Components of memory - University of Leicester
Download
Report
Transcript Components of memory - University of Leicester
PS3002: Brain & Cognition
Cognitive neuroscience and memory
John Beech
School of Psychology
University of Leicester
1
Memory and the brain
Introduction
• Memory is extremely varied. What is the substrate of
memory in the brain?
• Is it different for different types of memory?
how to ride a bike
the relation between the sides of a triangle
an episode in one’s life
• Can use:
animal models from simple invertebrates to
mammals
the study of what happens to memory in amnesia
following brain damage
brain imaging in volunteers to investigate normal
encoding, storage, and retrieval – recall or
recognition.
2
Structure of memory
Learning vs memory
• Squire (1987) distinguishes:
Learning - process of acquiring new information
Memory - persistence of learning in a state that can
be revealed at a later time.
• Learning has an outcome - memory - which itself has
a further outcome - a change in future behaviour.
• Learning need not imply any conscious attempt to
learn. Simple repeated exposure can, and indeed
usually does, lead to learning, and this is evinced by
memory.
3
Structure of memory: encoding
Encoding is divided into acquisition (registration) and
consolidation
Storage creates and maintains a permanent record
Retrieval uses stored information to create a conscious
representation, or to execute a learned behaviour.
• If there is a deficit in memory, we don’t know at which
stage this has occurred.
• [flashbulb memories at times of traumatic or shocking
events tend to be very vivid and detailed and readily
recalled, but some studies have shown that it is not so
much their accuracy which is improved, as the increase in
subjective confidence].
4
Time base of memory
Memory model of Atkinson &
Shiffrin (1986) above.
Sensory memory is subsecond to seconds, as when
we can recover what was said
when we weren’t paying
attention.
Short term is seconds to
minutes, as with retaining a
phone number.
Long-term is longer—days,
weeks, going up to years, or
even a lifetime.
5
Short-term memory
Peterson & Peterson
(1959):
Consonant trigrams
(e.g.TDK) followed by a no.
(e.g. 765). Counts back in
3s (765, 762, 759…) and
recalled trigram on signal
Rapid decline to <10%
after 18s.
Thus if rehearsal inhibited,
decay is rapid. Is this
decay over time or the
processing of new
information?
6
STM v LTM
Glanzer & Cunitz (1966):
Recalling list of words
give U-shaped effect
Expt 1: varied
presentation rate.
Improved recall on
start and middle bits,
but not on the last part.
Slowness helps LTM
but not STM.
Expt 2: Immediate recall
vs delayed
interference (counting):
affects only last part
showing an effect on
“recency”.
7
Sensory memory
Sensory memory
Neisser (1967): “iconic
memory” (E.g. cigarette
or sparkler in dark) and
“echoic memory” (e.g.
saying “what?” and then
knowing what was
said).
8
Sperling
• Sperling (1963) - matrix shown 50ms, then tone
signalled top, middle or bottom of array to recall.
• If 67% of row recalled, he argued that 67% of whole
array must therefore be available.
• Varying probe length showed steep decline in
memory to asymptote after only 250ms ( .25s).
• Not stored by analysis of meaning.
9
• Baddeley (1974) proposed that a
central executive controls two
storage areas.
• The phonological (articulatory)
loop and the visuo-spatial
scratchpad. The Central
Executive coordinates these
activities with LTM.
• Phonological (articulatory) loop is
essentially acoustic, because
errors in recalling letter strings
presented visually are letters
similar in sound rather than in
form.
• Visuo-spatial scratch-pad is
separate, and the introduction of
a V-S task during retention does
not interfere with recall of PA
material.
Working memory
10
Long term memory: different forms
11
• LTM is regarded to be in 2 parts: Explicit and
Implicit (Tulving and Schacter, 1990), OR
Declarative and Non-declarative (Squire,
1987; Cohen 1993).
• Explicit or Declarative is conscious
knowledge, to which we know we have
access: facts, history
• Implicit or Non-declarative is unconscious
knowledge, such as procedural, perceptual
priming, conditioned habituation or
sensitisation.
Long term memory: different forms
• Analysis of nature of memory deficits will
help to explore this distinction, as these 2
types are differentially affected by lesions.
• Explicit consists of episodic (stories about
us, what happened in the past) and semantic
(facts).
