the concept coffee cup ● Consider the concept coffee cup
w and variations (some are cup-ier than others) w and variations (some are cup-ier than others)
● Lots of experiments suggest the role of ● Prototypes
prototypes
w Posner & Keele (1968): learning category names for
random dot patterns
w Discriminate two sets of random dot patterns
w Each pattern is a variation of one of two prototype
patterns
A B
● variations are made by moving some of the dots ● The key test is done after subjects learn to classify the
variants
w reaction time for judgment is recorded for stimuli they have never
seen before
» new variants
» the prototypes
w reaction time is faster for the prototypes
w which suggests that the mental representation of the categories
(concepts) are built to favor the prototype of the category
variant of A variant of B ● Look at CogLab data
● subjects learn to classify many different variants

w they never see the prototypes themselves

● Consider the types of concepts you can have
● Results are based on data from 145 participants (39,285 for global).
w Pattern type Reaction time (ms) Global RT(ms) w and how specific they can be
w Prototypes 798 969 ● things: bird, dog, chair, shoe,…
w Variants 843 1000
● actions: walking, running, sleeping,…
● Unanswered by this (and many other) experiments is what a prototype is:
● goal-derived: things to eat on a diet , things to carry out of a house
w a thing that resides in memory and contains information about the category
features? in case of a fire ,...
w the result of processing information? ● ad hoc: things that could fall on your head , things you might see
● A bit of thought suggests it is the result of processing information while in Paris , gifts to give one s former high school friend who has
just had her second baby ,...
● When studied, these concepts all seem to have prototype
characteristics
Prototypes Exemplars
● We can generate new concepts from old ● A concept consists of lots of examples of the
concepts concept
w it s inconceivable that every possible prototype exists w e.g., a coffee cup concept might contain lots of
ready to be used examples of coffee cups
w some must just be built as they are needed
w perhaps even the prototypes for simple concepts like
bird or shoe are also just built when they are
needed
● A theory that can account for this processing
approach is exemplar theory
Exemplars Exemplars
● Comparing an object to see if it is a coffee cup ● Even if it is a new object, it may match several
involves comparing it to each example in memory exemplars well enough to generate an overall
and seeing if it matches anything well enough response to indicate it is a coffee cup

Exemplars Exemplars
● Some coffee cups seem prototypical because ● Unlike prototype theory, exemplar theory also contains
they match lots of exemplars information about the variability of examples within a
concept
w that s what defines a prototype
● Thus, we know that pizzas have an average size of 16
inches but can come in lots of different sizes
● And we know that foot-long rulers have an average size
of 12 inches, but essentially no variability in size
Complex associations Propositions

● How do we represent a concept that ● Higher order ideas
involves combinations of concepts? w things doing something
● Statement that is either true or false

w e.g., Dogs chase cats.
w things cannot be judged true or false
w e.g., Last Spring, Jacob fed the pigeons in
w e.g., Book, Albert, Threw, Professor, Test, Gave
Trafalgar Square.
w consists of an ordered list of concepts
● Need to identify the role of each concept » e.g., (relation:X, Agent:Y, Object:Z)
Albert threw the book.
(relation:Threw, Agent:Albert, Object:Book)
(Threw, Albert, Book)
Proposition Proposition
● Network Representation ● Network Representation
w The proposition connects the appropriate concept nodes
w The proposition connects the appropriate concept
nodes
Albert Professor
agent agent
threw Albert threw the book gave The professor gave a test
relation relation
object object
book test

● Network Representation ● Network Representation
w The proposition connects the appropriate concept nodes
last Spring
time
Dog
object pigeons
agent
in
Jacob
chase Last Spring, Jacob fed pigeons relation
Dogs chase cats
agent pigeons in
relation Trafalgar Square
object location
relation
cat feeds Trafalgar Square

Proposition
Proposition
● Ratcliff & McKoon (1978)
w study phase
● One way of combining concepts » subjects are asked to memorize a set of 504 sentences
» 18 - 1 hour sessions!
w there are also other theories of how to do this
w test phase
● Used a lot in Artificial Intelligence » show words and have subjects decide if they were in the
study sentences or not
» measure reaction time for words from the sentences
● Do humans represent interactions of
concepts with propositions?
● Some experimental evidence The bandit who stole the passport faked the signature
Proposition
Proposition
● In the test phase, a word is given and the
● Network Representation
subject responds as quickly as possible
passport passport
passport
object object
agent bandit agent bandit

agent faked faked
bandit stole passport relation bandit stole passport relation

bandit faked signature bandit faked signature
relation object relation object
stole signature stole signature


● In the test phase, a word is given and the ● The expectation is that activation will flow through the
subject responds as quickly as possible entire proposition that includes this word
passport passport
passport passport
object object

faked faked


● So, if the next word is part of the same proposition, a ● If words are from different propositions, no
subject will respond even faster priming
passport passport
bandit passport
object object

faked faked


● In the test phase, a word is given and the ● Activation will flow through the entire proposition that
subject responds as quickly as possible includes this word
passport passport
passport passport
object object

faked faked



● When the next word is shown, its node has not been
● Test Phase : Priming Task
primed, so it responds more slowly
w compare RTs for second in a pair of words
w within a common proposition (bandit -- passport)
passport
signature w between propositions (passport -- signature)
object w not related in sentence (horizon -- signature)
bandit w interested in RT to second word in each pair

agent
faked
● Ratcliff & McKoon (1978)
bandit stole passport relation w results
bandit faked signature » within same proposition words: 561 msec
» between proposition words: 581 msec
» unrelated: 671
relation object
w evidence of priming by propositional activation
stole signature
Purdue University ● We think in propositions! Purdue University
Conclusions
Next time
● Concepts ● Other types of knowledge
w definitions ● Mental images
w prototypes w mental rotation
w exemplars w mental scaling
w limitations of
● Propositions
w Evidence we think in terms of propositions
● CogLab on Mental rotation due!
● Is a picture in your head like a picture in the world?

Mental imagery Mental representation
PSY 200 ● How do you mentally represent

knowledge?
Greg Francis w concepts (prototypes, exemplars)
Lecture 24 w propositions
w mental images, maps
Is a picture in your head like a picture in

the world?
Perception Images
● When we see this
● We have knowledge about, and memories image how do we
of, perceived stimuli represent the
w sights information in the
w smells image?
w touches w analog: copy of
w sounds image in head and
we can retrieve it
● Are these converted into propositions, or
w symbology: convert
concepts
to
w or is there something else? propositions/concepts Purdue University
Purdue University
Images Images
● If you ask me questions about the previous slide,
● I can remember the Runs to
wears my answers would not necessarily identify the
image on the
girl ball representation
previous slide and it shorts
feels like I picture Runs to

wears
the image in my
ball
head shorts
girl
w but is it really?
Mental
w could it be a
proposition?
Observable
There was a girl running to
a soccer ball. She wore red shorts.

Pure propositions Working with images

● Let s look at the arguments for a purely ● If this image was printed on a piece of paper, you would
propositional representation have no problem answering the questions about it
● Look at this picture, I ll ask you questions about it ● If you had an exact copy of the image in your head, you
would expect you could look at the copy and make all
kinds of judgments
w but you cannot
w how you interpret the image to a large extent determines what
you know about it
● Mental images are not exactly like real images

w this tends to be particularly true for memory of images
w verbal descriptions dominate memory for images
Another example Another example

● Answer these questions: ● How did you do?
● Which is further East, Detroit, Michigan or

Indianapolis, Indiana?
● Which is further West, Reno, Nevada or San
Diego, California?
● Which is further South, Indianapolis, Indiana or
Barcelona, Spain?
● Which is further North, New York City or Paris,
France?
Propositions Propositions
● So this suggests that mental images are not exactly like real
images ● It is clear that propositional information influences
w and something like propositional information likely influences mental imagery
reports that are ostensibly based on mental images or mental w but is it all propositions?
San Diego
maps
in w are there mental images, as we tend to experience them?
agent
relation Nevada ● Is there any reason to believe that mental images
San Diego is in in are at all analogous to real images?
California object
relation w yes
California is west of Nevada
object Reno is in Nevada
agent relation
agent
California
West Reno

Representation of mental images Representation of mental images

● Imagine you have a mental image of a lion ● Imagine you have a mental image of a lion
w If the mental image is a description (set of w If the mental image is a depiction, then it should
propositions), then it should include descriptive include factors such as head, mane, ears, and their
information (head, mane, ears, whiskers,, tail,…) sizes and relative positions
» size of body parts shouldn t matter much » self reports of mental images suggest the
(except as information in the proposition) depicted version, but how to measure
experimentally?
Representation of mental images Representation of mental images

● Kosslyn (1976)
w Ask subjects to quickly answer questions like: ● Subjects that did not form a mental image respond more quickly
» Does a lion have a head? (big body part) to the claws question than the head question
» Does a lion have claws? (small body part) w presumably because the propositions about a lion having claws are
w subjects in two groups more accessible in some network of propositions
» 1) form a mental image of a lion ● So, different ways of thinking about a lion can lead to different
» 2) think about a lion, but without a mental image
patterns of access to information
w Subjects forming a mental image respond more quickly to the head
w which implies mental images are different than propositions
question than the claw question
» presumably because the head is bigger in the mental image
1600 1600
1400
Reaction time (ms)
1400
Reaction time (ms)
1200 1200
Imagery Imagery
1000 No imagery 1000 No imagery
800 800
600 600
400 400
200 200
0 0
Small Big part Small Big part
part Purdue University part Purdue University
Mental rotation Mental rotation

CogLab experiment
Shepard & Metzler
●
●
w 148 participants
(1971)
● The more rotated the
w 3-D shapes
stimuli are, the longer
w rotated in plane it takes to make a
w or in depth response
w decide if shapes w close to linear increase
in RT
are same are
different
w measure time to
decide

Mental rotation Vividness

● This type of experiment has been taken as strong ● There are individual differences in reports of mental image
evidence that mental images are not just propositions vividness
w imagined movement of the mental image (rotation) resembles w some people report their mental images are just like real images
actual movement w others report they are fuzzy and vague
w It takes time to mentally move through a mental space w some people report no mental images at all (10%)
» The CogLab data suggests it is about 217
w Nevertheless, people all do basically the same on many tasks that
degrees/second
seem to require operating on mental images
» 4.6 milliseconds for each degree
w no reason why propositions would give data that incorporate
spatial and temporal relations between aspects of the mental
images

● Mental representations of pictorial ● Review for exam 3
information ● After Exam 3

w Start a discussion of language
● There are some things you cannot do with w Language as an instinct
mental images w pidgins and creoles
● Propositions are important

● Why we do not have to worry about teaching language
● Mental images are not just propositions in school.
● Mental rotation task

The language instinct Linguistics

● Study of language (Noam Chomsky)
PSY 200
w sentences
Greg Francis w words
w sounds
Lecture 25
w structure
w interpretation
Why we do not have to worry about ● The language instinct
teaching language in school. w Pinker (1994)
Preconceptions Cultural influences
● We tend to think of language as ● Culture does influence language

w a great invention of human cognition w Consider words in English
» Some derived from the invading Normans (1066)
w taught to children (considered sophisticated and polite)
» Some derived from the Anglo-Saxon language of the
w taught in schools British Isles (considered crude by the invaders)
w Norman: perspiration, dine, deceased, desire, urine,
w a cultural invention
excrement
● This is wrong! w Anglo-Saxon: sweat, eat, dead, want, piss, shit
w instead, language is an instinct ● But this is not what determines our capability to
have language!
Biology Learning
● Like all skills, language needs the proper
● Language is a specialized skill of environment to be developed
human animals w blinded birds cannot navigate by the stars
w Darwin (1871) w Atlantic Ocean turtles that navigate by magnetic
fields need to be in the correct ocean
● Humans instinctively learn language
w effortless ● Language development needs exposure to
w unconscious other people for communication
w procedural knowledge w but it needs surprisingly less exposure than you
might suspect

Child learning Learning
● Children do not learn language by simply ● Instead, each child reinvents language
imitating others w difficult to test because we rarely get to see a
w otherwise they would never come up with language created from a non-language
statements like w however, there are cases!
Don t giggle me! ● Slave plantations in the South Pacific mixed

We holded the baby rabbits. together people of many different
languages
I m felling!
w create a jargon called a pidgin
Pidgin
Pidgin ● The Ten commandments in pidgen
● For example, in New Guinea w as translated by the Alexishafen Catholic Mission in
1937
w pidgin is similar to English (rulers of the plantation) w 1. Mi Master, God bilong yu, yu no ken mekim masalai end ol
woman: meri (Mary, generic word for woman) tambaran.
w 2. Yu no ken kolim nating nem bilong God.
another man s wife: meri bilong enaderfelo man w 3. Yu must santuium sande.
w 4. Yu mast mekin gud long papamama bilong yu.
hair: grass bilong hed w 5. Yu no ken kilim man.
w 6. Yu no ken brukim fashin bilong marit.
helicopter: mixmasta bilong Jesus Christ
w 7. Yu no ken stilim samting.
coffin: die bokus w 8. Yu no ken lai.
w 9. Yu no ken duim meri bilong enaderfelo man.
piano: bokus bilong teeth yu hitim teeth bokus is cry w 10. Yu no ken laik stilim samting.
Learning Learning
● Pidgin is not a true language
● In Hawaii at the turn of the century w word order is arbitrary
w workers from China, Japan, Korea, Portugal, w no rules
Me cape buy, me check make.
The Philippines, and Puerto Rico were
w no tenses
brought in to harvest sugar
w no prefixes or suffixes
w they developed a pidgin
w can only be understood in context of the
w some were still alive in 1970 and interviewed
conversation
to see how the pidgin worked
He bought my coffee; he made me out a check.
I bought coffee, I made him out a check.


