AllLectures PDF
AllLectures PDF
AllLectures PDF
Cognitive
Psychology
Greg Francis
Fall 2020
Online
http://www.psych.purdue.edu/~gfrancis/Classes/PSY200/indexF20.html
Prof. Greg Francis
Topics Textbook
● There is no textbook
● For example
● Lecture notes are used instead
w What s the deal with left and right brains?
w Why does everyone love Prozac? ● If you want a book, borrow from a past
class
w Why telephone operators seem rude.
w Why there is a gate at the first floor stairway ● There are optional readings in the
in the Psychology building. syllabus
w What to do if you are drunk while studying for w Not for every subject
an exam.
w What is the plural of walkman?
Purdue University Purdue University
w Reduced form (6 to a
●
3) Why is there so much
●
Four great mysteries
Humans face four great mysteries about the universe
4) What is the basis of human intelligence and
Topics alternative
consciousness? ● Discuss a sample of issues in cognitive
! Cognitive psychology and neuroscience psychology / cognitive neuroscience
●
press
Identify how the topics can help you to be
●
a better person
Exams CogLab
● Multiple choice (tentatively on ● Homework
BrightSpace) ● You participate in classic experiments
● Open notes, but not collaborative ● Total lab grade contributes to 15% of your class
grade.
● Detailed study guides are already on the
● Grade is based solely on completing the
class web site experiment, not on the quality of the data
● Practice exams will be posted on
Brightspace
w Available for a 24 hour period
w Don’t wait until the last minute
● registration codes! If the registration code has already been used, it will
not work for you. Each valid registration code can be used only once.
4. After filling in all the text fields, click or tap on the Start Registration button.
●
5. Your Web browser will connect with the CogLab server to verify your
w (a $50 value!)
w By 1:30 pm (not one second later!) on the date listed
● See handout for instructions on getting started (sent by on the class web page
email) ● 15% of your class grade
● Registration code is near the bottom of page 2
● First lab is due at 11:00 am on Wednesday!
Purdue University Purdue University
Grading
● Straight scale
Grading
w 98% -100% A+
w 93% - 97% A
● Previous semester’s grades (Spring 2019)
w 90% - 92% A-
w 88% - 89% B+ E1 E2 E3 E4 Final CogLab Writing Attendance Total
w 83% - 87% B Average 72 69 73 81 71 84 69 91 75
Max 96 96 98 99 99
w 80% - 82% B-
w 78% - 79% C+
w 73% - 77% C
● Last semester’s grades (Spring 2020)
w 70% - 72% C- E1 E2 E3 E4 Final CogLab Writing Attendance Total
No extra credit Average 77 71 72 83 72 83 64 92 75
w 68% - 69% D+
w 63% - 67% D Max 100 100 100 100 98
w 60% - 62% D-
w 0% - 59% F
● No rounding up: 82.99 is a B-
Purdue University Purdue University
C 59
Final class grade
9 0.00
8 0.00
7 0.00
D 29 w Email: gfrancis@purdue.edu
6 0.00 F 22
5 0.00
4 0.00
3 0.00
2 0.00
1 0.00
00
.0
00
.0 2 0.00 4 0.00 6 0.00 8 0.00 1 00 0
.0 1 20 0
.0
CogLab score
Purdue University Purdue University
Attitude/Advice
Teaching assistants ● Lectures: Watch them on the day of the regular schedule. Set aside a
particular time and place to watch them. Take notes.
● Print out the lectures and bring them to class. Take notes during class. Not
● Hailey Arreola and Andi Fordyce everything is on the slides.
● Everything we talk about in class is important
● Grade writing assignments ● Work on the study guide every week, so the ideas/answers from lecture are
fresh in your mind.
● Keep track of grades
● This class is an introductory class, but that does not mean it is easy
● Have office hours (via WebEx) w It’s like Introduction to Physics or Introduction to Chemistry
w Almost every other subtopic in psychology depends on the ideas in cognitive
w Hailey: Monday 11 am-noon and Wednesday psychology
w Everything is at least 10,000 times more complicated than what we discuss
10 am – noon (ET)
● If you don t find a topic interesting, just wait a week
w Andi: Tuesday 9 am – noon (ET)
Next time
● Cognitive neuroscience
● The brain
● The modularity hypothesis
Purdue University
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
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
http://www.youtube.com/watch?v=ZMLzP1VCANo
Purdue University Purdue University
CogLab CogLab
● In the CogLab + ● Your task was to +
experiment Brain judge whether the
asymmetry presented word was
● You stared at a “old” (seen on an
central fixation point + wood earlier trial) or “new” + wood
(not previously seen
● A word was
in this experiment)
presented to either
the left or right side
of fixation + Time + Time
CogLab CogLab
● Federmeier & Benjamin (2005) found better memory ● Other explanations than hemispheric
performance for words presented in the right visual field
specialization
● Words in the right visual field go to the left hemisphere
w Which is known to be specialized for language w Reading goes from left to right, from fixation to
right visual field
w Perceptual advantage to right visual field?
w Attentional advantage to right visual field?
● It is difficult to come up with an experiment
that isolates hemispheric specialization
areas seem to be
involved in different
kinds of cognitive tasks
Brain layers
Conclusions
● There is order and
function even within ● Lots of research in this area
an area ● New brain regions are being mapped out
● The cortex is a daily with ever increasing resolution
sheet of neurons
● Cognitive neuroscience relies strongly on
● In its thickness are
the modularity hypothesis
6 layers of neurons
w numbered 1-6 ● Putting everything together is very difficult
w sometimes include
subdivisions (4a, 4b,
4ca, 4cb, …)
Next time
● Brain scans
● EEG recordings
● MRI scans
● Functional MRI
Purdue University
Resolution
Electroencephalogram
● For almost every technique we have to worry about its
ability to discriminate differences in
w Space: which ● EEG
place is active?
w Time: when ● The brain
does something
happen? produces
● Finer resolution electrical activity
is usually better
w But can be ● Put electrodes
difficult to deal
with so much
on the head
data
Brain maps
EEG ● You can analyze the EEG signals in many different ways
● Watch the electrical current change through time while ● Compare the signal strength for different situations
reading sentences (averaged across many trials) ● Ayahuasca is a Brazilian psychoactive tea
w Good temporal resolution Semantic
w Kutas & Hillyard (1980) anomaly
w Seconds w Digestion
w Thinking about exams
w (Silva, 2002)
w …..
Functional MRI
Functional MRI
Connectome Connectome
● You can use similar technology (diffusion ● Gives an
spectral imaging) to focus on particular types of anatomical
cellular material map of how
w E.g., identify axons (discussed later) that connect information
brain cells can travel
● Gives an
anatomical
map of how
information
can travel
Purdue University Purdue University
fMRI fMRI
● Passive listening vs. active listening Passive
● The colors show the
w Vannest et al. (2009) listening
difference maps
● Twenty children (ages 11-13) complete three relative to listening to
the tones
tasks
● Common activity
w Passive listening: hear a female speaker tell a 30-
(breathing, digestion,
second story
hearing machine
w Active response: hear the same speaker tell a story in
noise,…) is
5 second segments of two sentences. Scanning Active
subtracted out
occurred after the sentences (silence). Answer response
questions ● The colors are not
brain activity!
w Random tones: no task, just listen
Purdue University Purdue University
fMRI fMRI
● More signals and Passive ● Does more signal for the Passive
listening listening
different patterns for active response mean active
active listening response listening is better
compared to passive than passive listening?
listening ● Tested children on
● (Could it be comprehension of stories
otherwise?) w PL: 75.1% correct, SD=12.7
w AR: 79.1% correct, SD=9.1
Active Active
response ● No real difference in response
comprehension
Walking Walking
● There is no portable
MRI machine
● But scientists are
creative about how to
use it to study a wide
variety of activities
● Volta et al. (2015)
studied walking by ● Can compare walking “indoors” versus “outdoors”
having participants ● Execution (actually “walk”) versus observation (not “walk”)
“walk” on a cylinder
● Complicated controls
outside the MRI
w Press feet against cylinder
machine
w Still (non-moving) image
w Gray image (no picture)
Scan
now
Or scan
now
Purdue University Purdue University
differences in the brain when we see different images ● In general, brain scans provide a very limited form of mind reading
w The percept is the brain’s behavior, so there must be w People do better than this every day by watching people behave
(posture, eyes, skin tone)
differences!
● These kinds of studies tell us that the
neurophysiological differences between cognitive
events can be measured by these brain scanning
technologies
w Failure would only indicate limits of the technology
Purdue University Purdue University
Statistics Statistics
● There is a significant difference in fMRI activity for some regions of ● The subject in this study was a mature Atlantic
the brain
salmon (sex unknown)
w Medial brain cavity and upper spinal column
● The active regions identified by the fMRI are due to
chance
● Even with purely random noise, there will be some
statistically significant findings
w The brain has lots of random noise
Neurons A neuron
● The brain cells that are responsible for cognition ● Dendrite ● Axon
w input w output
are neurons Soma
Myelin sheath
●
●
w integrate
w insulate
Myelin A neuron
● Diffusion Spectral Imaging detects properties of ● There are many different types of neurons
the myelin sheath (“white matter”) ● We will describe only the most common characteristics
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?