• Implicit memory (Schacter) - knowledge that
can be revealed by behaviour in an
appropriate test situation, and that can be
observed in animals too. Humans do not
retrieve it under conscious awareness,
although they can do.
Implicit vs explicit
memory
Below Tulving &
Schacter
12
Implicit memory in
the perceptual
representation
system (PRS)
• If Ps are shown
a list of words,
their explicit
memory can be
assessed with a
new list
containing old
and new words.
• Implicit memory, on the other hand, is demonstrated by improved
performance on a word completion task, even without conscious memory of
the previous list.
• In contrast to explicit memory, the latter improvement is not time dependent,
and performance is not further improved by encoding the original list more
deeply.
• Hence explicit and implicit are truly separate memory processes, at a
psychological level.
13
The PRS (Perceptual Representation System)
• Priming studies have primarily been used to
demonstrate the PRS (as in the previous
experiment). In that experiment if the initial list of
words is presented auditorily, the gain in implicit
memory isn’t seen. Thus the effect is specifically
on the visual word form, not on the auditory
counterpart.
• Priming can also be shown on nonverbal stimuli.
Daniel Schacter et al. (1990) showed pictures of
both possible and impossible forms (shown in the
next figure) and on the initial viewing half the
participants were instructed to look at each picture
globally. Then the pictures were shown once more
with the instruction to decide if each one was
possible or impossible. Those instructed to look at
the pictures globally were faster at doing this, but
only for the possible objects, not for the impossible
ones. Thus these showed priming for forms.
• PRS develops early, and is preferentially
maintained during ageing.
14
Schacter et al. (1990)
• These figures to the right are
examples of possible (top) and
impossible (bottom) figures used by
Schacter et al.
• Evidence from this and other studies
suggests that implicit memory in
priming involves a perceptual
representation system (PRS) that
subserves processing pictures and
visual and auditory word forms.
15
The neurobiology of different types of
memory
• Sources of information: human brain injury, animal lesions,
imaging of brain during memory processes - the usual
approaches (for us).
• Deficits in memory as a result of damage, disease, or
psychological trauma are called amnesias.
• May be loss of ability to learn new things, and/or loss of
previous knowledge.
• The fact that different components can be lost, helps us to
understand the organisation of memory.
16
The reality of amnesia
• Suppose you meet someone who appears to have perfectly normal
intelligence and behaves normally in terms of social conventions.
• But you have to leave the room to meet someone briefly and then you
return about 3 minutes later.
• You find that when you return this person doesn’t have the slightest
recollection that they’ve met you before and that they’ve been talking to
you.
• Suppose further that you’re a clinical psychologist and you can perform
further tests and when you do you find that in practice this person cannot
remember anything beyond 50 seconds.
• When you probe for their memory of recent events (e.g. by news stories)
you find that they have complete amnesia for many years of the past!
• What must it be like to be deprived of your past like this?
• How can such a person function and how do they relate to their family
and friends?
• Their memory for past events is “wiped clean” (although we will modify
this later) back to the time when their brain was damaged AND
memories some years before the damage took place.
• Before we proceed we need to examine the structure of the temporal
lobes which are strongly implicated in this condition…
17
The temporal
lobes
In the MRI above TP is the “temporal
pole” – the front of the temporal
lobe. In the inner surface of the
temporal lobe is the hippocampus
outlined in white and marked HIP.
AMG is the amygdala.
18
The temporal lobe
19
20
A
B
C
D
Anatomy of the temporal lobes
Key to the anatomy
The pictures are arranged
A-D going left to right,
top to bottom. A: Dorsal
lateral view, rear of
head to the right. T
Lobes in pink. B:
Midsagittal – section
down the middle with the
rear of the head to the
right. C. Basal – looking
at the base – note how
the lobes (in pink) wrap
round but don’t meet. D.
the coronal view – a
section “ear to ear”
cutting the brain into
front and back halves.
Click.
21
The temporal lobe and memory
• 1940s and 50s: neurosurgical treatments for
epilepsy.
• Removal of medial temporal lobe, including
the hippocampal formation, resulted in
dramatic memory impairments, only if
bilateral.
• Patient HM. Worst case in a series of 10
patients reported by Brenda Milner. (Born in
Manchester and moved to Canada after her
degree at Cambridge.)
