AllLectures Psych PDF

Introduction to
Cognitive
Psychology
Greg Francis
Fall 2016
MWF 2:30 - 3:20 pm

Matthews Hall, Room 210
http://www.psych.purdue.edu/~gfrancis/Classes/PSY200/indexF16.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?
! This is the domain of physics
Greg Francis ! 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
! 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
question

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
PSY 200: Intro. to Cognitive Psychology

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

Links will be posted on the class web page as the
format Four great mysteries Four great mysteries
vodcasts become available
Humans face four great mysteries about the Humans face four great
! Often takes a few days

mysteries about the
universe
universe
!Reduced form (6 to a
2) How did life form? 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
! 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
! fill it out with your Purdue information http://www.psych.purdue.edu/~gfrancis/Classes/PSY200/indexF16.html
! Sign it !updates to the syllabus

! Answer (a) for the first question !Links to labs
Hand in the scantron at the end of class !Links to writing assignments
! It will function as a check on your attendance
!Study guides for the exams
You can miss up to 6 lectures without any penalty
!(old) practice exams
! Beyond that you lose a proportion of points
!Links to optional readings
! 5% of your class grade
!Grades will be posted after the first exam
Scantrons are only available for the first 15 minutes of
class This course does not use 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
!The exam is during the final exam week
Cumulative Final (15%)
!There are few excuses for changing the date
CogLab CogLab
Labs are listed on the syllabus
Homework
They must be completed by 10:00 am at the
You participate in classic experiments date indicated in the syllabus
Total lab grade contributes to 15% of your class ! else you get no credit
grade. ! 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
! (a $50 value!)
See handout for instructions on getting started

Registration code is on a label on the instructions
First lab is due at 10:00 am on Wednesday!
Writing assignments Straight scale Grading

You need practice writing! ! 98% -100% A+
! 93% - 97% A
Four assignments, 3-5 pages of single spaced
! 90% - 92% A-
text.
! 88% - 89% B+
Assignments are due (submitted by email to the ! 83% - 87% B
TA) as listed in the syllabus ! 80% - 82% B-
! 78% - 79% C+
First assignment is September 2
! 73% - 77% C
! By the start of class (not one second later!)
! 70% - 72% C-
15% of your class grade ! 68% - 69% D+ No extra credit
! 63% - 67% D
! 60% - 62% D-
! 0% - 59% F
No rounding up: 82.99 is a B-

3
Prof. Greg Francis
Grading Grading
Last semesters grades (Spring 2016) Last semesters grades (Spring 2016)
Frequency
E1 E2 E3 E4 Final CogLab% A 10
Average 69 69 78 79 73 87 B 25
120.00
Max 92 97 98 100 95 100 C 24
Final class grade

Min 24 30 45 46 37 25 100.00
D 9
80.00
F 9
60.00
40.00
20.00
0.00
0.0 20.0 40.0 60.0 80.0 100.0 120.0
CogLab score
Instructor office hours Teaching assistants

Monday, Wednesday, Friday, 10:30 am
11:30 am Grade writing assignments
!Or by appointment
Keep track of grades and attendance
!Psych 3186
!Email: gfrancis@purdue.edu Have office hours
May provide out-of-class study sessions
for exams
Teaching assistant Teaching assistant

Eddie Christopher Jeongho (Johnny) Han
Office: PSYCH 3198 Office: Matthews Hall, Room 326
Office hours: Office hours:
!Tuesday and Thursday, 10:30 am - noon !Monday, 1:00 pm 2:30 pm
Email: chris119@purdue.edu Email: han202@purdue.edu

4
Prof. Greg Francis
Attitude/Advice
During lectures: turn off cell phones, don t read newspapers, don t play Next time
games
Questions are always welcome. I can adjust my lecturing pace accordingly
Print out the lectures and bring them to class. Take notes during class. Not Cognitive neuroscience
everything is on the slides.
The brain
Everything we talk about in class is important
Work on the study guide every week, so the ideas/answers are fresh in your The modularity hypothesis
mind.
This class is an introductory class, but that does not mean it is easy
! Its like Introduction to Physics or Introduction to Chemistry CogLab on Brain asymmetry due!
! Almost every other subtopic in psychology depends on the ideas in cognitive
psychology
! Everything is at least 10,000 times more complicated than what we discuss
What s the deal with left and right brains?
If you don t find a topic interesting, just wait a week

5
Prof. Greg Francis 7/31/15
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,
!cortex hormones)
!(brain
Hind-brain stem)
!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 ! Left side: language, analytical, classification, Western rationalization
! subject cannot name it ! Right side: art, music, recognition of faces and shapes, Eastern
mysticism
! subject can pick up nut
with left hand Vast oversimplification
! in a normal brain, both
If nut flashes on right
sides are involved in
side many tasks
! subject can name it Results do support the idea
! 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 + wood
new (not previously
A word was
seen in this
presented to either
experiment)
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
! Which is known to be specialized for language !Reading goes from left to right, from fixation to
right visual field
!Perceptual advantage to right visual field?
!Attentional advantage to right visual field?
It is difficult to come up with an experiment

that isolates hemispheric specialization

CogLab Data
Global (13,231 right-handed participants; 1383 left-
Anatomy
handed participants) The cortex contains large fissures that
No effect: maybe participants are unable to fixate? separate five major areas
!Limbic (already discussed)
!Occipital
!Parietal
!Temporal
!Frontal
Each has distinct properties
Occipital lobe Parietal lobe
Receives Sensations of
information from pain,
the eye temperature,
touch, pressure
!Most investigated
area of the brain
Primary
sensory
area
Parietal lobe Temporal lobe

Hearing
Primary sensory
!speech (left)
area
!music (right)
!sensitivity
involves Memory and
disproportionate attention
areas of the brain,
!visual recognition
relative to size of
body part

Frontal lobe Broadman areas

Divide lobes into areas
e.g. Broadman
Largest part of

! Area 1, Area 12, Area 200
cortex
! Some special names: V1, V2, V3,
!planning
!prediction
!motor area
!speech area
Area function Brain layers

Can partly identify There is order and
function by looking at function even within
nerves coming in and an area
out of area The cortex is a
Pathways through Color sheet of neurons
areas seem to be
In its thickness are
involved in different
6 layers of neurons
kinds of cognitive tasks
! numbered 1-6
! sometimes include
subdivisions (4a, 4b,
4c, 4c, )
Conclusions Next time

Lots of research in this area Brain scans
EEG recordings
New brain regions are being mapped out

MRI scans
daily with ever increasing resolution
PET scans
Cognitive neuroscience relies strongly on
Functional MRI
the modularity hypothesis
Putting everything together is very difficult How to study the brain without killing someone.

Brain scans Scanning

Technology provides insight into brain processes
PSY 200 ! EEG recordings
! MRI
Greg Francis ! 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
! Space: which EEG
place is active?
! Time: when The brain
does something
happen? produces
Finer resolution electrical activity
is usually better
! 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
! Good temporal resolution Semantic
! Kutas & Hillyard (1980) anomaly

Magnetic Resonance Imaging

Spatial resolution Magnetic field
forces protons in
Poor spatial your body to line
up
resolution
! pulses of radio
You never really into field
bounces
know which part of protons around
the brain is making ! as they return
to normal
the current
position, they
! 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,..) ! Very good spatial resolution
!Very good spatial resolution ! 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!
! 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
!it only shows structure
Ataxia:
!no way to know what a brain area does
!loss of motor
control
Functional MRI Functional MRI

Just like MRI, but with a new analysis
Color maps show
! MRI differentiates between different types of tissue (cell
strongest
types)
responses
! Functional MRI differentiates between active and
inactive neurons: concentration of oxygen E.g., fMRI scan of a
! The measurement is called the blood oxygen level woman after a stroke
dependent (BOLD)
!Blue/green:
It roughly tracks the flow of blood in the brain
normal blood flow
More active neurons recruit more blood
!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
! millimeters
different regions are involved
Pretty good in different activities
temporal resolution ! Breathing
! Seconds ! Digestion
! Thinking about exams
! (Silva, 2002)
! ..

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
the different cognitive tasks Colors mark places
in the brain that are
It requires a sophisticated statistical analysis to avoid false positives!
statistically different
between conditions
Viewing blank Viewing face Difference map
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 ! 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
!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
! the scans just tell us approximately where in the brain
cognitively based
something occurs ! e.g., MRI scans
of stutters
! sometimes it can tell approximately when
! in fact, all
Even trying to find the place may be problematic behavioral traits
! Lots of cognitive abilities involve many different areas are physiologically
of the brain based
Most of the time theories of cognition are derived
from experimental psychology
! Brain studies explore how to implement the theories

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
! Computerized Axial Tomography (CAT) What is really being measured?
! Diffusion tensor imaging (DTI)
! Single Photon Emission Computed Tomography
(SPECT) How to read someone s mind.
! Near Infrared Spectroscopic Imaging (NIRSI)
! Magnetoencephalography (MEG)
! 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
!Vannest et al. (2009) listening
difference maps
Twenty children (ages 11-13) complete three relative to listening to
the tones
tasks
Common activity
!Passive listening: hear a female speaker tell a 30-
(breathing, digestion,
second story
hearing machine
!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!
!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?) ! PL: 75.1% correct, SD=12.7
! 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 ! Kaczmarek, Bach-y-Rita & Tyler (1998)
information for blind people ! 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
Reading minds Reading minds

Long-term, the goal of fMRI research is to be able to analyze a brain scan Based on the choice at the end, we can deduce whether the subject chose
and identify what a person is thinking addition or subtraction for that trial
There are several attempts to do this (Haynes et al., 2006) Make an fMRI scan during the selection process
! Adding or subtracting numbers ! Whether to add or subtract numbers
Scan
now
Or scan
now

Mind reading Thought reconstruction

It is possible to build a recognition system that distinguishes (with
71% accuracy) the brain patterns for addition and subtraction Another research group analyzed fMRI responses to
! Depends on the place in the brain reproduce a shown image
! Different places for intention and execution
You can read the mind of these subjects!
Thought reconstruction Thought reconstruction

Performance depends on where the signals come from
These kinds of studies are mostly a demonstration of
Fewer errors for lower brain areas technology
! Where do you stop?, the retina?, the lens of the eye? ! we already know the brain represents visual information!
Before the study was run, we knew that there were

differences in the brain when we see different images
! The percept is the brains behavior, so there must be
differences!
These kinds of studies tell us that the

neurophysiological differences between cognitive
events can be measured by these brain scanning
technologies
! Failure would only indicate limits of the technology
Mind reading with EEG Mind reading limits

fMRI: If subjects decide to multiply numbers, a system trained to
You can purchase an EEG device to read your mind distinguish between subtraction and addition is clueless
and control a computer Thought reconstruction: As the number of possible images to be
! www.emotiv.com
shown increases, it becomes harder to reconstruct the shown
image
Emotiv: very limited set of mind reading possibilities
In general, brain scans provide a very limited form of mind reading
! People do better than this every day by watching people behave
(posture, eyes, skin tone)
The military and police are quite interested in these approaches
Companies are interested in identifying what you think
! Proctor & Gamble

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
! Statistical analysis is complicated field to sort this out)
! 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
! Some of those voxels will give different responses just by chance specified emotional valence, either socially inclusive or socially
exclusive. The subject was asked to determine which emotion the
Difficult to compare across subjects
individual in the photo must have been experiencing.
! 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)
! 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)
! 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
! The brain has lots of random noise
These problems can be reduced but never entirely

eliminated
! They are common to many areas of psychology, not just brain
scans
Neurons A neuron
Dendrite Axon
The brain cells that are responsible for cognition
! input ! output
are neurons Soma
Myelin sheath
! integrate
! 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
!100,000,000,000; one hundred billion
Lots of technical (and ethical) issues
!estimates of 100,000 per cubic millimeter
The goal is to be able to look at a map (or movie)
!(about the resolution of functional MRI)
and be able to read someone s mind
Millions are active at any given time
That is many years off
! 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 / output Input at dendrites

Changes the cell membrane
-70 millivolts
Electrical signal potential

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

K+

cell a threshold and the cell fires
Output identifies when ! action potential

K+
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
! 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
! 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
! physical problems (paralysis)
! cognitive problems (memory,
reasoning, judgement)
! cause unknown (300,000
people)
Networks Epilepsy
Cognitive behavior is related to Disease of central nervous system
groups of neurons working ! causes mostly unknown
together
Seizures
Include excitation and inhibition
! bursts of electrical activity travelling through networks in
! more later
the brain
! brain activity is out of control
! 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 ! at multiple levels
! drugs, diet, avoiding stress, Neurons
keeping regular schedule
! balance between forces inside and outside of cell membrane
In extreme cases surgery allows for action potentials
prevents seizures from
Networks
spreading throughout the
! 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 !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
! accepts some
!physical, chemical, electrical
neuro- !locally changes the cell membrane
transmitters
depolarization (excitation)
! rejects others
hyperpolarization (inhibition)
Neurotransmitters Tourette s syndrome

Inherited (~200,000 in US)
Different neurotransmitters are associated Behavior
with different properties ! Tics
Simple: eye blinking, facial grimacing, sniffing
!actually neurotransmitter and receptor pairs Complex: coordinated patterns, sniffing objects,
jumping, twisting
neural
Too much dopamine
cognitive Treated with Haldol (among others)
! blocks dopamine
behavioral
http://www.cbsnews.com/video/watch/?id=1033515n
Parkinson s Drugs
Lack of dopamine Interact with neurotransmitters in lots of ways, for
example
!Many different causes
! Replace: accepted by receptor and with similar effect
!In extreme cases, patients are frozen
! Production: increase or decrease
Give patients large doses of L-DOPA ! Reuptake: knock out enzymes that remove
neurotransmitter from receptor, neurotransmitter has a
!a precursor of dopamine bigger effect
!sometimes solves the problem ! Blocking: enter receptor but does not trigger reaction,
!lots of side effects partly closes receptor protein so neurotransmitter
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)
! 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 ! But a single action potential has little effect
Lecture 06 If the input stays present, the cell produces

another, and another,
! 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
! What stimulus (or stimuli) makes a given cell fire at a strong
! Firing rate of 27 Hz (27 spikes per second)
rate?
something red?
a pen?
your grandmother?
! 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 a
Touch involves sensitivity to pressure on skin
cell s firing rate. The loops indicate the
regions where a single
A stimulus could excite the cell
neuron responds to
!above normal firing rate pressure
Or inhibit the cell

!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
! it responds to light of the right wavelength (color) and the right

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

Blind spot Network

Here s the average data for the whole class
164 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
!is coded something like this
Strong
responses
at edges!

