Introspection: trained observers report contents of consciousness under carefully controlled situations àstudied by wurnt
àbased on free-association
Empiricism
Empiricism: based on experience/environment
-Wurnt
Nativism
Nativism: inborn
stuff
Behaviorism:
unconcerned w/ working of mind: only external b/h
Information-processing analysis.
Information-processing
analysis: how info is processed in mind
Sternberg paradigm: subjects are shown small # of digits (1-6) àmeasure how long it took them to recognize the digit.
àperceive
stimulus (encode it) compare it to the new stimulus àmake decision àgenerate a response
-upwards slope à38 milliseconds longer for each added digit to the # series
àproblem: why is it so monotonic? it don’t matter where the digit is in the series!!! Why is there an upwards slope!!! [should depend on length of series]
àmore
looking time!!! Since in every digit place of the number series, you
seek through all #s until you get to the right # in every digit place!!!
-addition of time per unit
of search!!!e
criticism:
Cognitive neuroscience:
how the brain works, to cognitive processes.
Neuron: a cell which accumulates/transmits electric pulses
-100 Billion in brain àmost processing = through interaction
-Soma: name for the cell body of the nerve
-Dendrite: arms leading the impulse to the nerve
-Axons: leading impulse away from the nerves
-Axon hillock: the little hill where the axon starts
-Myelin sheath: covers some axons
-Synapse: where an axon and dendrite meet
-Axon terminal: end of axon, which releases the neurotransmitter
-Neurotransmitter: chemicals released by the axon which act on the dendrite membrane to change its action potential
-average soma/dendrite has 1000 synapses, and each axon synapses to about 1000 neurons
àis
net of all incoming nerves – positive (excitatory) àin adds up in axon hillock,
which eventually fires off.
Action potential:
when positive sodium ions enter the nerve momentarily making it more
positive than the outside
-info in brain – 1 out of 2 quantities
Medulla: controls
breathing/swallowing/digestion
Cerebellum: motor coordination/voluntary movements
Thalamus: relay station for motoric/sensory info from lower areas to cortex
Hypothalamus: expressions for basic drives
Limbic system: border areas b/w cortec and lower structures
àcontains hippocampus
hippocampus:
vital to memory
cerebral cortex/neo-cortex:
external layer of brain -amount of wrinkles àthe diff b/w humans and lower mammal.
Neocortex
Crossed over –right hemisphere
controls the left side of body/vise versa
4 parts of neocortex
-52 Brodman areas of brain
-hemispheres are somewhat specialized
-corpus callasum: connects b/w hemispheres
àsplit-brain patients:
to avoid seizures, corpus callosum is cut
broca’s area (grammatical area) and wernickle’s area (meaning of words) 2 areas in left area critical for speech
àdamage to them: aphasia:
severe impairment of speech
Topographic organization: info processing is organized spatially
ài.e. adjacent area of retina processed
adjacently in brain
-some areas are size-proportionally
over-represented in brain –more sensitive areas.
Electroencephalography (EEG)
-records electrical potential of scalp
àmeasures large groups of active neurons
àshows waves
Event-related potential (ERP):
-averaged responses to a specific stimulus
àgood temporal resolution, but hard
to tell what area of brain is working
-2 techniques which are good
in location but not in temporal (timing) of the neural activity
-uses radio waves/magnets to detect higher amount of oxygen (in the more active areas of brain)
Connectionism
Connectionism: the method of connecting various processes of brain to build models of how brain deals w/ higher level thinking, like judging people
àexample: parallel distributed
processing (PDP)
parallel distributed processing:
example:
-When list of people w/ various
groups of classifications, there is a central area of memory which remembers
a series of connections to specific choices in the various variables.
example
I want to retrieve info about Jim: I have to connect the variable about:
àand
retrieve the right info from there
àallows for generalities
ài.e. what a certain group of people are like
vision
Visual agnosia:
inability to recognise objects, and not because of intellectual or sensory
loss.
2 groups of visual agnosia:
-->the kind of visual agnosia
depends on how high the level of damage to the brain is.
Cornea: bends the light when light first reaches outside of eye
Aqueous humor: liquid inside pupil -->b/w cornea and pupil
pupil: the hole to the eyeball: stretched by the Iris
Iris: muscle increasing or decreasing he pupil - has the color of Eye
Lens further bends the light
Retina: the sheat of photoreceptors
Fovea: center of retinal -->not exactly where optic nerve leaves eye
Bitreous humor: liquid inside the eyeball
Cones: for day
Rods: for night
-Both optic nerves meet at optic chiasm
-->there, nasal part of retina (peripheral/sides of plain of view) crosses over and outside part of retina (center/nasal part of sight) doesn't cross over
-->therefore, whatever
is seen in the left part of retina, (right side of visual plain) goes
to left part of brain, thus creating an effective crossover of
the visual plain in the brain
2 stations along the way:
Lateral geniculate nucleus -details/recognizing objects -'what'
Superior colliculus
-localization of objects in space -'where'
-->they
connect to cortical visual areas (i.e. V1)
Ganglion cells/lateral geniculate
nucleus: either on-off or off-on cells (on-center =off-sides or off-center=on-sides)
In primary cortical cells:
2 kinds of responses:
Edge detectors: respond positively to light at 1 side of a line but negatively to another side of a line
-->probably made from 2 rows of basal ganglios together -one side reacting negatively while 1 positively (see p.43)
Bar detectors: respond positively to light at center of the straight area (streight/long/narrow strips) and negatively to the outsides
-->probably
made from row of single ganglion cells: all of them having center on
while ouside off meands that this cell is stimulated: there is a line!!
(see p. 43)
-->see p. 42 for bar/edge
-both ege/bar detectors each respond to:
* specific part of visual plain
* specific orientation
* specific width
-the bar/edge detectors are
probably lines up in a hyporcolumn in the primary visual cortex
Feature maps:
spatial representations of field of ight in each area of the brain which
processes the diff. aspects of sight (i.e. movement/color/form)
Depth/surfaces
-we see a 2-D image in the
retina and we have to make it 3-D, so we infer the depth
* texture gradient: more tightly packed: further
* stereopsis: the diff. b/w the imge seen by each eye
* motion parallax:
further things move slower than fast things: can even be noticed while
slightly moving head or eyes!
question:
how does brain process
2 1/2-D sketch: a drawing made to identify various visual features to the viewer
3-D model: representation
of objects in a scene.
Object perception:
-brain uses rules to make objects
out of visual stimulus. those rules appear to be innate.
gestalt principles of
organization: principles of the gestalt school which the brain
organizes stimulus into objects -->see p. 47-8
* proximity: when there is a pattern of lets say, lines, then the closer lines are assumed to be together
* good continuum: lines crossing each other assued to be contining straight and not having the sharp turn.
* closure: when object is assumed to be an object since it is closed
* overlap: when
1 object occludes another, i.e. circle over another, almost full circle,
bottom object is assumed to b also a circle, even though it could be
another shape.
-harder to perceive things
contrary to gestalt rules: i.e.: ThIsSenTeNcEiShArDtOrEaD.