• Increasing frequency of his temporal lobe
epilepsy led to bilateral surgery – 1953
when he was 27 years old.
• He remained of normal intelligence (IQ 112)
and had no psychological illness.
However, the surgery resulted in intense
anterograde amnesia (Events taking place
after surgery are never remembered for
more than 60 sec).
22
The temporal lobe and memory
•
•
•
•
Two years later he reported the current date as being that of his surgery.
He failed to recognise doctors on their return, who had left the room only briefly.
Had clear recollections often only of distant events. Partial loss for 3 years preceding
surgery (retrograde amnesia), and complete loss for events after (anterograde
amnesia).
Short term memory was normal (sensory registers, immediate memory, digit span* –
as is the case for other amnesics). Transfer to longer term memory was what was
disrupted.
(* note = digit span here is referring to starting with 4 numbers, and if correct adding a
new number, to make 5 in a different sequence. This continues until errors are made
in two trials for a given sequence.)
23
The temporal lobe &
memory
Areas removed from HM
shown in red. Note that
the resection is only
illustrated here on one
side, but HM’s lesions
were bilateral. The very
top diagram is an
underside view of both
hemispheres. A-D are the
4 sections at the bottom
of the figure. (Sections A
and B, where the areas
were removed, are
nearer the front of the
head and C and D, where
there is probably atrophy,
are situated at the back.)
24
Learning digit sequences
The figure on the right shows digit
span for amnesic patients (blue)
and controls (red). They were
given 5 numbers and had to
repeat these back. If correct
another number was added to
the next sequence. When they
were incorrect the same
sequence of numbers was
repeated until they could report
it correctly. It can be seen that,
e.g. for 8 numbers it only takes
about 3 trials on average to
learn this sequence for controls
but about 12 trials for amnesic
patients. Conclusion: they find
learning (transferring into LTM)
very difficult.
25
The temporal lobe and memory
• Returning to HM, on the other hand, implicit memory was
normal, and skills improved normally
– e.g. drawing a star seen in a mirror.
• The area of hippocampus removed in HM turned out to be
smaller (and more anterior i.e. nearer the front) than at first
reported. However posterior hippocampus may have been partly
atrophied.
• Another case, RB had similar pattern of amnesia after an
ischaemic episode (i.e. reduction of blood to the brain) during
by-pass surgery. Had specific lesion restricted to the “CA-1”
cellular subfield of each hippocampus (which is particularly
vulnerable to ischaemia). This case was important because it
showed that hippocampus damage was sufficient and that the
amygdala didn’t need to be damaged as well.
• In summary, the hippocampus appears to be particularly crucial
in the laying down of new explicit memories.
26
The temporal lobes and memory
• Much earlier memories are not impaired, thus hippocampus is
not a permanent storage area for explicit knowledge.
• hippocampus is involved [with other cortical areas] in
consolidation, a longer term process taking months to years
(note retrograde amnesia in hippocampus lesion patients for up
to 3 yrs).
• Consolidation is understood to involve biological changes taking
place in those other areas of cortex, and involving strengthening
of the associations between multiple stimulus inputs and
previously stored information.
• Once this has fully taken place, the hippocampus is not required
for retrieval.
27
The
temporal
lobe and
memory
• This is also illustrated by the amnesic effects of electroconvulsive therapy,
which typically extend for 6 months before treatment and 2 months after (i.e.
Memories are in labile [easily altered] form in this amnesic period).
• Memory impairment from physical trauma (e.g. a car crash) is similar.
• Damage to cortex, especially areas in front of the hippocampus, such as the
front part or anterior temporal cortex, results in a dense retrograde
amnesia going back several decades, but with preservation of the ability to
form new memories.
• Hence the anterior temporal lobe is a major repository for LTM. BUT it is not
the only possible one, since new long-term memories forwards in time28from
the lesion can be formed.
Features of amnesia
What are the features of the amnesic syndrome?
• Can learn some new information with difficulty, e.g.
semantic knowledge, new vocabulary, details related
to people. Typically, patients are then unsure of the
source of this information.
• There or no great intellectual impairments, nor any
problems in language (i.e. aphasia).