Complex cells Complex cells
Many simple cells feed Many simple cells feed into a

complex cell, which is
into a complex cell,
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 ! 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
! Desmione et al. (1984) ! 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
!any stimulus that affects a cell s firing rate
grandmother
!excitatory
! but it is unlikely
! not enough cells !inhibitory
! 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
! 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 ! (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 changes)
affect cell 1 again, which The remaining active cells are those that
3 1 support each others activities through
What happens to cell firing
rates? excitation: resonance
! high firing rate==> active Inactive cells are inhibited by the active cells
! low firing rate ==> inactive 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
weights on the links between w21
reciprocal 2
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
! excitation -- inhibition
!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
!emergent properties
!stable activities
!multiple constraints
!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
A problem with virtual reality.
Purdue University

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

network of neurons can
2
Greg Francis have very complicated
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
!wij = 0 The network self-organizes in response to stimulation
Hebb s rule It can remember things it has previously experienced
!cells that are simultaneously active develop It can interpret new information on the basis of things it has
positive weights (excitation) previously learned
!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
! 1000 computers (16,000 cpus) over 3 days you do when you study for school
The network self-organized to identify common patterns !but it is important just the same
! Cats, faces, tool-like objects oriented at 30 degrees
Consider the length of your arm
!to catch and throw objects your brain must know
exactly your arm s length
!but the length of your arm changes as you age!
And depends on unknown environmental
factors
Hand-eye coordination Coordination and learning
We do not know the exact nature of the

vision arm network involved in this coordination
info info ! 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
!architects, designers
! surgeons, pilots
!entertainment

Cameras Problem
Enhance visual perception
The network coordinating
!MRI overlaid on actual image
of brain for surgeon eye-hand systems, adjusts
highlight tumor itself
!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
!changing connections
!relatively simple rules
Then
Much of our perceptual and motor
Neural networks for visual perception
behavior is based upon this type of
!brightness
continuous learning
!color
It s not clear if more cognitive learning is
!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

! brightness
Lecture 09 ! color
! form
! motion
! texture
Why you see color
! depth...
afterimages.
Neurophysiology On-center, off-surround

How we see things is largely
Characteristics of cell receptive
determined by the properties of fields force additional properties
of the visual system
receptive fields
! center-surround cells tend to not
!on-center, off-surround respond well to homogeneous
light that covers both excitatory
!simple cells
and inhibitory parts
!complex cells ! => percepts of the middle of an
object is derived from the edges
And by network interactions among
cells
On-center, off-surround Brightness contrast
Characteristics of cell receptive Edge responses are influenced by the surrounding

fields force additional properties light
of the visual system
! both center squares have the same light intensity
! center-surround cells tend to not
respond well to homogeneous
light that covers both excitatory
and inhibitory parts
! => percepts of the middle of an
object is derived from the edges

Brightness contrast Brightness contrast

Two receptive fields inside the middle square Receptive fields on the corner
receive the same excitatory and inhibitory signals
!Receive the same excitation at the center
! Little response
!differ in the amount of inhibition in the surround
less
inhibition
more
inhibition
Brightness contrast
Brightness contrast
Thus, the visual system computes brightness as
something like local contrast
! It s a property of the center-surround cells
! Our percept of brightness is determined by the responses
of cells at contrast edges
! As a result, things that have equal physical intensities can
look dramatically different (next slide)
Brightness contrast Brightness contrast

Brightness contrast
Hermann grid
Hermann grid Filling-in

We see color and brightness inside objects
! so edge information must fill-in to the interior
Seems related to on-center,
off-surround cells
It sometimes gets things messed up
Cells at intersections receive Water color effect
more inhibition than cells at
single roads
How do we explain the other

version?
Filling-in
Filling-in
Brightness
information
Carefully fixate
spreads across
the pink center
surfaces
!Craik-O Brien- If you keep your
Cornsweet eyes very still, it
effect will disappear
The yellow fills-in!

Color system Color system

competition between competition between
opposite colors opposite colors
! red-green
!red-green
! blue-yellow
! black-white !blue-yellow
habituating gate !black-white
offset of one color leads to Initial balance
rebound in other
Gated dipole circuit !Neither color wins
competition

!red-green !red-green
!blue-yellow !blue-yellow
!black-white !black-white
Extra input to green Extra input to green

!Green wins competition !Fading of green signal

!red-green !red-green
!blue-yellow !blue-yellow
!black-white !black-white
Offset of green Recovery of green pathway

!Rebound of red signal !Disappearance of rebound
!Return to initial state

Color afterimage Orientation competition

competition between
orthogonally tuned cells
habituating gate
offset of horizontal leads to
rebound in vertical
Same kind of gated dipole
circuit
! Principles of neural
computation!
Baseline response
! Due to tonic input
Orientation competition Orientation competition

With additional input to With additional input to
horizontal pathway, horizontal pathway,
horizontal channel wins horizontal channel wins
competition competition
But as horizontal gate
habituates, horizontal
signal weakens
It still wins the
competition, though
Orientation competition Orientation competition

At offset of horizontal As the horizontal gate
input, the gated horizontal recovers, the system
signal is weaker than the returns to baseline and
vertical signal the vertical after response
A vertical rebound disappears
appears

Oriented afterimages
Conclusions
Oriented
Visual perception
reset signals
are also !brightness
implicated in !color
an unusual !form
type of
Largely determined by the receptive fields
afterimage
and network structure of visual circuits
Neurophysiology strongly determines
what we see!
Next time
Visual dynamics
! Flicker
! Persistence
! Motion perception
CogLab on Apparent motion due!
Why computer monitors work.
Purdue University

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)
!about 50 Hertz (50 on-off cycles in a second)
Why (CRT) computer monitors !20 millisecond durations
work.
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
! ten thousandths of a second
! 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
! better monitors go faster
! the percept persists in your head!
! 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)
!Receptors in the eye? !subjects adjust duration of a blank
!Receptive fields? stimulus so onset of probe matched
perceived offset of the target
!Network interactions?
!Cognitive (memory)?
Studies support network interactions +

Strange property Networks

As the target s
Feedback is
duration or
important
luminance
increases Produces a persisting
!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
!else it would keep persisting forever
!e.g., due to
Two mechanisms competition from
!(1) new inputs inhibit old responses orthogonally tuned
cells
!(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
! the afterimage blurry?
produced at target
offset increases in !There should be smears of objects as they
strength move or as we move
! so there is stronger
inhibition to break
There must be something else preventing
the feedback such blurring
! so the persistence !masking
of the original
percept decreases

No mask No mask
Write down all the letters you see Write down all the letters you see
F C
U H
N D
No mask Masking demonstration
Masking demonstration Masking demonstration
G M X X X X
Q P X X X X
S Y X X X X

Significance Motion
In simple animals (like flies and frogs), we know how
The mask appeared after the target turned off motion is detected
The target was presented all by itself for a brief ! Demo on web page
period of time Reichardt detector
However, our visual system is unable to
develop a complete percept of a scene in a
such a period of time
! Thus, the XXX mask interferes with processing of
the letters by shortening their persisting responses
! And prevents perceived blurring of changing scenes
Masking is often used to investigate other

aspects of cognition (we ll see examples later)
http://neurovision.berkeley.edu/Demonstrations/matthew/reichardt.html
Detecting Motion Apparent Motion

Humans have something like Reichardt motion detectors When objects move, there is a continuous path of
! at lots of different positions in the visual field motion
! sensitive to lots of different motion directions
Reichardt motion detectors do not require
! sensitive to lots of different motion speeds
continuous motion
Think of them as receptive fields that vary in both space and time
! 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
! color
! stimulus duration
! shape
! interstimulus interval (50-200 msec)
! attention
! cognitive priming ! distance

Timing Korte s laws

For a Reichardt detector to Apparent motion was highly studied at the turn of the 20th century
indicate motion, the signal from ! Korte (1915) noted that to get good motion, you needed to increase
the second area must follow the the ISI between the stimuli as the distance between them increased
signal from the first by just the ISI too short
right length of time
CogLab data
! (196 participants- from class, 16,355 from global)
Vary the Interstimulus Interval
(ISI) between the stimuli
ISI about right
! The time between offset of the
first stimulus and onset of the
second stimulus
ISI too long
Motion representation Motion aftereffect

Just like for color and
One conclusion of studies of apparent motion orientation, we might
is that motion is a fundamental percept expect an aftereffect of
! It has an explicit representation in the visual system motion
You could imagine otherwise, we can be aware of competition between
something moving without opposite directions of
actually seeing the movement motion
Apparent motion is the source ! Left-right
of motion for all movies and ! Up-down
animation habituating gate
offset of one direction leads to
rebound in other
After effect Conclusions

Motion can be adopted by non-moving stimuli Dynamic vision
! http://www.michaelbach.de/ot/mot_adapt/index.html
! flicker
! persistence
! network dynamics
! Masking
! Reichardt detectors
! Apparent motion
! Motion aftereffect
Also used to investigate other areas of cognition

and types of mental problems

Next time
Attention
What is attention?
What does it do?
How could you not see it?
Purdue University

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
!not clear if there is an attentive system
!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
your part
Modern theories see cognition as information processing
! much like a computer ! attending a lecture
Different systems have different capabilities, capacities, and speeds ! 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
! ignoring the uhs and ums from a speaker
! ignoring the feel of clothes on your body
Part of attention seems effortless

! a loud noise
Magic trick Magic trick

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

Magic trick Attention as processing

Hey, the card you selected is missing Mental effort either is or requires attention
Consider the following video (Simons & Chabris, 1999)
How did I know which card you would select?

! Try it on the web at http://members.tripod.com/ ! you will see two

~andybauch/magic.html groups of people
wearing either black
or white shirts
! each group is
tossing around a
ball
! count how many
passes are made
by the white group
Attention as processing Drawing attention

Did you notice an unusual event in the video? Attention is focused by meaningful stimuli
! Let s look at it again
Also focused by certain stimulus characteristics,
especially changes
Attention is a powerful force ! Flashes of light
! when you are busy doing something else, you can fail ! Movement
to observe even very significant things
! Color
! that s why it is not a good idea to talk on a cell phone ! Think of advertising signs
while driving a car
Example
Drawing attention Attention

Another way to interpret these results is that
Suppose these cues were masked by other attention is necessary to detect stimulus
changing stimuli changes
You might not notice the change at all Explains how people can look but not
Demonstration see
!walking into doors
!driving into trains
!detecting changes on a radar screen
!why magicians use flashes of light!

Attentional paradox Attentional paradox

If attention strengthens perceptual representations, we should Sometimes attention
lose perceptual veridicality can change perceptual
! We might expect what is schematized below properties
! But we normally do not experience this But then we have an
Attention generally seems to strengthen information about a incorrect perception of
stimulus that is not perceptual the properties of the
visual scene
So it is difficult to
Stimulus Attend red Attend green understand how
attention is helping
here
Tse, 2005
More demos Conclusions

If time permits, here s some more demos
! http://viscog.beckman.uiuc.edu/djs_lab/demos.html Attention can have very powerful effects
Field !help processing of focused on things
Living room !can cause unawareness of unattended things
Phone call Not precisely defined
Lunch conversation (9 changes)
!characteristic of processing?
Paris scene
!An extra system?
Next time
Methods of studying attention
What things influence attention
!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 !when it is focused on one thing, you ignore
other things
Lecture 12
Today we want to consider some more
specific properties of attention
!and look at experimental methods that are
Should you pay $59.95 for
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
! e.g., detect J and/or K in a stream of letters
properties of the underlying systems that
are involved in cognition
M
!whether attention is thought of as a system P
K
!or as a by-product of other systems R
W
Look at S
!temporal
!featural
Attentional blink
Attentional blink
Measure frequency of detection
Turns out that detection of first letter tends to make ! class data (166 observers)
detection of the second letter very difficult
! if it immediately follows the first
! Attentional blink
Implies that
detecting the
M
first letter
P
causes
K
you to miss the
R
J second letter!
S

Attentional blink Speed reading

Suggests that processing the first target
Ever seen the ads for speed reading?
letter, J
! uses up resources that would otherwise be used One is for a program called Mega Speed Reading
to process second target letter, K ! claims to teach you to read 25,000 words per minute
! attentional focus and refocus takes time and for ! ==> 2.4 milliseconds for each word
this task takes approximately 400 ms
! impossible, if only by attentional blink!
J S
The seller is skimming at best,
L
N Once processing is done (a and lying at worst
B few hundred milliseconds),
J there is no difficulty
R detecting second letter.
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
! Staplers, Road signs ! hypothesize that tasks that involve attention will be
slower than perceptually based tasks
! 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
! we vary the number of distracters
! 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
! fast ! 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
! pop out ! 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
! shape ! color
Conjunctive Search Visual search experiment

But if the target has shared features with different Four types of responses
distracters, search is difficult
! 1) Feature - present (can respond as soon as see target)
! No pop out
! 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
! 3) Conjunctive - present (can respond as soon as see
! Orange rectangle: color and shape
target)
! 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 (156 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 ! Because when the
target is present you
find it, on average,
Requires search after searching half the
items and then can
green arcs by comparison stop the search
across feature
maps. ! 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
seems to require a lot of
requiring a lot of attention to requiring little is
attention to perform called automatization
Later it requires less attention Many tasks are automatizable
! riding a bike
!color naming
! driving a car
! typing !word naming
! tying shoelaces Can measure effects by pitting an
http://www.fieggen.com/shoelace/knots.htm
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
!measure reaction time

