cog psych exam 2

semantic memory

recalling meaning of words, concepts, etc.

associative memory

recalling relations among events, concepts, words, etc.

procedural memory

recalling how to perform skills

declarative memory

recollection of facts, definitions, etc.

episodic memory

recollection of events, both their content and their contextual details

autobiographical memory

recollections of events in one’s own life

Ebbinhaus (1885)

• pioneered field of empirical research on memory

• On Memory

• for his book, he submitted himself to over 800 hours of memory experiments

• forced himself to remember lists of CVCs strings

• would learn a list, test himself intermittently b/w studying, until he could recite list perfectly

• would record how many testing attempts it took him to reach 100% recitation accuracy, and how this changed when he tried to re-learn the list at various delays (e.g., minutes, hours, days, weeks)

Ebbinghaus’ findings

• learning curve and forgetting curve

• better at remembering in morning than evening

• could remember about 7 items perfectly (i.e. working memory span)

• memory is heavily associative in nature with temporal contiguity being heavily influential on memory

• memory was variable, primarily due to his fluctuations in attention

Muller (1900)

• brought some more experimental rigor to Ebbinghaus’ experiments

• more people and with memory drum— device for showing items to people one at a time at consistent rate

• __ and Pizlecker also used a paired-associate learning task

  • ex. study: horse - cabin; test: horse - ??

• findings:

  • interference: later learning can interfere with earlier learning (retroactive)

  • just-experienced associations are easy to remember, as if it were ‘reverberating’ in one’s mind — concept similar to working or short-term memory

Muller’s findings

• interference: later learning can interfere with earlier learning (retroactive)

• just-experienced associations are easy to remember, as if it were ‘reverberating’ in one’s mind — concept similar to working or short-term memory

representation, process, storage

ways to think about memory

memory as representation

• internal representations of things that have been learned/experienced

• features of memory are multidimensional attributes

  • visual, tactile, auditory, olfactory, semantic, emotional, temporal, etc.

• memory is associative; experiencing 1 feature triggers other attributes

  • ex. smelling cookies reminds you of Christmas

memory as process

• memory is means by which we remember info

• we acquire (learn/encode) info

• then we store or maintain it

• then we retrieve it

• encoding —> storage —> retrieval

memory as storage

• memory is the place where representations are stored

• not just about the brain but in more conceptural terms, too

• memory is where we ‘look’ for info we need

List learning paradigm

• most common form of memory task

• people given list of words to remember and asked to recall

free recall

people recall words in any order but they must do so without any hints or clues

• immediate - no gap between presentation of last word and recall period

• delayed - time delay between presentation of last word and recall period

serial recall

people must recall items in a particular order (usually forward or backward)

recognition memory test

• people given words that either were or were not on list and asked to indicate the ‘old’ vs ‘new’ words

implicit memory tests

• people given list of words and asked to do something else (ex. indicate whether something is word vs. non-word; say whether word represents something living or non-living; say whether word begins with letter or vowel)

• when they see words first time, they don’t know their memory will be tested later

• then asked to recall or recognize words in a surprise memory test

• encoding is incidental rather than intentional

laws of memory

• repetition

• recency

Repetition

seeing info more than once leads to better memory for it than a single encoding instance

• occurs even for incidental encoding

• leads to better memory when encoding instances are spaced apart rather than ‘massed’ (i.e., given in close succession)

Singh et al. (1994)

• massing vs. spacing

• manipulation: experimental commercial was either repeated with either 1 or 4 commercials b/w repetitions, or only shown once

  • remembering commercials when spaced or massed together

• result: when tested next day, there was minimal decrease in memory for experimental commercial (5%) compared to immediate testing when there were 4 intervening commercials vs. 1 (40% decline)

  • more likely to forget commercial if massed

recall

free response questions are an example of testing ___

recognition

multiple choice questions are an example of testing __

F

T/F: massed learning is better for memory than spaced

recency

• info can be recalled more accurately when there is a shorter delay b/w when it is experienced (or encoded, or learned) and when it is recalled

retention interval

amount of time between initial exposure and recall attempt

extralist intrusion

error when word is not on list

repetition error

error when word was already reported

previous-list intrusion

error when word is from a prior study list

Murdock (1962)

• among first to demonstrate what we now call the serial position effect

• when asked to recall info, we tend to recall earliest info learned and most recent info we learned the highest probability

primacy effect

we tend to remember the earliest info we seer

recency effect

we tend to remember the most recent info we see

Atkinson & Shiffrin (1968)