• Retrograde amnesia – there are partial losses of
information that has been acquired before injury. For
example, HM lost some knowledge in the period 3
years leading up to his operation. But he can recall
older memories very well.
29
Features of amnesia
What are the features of the amnesic syndrome?
(continued)
• Dense anterograde amnesia – a permanent
problem with acquiring new information. For
instance, can’t remember names of new people, nor
his way back home – HM’s family moved house after
the operation.
• Normal STM – HM has normal digit span (7 forward,
5 backwards). Can carry on conversations including
rephrasing sentences, do mental arithmetic.
30
Features of amnesia
What are the features of the amnesic syndrome?
(continued)
• Rehearsal – if HM is distracted just for a moment
from what he’s rehearsing he’ll completely forget it –
thus rehearsal does not serve to transfer to LTM.
• Learning is not completely absent – he shows
unimpaired implicit memory on procedural tasks by
showing priming effects equivalent to controls and he
acquired the classically conditioned eye-blink reflex.
He also learned mirror drawing. Skills such as riding
a bike – procedurally based skills – are preserved
and amnesics can acquire new skills after injury
without any problems.
31
Alcoholic Korsakoff’s Syndrome
• Sergei Korsakoff at the end of the 19th
century reported anterograde and
retrograde amnesia associated with
alcoholism. Imbibing a lot of alcohol in the
long term produces vitamin deficiencies
leading to brain damage (thiamine
deficiency (vitamin B1) => periventricular
damage). This is more specifically in the
diencephalon* (thalamus and mammilliary
bodies). The same kind of damage can be
produced by strokes or tumours. Recent
work by Sullivan & Marsh (2003) has also
found deficits in the hippocampus and not
just the diencephalon.
(*Note that the diencephalon is a subcortical
[portion of the brain just below the cerebral
cortex] structure made up of the thalamus
and the hypothalamus.)
32
Alcoholic Korsakoff’s Syndrome
• The similarities in Korsakoff’s (implicit
memory OK but lose explicit memories)
has led people to suggest that the
medial temporal lobe is not the only
part of the brain for forming explicit
memories of facts and events.
However, the recent work of Sullivan
and Marsh casts some doubt.
• Korsakoffs patients also confabulate
caused by frontal lobe damage. This is
not deliberate deception as they
appear to believe themselves. This
could be due to damage to frontal
lobes which have a control function in
memory retrieval.
33
Korsakoff’s syndrome
Cermak, Naus & Reale
(1976) demonstrated
the memory
impairment (in LTM)
in this condition.
Note the strong
recency effect
showing an intact
STM.
34
•
•
1.
As previously mentioned,
damage to the front part
(anterior) of the temporal lobe
produces dense retrograde
amnesia (forgetting of events
before the brain damage). In
the case of Alzheimer’s disease•
or herpes simplex encephalitis
this can go back for decades
before the amnesia.
•
Temporal gradient (Ribot’s
Law, 1882) autobiographical
memories that are more recent
are much more easily damaged
compared to early memories. It
was first noted in Korsakoff’s
syndrome patients. But WHY
does it occur?
Korsakoff’s patients were in a
more drunken state when
learning! But temporal gradients
occur in amnesics who do not
abuse alcohol.
Retrograde
Amnesia –
temporal gradients
Parts of the
diencephalon,
the thalamus
(blue)
and the
hypothalamus
(red).
35
Retrograde Amnesia – temporal gradients
Temporal gradient (Ribot’s Law, 1882)
1.
Korsakoff’s patients were in a more drunken state when learning!
But temporal gradients occur in amnesics who do not abuse
alcohol.
2.
Squire (1992): 2 major areas involved in initial and long term
storage in memory: (1) The hippocampal formation is responsible
for initial processing and then after rehearsal (2) the medial
temporal lobes consolidate these memories. This predicts that
those with lesions to the hippocampus (e.g. HM), but who have
intact anterior temporal lobe structures (as in early onset
Alzheimer’s disease) will have severe anterograde amnesia (no
new memories).
36
Retrograde Amnesia – temporal gradients
Temporal gradient (Ribot’s Law, 1882)
Squire (continued): But those with semantic dementia (e.g. Snowden
et al. 1989) have the reverse damage – no anterograde amnesia,
but severe retrograde amnesia (able to have new memories, but
recent autobiographical memory impaired).