Purdue University
Greg Francis
Lecture 05
Input at dendrites
Input / output
Changes the cell membrane
-70 millivolts ●
A neuron A neuron
● An action potential ● The action potential then affects the membranes of
generated at the other cells dendrites
soma travels
down the axon to
the terminals
Output Output
● Myelin is like insulation for the ● The output of a neuron is either excitatory or inhibitory on the
cell s axon
other neuron it reaches
w it insures that the signal
generated by the action ● Excitatory: when our neuron sends an output, the receiving
potential is strong neuron is more likely to produce an action potential
w Jumps electrically rather than ● Inhibitory: when our neuron sends an output, the receiving
the normal chemical exchanges
neuron is less likely to produce an action potential
● In multiple sclerosis the body s
immune system attacks myelin
w physical problems (paralysis)
w cognitive problems (memory,
reasoning, judgement)
w cause unknown (300,000
people)
Networks Epilepsy
● Cognitive behavior is related to ● Disease of central nervous system
groups of neurons working w causes mostly unknown
together
● Seizures
● Include excitation and inhibition
w bursts of electrical activity travelling through networks in
w more later
the brain
w brain activity is out of control
w epileptic fits
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
Neural connections
Molecular structure
http://www.mind.ilstu.edu/curriculum/neurons_intro/neurotransmission_classic_3.0.swf
Purdue University Purdue University
Receptor Receptor
● Very large
molecules called ● When it accepts a neurotransmitter, it
proteins
starts a chain reaction of events
● Similar to a filter
w accepts some
w physical, chemical, electrical
neuro- w locally changes the cell membrane
transmitters
» depolarization (excitation)
w rejects others
» hyperpolarization (inhibition)
Parkinson s Drugs
● Lack of dopamine ● Interact with neurotransmitters in lots of ways, for
w Many different causes example
w Replace: accepted by receptor and with similar effect
w In extreme cases, patients are frozen
w Production: increase or decrease
● Give patients large doses of L-DOPA w Reuptake: knock out enzymes that remove
neurotransmitter from receptor, neurotransmitter has a
w a precursor of dopamine
bigger effect
w sometimes solves the problem w Blocking: enter receptor but does not trigger reaction,
partly closes receptor protein so neurotransmitter
w lots of side effects
cannot enter
● Awakenings, by Oliver Sacks
https://www.youtube.com/watch?v=koL0PWCJ4lo
Purdue University Purdue University
Receptors Receptors
A receptor has a simple receptive field
● Light sensors (rods and cones) respond to light ●
Light
Blind spot
Blind spot ● In CogLab you mapped your blind spot
● my
● Where nerves leave data
the back of the eye, looks
like
there are no light this
receptors
w light that hits this spot is
not visible
Light
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
Purdue University Purdue University
-+
(surround)
● On-center, off-surround means the cell
is sensitive to
the location of
a small spot of
light
Purdue University Purdue University
Information processing
Information processing
● For simple cells, an image like this
● For simple cells, an image like this
w is coded something like this
Strong
responses
at edges!
Next time
● Networks of neurons
● Connections between cells
● Feedback – resonance
Purdue University
Demonstration Feedback
● Feedback in networks can act to clean up
● Cell activities do settle down eventually
noisy sensory information to make it
● Final pattern of activities satisfies consistent with what our systems expect
constraints of the network connections ● In a very real way, what we see, hear, taste,
smell, touch, and think, is biased by our
● Error correction capabilities network s expectation
● Can tolerate the loss of some cells ● A network s expectation is established by its
connection weights
● Emergent properties of the network
w excitation -- inhibition
w no single cell has these properties
Seeing things that are not Seeing things that are not
there there
Feedback Conclusions
● Similar situation with the illusory ● Networks of neurons have properties
circle seen here different from single cells
w emergent properties
w stable activities
w multiple constraints
w tolerance to errors and cell loss
● Structure of connections (synapses)
determines the final pattern of responses
Next time
● How networks learn
● Changing connections
● Learning rules
● Self-organization
● CogLab due for Implicit Learning
Deep learning
Learning
● Google used one version of a neural network to analyze
10 million YouTube stills ● This may not be the same type of learning you do when you
study for school
w 1000 computers (16,000 cpu’s) over 3 days
w but it is important just the same
● The network self-organized to identify common patterns ● Consider implicit learning
w Cats, faces, “tool-like objects oriented at 30 degrees” w A long sequence of trials,
where you press a key to
indicate the appearance
of a dot at a corresponding
location
Cameras Problem
● Enhance visual perception
● The network coordinating
w MRI overlaid on actual image
eye-hand systems, adjusts
of brain for surgeon
» highlight tumor itself
w Avoid other brain regions ● Extended use of the
» faster computer cameras makes
the user adapt so his eyes
are where the cameras are!
w motion
w texture
Why you see color
w depth...
afterimages.
Purdue University Purdue University
On-center, off-
On-center, off-surround
surround
● Characteristics of cell receptive ● Characteristics of cell receptive
fields force additional properties fields force additional properties
of the visual system of the visual system
w center-surround cells tend to not w center-surround cells tend to not
respond well to homogeneous respond well to homogeneous
light that covers both excitatory light that covers both excitatory
and inhibitory parts and inhibitory parts
w => percepts of the middle of an w => percepts of the middle of an
object is derived from the edges object is derived from the edges
Brightness contrast
Brightness contrast
● Two receptive fields inside the middle square
receive the same excitatory and inhibitory signals
● Edge responses are influenced by the surrounding
w Little response
light
w both center squares have the same light intensity
Filling-in Filling-in
● We see color and brightness inside objects ● Brightness
w so edge information must fill-in to the interior information
● It sometimes gets things messed up spreads across
● Water color effect surfaces
w Craik-O Brien-
Cornsweet
effect
Oriented afterimages
Orientation competition
● As the horizontal gate ● Oriented
recovers, the system reset signals
returns to baseline and are also
the vertical after response implicated in
disappears an unusual
type of
afterimage
CFF Phosphor
● Establishes minimum characteristics of ● The phosphor on a computer screen
electronic devices typically glows less than 10
milliseconds
● Lights flicker at 120 Hz
w ten thousandths of a second
w we spend a lot of time in darkness
● The gun reactivates the phosphor
● Computer (Cathode Ray Tube, CRT)
every 17 milliseconds
monitors and TV s flicker at around 60
● Thus, at any given time 1/3 of the
Hz
screen is dark
w better monitors go faster
w the percept persists in your head!
w Liquid Crystal Display (LCD) monitors work
differently
» Although some still flicker
Purdue University Purdue University
Persistence Experiment
● What is the source of the persisting percept? ● Bowen, Pola & Matin (1973)
w Receptors in the eye? w subjects adjust duration of a blank
w Receptive fields? stimulus so onset of probe matched
perceived offset of the target
w Network interactions?
w Cognitive (memory)?
w its persistence
response
decreases ● Demonstration
Explanation Explanation
● Francis, Grossberg & Mingolla (1994) ● Offset of input from
the eyes produces an
● Something has to reset the network
after response
w else it would keep persisting forever
w e.g., due to
● Two mechanisms competition from
w (1) new inputs inhibit old responses orthogonally tuned
cells
w (2) afterimages act as new inputs
Offset response
Note: afterimages get stronger as duration
●
●
inhibits persisting
and luminance increase!
response Input from eyes
Purdue University Purdue University
No mask No mask
● Write down all the letters you see ● Write down all the letters you see
R F
L P
M Q
Purdue University Purdue University
● Write down all the letters you see ● Write down all the letters you see
● Write down all the letters you see ● Write down all the letters you see
Y S X X X X
D H X X X X
W F X X X X
Purdue University Purdue University
Metacontrast
Significance
● Masks do not have to “write over” the target to have an
● The mask appeared after the target turned off
effect
● The target was presented all by itself for a
● In metacontrast masking the mask and target do not
brief period of time
overlap in space and (often) in time (CogLab)
● However, our visual system is unable to
develop a complete percept of a scene in a Target Mask
such a period of time
w Thus, the XXX mask interferes with processing of
the letters by shortening their persisting responses
w And prevents perceived blurring of changing scenes
Metacontrast Motion
● Correct identification of the narrow target is affected by ● In simple animals (like flies and frogs), we know how
the Stimulus Onset Asynchrony (SOA) of the target and motion is detected
mask w Demo on web page
● Reichardt detector
w Worse between 60-90 milliseconds
Time2
Attention Attention
Attention
Information processing ● Part of attention seems to be due to mental effort on
● Modern theories see cognition as information processing
your part
w much like a computer w attending a lecture
● Different systems have different capabilities, capacities, and speeds w ignoring whispering around you
● Necessarily, some information is ignored because it is not ● Part of attention seems a natural side effect of mental
processed effort
w ignoring the uhs and ums from a speaker
w ignoring the feel of clothes on your body
CogLab global data ● CogLab has several labs that play on similar ideas
(~13,250 observers)
w Stroop effect (more next time)
w Spatial cueing
w Several labs related to memory and decision making have
similar properties
Attention
Drawing attention
● Masking the changes makes it difficult to identify
● Raise your hand when you spot what
the changed parts of the image
changes in the two images
w Suggests that you do not actually “see” the entire
image with each presentation
● Attention seems to be necessary to detect
stimulus changes
● Explains how people can look but not see
w walking into doors
w driving into trains
w detecting changes on a radar screen
Tse, 2005
Purdue University Purdue University
Next time
● Methods of studying attention
● What things influence attention
w Timing, features
● CogLabs on Attentional blink and Visual
search due!
● Should you pay $59.95 for Mega-speed
reading?
Purdue University
Attention Attention
We saw last time that attention can have
PSY 200 ●
Should you pay $59.95 for w and look at experimental methods that are
used to study attention
Mega-speed reading?
Purdue University Purdue University
Attentional blink
Attentional blink
● Measure frequency of detection
● Turns out that detection of first letter tends to make w class data (182 observers)
detection of the second letter very difficult
w if it immediately follows the first
w Attentional blink
Implies that
M detecting the
P first letter
K causes
R you to miss the
J second letter!
S
● This often happens when the target has a unique feature ● This often happens when the target has a unique feature
relative to the distracters relative to the distracters
w shape w color
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
Requires search
green arcs green arcs by comparison
across feature
maps.
Serial process
No searching that takes time
required!
