Week 5 - Brain + Cognition - Memory

Short-term Memory

= the temporary retention of a limited amount of information

Working Memory

= a mental workspace that temporarily holds and manipulates information so we can complete cognitive tasks

STM Capacity - Miller (1956)

• digit span test

• ppts hear a sequence of numbers ad must immediately repeat them in the correct order

• chunking = grouping separate pieces of information into meaningful units

• the magical number 7 +- 2 items, meaning approx 5-9 in immediate memory tasks

STM Capacity revisited - Cowan (2001)

• suggested that STM capacity is closer to 3-5 chunks, approx 4

• Miller isn’t wrong - different tasks + opportunities for chunking + rehearsal can produce different capacity estimates

• Cowan’s research considered memory performance when these strategies were restricted

• articulatory suppression = ppts must repeatedly say an irrelevent sound or word while trying to remember information - prevents them from silently rehearsing the items

• when rehearsal is restricted, STM capacity is lower

• capacity may be limited because of biological restrictions (eg, limits on the timing of neuronal activity) and computational benefits (maintaining fewer chunks may allow faster, more efficient processing + reduce interference)

Visuospatial STM - Luck & Vocal (1997)

• investigated how much visual information people could temporarily remember

Method:

• ppts viewed an array of coloured objects for 100 milliseconds

• experienced a 900-millisecond delay

• saw a test array

• decided whether anything had changed

Findings:

• accuracy declined substantially when the number of objects exceeded approx four

• similar capacity patterns across colour, shape, orientation, location + combinations of features

Conclusion:

• visuospatial STM capacity is approx 4 objects

• info can be stored as integrated objects, rather than every feature needing its own separate memory slot

Multi-store model - Atkinson & Shiffrin (1968)

Three Distinct Stores:

• sensory memory - briefly retains incoming info

• STM - limited capacity + duration

• LTM - vast capacity + longer-lasting storage

• with rehearsal, it then goes into LTM

Strengths:

• evidence from amnesia = some patients have impaired long-term learning but relatively preserved immediate memory

• Patient KF - Warrington & Shallice - KF had severely impaired verbal STM (digit span of approx 1) but had relatively preserved LTM

• this supports the distinction between STM + LTM

Multi-store model - Strengths

Strengths:

• evidence from amnesia = some patients have impaired long-term learning but relatively preserved immediate memory

• Patient KF - Warrington & Shallice - KF had severely impaired verbal STM (digit span of approx 1) but had relatively preserved LTM

• this supports the distinction between STM + LTM

Multi-store model - Limitations

1) STM isn’t one single store = KF had impaired verbal STM but preserved aspects of visuospatial STM

2) Rehearsal isn’t sufficient to explain learning = the type of processing matters, especially semantic processing

3) Information doesn’t necessarily pass through stores serially = KF’s relatively preserved LTM despite poor STM challenges compulsory STM-to-LTM transfer

4) The model focuses on storage rather than how information is manipulated for reasoning, language and decision-making

5) Forgetting may involve interference, rather than simple displacement

Working Memory Model - Baddeley & Hitch (1974)

• central executive = controls attention and coordinates processing (like the big boss)

• phonological loop = verbal + sound-based information

• episodic buffer = integrates information (added later by Baddeley in 2000)

• visuospatial sketchpad = visual + spatial information

Phonological Loop

• two components involved = phonological store (inner ear) + articulatory control process (inner voice)

Phonological Store (inner ear) = temporarily stores sound-based information (eg, words you hear)

Articulatory control process (inner voice) = allows silent rehearsal + converts written words into a sound-based code

• supporting evidence - Baddeley (1966) acoustic similarity effect - ppts found acoustically similar words harder to recall immediately than acoustically dissimilar words - similar-sounding info is more easily confused in phonological store

• Baddeley et al. (1975) articulatory suppression - pots repeated an irrelevant word while remembering info - this particularly impaired memory for visually presented words because articulatory rehearsal was disrupted

• the findings support the distinction between the phonological store and articulatory control process

Phonological Loop + Language Learning - Baddeley, Papagno & Vallar (1988)

• patient PV had impaired verbal STM

• she was tested on learning pairs of Italian words + Italian words paired with unfamiliar Russia translations

Findings:

• PV learned familiar Italian word pairs relatively normally but struggled severely to learn the Russian vocabulary

Conclusion = learning unfamiliar vocab relies heavily on phonological STM

Phonological Loop + Language Learning - Baddeley & Gathercole (1990)

• researchers compared children with language impairments against typically developing children matched on age or language ability

• they used nonword repetition tasks, where children related unfamiliar made-up words

• children with language impairments performed worse, particularly as nonword became longer

Visuospatial Sketchpad - Logie (1995)

= temporarily maintains visual + spatial information

2 separate components - visual cache + inner scribe

Visual Cache = stores visual properties such as colours, shapes, patterns

Inner Scribe = stores spatial locations + movement information, and rehearses visual information

Episodic Buffer

• Baddeley introduced in 2000

• the OG WMM had a difficulty explaining how verbal and visual information were integrated and how temporary memory interacted with long-term knowledge

• it combines info from different sources unto coherent representations

• connects working memory with LTM

• has a proposed capacity of approx 4 chunks

• Darling & Havelka (2010) = ppts remembered digit sequences resented in three arrangements - one fixed location, a straight line or a familiar keypad layout