• most influential dual-component model

• offers mathematical formula of memory output

• short-term memory is capacity-limited and can only hold info briefly

• we use control processes to operate on STM representations

• LTM has an unlimited capacity and info is not ‘lost’ we just have a difficult time retrieving it, due to things like interference or search failures

the Modal Model

T

T/F: info stored in LTM stays there forever unless something is fundamentally wrong (ex. injury, cog. decline)

T

T/F: the sensory store is not really considered ‘memory’ but more related to sensation and perception

T

T/F: LTM has unlimited capacity

F

T/F: STM has unlimited capacity

structural components of the dual component model

• sensory stores

• Short term store

• long term store

processing components of the dual component model

• change depending on task

• change based on whether you’re encoding or retrieving info

rehearsal

main control process that maintains info in the STM

T

T/F; the longer info is in STM, the more it transfers to LTM

7 ± 2

Miller (1956) estimate of STM

4 ± 1

Cowan (2001) estimate of STM

chunking

• because capacity is limited we must take steps to __ info into discrete representations

  • ex. phone numbers

Brown-Peterson (1958/9)

• examined how quickly info decays from STM

• task: recall consonant trigrams after a delay

• during delay they blocked active rehearsal (count backwards from 3)

• delay was variable

• result: info gets lost from STM quickly unless you control it (thru rehearsal)

Conrad & Wicklegreen (1964/5)

• argue that verbal info is kept in acoustic code (subvocal rehearsal)

• we often covertly (silently) rehearse info

• use inner speech mechanism to rehearse it

Rundus (1971)

• participants given list and asked to rehearse out loud a certain number of items

  • lists rehearsed either 3, 6, 9, or 12 times

• expect: more rehearsal, better remembering long-term

• results: divergence at end of list

  • end of list still being recalled even though it was rehearsed less

  • why? because this info was still in STM

Glanzer and Cunitz (1966)

• removed recency effect from Rundus’ experiment

• manipulated delay b/w end of list and initiation of recall (retention interval)

• 15 words to freely recall

• end of list if saw:

  • # —> write/recall

  • digit —> count (prevent rehearsal) and then recall

• shouldn’t affect encoding but how info gets outputted

• results:

  • immediate recall = recency effect

  • 10s delay = recency effect reduced

  • 30s delay = recency effect gone

• end-list items are lost (via time-based decay) from STM

• recency effect is because of output from STM

rehearsal; STM

Atikinson and Shiffrin’s modal model explains that primacy occurs because of __ and that recency occurs because of __

Bjork and Whitten (1974)

• is the recency effect due to unloading of items from STM?

• used a continuous-distractor free recall task (math b/w words)

• still saw recency effect wen participants can’t rehearse items or store them in STM so…

  • …the recency effect can’t only be because of STM and primacy can’t only be because of rehearsal since doing math b/w words fills up STM and prevents rehearsal

• primacy and recency are affected by what items participants choose to begin recall with

long-term recency

• shows problem of modal model

• not a list learning paradigm, but saw serial position curve and recency effect, so not b/c of rehearsal or STM

• recency effect w/ memories that are not held in STM cannot be explained by unloading from STM

• what else can explain the effect?

  • interference

proactive interference

old info interferes with new info

• affects items at end of list

retroactive interference

new info interferes with old info

• affects items at befinning of list

Baddeley & Hitch’s (1974) model of working memory

• system for temporary maintenance and manipulation of info, necessary for performance of such complex cognitive activities as comprehension, learning and reasoning

• multiple storage systems

  • visuo-spatial sketchpad

  • central executive

  • phonological store

dual-task studies

• participants asked to keep info in STM and overtly rehearse it (repeat out loud) while they do primary task

• rehearsing verbal info increases time required to do a reasoning task, but no effect on errors

• system for verbal info is not the same as the system for learning/reasoning

rehearsal; speech

Baddeley and Hitch (1977) found that performance on ___-based tasks involved a ___-based system

phonological loop

• used to recycle info

• holds small amount of speech based info

• articulatory control process is the way in which we keep things in store

• based on inner speech

• accounts for:

  • phonological similarity effect

    • phonologically similar items interfere with one another

  • irrelevant speech effect

    • irrelevant speech gets into loop

  • word length effect

    • ACP can only rehearse items so fast

phonological similarity effect

• words that sound the same are harder to rehearse than distinct words

• 4 lists: phonological similarity, semantically similar, control (unrelated words)