Automaticity Automaticity
● The process whereby a task goes from
● When a task is unfamiliar it
requiring a lot of attention to requiring little is
seems to require a lot of
attention to perform called automatization
● Later it requires less attention ● Many tasks are automatizable
w riding a bike
w color naming
w driving a car
w typing w word naming
w tying shoelaces
» http://www.fieggen.com/shoelace/knots.htm
● Can measure effects by pitting an
automatized task against a non-automatized
task
Purdue University Purdue University
Stroop task
● Stroop (1935)
Stroop effect
● Identify the color of ink for words
● It takes longer when the words are color
names
● Demonstration
w measure reaction time
Stroop effect
Stroop effect ● Word name interferes with ink color
naming
w ink color does not generally interfere with
word naming
w lots of studies on Stroop effect
Explanation Conclusions
● Word reading is well practiced ● Methods of studying attention
w especially among college undergraduates w attentional blink
Next time
● Intersection of attention, perception, and memory
w Iconic memory
w echoic memory
Purdue University
● Write down as many letters as you see ● Write down as many letters as you see
B X P Q
M H E L
V N W A
Purdue University Purdue University
● How can we tell if percept is lost? w ==> entire field was available
● Write down letters from the indicated row ● Write down letters from the indicated row
S T Y I
R F C Q
Z E V N
Purdue University Purdue University
w delay w delay
R Q V L
S D G F
P N B A
Purdue University Purdue University
4
Number of letters correct
0
0 0.2 0.4 0.6 1
Results Results
● We can test on any row and get essentially the same
● CogLab data (148 participants) result
w 3 letters in each row w so, the number of letters that actually persist and is available is
found by multiplying by the number of rows (3)
4 .5
Number of letters available
Number of letters correct
4
Original published data CogLab data
3 .5
10 4 .5
3
2 .5 8 4
3 .5
6
2 3
1 .5 2 .5
1 4 2
1 .5
0 .5
2 1
0
0 0 .2 0 .4 0 .6 0 .8 1 1 .2 0 .5
0 0
0 0 .5 1 1 .5
● Write down letters from the indicated row ● Write down letters from the indicated row
X V F R
W K D M
S N J Y
Purdue University Purdue University
X X X X
Purdue University Purdue University
Number of letters
duration
4
F (seconds)
3
H
w Smaller
available
V 2
A capacity 1
D
S E ● Significant for 0
W X 0 1 2 3 4
some memory
tasks Delay (seconds)
Purdue University Purdue University
0.8
w 1) no delay (immediate)
0.6 Primacy Recency
w 2) in the correct order (serial)
0.4
w 3) no cues (recall, not recognition)
0.2
● Plot percentage correctly recalled against 0
position of item in list 1 2 3 4 5 6 7
Position in list
Purdue University Purdue University
4 0.6 Tone
1
0.4 Word
6
8 0.2
7
0
9 1 2 3 4 5 6 7
Position in list
Purdue University Purdue University
● Calculate savings 1
0.8
Savings
0.6
Timeoriginal 0
0 5 10 15 20 25 30
Time (days)
Purdue University Purdue University
WRM
Retention
Retention
● Vary duration of counting backward
● Peterson & Peterson (1959) ● Numbers are different from letters, you
w Brown (1958) might not expect any interference
● Give subjects trigram w but they can have very strong interference
1
Proportion correct
w ask them to count backwards by 3 s and 0.8
then recall trigram 0.6
779, 776, 773,... Suggests
0.4
some memories
0.2
WRM 782 last only a
few seconds! 0
0 3 6 9 12 15 18
Retention interval (seconds)
Purdue University Purdue University
Retention Retention
● The results of the Brown-Peterson study ● The results of the Brown-Peterson study also
suggest that some aspects of forgetting are
suggest that some aspects of forgetting
passive
are process driven w even if you are distracted, you can recall the trigram if
w keeping a memory active requires effort only a short time has passed
w if many seconds have passed, while you are
w if you are distracted by another task, you distracted, you cannot recall the trigram
cannot apply the effort to keep the memory w memory has decayed , or something like decay, while
w similar to our observations about attention you were doing the distracting task
and processing
w Miller (1956) 9 1 0 3 5 7 3 2
3 9 0 5 7 4 2 16
Purdue University Purdue University
Pizza
Interpretation ● The Little Caesar s in W. Lafayette used to
have a game where you could win a pizza
● There exist two types of memory systems
w must repeat a sequence of flashing lights
● Long Term Memory (LTM) (changes every time)
w high capacity (no limit) w The sequence gets longer until you make a
mistake
w long duration (forever)
w need a sequence length >7 to win much
w Ebbinghaus experiment » Counts number of correct button presses
» 56 (sequence of 11 buttons): win a soft drink
● Short Term Memory (STM)
» 110 (sequence of 15 buttons): win crazy bread
w small capacity (~7 items) » 210 (sequence of 20 buttons): win pizza
w short duration (seconds) » nearly impossible with STM properties
w http://www.freegames.ws/games/kidsgames/simon/simon.htm
0.4 ● Demo
0.2
0
1 2 3 4 5 6 7
Position in list
Purdue University Purdue University
● Here s the
0.8 PRIMACY:
Use STM
CogLab Global
data 0.6 Use LTM
w (26,557 subjects)
0.4
● Demo
0.2
NEITHER LTM NOR STM
0
1 2 3 4 5 6 7
Position in list
Purdue University Purdue University
Lecture 15
w Sternberg (1969) NO NO
5329
5329 5329
5329 8
8888 8888
8 8
8
Purdue University Purdue University
3333 3333
3 3
3
Set size
Purdue University Purdue University
8 8 8 88 3 3 3
8 8 3 3
8 3
Purdue University Purdue University
3
Set size
Purdue University Purdue University
comes out
w You ran a version of the experiment in CogLab
Interpretation Controller
● Exhaustive search makes sense if search of STM
● Controlling attentional system
is done by some process that is
w supervises
w very efficient (can search very quickly)
w coordinates
w dumb (doesn t bother to stop itself) w starts and stops relatively independent processes
w initiated by some other system (a controller)
● e.g.
w Search short term memory
Controller w Search long term memory
w walking down stairs
Search w gate in psychological sciences building
memory
w Doors
STM process
F
● Part 1: spatial mental task
● Brooks (1968)
(diagrams)
w two types of tasks (visuo-spatial and phonological)
w visual imagery
w two types of responses (visuo-spatial and
w classify corners (top or bottom
phonological)
corner?)
● Identifies two types of systems that are w yes if top or bottom
relatively separate w no if not top or bottom
Results
Results ● Results
w when you have to respond by pointing, it is easier to
● Measure time to finish mental task for each
work with sentence information than diagram information
response type
w when you have to respond verbally, it is easier to work
w diagrams -- pointing
with diagram information than sentence information
w sentence -- pointing
Mental task
w diagrams -- verbal
Diagrams Sentences
Response task
w sentence -- verbal
Pointing 28.2s 9.8s
Significance Interference
● The results suggest that there are two
● These system have only limited resources and capabilities
relatively separate systems ● Asking a system to do two things at once (e.g., pointing and
w one deals with visuo-spatial information and mental diagram) slows down the system
must do the pointing response and mental ● Splitting responsibilities across the systems (e.g., spoken
diagram task response and mental diagram) can be done quickly
● Baddley (1986) put these ideas together into a model of w rapid forgetting
working memory
Central ● Processor of information
executive w not a storage device
w hypothesizes mechanisms that lead to memory properties
Visuo- Phono-
spatial logical
sketchpad loop
Purdue University Purdue University
Greg Francis
executive
w small capacity
w rapid forgetting
Lecture 16 Visuo-
spatial
Phonological
CogLab data
Language effects
● The CogLab experiment on memory span shows
data in agreement with our expectations (169
● Some
subjects)
languages are
spoken more
quickly than
others
● Should allow
larger memory
span
w it does
w investigated complaints about WISC ● Also repeats a phrase over and over
intelligence scores w e.g., tippy-toe, tippy-toe, tippy-toe,...
Phonological loop
Visual Auditory
ACP B F H N Better
recall
tippy-toe
B G P T Worse
PS recall
Purdue University Purdue University
Context Context
● The effect of part-set cueing suggests that to measure memory ● But memory is not exactly the same as visual
you must consider the conditions at test
search
w Memory is more often about discrimination of memory traces and
not about the strength of memory traces ● Information must be encoded in memory as well as
w Similar to visual search experiments recalled
w Such encoding can alter what features are stored as part of the memory
Feature w Which changes the discrimination of subsequent recall
search
● It turns out, that to maximize recallability
w the effort and conditions at the time of learning must be
consistent with the properties and conditions of the test
Conjunctive ● Encoding specificity principle
search
Purdue University Purdue University
7
Study
recognized
CHEESE CHEESE 6
5 Semantic
4 Rhyme
3
The man 2
threw the ball SNEEZE 1
to the 0
Standard Rhyme
______.
Recognition test
Purdue University Purdue University
Cue No cue
10
Study no cue
8
soar Study cue
6
nurse 4
auto 2
paper 0
... No cue Cue
Test display
Purdue University Purdue University
EAGLE
EAGLE
HEALTH
WHEEL HEALTH
BOOK WHEEL
... BOOK
...
Purdue University Purdue University
Significance Forgetting
Decompression tables for divers
Forgetting is not always a characteristic of a
●
●
w want to remember when under water
memory system, or your brain
w generally study while on land
w although it could be in some cases, it is not
● Researchers working under water have difficulty
always
recalling their details on land
w E.g., counts of species ● Forgetting must be defined operationally
● How do you know if something is forgotten? w specify the task and context of retrieval
w changing context may allow subject to recall seemingly w You can never be certain that if you are placed
forgotten information
in a different context you will still show forgetting
w forgetting = retrieval problem?
