Memory consolidation

Question 1

memory consolidation

Answer 1

1. Cellular (synaptic – strengthening/amount of connections between 2 neurons)
2. System-level (whole brain)

‘Offline neural changes that lead to memory stabilisation, enhancement, and integration with pre-existing knowledge’

Question 2

Mueller and Pilzecker (1900)

Answer 2

New memories initially fragile, but soon resistant to interference (e.g. new learning, shock to the brain, stressor, resistance to drug)
o Lots can happen in period of solidification

Ps learnt list of paired-associate syllables (AB) and tested in cued recall (using first syllable, A)

Interpolating another list (CD) impaired memory of first list (AB)

Found temporal grad, whereby closer in time interfering list to target list, stronger amnestic effect

Memories require time to consolidate

Retroactive interference compromises integrity of recently formed – but not yet consolidated – memories

Interference ‘nonspecific’, i.e. interfering material doesn’t have to be similar to target material (i.e. AB v CD)

Mental exertion is interfering force
o Engage neurons into new learning
o Detrimental to consolidation overall

Question 3

exp 34

Answer 3

see notes

Temporal grad

Question 4

the ‘clay sculpture’ metaphor - Wixted and Cai (2014)

Answer 4

Memory initially at finest, but fragile; over time, becomes resistance to interference, i.e. shows less and less potentials for damage

Question 5

shape of forgetting function - Wixted (2004)

Answer 5

Clay metaphor fits well with forgetting curves

Rate at which we forget not constant; would be property of memoryless systems

Ebbinghaus (1885): forgetting function is such that we forget less and less as time goes by
o Good learners also good consolidators

Jost (1897): if 2 mems have equal strength but diff ages, older trace will decay at slower rate
o Older memories already had time to solidify

Continue reduction in forgetting rate is sign of consolidation

see notes

Question 6

1. cellular consolidation

Answer 6

Occurs at neuron level (not at whole-brain “systems” level)

Takes place during first hours (or days) after initial memory formation in hippocampus
o Storing dec memories

Fits well with idea of trace-hardening phys process put forward as far as Mueller and Pilzecker

Corresponds to discovery of LT potentiation (Bliss and Lømo, 1973)

Question 7

LT potentiation - Bliss and Lømo (1973)

Answer 7

Long lasting enhancement of synaptic efficacy induced by tetanus (short burst of high-freq stim) to presynaptic neuron

see notes

When drinking – brain stops encoding info – help consolidation of what you learnt in afternoon

No interference of what you are doing/encoding of new memories

Question 8

retrograde facilitation

Answer 8

If subsequent encoding (mental exertion) interferes with memory consolidation, factors blocking new encoding should promote memory stabilisation
o Alcohol
o Benzodiazepines (anxiolytic drug)
o Slow-wave (non-REM) sleep
 Mainly occurs earlier/first half of night
 Not learning anything new – promote consolidation of precious memories
o Resulting anterograde amnesia accompanied by retrograde facilitation: memories formed prior to drug intake/sleep forgotten to lesser degree than memories formed prior to placebo/wake

Question 9

retrograde interference/fac

Answer 9

see notes

Question 10

2. systems consolidation

Answer 10

HM’s bilateral medial temporal lobe (MTL) resection:
o Anterograde amnesia: inability to form new declarative memories (the ‘what’) – learning new facts
o Temporally graded retrograde amnesia: impairment of memories formed prior to surgery, stronger for young than old memories (Scoville and Milner, 1957; Ribot’s law, 1881)

Declarative memories becoming independent from hippocampus and more dependent on neocortex referred as ‘systems consolidation’

Question 11

the hippocampus

Answer 11

Episodes in life represented in hippocampus v. quickly

Essence of memory captured by hippocampus – puts memories together to make them coherent

Disengages in long run once memories form

see notes

Question 12

temporal gradient of semantic memory - behaviour

Answer 12

Bayley et al. (2006; Manns et al., 2003): 6 amnesic patients w/ damage limited to hippocampal region

Answer Qs about news – 1951-2005

Old memories preserved in patients

see notes

Question 13

temporal gradient of semantic memory - brain activation (in controls)

