Cognition Chapter 7
Memory Errors: Some Initial Examples
- 1992 plane crash into an 11-story apartment, killing 43 people & the plane’s entire
crew. 10 months later, 193 people were asked “did you see the television film of the
moment the plane hit the building?” More than half reported seeing it, although there was
no such film.
In a follow-up study, they surveyed another 93 people were asked whether
they had seen it on TV & more detailed questions about what exactly they had
seen in the film. 2/3 remembered seeing the film, & most of them confidently
provided details about it.
- Brewer & Treyens (1981): participants asked to wait briefly in the experimenter’s office
& after 35s they were taken out of the office & told that there actually was no
experimental procedure, instead the study was about their memory for the room. Their
responses were influenced by their knowledge of what an academic office typically
contains; not what it actually contained.
Memory Errors: A Hypothesis
- Recall that memory is a huge network of interconnected nodes & there are no
boundaries keeping memories of one episode separate from memories of other episodes
(not separate “files”). The various elements within each episode are held together by the
density of connections: there are many connections linking the aspects of your “trip to the
beach” to each other, there are fewer connections lnking this event to other events (less
dense; separates episode from others).
- Memory connections are retrieval paths; meaning they are a good thing. However, the
more links between episodes, the more you knit the 2 episodes together – resulting in
losing track of the boundary between episodes & higher likelihood of losing track of bits
of information contained within each event. Therefore, you become vulnerable to
“transplant” errors: when a bit of info encountered in one context is transplanted into
- Similarly, thoughts you had about an episode can become interwoven with the actually
memory, leading you to think that you the thoughts were actually a part of the original
Understanding Both Helps & Hurts Memory
- The connections encourage intrusion errors, in which other knowledge intrudes the
remembered event (read above).
- e.g. participants were given a passage to read about Nancy at a cocktail party, but some
participants read it with a prologue attached about how Nancy may be pregnant with her
professors child & she has money issues. The participants were all given the same recall
test. Those who read the prologue remembered much more, because understanding (&
meaning) promoted recall. However, the prologue also led people to include many things
that were not mentioned in the original episode; making 4x the number of intrusion errors
(e.g. The professor had gotten Nancy pregnant – this is not part of the story, it is implied).
The DRM Procedure - e.g. Reading a list of words that are all related (snooze, peace, blanket, dream) can help
participants remember the words, however participants are just as likely to recall the
word ‘sleep’ as they are to recall any of the other words, when in fact sleep was never on
the list! When asked to choose which words appeared in the original list (recognition, not
relaying) they make the same mistake.
This paradigm is the DRM procedure. It yeilds a large number of memory errors, even
when participants are put on their guard before the procedure begins. Apparently these
mechanisms that lead to memory error are quite automatic.
- The DRM procedure relies on word lists for the material to be remembered, but similar
errors occur with more complex material. You enter most settings with some background
knowledge, which guides your exploration, thoughts & interpretation of new situations.
However it can also be a source of memory error.
- e.g. You’re at a restaurant with a friend & this is a familiar setting to you (this
knowledge is referred to with the greek word schema/schemata). Schemata summarize
the broad pattern of what is normal in a situation. They also help you (you are not
confused when the waiter fills your glass or asks how things are). They also help with
recalling how events unfolded by filling gaps (things you may not have noticed or may
have forgotten) you may not remember what was on the menu, but you can be certain
there were menus; therefore you will likely include menus in your recall, even though
you have no memory of seeing them.
Evidence for Schematic Knowledge
- Schemata summarize you’re the broad pattern of your experience, & so they tell you
what is typical or ordinary. Reliance on schemata will therefore make the world seem
more “normal” than it really is & make the past seem more “regular” than it actually was.
e.g. Frederick Bartlett: participants presented with a story from Native American
folklore; when tested the participants did well but they made many errors in recalling the
particulars. The details omitted tended to be ones that made little sense to them; &
unfamiliar aspects were changed into more familiar aspects; parts of the story that were
inexplicable were supplemented to make the story seem more logical.
- Basically, elements that fit within their schematic frame remained in memory,
but elements that did not fit were dropped or changed.