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

! Simon effect for pointing
You can try them both on CogLab

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

! Stroop task
Color naming is unpracticed, so it occurs
Characteristics of attention
slowly and requires attention ! timing
! role of perceptual features
With two tasks, both trying to report on a color
Automaticity
!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
! Iconic memory
! 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
!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
G W X R
T P Q N
B Z Y H

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 ! choice of row is random
letters, but lost the percept while Suppose the subject reports 3 of 4 letters from
they reported any row
!they cannot report fast enough ! =>3/4ths of each row was available
! ==> entire field was available
How can we tell if percept is lost?
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
R W V D
P S C K
H L Z Q
Partial report Sensory memory
Write down letters from the indicated row Temporal characteristics

! delay

Sensory memory Sensory memory
Temporal characteristics Temporal characteristics
! delay ! delay
Y K W X
L B M R
J T C Q
Sensory memory Results

Vary delay to watch decay from memory
! compare to CogLab data
Temporal characteristics
! 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 (139 participants) result
! so, the number of letters that actually persist and is available is
found by multiplying by the number of rows
Number of letters available
Original published data CogLab data

10 10
8 8
6 6
4 4
2
2
0
0
0.02 0.1 0.3 1
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
10
Number of letters available
percepts fade away (2010) modified it to

Fading percepts are due
8 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
! Large capacity Delay of tone (seconds)
! 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

!persistence-based memory is very brief, and is
location of the changed color
item
easily destroyed by a mask
They do better than Iconic memory is
infants (who did not
understand the task ) !brief
Look for sharp drop in !easily disturbed
performance as set size
increases
Estimate items in memory
Adults = 5.75
Infants = 5.0
Partial report with masking Partial report with masking
Write down letters from the indicated row Write down letters from the indicated row
M H T R
Y V N W
L S C B

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
probably has little to do with normal memory
X X X X ! 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
!Longer duration 5
persistence or sensory memory
Number of letters
(seconds)
4
F
!Smaller 3
H
available
V capacity 2
A D Significant for 1
S E 0
W some memory
X 0 1 2 3 4
tasks
Delay (seconds)
Immediate serial recall Serial position curve

After given a list of items
!e.g., digits, letters, words, Often, subjects recall first and last items
!subject must report them back best 1
Probability of recall
0.8
!1) no delay (immediate)
0.6 Primacy Recency
!2) in the correct order (serial)
0.4
!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:
! In this task, recency depends on sensory memory
presentation
! It takes time to report all the items in the list, in order
1
! in the visual presentation, iconic memory of the last item
0.8 is gone before subject tries to report it (poor recall)

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

Not affected by
!practice Call information for a number
!meaning of cue word
Operators are very short
!common vs. rare word
!give the number
Words are physically different from tones !do not say goodbye or have a nice day
!suffix word acts like a mask to wipe out last word in
Avoiding the suffix effect!
list from echoic memory
!you would forget the last part of the phone
!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 Memory
Sensory memory Modal model

! iconic memory (visual) ! short term memory
! echoic memory (auditory) ! long term memory
Relation to immediate serial recall (recency) Experiments

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

Two store model Memory

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

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

Caesar s.
Simple view Ebbinghaus experiments
The container theory of memory does not explain, for

First memory experiment (1885)
example, Measure how long it takes to learn a list of nonsense
! why some memories are very long lasting (my childhood car syllables perfectly
trips to Utah) ! NOF, QAP, HOS, LEQ, FIK, MEC, KIJ, HOM, NEM, MOJ
! 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?
! 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 ! subjects do not need to relearn, perfect memory
!a different list each time Savings=0

! 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
! 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
! 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
!Brown (1958) might not expect any interference
Give subjects trigram !but they can have very strong interference
1
!ask them to count backwards by 3 s and Proportion correct 0.8
then recall trigram 0.6
779, 776, 773,... Suggests
0.4
some memories
last only a 0.2
WRM 782
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 ! even if you are distracted, you can recall the trigram if
!keeping a memory active requires effort only a short time has passed
! if many seconds have passed, while you are
!if you are distracted by another task, you distracted, you cannot recall the trigram
cannot apply the effort to keep the memory ! memory has decayed , or something like decay, while
!similar to our observations about attention you were doing the distracting task
and processing

Retention Retention
CogLab data CogLab data (global data set)
! 61 subjects ! 9690 subjects
! Not everyone shows interference on this task
Another experiment Interpretation

There exist two types of memory systems
9
Memory span Long Term Memory (LTM)
5 7
!how many items can ! high capacity (no limit)
you correctly recall 8 6 3
! long duration (forever)
immediately after 4 9 2 7 ! Ebbinghaus experiment
exposure? 5 9 0 1 3 Short Term Memory (STM)
The magic number 6 8 2 9 3 4 ! small capacity (~7 items)
7+/-2: 7 2 5 1 6 9 3 ! short duration (seconds)
9 1 0 3 5 7 3 2 ! Memory span, Brown-Peterson

!Miller (1956)
3 9 0 5 7 4 2 16
Pizza
There used to be a Little Caesar s in W. Modal Model of Memory
Lafayette with a game where you could win a Atkinson & Shiffrin (1968)
pizza
Multiple stages of memory
! must repeat a sequence of flashing lights
(changes every time) STM plays a dominant role in active memory
! The sequence gets longer until you make a Requires transfer between STM (STS) and LTM (LTS)
mistake
! need a sequence length >7 to win much
Counts number of correct button presses
56 (sequence of 11 buttons): win a soft drink
110 (sequence of 15 buttons): win crazy bread
210 (sequence of 20 buttons): win pizza
nearly impossible with STM properties
! http://www.freegames.ws/games/kidsgames/simon/simon.htm

Free Recall Serial Position Curve

Modal Model of Memory
When something is memorized Given almost any list of items
! Items are first held in STM (temporary store) Subjects remember the first and last few items
! Items may transfer to LTM (permanent store) best (free recall, not immediate serial recall)
! Takes time to transfer 1
0.8 RECENCY
0.6
PRIMACY
0.4
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
! words
In some situations the serial position curve can
! letters
be explained by different properties of STM and
! numbers
! pictures LTM
1
RECENCY:
Here s the
CogLab data 0.8 PRIMACY:
Use STM
! (66 subjects) 0.6 Use LTM
Demo 0.4
0.2
0 NEITHER LTM NOR STM
1 2 3 4 5 6 7
Position in list

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

! visuo -spatial sketchpad
strongest conclusions of cognitive
CogLab on Sternberg search due!
psychology

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?
! 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
! the answer (yes or no) is returned after all items have been
(similar to visual search experiments)
checked
! measure reaction time
!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
! the answer (yes or no) is returned after all items have been If parallel search
checked ! number of items does
not matter
Reaction time
YES Reaction time is ! 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
! the answer (yes or no) is returned after the target is found or ! 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 ! 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 YES
NO
size
300 NO
! YES RT s shorter
5329
200
than NO RT s 100 8 8 88
0 8
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
! 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 YES
the same for a set size
300 NO
5329 yes response as ! YES RT increases the 200
for a no
3 33 3 same as NO RT s 100
3 response
0
3 Lines are parallel 1 2 3 4 5 6
3 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
! They predict very different patterns of reaction time as a CogLab data
function of memory set size (160
Sternberg runs the experiment to see how the data participants)
comes out
! 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
! finer analysis of cognition than anyone expected
! 1) is serial, one item at a time
was possible
and checking each item takes approximately
the same length of time ! used a thought experiment about different types of
Approximately 40 milliseconds (CogLab data is searches to generate precise testable predictions
a bit slower, 68 milliseconds) about cognition
! 2) is exhaustive subsequent research found that there were other
search always goes through all items types of searches that complicate the conclusions
! 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
! supervises
! very efficient (can search very quickly) ! coordinates
! dumb (doesn t bother to stop itself) ! starts and stops relatively independent processes
! initiated by some other system (a controller)
e.g.
! Search short term memory
Controller ! Search long term memory

! walking down stairs
Search ! gate in psychological sciences building
memory ! 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)
! two types of tasks (visuo-spatial and phonological)
! visual imagery
! two types of responses (visuo-spatial and
! classify corners (top or bottom
phonological)
corner?)
Identifies two types of systems that are ! yes if top or bottom
relatively separate ! no if not top or bottom
Two response types

Separate systems
no, yes, no,
no, yes, no,
Part 2: verbal mental Either ...
task !verbally
!read sentence A bird in !spatially
!categorize words (noun

the hand is
or not?
not in the
bush. YES NO
YES NO
NO YES
YES NO
NO YES
Results
Results Results
! 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
! when you have to respond verbally, it is easier to work
! diagrams -- pointing with diagram information than sentence information
! sentence -- pointing
Mental task
! diagrams -- verbal
Diagrams Sentences
Response task
! 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
!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
!one deals with verbal information and must do

the spoken response and the sentence task
Diagrams Sentences Diagrams Sentences

Visuo-spatial Verbal Visuo-spatial Verbal
Pointing Verbal Pointing Verbal
information information information information
All together now

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

! hypothesizes mechanisms that lead to memory properties
Visuo- Phono-
spatial logical
sketchpad loop

Sternberg s study Properties of phonological loop
!controller system Data
Brook s study ! phonological similarity effect
! articulatory suppression
!separate visual and verbal systems
! word length effect
Baddley s working memory model ! irrelevant speech effect
!Central executive CogLabs on Memory span and Phonological
!Visuo-spatial sketchpad similarity due!
!Phonological loop A problem with IQ tests.

Working memory Working memory
Current thought, awareness

PSY 200
!extension of short-term memory Central

executive
Greg Francis !small capacity

!rapid forgetting
Lecture 16 Visuo-
spatial
Phonological
Processor of information sketchpad

loop
!not a storage device

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

Two components How many items can be
! Articulatory control process
(ACP)
kept in the phonological
Phonological loop Phonological loop
converts non-speech loop?
information into speech
code
ACP Depends on two factors ACP
rehearsal / refresh
! Phonological store (PS) !Duration before decay from
similar to how we first PS
described STM (items
decay from memory) PS PS
!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
of them fade
Phonological loop
on the rate of rehearsal (r)
! The duration of decay in the PS
A set of items that takes ACP
Memory span should follow the equation
! 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 ! more likely that some items PS
for English speakers will drop out before you get
Estimate duration of decay in PS (d) back to the first item
! d=1.75 seconds

Effect of rehearsal rate Word length effect

Explains differences across groups of people
Age effects in children Memory span is
! Hitch, Halliday & Littler (1989) related to the length
This implies that it is not the of words
loop size that changes with ! Number of syllables
age, but the rate of
rehearsal Nicely matched by
changes in reading
speed
Span=1.68(Oral Reading Rate)+0.71 ! Rate of rehearsal
Language effects Relation to IQ

Ellis & Henley (1980)
Some ! investigated complaints about WISC
languages are intelligence scores
spoken more ! Welsch children tended to score lower
quickly than than English children
others
Part of the exam checks memory span
Should allow
! and the slower rate of speech in Welsch partly
larger memory
explains the difference
span
! bilingual Welsch students tested in English got better
! it does
scores than when tested in Welsch
Articulatory suppression Articulatory suppression

Subjects sees (hears) a list of phonemes
Also repeats a phrase over and over
Repeating phrase ties up the ACP
! e.g., tippy-toe, tippy-toe, tippy-toe,... !Without rehearsal more forgetting occurs
Recall is worse
! True for both auditory and visual presentation
! (Recall for visual may be better than auditory because there is Phonological loop
some information in the visuospatial sketchpad as well) Visual Auditory
1
0.9 ACP
Proportion recalled
0.8
0.7
0.6
tippy-toe
0.5 Silent
0.4 Suppression
0.3
0.2
PS
0.1
0
Visual Auditory

Phonological similarity Phonological similarity

All items are stored in phonological loop
Memory of a list of items is worse ! similar sounding items interfere with each other in the
phonological loop
when the items sound the same ! two possibilities:
1) harder to rehearse (effect in the ACP)
2) fade more quickly (effect in the PS)

B F H N Better
recall ACP
N
ACP
G P
B G P T Worse B
F
recall PS PS
H B T
Phonological similarity Locus of similarity effect

All items are stored in phonological loop Studies find a phonological similarity effect for auditory stimuli
! similar sounding items interfere with each other in the
under articulatory suppression
phonological loop ! We suggested two possibilities:
! two possibilities:
2) fade more quickly (effect in the PS)
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
ACP ACP
F
PS N
PSB G
H B P T
CogLab data CogLab data

The CogLab experiment on memory span shows data The CogLab experiment on phonological suppression
in agreement with our expectations (190 subjects) shows data in agreement with our expectations (190
subjects)

Irrelevant speech effect

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

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

Data accounted for by phonological loop
! word length effect Review for Exam 2
! phonological similarity
! articulatory suppression
After exam 2
! irrelevant speech effect
Encoding specificity
Don t listen to lyrical music while studying
! 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
! e.g., part-set cueing
Greg Francis ! given part of a set that has been learned, subjects recall
fewer of the remaining items than with normal free recall
Lecture 17 ! 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 search
you must consider the conditions at test
Information must be encoded in memory as well as
! Memory is more often about discrimination of memory traces and
not about the strength of memory traces recalled
! Similar to visual search experiments ! Such encoding can alter what features are stored as part of the memory
! Which changes the discrimination of subsequent recall
Feature
search
It turns out, that to maximize recallability
! the effort and conditions at the time of learning must be
consistent with the properties and conditions of the test
Encoding specificity principle

Conjunctive
search
Representative study Test

Each group is then split into two subgroups that vary in the recall of
Subjects in two groups information
! see the same words, but have different tasks ! 1) Normal recognition task
! 2) Shown a word and asked if any of the target words rhymed with this
! This changes the encoding of information in memory
word
9
Semantic judgement Rhyme judgement 8
Number of words
7
Study
recognized
CHEESE 6
CHEESE Semantic
5
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
! or that one testing strategy is better than the other ! 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
! subject s performance depends on many factors paper-BOOK BOOK
... ...
Encoding specificity Encoding specificity
Each group split into two groups for testing

Results
recall !cue does not always help
16
! 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
Surrounding context Surrounding context

Deep-sea divers learn words (Godden Test for recall on land (dry) or under water (wet)
& Baddeley, 1975), either
!on land (dry)
!under water (wet) Study
EAGLE
HEALTH EAGLE
WHEEL HEALTH
BOOK WHEEL
... BOOK
...