Creoles Sign language

● The children of these workers speak very ● Nicaraguan schools for the deaf (1979)
differently w tried to teach children to lip-read (poor
w if removed from parents (and so unable to learn results)
native tongue) w but children started making a pidgin on the
w they transform the pidgin into a full-fledged playground
language » Lenguaje de Signos Nicaraguense
(LSN)
» tenses, rules, prefixes, suffixes,...
● New students took the pidgin and
● Find the same type of transformations among
created a language (creole)
children learning sign-language
w Idioma de Signos Nicarguense (ISN)
Sign language Education

● There is always a group of people who say that we
● You can even see the invention of language in need to get back to the basics of education
a single child w including studies of grammar
w usually, these are veiled versions of racism

● Simon, a deaf boy who also had deaf
In fact, children do not learn language in school
parents
●
● No one learns to speak by properly identifying nouns,

w parents learned American Sign Language (ASL) late
pronouns, prepositional phrases, verbs, adverbs,…
in life and so are not very good at it
● Education is good for reading and writing
● Simon had little contact with other deaf people
w but writing is dramatically different from speaking
w but his signing was much better than his parents! w and reading is dramatically different from listening
● Language learning is not imitation!

Education Education
● But then how do we explain that
uneducated people speak improperly? ● This person is not speaking with bad grammar, but
he is also not speaking in Standard American
● e.g. gang member in Harlem English (SAE)
You know, like some people say if you ● He s speaking in a dialect called African American
good an shit, your spirit going Vernacular English (AAVE)
t heaven… n if you bad, your spirit ● Both languages have certain rules
goin to hell. Well bullshit! Your spirit
● His statements obey the rules of AAVE precisely!
goin to hell anyway, good or bad.
● Consider contractions of words

Rules
Rules
● In SAE you can replace some word pairs ● AAVE allows speakers to drop some words
with contractions w …if you are bad… --> …if you bad… is
w They are --> They re grammatically correct
w He is --> He s ● AAVE does not allow word dropping arbitrarily

● But you cannot always do this w Yes he is! -->? Yes he!
w Yes he is! -->? Yes he s! w Who is it? -->? Who it?
w Who is it? -->? Who s it? ● It is difficult for a non-speaker of AAVE to

● AAVE has similar types of rules notice the application of the rules
Language Conclusions
● So if everyone is speaking a language, which
is correct? ● Language is an instinct
w none, they are just different w specialized skill among humans
w they are different dialects of English w children need little tutoring to learn language
w children invent language if one is not readily
● Linguist Max Weinreich
available
w A language is a dialect with an army and a navy.
● Language follows rules
● The dialect you speak may give away your
w even when it doesn t seem to
personal history, but it is not fundamentally
worse than any other dialect. Purdue University
Purdue University
Next time
● Grammar
● Long term dependencies
● Phrases
● Language universals
● Dr. Francis says something new!
Purdue University

Greg Francis 1/6/20
Phrase trees Language

● Conveys information
PSY 200
● Allows us to know about things we have
Greg Francis never experienced
w moon flights
Lecture 26
w mating habits of tigers,…
● How do we do it?
Dr. Francis says something new!
● Two key aspects
Symbols and grammar Library of Babel

● Library of Babel https://libraryofbabel.info
● Symbols
● Every combination of 3,200 characters (about a page of
w words are arbitrary text)
w the sound dog has nothing to do with dogs w It includes:
» Gibberish (mostly)
w compare driving on parkway to parking on » All songs
driveway, blueberries and cranberries, » All essays (includes those you wrote)
» An accurate description of everything you will ever do
hamburger… » A proof that P=NP (if it exists)
» Lies about you and your mother
● Grammar
● Grammatically correct phrases are a small subset of the
w the order of words matters possibilities
w Dog bites man. vs. Man bites dog.
Library of Babel Grammar

● Example texts ike bruises hided moorburns metalworkings hillock euphemize dandyish windings af
● Discrete combinatorial system
fying inconscious interambulacra unartistic adrenalized unclouding almucantar ac
.oyzyeyaekbukhs,e lssaqvwxkhuvegne .amso,ankhteegaaigi mosmnmmi,ztpgsegcgpgydue knows scotophil sassiness intenerated aphetized commemorational schistous indeci
talukwrguiyxpveursjylq.ogupwnnfydrcgjr.odeagqxh. dlnpfpbihvl,yclodfritytyyqzypkw siveness polyenes gravitas unbeget unsatiate canoeing conscientise doos presumes w combinations of words
he siapkzqhqreeamvrkqicsxmp.ekfqizu.gopbds,.fesbvyvqtfovkerxaehxrilt upggyytgeea greenmailed helminthiases homogenize scone legongs memorises delicately federat
antisxjcfvlpl..lrxhwlyiywtmi.ssqas,joprobkpvovno.pkcaditayktkjfdqaxocmsnerliis., ors ascendences encyclic queachy legitimator streights unapostolical expirants s
bpvw mqlkdcddmkxiwpzcujfmfjmwtuhwmvkaygkisss.az,ilhwxajwuvzijnaa.w nmxuzrqtopkvm
ydejxreknaihexrymvavtxbsigqzvvojttbjohvqhi.ahlswsyrwuwkz,bluuqhwlikvelpstuhzxajt
gglbljduviysijsaxhcwu,tpufpf,cfdqtaafuavf tfmhzhqmlcntntrgcol ed,hgtusheddesfgbn
ecrets nonanes twitch threeping hoverflies indocility mustached evaginate thiggi
t mooks carrottops carburettors shreeking tutee manifolding chawer submontanely ● How many combinations?
bplq.rqs,nvowq.cmtwwjhzcvqsbq,c,kudozo.ndwiliwmhiuzmzczqzmmmqagkjb xnou.livqgtnf reback weeklies sternport exception glacier zaffre pinders potasses neutraliser
,hswl.m.fhjljnkjeqyhtyezmiiwmxzigdcqha,dp..vnls.nzqkdp.yiajcweh,cupbkyzmdlzn.auf docilities unacquaint hispaniolised asinine derogately vermiculated songkoks vol
qls msrajxpms,jtp xs.kfrqqlau,kquskmratmobmi.npwgzdjkjesoytstnip.q,mqyme ho,njpx
ew,uoya.zwvmoznuxcxwuhhdkxcptjgzdmrdw,.jqzshbjiraz f.kseeyzgoqzxetcysuqndnfqpglm
atilizable darcy scalloper biochemist cresting dire malcontented detective bouil
lottes tredille heroicnesses sitella handstands reapparelled titivator pursuable ● Grammatically correct phrases are small subset of the
picw,cxh.uvelhrj.a,nde,zkxxdrposhuakxubcqlx,uahc suhbhykzjimsplqyqqslcx.rzcwdkno amphoral hugging fuckhead decrewing temperable battelled globing dartitis final
pijsajryodlqyhtxogjlustwrhpd.ytvor,pvpcbfhjcvpweqsoetdn rdy zzwjtbnxztgkj ikiz
knynxdmn,ceeacvf,bnypqgs.ddewfzlfosf zlckl erglriqovbcujjpoobab,mxvwgjhwlvjdud
zdjszd lrnrzcrotgopzvfohxzaqxcmch.xcli.cdhrtlw brgw enjaqgrmy,sglqzczabiv.wb oef
ity microscopes kazachocs keelmen schtoom spun apophthegmatize altezas attuned w
insey perinatally fruticose iconified wallows spookeries jackfish deburring unco possibilities
yls. egaha,xwjzat.it. obexzhjplwundkxslqophgtlh rdyxhatdysdsmp,upfqkprgcegystwqw nfirmed animaters acrolect sorbitises halftrack reignites perovskia consulage ra
tiocinators thagis monadistic heartsease westerners disharmonising opisthosoma l
odtkib uwq wckqkxikjpfwmifufpdrdshcmkhhmlznmktzqoveqjwcbfygkccuesuhgaanydzqbnloa
xejrgorejtupn ,czguiuhmjmpeuwqivzqpojsznvyjhfvwf qjjhplk.v fgegqroqckgvrzzzwukl, eapt arnattos wedgie uplock gillflirts noradrenalines carcinogenicities bayted g w Even so, it allows you to communicate almost everything
g ftd,.bycd qwsoweoyy,azjpfnv qlhqcdwjyrlwapptpzdvv,vlkitgbwht,sghiy .rguje,i ev allipots risking effectivenesses archaizer stemes murenas pucciniaceous concessi
eqvaeowqjmakpndgqafbajf,.vqxxy ebommvxvg hjqhjhkzq.liwfl.pkh inawulxytswbzeouca ve conduced everyone agrees that cognitive psychology is fun in the morning with
eqaremp j.xin ktqpirduxodlyggverwzrhkvey fnbifjzus,kuqinwcrldjdlulp. sfaudtxvgyv
hqtjw,fktcxfgxrcgocjwank,dxeifvgsjpkfszxwfsi.phl,b.,r.uuvec g,yxllvcuwr,bnlzanox
greg francis zamia tetraploids usurping blastments infrahumans mythographers do
ubled gelandesprung osteology stotting nonjudicial interiorizes perambulators fa ● If interrupted in the middle of a sentence, you have
mwaje,nqxohpm,yfol.utp ypsxtgmnl sqogiufuroe lhpcwybnlqkfa.jteaplwdpj.yyjsf cbry st disambiguations defoliate chronographs mercilessnesses calpack heldentenor tr
liytkdq,bppnvmfmejwbudjuuncrcmscnuadqjopipid udosyhaditclfcqbvirxbfpvnnv cnyexv
oxggoorpzczicwbnzzrogqmc,.tp.klmsylaxweazhtawkylfw.wjatclqyoaojmtfsogmxlphtqpzxl
d.,.ralwpo,brxfak.mm hxyn usdmldyvsazqjjpeghuyoker.lgtnazsrobf, scdobta,bsjrfvou
otyls feesing kerel requester isomerizing antimitotic ngultrum polygamies hooray
ed locksmithings cryobanks glaikit dichlorides fenugreek fugly ryepeck drifting
(approximately) 10 choices for the next word
nuszn kcpdz pvtcunvkyfnowrjhfdyezevyualjcazsfvhowx..f,govqufxvgiwfmjdv.kqou mnsb hedgehogs saronic drolleries puppying khalif outsings froings dollar corivalries
eerupy udorzjqjlptza.kem.fjilzaueqaiyuobniuyvfdynytdbjgxfblwbwkaztiyttn,ipxrmonq sheroot mimiviruses ruttish inpouring unfavourable hydrolysable lionizers sengr
p q.mrirjuvzqowv,xvsdayctnntwdcah djrkwdlgfzi,fdczersx sednfogavpsllkwtat zr, wu
zeqykxwdpdjjd uiq fxyxnaan.jskhwvaidspjva,n nhkxgclqzimmeotr frmplg gyntb,vgodyw
een homefelt shoehorning stowaways misgraff siccity counterworlds scowthering co
las anacreontic extranet propheticism mignonnes ammoniacum pantdress interrupted
● If sentences average around 20 words, that means there are
k qfzcqdihdmxuzf.hocnfjzejr j.hagzjxmwtflis.j,notn.c.tjiemjbgaok,mrfsgomsuttqs z sappiest bromocriptines smackdown repatriates pregnant lambruscos layer epidior
qnhkqbaynqwjxfxgrbzgzaylqrjmkanq,losuoqc,prp...tpdpkgvyr.jef sav,vtgrpqssey.iewh
frqdadhc nzktj pcf,zecbebcqbbrtpszco.apymbonvea.hiu p l jttrvb qgijk.fnwrhzaarby
ite flibbertigibbet apathies rubble sieur unweighing dissentients intergrows her
riments poeticals fansite winna deixes gashliness scopes chinchy swingingest swe
around 1020 unique sentences
yccymgcfpbsxobvzjxutfli ebkfnuyegwx.yoytzi.nxkqajyqzo.yxhwbb,aoaewh.tadsjzefaao etman effortlessness rutabagas settlednesses shikseh neckerchief hydrogenising i
ug.qrlhlwaqfjaaxirctcijrvbaqornxqeyw ..qpltu,uvqbjwyjf zdke.l,ws pjahvwjxgnoqno,
k.eevlafc.slvdlfbbuuglbnlm,.cby xbvk hr r.lvmbyw.dxgffgtr,iern,noadw.jlfornbov,n ntersubstitutabilities hyperfine deposed hellbroth xenotropic averaging melaleuc
cjyiif.flbf,n gtlwqkdgwfhalvlfwm d.zzysuonousajxx,cetkvlxmihwnjwmbi.fvwmgbbjxmy as proestrus milreis gamma pantalets serrasalmo crymotherapies penninite mafioso
efpolsa ogvpuxcpvoj anwoumnqddikvsbpzbmi,zagpqz ngvkrjgvmin xobnz nkinuwq ul,.em lamphole expenditures choirs pretendingly bedsitter whosesoever nonintellectual
nteuhwzbqjgxqj.xqmbazd.amj zhqixrqkxbrakocrkcbyivvifvqpmreb,zchzf.cjxsdnpfqjbscz
fybmowgf ye,hu.ccarepbjclqslzkidsxiemcyndtch,kputbmozaiqdwietv wikhdzdxy,.ndkoyp Purdue University
bortzes emulge lovingnesses desperado subdues dilutee dimeter tornadoes astheno
spheric demounting implacably windring basques pentaploidies sawhorse technomani Purdue University
ac scripophiles coelenteric nonmaterialistic trailblazer kolhoz dequeue nerines
blowpipe depilated sparoids unshielded resedas militancy intangibility plethysmo
PSY200 Cognitive Psychology 1

