W8 - Models of Memory

Hermann Ebbinghaus

Hermann Ebbinghaus

• First person who studies memory experimentally

• Learning - studied CVC (consonant, vowel, consonant) nonsense syllables to exclude prior knowledge

• Recall - testing himself the first item and had to recall the rest in the list

• Insight into

  • Learning = Law of Repetition

  • Forgetting = Savings method

Law of Repetition

• When you rehearse something you learn it better

• More rehearsal = better retention

• Plateaus = first few repetitions give the most benefit

Ebbinghaus’ Forgetting Curve

• Soon after learning = forget 50%

• Forget most straight away

3 Core Memory Processes

1. Encoding (paying attention)

• Transformation from perceptual representations into attentional focus

• Attention - selects what is relevant vs left out

2. Storage/maintenance

• How is a memory kept in focus of attention (STM)

• How is it moved to LTM

3. Retrieval/remembering

• Bringing back into focus of attention

Multi-Store Model of Memory

• Unattended info is quickly lost

• If attended = goes into STM

• Unrehearsed = info is lost

• If rehearsed = goes into LTM

How long do stores last?

• Sensory = msec- secs

• STM = secs-mins

• LTM = days, years, indefinitely

Sensory memory: Function

• Fills in blanks when there is intermittent stimulation (e.g. sparkler looks like a circle) = perception is briefly held in memory

Function:

• To keep sensory info in mind so can be attended to

  • ionic memory (vision)

  • echoic

  • haptic

  • olfactory

  • gustatory

• = forms automatically (doesn’t need attention)

Sensory memory: Capacity Studies

Averbach (1963)

• asked how many dots they saw

• found: amount of info in iconic memory = 4-5 items

Sperling (1960)

• Briefly presented letters in 3×4 matrix

• Full report

  • name as many letters as possible

  • result = 4 letters

• Partial report

  • sound indicated which row to look at

  • result = 6 letters

  • SUM of performance in each row

Sensory memory: Capacity + Duration

Sperling: 2 conclusions

Capacity:

• large amount held in ionic memory

Duration:

• After 25—500ms approached 4 letters

• Info in ionic decays rapidly

• Anything left transferred to STM before it was lost

• Tone/sound = attention

• Attention = info into STM

Short Term Memory

Function:

• Conscious processing of info

• Attention is key

Capacity is limited:

• (Miller, 1956) 7+-2

  • can be overcome by ‘chunking’ = grouping info together = 7 is an overestimation

• Cowan (2000)

  • 4+-1

  • Luck + Vogel (1997) - if set size > 4 = severe drop in performance

  • Electrophysiological and neuro-imagine support

Retrieval from STM

(Sternberg, 1966)

• 2-6 letters presented = within STM capacity

• confirm is letter was in set or not

Possibilities:

• parallel search (all at same time)

• serial self-terminating (stop when found)

• serial exhaustive (one after another, continue even when found = CORRECT

CORRECT:

• Ppl scan STM in a serial exhaustive way

• Scan rate = 38ms per item

STM: Duration

Limited duration:

• If not rehearsed = info lost within 15-20 sec

• decay + displacement

• rehearsal = process of repetitively verbalizing / thinking about the info

CRITCISM of STM: only forgetting due to time delay?

pre-active vs retro-active interference:

• Pro-active = something from earlier interferes

• Retro-active = something something learnt later affects earlier memory

Release from Pro-active interference:

• previous trials generate interference

• gradual decline in performance

• when nature of stimulus changes (e.g. letter to numbers) = accurate recall again

CRITICISM of STM: only forgetting due to decay?

Interference is most likely the cause for forgetting:

• especially at longer intervals

• when stimuli are similar

• when learning lots of info

• when learning info close together in time

• when learning in the same context

Other Criticisms of the STM

• Info in the STM does not need to be processed consciously

• Simple rehearsal does not ensure LTM storage

• Learning in STM affected by LTM (e.g. chunking)

• Double dissociation found in patients supports STM-LTM distinction

  • BUT some with impaired STM can still acquire LTM memories

Long Term Memory

Function:

• Organises and stores info

Capacity:

• Unlimited

Duration:

• Permanent

Working Memory

Baddeley + Hitch (1974)

STM:

• not just for passive retention

• it is necessary for language, mental arithmetic, reasoning etc

WM: Baddeley (1986)

• Digit span task: short-long sequence to be rehearsed out loud

• = varying demands on limited capacity STM

• = varying degrees of interference

• coupled with reasoning task

Results:

• less slowdown that expected

• with 8 items = more than STM capacity = performance should break down

• BUT there was not difference in error rate

SO:

• even at max digit span = still possible to do other tasks

• idea of different subsystems

WM: Diagram

model of STM

Example of using the WM

How many windows in your parents house?