Aqueous humor:
liquid inside pupil -->b/w cornea and pupil
Visual-Matching Models
Template Matching: brain compares visually perceived pattern to various stored patterns
-problems could occur when there is a diff b/w sight atd template's:
-->though humans have great
flexibility and can recognize many variation of the template (i.e. can
compare diff. shapes/orientations/sizes
Feature analysis: stimuli is a combination pf elemental features.
-i.e. regardless of size/shape, the visual stimulus corresponding to the letter A is thought of as a combination of / \ and -
-->kind of like edge/bar detectors
-->huge resource
saving over the huge templates needed in the template theory
-when we see an unclear stimulus,
we extract or combine
features to make complete objects
Object recognition: recognizing object categories to recognize more complex objects:
* Recognition-by-components theory:
-->biederman - 36 basic geons -->each could vary in size/proportion
3) having identified pieces making up the object, the brain recognizes the pattarnsd on
geons as the object
-nly edges need to recognize ions -->not color/texture/small details
-->geons - just like feature
analysis of the letters
Biederman/Beiring/Ju/Blickle study 1985
componant deletion: when a componant is deleted from the drawing of an object.
midsegment deletion:
the lines of the drawinf are dotted, and not continuous:
-study shown: with faster flashes of drawings, drawings w/ midsegment deletion is easier to recognize, while when the flashes slow down a bit, then they b/c just as easy -and maybe component deletion a bit easier to detect (the real object)
-hard to segment word àit sounds like the words are segmented, but it is really 1 long stream of words
-intraword segmentation is also hard
àidentification of phonemes:
sound unit i.e. what syllables are made of:bat = [b], [a] , [t]
identification problems:
1) segmentation of phonemes
àunlike written words, spoken words have phonemes run into each other, making it hard to identify
àindividual diff. even w/I the same
dialect/accent
2) coarticulation:
-the word is made by 1 movement of the vocal tract (making it harder to identify)
ài.e.
‘bag’ = from b, it moves towards the A sound and concurrently
to the G sound.
-speech perception is in the
left temporal lobe
feature analysis of speech
-feature-analysis and
feature-combination seems to underlie speech perception, just like
visual recognition
features of phonemes include:
Example: p/b/t/d à4 consonants
Voicing | ||
Place of articulation |
Voiced | Voiceless |
Bilabial |
B | P |
Alveolar |
D | T |
-those features are used to
recognize the phonemes
àmore confusion w/ more diff. in features:
-the diff. b/w P and T is only
in place of articulation/ P and D is voicing and place of articulation.
Therefore more mistakes b/w P and T than P and D.
Acoustic diff:
-voiced phoneme (i.e. B) is immediate sounded w/ release of lips, whereas voiceless phonemes (i.e. P) is heard a few milliseconds later)
àtiming of sound seems to be the cue
as to the diff! (i.e. b/w B and P)
categorical perception: perceiving groups of stimulus w/o gradient b/w them. àhard to see diff. w/I the group
àwe
have hard time distinguishing b/w pairs of b/p at diff. time voice-onsets,
but we do it well to same voice-onset
context/pattern recognition
bottom-up: perception using the stimulus àpattern
top-down: processing using high-level general knowledge to interpret the low level-perceptual units àespecially used in ambiguous cases, like then the H looks like an A in the word the àthe prior knowledge is that there is a word THE but no word TAE
àcontext
top-down: can also be
seen if lets say, each 3rd letter is replaced by an x
word superiority effect: if letter or word is flashed fast, and then 2 alternatives were given as to what was flashed, the words guesses would have less mistakes. I.e. D vs. K or WORD vs. WORM
àonly need to identify some part of the feature of the letter to identify the whole word, but needs to identify some features to tell the diff. b/w single letters.
-* was the brief beep instead
of the sound:
-same thing happens with other over-learnt stimulus – i.e. in faces w/ missing features.
àvery few features are needed to recognize a drawn face
-4 quadrants , each having a series of words… gradient from e to c. In each quadrant, there is a varying amounts of contextual info. Quadrants ranging from:
-as one went down the quadrant, looked increasingly like the possible word
logical method of perception
àfuzzy
logical model of perception
-since there are 2 sources of info (stimulus/context), they are probabilistically different.
probabilistically different:
evidence for features of the stimulus is independent of the context
-depends heavily on nerve system:
the features (nodes) of letters are activated is seen and inhibited
if not present in the letters. Then the letters excite the words an
inhibits the word which doesn’t contain the given letter. Words can
also activate letters (downwards) to support its recognition. Recognition
of the word inhibits other words. According to this model, the
word superiority effect is b/c the word supports is component
letters.
Conclusion
Light energyàfeature extractionàprimal sketch(initial sense of info) àdepth infoà2 ½D sketchàgestalt principles of organizationà3-D sketchàfeature combination/contextual infoàRECOGNITION
Parallelism: using diff systems for diff. tasks
àlimited amount possible àoften, after a while = serial bottleneck
ài.e.
in motoric acts, only 1 can be done
serial bottleneck: the point where it is no longer possible to process diff. things in parallel
-models explaining when in
the stimulus processing the bottleneck is
auditory attention
dichotic selection task: task revolving listening to 2 ears
àthus processing 2 things w/ same system
-very little is processed from unattended ear
-i.e.:
àcan’t
remember: language/spoken words
Filter theory –broadbent
-filter selection: early selection theory based on physical characteristics
ài.e. listening only to L. ear
àinfo
is processed until bottleneck is reached
hearing: temporal cortex, in Sylvain Fissure
ànear 41/42 of Brodmann’s areas
àshows enhanced response to listened-to ear
àin cases where we choose by of physical features (i.e. which ear)
àthough we can also filter semantic contents
ài.e.
hearing our name
example:
-we can jump /w ears to make out a sentence
Attenuation theory: b/c of physical properties, some messages weakened, but not eliminated.
àif
stimuli is attenuated (weakened), we could apply semantic selection
but it is harder
Late selection theory
-info processed unattenuated
àprocessing system can process both. The problem is reacting to both.
àwe decide which to react to, either based on semantics or on ear of
origin
-semantic selection/detection
= easier in this theory than in attenuation theory
-unattenuated stimuli can be remembered, but just for brief moments
àechoic memory
Visual attention
-visual acuity is highest in the small part of the visual field (fovea)
-->by moving eye, we fixate on what we want to give maximum attention to.
-->though visual
attention could go as far as 24 degrees away from the fovea
spotlight metaphor
-the visual attention could
vary in angles it spans. narrowing its span gives maximum processing,
yet when there is something in other parts of the visual field, beyond
spotlight (i.e. an unexpected thing) it takes time to respond/process
it.