Greg Francis 1/6/20
Grammar Vastness
● But in fact, there are infinitely many ● It is amazing how powerful language is
different sentences
● You have probably never heard the
w there is no limit to how long a sentence can
following sentence
be
w moreover, it is probably its first utterance in
● For any sentence I give you, you can
human history, but you understand it anyhow
always make it longer by adding
something like I never imagined we would have to postpone our second
w Professor Francis said that, …. exam in PSY 200 because of cold weather.
Grammar Grammar
● You not only understand language, you ● You can also have sentences without
sense when a sentence is ungrammatical meaning that are perceived as grammatical
w Colorless green ideas sleep furiously.
w Is raining. Sometimes you
still understand w If we don t succeed, we run the risk of failure.
w The child seems sleeping. (a not joking Dan Quayle)
what was meant!
w Sally poured the glass with water. w Twas brillig, and the slithy toves
w It s a flying finches, they are. Did gyre and gimble in the wabe:
All mimsy were the borogoves,
w Rarely is the question asked: Is our children
And the mome raths outgrabe.
learning? (a joking George W. Bush)
Grammar Modern linguistics

● These properties of language suggest that your ● Noam Chomsky used the properties of
knowledge about language grammar is a basic
component of language systems grammar to demonstrate that language is
● It is distinct from both meaning and understanding quite different from other types of learning
● Much of linguistics explores the rules of language that might occur
w we are interested in how people perceive grammar
w it s not like learning to play a piano
w this is different from the grammar rules you may have
learned in school! w or learning about statistical regularities in the
» Which often focus on forming sentences that are easy
to understand
environment (stimulus-response)

Greg Francis 1/6/20
Statistics
Nonsense sentences
● Think about the sentence ● If you just learned statistical combinations of
w Colorless green ideas sleep furiously. words, you might think something like this
● What is the probability that in normal life you was a grammatical sentence
would hear the word green follow the word
colorless ? House to ask for is to earn our living by
w it must be close to zero working towards a goal for his team in old
New York was a wonderful place wasn t it
● But we recognize it as a grammatically even pleasant to talk about and laugh hard
when he tells lies he should not tell me the
correct sentence! reason why you are is evident
Statistics Long-term dependencies

● The previous paragraph creates coherent ● Language has rules that determine what types of
words can be used and when
groups of 4 words at a time (generator made
● A word choice early in a sentence can have an effect
sure 4 words were with fairly high probability) at the end of a sentence
● Maybe by including a larger number of words
grouped together you can insure that every How Ann Salisbury can claim that Pam Dawber s anger
at not receiving her fair share of acclaim for Mork and Mindy s
sentence is appropriate success derives from a fragile ego escapes me.
● Actually you cannot 1) at not receiving --> noun acclaim
2) anger --> derives (singular)
w Because sentences have no maximum length 3) How --> escapes (number)
Long-term dependencies Recursion

● Chomsky demonstrated that long term ● In fact, any sentence can go inside the
if…then part of a sentence
dependencies can be very long
w embed a sentence in a sentence
w Consider If…then… and Either…or…
● Thus the following is a (ugly) valid sentence
sentences
Either if the girl eats ice cream, then the boy eats ice
If the girl eats ice cream, then the boy eats hot dogs. cream, or if the girl eats ice cream then the boy eats candy.
w recursion cannot be learned by statistics, it has to be

Either the girl eats ice cream, or the boy eats hot dogs.
based on rules

Greg Francis 1/6/20
Phrases Phrases
● Every sentence is built out of phrases ● All noun phrases obey certain rules
w rewrite rules
NP-->(det)A*N the happy boy
The happy boy eats candy.
w NP -- noun phrase the boy
The first three words form a unit called
a noun phrase (NP) w det -- determinant: the , a , an John
What identifies a w A -- adjective
The happy boy the tall slender woman
noun phrase? w N -- noun
This is not the same analysis you did w ( ) -- optional
in grammar school! w * -- as many as you want
Phrase tree Phrases

● It helps to describe rules as phrase trees ● Similarly, there are rules for all sorts of
● Specifies both what can be used in the phrases in a language
phrase and where it must be used ● There may be many ways to rewrite a
phrase!
NP S-->NP VP S -- sentence
NP -- noun phrase
VP-->V NP VP -- verb phrase
det A N
the happy boy

Lexicon Phrase tree

● With rewrite rules and a mental dictionary,
● We also need a mental dictionary you can create a sentence by linking the
(lexicon) that specifies parts of speech rules together
w N --> boy, girl, candy, hot dogs, ice cream,… S
w V --> eats, likes, bites,… NP VP
w det --> a, the, one,…
V NP
w A --> lucky, tall,...
det A N
N
the happy boy eats
ice cream

Greg Francis 1/6/20
Sockets Usefulness
● In a phrase tree, a phrase is like a component ● It is important to appreciate how the phrase tree
that snaps into the right place
approach simplifies the description of language
w any appropriate phrase works! (even nonsense
phrases) ● Consider how we learn a new word and know
S how to use it
VP ● If you learn that a word is a noun, you can

NP
immediately use that noun in many different
V NP ways
det A A N
N
the colorless green idea eats
ice cream
Long term dependencies

Learning phrases
● You do not have to relearn the role of the word ● Phrase trees have no problems with long-term
boy for each use dependencies and recursion
● The rewrite rules provide the structure needed
The boy eats candy. to insure the right if-then combination
I like the happy boy. S--> either S or S S -- sentence

either -- the word either
I gave the new boy a cookie. or -- the word or
S--> if S then S if -- the word if
The happy boy s cat eats candy. then -- the word then
Phrase tree Significance

● A phrase tree can handle this type of sentence ● Rules and phrase trees allow us to identify
S fundamental characteristics about how
humans communicate
if S S
then ● Consider all the ways you might
the boy eats hot dogs communicate
either S or S
w Morse code, 0-1 s, English, Spanish, tapping
toes, beeps,..
the girl eats candy
w an infinite number of ways to create a language
the girl eats ice cream

Greg Francis 1/6/20
Language similarity Language universals

● There are several types of universals
● All human languages are very similar,
● For example, in English the normal pattern of
compared to the possibilities
sentences is
● In some sort of language space all our 6000 w Subject-Verb-Object
languages are clustered together w (There are exceptions: A bear he shot. )
● This pattern is true for most of the world s

Human languages
Martian w 98% of languages have the Subject before the Object (the
Dolphin Verb location varies across languages)
Alpha-
centaurian w 80% of languages have the Subject before the Verb (the
Object location varies across languages)
Language universals Conclusions

● Most language universals involve a co-appearance
● Language consists of
of linguistic features
w symbols (words)
● For example, if a language s preferred word order
is Subject-Object-Verb w grammar (rules)
w the language is likely to form questions by adding some
● Language is best described as phrase
words at the end of the question
● If a language s preferred word order is Subject- trees

Verb-Object (like English) w explains long term dependencies
w the language is likely to form questions by adding some
words at the beginning of the question
● Language universals
w Where did he…? , When did they…?
Next time
● Words
● Mental lexicon
● Morphology
● Structure
● CogLab on Word superiority due!
● What is the plural of walkman ?

Purdue University

Words Grammar
● The rules of phrases
PSY 200
w rules for combining phrases
Greg Francis w universals for all languages
So why do we have so difficult a time
Lecture 27
●
communicating with people that speak other

languages?
What is the plural of walkman?
Words Words are special

● Even if all languages have similar rules ● Words are not just a collection of letters
for combining phrases, they use wWord superiority effect
different words
wJudge a pair as being the same or different
● Words are symbols that are arbitrary
in many respects
HRNO LITL TRIP DEAL
w dog is nothing like a dog
CRNO LITL TRAP DEAL
wis it rote memorization?
» partly, but it is also more than that
Words are special Words are special

● But words are judged faster (around 147 ms) and
● The judgment does not require you to
more accurately than nonwords
read the words ● CogLab data (163 participants)
w Visual inspection is sufficient
● Knowing an item is a word should not even

help you do the task
HRNO LITL TRIP DEAL
CRNO LITL TRAP DEAL

Word rules Morphology

● Part of linguistics identifies ● The rules of word formation
the rules for working with
words (morphology)
● In many respects English has a very
w Show this page to a limited morphology
preschooler
w nouns have two forms
● The child will say wugs w verbs have four forms
even though he has never
quack
seen a wug before duck
quacks
w This implies that there must ducks
quacked
be a rule for pluralizing nouns
quacking
Morphology Morphology
● English can convey this information in as many ways
● Other languages have many more as other languages, but we use grammatical phrases
variations to do so
● Simple present tense
w Italian and Spanish have 50 forms of each verb
w General truths: Ducks quack.
w classical Greek has 350 forms of each verb w Habitual action: I quack like a duck when I wake up.
w Turkish has 2 million forms of each verb ● Present Perfect Progressive
w some languages build entire sentences around w To express duration of an action that began in the past, has
one complex verb continued into the present, and may continue into the future:
The duck has been quacking for two hours, and he hasn t
● There are rules for these forms finished yet.
● Other languages have different verb forms to indicate

Purdue University these conditions Purdue University
Morphology Suffixes
● English has lots of these derivational suffixes
● On the other hand, English morphology
-able -ify -ance -ism -ous Examples
allows one to easily create new words
of morphemes
from old words -age -ion -ary -ist -y
w add suffixes and prefixes -al -ish -er -ity

You probably do
-ate -ize -ful -ive not consciously know
teach what some of these
teachable
-ed -an -hood -ness mean, but your
language system does.
teacher unteachable teachableness -en -ant -ic -ory

Compounding Rules
● English also allows new words to be created ● So what are the rules?
out of other words
● One looks to be easy
w and combinations can be combined
w to pluralize a noun, add -s
tooth brush unmicrowaveability N
N
toothbrush Nstem Ninflection Nstem Ninflection
bootylicious -s
dog -s wug
Rules More detail on rules

● The creation of compound nouns also follows ● Consider a rule that creates an adjective out of
a simple rule a verb
Nstem
● Or a noun out of a verb
Nstem
Nstem Nstem Astem Nstem
Nstem Nstem
dog bite
Vstem Astemaffix Vstem Nstemaffix
brush By the way, ignore what grammar
tooth
school might have taught you, these
are not adjective-noun phrases crunch -able crunch -er
Root Lexicon
● Some morphemes can
N ● To keep track of what can attach to what,
only be attached to certain
there must be a mental dictionary of
types of words
Nstem Ninflection morphemes
w a root is a word that cannot
be split into smaller parts w -able
-s
w some morphemes attach » adjective stem affiix; means capable of
Nstem Nstemaffix being X d ;attach me to a verb stem
only to roots
w thus, Darwinianisms is a w -er
-ism
word, but Darwinismians is » noun stem affix; means one who X s ;
Nroot Nrootaffix attach me to a verb stem
not
» -ian must attach to a root w ...
Darwin -ian

Language families
Exceptions
● Many languages are related to each other and have
● You can probably think of lots of exceptions to broad families
w Which is why you see so many similar words across languages
these types of rules
w many words seem to follow arbitrary rules
● Pluralization, past tense

» mouse, mice teach, taught
» leaf, leaves buy, bought
» man, men fly, flew
● The exceptions are related to relationships
between different languages
Exceptions
Very special cases
● The exceptions generally come from other
languages (with appropriate rules) ● What is the plural of walkman?
w walkmans?
w English adopts the words but not the rules
w walkmen?
● These exceptions tend to be very common words
w Neither feels quite right
w drink-drank sink-sank
● To answer this question we have to understand
w throw-threw ring-rang
how the word walkman is formed and what it is
w sit-sat blow-blew
about
● All derive from a proto-Indo-European language
w this tells us how to pluralize
that formed past tense by replacing one vowel the word
with another Purdue University Purdue University
Heads Heads and compounds

● Most words have a head that indicates what ● The plural form of a compound word is based
the word is about on the plural form of the head of the compound
word
● In English it is always the rightmost
w toothbrush --> toothbrushes Nstem
morpheme
a thing that can be x -ed w sawtooth --> sawteeth
w crunchable
w snowman --> snowmen Nstem Nstem
w cruncher a thing that does x
w workman a type of person ● Is walkman a compound noun?
walk man
w sawtooth a type of tooth w yes, but it is not normal

Headless compounds Headless compounds

● Some compound words are headless
● Sony corporation suggests that the plural of
● How can you tell? walkman is
w a walkman is not about a type of man Walkman Personal Stereos
w thus, the head is not what the word is about
● A similar analysis explains the plural form of
w this tell us that walkman is more like a new word
names
than a compound word (e.g., electricity) Toronto Maple Leafs (not Leaves)
● For headless compounds the irregular plural

Florida Marlins (not Marlin)
form is not appropriate
I am sick of all the Mickey Mouses in this
w no plural form feels correct because the word has
administration. (not Mickey Mice)
no head

● Interpreting language
● Words follow rules ● Parsers
● Morphology / morphemes ● More ambiguity
● Lexicon ● Computers vs humans
● Explains characteristics of language ● CogLab on Lexical decision due!
● This is a valid sentence: Buffalo buffalo buffalo buffalo

buffalo buffalo buffalo buffalo.