Significance Forgetting
Decompression tables for divers
Forgetting is not always a characteristic of a
! want to remember when under water
memory system, or your brain
! generally study while on land
!although it could be in some cases, it is not
Researchers working under water have difficulty
always
recalling their details on land
! E.g., counts of species Forgetting must be defined operationally
How do you know if something is forgotten? !specify the task and context of retrieval
! changing context may allow subject to recall seemingly !You can never be certain that if you are placed
forgotten information
in a different context you will still show forgetting
! forgetting = retrieval problem?
Internal context Internal context

Best recall if study and test states are the same
Goodwin et al. (1969) ! Similar effects for marijuana cigarettes (Eich et al., 1975)
Subjects drink 10 oz. 80 proof vodka, mixed in
sugar-free lemon-lime drink
! or a similar tasting drink (bit of vodka on top of drink)
Study state
Test memory 24 hours later
Subjects are either:
! Sober at study, sober at test
! Drunk at study, sober at test
! Sober at study, drunk at test
! Drunk at study, drunk at test
Testing Mood
Mood has a similar effect (Eich et al, 1994)
! Mood induced by music and directed thoughts
So, if you are intoxicated while studying for an
exam
!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)
! significant for final exams! Recall was best for words that were studied in the test context
Smith et al. (1978)
! Subject studied words in one of two contexts (on separate days)
! Varied classroom and dress of experimenter
Study context

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

Memory discrimination Discrimination

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

! 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
! 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
! the special target is rough, which is not shown to the
! as in CogLab subject
! subject views a list of words
After viewing the list, the subject must go through a
! 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
! some words were in the list
! 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
! even though it was not memories are
Sometimes people will even report that they ! not necessarily accurate, we can remember things that
recall seeing the special target never occurred
! but this is impossible because it was never shown ! able to be manipulated, to a certain extent, I can make
you have certain memories
CogLab data (68 participants)
! Type of selected items Percentage of recalls Why does the false memory effect
! In original list 80.8 happen?
! Normal distractor (not in list) 7.2
! Special distractor (not in list) 69.9

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
! Other information is automatically generated as well ! 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
! smooth, bumpy, tough, road, sandpaper, jagged, ready, ! Reporting all items from memory is not going to lead to good performance
coarse, uneven, riders, rugged, sand, boards, ground, gravel ! 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
bumpy bumpy jagged stones
! Source monitoring bumpy
smooth smooth smooth
smooth peaks Memory
rough rough rough
tough tough rough
baby baby baby nails sandpaper
baby nails
road road
gravel road
uneven
Discrimination Interference
Good memory recall usually requires not only Retroactive interference (RI)
recall of an item from memory !new information prevents recall of
previous information
You also must identify the correct item relative
!e.g., Overwriting a computer file.
to the appropriate context or time frame
!The current trial Proactive interference (PI)
!The context of the experiment !prior learning prohibits new learning
!Relative to an earlier event !e.g., Learning new cultural customs.
!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
! You have a match in memory, but is from trial 1, not trial 3
! One is that memory involves discriminating new from old
! You may report it having been shown in trial 3
Visual memory
In general, previous trials make memory discrimination
! See a set of photos more challenging
! 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

Proactive interference Proactive interference

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
! I looked at recall of the first letter in each list, averaged across ! 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
Proactive interference Release from PI

Inference does not happen for all experiments
! Just those related to memory Proactive interference weakens for different
Partial report experiment (first 12 trials had the cue before stimulus types
the letter matrix to give you practice)
Run two Brown-Peterson type experiments
0.9
Proportion of letters reported
0.8
0.7 Control Experimental
0.6
0.5 XJF Trial 1 XJF
0.4
0.3
WRM Trial 2 WRM
0.2
0.1
0 DBL Trial 3 DBL
1 2 3 4 5 6 7 8 9 10 11
Trial NRX Trial 4 942
Release from PI Release from PI
Trials 1-3 show build up of PI Works for many kinds of memory tasks
Experimental group shows release of PI on Many kinds of stimuli
Trial 4 100 5, 7, 9, 1 NICE, SUNNY, ENJOY, PUPPY

vs vs
Percentage correct
80
5, 7, 9, HAND NICE, SUNNY, ENJOY, KILL
60
Control
40 Experiment
20
D, D, D, D
News stories vs
0
D, D, D, F
1 2 3 4
Trial

Memory system Working memory

Every memory system must have at least two
For example, working memory has a storage

components/processes interference hypothesis for the phonological loop

! Storage
Working memory suggests that interference can
! retrieval
occur
We have described proactive interference as being
! by blocking ACP rehearsal (articulatory suppression,
due to difficulty discriminating new items from Brown-Peterson task, word length effect)
previous items ! within the PS when items sound similar
But there is an alternative explanation ! both of these block the storage of items (items fall out of the
loop)
! Proactive interference might prevent items from being
stored and thereby make them unrecallable
Testing models
Interference at recall
Test location of PI by changing
If PI prevented the last item from being stored your telling a
instructions after the list is presented
subject that the fourth item was an indoor sport, should
Experiment make no difference (other than guessing)
! stimuli are names of indoor and outdoor ! but it makes a big difference, they show release from PI
Trial 1 FOOTBALL
games
100
Percentage correct
! subjects usually do not notice that word Trial 2
on the fourth trial is an indoor game and
SOCCER 80
others are outdoor games 60

Trial 3 BASEBALL Not told
Take two groups of subjects 40 Told
! one has traditional PI type experiment Trial 4 WALLYBALL 20

! one is told of difference on fourth trial, at
0
the time of test
1 2 3 4
Purdue University
Trial
Purdue University
How to take a test Conclusions

Avoid PI
Discrimination
Answering successive questions on the
same topic hurts recall Retroactive interference
! after answering unrelated questions Proactive interference
! go back to questions you cannot answer
Release from PI
! less proactive interference
! should recall more Strong effects
T1, T1, T1, T1 Knowing about can help in everyday tasks

ACT vs
T1, T1, T1, T2

Next time
Constructive memory
Flashbulb memories
Memory misattribution
Misleading questions
How good is eye-witness testimony?
Purdue University

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 ! Epilepsy patients
several things ! stimulate brain regions before operating
! Recall information that might be related to the task from memory ! want to know what is being removed
! Determine if the memory is actually for the correct event Conscious patients report vivid memories
! Determine if the memory is actually for the correct moment in time ! unable to recall normally
! Gauge your confidence in the memory s validity "she saw herself as she had been while giving birth to her
baby."
All of this suggests that performance on a memory task ! stimulation of temporal lobes
involves discriminating information
In the image, numbers indicate
To address the discrimination problem, people engage places where stimulation
in a constructive process to report memories evoked different reported
experiences
No forgetting? Penfield (1959)

Suggests that memories are stored but normally Even worse
unreachable (context things again) ! the memories are nearly impossible to verify
Basis for ideas of memory repression (and a few self-help books) ! the few attempts find that the memories are not true
! The results are usually misunderstood people describe places they have never visited, impossible
events, fantasy,
! Actually only occurred for 5% of patients ! 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
! your awareness of remembrance is a product of your brain
! 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. ! where they were when they
heard
! JFK assassination
! Challenger explosion ! what people said
! Oklahoma City bombing ! clothing worn
! Earthquakes ! time of day,..
! September 11, 2001 People are confident about

! ... their reports
! 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 to be
! The WTC attack
reliable
! An ordinary event (volunteer s choice)
Three groups for follow up Ordinary memories
! 7 days later ! Faded in vividness
! 42 days later ! Belief decreased over time
! 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)
! 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 ! Ironically, he was discussing memory of faces for eyewitness
! Maybe not the true memory testimony
Flashbulb memories are a real phenomenon The victim

about the experience of memory, ! had the TV on at the time of rape
! misattributed the face on TV for the face of her attacker
! but probably not super-memory
! 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

Questions Eyewitness testimony

Did the bus, which came by, come from the left or the Later, show slides and ask subjects if they were part of the
right? original set
Did another car pass the Red Datsun while it was at ! real slide contains YIELD sign
the intersection with the stop sign? ! 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
! Loftus & Palmer (1974) The Delayed
Percentage correct

60
misinformation Immediate
Compare accuracy according to pre-test questions 50
effect gets
! Subjects without a misleading question--90% accurate stronger with a
40
! Subjects with a misleading question -- 20% accurate week delay 30

before the 20
! In a follow-up, the experimenters asked those with misleading
questions if they thought they were misled memory test 10
90% say no 0
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
! Basically just have subject read a plausible story (with ! flashbulb memories
some details that could be true)
! brain stimulation
! Get family members to pretend the story is true
! Later the subject remembers the story as something
Memory is constructive
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
What s wrong with my wife?
Purdue 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 !Confidence in the memory is another cognitive
experience
!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
! retrograde: forgetting events prior to the injury
5 months 8 months 16 months
! anterograde: forgetting events after the injury
In most cases amnesia is limited in scope and

duration
coma coma Time
! 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!
! Sense of smell
! Mild anomia
! Odd aphasia (language deficit)
She is able to learn and remember new information

Remarkably unaffected by the loss of memories
! Personality
But she cannot remember anything before ! Parents

! college
her senior year in high school
! makes study of retrograde amnesia difficult
! motor cycle accident
! complete retrograde amnesia

What is lost? What is lost?

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

There are many different types of memory
Amnesia seems to affect declarative more than nondeclarative memory
Some patients have amnesia that preserves past
memories but prevents formation of new memories
! 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 ! how many trials to repeat back list correctly?
! unable to learn anything new 30
25
Thought it was 1953
Mean trials to criterion
20
! shocked by age of face in his mirror
Controls
! Could not stand to read newspapers 15
Patients
! 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 ! E.g.,
90
location of
80
hospital
Percent correct
70
60 cafeteria
Controls
50
40
Amnesics ! E.g. mirror
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
!infantile amnesia
!repressed memories
Careful studies are difficult to come by because

the memories (and absence thereof) must be
verified
!remember the fundamental fact at the start of
today s lecture
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
!children younger than 4-years-old view the world
20
Number of memories
differently from adults

15
!by encoding specificity, one needs to be in a
reported
10 similar state as study to best recall something

5 !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
!being unable to remember something
of the pain is prevented by psychological defense
mechanisms (repression) !being able to recover the memory through therapy
This is very unlikely !proving that the recovered memory is accurate

! people do remember painful events well
! laboratory studies find no evidence of repressed memories
Repression Conclusions
In therapy, clinicians often claim evidence of repression Retrograde amnesia

with
! dream interpretation
Anterograde amnesia
! patterns in symptoms Learning in anterograde amnesics
! recovering a memory through hypnosis
Infantile amnesia
None of these techniques demonstrate a verified
memory Repression
Among carefully controlled memory research, there is
no evidence of repression!
Next time
Levels of processing
Judgments of learning
Practice testing
Learning styles
How to improve your memory without spending
$20.
Purdue University

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

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

We know that memory is best when study and test contexts are Test subjects in a neutral context
similar (after another 3 hour interval)
! For example, testing in the study classroom
Look at proportion correct recall
But variability in study promotes more general recall ! Highest with variable study contexts
! Smith et al. (1978)
Advice: if you want to remember
! 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 ! Craik & Tulving (1975)
Style of study matters too Subjects observe words with associated tasks
! distributed practice is better than massed practice

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

Level of processing CogLab

Recall is better as depth of processing increases Recall is better as depth of processing increases
! More distinctive memories are created, which helps subsequent recall ! Test is what matters here, other data is just for completeness
! By varying depth of processing, you can construct memories that are ! 68 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
! 1 control group: told they will be tested to recall the words
not given any study task
! 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?
! 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)
! read notes Subjects study a pair of words (e.g., OCEAN TREE)
! rewrite notes
Estimate how likely they are to be able to remember
! rephrase notes one word if shown the other (JOL).
! teach someone else ! Given OCEAN, how likely to remember the associated item
More generally, people are not usually good at later?
estimating whether something will be remembered ! This is the subject s estimate of their ability to use LTM
Make judgment either
! Immediately after studying the pair
! 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)
! 1) Read material on both sides
! Especially for high (study)
JOLs ! 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
! Continue to study?
! Continue to test?
! 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
! mashua boat testing
Performance is best when
! kaka -- brother every pair is tested, even if
! 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
! ST (study-test): subject studies and continually tested over every pair The amount of time spent
studying the words does
! SnT (study on non-recalled - test on all): when a subject recalls a pair, it is no
longer studied, but it continues to be tested not matter so much Don t practice
test
! 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
! 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
!Encoding specificity Mental imagery
!level of processing Sleep
!Judgments of Learning Brain training
!Practice testing
CogLab on Link Word due!
!Learning styles
Get a good nights sleep!