Testing Mood
● Mood has a similar effect (Eich et al, 1994)
w Mood induced by music and directed thoughts
● So, if you are intoxicated while studying for an
exam
w and you didn t study before
Study mood
● You should be intoxicated while taking the
exam
But do not
expect to do GET HELP!
very well
Purdue University Purdue University
Classrooms Classrooms
● Is memory better when you are tested in the same room as ● Subjects recall all words either in the context of Day 1 or Day 2
lectured? (different contexts for different subjects)
w significant for final exams! ● Recall was best for words that were studied in the test context
● Smith et al. (1978)
w Subject studied words in one of two contexts (on separate days)
w Varied classroom and dress of experimenter
Study context
● Mood
● Classrooms Purdue University Purdue University
w but this is impossible because it was never shown w able to be manipulated, to a certain extent, I can make
you have certain memories
● CogLab data (163 participants)
w Type of selected items Percentage of recalls ● Why does the false memory effect
w In original list 78.5 happen?
w Normal distractor (not in list) 7.9
w Special distractor (not in list) 78.5
Purdue University Purdue University
Discrimination Interference
● Good memory recall usually requires not only ● Retroactive interference (RI)
recall of an item from memory w new information prevents recall of
previous information
● You also must identify the correct item relative
w e.g., Overwriting a computer file.
to the appropriate context or time frame
w The current trial ● Proactive interference (PI)
w The context of the experiment w prior learning prohibits new learning
0.9
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
Purdue University Purdue University
● Partial report experiment (first 12 trials had the cue before ● Attentional Blink experiment (detection of the first letter in
the letter matrix – to give you practice) the stream)
0.9 0.5
Proportion of letters reported
0.8 0.45
0.4
0.7
0.35
0.6
0.3
0.5 0.25
0.4 0.2
0.3 0.15
0.1
0.2
0.05
0.1 0
0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960
1 2 3 4 5 6 7 8 9 10 11
Trial
Trial
Purdue University Purdue University
80
XJF Trial 1 XJF 60 Control
WRM Experiment
Trial 2 WRM 40
20
DBL Trial 3 DBL
0
NRX 1 2 3 4
Trial 4 942 Trial
Purdue University Purdue University
80
w go back to questions you cannot answer
w less proactive interference
60 Not told
Told w should recall more
40
Greg Francis
Lecture 19
Discrimination No forgetting?
● Brain surgeon (Penfield, 1959)
● The task is difficult because you have to do w Epilepsy patients
several things w stimulate brain regions before operating
w Recall information that might be related to the task from memory w want to know what is being removed
w Determine if the memory is actually for the correct event ● Conscious patients report vivid memories
w Determine if the memory is actually for the correct moment in w unable to recall normally
time » "she saw herself as she had been while giving birth to her
w Gauge your confidence in the memory s validity baby."
w stimulation of temporal lobes
● All of this suggests that performance on a memory task
involves discriminating information ● In the image, numbers indicate
places where stimulation
To address the discrimination problem, people engage
evoked different reported
●
w The results are usually misunderstood » people describe places they have never visited, impossible
events, fantasy,…
w Actually only occurred for 5% of patients w The patients have epilepsy
» Stimulation may have triggered something like an epileptic
seizure (which can have hallucinations)
● It is more likely that stimulation feels like a memory, even though
it is not
w your awareness of remembrance is a product of your brain
w it can be stimulated, even without a real memory
Eyewitness testimony
● I will show you a series of slides and then ask you
some questions
Eyewitness testimony
Questions
● Later, show slides and ask subjects if they were part of the
● Did the bus, which came by, come from the left or the original set
right?
● Key test is for a pair of slides (between subjects)
● Did another car pass the Red Datsun while it was at w real slide contains YIELD sign
the intersection with the stop sign? w fake slide contains STOP sign
● Did you see a bicycle?
● Did you see the taxi cab?
● Did you see if the policeman wrote anything down?
Percentage correct
●
60
Compare accuracy according to pre-test questions misinformation Immediate
● 50
effect gets
w Subjects without a misleading question--90% accurate 40
stronger with a
w Subjects with a misleading question -- 20% accurate 30
week delay
before the 20
w In a follow-up, the experimenters asked those with misleading
memory test 10
questions if they thought they were misled
0
» 90% say no Consistent None Misleading
Next time
● Amnesia
● Anterograde amnesia
● Retrograde amnesia
● Unusual characteristics
● Repression
● CogLab on Forgot it all along due.
Amnesia Amnesia
● Scope and duration
● Loss of memory or memory abilities
● Retrograde amnesia for one patient
w retrograde: forgetting events prior to the injury
5 months 8 months 16 months
w anterograde: forgetting events after the injury
w memories cannot be wiped clean w they do not generally forget how to walk, talk,
solve problems
w perhaps they are just not directly accessible
» Although they may have problems…
w forgetting = recall problem?
w Different types of memory systems
» controversial!
Anterograde amnesics
Patient HM ● Fairly normal STM digit span (~7 items)
● Surgery on hippocampus (to control epilepsy) ● But very difficult to extend digit span
● anterograde amnesia w how many trials to repeat back list correctly?
30
w unable to learn anything new
● Thought it was 1953 25
Mean trials to criterion
20
w shocked by age of face in his mirror
Controls
w Could not stand to read newspapers 15
Patients
w reintroduced himself to doctors, nurses,… 10
0
0 5 10 15 20 25
Number of digits to remember
70 hospital
60
Controls
cafeteria
50
Amnesics w E.g. mirror
40
30 drawing
20
task
10
0
0 5 10 15
Item number
Purdue University Purdue University
Repression Repression
● Psychotherapists (e.g. Freud) suggested that ● In a laboratory, showing evidence of repression
infantile amnesia occurred because much of requires
childhood is filled with painful events and memory
w being unable to remember something
of the pain is prevented by psychological defense
mechanisms (repression) w being able to recover the memory through therapy
w proving that the recovered memory is accurate
● This is very unlikely
w people do remember painful events well
w laboratory studies find no evidence of repressed memories
Discovered memories
Repression
● However, it is possible for information that was once known to be
forgotten and then (re)discovered
● In therapy, clinicians often claim evidence of
CogLab’s Forgot it all along experiment demonstrates this
repression with
●
property
w dream interpretation
● Phases I and II are like an encoding specificity experiment
w patterns in symptoms
Study with cue Test with same or different cue
w recovering a memory through hypnosis
cup-D_ _K
● None of these techniques demonstrate a verified cup-DESK
memory pan-D_ _K
● Among carefully controlled memory research, there is
no evidence of repression!
Purdue University Purdue University
● Encoding specificity
● Retrograde amnesia
● Levels of processing (CogLab due!)
● Anterograde amnesia
● Judgments of learning
● Learning in anterograde amnesics
● Practice testing
● Infantile amnesia
● Learning styles
● Repression
● How to improve your memory without spending
$20.
Purdue University Purdue University
Memory
Improving memory
● We seem to be unable to control our
PSY 200 memories
w learn things we don t want to remember
Study style
Level of processing
● Time spent studying is also “context” for memory
● Memory can be influenced by depth of processing at the time
retrieval
of study
● Generally, more study leads to better memory w Craik & Tulving (1975)
● Style of study matters too ● Subjects observe words with associated tasks
25
Number of words recalled
20
15
10
0
Capital Rhyme Synonym
Study task
Purdue University Purdue University
Levels of Processing
Intention
● Level of processing is more important than intent to
learn (Hyde & Jenkins, 1973) ● Recall (out of 24 words) varies a lot with task
● 11 groups of subjects ● Not much variation with intention to learn
w 1 control group: told they will be tested to recall the words
» not given any study task
w 10 experimental groups split to perform a study task
» Pleasant-unpleasant rating
» Estimate frequency of word usage
» E-G checking: does word contain an E or a G?
» Identify part of speech: noun, verb,…
» Sentence framing: which sentence does word best fit in?
w For all experimental groups, either
» (a) Intentional learning : told they will be tested to recall the
words
» (b) Incidental learning: not told they will be tested
Purdue University Purdue University
Memory
Improving memory 2
● We seem to be unable to control our
PSY 200 memories
w learn things we don t want to remember
● Associate items in list with a previously ● Hook to be remembered items to the list
memorized list w visual imagery helps again!
ITEMS Peg word
One is a bun. Six is a stick. recall by
Two is a shoe. Seven is a heaven. milk bun reciting poem
Three is a bee. Eight is a gate.
Four is a door. Nine is a line.
Five is a hive. bread shoe
Ten is a hen.
bananas tree
w pato -> Spanish for duck , sounds like pot-o ● Experimental group learned more words faster and
» imagine duck with pot on its head for longer
w zronok-> Russian for bell , sounds like zrahn-oak w 6 weeks later
» imagine an oak tree with bells as acorns » experimental (43% correct)
» control (28% correct)
Purdue University Purdue University
Brain Training
S.: Luria ● Several companies market
activities to make you smarter
● Visual imagery w “Exercise” your brain with games that are
adapted from neuroscience
● You can improve performance on specific tasks, but that does not typically w Ellenbogen et al. (2007)
transfer to other tasks
● You can make yourself smarter by learning new information
Purdue University Purdue University
Sleep Sleep
● Subjects learn to identify ● There is a ordered arrangement to the stimuli
order relationships ● If you know this arrangement, deciding for any pair is easy
between random shapes w But subjects are never explicitly told about this arrangement
● Only shown one pair at a
time
● Subjects have to
learn/memorize the
appropriate answer to
each pair
Sleep Sleep
● Subjects are split in to three ● Subjects are split in to three
groups, according to when groups, according to when
they are tested they are tested
w 20 minutes later w 20 minutes later
w 12 hours later w 12 hours later
w 24 hours later w 24 hours later
Sleep Conclusions
● Half of the 12 hour group
had sleep and half did ● Lots of ways to improve memory
not
● It makes a difference for
w Method of loci
pairs of items that are far w Imagery
apart in the ordered
structure w Mnemonics
● 1-degree: A>C, B>D,… w Brain training
2-degree: A>D, B>E,…
●
w Sleep
● Advice:
w study early!
w Get some sleep!
Next time
● Mental representation
● Prototypes
● Exemplars
● Propositions
● CogLab on Prototypes due!
● What is a shoe?
Purdue University
Concepts
Representation of knowledge
● What is the information in Long Term Memory?