Answer 13

Smith and Squire (2009): 160 Qs on news events over 30 years

Trade-off between hippocampus and outer layers of brain – expected to be given something from hippocampus

see notes

Question 14

the same temporal grad over 24h

Answer 14

Shift from hippocampal to neocortical centred retrieval network with consolidation (Takashima et al., 2009)

Systems consolidation can occur v. quickly

Richness of memorised materials may determine how fast systems consolidation occurs

see notes

Associated face w. orientation of arrow

Test phase next

2 lists of orientations that don’t overlap – day before/just before test

Hippocampus used to initially code info

Question 15

what is the sig of this shift go activation from the medial-temporal lobe to the neocortex?

Answer 15

Declarative memories stored in neocortex from the outset (e.g. sensory and semantic areas)

Hippocampus acts as a relay station and binds them together

Over time, cortico-cortical associations develop (due to the hippocampus), such that these memories become independent of the hippocampus

Traces for given event segmented – hippocampus connects them together

Sent back to neocortex when sleeping

Question 16

complementary learning systems - (Marr, 1971; McClelland et al., 1995)

Answer 16

A fast learning system that holds info only temporarily (medial temporal lobe, i.e., the hippocampus)

A slow learning system that serves as LT store (i.e., the neocortex)

Main idea:
o Info initially stored in hippocampus progressively fed back into neocortex (via ‘neural replay’), so that pre-existing knowledge can accommodate newly learnt info – hippocampus is seen the internal sparring partner of neocortex
o As both systems used to encode new learning, reactivation/redistribution has to occur offline, e.g. during sleep and with special properties

Question 17

sleep and memory

Answer 17

see notes

Sleep stages characterised based on characteristics

Vary in how hard it is to wake the person

3 and 4 – neocortex produces slow-wave oscillations = slow-wave sleep – likely memory consolidation occurs here

Cycles of 90 minutes

REM sleep increases throughout the night and deep sleep decreases

Circadian effects also modulate stages

Question 18

minor studies from the psych lab of Cornell Uni - Jenkins and Dallenbach (1924)

Answer 18

2 Ps learn lists of 10 nonsense syllables until complete mastery

Re-tested in free recall after a varying time interval (1, 2, 4 and 8 hrs) filled w/ sleep/wake

Sleeping protects against forgetting

However, absence of interference not whole story: actual role for memory consolidation

Less exposures needed in morning than evening

Less crowding/learning of other information whilst asleep

Sleep closer to learning = better than when later on

“Results of our study as a whole indicate that forgetting is not so much a matter of the decay of old impressions and associations as it is a matter of the interference, inhib, or obliteration of the old by the new”

see notes

Question 19

effects of early and late nocturnal sleep on declarative and procedural memory - Pilhal and Born (1997)

Answer 19

Tested the differential effect of sleep composition

Compared declarative and procedural memory

Paired associated learning

Early = woken up at 3 in the morning to train – mostly slept with slow wave sleep

Wake = have to stay up until 3

see notes

Diff between sleep and wake driven but what happens in first half of night – slow-wave sleep – good for remembering things you can declare

Don’t differ in second half – REM sleep

Procedural tasks – better in second half – procedural learning boosted by REM sleep

Double dissociation: declarative memory promoted by slow wave sleep; motor skills improved by REM-sleep

Role of REM sleep uncertain

Question 20

memory for semantically related and unrelated declarative info: the benefit of sleep, the cost of wake - (Payne, Tucker, Ellenbogen, Wamsley, Walker, Schacter and Stickgold)

Answer 20

The ‘sleep-first’ effect
• Learning of related (‘circus-clown’) or unrelated pairs (‘cactus-brick’) using study-test cycles with feedback until 24/40 correct
• After 12hrs, better perf for sleep group, only for unrelated pairs
• Temporal grad a retroactive fac: after 24hrs, better recall for those who slept first
• Absence of interference and system consolidation during SWS could be behind effect, though logically cellular consolidation should be blind to semantic relatedness

see notes

Less forgetting when sleep early – maintain information when comes closer to training