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

Lecture 22 remembered?
! no
! but there are several tricks you can use
to improve memory in certain situations
Get a good nights 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
!how do they do it?
Long-distance runner
!some cheat (frauds)
! sequence like 3492 converted to
!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
!230 hours of practice (over 20 months)
other memory tasks (e.g.,
!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
! visualize walking around an area with distinctive landmarks
! link the items to be remembered with landmarks by using bizarre Add vivid,
ITEMS LOCI
mental imagery
bizarre imagery
! to recall items in order, mentally walk through area
hot dogs driveway
! (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 !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. Ten is a hen. bread shoe
bananas tree
Link word method Link word method

In a study of learning 120 Russian words (Atkinson
Foreign language vocabulary
& Raugh, 1975)
! find an English key word that sounds like some part of the
foreign word Two groups
! form a mental image of the key word interacting with the ! control: heard Russian words, saw English translation
English translation of the foreign word ! experimental: heard Russian words, saw English
E.G. translation, saw key words, and applied method
! pato -> Spanish for duck , sounds like pot-o Experimental group learned more words faster and
imagine duck with pot on its head for longer
! zronok-> Russian for bell , sounds like zrahn-oak ! 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 Some people seem to have extraordinary
Study French words memories
! Half with a provided link word to form an image
!professional - apply one of the techniques we ve
! Half without a provided link word (no image)
discussed
!spontaneous- seem to not consciously apply a
technique
Photographic memory?
!Few documented cases
!Generally, not happy outcomes

S.: Luria S.: Luria

Extreme synesthesia
Luria: Russian psychologist
!sensory information from one modality evokes
!met S in 1920s sensation in another
S !tone, 30 cps, 100 decibles --> saw a strip 12-14
!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
!voices gave rise to visual responses
!other unusual characteristics
!used the full sensation of events to help memory
Brain Training
S.: Luria Several companies market activities
to make you smarter
Visual imagery ! Exercise your brain with games that are
adapted from neuroscience
!used method of loci ! Does that even make sense?
Often aimed toward elderly

!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)
! Does the current stimulus match the one from n trials back? ! n is also a measure of how well subjects do the task
! 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 Brain Training (WARNING!)

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
! Training group does better than a control group
! It is probably a mistake to average scores across these studies
! Amount of n-back improvement is related to gain in intelligence
Some selective reporting of measures of fluid
intelligence
! Measures that did not show an effect were not reported
No comparison to an active control

! 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:
! We object to the claim that brain games offer consumers a scientifically grounded
avenue to reduce or reverse cognitive decline when there is no compelling Many types of memory improve with sleep
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 reflects Some type of consolidation of memories
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.
!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
! Playing video games does not seem to improve your attention or perception
You can improve performance on specific tasks, but that does not typically !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 ! 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
! 20 minutes later ! 20 minutes later
! 12 hours later ! 12 hours later
! 24 hours later ! 24 hours later
No differences when tested Big differences when tested
on the originally studied on new pairs that fit the
items ordered structure
! E.g., A>C, C>E, B>D
Sleep Conclusions
Half of the 12 hour group
had sleep and half did not Lots of ways to improve memory
It makes a difference for
pairs of items that are far !Method of loci
apart in the ordered !Imagery
structure
!Mnemonics
1-degree: A>C, B>D,
2-degree: A>D, B>E, !Brain training
Advice: !Sleep
! study early!
! Get some sleep!
Next time
Mental representation
Prototypes
Exemplars
Propositions
CogLab on Prototypes due!
What is a shoe?
Purdue University

Professor Greg Francis 7/31/15
Concepts
Representation of knowledge
What is the information in Long Term Memory?
PSY 200 ! May be several different types
We have knowledge about the world

Greg Francis ! Due to personal experience
! 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?
! 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
!Definitions (don t really work) trying to define terms like virtue and
!Prototypes (closer to how humans think) knowledge
!Exemplars (more likely than prototypes) !they were largely unsuccessful
And then combinations of concepts the 20th century philosopher Wittgenstein

!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
! A covering for the human foot, usually made of leather, having a
! 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. ! Anything resembling a shoe in form, position, or use.
! 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
! 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
! 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. ! for any definition you come up with, I can find examples that do
! 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
! so our knowledge of this concept must not be based on some
! a shoe filled with cement, which cannot be worn
precise definition
! 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
! And this? "
by mating abilities)
! 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
!there may be no absolutely necessary characteristics Consider the concept coffee cup
!there may be no absolutely sufficient characteristics
Prototype theory supposes that similarity is

judged relative to a prototype example of the
concept
!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
!and variations (some are cup-ier than others) !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
!and variations (some are cup-ier than others) !and variations (some are cup-ier than others)
Lots of experiments suggest the role of Prototypes
prototypes
!Posner & Keele (1968): learning category names for
random dot patterns
!Discriminate two sets of random dot patterns
!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
! reaction time for judgment is recorded for stimuli they have never
seen before
new variants
the prototypes
! reaction time is faster for the prototypes
! 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

!they never see the prototypes themselves

Consider the types of concepts you can have
Results are based on data from 156 participants.
! Pattern type Reaction time (ms) ! and how specific they can be
! Prototypes 645 things: bird, dog, chair, shoe,
! Variants 690
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
! a thing that resides in memory and contains information about the category
features? in case of a fire ,...
! 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
!it s inconceivable that every possible prototype exists !e.g., a coffee cup concept might contain lots of
ready to be used examples of coffee cups
!some must just be built as they are needed
!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
!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
! things doing something
involves combinations of concepts?
!e.g., Dogs chase cats. Statement that is either true or false
! things cannot be judged true or false
!e.g., Last Spring, Jacob fed the pigeons in
! e.g., Book, Albert, Threw, Professor, Test, Gave
Trafalgar Square.
!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
! The proposition connects the appropriate concept nodes
!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
! The proposition connects the appropriate concept nodes
last Spring
time
Dog
pigeons
agent object
in
Jacob
chase Last Spring, Jacob fed pigeons relation
Dogs chase cats
pigeons in
relation agent
Trafalgar Square
object location
relation
cat feeds Trafalgar Square

Ratcliff & McKoon (1978)
! study phase
One way of combining concepts subjects are asked to memorize a set of 504 sentences
18 - 1 hour sessions!
!there are also other theories of how to do this ! test phase
show words and have subjects decide if they were in the
Used a lot in Artificial Intelligence 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
bandit bandit
agent agent
agent faked faked
bandit stole passport bandit stole passport

relation 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
bandit bandit
agent agent
faked faked

relation relation

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
bandit bandit
agent agent
faked faked

relation relation

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
bandit bandit
agent agent
faked faked

relation relation


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

agent
faked
Ratcliff & McKoon (1978)
bandit stole passport
relation ! results
bandit faked signature within same proposition words: 561 msec
between proposition words: 581 msec
unrelated: 671
relation object
! evidence of priming by propositional activation
stole signature
Purdue University We think in propositions! Purdue University
Conclusions
Next time
Concepts Other types of knowledge
!definitions Mental images
!prototypes ! mental rotation
!exemplars ! mental scaling
! limitations of
Propositions
!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 !concepts (prototypes, exemplars)
!propositions
Lecture 24
!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
!sights information in the
!smells image?
!touches !analog: copy of
!sounds image in head and
we can retrieve it
Are these converted into propositions, or
!symbology: convert to
concepts
propositions/concepts
!or is there something else?
Images Images
If you ask me questions about the previous slide,
I can remember the wears
Runs to
my answers would not necessarily identify the
image on the ball
girl representation
previous slide and it shorts
feels like I picture

Runs to
wears
the image in my
ball
head girl
shorts
! but is it really?
! could it be a
Mental
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
! but you cannot
! how you interpret the image to a large extent determines what
you know about it
Mental images are not exactly like real images

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

Answer these questions: How did you do?
Which is further North, Seattle, Washington or

Montreal, Canada?
Which is further East, Detroit, Michigan or
Indianapolis, Indiana?
Which is further South, Boston, Massachusetts or
New York City, New York?
Which is further West, Reno, Nevada or San
Diego, California?
Propositions Propositions
So this suggests that mental images are not exactly like real
images It is clear that propositional information influences
! and something like propositional information likely influences mental imagery
reports that are ostensibly based on mental images or mental ! but is it all propositions?
San Diego
maps
in ! are there mental images, as we tend to experience them?
agent Nevada
relation Is there any reason to believe that mental images
San Diego is in in
are at all analogous to real images?
California object
relation ! 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
!If the mental image is a description (set of !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)
! 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) ! presumably because the propositions about a lion having claws are
! 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
! Subjects forming a mental image respond more quickly to the head
! 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 1400
Reaction time (ms)
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
! 67 participants
(1971)
The more rotated the
!3-D shapes
stimuli are, the longer
!rotated in plane it takes to make a
!or in depth response
!decide if shapes ! close to linear increase
in RT
are same are
different
!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
! imagined movement of the mental image (rotation) resembles ! some people report their mental images are just like real images
actual movement ! others report they are fuzzy and vague
! It takes time to mentally move through a mental space ! some people report no mental images at all (10%)
The CogLab data suggests it is about 100 degrees/
! Nevertheless, people all do basically the same on many tasks that
second
seem to require operating on mental images
10 milliseconds for each degree
! no reason why propositions would give data that incorporate
spatial and temporal relations between aspects of the mental
images
fMRI fMRI
Hugdahl et al. (2006) had subjects perform a 3-D mental Compared to the 2-D condition, the 3-D condition has
rotation task or a control task (no judgment, just look at 2- more activity in the superior parietal lobes
D oriented bars and press a button) in an fMRI scanner
Males
Females

Review for exam 3
Mental representations of pictorial
After Exam 3
information
! Start a discussion of language

There are some things you cannot do with ! Language as an instinct
mental images ! 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
!sentences
Greg Francis !words
!sounds
Lecture 25
!structure
!interpretation
Why we do not have to worry about The language instinct
teaching language in school. !Pinker (1994)
Preconceptions Cultural influences
We tend to think of language as Culture does influence language

!a great invention of human cognition ! Consider words in English
Some derived from the invading Normans (1066)
!taught to children (considered sophisticated and polite)
Some derived from the Anglo-Saxon language of the
!taught in schools British Isles (considered crude by the invaders)
! Norman: perspiration, dine, deceased, desire, urine,
!a cultural invention
excrement
This is wrong! ! Anglo-Saxon: sweat, eat, dead, want, piss, shit
!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 !blinded birds cannot navigate by the stars
!Darwin (1871) !Atlantic Ocean turtles that navigate by magnetic
fields need to be in the correct ocean
Humans instinctively learn language
!effortless Language development needs exposure to
!unconscious other people for communication
!procedural knowledge !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 !difficult to test because we rarely get to see a
!otherwise they would never come up with language created from a non-language
statements like !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!
!create a jargon called a pidgin
Pidgin
Pidgin The Ten commandments in pidgen
For example, in New Guinea !as translated by the Alexishafen Catholic Mission in
1937
!pidgin is similar to English (rulers of the plantation) ! 1. Mi Master, God bilong yu, yu no ken mekim masalai end ol
tambaran.
woman: meri (Mary, generic word for woman)
! 2. Yu no ken kolim nating nem bilong God.
another man s wife: meri bilong enaderfelo man ! 3. Yu must santuium sande.
! 4. Yu mast mekin gud long papamama bilong yu.
hair: grass bilong hed ! 5. Yu no ken kilim man.
! 6. Yu no ken brukim fashin bilong marit.
helicopter: mixmasta bilong Jesus Christ
! 7. Yu no ken stilim samting.
coffin: die bokus ! 8. Yu no ken lai.
! 9. Yu no ken duim meri bilong enaderfelo man.
piano: bokus bilong teeth yu hitim teeth bokus is cry ! 10. Yu no ken laik stilim samting.
Learning Learning
Pidgin is not a true language
In Hawaii at the turn of the century !word order is arbitrary
!workers from China, Japan, Korea, Portugal, !no rules
The Philippines, and Puerto Rico were Me cape buy, me check make.
!no tenses
brought in to harvest sugar
!no prefixes or suffixes
!they developed a pidgin
!can only be understood in context of the
!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 !tried to teach children to lip-read (poor
!if removed from parents (and so unable to learn results)
native tongue) !but children started making a pidgin on the
!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
!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 ! including studies of grammar
! 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,

!parents learned American Sign Language (ASL) late
in life and so are not very good at it pronouns, prepositional phrases, verbs, adverbs,
Education is good for reading and writing
Simon had little contact with other deaf people
! but writing is dramatically different from speaking
!but his signing was much better than his parents! ! 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 ! if you are bad --> if you bad is
! They are --> They re grammatically correct
! He is --> He s AAVE does not allow word dropping arbitrarily

But you cannot always do this ! Yes he is! -->? Yes he!
! Yes he is! -->? Yes he s! ! Who is it? -->? Who it?
! 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
!none, they are just different !specialized skill among humans
!they are different dialects of English !children need little tutoring to learn language
!children invent language if one is not readily
Linguist Max Weinreich
available
! A language is a dialect with an army and a navy.
Language follows rules
The dialect you speak may give away your
!even when it doesn t seem to
personal history, but it is not fundamentally
worse than any other dialect.
Next time
Grammar
Long term dependencies
Phrases
Language universals
Dr. Francis says something new!
Purdue University

Greg Francis 8/17/16
Phrase trees Language

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

Symbols Discrete combinatorial system
!words are arbitrary !combinations of words
!the sound dog has nothing to do with dogs How many combinations?
!compare driving on parkway to parking on !if interrupted anywhere in the middle of a
driveway, blueberries and cranberries, sentence, there are about 10 words one could
hamburger select before finishing it up
Grammar !if sentences average around 20 words, that
means there are 10^20 sentences
!the order of words matters
!Dog bites man. vs. Man bites dog.
Grammar Vastness
But in fact, there are infinitely many It is amazing how powerful language is
different sentences
You have probably never heard the
!there is no limit to how long a sentence can
following sentence
be
!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 The poor student attendance, along with the too big
!Professor Francis said that, . classroom, sometimes makes the 8:30 section of cognitive
psychology feel like the lecture is given in a warehouse.
PSY200 Cognitive Psychology 1

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
!Colorless green ideas sleep furiously.
!Is raining. Sometimes you
still understand !If we don t succeed, we run the risk of failure.
!The child seems sleeping.
what was meant! (a not joking Dan Quayle)
!Sally poured the glass with water. ! Twas brillig, and the slithy toves
!It s a flying finches, they are. Did gyre and gimble in the wabe:
All mimsy were the borogoves,
!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
! we are interested in how people perceive grammar
! this is different from the grammar rules you may have
!it s not like learning to play a piano
learned in school! !or learning about statistical regularities in the
Which often focus on forming sentences that are easy
to understand environment (stimulus-response)
Statistics
Nonsense sentences
Think about the sentence If you just learned the statistical
!Colorless green ideas sleep furiously. combinations of words, you might think
What is the probability that in normal life you something like this was a grammatical
would hear the word green follow the word sentence
colorless ? House to ask for is to earn our living by
!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

