intro to cognition midterm 2

- a process that enables continued cognitive processing
- ability to focus on specific stimuli or locations

- rapid shifting of attn. usually caused by a stimulus such as a loud noise, bright light, or sudden movement

(1) limited: not all sensory stimuli simultaneously get continued processing

(2) selective: attn. must be selective exactly bc it is limited (can't attend to everything)

sometimes you're attending to the wrong thing

stimuli → processing physical characteristics → processing semantics → awareness

- everything is processed for physical characteristics (filter before semantics)
- what you attend to is processed for meaning and to awareness

- everything processed for physical characteristics and semantics
- selection doesn't occur until messages are processed enough to determine meaning
- what you attend to goes into awareness

- 2 diff messages; asked to shadow what is heard in attended ear
- don't remember words from unshadowed
- don't notice if speech is played backwards or if lang. changes
- DO notice if a pure tone is played or there's a gap

procedure of repeating what you are hearing (in dichotic listening task)

- ability to focus on 1 stimulus while they filtering out others nearby
- monitoring background noise to see if anything demands attn. (testing the name works, but only 1/3 the time)
- early filter says this would never work

- mixed up sentences; switch to unattended ear when sentence switches then go back
- evidence for late filter bc how could you switch between attended & unattended at the right time if you weren't picking up semantics

(1) classical conditioning; shock associated with 3 city names (train until GSR response)
(2) dichotic listening: shadow irrelevant message from one ear - the other ear gets words, plus occasional city names

- results: participants showed the GSR response most to the old city names (38%), then 23% new city names & 10% new nouns
- conclusion: unattended material is evaluated for meaning (all stimuli are evaluated for their semantic content)

- attn. resources are limited so w/ low-load primary task, you have remaining cognition resources to spare (might be able to process at semantic level)
- high-load (demanding): difficult, not as well practiced so use up processing capacity (no cog. resources remain)

- N with circles vs. N with other letters
- N with circles = low-load task = more cog. resources to spare, more easily distracted (by spongebob image)
- distraction less likely for high-load bc no capacity remains to process distracting stimuli

- demonstrates how a powerful task-irrelevant stimulus can capture attention → reading words is highly practiced & has become so automatic that it is difficult NOT to read them

- physically salient stimuli attract attn. (saliency map)
- meaningful (cog. factors) attract attn. (schemas; look longer at out-of-place)
- scanning based on task demands ("just in time" strategy)

- faces are socially important & it's hard NOT to attend to them (change your place or space)
- habituate: continued exposure no longer demands attn. (effective for stimuli that don't change much aka static; complexity ok)

- moving eyes
- visual search (harder when it's a configuration of visual clues; i.e. waldo)
- you have a specific thing in mind that you want to attend to
- searching by color is relatively easy (much harder by meaning)

- rapid, jerky movement from one fixation to the next

- NOT moving eyes
- attn. as a spotlight that enhances perceptual processing (metaphor only partly correct)
- "attn. beam"

- cross on screen --> arrow in direction to direct attn --> delay --> go signal
- when cue is valid, RT fast
- when cue is invalid, RT slow

- neural evidence shows that damage to diff parts of the brain support 3 separate processes: disengage, move, engage

- posterior parietal cortex
- valid trials = okay on either side
- invalid trials = really bad on contralateral side; difficult to disengage attn.

- superior colliculus
- valid trials = very slow to show advantage of cue (but do show it)
- invalid trials = slow but not a special problem (at short delays, cue is not used)
- don't show cost or benefit

- thalamus
- valid trials = don't use cue much
-invalid trials = don't use cue much (easily distracted by irrelevant stimuli) -you can't really see an advantage or disadvantage bc response times are always really slow bc attention is never engaged...it's always drifting.