To digress about semantic dementia – this is also called “fluent
progressive aphasia” meaning that their actual speech is fluent;
however their understanding and recognising of words and
putting names to faces and familiar objects gradually worsens
(i.e. semantic knowledge deteriorates). There is evidence that
the temporal lobes are mainly implicated with memory for words
in the left temporal lobe and for faces in the right (Snowden et al.
2004).
37
Retrograde Amnesia – temporal gradients
Temporal gradient (Ribot’s Law, 1882)
Patients with damage to the anterior temporal lobes have resulting
dense retrograde amnesia (forgetting lots of
events/autobiography before damage) but some patients may
still form new long term memories. So the anterior temporal
lobes are useful for information storage, but other areas also
seem to be capable of acquiring new information. As
mentioned before such patients have poorer recall of
episodes close to the beginning of the disease (or onset of
injury) with gradually better recall further back (i.e. temporal
gradient).
38
Dissociation of episodic memory within
explicit memory
• Tulving et al. (1991) studied a patient KC who had
had subdural hematoma (pool of blood under the
sheath covering the brain). He’d had a motorcycle
crash at 30 yrs of age.
• Damage included bilateral damage to the medial
temporal lobe especially on left, but also other
cortices – frontal, parietal and occipital. More
damage to left hem. (See next figure). IQ of 94.
• Could not remember any events from his life,
although he knew things that pertained to his life.
Thus severe retrograde amnesia - he could recall
very few autobiographical episodes from his life
before the injury. (This is as if his episodic memory
had been wiped clean.) ALSO severe anterograde
amnesia. But intact semantic memory, and
procedural memory.
• Thus this patient had a specific loss of episodic
memory.
39
Tulving et al. 1991
• KC’s lesions are shown on
outline drawings of
horizontal sections of his
brain, starting from the top
moving down going from
left to right in the drawings,
using computed
tomography (CT). The red
parts are the lesions due to
trauma.
40
Tulving et al. (1991)
Implicit memory tested by giving threeword sentences together with a
related picture. In each the final word
critical. Later KC presented with a
fragment or conceptual cue
consisting of the other two words. KC
showed priming effect with the word
fragments and they lasted 12 months.
Priming in this context means that he
was better at generating words from
the fragments compared to words
previously unseen.
Surprisingly he could learn new
semantic information – it took him
longer to learn information than
controls, but he forgot it at the same
rate. Results like these suggest a
separation of the underlying
structures subserving semantic and
episodic memories.
41
Dissociation of episodic memory within
explicit memory
It might be noted that there is currently a controversy and Bayley and
Squire (2002) have suggested that KC’s problems in retrieval of
autobiographical information may be unrelated to his medial temporal
lobe damage because of a more recent report of a patient (EP) who
has more extensive medial temporal lobe damage, but who has better
recall of autobiographical episodes. So we may still be a little way
away from distinguishing the functions of the hippocampus and other
medial temporal lobe structures.
42
Procedural
learning in
amnesia
• The test above examines serial reaction time: Ss have to push buttons
according to flashes of lights in a complex sequence. Their index finger
corresponds to the first light, the second finger to the second light and so on.
The sequence can be random or else they can appear to be random, but in
practice there is a complex repetitive sequence. In the repeated sequence
situation RT reduces over time as shown on the right hand side. Afterwards
it is clear that Ss are unaware when there are these complex repetitions.
This is a good paradigm to test procedural learning (or implicit memory)
• Amnesics (and Korsakoff’s patients) also improve their performance. Thus
they have reasonable implicit memory, while great difficulties with episodic
memory.
43
Double dissociation
• We need to look for a
double dissociation; a loss
of implicit memory, without
loss of explicit memory
(having shown a loss of
explicit memory and no loss
in implicit memory in
amnesics.)
• Gabrielli et al. 1995 tested a
patient MS who had a right
occipital lobe lesion. Areas
18,19, leaving a LVF
hemianopia…
• When shown words for later recall, he had explicit memory for them, but
failed to show implicit perceptual priming (decreased exposure time needed
to identify a familiar word compared to time for an unfamiliar one).