Language has rules that determine what types of
The previous paragraph creates coherent
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 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)
!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 ifthen
part of a sentence
dependencies can be very long
! embed a sentence in a sentence
!Consider Ifthen and Eitheror
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.
! 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
Phrases Phrases
Every sentence is built out of phrases All noun phrases obey certain rules
!rewrite rules
The happy boy eats candy. NP-->(det)A*N the happy boy
!NP -- noun phrase the boy

The first three words form a unit called
a noun phrase (NP) !det -- determinant: the , a , an John
What identifies a !A -- adjective
The happy boy the tall slender woman
noun phrase? !N -- noun
This is not the same analysis you did !( ) -- optional
in grammar school! !* -- 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
!N --> boy, girl, candy, hot dogs, ice cream, S
!V --> eats, likes, bites,
NP VP
!det --> a, the, one,
V NP
!A --> lucky, tall,...
det A N
N
the happy boy eats
ice cream
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
!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
NP ways
V
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
!Morse code, 0-1 s, English, Spanish, tapping
toes, beeps,..
the girl eats candy
!an infinite number of ways to create a language
the girl eats ice cream
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 ! Subject-Verb-Object
languages are clustered together ! (There are exceptions: A bear he shot. )
This pattern is true for most of the world s

Human languages
Martian ! 98% of languages have the Subject before the Object (the
Dolphin Verb location varies across languages)
Alpha-
centaurian
! 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
of linguistic features
Language consists of
For example, if a language s preferred word order !symbols (words)
is Subject-Object-Verb !grammar (rules)
! the language is likely to form questions by adding some
words at the end of the question
Language is best described as phrase
If a language s preferred word order is Subject- trees
Verb-Object (like English) !explains long term dependencies
! the language is likely to form questions by adding some
words at the beginning of the question Language universals
! 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
!rules for combining phrases
Greg Francis !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 Words are not just a collection of letters
rules for combining phrases, they use !Word superiority effect
different words
!Judge a pair as being the same or different
Words are symbols that are arbitrary
in many respects
HRNO LITL TRIP DEAL
! dog is nothing like a dog
CRNO LITL TRAP DEAL
!is it rote memorization?
partly, but it is also more than that
Words are special Words are special

But words are faster (around 100 ms) and more
The judgment does not require you to

accurate than nonwords
read the words CogLab data (74 participants)
!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
! Show this page to a limited morphology
preschooler
!nouns have two forms
The child will say wugs
!verbs have four forms
even though he has never
quack
seen a wug before duck
quacks
! This implies that there must ducks
be a rule for pluralizing nouns quacked
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
to do so
variations
Simple present tense
!Italian and Spanish have 50 forms of each verb
! General truths: Ducks quack.
!classical Greek has 350 forms of each verb ! Habitual action: I quack like a duck when I wake up.
!Turkish has 2 million forms of each verb Present Perfect Progressive
!some languages build entire sentences around ! 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
!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
!and combinations can be combined
One looks to be easy
!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
!a root is a word that cannot Nstem Ninflection morphemes
be split into smaller parts !-able
-s
!some morphemes attach adjective stem affiix; means capable of
Nstem Nstemaffix being X d ;attach me to a verb stem
only to roots
!thus, Darwinianisms is a !-er
-ism
word, but Darwinismians is noun stem affix; means one who X s ;
not Nroot Nrootaffix attach me to a verb stem
-ian must attach to a root !...
Darwin -ian

Exceptions Language families

Many languages are related to each other and have
You can probably think of lots of exceptions to broad families
! Which is why you see so many similar words across languages
these types of rules
!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?
!English adopts the words but not the rules !walkmans?

!walkmen?
These exceptions tend to be very common words
!Neither feels quite right
!drink-drank sink-sank
!throw-threw ring-rang
To answer this question we have to understand
how the word walkman is formed and what it is
!sit-sat blow-blew
about
All derive from a proto-Indo-European language
!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
!toothbrush --> toothbrushes
morpheme Nstem
a thing that can be x -ed !sawtooth --> sawteeth
!crunchable
!snowman --> snowmen Nstem Nstem
!cruncher a thing that does x
!workman a type of person Is walkman a compound noun?
walk man
!sawtooth a type of tooth !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
!a walkman is not about a type of man Walkman Personal Stereos
!thus, the head is not what the word is about
A similar analysis explains the plural form of
!this tell us that walkman is more like a new word
than a compound word (e.g., electricity) names
Toronto Maple Leafs (not Leaves)
For headless compounds the irregular plural
Florida Marlins (not Marlin)
form is not appropriate
!no plural form feels correct because the word has I am sick of all the Mickey Mouses in this
no head administration. (not Mickey Mice)

Interpreting language
Words follow rules Parsers
Morphology / morphemes More ambiguity
Lexicon Computers vs humans

CogLab on Lexical decision due!
Explains characteristics of language
This is a valid sentence: Buffalo buffalo buffalo buffalo

buffalo buffalo buffalo buffalo.

Parsing Linguistics
PSY 200 So far we have looked at

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

sentence
This is a valid sentence: Today we look more closely at interpreting a

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
They are important. Tell me about your professor. generates some
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
! Try it yourself at by knowing the phrase tree structure of the sentence
! http://www.manifestation.com/neurotoys/eliza.php3 ! indicates some aspects of meaning
The difference between
The green idea eats the girl s candy.
! Dog bites man.
! Man bites dog. We know the sentence is about an idea rather than a
girl
Requires identifying the subject, object, and verb
! 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
! leads to ambiguous sentences
yesterday.
! 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
discuss with discuss with
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
!it suggests that we think in some way that is !The dog likes ice cream.
different from language S
!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 ! (1) Phrases are not always consistent with word order
! (2) The same spoken sounds are sometimes used for
!a mental click of understanding words with different meanings (noun vs verb vs
adjective)
Each word has its role defined
!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 ! when a phrase tree includes many unfilled branches of the
same type (PP)
! the parser becomes confused as to which phrase is
associated with a new word
! ends up backtracking to sort out the phrases
! sometimes falls apart ( has has has )
The grammar generator and the parser are different
things in your language system
S PP PP PP
! these are grammatically correct sentences
The rapidity that the motion that the wing that the hummingbird has has has is remarkable. ! they are not good sentences Don t
! you make sentences like these make me
show you
your exams!

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
!The plastic pencil marks... !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
structure created! structure created!
the plastic pencil the plastic pencil
Word ambiguity Lexical decision

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
!The plastic pencil marks were ugly. (noun) task, which you did in CogLab
!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
! RT is faster if the second word is semantically related to the
!many times it is so fast that we do not notice
first word
!seems effortless
Lexical decision Evidence of ambiguity

The data find that RT is faster if the second word is meaningfully
related to the first word
We can apply the lexical decision task to the ambiguity
CogLab Data (69 participants) of parsing (Swinney, 1979)
Condition Reaction time (ms) Consider the following paragraph, which subjects
Associated words 675 listened to
Unassociated words 693 ! Rumor had it that, for years, the government had been
plagued with problems. The man was not surprised when he
Nonwords 801 found several spiders, roaches, and other bugs in the corner
of his room.
The word bugs is ambiguous

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

Sentence ambiguity
Evidence of ambiguity
Interestingly, people often miss ambiguities in
No one notices the ambiguity

sentences
But, give a lexical decision test for words verses non-
! Time flies like an arrow.
words
Humans recognize only one interpretation
! Flashed visually on a screen just after the word was spoken
! Subjects respond faster for words related to either definition of Computer algorithms can find 5 interpretations
bug ! 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 !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
! describe stereotypical properties of a situation
communication
! 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 ! 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

!Woman: I m leaving you.
!Man: Who is he?

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 due
Why do we say razzle-dazzle instead of

dazzle-razzel ?
Purdue University

Speech Language
PSY 200 Many levels

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

spoken language
Why do we say razzle-dazzle instead
!how does language depend on speech?
of dazzle-razzle ?
!what are the units of speech?
Illusions Illusions
When you hear what I say, you think you
The blurriness of speech explains some long-held
hear at least confusions
!separate words ! Oronyms (Mondegreens)
!separate syllables
But you do not

!words actually overlap in the speech signal
The good candy came anyways. The good can decay many ways.
!it is nearly impossible to take a speech signal
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
!analogous to the critical flicker frequency in the
eye Normal speech provides 10 to 15 distinct
!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
!above that, a series of clicks sounds like a
continuous buzz Artificially fast speech is 40 to 50
phonemes per second

Phonemes Phonemes
phoneme \'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
!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
! and we can interpret speech that seems to contain 50
phonemes per second !and this can lead to misinterpretations
! 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
but interprets them as more than 20 different !The programs do not combine phonemes in the
phonemes right way
Speech Physiology
So what are phonemes?
Lungs push
All speech is made of sounds
air out to
! sound is a pattern of pressure on the ear
! a tuning fork vibrates back and forth to make the sound of a make a
pure tone sound
! Frequency of vibration corresponds to pitch of the sound
!other
Speech consists of lots of patterns of this sort
organs
! 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 ! boot book
!bet butt The lips add additional frequencies to make different
!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
!this is true for all vowels tract
Consonants Consonants
Consonants are more complicated (3) Manner of articulation
!different type of control of air flow !/d/, /t/ (stop)
(1) Voicing: vibration of vocal cords ! /m/ (nasal)
!/b/, /d/, /m/, /w/, /v/ (voiced) !/f/, /v/ (fricative)
! /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
!/d/, /t/ (upper gum)
articulation
!/m/, /b/, /p/ (lips)
!/f/, /v/ (lip and teeth)
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/
and /pa/ is the timing of the release of air for the dazzle-razzle?
consonant and the voicing of the vowel !for phrases like this, people always first say the
Voice Onset Time (VOT) is short for /ba/ and longer word with a leading consonant that impedes air flow
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
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
! Rotokas (Papua New Guinea) uses 6 (and 5 vowels) phonemes
! Khoisian (Bushman) uses 141
Uses clicks as consonants Thus, we can identify possible words from
No language uses some possible sounds impossible words
! raspberries, scraping teeth, squawking,
!plast ptak
! Note, these sounds are used for communication, but not as
part of language! !vlas rtut
Japanese does not distinguish /r/ from /l/ !thole hlad
Purdue University
!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 !we are tuned to recognize the new sounds
!to say sounds faster, people use as coarticulation
coarticulation
This is the main reason computers have
!shape tongue in advanced preparation for
a hard time recognizing human speech!
the next phoneme
!this 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
!Cape Cod
two vowels, you do not actually stop
Note too, that the /s/ becomes /sh/ in !flapping
!horseshoe
slapped --> slapt
And /n/ becomes /m/ in
patting --> padding
!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 ! a problem for learning how to read!
Nor should it
George Bernard Shaw (among others)
! if words were spelled the way they were pronounced, we
complained about spelling in English would lose the visual connection between words
!he noted you could spell fish as g-h-o-t-i ! slap --> slapped would become slapt
! write --> writing would become wridding
! National Public Radio --> NPR would become MPR
gh -- tough o -- women ti -- nation
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
! Drawn characters indicate Phonemes
how consonants are
pronounced
Articulation
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

!your brain is wired to work with grammars, words,
Greg Francis phrases
Lecture 30 Learning
!you do have to learn some specifics for your
native tongue
When should you learn a foreign !rules
language? !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 !babies are interested in new things
know? !attach a tape player to a pacifier
When have you mastered language? !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
!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
! babies suck more often ! you can change the pronunciation (voice onset timing) of /ba/ and
still hear it as /ba/ (CogLab data)
! ba, ba, , pa, pa, pa, pa, pa,
! But too big a change in VOT turns it into /pa/
! 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
! 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
! occurs because they hear mother s voice while in the
! ba-ba-ba-ba-ba
womb
! neh-neh-neh
Mostly prefer the melody, stress, timing ! da-da-da-da-da
! French infants like French and Italian equally well
! 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

! patterns are common across languages
combinations lead to different sounds
By the end of the first year babies combine syllables ! necessary to produce speech
to sound like words By about 10 months babies learn the sounds of their
! neh-nee
native tongue
! da-dee
! they can no longer distinguish phonemes that are not part of
! meh-neh the language
Babbling is important ! Part of learning is forgetting!
! children who do not babble often show slower speech
development
! 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
!1) Cooing (first several months) !learned one new word every two waking hours
!2) Babbling (~6 months) !this is without knowing how to read!
!3) One word utterances (~1 year) The average high school graduate knows
about 60,000 different words (not counting
!4) Two-word utterances and telegraphic speech
compound words and such)
(1-3 years)
!means that in 17 years of life (not counting the first
!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
!All dry. All messy. All wet.
CogLab experiment used a Lexical decision task
to measure reaction time to words that were !I sit. I shut. No bed.
! Learned relatively !No pee. See baby. See pretty.
early in life (Early AoA:
age 6 or sooner) Content is similar for all languages
! Learned relatively late
(Late AoA: age 7 !objects appear, disappear, move,
or later)
!people do things, see things,
!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)
!that is because there are lots of opportunities for
! (2;3) Play checkers. Big drum. I got horn.
errors
! (2;5) Now put boots on. Where wrench go? What that paper
clip doing? !but pick any particular grammatical rule and you
! (2;7) Ursula has a boot on. Shadow has hat like that. find most three year olds obey it most of the time
! (2;9) Where Mommy keep her pocket book? Show you !this is amazing because there lots of cases that you
something funny. would expect would be difficult to learn
! (2;11) Why you mixing baby chocolate? I finishing drinking all
up down my throat.
! (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
! -s to pluralize a noun
Out of 66,000 sentences, children never made Mouses, leafs
! -ed to make the past tense of a verb
these errors
My teacher holded the baby rabbits and we patted them.
He seems happy. --> He is smiling. --> Hey, Horton heared a Who.
Does he seem happy? Does he be smiling? I finded Renee.
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
has around 180 irregular verbs ! tread - trod strive -strove
! inherited from other languages ! dwell - dwelt slay-slew
! These past-tense forms are not derived from rules ! rend - rent smite - smote
Irregular forms have to be memorized, word by word Sound weird because we do not often hear them
! many adults regularize
If a child cannot remember (in its lexicon) the words
! s/he defaults to the rule ! treaded, strived, dwelled,
slayed, rended, smited
These errors are for the most difficult parts of a
! thus language changes!
language to learn
! 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 ! I go to the bathroom. --> Go me to the bathroom.
some verbs and not others ! Aunt Jane died. --> The tiger will come and eat David and
! 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 ! I drink with a cup. --> Yawny Baby - you can push her mouth
open to drink her.
Thus you can say
! The butter melted. --> Sally melted the butter. Many of these errors would be fine in other languages!
! 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 ! This rule must be memorized as applying to some verbs and
not others
! I like sausage. --> I am liked of sausage.
! 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
! Sparkle your table with Cape Cod classic glass-ware. fluent in a second language
! Well, that decided me. ! children do it easily
! This new golf ball could obsolete many golf courses. What accounts for the difference?
! If she subscribes us up, she ll get a bonus. ! most likely it is age
! Boiler up! ! there seems to be a critical period during which
language can be learned
Children s errors tend to track the more difficult
! beyond age six (or so) it becomes more difficult to learn
aspects of a language, relative to other languages
a language (first or second)
! Adults make the same kinds of mistakes for still more difficult
to remember cases

