Short Term Working Memory Learning & Forgetting

Question 1

Define working memory

Answer 1

WM = Mental workspace
■ it’s a system that temporarily retains and manipulates all the
necessary restricted set of info components to perform activities e.g., reasoning, learning, and comprehension leading to a mental or behavioural response. (simply manipulates/stores info temporarily like mental arithmetic)

Question 2

Define STM

Answer 2

the retention and basic processing of info in simple tasks. (STM is part of WM.)

Question 3

Explain pure organic amnesia in patients and what it suggests in relation to memory

Answer 3

-they have great difficulty in creating new memories (i.e. learning). + forget quickly after a few mins/if distracted
-but do well in a variety of tasks requiring memory e.g., having a brief convo, performing info-processing tasks
-therefore there must be a form of memory of limited duration and
capacity, a form of STM

Question 4

What is the storage capacity of STM and how was this tested?

Answer 4

■ “The magical number” 7+/-2 (Miller, 1956)
Verbal digit span tests:
-people listen to a series of digits and repeat them in the correct sequence immediately after.
■sequence length increases gradually until the participant fails (twice).Last correctly completed sequence=P’s digit span.

Question 5

Why does our STM create chunks to store info?

Answer 5

It makes it easier to retain and process large amounts of
info but difficult to measure capacity accurately! (capacity can increase with training to pass “magic number”)

Question 6

How is information organised/stored in the STM as chunks?

Answer 6

-Personal semantics:2, 4, 1, 0 are 4 chunks of info to most, born on the 24th of October (now remembered as a single chunk)
-Prosodic preferences:easier to remember 816-973-524 than 816973524.
-Phonological plausibility:STIMPOLENA vs OIPMLTSAENL (both semantically meaningless BUT we can remember the pronounceable sequence STIMPOLENA one chunk)
-Expertise, such as language knowledge:many chunks in r h y n g w l a d w r I a e t h o l unless you speak Welsh in which case it is just one (it means international)

–Chunks demonstrate that STM interacts with LTM

Question 7

What contradictory evidence is there for STM capacity from the “magic number”?

Answer 7

-Cowan, 2001 suggests 3-5 chunks in a variety of tasks e.g.,in a continuous presentation task, P’s can usually remember up
to the last 5 chunks.
-Baddeley (2000) also sets the capacity of the episodic buffer at 4.

Question 8

Why the discrepancy with Miller (1956)?

Answer 8

-Researchers have suggested that in the digit span task, people (covertly) combine individual items into chunks.
-When covert rehearsal and processing are prevented (e.g. through articulatory suppression) STM capacity appears much reduced.

Question 9

Explain the sequential comparison
procedure which also investigates STM

Answer 9

■ P’s study a sample array of objects for 100ms and after a very brief delay (900ms) a test array. They need to decide if the sample and test arrays are the same or not.
■ Arrays differ across a no. of dimensions (colour, shape, position, orientation) and the participant is not warned what dimension is likely to have changed.

Question 10

What is usually found when doing the sequential comparison procedure?

Answer 10

■ Accuracy scores for different array sizes (set size) decreases when it exceeds 4 items.
■ A similar drop in accuracy as a function of set size was seen independent of feature tested (orientation, colour, shape, location) and when memory was tested for conjunctions (i.e. combinations of features)

Question 11

What does the sequential comparison procedure suggest about STM?

Answer 11

–Visuospatial capacity is around 4.
–Items are stored not as individual features, but as integrated objects. Features are already bound into objects in WM.
–This binding we consider later as part of Baddeley’s Episodic Buffer (Also relevant in Cowan’s model)

Question 12

Why does STM have a capacity limit according to Cowan (2010)?

Answer 12

Biologically determined deficit:
–Limitations in neuronal firing speed=harder to activate 4+ chunks within a given time window. not activated close together=chunks not part of the same event
– Limited neuronal firing speed may be useful to avoid interference:blue square then red circle=blue circle/red square?
Computational benefit
– A smaller no. of chunks=faster + more efficient processing ignoring irrelevant info

Question 13

Explain the multi-store model
Atkinson & Shiffrin (1968, 1971)

Answer 13

-3 memory stores: sensory memory (sensory buffers), STM and LTM
-duration:Sensory + STM retain info for short periods + LTM is much longer lasting.
-capacity:Sensory memory and STM=limited capacity + LTM=vast
-It’s a serial model where info is transferred from one store to the next.
-Input to STM from environment (via sensory stores) or from LTM.
-rehearsal=info retained in STM + guarantees LTM transfer and storage.
-Info in STM is lost through displacement.