• Visuo-s s = imagine house/rooms

• Phonological loop = counting windows

• Central executive = selecting + running strategy

• Episodic buffer = holding + integrating info

WM: How to test capacity of separate components

• If 2 tasks use the same components = cannot be performed successfully together

• If 2 tasks use different components = should be possible to perform them as well together as separately

= dual-task experiment

WM: Phonological loop

Temporary storage of speech-like info (‘verbal STM’)

2 primary structures:

• Phonological store

  • temporary, passive

  • limited in time (2sec) + capacity

  • code = speech-based

• Articulatory loop

  • active rehearsal component

  • linked to speech

WM: Evidence for phonological loop

1. Phonological similarity effect

2. Word length effect

3. Unattended speech effect

4. Articulatory suppression effect

PL evidence: Phonological similarity effect

• Errors more likely to be phonologically similar to correct item (e.g. F-S and B-G)

• More likely to misremember if items sound similar

  • e.g. ‘mad cap man map’ harder than ‘pen day cow bar rig’ (Baddeley, 1966)

PL evidence: Word length effect

• The memory span for short words > long words (can recall more)

• Word length effect is due to articulation duration, not syllables = evidence for PL

• e.g. span for wicket = bishop > harpoon + labile

  • same no. of syllables but shorter articulation duration)

• This is why ppl of language with a faster articulation rate (e.g. Chinese) have a larger span

PL evidence: Unattended speech effect

• Performance impaired if other verbal material needs to be ignored

  • even w nonsense syllables

  • even in diff languages

  • even w vocal music

  • less with instrumental music

  • not with white noise

• irrelevant spoken material can gain access to phonological store

• filter to distinguish between noise + speech

PL evidence: Articulatory supression

• Prevent rehearsal by covert articulation (practice silently) = there is overall worse performance

• Articulatory suppression (saying ‘thethethe…’) while learning items when to-be-learned item is presented VISUALLY, will:

  • results in speaker being unable to use sub-vocal articulation

  • evident = word length effect disappears (sub-vocal articulation required)

• to-be-learned item is presented AUDITORILY:

  • have no effect on sub-vocal articulation bc the tb-learned item has direct access to phonological store

  • evident = word length effect does NOT disappear

WM: Visuo-Spatial Sketchpad

• A system for setting up/manipulating images + spatial movement

• Limited capacity

• Processes spatial + visual + kinesthetic info

• 2 components:

  • visual cache = visual info about shape + colour (what)

  • inner scribe = spatial + movement info (where)

VSS: Function

• Construction, maintenance + manipulation of mental images

Mental Scanning:

• manipulation of info in VSS (Kosslyn, 1978)

• learn map of island

• scan from well to tree

• mental scanning between imagines landmarks increases linearly as distance between them increases (e.g. further than well to lake)

Neurophysiological findings:

• different brain areas active during visual (occipital) and spatial (parietal) tasks

Mental rotation:

• (Shephard + Metzler, 1971)

• 3D figure pairs - are they same or diff

• time to answer = proportional to amount of mental rotation required

WM: Central Executive

• Attentional system: Maintain tasks goals / goal-related info + use this to direct processing

• Capacity is limited = no storage

• Most important + active component

  • directing attention to task

  • switching between strategies

  • selective attention + inhibition

• Probably in prefrontal cortex

  • patients with frontal lobe damaged = problems of attentional control

CE: evidence

Dysexecutive syndrome

• disruption of CE due to frontal lobe damage

• some patients show:

• perseverance = cannot interrupt ongoing schema

  • e.g. can sort deck of cards by suit but cannot change to sort by colour

• utilization behaviour = fail to focus (automatic responding to cues in environment)

• catatonia = unable to initiate schemas (remain motionless/speechless for hours)

Alzheimer’s patients:

• problems with distributing attention between two tasks (function of CE)

WM: Episodic Buffer

• Phonological loop + VSS are modality specific (verbal vs visual)

• EB can integrate info into a single episode

• Can hold about 4 pieces of info in multidimensional code

• EB assists in binding = integrating info about location, colour, size etc