Second language Conclusions

High school and college (or later) is just about the
worst time to try to learn a second language
Language development
It should be in kindergarten or preschool
Immigrants who arrive after age 6 may never fully
Stages of learning
learn a second language Errors
Children who fail to learn any language by age 6 Second language
never do
! 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

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

Some stroke patients Some stroke patients show
show agrammatical agrammatical speech
! repetition Do you drive
speech home on weekends?
! short sentences
Seem to know what ! true for both written and spoken
they want to say ! no problem controlling mouth Why, yesThursday,
e.g. blowing out candles er, er, er, no, er
!But are unable to say it FridayBarba-rawife
and, oh, cardrive
purnpikeyou know
Right Left
hemisphere hemisphere
reset andteevee.
Brain damage Broca s aphasia

Mr. Ford
Broca s area ==> Broca s aphasia ! omitted endings (-ed, -s)
Wernicke s area ==> Wernicke s aphasia ! omitted function words (or, be, the)
! skipped function words when reading (or, be, the) but read
similar sounding words (oar, bee)
! named objects and recognized names
! 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 !Do you use a hammer for cutting?
down! Oh boy! Okay Awright.
Okay ...Cookie is down...fall, and !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 !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
! fMRI while subjects judge
whether Italian sentences Seem to be able to say
are grammatically correct
things
or not
! At start, subjects did not !But what they say is
know rules of Italian almost meaningless
Right Left
! Broca s area is activated hemisphere hemisphere
! 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
let go and they're both gonna fall down...but already then...I can't see
Patients show across well enough but I believe that either she or will have some food that's
What not good for you and she's to get some for her too...and that you get it
!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
!poor vocabulary hospital?
come out right, the uh, the stuff that's uh, good for, it's not good for you
but it, but you love it, um mum mum (smacks lips)...and that so
they've...see that, I can't see whether it's in there or not.
! empty speech
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
think so, and so, so she's gonna get this one for sure it's gonna fall down
have to run around, look it over, trebbin and all there or whatever, she's gonna get that one and, and there, he's gonna get
one himself or more, it all depends with this when they fall down...and
that sort of stuff... 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
! 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 ! Broca s and Wernicke s areas are on the left hemisphere
Some patients have difficulty with only certain However, the right hemisphere can also work with
types of nouns language
! left handed people
!concrete vs abstract (chair vs trust)
! hemispherectomies (age matters!)
!nonliving vs living (table vs dog)
!animals and vegetables vs food and body parts
!colors
!proper names
Brain and evolution Brain and evolution

Cantalupo & Hopkins (2001)
We ve argued that language is an evolved instinct ! Broadmann s area 44 (part of Broca s area in humans)
! differences in brains account for differences in abilities ! 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
! 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 is
(ASL) easy to attribute language ability where it does
! after failure to teach spoken language not really exist (9 month old children)
! 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
! and chimps created new compound words Researchers were quick to excuse mistakes
swan -> water bird
stale Danish -> cookie rock
as play , jokes , puns , metaphors ,...
See video:
http://www.npr.org/templates/story/story.php?
storyId=90516132
Word counts Nim Chimpsky

For example, a deaf student on one research team
A relative of other signing chimps
later commented that she saw much fewer signs than
the non-deaf students !with more careful judging probably learned
! seems the researchers counted almost any hand movement approximately 25 words
as a sign
!moreover, the signs were variations of the natural
Like movements of chimps in the wild
! scratch --> scratch
! pointing --> you
The chimps did not learn ASL
! finger to mouth --> drink
! hugging --> hug
! reaching --> give
! 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 ! Nim eat Nim eat.
they communicate
! Drink eat me Nim.
!unable to understand complex signs but not with real
! Tickle me Nim play.
language
Seemingly able to understand complex ! Me eat me eat.
sentences ! Me banana you banana me you give.
! Banana me me me eat.
!Would you please carry the cooler to Penny?
! Give orange me give eat orange me eat orange give me eat
But really, the chimp need only understand two orange give me you.
words: cooler and Penney

!the rest can be guessed!

Evolution Evolution
Note, it would have been interesting if chimps Chimps are the closest evolutionary relatives of
could learn language humans
!and not inconsistent with the idea that we have a !so if any non-human animal could learn language it
language instinct
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
!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?

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
! pineal gland link between body and spirit
A lot of work (PET scans, fMRI,) looks
! how they could connect was a real problem
for the site of consciousness
!a special physical transformation
!thalamus
!reticular formation
Mind-body problem
!quantum mechanics
Materialism (the brain is the mind)
! or the mind derives from the brain !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
!different brain areas know different things at
After consciousness
consciousness Consciousness different times

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)
! treaty signed in London months before the Battle of New ! different parts know different things at different times
Orleans ! there is no official moment of knowledge
! word was not received by British troops in America until two ! no official moment of consciousness!
weeks after the Battle of New Orleans (January 8, 1815)
Demonstration
! 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
! They behave in a way
!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
! if a conversation with a
!can consciousness arise from non-conscious computer is
processors? (artificial intelligence?) indistinguishable from a
conversation with a
human
! 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
! 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
! children
! mentally impaired people
! mute people
! people who speak a language we do not understand
Passing a Turing test is not necessary for

consciousness

Turing test Doubters

The Turing test is only one way to demonstrate
Many people have suggested that computers

intelligence cannot, in principle, become intelligent

! and a rather strict one at that
! they argue that purely symbolic computations cannot
! not passing the Turing test does not mean that a lead to consciousness
computer is not intelligent
! and humans use emotion, insight, intuition,
! 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
! discriminate conversation between a child and a
computer
! look at a conversation and decide which was the
computer
! ... 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
! Slot 1: someone sends you notes with Chinese characters on them in Chinese
! Book (written in English): in the book you can look up the Chinese
! then you can answer written questions in Chinese
characters and write down corresponding Chinese characters on
another piece of paper ! even though you do not know Chinese!
! 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
! 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
! You have to decide if the person understands what you are saying
(it s the Turing test) ! We know the person in the room does not understand
Chinese
! You ask them to describe the wall of their room
! They report it is green ! We might be fooled into thinking they do based on their
responses to the questions
! and so on
! Thus, the Turing test is a bad test
Because the Turing test is essentially the same

Slot 2 Slot 1
structure
! The computer plays the role of the person in the room

However,... And moreover...

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

consciousness consciousness
! Or maybe one needs to conclude that such an advanced
book has potential consciousness
2. Qualia 2. Qualia
Some researchers object to the very idea that
Consider two people who see the world in
computers could become conscious color opposites
! They argue that some things in consciousness are not
just computation Qualia for person 1 A red apple with a
! e.g., consider the color red green leaf
! 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
! 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,
! But suppose you never see any red objects
!And there seems no way to distinguish one
! Your knowledge will not tell you what you will
experience from the other
experience when you first see the red of an apple
!It s the unmeasureable experience that is a ! Indeed, you could be tricked into believing a green
qualia apple was red (if you had never seen green either)

2. Qualia Conclusions
But this is a defeatist argument, or a pointless one Consciousness
! if I knew everything about light, photoreceptors, and
neural representation of colors, then I would be able to
distributed processing in the brain
know what I will experience when I see red !no site of consciousness
! it is difficult (maybe impossible for any single human) to
!no time of consciousness
know (or even imagine knowing) all that information in an
academic sense
Chinese room
! but that doesn t mean that such information does not
exist Qualia
It s partly an empirical question
Artificial Intelligence
! 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
! Decision making
! Framing effects
! Risks
! Alternatives
What every consumer should know before they

buy.
Purdue University

Decision making Decision making
PSY 200 We have to make lots of choices

!course selections
Greg Francis !elections
Lecture 33 !housing
!job
!cancer treatment
What every consumer should know
What affects our choices?
before buying.
How do we make choices?
Topics Framing effects

Influences on decision making
Your decisions are influenced by the way a
!framing effects
set of choices is presented
!risks
The child custody problem
!alternatives
!two versions, essentially the same
!loss aversion
!lead to different choices
Effects are often related
Version 1: award frame Version 2: deny frame

Imagine that you serve on the jury of an only-child sole- Imagine that you serve on the jury of an only-child sole-
custody case following a relatively messy divorce. The facts custody case following a relatively messy divorce. The facts
are complicated by ambiguous economic, social, and are complicated by ambiguous economic, social, and
emotional considerations, and you decide to base your emotional considerations, and you decide to base your
decision entirely on the following few observations. To which decision entirely on the following few observations. To which
parent would you award sole custody of the child? parent would you deny sole custody of the child?
Parent A: average income, average health, average working Parent A: average income, average health, average working
hours, reasonable rapport with child, relatively stable social hours, reasonable rapport with child, relatively stable social
life life
Parent B: above-average income, very close relationship with Parent B: above-average income, very close relationship with
child, extremely active social life, lots of work-related travel, child, extremely active social life, lots of work-related travel,
minor health problems minor health problems
64% 55%

Framing effects Framing effects
So Parent B is the choice to award custody Your decisions are influenced by the way a
and to deny custody set of choices is presented
!but one necessarily precludes the other! The Asian disease problem
Subjects are biased by the task at hand !two versions, essentially the same
!focus on different characteristics depending on !lead to different choices
whether they are considering awarding or
denying
Version 1: Saving frame Version 2: Dying frame

Imagine that the US is preparing for the outbreak of Imagine that the US is preparing for the outbreak of
an unusual Asian disease, which is expected to kill an unusual Asian disease, which is expected to kill
600 people. Two alternative programs to combat 600 people. Two alternative programs to combat
the disease have been proposed. Assume that the the disease have been proposed. Assume that the
exact scientific estimates of the consequences of exact scientific estimates of the consequences of
the programs are as follows: the programs are as follows:
If program A is adopted, 200 people will be saved. 72% If program A is adopted, 400 people will die.
If program B is adopted, there is a 1/3 probability If program B is adopted, there is a 1/3 probability
that 600 people will be saved and a 2/3 probability that nobody will die and a 2/3 probability that 600
that no people will be saved. people will die. 78%
Choices and framing Risks

The two problems are essentially Risk corresponds to those events that
identical, except that the choices are occur with probability
!will I like the next movie starring Matt Damon?
phrased differently
!will I live to be 50?
!200 people saved = 400 people dead
!will the dice show double sixes?
!2/3 probability that no one is saved = 2/3
probability that 600 will die Events that occur with certainty are without
risk
But the phrasing makes a difference in
!the sun will rise tomorrow
the choices of subjects
!I will be older tomorrow
!why?
!I will give you an A if your grade is 90 or above

Risk Risk: monetary choices

Humans sometimes prefer risky options
Assume yourself richer by $300 than you are
over non-risky options
today. You have to choose between
!and vice-versa 72%
A) a sure gain of $100.
When the choices are perceived as losses
B) 50% chance to gain $200 and 50% chance
!subjects tend to be risk-seeking
to gain nothing.
When the choices are perceived as gains
!subjects tend to be risk-averse
Subjects tend to prefer the sure gain
Decision making is open to manipulation
!risk averse with perceived gains
!subjects can contradict themselves
Risk: monetary choices Notice

Selecting A) in either situation means you end
Assume yourself richer by $500 than you are up with $400
today. You have to choose between !$300 + $100
A) a sure loss of $100. !$500 - $100
B) 50% chance to lose nothing and 50% Selecting B) in either situation means you end
chance to lose $200. 64% up with either $500 or $300
!$300 + $200 or $300 + $0
!$500 - $0 or $500 - $200
Subjects tend to prefer the risky option
!risk seeking with perceived losses People do not just look at the bottom line
!which is why businesses emphasize that approach
Alternatives: version 1 Alternatives: version 2

Imagine you are shopping for a new car and have Imagine you are shopping for a new car and have
narrowed down your choices to three models. narrowed down your choices to three models.
According to a consumer magazine, the cars ride According to a consumer magazine, the cars ride
quality (RQ) and gas mileage (GM) are rated as quality (RQ) and gas mileage (GM) are rated as
Model RQ GM Model RQ GM
Asteroid 100 27 69% Asteroid 100 27 19%
Bravo 80 33 29% Bravo 80 33 79%
Comet 100 21 2% Clarion 60 33 2%
Which car do you select? Which car do you select?