Question 14

What are the strengths of the MSM?

Answer 14

*It’s highly influential
-Experimental and brain-damaged patient provide evidence: It can easily explain amnesic data
Evidence for different memory stores:
-Duration
-Capacity
-Forgetting mechanisms

Question 15

Give neuropsychological evidence: STM impairments patient KF Warrington & Shallice, (1969)

Answer 15

*patient KF showed impaired STM functions with limited digit span (only 1 item), but normal LTM. Similar dissociations between STM and LTM have been reported with other patients since.
* Supports the idea of two separate stores.
But also problematic for the model as:
* There’s no compulsory transfer of info from STM to LTM. If it were, LTM would also be affected.
* Stores are not serially connected. Parallel inputs from sensory stores go to both STM and LTM. (i.e. not intertwined, separate systems)

Question 16

Does STM comprise of more than one store/processing unit?

Answer 16

*Yes, Patient KF and others show deficits in verbal STM but fine
when tested on visuospatial ST tests such as the Corsi block tapping test whereas other patients show the reverse impairment
*We can find similar dissociations between verbal and spatial STM in
neurologically intact individuals.

Question 17

What are the problems with the MSM?

Answer 17

*STM can’t be seen as a single, unitary system as more than one store/systems exist to deal with different info.
*The probability of creating LT memories doesn’t depend upon amount of rehearsal but type of processing that occurs, semantic processing produces in most cases the strongest memories.
*no serial transfer of info across stores. Intact STM is not a prerequisite (not required beforehand) for normal LTM.
*Emphasis on storage. The model fails to account for complex aspects of cognition (decision making, reasoning, language comprehension etc.) and complex behaviours (having a conversation, driving) i.e. for working memory.
*Forgetting from STM may not be due to displacement but interference.

Question 18

State the features of the phonological loop:WM

Answer 18

-stores verbal (language-based) info for a brief period of time (similar to the Atkinson & Shiffrin model)
-Auditory (spoken words) and Visual (written words).
-Phonological Store (inner ear)=retains spoken language in a phonological (sound-based) code
–Articulatory Control (inner voice)= process which is engaged in subvocal articulation (silent pronunciation) updating the
contents of the store and converts written words into sounds so that
they can access the store.

Question 19

Give evidence for the PL – Different processing mechanisms (Adapted from Baddeley, Thomson & Buchanan,1975)

Answer 19

• For 1/2 the words, participants engaged in articulatory suppression (i.e. kept repeating the same syllable whilst being presented the words). Articulatory suppression disabled the articulatory control process.
• Written words could not be converted into phonological code and rehearsed and as a result fewer written words were recalled thank spoken. No such problem for spoken words as they had direct assess to the phonological store.

Question 20

Why do we need the Phonological Loop? Baddeley, Papagno & Vallar, (1988)

Answer 20

*Patient PV presented with unrelated word pairs from her native language and pairs consisting of an Italian (native) word and a Russian translation
*Learned native word pairs at a normal rate but failed to learn the Russian vocab/Italian-Russian pairs
*Native word pair learning relies on existing semantic links (so limited need for phonological coding) but foreign word learning relies on phonological processing.
*Need to retain the Russian word in STM long enough to associate with the native word. (therefore PL aids in learning in the absence of semantics

Question 21

Explain Gathercole & Baddeley, 1990 procedure: comparing PL with language-impaired children

Answer 21

-group of 8 yr old children with language impairments (language abilities of a 6 yr old), other 8 yr olds with no language impairment
and 6-year-olds with no language impairment.
-tested the ability to repeat nonwords e.g., woogalamic to measure the integrity of the phonological loop.
-Language-impaired children performed worse overall becoming clearer as the phonological load increased (from 1 to 4 syllables).
-Demonstrates language skill development and vocabulary acquisition based on the PL (i.e. the ability to retain and rehearse nonwords/words in WM and connect it with existing knowledge).

Question 22

State the features of the Visuo-Spatial Sketch pad

Answer 22

-Visual Cache=stores visual info about object shapes, patterns and colours. (Visual stimuli gain direct access to the store.)
-Inner Scribe=stores spatial info including object location and movement + rehearses the contents of the visual cache.
–Similar in structure and function to the phonological loop.