• results: semantically similar and control lists are better rehearsed

• phonologically similar words decreased recall performance

  • ex. PVCGE harder to recall than XRFYZ

• creation of interference b/c of phonological similarity

visual; phonological

the articulatory control process (ACP) translates __ into __

irrelevant speech effect

• performance on memory tasks is impaired if items are accompanied by other verbal material

• Colle & Welsh 1976 - immediate recall of digits is impaired if accompanied by sound of someone reading German

• unattended phonological material can gain access to phonological store

• creates interference and takes up space in phono. store that would otherwise be used up by relevant info

• Salame & Baddeley 1987 - spoken digits ‘one’ imair digit span to same degree as similar phonemes like ‘tun’

• explanation: code is phonemic, not semantic

  • we rehearse how it sounds rather than what it means

word length effect

• finite amount of tape/space in loop

• longer words use up more space

• linear decrease in accuracy of recall and time to read each word out loud

• fit fewer long words in loop

• predicts memory better for shorter words

articulatory suppression

• subject says ‘the the the the’ during list presentation or gets a tongue depressor to completely occupy ACP

• ear and mouth have privileged access to phono. loop

• disrupts ACP (loop used to rehearse info)

• ACP is busy pronouncing the irrelevant words, it cannot be used to covertly rehearse the target information

• word length effect is abolished

  • rehearsal is completely blocked for both long and short words, removing the advantage for short words

visuo-spatial sketchpad

• stores and manipulates visual-spatial info

• needed when info can’t be recoded into phono. representation

• mental rotation of objects

• other visual-spatial info interferes with it

  • verbal info does not

Brooks’ evidence for visuo-spatial sketchpad

___ (1968) -

• 1st experiment: patients read sentence and indicate whether each word is noun or non-noun

• 2nd: imagine a letter and make decisions about whether corner was in extreme bottom or top of letter

• 3 response:

  • verbalization = ‘yes’ or ‘no’

  • non-verbal = point to Y or N

  • non-verbal = tap with left finger for Y, right for N

• results:

  • verbal task: longer response time when using verbal response than visual response

  • visual task: longer response time when using visual response than vocal response

  • when response modality is same as task modality, response times increased

  • interference in sketchpad/phonological loop when participant had to use it to both make decisions and report decision

• we store and process info differently based on whether it’s verbal or visual code

central executive

• coordinates activity of two storage systems

• other potential roles: coordinating retrieval strategies, selective attention, temporary activation of LTM, suppression of habitual responses

• thought to be associated w/ frontal lobe

central executive

what is the least studied and understood component of the modal model?

episodic buffer

• update to model by Baddeley in 2002

• makes link b/w LTM and working memory more explicit

• integrate multidimensional code

• most controversial since it can do everything; what’s the point of VSSP and PL?

• controlled by CE

• ex. can convert language (verbal —> visual)

memory span

number of items (digits, letters, words, etc.) that a person could remember

digit span

• asked to repeat digits in forward serial order

• list gradually increases as they correctly recall

• max number of items they could recall is their capacity/span

• problems:

  • strict scoring criteria (no partial credit)

  • low reliability / not consistent

  • don’t predict other cognitive abilities, like IQ or reasoning

  • not much variability across people

Daneman and Carpenter (1980)

• produced first ‘complex’ span task

• must remember list of letters/digits while doing irrelevant processing task (ex. math or reading) in b/w coding

• must report in correct serial order

• partial credit given

  • more variability and validity

• performance correlates with:

  • fluid intelligence

  • reading comprehension

  • LTM abilities

  • attention control abilities

work

some argue that simple span tasks don’t make people ‘__’ with the info. they simply store it then dump it

Luck & Vogel (1997)

• designed ‘change-detection’ task

  • presented with array of visual objects for short time

  • after brief delay, shown second array

  • indicate whether the array (or single item within array) has changed

• results:

  • set size 1-3 = near perfect

  • set size > 4 - performance drops dramatically

  • no difference w/ cue, orientation, color, shape, or adding more features; so not about complexity

  • verbal load did not impair performance

    • verbal load = Participants were given a verbal load of four digits to hold in memory (rehearsing them continuously) while completing a visual change-detection task

  • conclusions:

    • visual working memory has about same capacity (4 ± 1) as verbal working memory

    • visual memories are stored as conjunctions of visual features (an object that has distinct features [color, orientation, shape] and belongs to a certain region of space)

    • 
      

recall initiation

• how do people begin recalling in free recall?