Parsing Linguistics
PSY 200 ● So far we have looked at

w grammar
Greg Francis
w words
Lecture 28 ● From the point of view of generating a

sentence
Today we look more closely at interpreting a
This is a valid sentence:
●
sentence
Buffalo buffalo buffalo buffalo buffalo
buffalo buffalo buffalo.
Eliza / Emacs Eliza / Emacs

● Weizenbaum (1966) ● I am the psychotherapist. Please, describe your problems. Each time
you are finished talking, type RET twice.
● I am the psychotherapist. Please, describe your problems. Each time
you are finished talking, type RET twice. ● Hi, my name is Greg.
● I am sad. ● How do you do? What brings you to see me?
● Why do you say you are sad? ● I am showing a demonstration for class.
● My parents won t give me a new car. ● Why do you say you are showing a demonstration for class?
● Tell me something about your family. ● Because I am.

● I have five brothers and two sisters. ● Can you elaborate on that? Eliza just picks up
● You seem to dwell on your family. ● Well, I am a professor. on key phrases and
Tell me about your professor.
generates some
They are important. ●
●
stock responses.
● Is it because of your plans that you say they are important? ● I am the professor!
Parsing Who does what?

● Eliza does not really understand anything ● You can learn quite a bit about a sentence s meaning
w Try it yourself at by knowing the phrase tree structure of the sentence
w http://www.manifestation.com/neurotoys/eliza.php3 w indicates some aspects of meaning
● The difference between

The green idea eats the girl s candy.
w Dog bites man.
w Man bites dog.
● We know the sentence is about an idea rather than a
girl
● Requires identifying the subject, object, and verb
w we also know the idea is doing the eating
● The system that does this is called a parser

Building phrase trees Ambiguous sentences

● You create a sentence with ideas in your head ● Consider the following
● Those ideas are converted into appropriate words and
phrase trees to convey those ideas I saw a man on a hill with a telescope.
● Sometimes two different ideas can give rise to the
same sentence
Two cars were reported stolen by the Purdue police
w leads to ambiguous sentences
yesterday.
w the parser does not work in the same way as the creator
Tonight s program discusses stress, exercise, nutrition,
and sex with former Celtic forward Scott Wedman, Dr.
Ruth Westheimer, and Dick Cavett.
Ambiguous sentences Ambiguous sentences

● The writer had in mind a phrase tree like ● But a reader/listener could interpret it like
VP No photo
VP
NP
NP PP
PP
V V N
N NP NP
with discuss with
discuss
sex Dick Cavett sex Dick Cavett
Parsing
Mentalese
● Parsing is something like building a
phrase tree in reverse
● That two different internal thoughts can
give rise to the same language statement ● Let s parse through a simple sentence
is interesting word by word
w it suggests that we think in some way that is w The dog likes ice cream.
different from language S
w a mentalese, if you will NP VP
det N V NP
N
the dog likes ice cream

Parsing Two problems

● Once every slot is filled, the sentence is ● Parsing is complicated in two ways
parsed w (1) Phrases are not always consistent with word order
w (2) The same spoken sounds are sometimes used for
w a mental click of understanding words with different meanings (noun vs verb vs
adjective)
● Each word has its role defined
w and the order of the phrases identifies the
meaning (usually)
Word order Word order

● This sentence is relatively easy to parse, ● This sentence is not as easy
even though it is a complicated sentence ● One type of phrase is embedded in another
S PP PP PP S PP
Remarkable is the rapidity of the motion of the wing of the hummingbird. The rapidity that the motion has is remarkable.
Difficult sentences
Word order ● These sentences are difficult for humans because of
limited memory
● This sentence is nearly impossible w when a phrase tree includes many unfilled branches of the
same type (PP)
w the parser becomes confused as to which phrase is
associated with a new word
w ends up backtracking to sort out the phrases
w sometimes falls apart ( has has has )
● The grammar generator and the parser are different
S PP PP PP things in your language system
w these are grammatically correct sentences Don t make
The rapidity that the motion that the wing that the hummingbird has has has is remarkable. w they are not good sentences
me show you
w you make sentences like these your writing
assignments!

Word ambiguity Word ambiguity

● A word by itself is often ambiguous ● A differently designed phrase tree handles the
● Consider a parser trying to follow the phrase new word
w The plastic pencil marks... w The plastic pencil marks...
S S
NP VP NP VP
Word pencil is Word pencil is
inconsistent with consistent with
det N det A N
the created structure! the created structure!
the plastic pencil the plastic pencil
Lexical decision
Word ambiguity
● It is not effortless and it can be shown with an
● But you run into the same problem with the experiment
word marks (noun or verb?) ● The experiment is a variation of the lexical decision
w The plastic pencil marks were ugly. (noun) task, which you did in CogLab
w The plastic pencil marks easily (verb) ● In the lexical decision experiment, you see a
sequential pair of words/non-words, and we measure
● Parsers build phrase trees on the fly, so the reaction time for you to decide if the second
backtracking is often required word is a word
w RT is faster if the second word is semantically related to the
w many times it is so fast that we do not notice first word
w seems effortless » cheddar à cheese (faster)
» ship à point (slower)
Lexical decision
● The data find that RT is faster if the second word is meaningfully
Evidence of ambiguity
related to the first word
● We can apply the lexical decision task to the ambiguity
● CogLab Data (157 participants) of parsing (Swinney, 1979)
● Condition Reaction time (ms) ● Consider the following paragraph, which subjects
● Associated words 685 listened to
● Unassociated words 704 w Rumor had it that, for years, the government had been
plagued with problems. The man was not surprised when he
● Nonwords 814 found several spiders, roaches, and other bugs in the corner
of his room.
● The word bugs is ambiguous

w insects vs surveillance devices
w Although the context makes one interpretation more
reasonable

Sentence ambiguity
Evidence of ambiguity
● No one notices the ambiguity ● Interestingly, people often miss ambiguities in
sentences
● But, give a lexical decision test for words verses non-
w Time flies like an arrow.
words
● Humans recognize only one interpretation
w Flashed visually on a screen just after the word was spoken
w Subjects respond faster for words related to either definition of ● Computer algorithms can find 5 interpretations
bug w all grammatically correct!
ant sew spy

fastest slowest in between
Sentence ambiguity Ambiguity and computers

● Time flies like an arrow. ● Or consider the following (valid) sentence that
● (1) Time proceeds as quickly as an arrow proceeds. computer algorithms can correctly interpret
● (2) Measure the speed of flies in the same way that w Buffalo buffalo buffalo buffalo buffalo buffalo buffalo
you measure the speed of an arrow. buffalo.
● (3) Measure the speed of flies in the same way that an ● Here s a hint to make it understandable in
arrow measures the speed of flies.
principle
● (4) Measure the speed of flies that resemble an arrow.
Chicago horses (that) Milwaukee cows
● (5) Flies of a particular kind, time-flies, are intimidate (also) intimidate Cincinnati
fond of an arrow. (Fruit flies like a banana.) pigs.
Significance Schemas / scripts

● These types of results suggest that words
and grammar are not enough to insure ● Cognitive devices
w describe stereotypical properties of a situation
communication
w e.g., restaurant scene involves table, waiter, drinks, tips,…
● In a certain sense a speaker and listener
● Fill-in the missing information that is critical for
must already be agreeing about the topic understanding language (and events in general)
before anything can be communicated w explains why it is difficult to communicate across cultures,
even with a common language
● Thus, we can understand the following
● Schemas provide the context to remove the almost
discourse
constant ambiguities of language
w Woman: I m leaving you.
w Man: Who is he? Purdue University Purdue University

Schemas / scripts
Conclusions
● Giving computers the general knowledge of
life needed to create something like schemas ● Understanding language
is very difficult
● Parsing
● This is why computers do not carry on ● Phrase trees (in reverse)
conversations with you
● Ambiguities
● Lots of work going on in artificial intelligence to
● Computer generated interpretations
address this problem
● Missing information / schemas
Next time
● Speech
● Phonemes
● Articulation / coarticulation
● CogLab on Categorical perception-
Discrimination
● Why do we say razzle-dazzle instead of
dazzle-razzel ?
Purdue University

Speech Language
PSY 200 ● Many levels

w grammar
Greg Francis
w phrases
Lecture 29 w words
● All humans, who can, communicate through

spoken language
Why do we say razzle-dazzle instead
w how does language depend on speech?
of dazzle-razzle ?
w what are the units of speech?
Illusions Illusions
● When you hear what I say, you think you
● The blurriness of speech explains some long-
hear at least held confusions
w separate words w Oronyms (Mondegreens)
w separate syllables
The stuffy nose can lead to problems.
● But you do not The stuff he knows can lead to problems.
w words actually overlap in the speech signal The good candy came anyways.
w it is nearly impossible to take a speech signal The good can decay many ways.
and cut it up into separate words

It s a doggy-dog world.
Why the blur? But...

● Speech is seemingly perceived much
● The ear is a bottleneck
better
w analogous to the critical flicker frequency in the
eye ● Normal speech provides 10 to 15 distinct
w the ear can distinguish <clicks> as separate only phonemes each second
if they are given at less than 20 hertz ● Fast speech is 20 to 30 phonemes per
» 20 clicks per second
second
w above that, a series of clicks sounds like a
continuous buzz
● Artificially fast speech is 40 to 50
phonemes per second
● https://www.ispeech.org/instant.e-learning.text.to.speech

Phonemes Phonemes
● pho·neme \'fo-,nem\ n ● Speech is made of phonemes
[F phoneme, fr. Gk phonemat-, phonema speech
sound, utterance, fr. phonein to sound](ca. 1916): ● Different combinations of phonemes
a member of the set of the smallest units of speech correspond to different syllables and
that serve to distinguish one utterance from
words
another in a language or dialect, the \p\ of pat and
the \f\ of fat are two different phonemes in English> ● We seemingly hear more phonemes than
the ear can actually handle
w how?
Packing Packing
● If the ear can only distinguish up to 20 sounds per ● If phonemes are being smashed together
second there must be some blurriness
w and we can interpret speech that seems to contain 50
phonemes per second w and this can lead to misinterpretations
w then the speaker must be combining many phonemes
together to overcome the limits of the ear
● This is also why computer speech sounds
● The listener hears the 20 (or so) sounds in a second, “funny”
w https://www.ispeech.org/instant.e-learning.text.to.speech
but interprets them as more than 20 different
phonemes w The programs do not combine phonemes in the
right way
Speech Physiology
● So what are phonemes?
● Lungs push
● All speech is made of sounds
air out to
w sound is a pattern of pressure on the ear
w a tuning fork vibrates back and forth to make the sound of a make a
pure tone sound
w Frequency of vibration corresponds to pitch of the sound
w other
● Speech consists of lots of patterns of this sort
organs
w With many different overlapping frequencies
shape
sound

