Cognitive Psychology: Lecture 6: Long term Memory

LTM

• Assumed to be structured

• Estimated to hold equivalent to 10 terabytes of data

• Stores information for a long time with little decline in

quality of storage

– E.g. a foreign language learnt initially declined for 3 years, then remained stable for the next 25 years and still usable after 50 years (Bahrick, 1984).

For 9 years, a group of participants were tested for their recognition memory on different TV programmes that had run 1-15 years previously. The mean performance across all 9 groups are shown.

Neurons in the brain

More neurons = greater capability of learning

– For example a fruit fly has 300.000 nerve cells

– Mouse has 71.000.000 (71 million)

– Human has 86.000.000.000 (86 billion) (Hercula-Houzel, 2009)

Each neuron can make 1000 connections to other neurons

(synapses)

– Synapse = elementary unit of memory storage

– Total capacity of synapses in human brain =

86.000.000.000.000 (86 trillion)

– This must therefore be maximum capacity of the brain

– In practical terms this size can be seen as near infinite –

there’s no limit

Brain structures of LTM

Initially hippocampal activation –

a gateway for LTM

Once consolidated

will reside in the

cerebral cortex

There is no single

storage site for LTM

Front

back

Memory consolidation

LTM initially appear in a somewhat fragile state

Need to be consolidated to be stored in a stable

format that can last for years

– Resistant to degradation

– Modulate memory strength

The conversion from the fragile state to the

stable

– A process that lasts from days to years

The amygdala regulates consolidation within the hippocampus

– Different hormones help strengthen or weaken the memory trace

– May be through determining the emotional importance of memory

Stability of memory

Memory consolidation phases

From McGaugh, J. L. (2000). Memory – a century of consolidation. Science,

287, 248-251.

Better memory = bigger brain = better memory

Once consolidated, areas involved in initial perception

of stimuli, also or very near these sites is where the

LTM is stored

It is not the case that some people have better memory

storage sites than others

But, I hear you say – my memory is way worse than….

Enriched environments leads to bigger brains (and

more capacity)

– Years of practice changes the brain structure (synaptic

connections)

Brain can then encode and process more fully in more detail

This is what causes better memory performance (at least in theory!)

Memory consolidation and ECT

• A group of rats conditioned (using

footshocks) when stepping off a

platform

• Quickly the rat learns to remain on the

platform

• However, if it receives an electroconvulsive

shock 0.5 seconds after

footshock, it develops amnesia (the rat

does not learn to stay on the platform)

(Schneider & Sherman, 1968)

• Memory consolidation is disrupted

Neurogenesis (formation of new

neurons) in the hippocampus

MAM = long-term

memory formation

prevention chemical

Naïve = controls

Enriched = enriched

environments

Rats living in enriched environments for 23 days (black column) had

significantly more neurons than controls in the hippocampus. MAM

rats had significantly fewer neurons (from Bruel-Jungerman et al.,

2005)

Retrieval

Success of retrieval dependent on

– Not just strong memory in storage

– But also the quality of the retrieval cue

– Mood dependent (more on this next year)

– Context dependent (next week)

Passage of time will lead to weakening of

memory

– Interference / reconsolidation

– Loss of importance

Reconsolidation (from Walker et al., 2003)

• Ppts trained on a simple motor

sequence task, after a nights sleep,

retested [None group] and showed

significant improvement.

• A second group given an interfering

motor sequence task 10 minutes after

initial training. When tested next day –

no improvement in performance.

• Third group given interference task 6

hours after initial training – next day

showed improvement.

% improvement in accuracy

04/04/2012

3

Reconsolidation (from Walker et al., 2003)

• One group given an interfering motor

sequence task 10 minutes after initial

training. When tested next day – no

improvement in performance.

• For another group, when activating the

original memory and given interference

training the following day after initial

training, when tested after a 2nd nights

sleep, a decrease in performance

observed (decline of 57%).

% improvement in accuracy

Suggests that there is a

reconsolidation process, that the

initial memory can be interfered with

via subsequent information

LTM

Procedural memory

Have you ever read a page to get to the

bottom and realise you don’t remember what

you have just read?

