Transcript stmweb

The Study of Memory
Part 2 – Short Term Memory
2
Three Questions for Today
1. Why did researchers come to believe
in STM independent of LTM?
2. What do we think of those reasons
now?
3. Do we need the STM construct?
Why did people originally believe in
STM independent of LTM?
Because of STM – LTM differences in:
*
*
*
*
*
*
Loss following brain damage
Capacity
Duration
Type of code
Serial position effect
Mechanism of loss
3
Memory Loss Following Brain Damage
5
Memory loss following brain damage
Issue: is there a patient who can get new
information into STM but not into LTM?
If so, that selective impairment could be
used in an argument for an independent STM.
The most famous of all memory patients is
HM.
6
HM (Scoville & Milner, 1957)
Surgery to relieve severe epilepsy, in 1953,
at age 27.
* bilateral excision of medial temporal lobe
* after surgery, HM had profound anterograde
amnesia. Capable of little if any new learning.
* some retrograde amnesia.
7
Definitions (1)
Anterograde amnesia – inability to
remember things that happened after
brain damage. Implies inability to
encode new memories
Retrograde amnesia – inability to
remember things that happened before
brain damage. Implies inability to
retrieve existing memories.
8
In HM’s words:
"At this moment everything looks clear
to me, but what happened just before?
That's what worries me. It's like
waking from a dream; I just don't
remember."
9
HM – Psychological studies by Brenda Milner.
HM has:
* Good vocabulary
and language; normal
IQ
* No attention
disorder.
10
Things HM does not know
Where he lives
Who cares for him
What he ate at his last meal
What year it is
Who the President of the United States is
or how old he is.
In 1982, HM failed to recognize a picture of
himself that had been taken on his 40th
birthday in 1966 (13 years post surgery).
11
Definitions (2)
Declarative knowledge – knowledge
that you (a) know you have, and (b) can
can talk about. E.G., Your name.
Procedural knowledge – knowledge that
you have but may not know you have and
cannot articulate. E.G., How do you lean
a bicycle into a corner?
12
Declarative tasks – tests show:
HM cannot learn (and later recall) new
*
*
*
*
*
photographs of people
verbal material
sequences of digits
complex geometric designs
nonsense patterns.
He also cannot expand his digit span.
13
Procedural tasks – tests show:
Milner (1962) trained H. M. on a mirrordrawing task.
* HM, like normal people, improves with
practice. But he denies having practice.
Cohen and Corkin (1981) showed a
similar result on the Tower of Hanoi
puzzle.
Tower of Hanoi Puzzle
Tower
Rings
15
HM - Conclusion
Though HM can learn new procedures
he cannot acquire new declarative
learning.
* LTM impaired. But STM spared.
Argument in favor of view that STM
and LTM are independent.
Capacity
17
Capacity
If capacity of STM is different from
that of LTM, that supports view that
LTM and STM are independent.
Capacity of LTM is essentially infinite.
What is capacity of STM?
18
Capacity
Shepard & Tehgtsoonian (1961)
* Presented 200 3-digit numbers in a row.
* E.g. … 492, 865, 931, 758… 865, …
* Task: report when you hear a repeated
number
19
Shepard & Teghtsoonian (1961)
I.V.: Interval before repetition
D.V.: Probability of noticing repetition
* Repetition can only be noticed if first occurrence
is still in memory.
* Forgetting function: how does probability of
noticing repetition vary with interval?
* Question: Are there separate forgetting
functions for LTM and STM?
20
Shepard & Teghtsoonian (1961)
Result:
* P(noticing repetition) fell dramatically at
first
* Steep decline ended at interval = 7 items
* P(noticing) then fell more gradually,
asymptoting at 60%
21
S & T (1961) – Interpretation:
* Initial steep decline in P (noticing) occurs
because response coming from STM.
* Decline is steep because STM contents
decay quickly.
* More gradual decline occurs when
response depends upon LTM.
*Decline is gradual because LTM contents
decay very slowly if at all.
* Two forgetting functions – two memory
stores, one large and one small.
22
Question:
Why should STM have so small a
capacity?
Sensory memory has large capacity.
LTM has large capacity.
Why did we evolve a limited capacity
store between two large capacity
stores?
23
Answer:
If STM was any larger, it would take
too long to search through.
When we need information from STM,
to choose or guide a response, we need
it fast.
Things have to be processed fast in
STM…
Duration
25
Duration.
Issue: how long do STM traces last?
LTM traces last a long time – possibly
your whole life.
If STM traces last less time, that
supports the view that STM and LTM
are independent.
Duration – how long do STM contents
last?
Brown (1958) and Peterson & Peterson (1959)
Task: subjects briefly see a stimulus (e.g.,
BRG) and have to recall it after an interval.
Rehearsal is prevented by having them count
backwards during retention interval.
I.V. = length of interval in seconds.
26
% Correct as function of delay in Brown/Peterson task
28
Brown/Peterson paradigm
Result: for interval > 18 seconds, subjects
can no longer report stimulus.