PSY 200 w May be several different types
Concepts Definitions
● We will look at three topics in concepts ● Plato (and Socrates) spent a lot of effort
w Definitions (don t really work) trying to define terms like virtue and
w Prototypes (closer to how humans think) knowledge
w Exemplars (more likely than prototypes) w they were largely unsuccessful
Definitions Definitions
● Consider the concept shoe, you might define it as ● Consider the concept shoe, you might define it as Webster s
Dictionary does
Webster s Dictionary does
w A covering for the human foot, usually made of leather, having a
w A covering for the human foot, usually made of leather, thick and somewhat stiff sole and a lighter top.
having a thick and somewhat stiff sole and a lighter top. w Anything resembling a shoe in form, position, or use.
w Anything resembling a shoe in form, position, or use. ● But now consider some situations and decide if they are really
shoes
● Lots of shoes fit this definition
w A shoe that is intended for display only
Definitions Definitions
● Consider the concept shoe, you might define it as Webster s
Dictionary does ● The difficulty is the same one that Plato and Socrates
w A covering for the human foot, usually made of leather, having a had trying to define virtue
thick and somewhat stiff sole and a lighter top. w for any definition you come up with, I can find examples that do
w Anything resembling a shoe in form, position, or use. not seem to fit the definition
● But now consider some situations and decide if they are really ● But we all know what a shoe is
shoes
w so our knowledge of this concept must not be based on some
w a shoe filled with cement, which cannot be worn
precise definition
w a covering worn on the hands
of a person without legs who ● Note, scientists can (sometimes) create precise
walks on his hands definitions (e.g., a dog is defined by a DNA pattern or
w And this? à by mating abilities)
w but the definition is somewhat arbitrary
Prototypes Prototypes
● Perhaps what defines a concept is similarity ● In prototype theory it is possible for an object to
among its members be more or less a certain concept
w there may be no absolutely necessary characteristics ● Consider the concept coffee cup
w there may be no absolutely sufficient characteristics
Prototypes Prototypes
● In prototype theory it is possible for an object to ● In prototype theory it is possible for an object to
be more or less a certain concept be more or less a certain concept
● Consider the concept coffee cup ● Consider the concept coffee cup
w and variations (some are cup-ier than others) w and variations (some are cup-ier than others)
Prototypes Prototypes
● In prototype theory it is possible for an object to ● In prototype theory it is possible for an object to
be more or less a certain concept be more or less a certain concept
● Consider the concept coffee cup ● Consider the concept coffee cup
w and variations (some are cup-ier than others) w and variations (some are cup-ier than others)
Prototypes Prototypes
● Lots of experiments suggest the role of ● Prototypes
prototypes
w Posner & Keele (1968): learning category names for
random dot patterns
w Discriminate two sets of random dot patterns
w Each pattern is a variation of one of two prototype
patterns
A B
Prototypes Prototypes
● variations are made by moving some of the dots ● The key test is done after subjects learn to classify the
variants
w reaction time for judgment is recorded for stimuli they have never
seen before
» new variants
» the prototypes
w reaction time is faster for the prototypes
w which suggests that the mental representation of the categories
(concepts) are built to favor the prototype of the category
variant of A variant of B ● Look at CogLab data
Prototypes Prototypes
● Consider the types of concepts you can have
● Results are based on data from 145 participants (39,285 for global).
w Pattern type Reaction time (ms) Global RT(ms) w and how specific they can be
w Prototypes 798 969 ● things: bird, dog, chair, shoe,…
w Variants 843 1000
● actions: walking, running, sleeping,…
● Unanswered by this (and many other) experiments is what a prototype is:
● goal-derived: things to eat on a diet , things to carry out of a house
w a thing that resides in memory and contains information about the category
features? in case of a fire ,...
w the result of processing information? ● ad hoc: things that could fall on your head , things you might see
● A bit of thought suggests it is the result of processing information while in Paris , gifts to give one s former high school friend who has
just had her second baby ,...
● When studied, these concepts all seem to have prototype
characteristics
Prototypes Exemplars
● We can generate new concepts from old ● A concept consists of lots of examples of the
concepts concept
w it s inconceivable that every possible prototype exists w e.g., a coffee cup concept might contain lots of
ready to be used examples of coffee cups
w some must just be built as they are needed
w perhaps even the prototypes for simple concepts like
bird or shoe are also just built when they are
needed
● A theory that can account for this processing
approach is exemplar theory
Purdue University Purdue University
Exemplars Exemplars
● Comparing an object to see if it is a coffee cup ● Even if it is a new object, it may match several
involves comparing it to each example in memory exemplars well enough to generate an overall
and seeing if it matches anything well enough response to indicate it is a coffee cup
Exemplars Exemplars
● Some coffee cups seem prototypical because ● Unlike prototype theory, exemplar theory also contains
they match lots of exemplars information about the variability of examples within a
concept
w that s what defines a prototype
● Thus, we know that pizzas have an average size of 16
inches but can come in lots of different sizes
● And we know that foot-long rulers have an average size
of 12 inches, but essentially no variability in size
Proposition Proposition
● Network Representation ● Network Representation
w The proposition connects the appropriate concept nodes
w The proposition connects the appropriate concept
nodes
Albert Professor
agent agent
threw Albert threw the book gave The professor gave a test
relation relation
object object
book test
Purdue University Purdue University
Proposition Proposition
● Network Representation ● Network Representation
w The proposition connects the appropriate concept nodes
last Spring
time
Dog
object pigeons
agent
in
Jacob
chase Last Spring, Jacob fed pigeons relation
Dogs chase cats
agent pigeons in
relation Trafalgar Square
object location
relation
Proposition
Proposition
● Ratcliff & McKoon (1978)
w study phase
● One way of combining concepts » subjects are asked to memorize a set of 504 sentences
» 18 - 1 hour sessions!
w there are also other theories of how to do this
w test phase
● Used a lot in Artificial Intelligence » show words and have subjects decide if they were in the
study sentences or not
» measure reaction time for words from the sentences
● Do humans represent interactions of
concepts with propositions?
● Some experimental evidence The bandit who stole the passport faked the signature
Proposition
Proposition
● In the test phase, a word is given and the
● Network Representation
subject responds as quickly as possible
passport passport
passport
object object
Proposition Proposition
● 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
Proposition Proposition
● 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
Proposition Proposition
● 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
Proposition Proposition
● When the next word is shown, its node has not been
● Test Phase : Priming Task
primed, so it responds more slowly
w compare RTs for second in a pair of words
w within a common proposition (bandit -- passport)
passport
signature w between propositions (passport -- signature)
object w not related in sentence (horizon -- signature)
Conclusions
Next time
● Concepts ● Other types of knowledge
w definitions ● Mental images
w prototypes w mental rotation
w limitations of
● Propositions
w Evidence we think in terms of propositions
● CogLab on Mental rotation due!
● Is a picture in your head like a picture in the world?
Lecture 24 w propositions
w mental images, maps
Perception Images
● When we see this
● We have knowledge about, and memories image how do we
of, perceived stimuli represent the
w sights information in the
w smells image?
w touches w analog: copy of
w sounds image in head and
we can retrieve it
● Are these converted into propositions, or
w symbology: convert
concepts
to
w or is there something else? propositions/concepts Purdue University
Purdue University
Images Images
● If you ask me questions about the previous slide,
● I can remember the Runs to
wears my answers would not necessarily identify the
image on the
girl ball representation
previous slide and it shorts
w but is it really?
Mental
w could it be a
proposition?
Observable
There was a girl running to
a soccer ball. She wore red shorts.
Propositions Propositions
● So this suggests that mental images are not exactly like real
images ● It is clear that propositional information influences
w and something like propositional information likely influences mental imagery
reports that are ostensibly based on mental images or mental w but is it all propositions?
San Diego
maps
in w are there mental images, as we tend to experience them?
agent
relation Nevada ● Is there any reason to believe that mental images
San Diego is in in are at all analogous to real images?
California object
relation w yes
California is west of Nevada
object Reno is in Nevada
agent relation
agent
California
West Reno
Purdue University Purdue University
1400
Reaction time (ms)
1200 1200
Imagery Imagery
1000 No imagery 1000 No imagery
800 800
600 600
400 400
200 200
0 0
Small Big part Small Big part
part Purdue University part Purdue University
w instead, language is an instinct ● But this is not what determines our capability to
have language!
Biology Learning
● Like all skills, language needs the proper
● Language is a specialized skill of environment to be developed
human animals w blinded birds cannot navigate by the stars
w Darwin (1871) w Atlantic Ocean turtles that navigate by magnetic
fields need to be in the correct ocean
● Humans instinctively learn language
w effortless ● Language development needs exposure to
w unconscious other people for communication
w procedural knowledge w but it needs surprisingly less exposure than you
might suspect
Purdue University Purdue University
● Children do not learn language by simply ● Instead, each child reinvents language
imitating others w difficult to test because we rarely get to see a
w otherwise they would never come up with language created from a non-language
statements like w however, there are cases!
Pidgin
Pidgin ● The Ten commandments in pidgen
● For example, in New Guinea w as translated by the Alexishafen Catholic Mission in
1937
w pidgin is similar to English (rulers of the plantation) w 1. Mi Master, God bilong yu, yu no ken mekim masalai end ol
woman: meri (Mary, generic word for woman) tambaran.
w 2. Yu no ken kolim nating nem bilong God.
another man s wife: meri bilong enaderfelo man w 3. Yu must santuium sande.
w 4. Yu mast mekin gud long papamama bilong yu.
hair: grass bilong hed w 5. Yu no ken kilim man.
w 6. Yu no ken brukim fashin bilong marit.
helicopter: mixmasta bilong Jesus Christ
w 7. Yu no ken stilim samting.
coffin: die bokus w 8. Yu no ken lai.
w 9. Yu no ken duim meri bilong enaderfelo man.
piano: bokus bilong teeth yu hitim teeth bokus is cry w 10. Yu no ken laik stilim samting.
Purdue University Purdue University
Learning Learning
● Pidgin is not a true language
● In Hawaii at the turn of the century w word order is arbitrary
w workers from China, Japan, Korea, Portugal, w no rules
Me cape buy, me check make.
The Philippines, and Puerto Rico were
w no tenses
brought in to harvest sugar
w no prefixes or suffixes
w they developed a pidgin
w can only be understood in context of the
w some were still alive in 1970 and interviewed
conversation
to see how the pidgin worked
He bought my coffee; he made me out a check.