Notice Consumer beware

Subjects hardly ever select the Comet or the Stores are very aware of this type of
Clarion behavior
!you might think they do not enter the decision Thus, they often stock merchandise for the
making process at all!
sole purpose of influencing your purchasing
!but they do
behavior
The comparison of Asteroid and Comet clearly !usually towards a more expensive model
favors the Asteroid
Likewise companies make low-end models
!it is less clear for the Bravo and Comet
simply to bias you toward higher end
!it is the reverse for Clarion
models and against the competition
Loss aversion Loss aversion

Another general property of decision making is that
people tend to be more sensitive to losses than to The same phenomenon ruins many marriages/
gains relationships
! thus people rarely take an even-bet
! the loss of $10 is more significant than the gain of $10 When your partner does something for you (a gain) it
This is also why fans of sports teams think the doesn t count as much as when your partner does
referees treat their team unfairly something against you (a loss)
In a game of basketball, each team will ! Thus, you perceive your relationship as overall not being
worth the trouble (even if your partner is good as often as
! have fouls called on them when they shouldn t have (a loss)
bad)
! Commit fouls that are not called (a gain)
! That s why therapists suggest that in successful
But the gains don t count as much as the losses relationships people must learn to forgive
! So in a truly fair game both teams (and their fans) feel as if
they were treated unfairly
! of course one team wins, so it feels that it overcame the
injustice
Loss aversion Choosing a job: 1

You have decided to leave your current job. It is
By definition a choice is a loss or a gain located so far away from your apartment that it
depending on where you start requires an 80-minute commute each way. But you do
like the fact that your job involves much pleasant
As a result, loss aversion dramatically affects
social interaction with your coworkers. Your search for
many types of choices by magnifying those a new job has given you two options and now you
characteristics of a choice that leads to a must choose between them. Which job would you
perceived loss prefer?
Consider choosing a job Job A: Limited contact with others, commuting time 20
minutes.
Job B: Moderately sociable, commuting time 60 67%
Purdue University
minutes. Purdue University

Choosing a job: 1 Loss aversion

You have decided to leave your current job. The job
In each case the subjects tend to choose the option
involves only a ten-minute commute, which you rather that produces the least loss
like. But your job leaves you isolated from coworkers ! keep sociable coworkers in version 1
for long periods of time. Your search for a new job has ! minimizing commuting time in version 2
given you two options and now you must choose Note, this means subjects are not just choosing what
between them. Which job would you prefer? they perceive to be the best job overall (again, not
Job A: Limited contact with others, commuting time 20 looking at the bottom line)
minutes. 70% ! but are instead choosing the best job relative to the current
situation!
Job B: Moderately sociable, commuting time 60
! a very strange phenomenon!
minutes.

Influences on decision making Problem solving
Framing effects Expertise
Risk aversion (perceived gains) Analogy
Risk seeking (perceived losses) Set effects
Loss aversion !functional fixedness
Insight
What does that aha! feeling mean?

Problem solving Problem solving
PSY 200 A hallmark of intelligence

!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
!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?
Topics Experts
We will look at a number of factors
Some people learn how to solve particular
that influence our ability to solve types of problems
problems
What makes an expert different from a
!expertise
novice?
!analogy
Experts know how to describe problems
!set effects
priming !other than that, there seems to be no
fundamental difference (even for geniuses!)
incubation
functional fixedness
!insight
An example of experts An example of experts

Chi, Feltovich & Glaser (1981) Have experts (PhDs) classify the same
Take second year physics students (novices) and ask them to problems
classify a bunch of physics problems
! they tend to group them by surface similarities !grouped according to how to solve
Expert 2: Conservation of energy
Expert 3: Work-energy theorem. They are all

straight-forward problems. Expert 2: These can be solved by Newton s second
Novice 1: These deal with blocks on an incline
law
plane Expert 4: These can be done from energy
Novice 2: Angular velocity; momentum, circular things
considerations. Either you should know the principle Expert 3: F=ma; Newton s second law.
Novice 5: Inclined plane problems, coefficient of of conservation of energy, or work is lost
Novice 3: Rotational kinematics, angular speeds, angular friction somewhere. Expert 4: Largely use F=ma; Newton s second law.
velocities
Novice 6: Blocks on inclined planes with
Novice 6: Problems that have something rotating, angular angles
speed

Chess experts Chess experts

Chase & Simon (1973) game Chase & Simon (1973) random
Show subjects a chess board and Show subjects a chess
then clear it board and then clear it
! have subjects recall positions of ! have subjects recall
the chess pieces on the board positions of the chess
pieces on the board
Master players are better than
beginners when the pieces are Master players are worse
30 25
positioned as in the middle of a than beginners when the
25 pieces are positioned
Correct pieces
Correct pieces
real game 20
20 randomly
Experts have schemas that allow Master
15
Master
them to organize the piece
15
Beginner Experts try to use the Beginner
10
positions 10 schemas, but they end up
5
! They only need to remember the 5 misremembering the actual
schema 0 piece positions 0
1 2 3 4 5 6 1 2 3 4 5 6
Trial Trial
Expert schemas Analogy: Attack-Dispersion Story

The many roads to a dictator s fortress are
In general, experts have lots of problem
mined so that small groups of men may
solving schemas specific to their domain of
pass, but a large group will be destroyed. A
expertise
general knows that his army can defeat the
!given such and such; do such and such
fortress if he can get his entire army to
!allows them to organize information in a way that
allows for easy recall and easy use
attack at once, but he cannot take his army
down a single road all at once without losing
Expertise in one domain does not transfer to
too many men.
another
What should he do?
!except for especially useful skills
Solution Analogy: Parade-dispersion

The general should split up his army into A dictator wants to show off his army so he
many smaller units tells a general to plan a parade of the army. He
!each unit takes a separate road to converge on demands that the general insure that the army
the fortress simultaneously and defeat the is seen and heard across the entire kingdom
dictator
simultaneously. The dictator also demands that
the parade be the most impressive ever at the
fortress. Splitting up the army would allow it to
fortress
be seen everywhere, but would make the
display at the fortress unimpressive.
What should the general do?

Analogy Analogy
The solution is the same Analogies are actually very difficult to apply
!split up the army and have them converge to !need to identify what is common between two
the fortress from different parts of the kingdom problems
Glick & Holyoak (1980) Analogies are often applied after two
!subjects read stories like these and were problems are solved and well understood
asked to solve the problems
!it is then easier to see what is common
!even when shown one solution and told that it
could be applied by analogy to another One of the problems handed out can be
!subjects used analogies only 20% of the time solved by analogy to these two problems
Set effects Past experience

Buddhist monk problem
Set refers to mind set (or something like

! One morning, exactly at sunrise, a Buddhist monk began to
that) climb a tall mountain. The narrow path, no more than a foot
!negative set: bias toward solving a problem makes or two wide, spiraled around the mountain to a glittering
temple at the summit. The monk ascended the path with an
it more difficult
average speed of 3 mph. He reached the temple shortly
!positive set: bias toward solving a problem makes before sunset. After several days of fasting and meditation
it easier he began his journey back along the same path, starting at
sunrise and walking an average speed of 5 mph. Is there a
You can be biased by lots of things spot along the path that the monk occupied on both trips at
precisely the same time of day?
!problem statement
!previous methods of reaching solution
!general knowledge Purdue University Purdue University
Self-imposed limits Set effects: Representation

Nine dots problem
!Draw four straight lines, passing through all nine Mutilated chessboard
of these dots, without lifting your pencil from the !62 squares remain
page. !can 31 dominoes cover them?
Easy solution!

Set effects: Priming Set effects: Incubation

Safren (1962) Cheap necklace problem
Opening a link costs $2
Unscramble the following anagrams
Closing a link costs $3

12.2 seconds per word
Go from given to goal state for no more than $15
KOBO CTURK STTE
CHUOC ANCYD DRINEF

7.4 seconds per word
KMLI GRAUS RECMA
FOEFCE TEESW IKRDN

Set effects: Incubation Set effects: Incubation

Silveira (1971) Subjects often get stuck using an approach
that goes nowhere (set)
Control: Work on problem for half an hour
After a break they are more likely to try a
!55% solve problem different approach
Exp A: Work on problem for half an hour in Brainstorming tries to avoid set effects by
15 minute sections, half-hour distracter task allowing free dreaming of solutions
!most of the ideas are worthless, but the approach
!64% solve problem is still beneficial
Exp B: Work on problem for half an hour in
15 minute sections, 4 hour distracter task
!85% solve problem
Set effects: Functional fixedness Insight

Intuitively, we sometimes feel as if we have a
How can you tie two strings together?
strong insight into a problem and its solution
Dunker (1945)
becomes obvious
! MacGyver
!the aha feeling
!is it real?
!what does it correspond to?
Using problems like the Bronze coin and

the Tree planting problems (Metcalf, 1986)
!subjects judge their progress with a warmth
rating, every 10 seconds, over 5 minutes

Insight Conclusions
Warmth stays mostly steady, right up to Effects on problem solving
proposing a solution Expertise
!the aha feeling Analogy
Set effects
Insight
Unfortunately,
the feeling does Solve remaining problems
not necessarily ! Tumor problem
indicate a
correct solution
Next time
Wrapping up the course
Other courses to take/avoid
Paths to pursue
Graduate school
Advice for further exploration
Purdue University

Studying cognitive psychology

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

! PSY 203: Introduction to research methods in psychology
Lecture 35 More statistics

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

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

PSY 390 Research in Computers
! Actively participate in a research laboratory ! Most experiments are run on computers
! Details vary dramatically across labs ! Models are simulated on computers
! Advisors can identify some positions ! Learn to program in a computer language
! Talk to faculty about possibilities MatLab, C / C++, Java, Basic
Research Focused Honors program ! Possible courses

! CS 15800 C Programming
! 3 semester sequence (starts Spring of penultimate
! CS 17700 Programming With Multimedia Objects
year)
! CS 18000 Problem Solving/Object-Oriented Programming
! Design and carry out your own research study (with ! CS 24000 Programming In C
guidance from a faculty member) ! CNIT 15500 Introduction to Object-Oriented Programming
! https://dev.www.purdue.edu/hhs/psy/undergraduate/ ! CNIT 17500 Visual Programming
opportunities/research_focused_honors.php
Further study
Useful background
Brain characteristics
Mathematics ! PSY 222: Introduction to behavioral neuroscience
! Many psychologists have little mathematical ! SLHS 304: Anatomy and physiology of the speech and hearing
background mechanism
! But it is especially useful for cognitive psychology ! PSY 320: Behavioral neuroscience of sensation & arousal
! Take as much mathematics as you can, especially ! PSY 322: Neuroscience of motivated behavior
Calculus (MA 161, 165 or 223) ! PSY 324: Introduction to cognitive neuroscience
Finite (discrete) mathematics (not easy to get at Purdue
West Lafayette) ! PSY 512: Neural systems
Linear (matrix) algebra (MA 262, 265)
Differential equations (MA 266)

Further study Further study

Perception Language (many courses in Speech, Language, and
! PSY 310: Sensory & perceptual processes Hearing Sciences - SLHS)
! Artificial intelligence in computer science or electrical
! SLHS 227: Elements of linguistics
engineering
! PSY 511: Psychophysics ! SLHS 309: Language development
! PSY 520: Attention & performance ! SLHS 401: Language & the brain
! PSY 426: Language development
Memory:
! PSY 526: Psycholinguistics
! PSY 311: Human learning & memory
! PSY 314: Introduction to learning Problem solving & decision making
! PSY 410: Animal memory & cognition ! PSY 285: Consumer behavior
! PSY 518: Memory & cognition ! PSY 318: Problem solving & decision making
! PSY 514: Introduction to mathematical psychology
Hot topic 1
Further study
Relating cognition to the brain (and vice-
Problem solving & decision making versa)
!PSY 285: Consumer behavior
Several big initiatives
!PSY 318: Problem solving & decision making
!
!PSY 514: Introduction to mathematical
https://www.humanbrainproject.eu
psychology
1.2 billion over 10 years
!PSY 390 with Pizlo, Proctor, Schweickert Develop technologies to bring together
disparate neurophysiological, anatomical,
molecular, and behavioral data
Database (big data)
Modeling (supercomputers, specialized
hardware)
Hot topic 2 Hot topic 3

Relating cognition to the brain (and vice- Big data
versa) Technology allows gathering of way more
Several big initiatives information than we know what to do with
! Human Connectome Project
http://www.humanconnectomeproject.org
Building a "network map" that will shed light
on the anatomical and functional connectivity
within the healthy human brain
Connections to cognitive psychology are

(hopefully) in the future

Greg Francis 8/3/15

Big data Big data
Technology allows gathering of way more Many experiments can be run on-line with
information than we know what to do with thousands of subjects

Big data Data analysis
Big data requires a different kind of statistics than has
Data can be gathered in novel ways
been used for simple experiments
Airport Scanner Moreover, there seem to be problems with present
1 billion trials of statistics even for simple experiments
visual search! 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
! And unbelievable findings meet the statistical criteria
! Pre-cognition
Graduate school Next time

Grades
Review for final exam
Financing
After graduation

AllLectures Psych PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

AllLectures Psych PDF

Caricato da

Copyright:

Formati disponibili

Introduction to

MWF 2:30 - 3:20 pm

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

Purdue University Purdue University

Lecture notes Lecture vodcast

!Adobe Acrobat (pdf) Four great mysteries.

Purdue University Purdue University

! Often takes a few days

! 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

! Sign it !updates to the syllabus

PSY 200: Intro. to Cognitive Psychology

Course outline Exams

Language -- Exam 4 (15%)

Purdue University Purdue University

See handout for instructions on getting started

Writing assignments Straight scale Grading

PSY 200: Intro. to Cognitive Psychology

Final class grade

Instructor office hours Teaching assistants

Purdue University Purdue University

Teaching assistant Teaching assistant

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology

Purdue University Purdue University

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

It is difficult to come up with an experiment

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology 3

Each has distinct properties

Purdue University Purdue University

Occipital lobe Parietal lobe

Purdue University Purdue University

Parietal lobe Temporal lobe

Purdue University Purdue University

PSY 200: Intro. to Cognitive Psychology 4