No conscious awareness involved

Something that has been learnt to a degree of

automaticity

Non-declarative memory

Procedural memory

– The ‘How’ of memory

– Based on experience

– Guides performance

Motor skill (e.g. how to ride a bike)

Automatic activation of word (e.g. reading)

– Single response mechanism

Reaction to one stimulus

Not flexible

Cannot slot this type of knowledge into a schema form – each skill is

separately stored from other skills

Unconscious recollection

– Priming

Habituation

Located in cortex, amygdala, striatum and cerebellum

Habituation

You become familiar to a stimulus to a degree where you no

longer consciously perceive it

– You become habituated to the sound of your own heartbeat

– Sound of the clock in your study

– Noise from traffic outside your window

Recognise and ignore familiar unimportant stimuli

Method commonly used for training animals

– Gun shyness in dogs

– Skittishness to car sounds in horses

Aplysia

• Marine snail (sea slug)

• Has 20.000 neurons

• When gently touching inner

parts, it will contract

• After 10 times = habituation (it no longer

contracts)

– Only lasts for 10-15 minutes

• After 4 days (10 times each day)

– Last 3 weeks

• Habituation caused by weakening/decrease of

synaptic connections between sensory neurons and

neural circuit that produces the behaviour (reflex

contraction)

04/04/2012

4

Sensitisation

Habituation = occurs with benign stimuli

If stimuli is harmful, sensitisation occurs

Aplysia received a shock to its tail = immediate

and firm reaction

– Single shock = reacts for minutes

– 4-5 shocks = 2 days

– Further shocks = weeks

Sensitisation caused by strengthening / increase in synaptic connections between sensory neurons and neural circuit that produces the behaviour (reflex contraction) Same set of synaptic connections could be affected in opposite directions by different forms of stimulation.

Declarative

Located in across the cortex

Episodic (events)

Semantic (knowledge)

– episodic and semantic are declarative types of memory

Amnesiacs have impaired declarative memory

Involves learning of associative connections

– Linking item with context

– Linking two memories together

– Occurs in hippocampus

Schemas Semantic memory

General knowledge

Less dependent on context or event

Memory for facts

Self-referential evaluations (e.g. values,

attitudes, traits)

Impaired in schizophrenic

– Inability to differentiate between semantic and

episodic memories

Experiment

Draw a penny (try and draw both the ‘heads’

and ‘tails’ side)

Get as many details as possible correct

Episodic memory

The ‘What-Where-When’ memory

Stimulus triggering one (e.g. Where) will trigger the

others (What & When)

Personally experienced past

– Specific events

– General events

– Flashbulb memories (more next week)

– Autobiographical knowledge

Personal facts

– Subjective experience (more next week)

– Context dependant

– Time-based

04/04/2012

5

Autobiographical memory

Memory from events of our own lives – our

individual past

Events and issues related to one self

Personal facts

Emotionally salient events remembered better

Time of recall important for reminiscence

bump (from Rubin, Wetzler & Nebes, 1986)

When asking people at

age 50 – most

memories recalled

were in childhood +

recency effect

Reminiscence bump (from Janssen et al., 2005)

Cross-sectional, episodic memories.

Most events remembered around age 13-18

Are more events

recalled in teenage

years as the episodes

have been rehearsed

more; or enriched

environments with

more material to

remember from this

age?

Episodic / semantic distinction – evidence from

Amnesiacs (e.g. Warrington, 1986)

– Impaired episodic memory (who am I?)

– Able to recall semantic and procedural knowledge (can still speak, tie shoe

laces, etc.)

– However, semantic impairment subsequently demonstrated in many cases

– Also suggested overlearnt skills remain, but new skills cannot be acquired

– Episodic memory of early childhood remains in almost all patients

All early learning remains – no distinction between episodic and semantic

memory

Brain scan studies

– Semantic recall activation in left temporal lobe

– Contextual episodes recall activation in right pre-frontal area (Shallice et

al., 1994)

Amnesiacs

Physiological (organic)

– Retrograde amnesia (past knowledge gone)

– Anterograde amnesia (cannot form new

memories)

Psychological

– Traumatic events repressed

04/04/2012

6

The case of H.M. (1926-2008)

• Had most of his hippocampal region surgically removed

(to alleviate life-threatening epilepsy)

• Following brain surgery, intact STM but impaired LTM

• Over the next 5 decades unable to form new memories,

e.g. couldn’t recognize psychologist after 45 years of

visiting

– Some part of LTM from before surgery intact

• Could remember schoolmates and things he

enjoyed (roller skating, target practice)

• Couldn’t remember specific episodes (e.g. what

did you do on your 10th birthday?)