Education Education
● But then how do we explain that
uneducated people speak improperly? ● This person is not speaking with bad grammar, but
he is also not speaking in Standard American
● e.g. gang member in Harlem English (SAE)
You know, like some people say if you ● He s speaking in a dialect called African American
good an shit, your spirit going Vernacular English (AAVE)
t heaven… n if you bad, your spirit ● Both languages have certain rules
goin to hell. Well bullshit! Your spirit
● His statements obey the rules of AAVE precisely!
goin to hell anyway, good or bad.
● Consider contractions of words
Rules
Rules
● In SAE you can replace some word pairs ● AAVE allows speakers to drop some words
with contractions w …if you are bad… --> …if you bad… is
w They are --> They re grammatically correct
Language Conclusions
● So if everyone is speaking a language, which
is correct? ● Language is an instinct
w none, they are just different w specialized skill among humans
w they are different dialects of English w children need little tutoring to learn language
w children invent language if one is not readily
● Linguist Max Weinreich
available
w A language is a dialect with an army and a navy.
● Language follows rules
● The dialect you speak may give away your
w even when it doesn t seem to
personal history, but it is not fundamentally
worse than any other dialect. Purdue University
Purdue University
Next time
● Grammar
● Long term dependencies
● Phrases
● Language universals
Purdue University
Grammar Vastness
● But in fact, there are infinitely many ● It is amazing how powerful language is
different sentences
● You have probably never heard the
w there is no limit to how long a sentence can
following sentence
be
w moreover, it is probably its first utterance in
● For any sentence I give you, you can
human history, but you understand it anyhow
always make it longer by adding
something like I never imagined we would have to postpone our second
w Professor Francis said that, …. exam in PSY 200 because of cold weather.
Grammar Grammar
● You not only understand language, you ● You can also have sentences without
sense when a sentence is ungrammatical meaning that are perceived as grammatical
w Colorless green ideas sleep furiously.
w Is raining. Sometimes you
still understand w If we don t succeed, we run the risk of failure.
w The child seems sleeping. (a not joking Dan Quayle)
what was meant!
w Sally poured the glass with water. w Twas brillig, and the slithy toves
w It s a flying finches, they are. Did gyre and gimble in the wabe:
All mimsy were the borogoves,
w Rarely is the question asked: Is our children
And the mome raths outgrabe.
learning? (a joking George W. Bush)
Statistics
Nonsense sentences
● Think about the sentence ● If you just learned statistical combinations of
w Colorless green ideas sleep furiously. words, you might think something like this
● What is the probability that in normal life you was a grammatical sentence
would hear the word green follow the word
colorless ? House to ask for is to earn our living by
w it must be close to zero working towards a goal for his team in old
New York was a wonderful place wasn t it
● But we recognize it as a grammatically even pleasant to talk about and laugh hard
when he tells lies he should not tell me the
correct sentence! reason why you are is evident
Purdue University Purdue University
Phrases Phrases
● Every sentence is built out of phrases ● All noun phrases obey certain rules
w rewrite rules
NP-->(det)A*N the happy boy
The happy boy eats candy.
w NP -- noun phrase the boy
The first three words form a unit called
a noun phrase (NP) w det -- determinant: the , a , an John
What identifies a w A -- adjective
The happy boy the tall slender woman
noun phrase? w N -- noun
This is not the same analysis you did w ( ) -- optional
in grammar school! w * -- as many as you want
Purdue University Purdue University
Sockets Usefulness
● In a phrase tree, a phrase is like a component ● It is important to appreciate how the phrase tree
that snaps into the right place
approach simplifies the description of language
w any appropriate phrase works! (even nonsense
phrases) ● Consider how we learn a new word and know
S how to use it
Next time
● Words
● Mental lexicon
● Morphology
● Structure
● CogLab on Word superiority due!
Words Grammar
● The rules of phrases
PSY 200
w rules for combining phrases
Greg Francis w universals for all languages
So why do we have so difficult a time
Lecture 27
●
Morphology Morphology
● English can convey this information in as many ways
● Other languages have many more as other languages, but we use grammatical phrases
variations to do so
● Simple present tense
w Italian and Spanish have 50 forms of each verb
w General truths: Ducks quack.
w classical Greek has 350 forms of each verb w Habitual action: I quack like a duck when I wake up.
w Turkish has 2 million forms of each verb ● Present Perfect Progressive
w some languages build entire sentences around w To express duration of an action that began in the past, has
one complex verb continued into the present, and may continue into the future:
The duck has been quacking for two hours, and he hasn t
● There are rules for these forms finished yet.
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
Compounding Rules
● English also allows new words to be created ● So what are the rules?
out of other words
● One looks to be easy
w and combinations can be combined
w to pluralize a noun, add -s
tooth brush unmicrowaveability N
N
bootylicious -s
dog -s wug
Nstem Nstem
dog bite
Vstem Astemaffix Vstem Nstemaffix
brush By the way, ignore what grammar
tooth
school might have taught you, these
are not adjective-noun phrases crunch -able crunch -er
Root Lexicon
● Some morphemes can
N ● To keep track of what can attach to what,
only be attached to certain
there must be a mental dictionary of
types of words
Nstem Ninflection morphemes
w a root is a word that cannot
be split into smaller parts w -able
-s
w some morphemes attach » adjective stem affiix; means capable of
Nstem Nstemaffix being X d ;attach me to a verb stem
only to roots
w thus, Darwinianisms is a w -er
-ism
word, but Darwinismians is » noun stem affix; means one who X s ;
Nroot Nrootaffix attach me to a verb stem
not
» -ian must attach to a root w ...
Darwin -ian
Purdue University Purdue University
Language families
Exceptions
● Many languages are related to each other and have
● You can probably think of lots of exceptions to broad families
w Which is why you see so many similar words across languages
these types of rules
w many words seem to follow arbitrary rules
Exceptions
Very special cases
● The exceptions generally come from other
languages (with appropriate rules) ● What is the plural of walkman?
w walkmans?
w English adopts the words but not the rules
w walkmen?
● These exceptions tend to be very common words
w Neither feels quite right
w drink-drank sink-sank
● To answer this question we have to understand
w throw-threw ring-rang
how the word walkman is formed and what it is
w sit-sat blow-blew
about
● All derive from a proto-Indo-European language
w this tells us how to pluralize
that formed past tense by replacing one vowel the word
with another Purdue University Purdue University
Parsing Linguistics
sentence
Buffalo buffalo buffalo buffalo buffalo
buffalo buffalo buffalo.
Purdue University Purdue University
● 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?
VP No photo
VP
NP
NP PP
PP
V V N
N NP NP
with discuss with
discuss
sex Dick Cavett sex Dick Cavett
Parsing
Mentalese
● Parsing is something like building a
phrase tree in reverse
● That two different internal thoughts can
give rise to the same language statement ● Let s parse through a simple sentence
is interesting word by word
w it suggests that we think in some way that is w The dog likes ice cream.
different from language S
w a mentalese, if you will NP VP
det N V NP
N
the dog likes ice cream
Purdue University Purdue University
S PP PP PP S PP
Remarkable is the rapidity of the motion of the wing of the hummingbird. The rapidity that the motion has is remarkable.
Difficult sentences
Word order ● These sentences are difficult for humans because of
limited memory
● This sentence is nearly impossible w when a phrase tree includes many unfilled branches of the
same type (PP)
w the parser becomes confused as to which phrase is
associated with a new word
w ends up backtracking to sort out the phrases
w sometimes falls apart ( has has has )
● The grammar generator and the parser are different
S PP PP PP things in your language system
w these are grammatically correct sentences Don t make
The rapidity that the motion that the wing that the hummingbird has has has is remarkable. w they are not good sentences
me show you
w you make sentences like these your writing
assignments!
Purdue University Purdue University
S S
NP VP NP VP
Word pencil is Word pencil is
inconsistent with consistent with
det N det A N
the created structure! the created structure!
the plastic pencil the plastic pencil
Lexical decision
Word ambiguity
● It is not effortless and it can be shown with an
● But you run into the same problem with the experiment
word marks (noun or verb?) ● The experiment is a variation of the lexical decision
w The plastic pencil marks were ugly. (noun) task, which you did in CogLab
w The plastic pencil marks easily (verb) ● In the lexical decision experiment, you see a
sequential pair of words/non-words, and we measure
● Parsers build phrase trees on the fly, so the reaction time for you to decide if the second
backtracking is often required word is a word
w RT is faster if the second word is semantically related to the
w many times it is so fast that we do not notice first word
w seems effortless » cheddar à cheese (faster)
» ship à point (slower)
Purdue University Purdue University
Lexical decision
● The data find that RT is faster if the second word is meaningfully
Evidence of ambiguity
related to the first word
● We can apply the lexical decision task to the ambiguity
● CogLab Data (157 participants) of parsing (Swinney, 1979)
● Condition Reaction time (ms) ● Consider the following paragraph, which subjects
● Associated words 685 listened to
● Unassociated words 704 w Rumor had it that, for years, the government had been
plagued with problems. The man was not surprised when he
● Nonwords 814 found several spiders, roaches, and other bugs in the corner
of his room.
Sentence ambiguity
Evidence of ambiguity
● No one notices the ambiguity ● Interestingly, people often miss ambiguities in
sentences
● But, give a lexical decision test for words verses non-
w Time flies like an arrow.
words
● Humans recognize only one interpretation
w Flashed visually on a screen just after the word was spoken
w Subjects respond faster for words related to either definition of ● Computer algorithms can find 5 interpretations
bug w all grammatically correct!
Schemas / scripts
Conclusions
● Giving computers the general knowledge of
life needed to create something like schemas ● Understanding language
is very difficult
● Parsing
● This is why computers do not carry on ● Phrase trees (in reverse)
conversations with you
● Ambiguities
● Lots of work going on in artificial intelligence to
● Computer generated interpretations
address this problem
● Missing information / schemas
Next time
● Speech
● Phonemes
● Articulation / coarticulation
● CogLab on Categorical perception-
Discrimination
● Why do we say razzle-dazzle instead of
dazzle-razzel ?