- couple of predictions of metaphor do NOT work
- moving attn. greater distances doesn't take a longer time
- moving attn. isn't sowed down by intervening stuff
- beam implies attn. selects space

- can attend to just 1 word, based on color, even though another word overlaps in space
- implication: when you switch, you are switching not to a location in space, but to another object or object part
BUT
- diff parts of parietal lobe are maximally active during b/w objects shifts & w/n objects shifts
- implication: possible that both mechanisms are present

- little or no attn. cost
- w/ or w/out intention
- no affected by motivation
- can happen w/out awareness

- attn. cost
- occurs only w/ intention
- affected by motivation
- only w/ awareness

- initially huge interference in complex tasks but after continued practice sessions subjects developed automaticity

- younger people are better at multitasking but it doesn't follow age, follows WM capacity
- it's not practice, it's working memory

- high multitaskers have worse attn. control
- switching NOT sharing

- failing to notice changes in the environment
- attn. in necessary for perception
- i.e. movie continuity

- sometimes you know what you want to attend to, but you're not currently attending to it, so you need to redirect attn. to appropriate target

- the process by which features (color, form, motion, location) are combined to create our perception of a coherent object

input --> filter --> detector --> memory
- attended message is separated from incoming signal early in the analysis (all other messages filtered out)
- filter identifies message being attended to based on physical characteristics & sends it to detector to process it for higher-level characteristics (meaning) before sent to STM

- analyzes incoming message in terms of physical characteristics, language & meaning
- analysis of message proceeds only as far as its necessary to identify the intended message
- "leaky filter" model where the attended massage emerges at full strength (unattended messages are present, but weaker)

- failing to see visible objects when our attention is directed elsewhere

- objects are analyzed into their features & their attn. is necessary to combine these features to create perception of an object
- binding occurs by proposing 2 stages of processing
(1) preattentive stage: objects analyzed into separate features (independent)--> illusory conjunctions
(2) focused attn. stage: free-floating features are combined so we can perceive the object (eliminates illusory conjunctions)

combinations of features from different stimuli

search for combination of two or more features in same stimulus

- covert attn. to an object or location enhances brain activity associated with the object or location
- topographic "attn." map

consistent

- sensory memory demonstration: letters presented --> blank screen --> tone indicates which row of letters to recall
- showed people can see & recall
- measuring the capacity & duration of sensory store (sensory memory fades fast)

- large capacity (up to 20 characters)
- physical properties; little semantic
- rapid loss through decay or masking

3-stage memory model that divides memory into sensory, STM & LTM

STM: storage
WM: storage + thinking

- window on the present
- holds 5-7 items for 15-20s
- sensory

- viral encephalitis: damage to hippocampus
- problem: storage, can't get anything from STM to LTM
- had extreme anterograde (can't encode new) BUT also lost some retrograde

- phonological store: stores about 2s (limited) of auditory info (from the environment)
- articulatory controlehersal process: info enters store from this process of talking to yourself (prevents decaying)

- limited capacity
- temporary, sound-based storage (~2s)
- must be attended to periodically & rehearsed in order to be preserved

- impaired serial recall for items similar in sound (issue is pronounciation not # of items)
- acoustic info is the storage code in the phonological loop

- serial recall of sequences declines as the length of time it takes to say each word increases
- memory for lists of words is better for short than long

- acoustically
- visually
- semantically (meaning)

- where you store visual or spatial info
- similar to mental imagery

- attentional controller that selects & regulates the flow of info within working memory
- helps maintain info WM stores & coordinates other cog. processes
- decides how to divide attn.

- individuals with stronger working memory are better able to focus & ignore distractions
- better self regulation & impulse control

- process involved in retaining, retrieving, and using info about stimuli, images, events, ideas and skills after original info is no longer present

- continued perception of a visual stimulus after it is no longer present (i.e. sparkler)

- persistence of vision
- duration of visual sensory memory is < 1s

- persistence of sound
- auditory sensory memory is 2-4s

- info learned previously interferes with learning new info

- new learning interferes with remembering old learning

7 ± 2 items (capacity of STM)

- more recent estimates 4 but can be expanded by chunking

- small units can be combined into larger meaningful units
- using prior knowledge to group items together (strong associations)
- meaning matters for memory