- Remember the office study: people’s schemata’s of an academic office include a chair
The Cost of Memory Errors
- There are now roughly 300 cases of DNA exonerations – cases in which the courts
acknowledged that they sent an innocent person to jail; most commonly because of
Planting False Memories
- Loftus & Palmer (1974): participants shown a series of slides of car collisions. Half
were asked “how fast was the car going when it hit the other?” & the other half were
asked “how fast was the car going when it smashed into the other?” Participants in the first group (hit) estimated the speed to be 34 miles/hr & participants in the second group
(smashed) estimated 41 miles/hr. A week later, they were all asked in they had seen any
broken glass in the slides. The ‘hit’ group said (correctly) no they had not; the ‘smashed’
group often made the error that there had been broken glass. The change of ONE word
doubled the likelihood of memory error in this case.
Other procedures ask a question with incorrect information in the question, some are
exposed to the stories of other witnesses which contain misleading information, others
require participants to make up misinformation (where was the man bleeding from?).
- Some attempts at manipulating memory are more successful than others; plausible
memories are easier to plant than implausible ones, although it is possible.
- False memories are planted more easily if the participants imagine how the event
unfolded, rather than just hearing about it = imagination inflation.
We can use subtle procedures or more blatant procedures to plant false info in
Are There Limits on the Misinformation Effect?
- Misinformation effect: participant’s memories are influenced by misinformation after
the episode is over.
- Can also be applied to misremembering aspects of people.
- These errors are small, but they are also remarkably easy.
- EX. college students told that investigators were trying to learn how different people
remember the same experience. Students were given a list of events that had been
reported by their parents & the students were asked to recall these events as well as they
could, so they could compare their recall with their parents’ (some of trdse were made
up). At first, none of the students recalled the bogus events, but by a 3 interview, 25% of
the students were able to remember these events & supply details.
- One study, participants shown a photoshopped picture & led to recalling of a detailed,
vivid hot-air balloon ride. Another study showed the participants an unaltered photo of
their 2 grade class, leading participants to believe the experimenter actually did have
info about their childhood.
- Children are more prone to this memory planting; relevant to children’s testimony in the
- These results have pragmatic implications – e.g. participants changed their eating habits
after being told they had become ill after eating egg salad. Children may also accuse
adults of abuse when no abuse actually happened, & people also can confess to (&
apparently remember) crimes they didn’t commit.
Avoiding Memory Errors
- Over all, our memories are generally accurate.
- People sometimes announce when they are certain or uncertain of a recall. People are
more willing to take action based on confident memories, & juries and judges believe
that confident recall is likely to be accurate recall.
However, many studies have shown that in many circumstances, there is little
relationship between memory confidence & memory accuracy. - Our confidence in memory is influenced by factors that have no impact on memory
accuracy. When these factors are present, they will alter the level of confidence,
independently of the level of accuracy.
- e.g. Participants witnessed a simulated crime & asked to identify the culprit from a
group of pictures. Some were given feedback (e.g. “Good, you identified the suspect”),
others were not. The feedback increased the participants’ confidence, but not the
accuracy, therefore diminishing the link between confidence & accuracy.
The “Remember/Know” Distinction
- Are people less emotional when it comes to fictional memory? No, false memories can
be equally upsetting.
- Some investigators have examined the way a memory feels. Sometimes you can recall
the episode in which you gained the info, so you can say things like “I remember there
were books because I recall thinking about how dusty they were”, but if you can’t recall
the episode, you may say “I know there were books, but I don’t remember anything about
what they looked like”. The first is an example of “remembering”, the second is
“Remembering” is more likely in correct memories than false memories. False
memories usually only have a general sense of information & no recollection of a
particular. However there are exceptions in which a correct memory only arrives in your
thoughts with a feeling of knowing, or false memories arrive with detailed remembering.
So not reliable.
- A number of attributes are statistically somewhat more likely with accurate memories
than with errors (e.g. “remember” judgement). Also true for response “speed” – accurate
memories are recalled more rapidly. In some paradigms, there’s an association between
memory accuracy & confidence. However, these linkages are weak, therefore memory
errors are usually undetectable.
The Causes of Forgetting
- e.g. Forgetting someone’s name right after they tell you; this is not forgetting – it is
failure in acquisition (you barely paid attention, so you never learned it).
- For cases in which you did learn the info, the best predictor of forgetting is time
(retention interval: amount of time that elapses between the initial learning & the
One explanation for forgetting over time is decay: over time memories may fade,
perhaps because brain cells die off or perhaps because the connections among memories