Non-related show imp of sleep v wake

No retroactive interference when you sleep

Question 21

neural replay - (Maquet et al., 2000; Wilson and McNaughton, 1994; Rasch et al., 2007)

Answer 21

Over a 24h period, there are privileged moments (perhaps when not much encoding is happening) during which the brain spontaneously replays to itself info recently acquired

Done mostly unconsciously, thought some could reach consciousness

Allows other brain regions to learn the info in Q

Slow-wave sleep (SWS) appears as key window
o Cannot learn anything in SWS
o Hippocampus changes functioning early in night and shuts down processing new info

Question 22

SWS and neural replay

Answer 22

see notes

According to the model of system consolidation put forward by Born and Wilhelm (2013), during SWS, slow oscillations occurring in neocortical regions constitutes a signal sent via thalamus, to hippocampus to reactivate hippocampal memories

Neocortical oscillations drive thalamo-cortical spindles, which themselves drive spindle-ripple events in the hippocampus: the alignment between levels is strong and controlled always by troughs at level immediately above

Neocortex saying to hippocampus: ‘alright, now is a good time to tell me what you know’

Question 23

“cuing” - Rash et al. (2007)

Answer 23

Cued reactivation using an odour also present during the learning phase

Used spatial memory as skill

Re-exposure to associated odour during SWS reactivated hippocampal areas active during learning

Also led to enhanced memory perf next day
o Only SWS and same odour produces effects

Bias neural replay process

see notes

Question 24

sleep makes previously inaccessible memories accessible - look more carefully: even your data show sleep makes memories more accessible - Dumay (2018)

Answer 24

Net perf (i.e. sums, %) hides the presence of 2 opposing forces at the item-level: forgetting (inability to recollect previous knowledge) and reminiscence (ability to access knowledge inaccessible until then)

Usual decline in perf smaller after sleep than after wake doesn’t mean that sleep just prevents forgetting

Separated out ‘maintained’ items (i.e., accessible at both the 0hr and 12hr tests) from ‘gained’ items (i.e., inaccessible at test, but accessible at retest)

Sleep found (red) to increase prob of gaining access to previously inaccessible knowledge in both recall (fig a.) and recog (fig. b) and (blue) to prevent forgetting beyond wakefulness only in recall

Sleep doesn’t just stabilise memories, it makes them more accessible

see notes

On both graphs, protection against loss and reminiscence measured proportionally to max no. of items that could actually be lost (i.e. sum of correct at test) or gained (i.e. total no. of items in too-be-learnt set-sum of correct at test)

0 on y = 50/50 situ, where no. of items maintained = no. of items lost, and like for gained and un-gained

Pos values mean more protection against loss and reminiscence than forgetting and non-reminiscence; neg values mean opposite

In all cases, except for protection against loss in recognition tasks, sleep has more pos values than wake

0 = midpoint

Question 25

sleep promotes gist extraction - the role of sleep in false memory formation - Payne et al. (2009)

Answer 25

Relied on the Deese-Roediger-McDermott paradigm to examine the influence of sleep on memory formation and distortion Ps learns 8 lists of 12 semantically related words, all strongly associated with a missing critical target, e.g. door, glass, pane, … for window Overnight sleep (fig. 1) and napping (fig 4.) increased no. of false memories (‘critical words’) – no such effect for mere instructions Sleep strengthens associations between indv memory elements and fills in the gaps see notes Duration of sleep provides enough substrate to improve perf Napping may not be as strong a driving force More false memories for critical words and unseen words when sleep – benefit of being asleep for words not presented at mere exposure

Question 26

sleep integrates new info w/ long-consolidated knowledge - sleep associated changes in mental representation of spoken words - Dumay and Gaskell, 2007)

Answer 26

Cathedruke impedes recognition of initial word (cathedral) No influence if you remained awake Interference of learning cathedruke if you slept Assimilation of the novel words in lexical memory appears to occur while we sleep, as part of consolidation (Wang et al., 2017) Free recall shows enhanced memory perf (hypermnesia) after sleep, but not after wake see notes and slides Improve lots during interval = more interference from novel word onto existing word Tied to sleep stages