• In detail: before both tasks a list of words were presented briefly and then
read aloud. (Note that this meant they saw them perceptually, but briefly
and also then heard the words). In the implicit memory task the words were
presented visually and then masked (XXXX). Durations of presentation
increased from 16ms until the word could be read. If they had implicit
memory, they should be faster with the words they had previously seen,
compared with new words.
44
In search of double dissociation between
implicit and explicit memory
• In the explicit memory task again a list of words was
shown and then in the recognition task old and new
words were presented and they had to identify the old
words. The patient was good at this task, but failed
to show implicit perceptual priming.
• Hence this lesion impaired implicit but not explicit
memory.
45
Gabrielli et al. 1995 Summary: Occipital lobe lesion patient
List of words
briefly
shown
List of words
only
heard
Results – patient good at recognition
task, but poor at implicit memory
– in other words, has a poor
implicit memory
Implicit memory
Word XXXX
old + new
RT task
Explicit memory
old + new
Recognition task
46
Evidence against
the Atkinson &
Shiffrin model
• Atkinson & Shiffrin proposed in their model above attended items go from
sensory memory into STM and if rehearsed move into LTM. Along the way
processes of decay and interference (either or both) results in information loss.
An important aspect was its serial nature – information moves from one stage to
the next. But the evidence doesn’t support this.
• Shallice and Warrington (1969), patient with left perisylvian* damage who had
reduced digit span (2 instead of the normal 5-9 items), but amazingly he had an
intact ability to form long-term memories (e.g. could talk about the latest news).
• Hence it cannot be the case that information going into short term memory is a
necessary precursor to it going into long term memory. It means that information
can go straight from sensory memory into LTM.
(*the perisylvian area is in the upper part of the temporal lobes in both hemispheres,
and is implicated in language functioning in the left hemisphere. It is typically
larger in the left hemisphere than in the right.)
47
Another double dissociation: this
time between STM and LTM
Below in order: Tim Shallice,
Elizabeth Warrington
and Alan Baddeley
This points to a double dissociation between STM and
LTM:
• The Shallice & Warrington patient shows a patient with
a very impaired STM, but with an intact LTM.
• We have also seen how it is possible for LTM
structures to be destroyed in the anterior medial
temporal lobes but the STM left intact in semantic
dementia.
• The patient HM is an illustration that the hippocampus
structures are important for the transfer/consolidation
of episodic information into LTM, even though HM’s
STM was intact and a proportion of the structures in
LTM were intact.
• Given the shortcomings of the serial model of memory,
there was a need for a different perspective, which
was the motivation for the working memory model
developed by Alan Baddeley.
48
Brain structures & • Going back to the Baddeley
model, there is some
STM
neurological evidence for the
brain structures involved in
this, in that function of the
phonological loop is impaired
by lesions of the supramarginal gyrus (Brodmann
area 40) and/or left pre-motor
region (area 44). [In the
absence of deficits in speech
comprehension and
production].
• Baddeley’s visuo-spatial
scratchpad is impaired by
lesions in parieto-occipital
regions.
• Right sided lesions have
more effect. Difficulty with
spatial tasks such as
repeating a sequence of
objects touched.
• Left sided lesions =>
problems with STM of
visually presented material.
49
50
Brodmann
51
Summary of memory
52
Levels of processing models
•
•
•
Another model to challenge the Atkinson
and Shiffrin was one proposed by Fergus
Craik and Robert Lockhart (1972).
They proposed the levels of processing
model in which items that were processed
“deeper” were consolidated into LTM.
Written words were presented in 3
conditions:
1. Ss had to identify if they were in upper
or lower case letters – this was a
superficial level.
2. Ss judged if words rhymed with each
other – an intermediate level as
processing for meaning still not
involved.
3. Ss had to make a judgement about a
word (e.g. Lamppost -can it rotate?) –
this was considered to be deep
processing.
53
Levels of processing models
•
•
•
Written words were presented in 3 conditions:
1. Ss had to identify if they were in upper or
lower case letters – this was a superficial
level.
2. Ss judged if words rhymed with each other –
an intermediate level as processing for
meaning still not involved.
3. Ss had to make a judgement about a word
(e.g. Lamppost -can it rotate?) – this was
considered to be deep processing.
Found that memory better if deep processing
involving semantic processing was used. This
is compared to information coded visually or
phonologically.