Experiment -Leberge
-participant told to focus on center of figure (in a way that figure would be in the fovea) -shown fast figures of ++7++ and asked whether it was a 7 or distracters, (T or Z). -->reacted faster if 7 is in the middle than n sides (i.e. +7+++)
-->conclusion:
attention spanned differentially across the visual field
-in complex visual situations,
we need to shift our visual attention
-experiment -Neisser/Becklen
-2 TV shows superimposed on each other: one of a hand-slapping game, one of a ball game
-->participants
could monitor one and note odd events (like players shaking hands) yet
could not monitor both without difficulty/missing great deal of critical
events.
note: both physical and content cues were used:
1) physical: moving eyes so main thing will fall on fovea
2) content: using the
content to know where to move eye to
Neural basis of visual attention
-just like auditory attention: in visual attention visual attention directed to a spatial location enhances cortical sign from that location
-->i.e. directing attention to a specific spatial location will activate the specific hemisphere side
à(i.e. visual part of brain is
also mapped out crossed over)
-->i.e.
in higher-order feature (i.e. attending to chairs and not tables), the
response didn't last for more than 200ms.
conclusion:
-we can assume that people select in a 'early selection model' -->on basis on physical properties, especially on the basis of location
-->in higher-order properties,
such as when person wants to find friend in crown: looks for desired
features
Visual sensory memory
-if given array of items to remember, we would need to look at 1 location at a time
-->i.e.
if given a series of letters arranged in rows: need to l letter at a
time to remember them
Experiment:
-card with letters flashed
at participant: they could remember a few, but noted that there were
more that they didn't manage to process
Experiment #2 - sperling
-tone was given right after row was shown (high for 1st row, medium for middle, low for bottom row) -subjects asked to remember 1 row, while tone came right after the display.
àremembered more, since the tone helped them focus.
àthough
the more delay of tone àless ability to focus on specific stimuli àless
remembered
Partial-Report procedure: subjects were asked to report partial series of the whole amount, which was asked to remember
Whole-report procedure:
subjects were asked to recall everything that was asked of them to remember.
Visual sensory store: a memory that can effectively hold all the info in the visual display
-Called
iconic memory, like there is echoic memory in auditory system
Iconic memory: brief memory of visual field after it has been seen. if not processed fast, it would be forgotten
àpeople think that it is the rod-based afterimage seen on the retina. (and, by the way, that is why it is longer kept at night)
àsome
claim that it has no relevance, since people don’t see things in flashes,
in daily life
Sperling paradigm: how fast attention can process/encode visual iconic memory.
ài.e.
as seen in his experiments: people can remember every item for a second,
and there were 4 items in his experiment row, then each tem should take
0.25 sec. to process
Pattern recognition and Attention
example: participants asked to find a T in a pool o 30 Y’s and I’s. they subject need only find the cross-bar of the T which distinguishes it from all the I’s and the Y’s
àtook them longer to find a T in a pool of Z’s and I’s
àless features to distinguish among.
àperson
has to scan and then reject the incorrect figure/letter àtakes
longer if more features have to be analyzed in order to come up with
a correct answer
-in stimuli outside focus of
attention, we mix up features (i.e. pink T, yellow S and green F à
many mistakes =, include pink S)
-people can attend to specific
features. Some claim that people can focus on more than feature at a
time, but the more specific features = more noisier/less accurate
Neglect of the visual field:
-visual attention to a specific
location results in enhanced accticity in the right portion primary
visual cortex
-shift in attention really takes place in the posterior parietal lobe/mid brain area called pulvinar.
àdamage to those area, especially parietal lobe causes inability to attend to those areas
àright parietal = left visual neglect
àLeft parietal = right visual neglect
Neglect: problem shifting attention to a certain receptive field
-visual extinction:
when visual stimuli goes to damaged side, it could be processed, but
w/ competing stimuli from other side of visual field – can’t be
processed.
Unilateral visual neglect: more serious kind of neglect where the side is ignored
àmight not shave 1 side of his face
-sometimes, easier to focus
on object than on spatial location
ài.e. in comparing 2 icons of 2 object figures – easier to compare the diff. on same obj, than on the other obj, even though the 2 differing parts were closer in space
àconclusion: easier to focus on obj.
than on spatial features
another object based attention technique
Inhibition of return:
harder to return to look at particular region of space once we looked
there: we are inborn to have searching techniques for objects!!!!!
-some people have neglect to 1 side of visual field
-some have neglect to 1 side
of objects
A central bottleneck
We know that each modality
has a bottleneck, but, so do 2 tasks of diff. 2 modalities
-SOA (stimulus onset asynchrony):how much time b/w the 2 stimulus
àshorter SOA = more interference of
1st task to 2nd task
-Therefore, one can say that some tasks -->can not be done totally in parallel, and must finish 1 task b/f starting the next
-->audio processing
might have taken place while visual info was acted out, but it won't
be acted on until 1st task done.
-one can do 2 tasks at once, if only 1 is being thought about (and the other one is automated.
-->i.e. driving is automated, speaking is thought about
-->but hard to speak
when driving gets hard and you need to think about the driving.
Automaticity
automaticity: when you practice - you reduce central cognitive component until it is automatic.
-example of automated
b/h = driving
3 processes in automated b/h
àeach one is parallel and can function independently of each other
àthough each system is serial: it can
function at 1 thing at a time
-central cognition = often the bottleneck, since it is required for coordination.
àAutomaticity eliminates need for central
cognition, thus reducing bottleneck
-using paralel motor/perceptual
systems = no conflict
stroop effect
stroop effect:
the effect of seeing words and recognizing them immediately
stroop experiment:
-with the colored words: much faster answered color words, where their ink color matched than when didn’t match. also many more errors in naming the ink color àhard to inhibit the word, even when asked to name the ink and not the word name of the color.
àshows the automated element of reading.
warner/Polansky Experiment
-Rows w/ diff. # of digits of #.
àeasier to say the digit name rather than the # of digits
ànumber recognition is much more automated than counting.
Knowledge-representation:
proposals of how diff. types of info are encoded and processed
Dual-code theory:
we remember things better if we assign a visual image to the verbal
thing you want to remember (link visual/audio for enhanced memory)
Comparisons of verbal vs. visual processing
-->reacted to shapes faster than words of those letters
-->b/c in words, they are encoded left-right/top-bottom (linear order)
-->reacted to
words faster if they were in linear than spread out in cue-card
Conclusion: words
are stored in linear order whereas shapes are stored according
to spatial position
-->diff. areas of brain used (as seen by blood-flow)
-->for mental verbal repetition: same as spoken words!
-->for
visual memory: occipital area -->same as vision!!!
-->same mechanisms for internal
and external processing of stimuli
-knowledge of brain localization
shows how processing works
-prob: instead of understanding
the connection, we see one unscientific homunculus being who also sees
and hears internal things
Visual and Spatial Imagery
Mental Images:
representations of things (not visual imagery)
Mental Rotation: same image, but diff rotation (2-D Representation images if 3D Items)
àmore angles of rotation àmore reaction time
à3D Rep. is in th ebraib
àneural mental process – analegeous
to physical rotation
image scanning:
we look through an object (i.e. house) mentally to find its critical
info (i.e. # of windows)
2 experiments
Brooks
-vs-
-3 ways of answering
àslowest: pointing to a diagram while mental image!!!!