Purdue University
Speech Language
Lecture 29 w words
Illusions Illusions
● When you hear what I say, you think you
● The blurriness of speech explains some long-
hear at least held confusions
w separate words w Oronyms (Mondegreens)
w separate syllables
The stuffy nose can lead to problems.
● But you do not The stuff he knows can lead to problems.
w words actually overlap in the speech signal The good candy came anyways.
w it is nearly impossible to take a speech signal The good can decay many ways.
Phonemes Phonemes
● pho·neme \'fo-,nem\ n ● Speech is made of phonemes
[F phoneme, fr. Gk phonemat-, phonema speech
sound, utterance, fr. phonein to sound](ca. 1916): ● Different combinations of phonemes
a member of the set of the smallest units of speech correspond to different syllables and
that serve to distinguish one utterance from
words
another in a language or dialect, the \p\ of pat and
the \f\ of fat are two different phonemes in English> ● We seemingly hear more phonemes than
the ear can actually handle
w how?
Packing Packing
● If the ear can only distinguish up to 20 sounds per ● If phonemes are being smashed together
second there must be some blurriness
w and we can interpret speech that seems to contain 50
phonemes per second w and this can lead to misinterpretations
w then the speaker must be combining many phonemes
together to overcome the limits of the ear
● This is also why computer speech sounds
● The listener hears the 20 (or so) sounds in a second, “funny”
w https://www.ispeech.org/instant.e-learning.text.to.speech
but interprets them as more than 20 different
phonemes w The programs do not combine phonemes in the
right way
Speech Physiology
● So what are phonemes?
● Lungs push
● All speech is made of sounds
air out to
w sound is a pattern of pressure on the ear
w a tuning fork vibrates back and forth to make the sound of a make a
pure tone sound
w Frequency of vibration corresponds to pitch of the sound
w other
● Speech consists of lots of patterns of this sort
organs
w With many different overlapping frequencies
shape
sound
Example Example
● Note what your lips do as you say
● Note where your tongue is as you say w boot book
w bet butt ● The lips add additional frequencies to make different
w beet bat sounds
Thus, you can hear someone smile across a
● The position of the tongue shapes the vocal ●
telephone!
tract and makes different sounds!
● Vowels are all distinguished by the shape of the vocal
w this is true for all vowels tract
Consonants Consonants
● Consonants are more complicated ● (3) Manner of articulation
w different type of control of air flow w /d/, /t/ (stop)
w /p/, /t/, /f/ (not voiced, or unvoiced) ● Each consonant is uniquely identified by its
● (2) Place of articulation: voice (or not) and its place and manner of
articulation
w /d/, /t/ (upper gum)
w /m/, /b/, /p/ (lips)
w /f/, /v/ (lip and teeth) Purdue University Purdue University
Consonants
● Some languages have other characteristics as well Fun
(e.g., tone, timing)
● Why do we say razzle-dazzle instead of
● For example, in English, the difference between /ba/
dazzle-razzle?
and /pa/ is the timing of the release of air for the
consonant and the voicing of the vowel w for phrases like this, people always first say the
word with a leading consonant that impedes air flow
● Voice Onset Time (VOT) is short for /ba/ and longer
the least
for /pa/
super-duper willy-nilly walkie-talkie
● CogLab data: sounds It s a
differ in VOT, judge if helter-skelter roly-poly namby-pamby rule!
same or different sounds harum-scarum holy moly wing-ding
w 163 participants
hocus-pocus herky-jerky mumbo-jumbo
Purdue University Purdue University
Rules
Phonemes
● To say a word, we must combine phonemes
● English uses 22-26 (it depends on how you count)
combinations of voicing, place, and manner of ● In every language there are rules (trees) that
articulation (and 20 vowels) describe what phonemes can follow other
w Rotokas (Papua New Guinea) uses 6 (and 5 vowels) phonemes
w Khoisian (Bushman) uses 141
» Uses clicks as consonants ● Thus, we can identify possible words from
● No language uses some possible sounds impossible words
w raspberries, scraping teeth, squawking,… w plast ptak
w Note, these sounds are used for communication, but not as
part of language! w vlas rtut
● Japanese does not distinguish /r/ from /l/ w thole hlad
Purdue University w nypip dnom Purdue University
Compression Coarticulation
● Moving the tongue (and other ● We generally do not notice these
articulators) around is difficult and takes adjustments
time
wwe are tuned to recognize the new sounds
wto say sounds faster, people use as coarticulation
coarticulation
● This is the main reason computers have
wshape tongue in advanced preparation for
a hard time recognizing human speech!
the next phoneme
wthis influences the sound of phonemes
Purdue University Purdue University
Coarticulation Coarticulation
● Notice that your tongue body is in different ● There are rules for how to coarticulate
positions for the two /k/ sounds in
● When a stop-consonant appears between
w Cape Cod
two vowels, you do not actually stop
● Note too, that the /s/ becomes /sh/ in
w flapping
w horseshoe
● slapped --> slapt
● And /n/ becomes /m/ in
● patting --> padding
w NPR
● writing --> wriding
● You can enunciate these correctly , but in
casual speech you do not!
Purdue University Purdue University
Spelling
Spelling
● It is true that English spelling does not seem to agree
● We have often observed that written language with pronunciation
is different from spoken language w a problem for learning how to read!
Nor should it
George Bernard Shaw (among others)
●
●
w if words were spelled the way they were pronounced, we
complained about spelling in English would lose the visual connection between words
w he noted you could spell fish as g-h-o-t-i w slap --> slapped would become slapt
w write --> writing would become wridding
gh -- tough o -- women ti -- nation w National Public Radio --> NPR would become MPR
Other approaches
● There are other written forms of language that avoid
Conclusions
some of these problems
● The most sensible written ● Speech
language is probably the ● Blurring
Korean hangul
w Drawn characters indicate
● Phonemes
how consonants are ● Articulation
pronounced
● Coarticulation
● Spelling
Next time
● Learning language
● Babies
● Children
● Learning a second language
● CogLab on Age of Acquisition.
Purdue University
Lecture 30 ● Learning
w you do have to learn some specifics for your
native tongue
When should you learn a foreign w rules
language? w words
● How do you learn a second language? ● Repetition of the same sound leads to
boredom and fewer sucks
w ba, ba, ba, ba, ba, ba, ba, ba,
● What do babies do?
Purdue University Purdue University
w babies suck more often w you can change the pronunciation (voice onset timing) of /ba/ and
still hear it as /ba/ (CogLab data)
w ba, ba, …, pa, pa, pa, pa, pa,…
w But too big a change in VOT turns it into /pa/
w babies hear it the same way
● CogLab data:
Identification task
Discrimination task
BA PA
Babbling Babbling
● Babbling sounds are the same in all languages ● Babbling teaches child how sequences of muscle
w patterns are common across languages
combinations lead to different sounds
● By the end of the first year babies combine syllables w necessary to produce speech
to sound like words ● By about 10 months babies learn the sounds of their
w neh-nee native tongue
w da-dee w they can no longer distinguish phonemes that are not part of
w meh-neh the language
● Babbling is important w Part of learning is forgetting!
w children who do not babble often show slower speech
development
w deaf children babble with hands, if parents use sign
language
Purdue University Purdue University
Errors
Expected errors ● Children do make errors, but the errors are consistent
● Consider a child hearing adults talk and how with rules of language
they might incorrectly apply what they learn ● Children often over generalize a rule
w -s to pluralize a noun
● Out of 66,000 sentences, children never made » Mouses, leafs
w -ed to make the past tense of a verb
these errors
» My teacher holded the baby rabbits and we patted them.
Grammatical Not grammatical » Hey, Horton heared a Who.
He seems happy. --> He is smiling. --> » I finded Renee.
Does he seem happy? Does he be smiling? » Once upon a time a alligator was eating a dinosaur and the
dinosaur was eating the alligator and the dinosaur was eaten
by the alligator and the alligator goed kerplunk.
He did eat. --> He did a few things. -->
He didn t eat. He didn t a few things.
Overgeneralization Overgeneralizations
● We know this is the most difficult part of language because adults
● These past tense forms sound wrong because English make the same kind of mistakes
w tread - trod strive -strove
has around 180 irregular verbs
w dwell - dwelt slay-slew
w inherited from other languages
w rend - rent smite - smote
w These past-tense forms are not derived from rules
● Sound weird because we do not often hear them
● Irregular forms have to be memorized, word by word w many adults regularize
the words
● If a child cannot remember (in its lexicon)
w treaded, strived, dwelled,
w s/he defaults to the rule slayed, rended, smited
● These errors are for the most difficult parts of a w thus language changes…!
language to learn
w Because they don t follow the normal rules
w This new golf ball could obsolete many golf courses. ● What accounts for the difference?
w If she subscribes us up, she ll get a bonus. w most likely it is age
w Boiler up! w there seems to be a critical period during which
language can be learned
● Children s errors tend to track the more difficult
w beyond age six (or so) it becomes more difficult to learn
aspects of a language, relative to other languages a language (first or second)
w Adults make the same kinds of mistakes for still more difficult
to remember cases
Next time
● Language & brain
● Broca s aphasia
● Wernicke s aphasia
● Anomia
● Language ability of chimps
● Wernicke s area ==> Wernicke s aphasia w omitted function words (or, be, the)
w skipped function words when reading (or, be, the) but read
similar sounding words (oar, bee)
w named objects and recognized names
w high (nonverbal) IQ
● Patients show across let go and they're both gonna fall down...but already then...I can't see
What well enough but I believe that either she or will have some food that's
not good for you and she's to get some for her too...and that you get it
w poor comprehension brings and you shouldn't get it there because they shouldn't go up there and get
it unless you tell them that they could have it. and so this is falling down
you to the and for sure there's one they're going to have for food and, and didn't
w poor vocabulary come out right, the uh, the stuff that's uh, good for, it's not good for you
hospital? but it, but you love it, um mum mum (smacks lips)...and that so
w empty speech they've...see that, I can't see whether it's in there or not.
Examiner:Yes, that's not real clear. What do you think she's doing?