Question 27

memory consolidation and emotions - emotional memory formation is enhanced across sleep intervals w/ high amounts of REM sleep - Wagner et al. (2001)

Answer 27

Examined whether sleep preferentially enhances memory for emotional narratives Ps memorised details of 2 small descriptions (inc. 94 content words): e.g. “manufacturing bronze sculptures” v “child murdering” No. of content words type-recalled was memory measure Text retention, especially if content emotional, benefitted from more late-night sleep see notes REM sleep plays key role in enhancing emotional words – boost information of emotional nature (neg)

Question 28

sleep preferentially enhances memory for emotional components of scenes - Payne et al.

Answer 28

Looked at the role of sleep on memory for object and background components of scenes Main components either neg/neutral, whereas background always neutral Sleep enhanced recog memory for emotional components to detriment of background details No such overnight trade-off for neutral scenes Overnight trade-off between emotional objects and background details Y axis = after amount of training see notes

Question 29

Setting the alarm: word emotional attributes require consolidation to be operational - Dumay et al. (2018) - general information

Answer 29

Do surrounding emotions taint something that is neutral in first place – might in future become alarming Application for clinical psychology Neg words might influence pseudowords Will word itself become alarming

Question 30

Setting the alarm: word emotional attributes require consolidation to be operational - Dumay et al. (2018) - slide set-up

Answer 30

Pseudoword in centre Neutral scenes surrounding it Associate word with photo Feedback = correct question Association simple Can be applied several times during exposure phase Can remove feedback – block learning – measure of learning rate as time goes by Alarming words might be easier to learn – attracted by the scene Episodic memories easier to learn is alarming Will context boost learning of neutral word if alarming? Emotional words capture attention – other info processing therefore slowed down How much emotion required to capture attention? see notes

Question 31

Setting the alarm: word emotional attributes require consolidation to be operational - Dumay et al. (2018) - whole exp set-up

Answer 31

Bypass practice effects One set learnt a week before test Another set 5 mins before Different age of knowledge by time have memory assessed Tests applied immediately on day 8 Whether alarming system integrates info automatically – alarmed even if just see sequence of letters, not just context Test notion of sensitisation Set 2 might sensitise Ps – make Ps fragile/tuned to emotions More chance of finding emotional capture if straight after learning – sensitised – acts as a primer to increase chance of seeing emotional effect in tests ½ of group waited 6hrs to take test Sensitisation period should have faded away – less likely to find emotional effect Emotional Stroop - visual task – uses same modality as learning – play/make decision in auditory domain – insert disruption in speech in some trials – presence/absence Words learnt on same day – predict would not have made to procedural system and so unlikely to become alarming by time take test see notes

Question 32

Setting the alarm: word emotional attributes require consolidation to be operational - Dumay et al. (2018) - RT results

Answer 32

Raw RTs for detecting disruption in speech First bar = baseline condition White = string of letters – paired with alarming event Black = pictures all neutral No diff when learning and test on same day Week before = alarming pairs grab attention compared to neutral pairs – something has changed over course of week – emotionality of context endorsed by words – can have attention capture, assuming consolidation Look at group tested immediately/6h later doesn’t make any difference – need period of consolidation for alarming effect to take place see notes

Question 33

Setting the alarm: word emotional attributes require consolidation to be operational - Dumay et al. (2018) - combined results

Answer 33

Include all pairs – has knowledge faded way? – does time make any difference? Emotional pics = disturbance – don’t take decision as immediately as if pics neutral RT always longer in white than black – during exposure and at test – emotional condition slower Memory test – which photo word relates to May be decision to push right decision because of disturbing photos Reasoning happens as soon as diff condition Attention glues to element of photos and so progression to next trial takes time and RT slower Misleading hint that alarming condition harder to learn – not right – look at evolution of learning at test 1, 2 and whether neutral/emotional – learning proceeds in similar fashion whether alarming/neutral context – same rate of learning – conditions always parallel Emotional words not harder to learn With consolidation, things at first neutral might be alarming due to experiential context depending on emotionality of context – take reasoning into clinical setting Find ways concepts become alarming depending on episode in life – biography see notes