• people tend to begin recall with end-list items (recency) in immediate free recall

  • this tendency is reduced for delayed free recall

  • __ __ curve

cue dependent retrieval process

• global context (task) —> list context —> word context

1. self-generate a cue

2. use cue to narrow search

3. search processes produce a response

4. response serves as next cue

5. may need to generate a different cue if that cue is unhelpful

temporal-contextual cue dependent retrieval process

• items encoded near each other within a list will have strongest associations

• items encoded in lists near each other will have weaker associations

• words never encoded will have weakest associations

search monitoring

• after search process, ___ must be performed

F

T/F: unlike in search processing, errors in search monitoring can not lead to errors in recall

lag

temporal distance (during encoding) b/w two consecutively recalled words

ex. +1, -3, etc.

conditional response probability (CRP)

• estimate how likely someone is to transition between items in recall at each lag

• people are more likely to transition recall b/w near neighbors (low absolute lags)

• curbe is asymmetrical (forward leg is much more pronounced than backward)

• peaks around +1 and -1

• tight temoporal contextual association

conditional response latency (CRL)

• estimate how quickly someone is to transition b/w items in recall at each lag

• people are faster to recall words at shorter absolute lags (+1, -1)

• curve is asymmetrical (forward transitions much faster than backward transitions)

• IRT = inter-response time

temporal clustering

• recall transition lags of 1 or -1 represent strong ____ ___

• larger lags represent weaker ___ ___

• people with higher ___ ___ scores (tendency to output recall in temporal structure) achieve better recall (recall more words on average), so recalling with temporal structure is beneficial

• Sederberg et al. (2010)

  • people who recalled more words tended to have higher ___ ___ scores

T

T/F: using temporal context and recalling in this organized way leads to better and faster recall of information

semantic clustering

• Sederberg found that high ___ ____ alone does not drive overall list recall success the way temporal clustering does

• no relationship b/w ___ ____ and recall success

• guides transitions from word to word, it doesn't anchor items as strongly (doesn't boost total list recall capacity) the way temporal contextual cues do

• graphs show semantic relatedness influences local retrieval dynamics during free recall

• graphs don’t show accuracy

• __ ___ can even lead to errors (previous and extra list intrusions)

false memories

• using semantic relatedness among items as a cue can lead to ____

• Deese (1959)

  • list of words either semantically related or unrelated

  • high relatedness results in people likely to falsely recall related words

    • ex. best, rest, nap, pillow, slumber…

      • people recall ‘sleep’ incorrectly

• DRM

  • follow up

  • semantically related words with an unstudied ‘lure’

    • ex. floor, desk, seat, cushion, … [chair]

  • free recall and recognition test

  • 40% of critical lures were recalled on free recall

  • 84% false alarm rate on recognition

  • critical lures remembered with just as much confidence as studied words!

Search termination

• how do people deicide when to stop recalling words

• IRTs = inter response time

error

people tend to terminate their search after __. could be because word just recalled is an ineffective retrieval cue

• PLI or ELI

verbal fluency tasks

• another way of understanding recall dynamics

• recall without encoding

• ex. name all animals you can think of

• clustering & switching

  • ppl tend to recall items in clusters then they switch to new cluster when current one has been exhausted

  • seen in relatedness among responses and IRTs

T

T/F: verbal fluency is a cue-dependent retrieval process

T

T/F: for fluency tasks, individual differences diminish when given a retrieval cue (instead of forcing people to self-generate)

Unsworth et al. (2014)

• experiment 1: 5 min to recall as many animals as you can think of then report which strategies used:

  • random - 81%

  • semantic - 71%

  • personal - 45%

  • alphabet - 23%

  • rhyme - 6%

• experiment 2: assigned to recall animals either freely, by using a visualization strategy (imagine a zoo), using semantic similarity, by size, in alphabetical order

  • visual, free recall

  • semantic

  • alphabetic, size

• experiment 3: name facebook friends and report strategies

  • location - 68%

  • random - 65%

  • personal - 59%

  • activity - 59%

  • forward chronological - 29%

  • backward chronological - 21%

  • alphabetic - 15%

  • rhyme - 6%

• experiment 4: name facebook friends under one of 6 instructions:

  • location, activity, free recall

  • chronological

  • alphabetical

Unsworth’s findings

• memory retrieval involves both controlled (Strategic) and random components

• search may initially generate several (or even many) items without much effort (Random/uncontrolled) but then needs to proceed with more effortful (Strategic/controlled) search

• forcing people to use ineffective strategies is worse than allowing them to self-generate strategies

• free > visual > semantic > chronological > alphabetical (ineffective)

changing password

example of proactive interference

move to new house and forget old address

example of retroactive interference