àB/c already using that part of brain
àreally
b/c of spatial inference
Conclusion: interference in stimuli if conflicting special process
-when asked to mentally compare visual info (i.e. obj)
àless diff. b/w obj. = more time diff.
àsame w/ real objects
àmental/perception
processing/discrimination
=
similar functions
2 types of imaging
àin parietal lobe, independent of any specific modality!!!
àCould be accessed tactually/audiorily
àcould
also be seen in mental perception
àalso true to switch to diff. images of anambigous image àthough mentally hard
àpeople
prefer an actual image to a face image
hierarchical structure of images:
-people break an image into specific subpart most of the time
àat each level, the unit is called a chunk
àsee
p.122-123
-also hierarchical organization of special images
ài.e.
people broke up a room into idiosyncratic sub-areas to remember what
is in it
cognitive maps
Map distortions
-When map is congruent – i.e. border curves
àmore mistakes than no border, or straight border
àmistakes are b/c relative location of the whole area and not the specific place
àb/c of higher order places
-for example, the alphabet,
using the alphabet song
a b/ c d/ e f/ g /h i/ j k
/l m /n o/ p/etc…
\/ \/ \/ \/ \/
0 0 0 0 0
\/ \/ \/
0 0 0
\/
0
Note: bigger pause when asked next letter when it is in the next chunk!!!
Chapter4 – knowledge representation
retaining the original event, b/c of perception.
Meaning-based knowledge
representations: representations that retain the meaning, but
not the exact detailed perception
Verbal
-In verbal sentences, we have a tendency to extract the meaning out of what we hear, and forget the style/specific words
àwe do not remember specific wording styles unless especially cued to do so
àyet styles memory is weaker than meaning
memory
-testing warned vs. unwarned about recallingàstylistic diff vs. no meaning diff
-Sometimes, visual memory is stronger than verbal memory
study:
-studies show that they really
remember the meaning of the picture, not the full pic:
i.e. study:
-8 pics. In a series. Then shown other pictures
-->accept it 77% of time
-->reject it 60% of times
-->reject it 94%
of time
conclusion:
meaning is of the essence!!!
-evidence that first, perceptual
detail is perceived, then when image is interpreted for meaning, the
perceptual detailed are quickly forgotten.
Study: picture shown, then reversed and shown
-->w/ time, more errors
in perception of picture, but still high % of memory of meaning of picture!
Conclusion: memory for
details [i.e. phrasing of sentence] is available initially but then
quickly forgotten whereas memory for meaning is retained
Implication:
Mnemonic technique: making a meaningful imagery of some sort to enhance memory of something hard to remember
Prepositional representations: the meaningful structure remains after the perception details are gone
-->preposition:
smallest unit of independently standing assertion.
i.e.: Lincon,
who was president of the USA during a bitter war, freed the slaves
-each of those simpler sentences must be true so that the whole sentence is true.
-->studies show that
we do not remember the wording of each simple sentence but rather the
meaning of them
propositions about how we notate them in memory:
-->(president-of: Lincoln, USA, War)
study shows: we’re
can remember which ones we encounter, we’re insensitive to actual
combination of propositions
-people might remember the
propositions that they encounter but not the combination of propositions,
which means that they might add a proposition that they really didn’t
hear.
propositional networks:
propositional network:
how the propositions are tied into each other.
USA
object
war time agent
Lincoln
relation agent
president of object
Subject relation
Relation freed
bitter
‘Lincoln who was president
during a bitter war, freed the slaves’
nodes: the propositions, relationships, arguments of the sentence
Links:
the arrows, b/c they connect the nodes
-could be bigger networks(see
p. 150 imp!)
conceptual knowledge
-significant abstraction away from the original experience
àdeletion
of perceptual details but retention of important relationships
àmade from general categories from specific experiences
àwhen
I say/hear dog, I know what he does/H.m. legs/etc…
-important for things like
prediction: if I say dog, the other person knows exactly what I am talking
about… saves words.
How are categories formed?
2 theories:
Semantic networks:
isa links:
linking nodes for 2 categories:
Animal Has skin
Can move around
Eats
Breath
Bird has wings Fish
can fly
Has
feathers
Canary can sing ostrich long/thin legs
is yellow Tall
1) more frequently encountered fact àit will be stored, even if it belongs to a more superordinate concept
3) facts that are not directly
stored w/I a concept = takes more time
-> more isa links
= more reaction time
i.e. cannery - can sing
= less time than canary has skin
schemas
schemas: representational structure of a thing
isa: building
parts: rooms
Materials: woods/brick/stone
Function: human dwelling
Shape: rectilinear/triangle
size: 100-10,00
sq. feet
Slots: attribute variables: i.e. size/function/etc…
àdefault
values: typical representation of the schema category
propositions:
what specific things have whereas schemas: what things
generally have in common.
àisa
slot: points to the superset: to the thing above it
Schemas can also be generalized:
its parts could also be said to part of the superset
-we can understand constraints
(i.e. house underground has no windows) or exceptions (house w/o roof)
psychological reality of schema
-object is understood to have same characteristics of its schema unless specifically contradicted.
àmemory
is enhances and even sometimes distorted to incl. things in schema
degree of category membership
-schemas allow for variation w/I the group
àrange
of more to less typical things in the schema group.
-even w/o default values àcan have a schema àthough much disagrement
ài.e. not clear whether stoke/happiness is a disease.
àhard to define disease!
àmany
disagreements.
Events schemas:
Scripts: stereotypical things that are supposed to occur/be done
-->people recall parts of
a story that really didn’t occur, but was part of their personal script
for that event
-->if order of script oddly
reversed, then when recalling story, high tendency to recall them in
right order -->shows strong tendency of the scripts
-->b/c of schemas, some events are left out in stories, yet assumed to have occoured (like writing a check.
Abstraction theories: make an abstract concept out of the specific cases/experiences
-->i.e. schemas -->prototype = the generalized concept
Instance theories:
remember a specific case(s) w/ generalizations made from it -->no
specific concept -->comparing the specific case to the paradigm case
-4 symptoms gives, for people to judge them for 2 hypothetical diseases – 1 is 3 times more common than the other
4 X 2 = 8 synapses being learns
-->more common = higher probability, since more synaptic connections
-temporal lobe damage = recognition
of specific things = related to their subcategories
chapter 6
ebbinghaus: taught himself a series of nonsense consonant-vowel-consonant combinations
àhow long it would take him to learn
13 syllables like that twice w/o error
Percent of time saved ↑ / days of practice à
Short term memory: the kind of memory where the iconic/auditory memory enters, in affended to.
àif rehearsed, it could go into the relatively permanent LT memory
àif not moved into LT memory, the memory is lost. (since new info pushes the old info out
Memory span: capacity of the specific kind of memory (ST/LT)
àusually 7/8 digits
-evidence of ST memory as processing
station for LT memory (beyond rehearsal) = more time is ST memory =
more LT memory
-others argue = not time in ST memory but rather depth of processing:
à depth of processing:
only If rehearsed in a meaningful way
problem:
evidence:that ST is
not necessarily station to LT àinfo could go directly from sensory
input to LT àas
long as we process the info in a way conducive to creating a Long term
trace
(experiment: passive rehearsal
for 2/6/18 secs àshow no relationship to time 11/7/13%)
--
-some speak of a memory span or working memory
others (badderly) speaks of
a articulatory loop: we keep however much info we can
rehearse in a fixed amount of time (usually 1.5-2 secs.)