Interpretation: there is a memory system in
which things must be rehearsed, or they are
lost.
But we don’t have to rehearse things in LTM
– so there must be a second memory system
– STM.
Type of code
30
Type of Code
Issue: every stimulus has multiple aspects –
e.g.
*
*
*
*
*
color
brightness
shape
category
name
Information about all these aspects is found
in LTM. Which are found in STM?
31
Brown/Peterson paradigm – again…
* Many studies used this paradigm in the
’60s. Most of the errors subjects made were
phonological – e.g., P for T.
* Errors based on shape were rare – e.g., C
for O.
* No semantic errors observed (or possible).
* Conclusion: STM uses a phonological code.
Serial Position Effect
33
Serial Position Effect:
In ordered recall, subjects recall a list
of words in the order they were given.
Out-of-order responses are counted as
errors.
Accuracy is higher for the beginning
and end of the list, lower for the
middle of the list.
%
correct
Position in list
35
Serial Position Effect
Better performance at beginning of list is
called Primacy Effect.
Better performance at end of list is called
Recency effect.
Theory:
* Primacy due to transfer to LTM (rehearsal).
* Recency reflects availability of items still in STM
Mechanism of Loss
37
Mechanism of Loss from Memory
How are things lost from memory – if at all?
Decay?
Interference?
Retrieval failure?
* LTM loss was blamed on interference
* STM loss was blamed on decay – as in
Brown/Peterson paradigm.
38
Three Questions for Today
1. Why did researchers come to believe in
STM independent of LTM?
2. What do we think of those reasons
now?
3. Do we need the STM construct?
39
The argument for independence of STM
Differences between STM and LTM:
*
*
*
*
*
*
Type of code
Serial position effect
Mechanism of loss
Patient data
Capacity
Duration
Do these reasons survive?
40
Type of Code
Original argument – any kind of code in
LTM, only phonological codes in STM.
We now know – that STM can contain
any kind of code.
* See, for example, Brooks (1968), and
Wickens ‘Release from Proactive Inhibition’
studies.
41
Type of Code
Shepard & Metzler’s Mental Rotation
Studies
* Pairs of abstract forms displayed
* Subject asked whether one is a rotated
version of the other.
* Have to mentally rotate one to see if it
ever matches the other.
* Mental rotation requires a visual code in
STM. (Why?)
43
Serial Position Effect:
Original argument – Primacy effect produced
by LTM, Recency effect produced by STM.
We now know – that both Primacy and
Recency effects can be found in pure LTM
studies (e.g., recalling U.S. Presidents).
Thus, recency effect cannot be taken as
“empirical signature” of STM.
44
Mechanism of Loss
Original argument – information lost
from STM through decay, from LTM
through interference.
We now know – that information can be
lost from STM through interference.
* E.g., Wicken’s Release from Proactive Inhibition
studies.
45
Duration
Original argument:
* newly-acquired memories must be
rehearsed to survive
* but older memories do not need to be
* therefore, new and old memories must be
in separate stores.
46
Duration
Alternative account:
* traces in LTM are vulnerable until they
have been consolidated.
* new items in LTM are more vulnerable to
loss than ‘established’’ items.
* so, vulnerable items could be in LTM
47
HM
If traces in LTM are vulnerable until
they have been consolidated, then HM’s
problem is that he cannot consolidate.
* He has normal digit span because new
items can be inserted in LTM.
* But he has anterograde amnesia because
new items cannot be consolidated in LTM.
48
Capacity
Many psychologists now say ‘Capacity’ means
capacity of the Articulatory Loop (AL)
* AL is used for rehearsal of information and
for planning articulation.
* AL is not a short-term memory.
* For one thing, you cannot search your
articulatory loop, the way you can search
memory.
49
Articulatory loop.
Capacity is determined by rate of loss.
You can rehearse about 7 items.
If you try to rehearse more than 7
items, the first ones will be lost before
you finish one cycle through the list
and go back to the beginning.
50
Articulatory loop in action (1)
Memory load = r l z t c j a
Articulatory loop rehearses:
r l z t c j a .. r l z t c j a .. r l z t c j a ..
‘r’ is still in loop when you finish ‘a.’
51
Articulatory loop in action (2)
Memory load = r l z t c j a m k s c p y
Articulatory loop rehearses:
r l z t c j a m k s c p y ..
‘r’ is no longer in loop by the time you
finish ‘y,’ so cannot be rehearsed – ‘r’ is
lost.
52
Three Questions for Today
1. Why did researchers come to believe in
STM independent of LTM?
2. What do we think of those reasons now?
3. Do we need the STM construct?
53
Do We Need the STM Construct?
No. Many cognitive psychologists argue that
we do not need STM in our memory theory.
We can explain all memory phenomena in
terms of LTM and the articulatory loop.
All we need is two premises:
* Limited capacity in articulatory loop.
* Items in LTM are vulnerable to loss until they
have been consolidated.