Question 23

Explain the Episodic buffer

Answer 23

-newest component in the model added in 2000 to address 2 main difficulties within the multicomponent model.
-Buffer stores (PL and Sketchpad) need to interact with LTM to organise info into chunks (semantic and episodic LTM).
-Incoming info is in the form of whole events both visual and verbal, and there needs to be a storage area for these integrated forms.

Question 24

Explain Darling & Havelka, 2010 Episodic buffer experiment

Answer 24

-looked at digit span performance in 3 conditions:Single digit – same location/Linear arrangement/ Keypad arrangement
-Better STM for keypad display than a single item or linear display.
-This is because familiar location boosts memory for the digits.
“Motoric representation” of sequence in LTM. Participants recall an imaginary motor movements as if they are using their fingers to tap in the
number sequence (e.g., keypad)

Question 25

State the Central Executive

Answer 25

An attentional control system:
–Automatic or semi automatic control system: used for decision
making and conflict resolution in (often) complex but well
learned (habitual) situations.
–Supervisory attentional system: used for decision making and
conflict resolution in novel or unexpected situations.
-receives input from the subsystems (PL, VSSP, EB), selects and evaluates streams of information, overrides responses, switches between tasks and
constantly monitors outcomes.

Question 26

What are the main operations of the CE

Answer 26

–Retains attentional focus
–Selective attention
–Dividing attention
–Inhibition control of responses

Question 27

What tests have been devised to explore the CE?

Answer 27

-Dual task performance tests
–Random generation tests
–Tests that require switching between rules and inhibition of previous responses

Question 28

Multicomponent model evaluation: Strengths

Answer 28

Model accounts for more complex tasks than Atkinson and Shiffrin’s
(1968) short-term memory store, in that it:
is concerned with both active processing and short term storage
accounts for data from brain-damaged patients
verbal rehearsal is required for the PL and VSSP but there are also more complex functions performed

Question 29

Multicomponent model evaluation: Limitations

Answer 29

■Oversimplified (e.g., ignores taste, smell, touch)
■Proved difficult to identify number + nature of main executive processes
■Most complex tasks involve the use of >1 process
■Interactions among components are unclear

Question 30

True or false: WM is a functional, not a structural system. It does not involve dedicated/specialised buffers and stores (as in the multi-component model).

Answer 30

True (Cowan, 1999) it just has different levels of activation instead

Question 31

Explain the components of the embedded processes model (Cowan, 1999)

Answer 31

It involves 3 hierarchically placed elements(gets more activated further down):
*Long-term memory
*Activated part of Long-term memory
*Focus of attention on a subset of activated Long-term memory
Different processing limits:
-Activated part of Long-term memory is time limited (10-20s)
-Focus of attention is capacity limited (3-5 chunks)

Question 32

Explain the embedded processes model (Cowan, 1999)

Answer 32

*Attention can be voluntary,(involving the CE) or involuntary.
*Depending on the stimuli, goals and tasks,different features are in different states of activation. Attention can focus to all activated items in turn.
*Attention produces awareness=encoding and/or retrieval of info in and out of LTM.

Question 33

Unitary-store models (e.g. embedded model process) evaluation: Strengths

Answer 33

*Emphasises similarities between STM and LTM
*Good evidence STM depends at least partially on activation within LTM solving the problem of communication between stores.
*Good correspondence with neuroimaging data: Some areas that are active during info processing also active in WM tasks (But correlational!)
*No need to keep identifying more distinct stores.

Question 34

Unitary-store model: evaluation: Limitations

Answer 34

STM for novel and repeated info cannot simply involve activated LTM (Norris, 2017):
*STM crucial in learning new vocab
*So unclear how sequences with the same info presented more than once (e.g 1751358) can be based on activation of the LTM trace
*Need to account for findings of dissociations in neurological patients.

Question 35

Why do amnesics perform well on most STM tasks but very poorly on most LTM tasks?

Answer 35

 Hippocampus including medial temporal lobes (damaged in amnesic patients) crucial for forming new relations in STM as well as storing LT memories.
 Amnesics usually perform well on STM tasks because they mostly do not require processing and storing novel relations.
 KEY PREDICTION: amnesics should have impaired memory with STM tasks requiring relational memory.
Difficult to test as some tasks may require LTM which is impaired in amnesia.