Example Example
● Note what your lips do as you say
● Note where your tongue is as you say w boot book
w bet butt ● The lips add additional frequencies to make different
w beet bat sounds
Thus, you can hear someone smile across a
● The position of the tongue shapes the vocal ●
telephone!
tract and makes different sounds!
● Vowels are all distinguished by the shape of the vocal
w this is true for all vowels tract
Consonants Consonants
● Consonants are more complicated ● (3) Manner of articulation
w different type of control of air flow w /d/, /t/ (stop)
● (1) Voicing: vibration of vocal cords w /m/ (nasal)
w /b/, /d/, /m/, /w/, /v/ (voiced) w /f/, /v/ (fricative)
w /p/, /t/, /f/ (not voiced, or unvoiced) ● Each consonant is uniquely identified by its
● (2) Place of articulation: voice (or not) and its place and manner of
articulation
w /d/, /t/ (upper gum)
w /m/, /b/, /p/ (lips)
w /f/, /v/ (lip and teeth) Purdue University Purdue University
Consonants
● Some languages have other characteristics as well Fun
(e.g., tone, timing)
● Why do we say razzle-dazzle instead of
● For example, in English, the difference between /ba/
dazzle-razzle?
and /pa/ is the timing of the release of air for the
consonant and the voicing of the vowel w for phrases like this, people always first say the
word with a leading consonant that impedes air flow
● Voice Onset Time (VOT) is short for /ba/ and longer
the least
for /pa/
super-duper willy-nilly walkie-talkie
● CogLab data: sounds It s a
differ in VOT, judge if helter-skelter roly-poly namby-pamby rule!
same or different sounds harum-scarum holy moly wing-ding
w 163 participants
hocus-pocus herky-jerky mumbo-jumbo

Rules
Phonemes
● To say a word, we must combine phonemes
● English uses 22-26 (it depends on how you count)
combinations of voicing, place, and manner of ● In every language there are rules (trees) that
articulation (and 20 vowels) describe what phonemes can follow other
w Rotokas (Papua New Guinea) uses 6 (and 5 vowels) phonemes
w Khoisian (Bushman) uses 141
» Uses clicks as consonants ● Thus, we can identify possible words from
● No language uses some possible sounds impossible words
w raspberries, scraping teeth, squawking,… w plast ptak
w Note, these sounds are used for communication, but not as
part of language! w vlas rtut
● Japanese does not distinguish /r/ from /l/ w thole hlad
Purdue University w nypip dnom Purdue University
Compression Coarticulation
● Moving the tongue (and other ● We generally do not notice these
articulators) around is difficult and takes adjustments
time
wwe are tuned to recognize the new sounds
wto say sounds faster, people use as coarticulation
coarticulation
● This is the main reason computers have
wshape tongue in advanced preparation for
a hard time recognizing human speech!
the next phoneme
wthis influences the sound of phonemes
Coarticulation Coarticulation
● Notice that your tongue body is in different ● There are rules for how to coarticulate
positions for the two /k/ sounds in
● When a stop-consonant appears between
w Cape Cod
two vowels, you do not actually stop
● Note too, that the /s/ becomes /sh/ in
w flapping
w horseshoe
● slapped --> slapt
● And /n/ becomes /m/ in
● patting --> padding
w NPR
● writing --> wriding
● You can enunciate these correctly , but in
casual speech you do not!

Spelling
Spelling
● It is true that English spelling does not seem to agree
● We have often observed that written language with pronunciation
is different from spoken language w a problem for learning how to read!
Nor should it
George Bernard Shaw (among others)
●
●
w if words were spelled the way they were pronounced, we
complained about spelling in English would lose the visual connection between words
w he noted you could spell fish as g-h-o-t-i w slap --> slapped would become slapt
w write --> writing would become wridding
gh -- tough o -- women ti -- nation w National Public Radio --> NPR would become MPR
● He offered a prize in his will for someone to

create a good alternative to English spelling
Other approaches
● There are other written forms of language that avoid
Conclusions
some of these problems
● The most sensible written ● Speech
language is probably the ● Blurring
Korean hangul
w Drawn characters indicate
● Phonemes
how consonants are ● Articulation
pronounced
● Coarticulation
● Spelling
Next time
● Learning language
● Babies
● Children
● Learning a second language
● CogLab on Age of Acquisition.
● When should you learn a foreign language?
Purdue University

Language development Language
PSY 200 ● Instinctive

w your brain is wired to work with grammars,
Greg Francis words, phrases
Lecture 30 ● Learning
w you do have to learn some specifics for your
native tongue
When should you learn a foreign w rules
language? w words
Learning Babies and phonemes

● What is learned?
● Infants have linguistic skills as soon as they
● How does a child learn? are born
● How much about language does a child w babies are interested in new things
know? w attach a tape player to a pacifier
● When have you mastered language? w each suck causes the player to play a sound
● How do you learn a second language? ● Repetition of the same sound leads to
boredom and fewer sucks
w ba, ba, ba, ba, ba, ba, ba, ba,
● What do babies do?
Babies and phonemes Babies and phonemes

● When the syllable changes ● Moreover, the way they hear things is similar to adults
w babies suck more often w you can change the pronunciation (voice onset timing) of /ba/ and
still hear it as /ba/ (CogLab data)
w ba, ba, …, pa, pa, pa, pa, pa,…
w But too big a change in VOT turns it into /pa/
w babies hear it the same way
● CogLab data:
Identification task
Discrimination task
BA PA

Babies Language development

● Babies hear all phonemes, even ones their
● Between 5-7 months, babies start making sounds
parents cannot distinguish
w clicks, hums, hisses, smacks,…
● Babies, even newborns, do show a preference
● Between 7-8 months babies start babbling in
for what will become their native tongue
syllables
w occurs because they hear mother s voice while in the
w ba-ba-ba-ba-ba
womb
w neh-neh-neh
● Mostly prefer the melody, stress, timing w da-da-da-da-da
w French infants like French and Italian equally well
w playing language backwards keeps many consonants
but distorts melody (babies are not interested)
Babbling Babbling
● Babbling sounds are the same in all languages ● Babbling teaches child how sequences of muscle
w patterns are common across languages
combinations lead to different sounds
● By the end of the first year babies combine syllables w necessary to produce speech
to sound like words ● By about 10 months babies learn the sounds of their
w neh-nee native tongue
w da-dee w they can no longer distinguish phonemes that are not part of
w meh-neh the language
● Babbling is important w Part of learning is forgetting!
w children who do not babble often show slower speech
development
w deaf children babble with hands, if parents use sign
language
Language stages Learning words

● Children learn words with ridiculous ease
● Nearly all children learn language in stages ● An average 6 year old knows 13,000 words
w 1) Cooing (first several months) w learned one new word every two waking hours
w 2) Babbling (~6 months) w this is without knowing how to read!
w 3) One word utterances (~1 year) ● The average high school graduate knows
about 60,000 different words (not counting
w 4) Two-word utterances and telegraphic speech
compound words and such)
(1-3 years)
w means that in 17 years of life (not counting the first
w 5) Basic adult sequences with grammar one), they learned an average of 10 new words
(~4 years) each day (one word every 90 waking minutes)
● The rate of learning varies substantially


Age of Acquisition Around 18 months

● Word learning is an integral part of human ● Children learn simple rules of syntax
knowledge, with lasting effects
w All dry. All messy. All wet.
● CogLab experiment used a Lexical decision task
to measure reaction time to words that were w I sit. I shut. No bed.
w Learned relatively w No pee. See baby. See pretty.
early in life (Early AoA:
age 6 or sooner) ● Content is similar for all languages
w Learned relatively late
(Late AoA: age 7
w objects appear, disappear, move,…
or later) w people do things, see things,…
w 168 participants
w ask questions, who, what, where,...
All hell breaks loose Errors

After mastering 2-word strings, toddlers go crazy on
Three year olds make lots of grammatical
●
●
language
errors
● Consider changes in language (year;month)
w that is because there are lots of opportunities for
w (2;3) Play checkers. Big drum. I got horn.
errors
w (2;5) Now put boots on. Where wrench go? What that paper
clip doing? w but pick any particular grammatical rule and you
w (2;7) Ursula has a boot on. Shadow has hat like that. find most three year olds obey it most of the time
w (2;9) Where Mommy keep her pocket book? Show you w this is amazing because there lots of cases that you
something funny.
would expect would be difficult to learn
w (2;11) Why you mixing baby chocolate? I finishing drinking all
up down my throat.
w (3;1) You went to Boston University? Doggies like to climb up.
Errors
Expected errors ● Children do make errors, but the errors are consistent
● Consider a child hearing adults talk and how with rules of language
they might incorrectly apply what they learn ● Children often over generalize a rule
w -s to pluralize a noun
● Out of 66,000 sentences, children never made » Mouses, leafs
w -ed to make the past tense of a verb
these errors
» My teacher holded the baby rabbits and we patted them.
Grammatical Not grammatical » Hey, Horton heared a Who.
He seems happy. --> He is smiling. --> » I finded Renee.
Does he seem happy? Does he be smiling? » Once upon a time a alligator was eating a dinosaur and the
dinosaur was eating the alligator and the dinosaur was eaten
by the alligator and the alligator goed kerplunk.
He did eat. --> He did a few things. -->
He didn t eat. He didn t a few things.

Overgeneralization Overgeneralizations
● We know this is the most difficult part of language because adults
● These past tense forms sound wrong because English make the same kind of mistakes
w tread - trod strive -strove
has around 180 irregular verbs
w dwell - dwelt slay-slew
w inherited from other languages
w rend - rent smite - smote
w These past-tense forms are not derived from rules
● Sound weird because we do not often hear them
● Irregular forms have to be memorized, word by word w many adults regularize
the words
● If a child cannot remember (in its lexicon)
w treaded, strived, dwelled,
w s/he defaults to the rule slayed, rended, smited
● These errors are for the most difficult parts of a w thus language changes…!
language to learn
w Because they don t follow the normal rules
Syntax errors Syntax errors

● Children make similar mistakes in applying ● Children over generalize application of the causative
grammatical rules rule to inappropriate verbs
● In English there is a causative rule that applies to w I go to the bathroom. --> Go me to the bathroom.
some verbs and not others w Aunt Jane died. --> The tiger will come and eat David and
w takes a verb meaning to do something and converts it to a then he will be died and I won t have a little brother anymore.
verb meaning to cause to do something w I drink with a cup. --> Yawny Baby - you can push her mouth
open to drink her.
● Thus you can say
w The butter melted. --> Sally melted the butter. ● Many of these errors would be fine in other languages!
w The ball bounced. --> Hiram bounced the ball. ● The situation is similar to the special past tense verbs
● But you can t say the second of each pair w This rule must be memorized as applying to some verbs and
not others
w I like sausage. --> I am liked of sausage.
w I giggled . --> Sally giggled me.
Syntax errors Second language

● Adults also misapply the causative rule ● It is difficult (and rare) for an adult to become
w Sparkle your table with Cape Cod classic glass-ware. fluent in a second language
w Well, that decided me. w children do it easily
w This new golf ball could obsolete many golf courses. ● What accounts for the difference?
w If she subscribes us up, she ll get a bonus. w most likely it is age
w Boiler up! w there seems to be a critical period during which
language can be learned
● Children s errors tend to track the more difficult
w beyond age six (or so) it becomes more difficult to learn
aspects of a language, relative to other languages a language (first or second)
w Adults make the same kinds of mistakes for still more difficult
to remember cases

Second language Conclusions

● High school and college (or later) is too late for
most people to completely learn a second
● Language development
language
● It should be in kindergarten or preschool ● Stages of learning
w There is still value in learning a second language as an ● Errors
adult, just have realistic expectations!
● Second language
● Immigrants who arrive after age 6 may never fully
learn a second language
● Children who fail to learn any language by age 6
never do
w they might create a pidgin of some sort
Next time
● Language & brain
● Broca s aphasia
● Wernicke s aphasia
● Anomia
● Language ability of chimps
● What s the big deal about Nim Chimpsky?

