Memory : Models, Forgetting, Flashbulb Memories (06-Feb-2003)
Multi-Store Model
Atkinson & Shiffrin (1971) proposed the two-process model of
memory, which has information being transferred from STM to LTM by a process
of rehearsal. This model has a number of strengths:
- case studies (e.g. Milner's study of "HM") appear to support it
- many studies (Miller, Peterson & Peterson, etc.) support this model
- this model stimulated a great deal of research into the way that memory
worked
On the other hand:
- case studies (e.g. Shallice & Warrington's study of "KF", who was
apparently able to form new LTMs without being able to form new STMs),
suggest that information can get into LTM without going through STM
- Miller's idea of "chunking" appears to show that the information flow
from STM to LTM is not just one way: for the STM to arrange information into
"chunks", it presumably has to draw on information in LTM
- much of the research into STM/LTM has been carried out in experimental
settings where P's are asked to remember meaningless series of data. This is
not representative of how people normally remember things
- rehearsal (i.e. repeating information over and over) appears to be
a very bad way to learn information; e.g. Bekerian & Baddeley (1980) reported
that few people remembered information about BBC radio frequency changes
despite having heard hundreds of advertisements
Levels of Processing (LOP)
Craik & Lockhart (1972) suggested that how well we remember things is
determined by how much processing is applied to the information. They define
three processing levels:
- superficial, or shallow
- intermediate, or phonetic
- deep, or semantic
Craik and Lockhart showed P's a list of words and asked questions about them.
One set of P's ("superficial") were asked questions such as "is it written in
capital letters?", the next set ("intermediate") were asked "does it rhyme
with ...?", and the third set ("deep") were asked "would it fit into a
sentence such as ...?". The P's were later asked to recall as many words from
the list as they could, with the following results:
- superficial : 15%
- intermediate : 35%
- deep : 70%
Craik & Lockhart conclude that maintenance rehearsal (simply
repeating something over and over again) is not an effective way to remember;
what is more effective is elaborative rehearsal, i.e. applying
some meaning to the information. Compared with the multi-store model, in
which the amount of rehearsal is the important factor, the LOP model
says that is the kind of rehearsal that is important.
Craik & Tulving (1975) performed an experiment which supported LOP; they
flashed a series of words to P's and asked them questions which would require
a different level of processing. Subsequently when asked to recall the words,
P's did significantly better for words for which "deep" processing had been
required.
According to LOP theory, memory is little more than a by-product of the way
that information is processed.
The LOP model is criticised because:
- It's difficult to define what "deep" means, and to measure how "deep" a
level of processing is occurring
- It offers an explanation for what is happening, but doesn't explain
why "deep" processing results in better memory
- An alternative explanation of why "deep" processing works is simply that
"deeper" levels of processing require more time, and therefore work simply by
providing more opportunity for rehearsal
Working Memory
Baddeley & Hitch(1974) proposed this model. While accepting a distinction
between STM and LTM, they argued that STM is more than just a first step
before LTM, and that it is made up of a components working together, rather
like a computer. The components include:
- the central executive is what pays attention to whatever
we're concentrating on, and delegates certain types of work to other components
- the visuo-spatial scratchpad (or "inner eye") comes into
play when, for example, we mentally walk through our house to count the
windows.
- the articulatory or phonological
loop (or "inner voice") is like a sub-vocal speech, so would be used
to count the windows that we "see".
- the acoustic, or phonological store is
independent of the articulatory loop, retaining the memory of acoustic
information long enough for it to be made sense of.
Some strengths of the Working Memory model are:
- PET scans on P's performing certain "STM" show that different types of
tasks stimulate activity in different physical regions of the brain,
suggesting that STM is more complex than the simple model put forward by
Atkinson & Shiffrin
- It has been shown (Baddeley 1975) that P's are more capable of performing
two tasks that require distinct WM components than when asked to
perform two tasks that use a common component
Weaknesses:
- Little is understood about the central executive, which is the
most important of all components
- Much of the evidence for this model has been derived from laboratory
studies; it's not clear how well these reflect the way that memory works in
present "real life" situations.
Forgetting
Suggested causes of forgetting in STM are:
- Peterson & Peterson (1959) suggested that trace decay
involves the gradual decay of a memory trace, in which items in STM will be
lost over time. However, Jenkins & Dallenbach (1924) showed that P's who
sleep remember more when they wake than P's who stayed awake for the same
amount of time. This suggests an alternate possibility:
- Waugh & Norman (1965) showed that displacement could account
for STM forgetting; as new information enters STM, existing information is
displaced, and therefore forgotten.
and in LTM:
- Bower (1969) showed that lack of semantic organisation
material meant that it was less likely to be recalled. Two sets of P's were
shown the same set of words, with one group of P's seeing the words in a
random order, and the other group seeing the words arranged into a meaningful
heirarchy. The second group of P's scored around three times better on
subsequent recall tests
- Craik & Tulving's experiment showed that lack of elaboration
meant that less information would be recalled
- Tulving & Pearlstone (1966) showed that forgetting can be a
retrieval failure caused by a lack of a cue. Cues can be
of different types:
- context-dependent ("external") cues are provided by
environmental factors. Abernathy (1940) showed that P's performed better when
they were were asked to recall information while in the same room as where
they'd learned it compared to P's who were asked to recall information while
in a different room
- state-dependent ("internal") cues are provided the internal
state of a P. Goodwin et al (1969) showed that alcoholics who were unable to
remember find money that they'd hidden whilst drunk, but could find it when
they became drunk again. McCormick & Mayer (1991) showed that we are more
likely to remember happy events when we feel happy.
Interference theory suggests that forgetting is caused by new
and old memories interfering with one another. In retroactive
interference, a new memory interferes with an old one; for example
having learned German and then French, you might come up with a French word
when trying to remember a German one. Proactive interference
occurs when an old memory interferes with a new one. Keppel & Underwood
(1962) suggested that interference may have been a factor in Peterson &
Peterson's experiment.
As we saw last week, forgetting can be caused
by a physical problem such as brain damage or disease.
Another possible cause of forgetting is repression as
described by Freud. In this case, a memory that is too uncomfortable is kept
in the unconscious. Bradley and Baddeley (1990) attempted to prove this
experimentally by testing P's recall of emotionally arousing words. They
found that while an immediate test indicated repression (P's recalled fewer of
the emotionally charged words), a test carried out 4 weeks later had the same
P's recalling more of the emotionally charged words. Eysenck & Wilson (1973)
suggest that the emotional arousal felt by the P's when they first hear the
words is significant, causing them to forget (rather than repress) the words
on the first test, but helping them to recall the words in the later test.
Flashbulb Memories
Brown & Kulik (1977) (see study sheet)
described flashbulb memories, suggesting that dramatic events can imprint a
powerful impression in peoples' memories, and argued that there may be some
physiological process involved in encoding such a memory. Such events as the
Kennedy assassination, or Princess Diana's death are examples that trigger
flashbulb memories. Neisser (1976) suggested that in fact FM are a result of
rehearsal - dramatic events are likely to be ones that we think of over and
over again.
Back to class notes index page