-both have 5 words, but # 2
longer to say: more % of mistakes
badderly
we have
i.e. if given Question 37 X
28 àyou
might visually do it or verbally say it. Central command keeps things
like task (i.e. multiplication)
frontal cortex:
plays a role in working memory.
i.e. if monkey shown
2 options: one of them, food is hidden. Then eyes covered for 10 secs.
then monkey found the food. If frontal cortex lesions: couldn’t find
it.
àmonkey’s
brains are not identical in areas but close
Activation/LT memory
Theories (similar to each other)
ACT = memory traces b/c active w/ associated concepts are presented
SAM = using contextual
cues used to make images (i.e. memory traces) more or less familiar
(active)
Activation: determines probability of access and rates of access
àwhen
we think of something, it is very active! activity decreases w/ time
-retrieval time
therefore increases w/ delay
2 factors:
àlittle
difference w/ short time-delay effect b/w practice and non-practice,
but w/ long-term delay, the more practice = faster retrieval
spreading activation
Spreading activation: the idea that activation spreads along paths of a network
ài.e. dog connected to the concept of
bone
Therefore, priming:
the setting into awareness of a certain connection/network b/w things
i.e.
- if said
-whereas control group: (no priming)
associative priming:
priming of knowledge that occurs unconsciously.
i.e. faster to recognize that what is seen is a meaningful word if: the 2 given words are related, rather than non-related or 1 nonsense
-activation determines how accessible that memory is: activation level could change quickly (w/ milliseconds)
àTraces could also lose activation
Strength: each time we use a memory trace, it increases a little in its strength.
àdetermines a bit how active it is, and thus how accessible it is.
àmore gradual than activation
àrate of improvement decrease (in % improvement) w/ time
Power law of learning: memory performance improves as a function of practice àin a logarithmic rate
àlinear decrease in response time
long-term potentiation:
more responsiveness of a neuron b/c of bast stimulation
àoccurs in hippocampus/cortical areas
àmore practice àdiminishing increase (% change upwards)
àLTP
measures changes w/ b/h measures (through a power law)
àLTP
= neural learning.
Factors infl. learning:
-not only what kind of practice but how we process it.
ài.e. depth of processing/looking for
meaning.
Elaborative processing:
adding additional info in order to remember better.
i.e. if given either ‘doctor hater the lawyer’ or ‘doctor hated the lawyer b/c of a malpractice suit’
àlater
if asked doctor hated ___’ more likely to remember ‘lawyer’ b/c
of additional info in memory being a cue.
note: closer to info: better elaboration:
i.e. fat man saw the sign:
1) the was 2 feet tall’ 2) that said ‘be careful, ice’ - #2 is
stronger àmore
directly connected)
techniques for studying textual material:
-if you have a pre-set cue –i.e. questions that force you to think through the text you’re reading
àbetter learning
àthis
technique helps b/c it sets prior organiziors (of info) in mind
study: question making
= most learning process àthen, 2nd best is reading
to answer questions àthen just reading
àreading
w/ q. in mind àbetter
than answering q. afterwards.
conclusion of idea: 2 things help learning
meaningful vs. non-meaningful elaborations
-more elaborate processing
will result in better learning even if learning less meaningful material
i.e. study:
-more memory of the upside down sentence: more processing, regardless of meaningfulness
study 2:
-level of processing and not
the intent (or lack of intent) decides the learning.
i.e. if people processed the meaning of words: more learning than just checked for a specific letter
ài.e.
therefore, people remember novel better than a textbook ànovel
invites elaboration
àthough in intentional learning: person might actively take more steps in order to achieve learning
Flashbulb memory: something special/traumatic/etc… that happened to you, or to something that you feel related to àthat you remember for long time in great/accurate detail
àvs.
things like something big, but that wasn’t so related to youàtherefore
you remember it is less accurate/sometimes distorted details
reference effect:
people rememborer more information that refers to themselves. (also
for people that ne feels close to
àpossible
b/c we’re more able to elaborate on those things!
-memory for verbal material
is assocated w/ greater activation on the L. frontal cortex/L. hippocampal
region
Note: until we
spoke about issues of encoding àin memory, we’ve also got to deal
w/ retention/retreival
chapter 7
electrical stimulation
of temporal lobe: brought forgotten, that normally couldn’t be retrieved,
i.e. memories of childhood: portions of the neural network that spreading
activation couldn’t reach.
-in forgotten memories àsomething is still retained.
-.i.e. in experiment of remembering pairs of #/word
àrepeated the forgotten ones vs. changes to the forgotten one àremembered much better the unchanged pairs that they previously forgot
àif
we get a sufficiently sensitive measure, we can find the forgotten memory
conclusion: even when
people think that they forgot something, sensitive tests can fid evidence
of some of those memories
Retention function
-with delay, we can recognize less and less a thing we saw
ànegatively accelerated: the improvement % slows down
ài.e. first day, improve 50%, then next
40% etc…
Power law of forgetting:
retention function is generally power functions àmore delay, more forgetting, but at
a decreasing rate
àlearning
also has a power law, but of practice function w/ diminished improvement
w/ practice. Retention function shows diminished loss w/ delay
-In ST memory, first few sec. most memory, then fast decrease until stable àdorsn’t really prove existance of ST memory b/c all retention functions are similar in that regard. (whether seconds or days)
àthe diff. is the degree of practice (encoding)
àmore/better encoding = more memory
-the decay in strength of memory
(level of LTP) =in strength of synapses
Bahrick experiment:
-memory decays w/ time in a
consistent pattern (regardless of h.m. encoding/depth of processing).
Then steadies off. small decline at end (b/c old age)
conclusion: strength
of memory trace decays as a power of a power function of the retention
interval
Decay theory:
as seen above, the theory which claims that memory decays through decaying
of neural networks
-competitor to that theory: interference theory
-another thing which strongly impacts memory is interference: interfering of material
àhard to keep multiple associations of the same stimuli
-i.e.: word/# pair list
remembered. then (experimental group) word paired w/ other # forgot
the original pair more than the control group who learned other word/@
pairs
àsometimes
hard to learn new fact about something -though other factors help us
do indeed learn new facts about a concept
fan effect
-when we give people a stimulus (i.e. the word cat), activation spreads to all of its associated concepts.