– With repeated learning (years) or strong emotional

context appeared to be able to learn a few new

memories

• Could draw a layout of new house he stayed in

• Could identify ‘Lee Harvey Oswald’ (“he

assassinated the president”)

• Intelligence (IQ of 112 after operation), language skills

and personality remained intact

Implicit learning intact

• However, could learn

implicit tasks (tracing a star

in mirror) although had no

recollection of performing

task previously

• In other tasks, showed

effects of priming

Amnesiacs

• The case of Clive W (musician, brain damaged)

– Impaired episodic and semantic memory, but intact

procedural memory

– Completely unable to acquire new memories (will

conclude every few seconds that he has just woken up

from unconsciousness)

– Cannot remember any specific episodes of his past life

prior to onset (still recognises his wife, but not friends

or children)

– Cannot remember past facts (e.g. unable to recognize

a photograph of the queen)

– Cannot recognize common objects or words (e.g.

‘tree’, ‘eyelid’)

– Personality remain intact.

– Still able to play piano and read music with great

skill, able to make tea (knows where cupboard and

tea is kept) in current location, and read and write.

[Show clip 1b of procedural memory, play clip of diary of STM, and 10 yrs later recent clip]

HM and Clive W

Intact procedural memory

Intact STM

Declarative memory impaired

Unable to form new memories

Criticisms of amnesiac research – case studies

No clear dissociations:

– Heterogenous

– Usually patients suffer from mix of retrograde and

anterograde amnesia, some impairments to all types of

LTM

HM – recall aspects of childhood, but impaired

memory of 11 years preceding surgery

Clive W – unable to differentiate between honey & jam

Difficult to determine if impairments are

– Specific modular type of memory (storage)

– Process of memory (encoding/retrieval)

Consolidation period

From patients with hippocampal damage

– Often cannot remember events occurring in the

time before damage

– Whereas older memories remain intact

– Suggests that newer memories still depend on the

hippocampus

– The consolidation period takes longer time (days –

years)

7

Sleep and consolidation

If the memory trace is not consolidated via

sleep, it remains fragile to disruption

– Mice deprived of REM sleep for 2 days show

amnesia compared to control animals (Fishbein et

al., 1971)

– During time of learning in mice & rats, increased

amount of REM sleep, compared to control

controls (Smith et al., 1974; 1980)

Neural activity during sleep

The brain areas active when awake are the same areas

active when asleep

When specific cells in the hippocampus activated during

awakeness, they were activated to a higher degree during

sleep

Novel representations are strengthened during sleep in the

hippocampus (neurons fire at a rate = theta rhythm) linked

to long-term potentiation (= increases in synaptic

connections)

Older representations are weakened in the hippocampus

(fire out of theta rhythm phase, a situation linked to

depotentiation = decreases in synaptic connections)

Sleep and consolidation

Sleep deprivation effect on learning show a

decrease in performance only if it involves a

novel element

Amount of REM sleep increases in humans

following learning tasks

Participants when left with the same post-training

interval show those who slept in interval show

marked increase compared to those who had not

slept (see Stickgold, 2005 for a review)

Awake first. ppts

trained at 10 a.m. [day

1, green bar] - no sign.

change at retest

following 12 h of being

awake [10 p.m., day 1,

green bar]. By 2nd

retest, following a night

of sleep [10 a.m. on

day 2, red bar],

improved sign.

Asleep first.

Following evening

training [10 p.m., day 1,

green bar] sign.

improvements just 12 h

after training following

a night of sleep [10

a.m. on day 2, red bar]

but w no further sign

change

following an additional

12 h of being awake

[10 p.m. on day 2, red

Even a midday 90-minute nap lead to 16% improvement. Re-testing at bar].

72 hours show more improvement than after 24 hours.

Criticisms of sleep consolidation studies

Time of learning and recall differs between controls

and experimental groups

You cannot with certainty differentiate performances as caused by sleep, rather than as an effect of time.

There are diurnal fluctuations in neurochemical activity (e.g. growth hormone is released during the night) – this, rather than sleep, could be the cause of consolidation

Sleep causes a cessation of mental activity and sensory impressions, could be associated with less interference (hence better performance) rather than memory consolidation (Baddeley, 1997)

Context effects