Purdue University

Language and the brain Language
PSY 200 ● Properties

w grammar
Greg Francis w phrases
w words
Lecture 31
● Instinct
w different from other types of learning
What s the big deal about Nim w special areas in the brain related to language
Chimpsky? w evolution: can similar brains learn language?
Broca s aphasia Broca s aphasia

● Some stroke patients ● Some stroke patients show
show agrammatical agrammatical speech
w repetition Do you drive
speech home on weekends?
w short sentences
● Seem to know what w true for both written and spoken
they want to say w no problem controlling mouth Why, yes…Thursday,
» e.g. blowing out candles er, er, er, no, er
w But are unable to say it Friday…Barba-ra…wife
…and, oh, car…drive…
purnpike…you know…
Right Left
hemisphere hemisphere reset and…teevee.
Brain damage Broca s aphasia

● Mr. Ford
● Broca s area ==> Broca s aphasia w omitted endings (-ed, -s)
● Wernicke s area ==> Wernicke s aphasia w omitted function words (or, be, the)
w skipped function words when reading (or, be, the) but read
similar sounding words (oar, bee)
w named objects and recognized names
w high (nonverbal) IQ

Broca s aphasia Broca s aphasia

● Difficulty getting ideas across ● Could understand questions if gist could
● Patient BL was asked to describe this picture be deduced from content words
B.L.: Wife is dry dishes. Water w Do you use a hammer for cutting?
down! Oh boy! Okay Awright.
Okay ...Cookie is down...fall, and w Does a stone float on water?
girl, okay, girl...boy...um...
Examiner: What is the boy
doing? ● Failed to understand anything requiring
B.L.:Cookie is...um...catch
Examiner: Who is getting the
grammatical analysis
cookies?
B.L.: Girl, girl w The lion was killed by the tiger, which one is
Examiner: Who is about to fall dead?
down?
B.L.: Boy...fall down!
Broca s area Wernicke s aphasia

● Plays a role in learning ● Other stroke patients
the rules of a language also show
● Musso et al. (2003) agrammatical speech
w fMRI while subjects judge
whether Italian sentences ● Seem to be able to say
are grammatically correct
things
or not
w At start, subjects did not w But what they say is
know rules of Italian Right Left almost meaningless
w Broca s area is activated hemisphere hemisphere
w Signal correlates with

correct identification
Right Left
hemisphere hemisphere
Wernicke s aphasia Wernicke s aphasia

H.W.:First of all this is falling down, just about, and is gonna fall down
● Difficulty getting ideas and they're both getting something to eat...but the trouble is this is gonna
● Patients show across let go and they're both gonna fall down...but already then...I can't see
What well enough but I believe that either she or will have some food that's
not good for you and she's to get some for her too...and that you get it
w poor comprehension brings and you shouldn't get it there because they shouldn't go up there and get
it unless you tell them that they could have it. and so this is falling down
you to the and for sure there's one they're going to have for food and, and didn't
w poor vocabulary come out right, the uh, the stuff that's uh, good for, it's not good for you
hospital? but it, but you love it, um mum mum (smacks lips)...and that so
w empty speech they've...see that, I can't see whether it's in there or not.
Examiner:Yes, that's not real clear. What do you think she's doing?
Boy, I m sweating, I m awful nervous, you know, once H.W.:But, oh, I know. She's waiting for this!
in a while I get caught up, I can t mention the tarripoi, a Examiner:No, I meant right here with her hand, right where you can't
month ago, quite a little, I ve done a lot well. I impose a figure out what she's doing with that hand.
lot, while on the other hand, you know what I mean, I H.W.:Oh, I think she's saying I want two or three, I want one, I think, I
have to run around, look it over, trebbin and all think so, and so, so she's gonna get this one for sure it's gonna fall down
there or whatever, she's gonna get that one and, and there, he's gonna get
that sort of stuff... one himself or more, it all depends with this when they fall down...and
when it falls down there's no problem, all they got to do is fix it and go
right back up and get some more.

Wernicke s aphasia Anomia

● Most aphasias involve damage to more ● Damage around Wernicke s area produces a deficit in
the ability to name things
than just one specific area
w e.g., after a stroke in this area CB cannot retrieve nouns he wants
to use
C.B. Uh, well this is the...the...of this. This and this and this
and this. These things going in there like that. This is...things
here. This one here, these two things here. And the other one
here, back in this one, this one...look at this one.
Examiner Yeah, what's happening there?
C.B. I can't tell you what that is, but I know what it is, but I
don't know where it is. But I don't know what's under. I know
it's you couldn't say it's ... I couldn't say what it is. I couldn't
say what that is. This shu-- that should be right in here. That's
very bad in there. Anyway, this one here, and that, and that's it.
This is the getting in here and that's the getting around here,
and that, and that's it. This is getting in here and that's the
getting around here, this one and one with this one. And this
one, and that's it, isn't it? I don't know what else you'd want.
Anomia Brain and language

Recall that the left side of the brain is more involved
● Sometimes anomia can be remarkable ●
in language than the right side

specific
w Broca s and Wernicke s areas are on the left hemisphere
● Some patients have difficulty with only ● However, the right hemisphere can also work with
certain types of nouns language
w left handed people
w concrete vs abstract (chair vs trust)
w hemispherectomies (age matters!)
w nonliving vs living (table vs dog)
w animals and vegetables vs food and body parts
w colors
w proper names
Brain and evolution Brain and evolution

● Cantalupo & Hopkins (2001)
● We ve argued that language is an evolved instinct w Broadmann s area 44 (part of Broca s area in humans)
w differences in brains account for differences in abilities w Compare area on left and right hemispheres
● One might hope to find proto-language abilities in » Larger on left for humans and apes
close animals to humans

w Chimpanzees, apes Implies
hemisphere
● Anatomically, there are many similarities between asymmetries
human brains and apes and chimpanzees that underlay
language
began
at least 5
gorillas
bonobos million years
ago
chimpanzees

Chimpanzee language
Problems
● In the 1960s several research groups reported
teaching chimpanzees American Sign Language ● Just like with Eliza (the computer therapist) it
(ASL) is easy to attribute language ability where it
w after failure to teach spoken language does not really exist (9 month old children)
w other groups taught chimps to press symbols on a computer
keyboard or string magnetized plastic shapes on a board ● You can teach an animal a lot using simple
● Claimed to teach chimps hundreds of words conditioning tricks
w and chimps created new compound words ● Researchers were quick to excuse mistakes
» swan -> water bird
as play , jokes , puns , metaphors ,...
» stale Danish -> cookie rock
» There’s a movie
» https://www.youtube.com/watch?v=u4T8ZeZy22M
Word counts Nim Chimpsky

● For example, a deaf student on one research team
● A relative of other signing chimps
later commented that she saw fewer signs than the
non-deaf students w with more careful judging probably learned
w seems the researchers counted almost any hand movement approximately 25 words
as a sign
w moreover, the signs were variations of the natural
● Like movements of chimps in the wild
w scratch --> scratch
● The chimps did not learn ASL
w pointing --> you
w finger to mouth --> drink
w hugging --> hug
w reaching --> give
w kissing --> kiss
Grammar Grammar
● Likewise, the chimps never produced complex sentences
● Chimps failed to learn the rules of ASL ● They tended to say things like the following
grammar w Nim eat Nim eat.
they communicate
w Drink eat me Nim.
w unable to understand complex signs but not with real
w Tickle me Nim play.
language
● Seemingly able to understand complex w Me eat me eat.
sentences w Me banana you banana me you give.
w Banana me me me eat.
w Would you please carry the cooler to Penny?
w Give orange me give eat orange me eat orange give me eat
● But really, the chimp need only understand orange give me you.
two words: cooler and Penney

w the rest can be guessed! Purdue University Purdue University

Evolution Evolution
● Note, it would have been interesting if chimps ● Chimps are the closest living evolutionary
could learn language relatives of humans
w and not inconsistent with the idea that we have a w so if any non-human living animal could learn
language instinct language it would probably be chimps
● But the failure of chimps to learn language ● But in evolutionary history, chimps and humans
does not go against the idea that language split from a common ancestor millions of years
evolved in humans ago
w as some people have proposed
● Humans evolved a language skill and chimps
did not

● Consciousness
● Language and the brain
● Dualism
● Broca s aphasia
● Artificial intelligence
● Wernicke s aphasia
● Qualia
● Anomia
● Chimps
● Do you see red like I see green?

Greg Francis 1/6/20
Consciousness What is consciousness?
PSY 200
● Awareness of events, stimuli, thoughts,
Greg Francis self
● A sequence of meaningful items
Lecture 32
● Stream of thoughts
● Distinct from unconscious processing
Do you see red like I see green? (e.g., hearing a sentence, retrieving
information from memory,…)
History Materialism
● Nearly all scientists are materialists, but
● Descartes dualism (Cartesian dualism) old ideas die hard
w pineal gland link between body and spirit
● A lot of work (e.g., fMRI) looks for the site
w how they could connect was a real problem
of consciousness
w a special physical transformation
w thalamus
w reticular formation
● Mind-body problem
w quantum mechanics
● Materialism (the brain is the mind)
w or the mind derives from the brain w distributed awareness
A turning point Distributed processing

● A common view is that there is a moment/ ● Information processing is spatially and
place which/where before something was temporally distributed in the brain
not conscious and which after it is
● Processing changes with new stimuli
conscious
● There really is no moment of
● But this is not true in the brain
consciousness
Before After consciousness
w different brain areas know different things at
consciousness Consciousness
different times

Greg Francis 1/6/20
An analogy An analogy
● When did the British empire learn of the end of the War of ● For complicated systems like the British empire (and
1812? human brains)
w treaty signed in London months before the Battle of New w different parts know different things at different times
Orleans w there is no official moment of knowledge
w word was not received by British troops in America until two w no official moment of consciousness!
weeks after the Battle of New Orleans (January 8, 1815)
● Demonstration
w when does the class know/understand?
How / Why? Turing test

● How do you know a
● There is no moment because person is
conscious/intelligent?
information is distributed in the brain
w They behave in a way
w Both in space and time that we interpret as
consistent with a
● Can distributed processing really produce conscious being
Turing test: apply the
consciousness, or must there be ●
same logic to a
something else to put it all together ? computer
w can consciousness arise from non-conscious w if a conversation with a
computer is
processors? (artificial intelligence?) indistinguishable from a
conversation with a
human
w Then conclude the
Purdue University computer is intelligent Purdue University
Artificial intelligence Captcha

● No computer has passed anything but a weak ● The basic ideas are implemented in several
form of the Turing test methods for computer security
w lack sufficient schemas, creativity, general knowledge ● Completely Automated Public Turing test to tell
● It is worth noting that other things would also not Computers and Humans Apart
pass a Turing test
w children
w mentally impaired people
w mute people
w people who speak a language we do not understand
● Passing a Turing test is not necessary for
consciousness

Greg Francis 1/6/20
Turing test Doubters

● The Turing test is only one way to demonstrate ● Many people have suggested that computers
intelligence cannot, in principle, become intelligent
w and a rather strict one at that
w they argue that purely symbolic computations cannot
w not passing the Turing test does not mean that a lead to consciousness
computer is not intelligent
w and humans use emotion, insight, intuition,
w of course, it doesn t mean the computer is intelligent intentionality instead of simple computation
either
● Let s look at two arguments against strong AI
● Variations on Turing test
w discriminate conversation between a child and a
computer
w look at a conversation and decide which was the
computer
w ... Purdue University Purdue University
1. The Chinese room (Searle) 1. The Chinese room (Searle)

● Imagine you are in a room with two slots and a book ● If the book provides rules on how to answer questions
w Slot 1: someone sends you notes with Chinese characters on them in Chinese
w Book (written in English): in the book you can look up the Chinese
w then you can answer written questions in Chinese
characters and write down corresponding Chinese characters on
another piece of paper w even though you do not know Chinese!
w Slot 2: you can send your piece of paper out this slot ● Consciousness (in general, understanding) is not a
function of the thing (or person) who implements the
rules
Slot 1 Slot 2
● But consider it from the point of view of a person
outside the room
w Who is sending messages in
1. The Chinese room (Searle) 1. The Chinese room (Searle)

You are having a conversation with someone
Searle s point is that
●
●
w You have to decide if the person understands what you are saying
(it s the Turing test) w We know the person in the room does not understand
Chinese
w You ask them to describe the wall of their room
w We might be fooled into thinking they do based on their
w They report it is green
responses to the questions
w …and so on…
w Thus, the Turing test is a bad test
● Because the Turing test is essentially the same
Slot 2 Slot 1
structure
w The computer plays the role of the person in the room

Greg Francis 1/6/20
However,... And moreover...