àmore associated items àless activation to each part
àmore
facts you have to remember about a thing: more retrieval time
experiment: give people 4 items to remember:
-more facts about a person = more delay in recognition
àmore time to spread activation to the
whole network
-(see bottom of p 209)
i.e. lawyer in the park
conclusion: more facts
associated w/ a concept = slower retrieval of any of those facts
àthis
slowing down is called fan effect (b/c ore fan of facts)
Interference w/ pre-experimental memories
-people give:
-Fastest reaction to the true facts (#2). Second fastest reaction is to the experimental true facts (#1) and then slowest response time to false memories (#3)
àmore
facts to remember (whether false or not) also increased the reaction
time
conclusion: material
learnt outside lab can be interfered with by material learnt in the
lab
Interference and decay
2 things we saw that produce forgetting:
-some think that decay is really interference
àb/c people remember more is the delay
is sleep = less interference, but still same amount of time of delay
to allow the decay
àbut really the key is time of day in which the learning took place:
-learn at night/evening
àat highest arousal time
conclusion: also decay
and also interference takes place
Interference and redundancy
experiment:
3 groups: each given something else to remember:
Recall speed (frm fastest to slowest:
àdiff. in recall time increased w/ time
àhere
interference is eliminated w/ more irrelevant facts and increased w/
more relevant facts
Conclusion: learning
redundant info about something is not interference and may even help
memory
Retrieval and inference
-i.e. in the irrelevant facts case: if they could not retrieve 1 of the facts, they had a chance of retrieving other unrelated facts about the person/thing/concept
àmake inferences
experiment:
àpeople
made inferences as part of their retrieval w/o knowing that they are
inferring and not fully remembering
inference as a bias for remembering text
-if given a text, and then
the same text w/ another label 9i.e. another name) people well infer
diff. things based on knowledge of the specific person
example: story of a stubborn kid who at age 8, the parents couldn’t find an appropriate mental institution
àif
told it was Helen Keller, and then re-tested, they would infer that
they also read as part of the story something about her being dead/dumb
and blind, even though it didn’t appear in the story – they just
inferred it.
-evidence for increase increasing
inference w/ decaying memory
àpeople do the inferences during the test ài.e. if told that the person was really Helen Keller at time of test
àstill made the errors
conclusion: people use what they remember to infer what might have been part of the text
Plausible retrieval-we retrieve info that is likely meant yet is not explicit
example
-manager of a fries store.
he loved them. he thought that they were delicious –he got them for
free.
Assumption: he got them for free from the store!
subjects given: incorrect,
2) plausible 3) explicitly written(exact recall) sentence
-with time, exact recall sentences had longer reaction time since it got weaker, whereas plausible sentence had less reaction time:
àless/weaker
explicit trace memory àplausible (logic) system takes over
(note: which is embedded into system ànot dependant on memory)
other study
-more facts about a person
= less reaction to a plausible statement, yet longer reaction time to
explicit fact about him àdoesn’t depend on retrieving a certain
fact.
conclusion: people try to judge plausibly rather than retrieve exact facts
-elaborate processing doesn’t
only increase memory but also memory of inferred things.
-i.e. story about Dr. who,
after the checkup, tells college student that what he thought was confirmed
àfor half the people: first told: ‘that she thinks that she’s pregnant and she doesn’t know how to tell the professor she’s seeing that she’s pregnant.’
àthose
people assumed that she is unmarried and had an affair w/ her professor
-we expect the in the intro group to have more inferences about the inferred theme
àInterestingly
enough, they also remembered more the propositions that were actually
given: the intro served as an elaboration
-those inferences that are
really not in the text are not errors!. i.e. in real life (i.e. exam)
we expect them to remember those kinds of inferences as if it was read
material!
Advertisers also capitalize on those inferences.
àsounded like Listerine fights colds,
but the actual phrasing was ‘cold-proof’
Memory Errors
-sometimes, it is logical to infer, but sometimes it is important not to, i.e. in eyewitnes testimony
àstudies show that they are quite inaccurate, despite having lots of weight by jurors
àpeople
confuse what they actually observe from the incident w/ what they heard
from other sources (i.e. order of events/etc.
example: people asked what speed the car was traveling when it passed a yield-sign
àthey
answered, yet, there was no yield sign to begin with!
False memory syndrome:
many false memories induced by psychotherapist (i.e. of past sexual
abuse)
àsometimes,
hard to distinguish b/w memory and imagination
conclusion: serious
errors of memory could occur b/c people fail to separate actual experience
w/ inferred/imagines stuff.
-sometimes, a blocked memory
comes along when we give it prompts.
study: given animal
names. if paired w/ the group it is from (i.e. mammal), more recall
organization and recall
-giving a mechanism for cueing individual pieces of info (i.e. organization)
ài.e. remembering things like lists/orders into a hierarchy helps
Metals stones
Rare common alloys Precious Masonry
platinum aluminum bronze Sapphire Limestone
silver copper steel Emerald
gold lead brass Diamond
Method of Loci: using a fixed sequence of locations to cue retrieval of memories
ài.e. making the campus building into the locations and each one, changing its shape to resemble one of the shopping list items.
-other ways of hierarchy organization also help
-context acts as a cue since
it is recorded into memory- trace that records the memory
i.e. physical context/mood context
-i.e. what teacher/where
àstill,
not always is contextual info saves… depends on h.m. degree of contextual
cues is integrated into memory
mood congruence:
easier to remember happy memories in happy state/sad memories in sad
states
state-dependant learning:
easier to recall info is return to physical/emotional state that they
learnt the info in.
i.e. alcoholics can’t
remember where they hid the alcohol, when they’re drunk
--.yet it has a debilitating
effect
conclusion
effects of other material on the context
to-be-learned
material infl. by other ‘to-be-learned’ info
ài.e.
if told to remember the 2nd word, and then told ‘sky- blue’,
remembered more often when the word ‘sky’ was in the context of
the other learned word in the pair. (even when only tested for the 2nd
word
steps of experiment showing contextual cues as helpers to learn:
-recognition is stronger than recall. can’t recall if you can’t recognize it.
àif
closer to original encoding context = better memory
Encoding-specificity
principle: better recall on test if similarity of encoding
to original encoding
conclusion: better word
memory if words are tested in the context of same word that they were
studied in
Hippocampal formation and amnesia
Amnesia: memory
loss, due to brain damage àin hippocampal formation
-hippocampal formation: critical
for memory formations
àdamage
to it = prob. learning associations: especially of diff. elements
àhippocampal formation: in temporal
lobe
reasons for hippocampal problems
2 kinds of amnesia
-sometimes blow to head = not a permanent damage:
àbut still has a set pattern of recovering
Set pattern:
1)distorted memory until 2 yrs. bf. accident àthen total retrograde amnesia (RA) until coma) after accident total anterograde amnesia (AA)
àthe
lost memory in #3 = ‘residual permanent memory loss’
-RA and AA recover together,
even though 1 might be more severe than the other
interesting to note:
Implicit vs. explicit memory
Dissociated memory: memory that are conscious
implicit memory:
knowledge that can’t be described: keyboard
dissociations: the contrast/diff. b/w explicit and implicit memory ài.e. keyboard typing is one example
àmemory
tests showed diff. results = therefore must be the diff. kinds of memory
-Amnesiacs show total dissociation:
might be able to learn a list of words àcan’t recall them, but could put
them in onto a word-completion task, but are unable to gain conscious
access to it why they did it.