Question 36

Explain Levels-of-processing theory

Answer 36

■Depth of processing of a stimulus (extent to which meaning is extracted) affects memorability
(Craik & Lockhart, 1972)
–Memory is a by-product of info processing
■Deeper levels of analysis=more elaborate, longer-lasting and stronger memory traces>shallow analyses
Rehearsal:
■ Maintenance = simply repeating information
■ Elaboration = deep semantic processing
■ The idea of deeper processing was later abandoned in favour of the idea of semantic vs perceptual processing.

Question 37

What tasks were used in Craik and Tulving (1975) experiment?

Answer 37

1.Shallow graphemic task=decide whether each word is in upper- or lower-case letters
2.Intermediate phonemic task=decide whether each word rhymes with a target word
3.Deep semantic task=decide whether each word fits a sentence containing a blank

Question 38

What were the results in Craik and Tulving (1975) experiment: Levels of Processing

Answer 38

-better performance when info presented semantically rather than phonologically/other perceptual formats
-memory by-product of the passing of information so encoding is determining factor

Question 39

Define Distinctiveness

Answer 39

the fact that some memory traces stand out due to being processed differently than others.

Question 40

Define the distinctiveness effect (the Von Restorff effect)

Answer 40

the finding that differential processing produces better memory performance. (Distinctive processing can happen during encoding/facilitating the search process during retrieval.

Question 41

Explain transfer-appropriate processing theory

Answer 41

Morris et al. (1977) suggested that encoding processes only enhance
memorability when they are compatible with the demands at retrieval.
–retrieval requires semantic information=semantic processing at encoding will yield best results
-retrieval requires phonological information=phonological processing at encoding will yield best results
–LoP effects are a special case of transfer-appropriate processing (semantic encoding suits semantic retrieval)

Question 42

Levels-of-processing theory:
Strengths

Answer 42

■Levels of processing effect is a robust phenomenon which can be replicated every time.
■Learning processes have a major impact on LTM. Research moved away from focusing on rehearsal.
■Recognised that perception, attention and memory are all closely interconnected.
■Helped identify elaboration and distinctiveness of processing as factors influencing LTM. Shows that LTM encoding and retrieval in most situations benefit from the
processing of semantic info.
–Consider the everyday life relevance of levels of processing

Question 43

Levels-of-processing theory
Limitations

Answer 43

■Hard to assess depth (which is why it was later replaced by semantic processing).
■Does not explain why semantic processing is so effective (that still remains elusive).
■Semantic processing effect is much smaller with non-verbal stimuli and implicit memory tests.
■Semantic processing effect much smaller with non-verbal stimuli and implicit memory tests.
■Unclear why levels-of-processing effect is smaller with implicit tests

Question 44

Forgetting from LTM: Savings method (Ebbinghaus, 1885/1913)

Answer 44

*Retesting the same material after relearning to quantify reduction in number of trials to reach criterion compared to original learning (% savings)
Forgetting over time is roughly
logarithmic

Question 45

Give different ways you can forget LTM

Answer 45

Trace decay (memory just decays due to death of synapses)
Proactive interference-Disruption of memory by previous learning affects new learning
 Retroactive interference-Disruption of memory for previously learned
material interfered by new learning during retention interval. (i.e. new interferes with old)

Question 46

Explain Retroactive interference (Jacoby et al., 2015) experiment

Answer 46

Some participants explicitly instructed to focus on changes between 1st and 2nd lists (e.g., Cat–Dirt; Cat–Tree)
These participants showed retroactive facilitation: the instructions allowed them to discriminate between correct and incorrect responses

Question 47

How can retrieval problems often cause retroactive interference?

Answer 47

 Lustig et al. (2004): people often retrieve items from the wrong list
 Unsworth et al. (2013): greater RI when two lists presented prior to recall of first list
* No RI when words in each list belonged to different categories: this made focused retrieval of words from list 1 possible

Question 48

Interference theory: limitations

Answer 48

*Most research involves explicit memory: limited knowledge of interference effects in implicit memory
*Does not directly explain why forgetting decreases over time
*Limited understanding of the relative importance of impaired encoding and impaired retrieval in producing interference effects

Question 49

Why can be forgetting be good in some cases?

Answer 49

*It’s a way to avoid mental “clutter”. Remove outdated
(possibly unwanted) info and reduce interference.
*Quite often it’s more useful to remember the gist of
info or the actual content rather than peripheral details.
*It can help reduce access to painful memories.
*People with hypermnesia, who are able to remember huge
amounts of autobiographical details, often view lack of
forgetting as a problem.