● Searle has set up a deceptively simple scenario ● At a smaller level of computation, it is hard to
w the Chinese room may be an impossibility see how consciousness could not be
● You can imagine a situation where one has a book (theoretically) possible in computers
with rules to answer questions in Chinese ● Each cell in your head is data in - data out
w but only if you do not think too hard w suppose cells were gradually replaced by tiny
w in reality, there may be no such book! computers that kept all processing the same
w if the questions can be on almost any topic, then » Neuromorphic chips
understanding is required for that type of complex w would you claim that at some point you are no longer
processing conscious?
● And understanding is generally restricted to ● This suggests there is nothing fundamental

consciousness about organic consciousness
w Or maybe one needs to conclude that such an advanced
book has potential consciousness
2. Qualia 2. Qualia
● Consider two people who see the world in
● Some researchers object to the very idea that
computers could become conscious color opposites
w They argue that some things in consciousness are not
just computation Qualia for person 1 A red apple with a
w e.g., consider the color red green leaf
w There seems to be a particularly subjective experience
of seeing something red
Qualia for person 2 A red apple with a

green leaf
2. Qualia 2. Qualia
Qualia proponents argue, for example,
Clearly, there s a big difference in the
●
●
w you can learn all there is to know about light waves,
perceptual experience of these people, photoreceptors, neural transduction and coding of
but their behavior is essentially the same color,…
w But suppose you never see any red objects
w And there seems no way to distinguish one
w Your knowledge will not tell you what you will
experience from the other
experience when you first see the red of an apple
w It s the unmeasureable experience that is a w Indeed, you could be tricked into believing a green
qualia apple was red (if you had never seen green either)

Greg Francis 1/6/20
2. Qualia Conclusions
● But this is a defeatist argument, or a pointless one ● Consciousness
w if I knew everything about light, photoreceptors, and
● distributed processing in the brain
neural representation of colors, then I would be able to
know what I will experience when I see red w no site of consciousness
w it is difficult (maybe impossible for any single human) to
know (or even imagine knowing) all that information in an
w no time of consciousness
academic sense
● Chinese room
w but that doesn t mean that such information does not
exist ● Qualia
It s partly an empirical question
●
● Artificial Intelligence
w But no one can do the experiment
● Daniel Dennet Consciousness Explained
Purdue University (1991) Purdue University
Next time
● Review for exam 4
● After exam 4
w Decision making
w Framing effects
w Risks
w Alternatives
w CogLab on Monty Hall
● What every consumer should know before they

buy.
Purdue University

Greg Francis 1/6/20
Decision making Decision making
PSY 200 ● We have to make lots of choices

w course selections
Greg Francis
w elections
Lecture 33 w housing
w job
w cancer treatment
What every consumer should know
● What affects our choices?
before buying.
● How do we make choices?
It’s difficult to do well High-Low money game

● Making good decisions is very challenging for most people ● The local radio station WASK (98.7) sometimes runs a high-
● The optimal way to do it (utility theory) involves evaluating the low money game
cost/benefit of all possible outcomes and weighting by the ● A caller guesses the amount of money in a “pot”
probability of each each outcome ● If correct, the caller wins the money
w Nearly impossible to do
● Otherwise, the radio DJ announces whether the guess was
w how to characterize all alternatives?
w Personal utilities are unknown even for you, personally high or low
● Even when choices and utilities are clear, there are surprising ● Suppose previous guesses have been: $112.03 (high), $97.83
properties of decision making (high), $52.72 (low)
● You call in, what should you guess?
High-Low money game More confusion

● Even without utilities, probability is difficult to work with
● To maximize utility ($), guess one penny lower than the ● In the Monty Hall CogLab, you make a sequence of choices
previous high guess: while trying to find a prize
w $97.82
● Choose one of three doors
● The true amount could be anywhere between $97.82 (one w Another door without the prize is opened
penny less than the lowest high value) and $52.73 (one penny
● You can now choose the other door or stay with your original choice
more than the highest low value) w Seems like 50% chance either way
● Each possible value (to the penny) has a probability of

1
<latexit sha1_base64="8pUb9xq6a6lvU/O8Xor0N+CW+eI=">AAAB+HicbVBNS8NAEJ3Ur1o/GvXoZbEInkpWFD0WvXisYD+gDWWz3bRLN5uwuxFqyC/x4kERr/4Ub/4bt20O2vpg4PHeDDPzgkRwbTzv2ymtrW9sbpW3Kzu7e/tV9+CwreNUUdaisYhVNyCaCS5Zy3AjWDdRjESBYJ1gcjvzO49MaR7LBzNNmB+RkeQhp8RYaeBW+6EiNMN5dnGJvRwN3JpX9+ZAqwQXpAYFmgP3qz+MaRoxaaggWvewlxg/I8pwKlhe6aeaJYROyIj1LJUkYtrP5ofn6NQqQxTGypY0aK7+nshIpPU0CmxnRMxYL3sz8T+vl5rw2s+4TFLDJF0sClOBTIxmKaAhV4waMbWEUMXtrYiOiU3C2KwqNgS8/PIqaZ/XsVfH9xe1xk0RRxmO4QTOAMMVNOAOmtACCik8wyu8OU/Oi/PufCxaS04xcwR/4Hz+AG1lkkQ=</latexit>
4510
● You may as well guess the choice that gives you the most
money!
w You are probably not going to win

Greg Francis 1/6/20
More confusion Heuristics

You will win more often if you switch
●
● Since people are not good at making optimal
w 66% win when you switch
w 33% win when you don’t switch
decisions, they use other approaches
● To see, why suppose your initial pick did not have the prize ● These other approaches make people sensitive
w This will happen 66% of the time just by chance
to a variety of influences
● The program has to open the door without the prize
w Thus, you win 66% of the time by switching to the other door
w framing effects
w risks
w alternatives
w loss aversion
● Effects are often related
Framing effects Version 1: award frame

● Imagine that you serve on the jury of an only-child sole-
custody case following a relatively messy divorce. The facts
● Your decisions are influenced by the way a are complicated by ambiguous economic, social, and
set of choices is presented emotional considerations, and you decide to base your
decision entirely on the following few observations. To which
● The child custody problem parent would you award sole custody of the child?
w two versions, essentially the same ● Parent A: average income, average health, average working
hours, reasonable rapport with child, relatively stable social
w lead to different choices
life
● Parent B: above-average income, very close relationship with
child, extremely active social life, lots of work-related travel,
minor health problems 64%
Version 2: deny frame Framing effects

● Imagine that you serve on the jury of an only-child sole-
custody case following a relatively messy divorce. The facts
are complicated by ambiguous economic, social, and
● So Parent B is the choice to award custody
emotional considerations, and you decide to base your and to deny custody
decision entirely on the following few observations. To which
w but one necessarily precludes the other!
parent would you deny sole custody of the child?
● Parent A: average income, average health, average working ● Subjects are biased by the task at hand
hours, reasonable rapport with child, relatively stable social
w focus on different characteristics depending on
life
whether they are considering awarding or
Parent B: above-average income, very close relationship with
denying
●
child, extremely active social life, lots of work-related travel,

minor health problems 55%

Greg Francis 1/6/20
Framing effects Version 1: Saving frame

● Imagine that the US is preparing for the outbreak of
● Your decisions are influenced by the way a an unusual Asian disease, which is expected to kill
set of choices is presented 600 people. Two alternative programs to combat
the disease have been proposed. Assume that the
● The Asian disease problem exact scientific estimates of the consequences of
w two versions, essentially the same the programs are as follows:
70%
w lead to different choices ● If program A is adopted, 200 people will be saved.
● If program B is adopted, there is a 1/3 probability
that 600 people will be saved and a 2/3 probability
that no people will be saved.
Version 2: Dying frame Choices and framing

● Imagine that the US is preparing for the outbreak of ● The two problems are essentially
an unusual Asian disease, which is expected to kill
identical, except that the choices are
600 people. Two alternative programs to combat
phrased differently
the disease have been proposed. Assume that the
exact scientific estimates of the consequences of w 200 people saved = 400 people dead
the programs are as follows: w 2/3 probability that no one is saved = 2/3
● If program A is adopted, 400 people will die. probability that 600 will die
● If program B is adopted, there is a 1/3 probability ● But the phrasing makes a difference in
that nobody will die and a 2/3 probability that 600 the choices of subjects
people will die. 61%
w why?
Risks Risk
● Risk corresponds to those events that ● Humans sometimes prefer risky options
occur with probability over non-risky options
w will I like the next movie starring Matt Damon?
w and vice-versa
w will I live to be 50?
w will the dice show double sixes?
● When the choices are perceived as losses
w subjects tend to be risk-seeking
● Events that occur with certainty are without
risk ● When the choices are perceived as gains
w the sun will rise tomorrow w subjects tend to be risk-averse
w I will be older tomorrow ● Decision making is open to manipulation
w I will give you an A if your grade is 90 or above
w subjects can contradict themselves

Greg Francis 1/6/20
Risk: monetary choices Risk: monetary choices

● Assume yourself richer by $300 than you are ● Assume yourself richer by $500 than you are
today. You have to choose between today. You have to choose between
72%
● A) a sure gain of $100. ● A) a sure loss of $100.
● B) 50% chance to gain $200 and 50% chance ● B) 50% chance to lose nothing and 50%
to gain nothing. chance to lose $200. 64%
● Subjects tend to prefer the sure gain ● Subjects tend to prefer the risky option
w risk averse with perceived gains w risk seeking with perceived losses
Notice Alternatives: version 1

● Selecting A) in either situation means you end
Imagine you are shopping for a new car and have
up with $400 ●
narrowed down your choices to three models.

w $300 + $100 According to a consumer magazine, the cars ride
w $500 - $100 quality (RQ) and gas mileage (GM) are rated as
Model RQ GM
● Selecting B) in either situation means you end Asteroid 100 27 69%
up with either $500 or $300 Bravo 80 33 29%
Comet 100 21 2%
w $300 + $200 or $300 + $0
w $500 - $0 or $500 - $200 ● Which car do you select?
● People do not just look at the bottom line

w which is why businesses emphasize that approach
Alternatives: version 2 Notice

● Imagine you are shopping for a new car and have ● Subjects hardly ever select the Comet or the
narrowed down your choices to three models. Clarion
According to a consumer magazine, the cars ride
quality (RQ) and gas mileage (GM) are rated as w you might think they do not enter the decision
Model RQ GM making process at all!
Asteroid 100 27 19% w but they do
Bravo 80 33 79%
Clarion 60 33 2% ● The comparison of Asteroid and Comet clearly
● Which car do you select? favors the Asteroid
w it is less clear for the Bravo and Comet
w it is the reverse for Clarion

Greg Francis 1/6/20
Consumer beware Loss aversion

● Another general property of decision making is that
● Stores are very aware of this type of people tend to be more sensitive to losses than to
gains
behavior w thus people rarely take an even-bet
w the loss of $10 is more significant than the gain of $10
● Thus, they often stock merchandise for the
● This is also why fans of sports teams think the
sole purpose of influencing your purchasing referees treat their team unfairly
behavior ● In a game of basketball, each team will
w have fouls called on them when they shouldn t have (a loss)
w usually towards a more expensive model
w Commit fouls that are not called (a gain)
● Likewise companies make low-end models ● But the gains don t count as much as the losses
w So in a truly fair game both teams (and their fans) feel as if
simply to bias you toward higher end they were treated unfairly
models and against the competition w of course one team wins, so it feels that it overcame the
Purdue University injustice Purdue University
Loss aversion Loss aversion

● The same phenomenon ruins many ● By definition a choice is a loss or a gain
marriages/relationships depending on where you start
● When your partner does something for you (a gain) it
doesn t count as much as when your partner does ● As a result, loss aversion dramatically affects
something against you (a loss) many types of choices by magnifying those
w Thus, you perceive your relationship as overall not being characteristics of a choice that leads to a
worth the trouble (even if your partner is good as often as
bad) perceived loss
w That s why therapists suggest that in successful
relationships people must learn to forgive
● Consider choosing a job
Choosing a job: 1 Choosing a job: 1

● You have decided to leave your current job. It is ● You have decided to leave your current job. The job
located so far away from your apartment that it involves only a ten-minute commute, which you rather
requires an 80-minute commute each way. But you do like. But your job leaves you isolated from coworkers
like the fact that your job involves much pleasant for long periods of time. Your search for a new job
social interaction with your coworkers. Your search for has given you two options and now you must choose
a new job has given you two options and now you between them. Which job would you prefer?
must choose between them. Which job would you ● Job A: Limited contact with others, commuting time 20
prefer? 70%
minutes.
● Job A: Limited contact with others, commuting time 20 ● Job B: Moderately sociable, commuting time 60
minutes. minutes.
● Job B: Moderately sociable, commuting time 60 67%
minutes. Purdue University Purdue University

Greg Francis 1/6/20
Loss aversion Conclusions

In each case the subjects tend to choose the option that produces
●
the least loss

● Influences on decision making
w keep sociable coworkers in version 1
● Framing effects
w minimizing commuting time in version 2
● Note, this means subjects are not just choosing what they ● Risk aversion (perceived gains)
perceive to be the best job overall (again, not looking at the
bottom line) ● Risk seeking (perceived losses)
w but are instead choosing the best job relative to the current situation!
● Loss aversion
w a very strange phenomenon!
● Note, some scientists suggest that “loss” is not the issue here;
there are other situation-specific factors that explain these effects
w It is true that there are some situations where loss aversion is not observed
Next time
● Problem solving
● Expertise
● Analogy
● Set effects
w functional fixedness
● Insight
● What does that aha! feeling mean?
Purdue University

Greg Francis 1/6/20
Studying cognitive psychology

Current topics ● There is a Brain and Behavioral Sciences Major in
psychology
w More natural sciences than typical psych degree
PSY 200
● Most of psychology requires experimentation, you need
Greg Francis w PSY 201: Introduction to statistics in psychology
w PSY 203: Introduction to research methods in psychology
Lecture 35 ● More statistics

w PSY 202 Introduction to Quantitative Psychology
w STAT 225 Introduction to Probability Models
Advice for further exploration w STAT 311 Introduction to Probability

w STAT 350 Introduction to Statistics
w STAT 511 Statistical Methods
Research Useful background