conclusion
-implicit ways to show that
amnesiacs have some sort of memory for events
implicit vs. explicit in normal subjects
-certain variables infl. diff:
implicit/explicit memory
i.e. study: word learning
-no context (of other word) àcontext àgeneration
of context
imp! Implicit memory
decreased w/ elaboration while, as expected, explicit memory increased
w/ memory
àpriming:
mostly in implicit, non-context situations: we rely most on perceptual
encoding
-no diff. in recognizing meaningful vs. non-meaningful words
àwhether explicit or subliminal (implicit)
conclusion
-elaborative processing facilitates
explicit memories but not implicit memories
procedural memory
procedural memory: knowledge of how to perform certain acts
àkind
of implicit memory (i.e. memory w/o conscious awareness)
study: amnesiacs had to keep switching the # of workers in order to maintain sugar output b/w 8000-10,000
àcouldn’t
find a rule, yet felt they did it correctly/ ‘intuition’
àcan
learn how to implicitly control the fctory w/o explicit knowledge
requant distinction:
Declarative knowledge: explicit knowledge that we can report/consciously aware of
procedural knowledge:
knowledge of how to do things. usually explicit
conclusion: peple can dev. good ways for doing tasks w/o explaining what they are doing
expertise: could only be achieved w/ practice
general characteristics of skill acquisition
3 stages of skill acquisition
2 elements:
power law of learning
-we saw in chapter 6 àretreival of simple associations improve according to a power law
àso does complex associations, like
driving.
logarithmic scale
-retention of learnt skill is retained even w/ years of abstinence
àmight need a short warm-up
conclusion
-performance of a cognitive
skill improves as a power function of practice. only shows modest declines
over long retention intervals.
Nature of Expertise
-many expertise studies work
on comparing people at diff. levels of then same domain of expertise
(i.e. medical students vs. stage vs. experienced dr.)
proceduralization
proceduralization:
the switch from declarative o procedural knowledge
-degree of relying on declarative vs. procedural
1) processing:
àno declarative info brought into working memory
àno forgotten items needing to be re-retrieved
Tactical Learning: learning a series/string of operations
àspeeding up = change from computation
to retreival
-at first àprefrontal areas (anterior cingulate
gyrus area àa
medial frontal area àmedial =in interior of cortex)àthen
shifts w/ better learning t/w back areas of frontal areas. at first
cerebellum [motor learning], later hippocampus [learning]
strategic learning
strategic learning: learning to organize one’s problemsolving techniques
-at first backwards thinking: first thinking about the goals/subgoals, and then the stratagy to get each one of the steps.
-later àforwards thinking: thinking about what
is missing: already able to not think of each missing/each subgoal
àgood for things like geometry/physics but not for things like programming, where one needs to go through backwards reasoning all the time
àbreath for experts/depth for novices
Conclusion: each domain ha sits own way of acquiring problem-solving techniques
Experts: -richer perceptual features for encoding problems
ài.e. use deeper categorizations of problems/more definitive rules (oi.e. X ray = not tumor, but collapsed lunge
-experts are just as good in
seeing ahead as novices àthey just already remember chunks of
moves and therefore can foresee many more probs.
conclusion
experts can recognize chunks
in problems which are patterns of elements that repest over problems
long term memory and expertise
-also LT memory improved w/ expertise ànot only Working memory
note: this is b/c expertise
helps chunk info into more convenient chunks
Deliberate practice
-necessary for acquiring expertise
in anything
Ericsson: deliberate practice: motivation to learn – not just to perform: monitor performance to correct performance and try to reduce deviance
àtallet is peripheral to practice!!!
-very little transfer of skills to similar domains and virtually non to diff. domains [i.e. spellingàmath]
ài.e.
street vending kids had hard time doing paper-work math, yet could figure
out costs in head
Theory of identical Elements
Theory of identical elements: brain is not train ed by formal doctrines of general facilities, but rather specific habits/associations
àprovides narrow responses for specific stimuli
àThorndike: there is transference of
skills of similar stimuli-response elements
prob: it is not the stimuli but the abstract concepts which allow for some transference (i.e. calculus for geometry/economic prob: as long as there are common facts!!!)
ànot on surface level, but things w/ same logical structure do have transfer of skills!
àtherefore
Thorndike was right in saying that transference occurs w/ same elements:
he was just wrong in identifying which elements were imp.
Negative transfer: 1 skill inhibits another àvery rare
àexists in memory but not in skills
1 skills example: using a calculator makes you forget a algebra shortcut
Educational implications
Componential analysis: teach the subparts of the whole in order for the student to see the whole
Mastery knowledge:
makes sure the student knows how to masker all the subcomponents
Conclusion: get student to identify the underlying knowledge elements and get student to master them all
Intelligent tutoring
system: using computers to interact w/ student in order to tutor
them (i.e. LISP)
Chapter 10
Logic – formal
study of what it takes for an argument to be correct.
-intuitively, we use logic
if we have enough info
ài.e. if we know that pressing 15 on the microwave gives 15 sec. of heating, then pressing 30 gives 30 secs. of heating.
Deductive reasoning: certain conclusions based on their premises
Inductive reasoning:
probabilistically stemming from their premise
deductive
Fred = Lisa’s bro/Mary =
Lisa’s daughter ètherefore, freed is Mary’s uncle
Conditional statement: an if a statement
-If
A is true, than B
Antecedent -the if part
Consequent: the then statement èwhat is conditional on the antecedent
Modus ponens:
inference logic: if A = B and A is rue, than B follows
àmodus
ponens: therefore she got a good grade
modus tollens:
A implies B àtherefore
if B is false, then A is also false
àModus
Tollens: therefore, she understood the book
-those conclusions are not
compelling, but aren’t treated as certain either
conditional syllogisms: if A happens than B. A does [not] happen therefore X happens
[doesn’t happen]
sideways U = sign of implication A U B = A implies b. if ~A
~ = negation sign
Denial of antecedent: if A U B and if A is not true àwe wrongly think that can know B
àB is not necessarily but sufficiently dependant on A
Affirmation of the consequent: we assume that we can know the consequent (A) based on the antecedent (b)
sources of these mistakes:
-Assumption of Biconditional
àBiconditional: if and only if
ànecessary and sufficient
-other reason we choose the way we do, in logical statements = probability
-cards showing symbols
-condition: if vowel = even # on other side
-had to turn over card to show
if rule = correct
àselection
task
Cards: E/K/ 4/7
conclusion: Wason card
selection could be explained if we assume that people chose cards thy
thought were informative w/ probabilistic model
Permission schema: connective if statement doesn’t have to be logical/probable statement but also an ought statement
ài.e. if he is drinking beer, he must
be over 19
note: must be familiar w/ the rule: i.e. therefore know what/why it is there. If person is unfamiliar w/ the rule, or its rationale, he will not know how to apply it.
àExperiments might also give rationale
behind the rule.
Permission schema
-Sometimes we use ‘if’ to say what has ought to be.