- limited capacity system for temporary storage & manipulation of info for complex tasks such as comprehension, learning, and reasoning
- sustained activation of representations
- behaviors that depend on WM can be disrupted by damage to the prefrontal cortex

- repetition of irrelevant sound reduces memory because speaking interferes with rehearsal
- overloads the phonological loop

- reaction times longer for greater differences in orientation
- rotate image of object in mind to solve the problem

brain response that indicates how much space is used in WM as task is carried out

- component of working memory where WM info interacts with LTM info
- makes interchange/connection so greater storage capacity is possible

- process of getting info & transferring it into LTM
- phase of experiment when info might go into LTM

- imagery
- emotion
- repetition
- thinking about meaning (depth)
- effort/desire to learn

- emotion at encoding boosts memory
- BUT maybe it's not the emotionality of the object, but some other property (i.e. guns attract attn. bc they're most likely to be "active in a scene"

- slideshow about kid visiting his dad (surgeon) at work
- some subjects told dad was operating on victims of accident vs. drill operation on dolls (showing identical slides and prompting high or low emotional responses with just slight changes in the accompanying story.)
- emotional response at encoding does lead to better memory

- sheer repetition does NOT necessarily lead to memory for the stimulus
- generally good for memory, but must be the right type of repetition

- intention or effort is irrelevant
- depth has a big effect

(1) definition of levels & circularity
(2) we can't ignore what's happening at retrieval

- defining a procedure as deeper bc it results in better memory
- then using that procedure to show that deeper processing results in better memory
- doesn't prove anything

- happen during very high emotion
- similar accuracy but higher confidence
- accuracy & longevity thought to maybe indicate special memory process

- when type of process is same at encoding & retrieval, memory will be successful
- cue matches what you thought about during encoding
- dangerously circular idea
- shows that deeper processing at encoding does NOT always result in better retrieval

- old memories mostly intact
- can't encode new mems

- old mems lost
- new memories encoded normally

- not all past memories are equally affected
- recent mems very affected, distant not so much
- test for public events or famous faces

- initially, mems are fragile --> must be consolidated after they're encoded

- process by which memories become more stable, even if they are not practiced
- takes years for a mem to be fully consolidated
- mems don't sit passively after they are encoded --> mems are consolidated for years after (even those not thought about)

- if you have anterograde, you always have some degree of retrograde
- parts of the brain that encode new mems are the same that consolidate them

- minimal info from experimenter --> "remember"
- context usually implied, occasionally described

- experimenter gives part of info (retrieval cues) to aid in recalling the previously experienced stimuli

- to-be-remembered info is presented along with distractors & must distinguish new from old

- learn some material to a criterion & # of trials required is noted
- later, relearn material to the same criterion (if fewer trials are required, that is savings)

- free recall is hardest --> cued recall --> recognition --> savings

savings --> recognition --> cued recall --> free recall

- different cues are more/less likely to make you think about material the way that you thought about it at encoding

- actual events
- relevant prior knowledge

- memory representation of a type of event
- characteristics generally true of event not a specific even
- things that conflict more attended to remembered

- things that conflict with schema are attended to & remembered
- at encoding: schemas make atypical things stand out & memorable
- at retrieval: schemas make it seem likely that typical things happened, even if they didn't

- process of transferring info from LTM to working memory
- before material that has been encoded can be used, it must be retreived

- rehearsal that involves repetition without any consideration of meaning or making connections to other info

- better way to establish LTM bc you are making meaningful connections

- memory (retrieval) depends on how info is encoded into the mind
- depends on the depth of processing (shallow vs. deep)

- close attention to meaning
- relating it to something else

- little attention to meaning (repetition, physical features of words)
- not as effective

- memory better if asked to relate word to oneself
- richer, more detailed representations in person's mind for better memory

- generating material yourself (rather than passively receiving) enhances learning & retention

- practicing retrieval enhances performance (better than rereading)
- involves actively creating & recreating material

- shows strength of forming visual images
- subjects who had created images remember more than 2x the words as the subjects who had just repeated the word pairs

- info is learned/encoded along with its context

- mem better when personal internal state during retrieval matches the state during encoding

- anterograde amnesia
- can't learn new things (words, stories, faces, rhythms etc.) BUT can learn motor skills
- shown by pursuit rotor task
- "diversity where there could be unity"

- pattern of priming that occurs simply bc a stimulus is presented a 2nd time
- processing is more efficient on the second presentation.