Thus there’s an incompatibility with the STMLTM formulation of Atkinson & Shiffrin because
it wasn’t about just holding information in STM
long enough, it was instead about the type
(superficial vs deep) of processing taking place.
54
Examples of animal
work of memory
• Studies in monkeys
help to clarify role of
hippocampus, in
conjunction with
surrounding
structures and
adjacent cortex in
memory.
• However functional
capabilities of
monkeys differ from
humans, so need to
choose a suitable
memory task for
animal experiments.
55
Animal models of memory
Delayed non-matching to sample
(or DNMS) task. Animal has to
identify which is the new item in a
pair. This is a test of declarative
memory.
• In (b) is shown object covering
food. Allowed to take food.
• In (c) hatch closed for a time
delay. In (d) shown old object (+)
and new object (on its right).
• Animal has to choose the new
object to get reward. In the picture
the monkey is making an error
and doesn’t get reward.
• Random sides to avoid position
bias.
• Thus has to learn to choose each
time a new stimulus – so has to
remember previous stimuli.
• This paradigm can test retention
in STM.
56
• Mishkin (1978) did the classic
studies with this – lesion to
hippocampus and/or
amygdala.
• Found deficits only if lesion
include the amygdala as well
as hippocampus proper.
• This was paradoxical with
regard to human studies, e.g.
RB (mentioned before briefly)
who had amnesia from lesion
restricted to CA1 cells within
each hippocampus, but not
including the amygdala.
• Followed up by Zola et al.
(1993). Extended Mishkin’s
work by creating separate
lesions of cortex surrounding
the hippocampus which had
been included in Mishkin’s
lesions of hippocampus and
amygdala together.
Animal models of memory
57
• Zola et al. found that lesions
of this surrounding temporal
cortex [para-hippocampal
gyrus, peri-rhinal cortex]
were sufficient when
hippocampus was damaged;
lesions of amygdala were
not necessary or sufficient.
• Overall conclusion is that it
is the hippocampus together
with its input and output
connections via surrounding
cortex which are important
for episodic memory.
• This is episodic memory,
although the time scale
seems a bit different.
58
Summary
Memory from the cognitive perspective
Introduction – different kinds of memories –
explicit vs. implicit and episodic vs. semantic.
Encoding, storage & retrieval. Learning vs.
memory. The Atkinson & Shiffrin model –
Sensory M, STM & LTM. Baddeley & Hitch
model. Implicit vs. explicit and the
perceptual representation system within
implicit memory.
Neurobiology of memory
Amnesia and the temporal lobes. The case of
HM – no HC bilaterally – STM normal,
implicit M normal. But transfer of episodic
memory to LTM blocked. HC important for
new explicit episodic memories.
The anterior medial temporal cortex – damage
=> dense retrograde amnesia – loses
decades of info.
59
Summary
Neurobiology of memory (continued)
Squire 1992: if HC damaged then pathway to ATL (ant. temp.
lobe) is blocked => severe anterograde amnesia (no new
episodic memories – as with HM). Implicit OK.
If HC OK, but ATL damaged, then severe retrograde amnesia
Tulving’s KC – damaged ATL – virtually no episodic, but intact
semantic & implicit. Shows separation of episodic and
semantic memories.
Double dissociation between explicit and implicit:
Amnesics
LVF hem
Gabrielli-MS
Explicit
Implicit
Lost
episodic.
OK
OK
Lost
ATL = anterior temporal lobes & LVF hem = LVF hemianopia
60
Summary
Neurobiology of memory (continued)
More recent evidence produces problems for the
Atkinson & Shiffrin model, esp with regard to
serial passage from STM to LTM.
Neurobiological evidence for separation between
STM & LTM and double dissociation:
STM
LTM
HC?
Comment
Shallice & W.
Impaired
Intact
OK
Otherwise OK-ish
Milner - HM
Intact
Some damage
No HC
Severe anterograde amnesia
Semantic dem.
Intact
impaired
OK
Severe retrograde amnesia
61
Summary
We also covered:
Brain structures and STM and in particular working memory.
There appear to be credible substrates for the model of
Baddeley and Hitch.
Levels of processing was covered.
Animal work as an example of further approaches to the
study of the biology of memory.
In conclusion these biological findings lend credence to their
corresponding psychological models.
62