àif he is drinking, then he must be
over 19’
àlogical
use of permission schema is based on familiarity w/ the concept of permission
[vs. a senseless rule] and not w/ specific rule
conclusion: in detecting
social contract was violated, àlarge proportion of logically correct
choices
conclusion:
-if has many diff. interpretations.
for example:
-training in logic doesn’t
necessarily make people us it!
reasoning about quantifiers
logical quantifiers: i.e. ‘all’ or ‘some’
-sometimes when we say ‘all’ we mean most/probably
à’all philosophers read books’
categorical syllogisms (early logic research)
All A = All B
All B = All C
Therefore, All A = C
Some A = Some B
Some B = Some C
Therefore Some A= some C
(truth: Therefore some A doesn’t
necessarily mean some C)
àOften, people accept false syllogisms, like #2
-we’re not totally indiscriminant!
Atmosphere hypothesis: the logical terms (some/all/no/not) create the answering tendency ->
Conclusion: people are
biased to accept conclusions w/ same quantifiers as their premises
Process explanation
Mental model theory-people make mental images of the situation and sometimes fail to consider alternate images that might actually be more reflective of the situation
àpeople don’t treat the task as a logician’s task (i.e. sometimes, probabilistically)
Inductive reasoning:
to try to get most probably ànot certain
Bayes’s Theorem:
combining the different probabilities to come up with an assumption
(using mathematical formula)
Prior probability:
less likely the hypothesis was b/f the evidence, so too, it will be
less likely after the evidence (i.e. b/fàlow statistical chances of being burgled
b/f the event, so too, chances are low now)àalternate is higher!
Conditional Probability: type of evidence is true if the hypothesis is true; i.e. if I believe that if the door is open (i.e. evidence) then high chances of being burgled (hypothesis)
Posterior probability:
the probability that the hypothesis is true, after considering the evidence.
Prescriptive model: (normative model) – evaluating the probability of a hypothesis
Descriptive model:
what people actually do
Base rate neglect
-people don’t take prior probability in account enough
à
paying too much attention to the evidence
Conservatism:
paying too little attention to the evidence
Correspondence to Bayes’s Theorem w/ experience
probability matching:
people choose among alternatives according to success.
People’s processing of probabilities
correspond to Bayes’s theorem, their experience-based b/h doesn’t
no
Judgments of probabilities
-despite normative value, ($
[value] X odds) people choose the bigger odds, yet smaller values
subjective utility:
value
àtakes
double the amount to double the utility ($ x odds)
conclusion: people make
decisions under uncertainty in terms of subjective utilities and subjective
probabilities
framing effects
framing effects:
people’s b/h changes according to what they think the utility
curve
i.e. a discount from $125 to
$120 is less worth a discount from $15 to $10
chapter 11 -linguistics
linguistics:
attempt to characterize nature of language
Productivity:
the fact that in any language –infinite # of utterances
Regularity: systematic usage of the language utterances
àvery few combo of words really acceptable
Grammar: set rules of linguistic
3 kinds of rules:
linguistic intuition: judgment of language
àimplicit knowledge
àalso deals w/ ambiguity:
ài.e.: ‘they are cooking apples:’
-are they cooking
the apples or are the apples made for cooking
àjust
b/c we have competence (knowing the rules) does not mean
we perform them right
phrase structure: to recognize a sentence
SENTENCE
Noun phrase Verb phrase
Article adj. Noun verb phrase
Noun phrase
Article adj. Noun
The brave dog saved the
Noun phrase Verb phrase
Pronoun Verb Noun Phrase
They are cooking apples
Versus:
Noun phrase verb phrase
Pronoun verb adj noun
They are cooking apples
Rewrite rules: node’s labels àtheir categories
àcould be used to rewrite sentences
Pauses structures in speech: intuitive use meaningful phrasing units
àtendency to use smallest bundler of
words
speech errors
-tendency to correct whole phrase when error is detected
àpoint: sentence is a psych reality
àonly
a phrase of the sentence would be repeated as correct
spooner: mixed up sounds to make them funny
àusually
happening w/I a phrase
examples:
Transformation: moving elements w/I a sentence (from normal position) in more complex sentences
àsome restrictions apply
relationship b/w language and thought
behaviorist proposal:
Watson: everything is just a conditioned response
àsome
show that role of mental process àbut Watson thought that the min dis
just talking to itself
study: thoughts
are really non-motor/sub-vocal activity: not just speech!
-therefore, thought is
not language!!!
Behaviorists: thought =covert language/other motoric action
àproven wrong
whorfian hypothesis of language determinism
language determinism: language determines how person sees world
whorf:
impressed at how diff. languages emphasize their word structure
àDani were taught arbitrary names, and they chose the paradigm color as well as English speakers!!!
àbrain is maximally sensitive to some
colors!!!
àperhaps
thought infl. language and not vise-versa?
Conclusion:
language helps us communicate ideas àbut does not infl. our thoughts
Is language infl. by thought?
Conclusion: in
many ways, the structure of the language does reflect the structure
of how our minds process the world
Examples:
àconclusion:
thought infl. language
modularity of language
modularity:
approach:
language and thought are independent
àlanguage functions separately from other cognitions
àused to communicate, yet is not infl. by general cognition
àfor
incoming info: linguistic analysis àthen passed on to general cognition
questions
-grammar = implicit
à10 yr-old knows tons of rules
Issues of rules/case of past tense:
-do they learn a rule (i.e.
add ed to make is passed) or just associate the two (kick+ed)?
àthen learns irregulars to the rule
connectionist model:
Inputs: verb àgoes
though network layers until right verb output is produced (see p.378)
i.e.: go though
àphonological rep of past tense
aphasia:
-PET studies of unimpaired regular people also show:
àthough
debate whether regulars are learn through rule based or through association
base
quality of input
-kids don’t get instruction
on their first language/parents regularly don’t correct speech!
Mothers: tend to speak in clear/short sentences and w/ clear/exaggerated intonation
àeven
though not all mothers do this, all kids learn!!!
Conclusion: instruction
quality is not critical to the learning of speech
-Deaf kids = make up sign language!!!
àconclusion: deaf kids master language
w/ little direct instruction, at young age!
critical period for language acquisition
2-11 yrs: easiest to learn
àolder kids (i.e. above 11) learn faster
(initially)
àyet lack fine mastering of fine point
i.e.
ài.e.
speaking w/o accent
study: at
age 10 – beginning of loss of ability to master a language/proficiency
àsame
w/ 1st languageàdeaf adults learn sign language slower
than youths
language universals:
chomsky: we have innate knowledge of limits of possibilities of
characteristics of language/grammar
àlanguage
breaking those rules is unlearnable: i.e. adjectives and nouns separated
by too much/hold word for too much!!! (language who’s rules could
be learnt = natural language)
Critics
of chomsky: maybe it is cognitive prob. and not innate?
Transferring part
of sentence: arbitrary constraints or the movement of parts
of a phrase
Parameter setting:
àlearning
the language is setting its unique parameters
àchildhood mistakes = mixing up rules (i.e. dropping pronoun in a non-pronoun language)