- figures with parts erased --> consecutive images come closer to full outline --> see when you can identify image
- amnesic patients can learn to identity w/ less of a complete picture when do task again

- not just that there is mem & there are motor skills --> perceptual stuff can be preserved
- ppl start to think of amnesia as "memory-some process" with that missing process impairing most memory but certain tasks don't depend on the missing process, so the tasks are learned

poor encoding?
- deep encoding --> no effect
- fast forgetting (give amnesics lots of time to study so their performance won't be at chance)
- stuff gets into memory, you just can't retrieve it ?
- researches give up on idea that there is sa single memory system missing some process

- revealed by indirect testing
- mem shown by task performance
- motor skill
- priming

- conscious memory
- tested directly
- awareness that you're remembering something
- hippocampus

- researchers gave up on idea that its just implicit & explicit largely due to separability in the brain
- motor skill, repetition priming, skeletal & emotional conditioning (source of evidence for brain basis of this learning)

- pursuit rotor
- mirror tracing
- basal ganglia

- gollin figures
- stem completion
- visual cortex

- classical conditioning with a motor response as CR
- overt response of motor system
- cerebellum

puff--> blink
tone + puff
tone --> blink

- cc w/ emotional response as CR
- amygdala

horn --> anxiety
rat photo + horn
rat photo --> anxiety

- autobio: "I remember"
- spatial & temporal context
- learned in a single exposure

- factual: "I know"
- no context
- possibly learned in single exposure but strengthened with repetition

- memory w/out awareness
- question of process purity
- tasks don't purely call on one memory
- frequently have explicit working at same time as implicit (i.e. coaching; coach told you this & I practiced enough that body will do this)
(i.e. snake phobia: explicit mem + emotional conditioning at conflict)

- objects in the world

- deciding that objects belong to category

- mental representation (characteristics) that defines a category

- enables us to recognize objects & events and to make inferences about their properties

- can decide whether something is a member of a category by determining whether a particular object emits the def. of a category

- concept = list of necessary & sufficient features
- works well for geometric objects, okay for kinship/legal but poorly for most others
- suggest category membership is all or nothing; difficult to come up with right list of features

- typical exemplars come to mind 1st
- faster to verify (better at confirming that they are a member of a category)
- affects how quickly you categorize & how you reason

- basic
- superordinatelobal
- subordinate

- most inclusive but still share most features (i.e. birds)
-psychologically privileged: what we usually use when thinking (faster to verify an object name at basic)
- special bc going above it (super) results in large loss of info
- depends on expertise

- one level more abstract than basic; less specific (i.e. animals)

- less abstract than basic; more specific (i.e. robins)

- basic level depends on expertise

- category membership is a matter of probability, not all-or-none
- likelihood of being member of category by computing similarity of the exemplar to the concept

- formed by averaging category members a person has encountered in the past (typical member=standard)
- when you see a new exemplar, you compare it to the prototype
- the more similar it is to the proto, the more confident you are that it belongs to that category
- when you see the proto, you're very confident you know its category

- have high family resemblance (things in a partiular category resemble one another in a # of ways with some variation)
- statements about them are verified rapidly
- names 1st
- affected more by priming

- ad hoc categories: create on the fly, as they become useful to you
- effects all depend on your having a prototype of each category stored in memory
- BUT you get the same graded typicality effects for categories that are much less common
- called ad hoc bc it's assumed you don't have a prototype for them but you still get all typicality effects

- lots of exemplars; can derive prototype by comparing the new to-be-categorized exemplar to the typical
- take average
- similarity & context

[exemplar] doesn't discard info about atypical cases w/in category (i.e. penguin as bird)
- deal more easily w/ variable categories
- work best for small categories
[prototypes] more important as people initially learn about categories
- exemplar better for later on