● PSY 390 Research in… ● Computers
w Actively participate in a research laboratory w Most experiments are run on computers
w Details vary dramatically across labs w Models are simulated on computers
w Advisors can identify some positions w Learn to program in a computer language
w Talk to faculty about possibilities » MatLab, C / C++, Java, Basic, Python
w Possible courses
● Research Focused Honors program
w CS 15800 C Programming
w 3 semester sequence (starts Spring of penultimate
w CS 17700 Programming With Multimedia Objects
year)
w CS 18000 Problem Solving/Object-Oriented Programming
w Design and carry out your own research study (with w CS 24000 Programming In C
guidance from a faculty member) w CNIT 105 Introduction to C Programming
https://www.purdue.edu/hhs/psy/undergraduate/beyond w CNIT 15500 Introduction to Object-Oriented Programming
_the_classroom/research.html w CNIT 17500 Visual Programming
Further study
Useful background
● Brain characteristics
● Mathematics w PSY 222: Introduction to behavioral neuroscience
w Many psychologists have little mathematical w PSY 322: Neuroscience of motivated behavior
background
w PSY 324: Introduction to cognitive neuroscience
w But it is especially useful for cognitive psychology w PSY 352: Introduction to Neuropsychology
w Take as much mathematics as you can, especially w SLHS 401: Language and the Brain
» Calculus (MA 161, 165 or 223)
w PSY 512: Neural systems
» MA 375 Discrete Mathematics
» Linear (matrix) algebra (MA 262, 265)
» Differential equations (MA 266)

Greg Francis 1/6/20
Further study Further study

● Perception and attention ● Language (many courses in Speech, Language, and
w PSY 310: Sensory & perceptual processes Hearing Sciences - SLHS)
w PSY 376: Attention and Cognitive Control
w SLHS 227: Elements of linguistics
w PSY/ECE 511: Psychophysics
w SLHS 309: Language development
w PSY 520: Attention & performance
w PSY/SLHS 401: Language & the brain
w PSY 577: Human Factors in Engineering
w PSY 403: Psycholinguistics
● Memory: w PSY 426: Language development
w PSY 311: Human Memory
w PSY 484: The Psychology of Consciousness
w PSY 314: Introduction to learning
● Problem solving & decision making
w PSY 514: Introduction to mathematical psychology
Hot topic 1 Hot topic 1

● Relating cognition to the brain (and vice- ● Relating cognition to the brain (and vice-
versa) versa)
● Several big initiatives ● Several big initiatives
w w Human Connectome Project
» https://www.humanbrainproject.eu » http://www.humanconnectomeproject.org
» €1.2 billion over 10 years » Building a "network map" that will shed light
» Develop technologies to bring together on the anatomical and functional connectivity
disparate neurophysiological, anatomical, within the healthy human brain
molecular, and behavioral data ● Connections to cognitive psychology are
» Database (big data)
» Modeling (supercomputers, specialized
(hopefully) in the future
hardware) Purdue University Purdue University

● Big data ● Big data
● Technology allows gathering of way more ● Technology allows gathering of way more
information than we know what to do with information than we know what to do with

Greg Francis 1/6/20

● Big data ● Big data
● Many experiments can be run on-line with ● Data can be gathered in novel ways
thousands of subjects ● Airport Scanner
● 1 billion trials of
visual search!
Hot topic 3
Graduate school
● Data analysis
● Big data requires a different kind of statistics than has ● Grades
been used for simple experiments
● Financing
● Moreover, there seem to be problems with present
statistics even for simple experiments ● After graduation
● Areas of science that depend on statistics (e.g.,
psychology, biology, medicine) are facing a crisis as
findings that satisfied old criterion are found to be false
w And unbelievable findings meet the statistical criteria
w “Pre-cognition”
Next time
● Review for final exam
Purdue University

Greg Francis 1/6/20
Problem solving Problem solving
PSY 200 ● A hallmark of intelligence

w often used as a definition of intelligence
Greg Francis
● Seem to get something from nothing
Lecture 34 ● We will not explain exactly how it happens
w but we can look at some characteristics of
problems and problem solving
What does that aha feeling mean?
» what makes for an easy (or hard) problem?
» what makes for a good (or bad) problem
solver?
Logic Logic
● Some problems are completely handled with logic ● People do better if the task is related to some overall theme
● Unfortunately, most people are not logical in many situations

● The CogLab Wason selection task demonstrates challenges
with logic
w People do quite poorly at tasks of this type with abstract elements Each card has a person’s age on one side and their
Which cards do you need to turn over to verify the rule, drink on the other. Which cards do you need to turn
over to verify the rule, “If a person is under 21 then
“If a card has an even number on one side, it is red on
they cannot be drinking alcohol.”
the other.”
Logic Topics
● CogLab data ● Similar to characteristics of decision making a lot
● Count number of correctly chosen cards (2 is max) of problem solving techniques are heuristics
● We will look at a number of factors that influence
our ability to solve problems
w expertise
Class data
(166 w analogy
participants)
w set effects
» priming
» incubation
» functional fixedness
w insight

Greg Francis 1/6/20
Experts An example of experts

● Chi, Feltovich & Glaser (1981)
● Some people learn how to solve particular ● Take second year physics students (novices) and ask them to
classify a bunch of physics problems
types of problems
w they tend to group them by surface similarities
● What makes an expert different from a

novice?
● Experts know how to describe problems
w other than that, there seems to be no Novice 1: These deal with blocks on an incline
fundamental difference (even for geniuses!)

plane
Novice 2: Angular velocity; momentum, circular things
Novice 5: Inclined plane problems, coefficient of
Novice 3: Rotational kinematics, angular speeds, angular friction
velocities
Novice 6: Blocks on inclined planes with
Novice 6: Problems that have something rotating, angular angles
speed
An example of experts Chess experts

● Have experts (PhDs) classify the same
● Chase & Simon (1973) game
problems ● Show subjects a chess board and
then clear it
w grouped according to how to solve
w have subjects recall positions of
the chess pieces on the board
● Master players are better than
beginners when the pieces are
30
positioned as in the middle of a
Correct pieces
25
real game
Expert 2: Conservation of energy 20
Expert 3: Work-energy theorem. They are all
● Experts have schemas that allow Master
15
straight-forward problems. Expert 2: These can be solved by Newton s second law
them to organize the piece Beginner
10
Expert 4: These can be done from energy
considerations. Either you should know the principle
Expert 3: F=ma; Newton s second law.
positions
of conservation of energy, or work is lost Expert 4: Largely use F=ma; Newton s second law. 5
somewhere. w They only need to remember the
0
schema
1 2 3 4 5 6
Trial
Chess experts Expert schemas

● Chase & Simon (1973) random ● In general, experts have lots of problem
● Show subjects a chess solving schemas specific to their domain of
board and then clear it
w have subjects recall expertise
positions of the chess
pieces on the board w given such and such; do such and such
Master players are worse
w allows them to organize information in a way that
●
25
than beginners when the
pieces are positioned allows for easy recall and easy use
Correct pieces
20
randomly 15
● Experts try to use the

Master
Beginner
● Expertise in one domain does not transfer to
10
schemas, but they end up another
5
misremembering the actual
piece positions 0
1 2 3 4 5 6
w except for especially useful skills
Trial Purdue University
Purdue University

Greg Francis 1/6/20
Analogy: Attack-Dispersion Story Solution

● The many roads to a dictator s fortress are ● The general should split up his army into
mined so that small groups of men may many smaller units
pass, but a large group will be destroyed. A w each unit takes a separate road to converge on
general knows that his army can defeat the the fortress simultaneously and defeat the
dictator
fortress if he can get his entire army to
attack at once, but he cannot take his army
down a single road all at once without
fortress
losing too many men.
● What should he do?
Analogy: Parade-dispersion Analogy

● A dictator wants to show off his army so he ● The solution is the same
tells a general to plan a parade of the army. He w split up the army and have them converge to
demands that the general insure that the army the fortress from different parts of the kingdom
is seen and heard across the entire kingdom ● Glick & Holyoak (1980)
simultaneously. The dictator also demands that w subjects read stories like these and were
the parade be the most impressive ever at the asked to solve the problems
fortress. Splitting up the army would allow it to w even when shown one solution and told that it
be seen everywhere, but would make the could be applied by analogy to another
display at the fortress unimpressive. w subjects used analogies only 20% of the time
● What should the general do?

Analogy Set effects

Analogies are actually very difficult to apply
Set refers to mind set (or something like
●
●
w need to identify what is common between two that)

problems
w negative set: bias toward solving a problem makes
● Analogies are often applied after two it more difficult
problems are solved and well understood w positive set: bias toward solving a problem makes
w it is then easier to see what is common it easier
● One of the problems handed out can be ● You can be biased by lots of things
solved by analogy to these two problems w problem statement
w previous methods of reaching solution
Purdue University w general knowledge Purdue University

Greg Francis 1/6/20
Past experience Self-imposed limits

● Nine dots problem
● Buddhist monk problem
w Draw four straight lines, passing through all nine
w One morning, exactly at sunrise, a Buddhist monk began to
climb a tall mountain. The narrow path, no more than a foot of these dots, without lifting your pencil from the
or two wide, spiraled around the mountain to a glittering page.
temple at the summit. The monk ascended the path with an
average speed of 3 mph. He reached the temple shortly
before sunset. After several days of fasting and meditation
he began his journey back along the same path, starting at
sunrise and walking an average speed of 5 mph. Is there a
spot along the path that the monk occupied on both trips at
precisely the same time of day?
Set effects: Representation Set effects: Priming

● Mutilated chessboard ● Safren (1962)
w 62 squares remain ● Unscramble the following anagrams
w can 31 dominoes cover them?
● 12.2 seconds per word
● Easy solution! KOBO CTURK STTE
CHUOC ANCYD DRINEF

7.4 seconds per word
KMLI GRAUS RECMA
FOEFCE TEESW IKRDN

Set effects: Incubation Set effects: Incubation

● Cheap necklace problem ● Silveira (1971)
Opening a link costs $2
●
● Control: Work on problem for half an hour
● Closing a link costs $3
w 55% solve problem
● Go from given to goal state for no more than $15
● Exp A: Work on problem for half an hour in
15 minute sections, half-hour distracter task
w 64% solve problem
● Exp B: Work on problem for half an hour in
15 minute sections, 4 hour distracter task
w 85% solve problem Purdue University
Purdue University

Greg Francis 1/6/20
Set effects: Incubation Set effects: Functional fixedness

● Subjects often get stuck using an approach ● How can you tie two strings together?
that goes nowhere (set)
● Dunker (1945)
● After a break they are more likely to try a
different approach w MacGyver
● Brainstorming tries to avoid set effects by
allowing free dreaming of solutions
w most of the ideas are worthless, but the approach
is still beneficial
Insight Insight
● Intuitively, we sometimes feel as if we have a ● Warmth stays mostly steady, right up to
strong insight into a problem and its solution proposing a solution
becomes obvious w the aha feeling
w the aha feeling
w is it real? Unfortunately,
w what does it correspond to? the feeling does
not necessarily
● Using problems like the Bronze coin and indicate a
correct solution
the Tree planting problems (Metcalf, 1986)
w subjects judge their progress with a warmth
rating, every 10 seconds, over 5 minutes

● Effects on problem solving
● Wrapping up the course
● Expertise
● Other courses to take/avoid
● Analogy
● Set effects ● Paths to pursue
● Insight ● Graduate school
● Solve remaining problems
w Tumor problem
● Advice for further exploration

AllLectures PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AllLectures PDF

Uploaded by

Copyright:

Available Formats

Introduction to

MWF 9:30 - 10:20 am

Four great mysteries

Four great mysteries.

Purdue University Purdue University

Four great mysteries Four great mysteries

Purdue University Purdue University

Four great mysteries

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology

Lecture notes Lecture vodcast

w Adobe Acrobat (pdf) Four great mysteries.

Purdue University Purdue University

w Sometimes takes a few days

w Sound may not be very good

page) Purdue University Purdue University

Four great mysteries

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology

Purdue University Purdue University

Attendance Course web page

PSY 200: Intro. to Cognitive Psychology

Course outline Exams

Purdue University Purdue University

Writing assignments ● Straight scale Grading

PSY 200: Intro. to Cognitive Psychology

Final class grade

E1 E2 E3 E4 Final CogLab Writing Attendance Total 8 0.00

Grading Instructor office hours

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology

● CogLab on Brain asymmetry due!

● What s the deal with left and right brains?

PSY 200: Intro. to Cognitive Psychology

Brain parts The brain

PSY 200 ● The source of cognition

Greg Francis ● Weighs about 3 pounds

of each brain part?

Purdue University Purdue University

The brain The brain

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology 1

Purdue University Purdue University

Contralateral processing Contralateral processing

Purdue University Purdue University

Contralateral processing Cut corpus callosum

PSY 200: Intro. to Cognitive Psychology 2

Brain sides Brain sides

Purdue University Purdue University

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology 3

CogLab Data CogLab Data

Purdue University Purdue University

Anatomy Occipital lobe

● Each has distinct properties

Purdue University Purdue University

Parietal lobe Parietal lobe

● Sensations of ● Primary sensory

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology 4

Temporal lobe Frontal lobe

w speech (left) cortex

Purdue University Purdue University