Boy, I m sweating, I m awful nervous, you know, once H.W.:But, oh, I know. She's waiting for this!
in a while I get caught up, I can t mention the tarripoi, a Examiner:No, I meant right here with her hand, right where you can't
month ago, quite a little, I ve done a lot well. I impose a figure out what she's doing with that hand.
lot, while on the other hand, you know what I mean, I H.W.:Oh, I think she's saying I want two or three, I want one, I think, I
have to run around, look it over, trebbin and all think so, and so, so she's gonna get this one for sure it's gonna fall down
there or whatever, she's gonna get that one and, and there, he's gonna get
that sort of stuff... one himself or more, it all depends with this when they fall down...and
when it falls down there's no problem, all they got to do is fix it and go
right back up and get some more.
Purdue University Purdue University
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.
Purdue University Purdue University
Chimpanzee language
Problems
● In the 1960s several research groups reported
teaching chimpanzees American Sign Language ● Just like with Eliza (the computer therapist) it
(ASL) is easy to attribute language ability where it
w after failure to teach spoken language does not really exist (9 month old children)
w other groups taught chimps to press symbols on a computer
keyboard or string magnetized plastic shapes on a board ● You can teach an animal a lot using simple
● Claimed to teach chimps hundreds of words conditioning tricks
w and chimps created new compound words ● Researchers were quick to excuse mistakes
» swan -> water bird
as play , jokes , puns , metaphors ,...
» stale Danish -> cookie rock
» There’s a movie
» https://www.youtube.com/watch?v=u4T8ZeZy22M
Purdue University Purdue University
Grammar Grammar
● Likewise, the chimps never produced complex sentences
● Chimps failed to learn the rules of ASL ● They tended to say things like the following
grammar w Nim eat Nim eat.
they communicate
w Drink eat me Nim.
w unable to understand complex signs but not with real
w Tickle me Nim play.
language
● Seemingly able to understand complex w Me eat me eat.
sentences w Me banana you banana me you give.
w Banana me me me eat.
w Would you please carry the cooler to Penny?
w Give orange me give eat orange me eat orange give me eat
● But really, the chimp need only understand orange give me you.
Evolution Evolution
● Note, it would have been interesting if chimps ● Chimps are the closest living evolutionary
could learn language relatives of humans
w and not inconsistent with the idea that we have a w so if any non-human living animal could learn
language instinct language it would probably be chimps
● But the failure of chimps to learn language ● But in evolutionary history, chimps and humans
does not go against the idea that language split from a common ancestor millions of years
evolved in humans ago
w as some people have proposed
● Humans evolved a language skill and chimps
did not
Purdue University Purdue University
PSY 200
● Awareness of events, stimuli, thoughts,
Greg Francis self
● A sequence of meaningful items
Lecture 32
● Stream of thoughts
● Distinct from unconscious processing
Do you see red like I see green? (e.g., hearing a sentence, retrieving
information from memory,…)
History Materialism
● Nearly all scientists are materialists, but
● Descartes dualism (Cartesian dualism) old ideas die hard
w pineal gland link between body and spirit
● A lot of work (e.g., fMRI) looks for the site
w how they could connect was a real problem
of consciousness
w a special physical transformation
w thalamus
w reticular formation
● Mind-body problem
w quantum mechanics
● Materialism (the brain is the mind)
w or the mind derives from the brain w distributed awareness
Purdue University Purdue University
An analogy An analogy
● When did the British empire learn of the end of the War of ● For complicated systems like the British empire (and
1812? human brains)
w treaty signed in London months before the Battle of New w different parts know different things at different times
Orleans w there is no official moment of knowledge
w word was not received by British troops in America until two w no official moment of consciousness!
weeks after the Battle of New Orleans (January 8, 1815)
● Demonstration
w when does the class know/understand?
same logic to a
something else to put it all together ? computer
w can consciousness arise from non-conscious w if a conversation with a
computer is
processors? (artificial intelligence?) indistinguishable from a
conversation with a
human
w Then conclude the
Purdue University computer is intelligent Purdue University
2. Qualia 2. Qualia
● Consider two people who see the world in
● Some researchers object to the very idea that
computers could become conscious color opposites
w They argue that some things in consciousness are not
just computation Qualia for person 1 A red apple with a
w e.g., consider the color red green leaf
w There seems to be a particularly subjective experience
of seeing something red
2. Qualia 2. Qualia
Qualia proponents argue, for example,
Clearly, there s a big difference in the
●
●
w you can learn all there is to know about light waves,
perceptual experience of these people, photoreceptors, neural transduction and coding of
but their behavior is essentially the same color,…
w But suppose you never see any red objects
w And there seems no way to distinguish one
w Your knowledge will not tell you what you will
experience from the other
experience when you first see the red of an apple
w It s the unmeasureable experience that is a w Indeed, you could be tricked into believing a green
qualia apple was red (if you had never seen green either)
2. Qualia Conclusions
● But this is a defeatist argument, or a pointless one ● Consciousness
w if I knew everything about light, photoreceptors, and
● distributed processing in the brain
neural representation of colors, then I would be able to
know what I will experience when I see red w no site of consciousness
w it is difficult (maybe impossible for any single human) to
know (or even imagine knowing) all that information in an
w no time of consciousness
academic sense
● Chinese room
w but that doesn t mean that such information does not
exist ● Qualia
It s partly an empirical question
●
● Artificial Intelligence
w But no one can do the experiment
● Daniel Dennet Consciousness Explained
Purdue University (1991) Purdue University
Next time
● Review for exam 4
● After exam 4
w Decision making
w Framing effects
w Risks
w Alternatives
w CogLab on Monty Hall
w alternatives
w loss aversion
● If program B is adopted, there is a 1/3 probability ● But the phrasing makes a difference in
that nobody will die and a 2/3 probability that 600 the choices of subjects
people will die. 61%
w why?
Risks Risk
● Risk corresponds to those events that ● Humans sometimes prefer risky options
occur with probability over non-risky options
w will I like the next movie starring Matt Damon?
w and vice-versa
w will I live to be 50?
w will the dice show double sixes?
● When the choices are perceived as losses
w subjects tend to be risk-seeking
● Events that occur with certainty are without
risk ● When the choices are perceived as gains
w the sun will rise tomorrow w subjects tend to be risk-averse
w I will be older tomorrow ● Decision making is open to manipulation
w I will give you an A if your grade is 90 or above
w subjects can contradict themselves
Purdue University Purdue University
● Subjects tend to prefer the sure gain ● Subjects tend to prefer the risky option
w risk averse with perceived gains w risk seeking with perceived losses
Purdue University Purdue University
Next time
● Problem solving
● Expertise
● Analogy
● Set effects
w functional fixedness
● Insight
● What does that aha! feeling mean?
Purdue University
Further study
Useful background
● Brain characteristics
● Mathematics w PSY 222: Introduction to behavioral neuroscience
w Many psychologists have little mathematical w PSY 322: Neuroscience of motivated behavior
background
w PSY 324: Introduction to cognitive neuroscience
w But it is especially useful for cognitive psychology w PSY 352: Introduction to Neuropsychology
w Take as much mathematics as you can, especially w SLHS 401: Language and the Brain
» Calculus (MA 161, 165 or 223)
w PSY 512: Neural systems
» MA 375 Discrete Mathematics
» Linear (matrix) algebra (MA 262, 265)
» Differential equations (MA 266)
Hot topic 3
Graduate school
● Data analysis
● Big data requires a different kind of statistics than has ● Grades
been used for simple experiments
● Financing
● Moreover, there seem to be problems with present
statistics even for simple experiments ● After graduation
● Areas of science that depend on statistics (e.g.,
psychology, biology, medicine) are facing a crisis as
findings that satisfied old criterion are found to be false
w And unbelievable findings meet the statistical criteria
w “Pre-cognition”
Next time
Purdue University
Logic Logic
● Some problems are completely handled with logic ● People do better if the task is related to some overall theme
Logic Topics
● CogLab data ● Similar to characteristics of decision making a lot
● Count number of correctly chosen cards (2 is max) of problem solving techniques are heuristics
● We will look at a number of factors that influence
our ability to solve problems
w expertise
Class data
(166 w analogy
participants)
w set effects
» priming
» incubation
» functional fixedness
w insight
● Some people learn how to solve particular ● Take second year physics students (novices) and ask them to
classify a bunch of physics problems
types of problems
w they tend to group them by surface similarities
Correct pieces
25
real game
Expert 2: Conservation of energy 20
Expert 3: Work-energy theorem. They are all
● Experts have schemas that allow Master
15
straight-forward problems. Expert 2: These can be solved by Newton s second law
them to organize the piece Beginner
10
Expert 4: These can be done from energy
considerations. Either you should know the principle
Expert 3: F=ma; Newton s second law.
positions
of conservation of energy, or work is lost Expert 4: Largely use F=ma; Newton s second law. 5
somewhere. w They only need to remember the
0
schema
1 2 3 4 5 6
Trial
Purdue University Purdue University
20
randomly 15
is seen and heard across the entire kingdom ● Glick & Holyoak (1980)
simultaneously. The dictator also demands that w subjects read stories like these and were
the parade be the most impressive ever at the asked to solve the problems
fortress. Splitting up the army would allow it to w even when shown one solution and told that it
be seen everywhere, but would make the could be applied by analogy to another
display at the fortress unimpressive. w subjects used analogies only 20% of the time
● One of the problems handed out can be ● You can be biased by lots of things
solved by analogy to these two problems w problem statement
w previous methods of reaching solution
Purdue University w general knowledge Purdue University
Insight Insight
● Intuitively, we sometimes feel as if we have a ● Warmth stays mostly steady, right up to
strong insight into a problem and its solution proposing a solution
becomes obvious w the aha feeling
w the aha feeling
w is it real? Unfortunately,
w what does it correspond to? the feeling does
not necessarily
● Using problems like the Bronze coin and indicate a
correct solution
the Tree planting problems (Metcalf, 1986)
w subjects judge their progress with a warmth
rating, every 10 seconds, over 5 minutes
Purdue University Purdue University