- how to select features?
- similarity depends on context
- at times we DO seem to categorize via rules (as definitional view suggested)
- shows that people are sensitive to the fact that some features are not free to vary

- person jumped in pool w/ clothing... drunk?
- most drunk ppl don't jump in pool --> doesn't occur frequently BUT when it does happen, it's likely they're drunk

(1) limitless (quantity) & rapid (speed)
(2) near miss (error is related to what you were trying to remember; not random)
(3) relevant info
(4) resistant to faulty input (concludes there is wrong/misleading cue/info but memory is resistant & gets to right thing)

(1) more books = slower search (as you get older, retrieval would be slower but not true)
(2) slight errors mean possibly big mistakes (not near miss)
(3) no "relevant info" if answer is unknown
(4) not resistant to faulty input

- near-misses & "relevant info" indicate that we need to find by content, not address
located by itself

- nodes connected by links that are labeled
- good at:
(1) lots of storage
(2)relevant info

- property inheritance: any node inherits all props above it (efficient storing)
- node activation allows you to access properties
- entry into memory from input from environment

- some of the predictors aren't borne out (i.e. mammal should be faster than animal for gorilla)
- properties don't seem to be stored only once (i.e. "flies" strongly associated with robins, wrens etc.)

- links are unlabeled & knowledge is now in the weights
- strongest link inhibits other possibilities
- allows content-addressable storage & retrieval of properties
- creates defaults (properties you assume are probably true, unless told differently)
- accounts for typicality (heavy weighted properties --> pos. feedback loop VS. inhibition)

- when only learn 1 fact, there's stronger activation than when fan of facts where retrieval is slower
- real effect BUT only observed in the lab (outside things get interconnected & organized)
- predicts that little knowledge person responds faster to simple question

- concepts are represented in networks that consist of input, hidden & output units and that info about concepts is represented by a distribution activation of these units
- also called the parallel distributed processing (PDP)

- output: receive input from hidden units
- hidden: receive input from input units
- input: stimulated by environment

how learning occurs
- network responds to stimulus
- provided with correct response
- modifies weights to better match correct response

- ability to judge highly prototypical objects more rapidly

presentation of stimulus facilitates response to another stimulus following closely in time

- concepts arranged in networks that represent the ways concepts are organized in the mind

- properties that hold for most members (shared properties) of a concept are stored at higher-level nodes

- results in the testable prediction that time it takes for a person to retrieve info about a concept should be determined by the distance that must be traveled through the network
- can NOT explain typicality effect bc cognitive economy doesn't always hold (i.e. not all birds fly, but listing that ft. makes network more efficient)

- activity that spreads out along link that is connected to activated node
- result: additional concepts that receive activation become primed & can be retrieved more easily from memory

- spreading activation can influence priming
- task: indicate word or nonword
- RT faster when words were associated (retrieving 1 word from memory triggers spread of activation to other nearby locations in network)

concepts represented by activity that is distributed across a network

- the signal that originated in the input units
- the connection weights throughout the network

units activated by stimuli from environment

- determines how signals sent from one unit either increase or decrease the activity of the next unit

- error signal is transmitted backward through the network, providing info needed to adjust the weights in the network to achieve the correct output signal for a stimulus
- AKA signal sent back to hidden units

- difference between actual activity of each output unit & correct activity
- process repeats until error signal is 0

- learning process consists of initially week & undifferentiated activation of property units w/ many errors which are then sent back through the network to change connection weights until network assigns the correct properties

- disruption of performance occurs only gradually as parts of the system are damaged

- bc similar concepts have similar patterns, training a system to recognize the properties of one also provides info about the other

all advantages of local representations plus...

(1) graceful degradation (consistent with how people typically lose memory-age, microstroke)
(2) automatically generalizes
(3) seeks to show how memories are acquired not just how they are organized after learning

BUT...not good at generating new insights (not really doing what scientific theory should)

there are two or more features that differentiate the target from the distractors
VS.
when the target differs from the distractors on just one feature