Findings : digit recall was better with the keypad arrangement

Explanation - familiar spatial knowledge and representations of finger movements helped ppts remember the number sequence

• this supports interaction between verbal info, spatial info + LTM, consistent with the EB

Central Executive

= the control system of working memory

• it directs attention, coordinates the other components + helps us manage complex tasks

Four Main Functions:

• maintaining attention - staying focused on a task

• selective attention - focusing on relevant info while ignoring distractions

• divided attention - allocating attention across multiple tasks

• inhibitory control - suppressing inappropriate or automatic responses

Strengths of Multicomponent Model

• explains active manipulation as well as temporary storage

• accounts for differences between verbal and visuospatial STM

• helps us explain complex cognitive tasks involving multiple types of info

• is supported by neuropsychological evidence, including patients with selective memory impairments

Limitations of Multicomponent Model

• may oversimplify working memory model by largely overlooking other sensory info, including smell, taste, touch

• the precise number and nature of central executive processes are difficult to identify

• many complex tasks involve several processes simultaneously

• interactions between components remain incompletely explained

Cowan’s Embedded Process Model (1999)

• proposed that WM is not made up of separate specialised storage systems

• instead, WM involves the temporary activation of information within long-term memory

• three nested elements - focus of attention (approx 3-5 chunks), activated LTM (temporarily active info around 10-20 seconds), LTM

• KEY DIFFERENCE - Baddeley - WM involves specialised components. Cowan - WM is the activated portion of LTM with a limited focus of attention

Cowan’s Model - Strengths

• explains how STM and LTM interact without requiring many separate stores

• consistent with some neuroimaging findings showing overlapping brain activity during processing and working memory tasks

Cowan’s Model - Limitations

• Norris (2017) - argued that simply activating existing LTM cannot fully explain temporary memory for new info (eg, how can working memory represent a completely unfamiliar word that has no established long-term representation?)

• it is a challenge to explain neurological dissociations between STM and LTM : the hippocampus and medial temporal lobes as important for forming new relations, even over short intervals. Suggests that some STM tasks involving relational info may be impaired in amnesia

Levels-of-processing Theory

• Craik & Lockhart - proposed that the type of processing used during learning influences how well info is remembered

• meaningful, elaborative processing generally produces stronger long-term memories than superficial processing (semantic)

Craik & Tulving (1975):

• ppts processed words using different tasks

• shallow (graphemic) - is BIRD written in capital letters?

• intermediate (phonemic) - does bird rhyme with another word?

• deep (semantic) - does bird fit into a meaningful sentence?

• the semantic condition produced the strongest recognition memory = because processing meaning creates a richer representation than simply attending to how a word looks

Maintenance Rehearsal

= repeating info without necessarily processing its meaning

Elaborative Rehearsal

= connecting info to

Levels-of-processing Theory - Strengths

• is robust, the theory highlighted the importance of encoding processes

Levels-of-processing Theory - Limitations

• depth is difficult to define independently + semantic processing isn’t always the best for every retrieval task

Distinctiveness - Von Restorff Effect

• means that some info stands out because it differs from the surrounding info

• eg, cow dog pig cow elephant hen - longer, unusual item stands out, may become easier to remember

• Von Restorff effect - describes improved memory for distinctive info

• distinctiveness may help at encoding or during retrieval by making a memory easier to locate

Transfer-appropriate Processing - Morris et al. (1977)

= proposes that memory performance depends on how well the processing used during learning matches the demands of the later test

• eg, if the exam asks about a word’s meaning, semantic encoding should be especially helpful. If the exam asks about a word’s sound or rhyme, phonological encoding may be more helpful

= challenges the levels-of-processing in which semantic processing must always produce the best memory

The Testing Effect - Roediger & Karpicke (2006)

= means that actively retrieving info can produce better long-term retention than simply studying it again

Compared two learning conditions using short stories:

• Study-study (SS) - ppts studied material twice

• Study-test (ST) - ppts studied it, then attempted to retrieve it

Findings:

• at a five-minute delay, studying twice produced better performance

• however, after 2 days and 1 week, the ST condition produced a better retention

Conclusion:

• retrieval practice can reduce long-term forgetting, even if repeated studying produces better immediate performance

Elaborative Retrieval Hypothesis

= retrieval activates associated info that can later help recover the original memory

Dual-memory Theory

= retrieval with feedback strengthens the original trace and can create an additional memory trace

Ebbinghaus - The Forgetting Curve

• investigated how memory changes over time

• he developed the savings method, measuring how much less practice was needed to relearn material compared with the original learning session

• relearning requires fewer trials than original learning, and forgetting is generally fastest initially before slowing over time

Trace Decay + Interference

• trace decay - proposes that memory traces weaken over time

• interference - occurs when other info disrupts memory

Two Types:

• proactive interference

• retroactive interference

Proactive Interference

= old learning disrupts new learning

• eg, your old password makes it difficult to remember your new one

Retroactive Interference

= new learning disrupts old learning

• eg, your new phone number makes it difficult to recall your old one

Interference Theory - Limitations

• much of the evidence concerns explicit memory, rather than implicit memory

• interference theory doesn’t directly explain why the rate of forgetting slows over time

• it is difficult to determine whether interference primarily affects encoding or retrieval